Fuzzy Control Systems for Power Quality Improvement—A Systematic Review Exploring Their Efficacy and Efficiency

Abstract

Fuzzy-based control systems have demonstrated a remarkable ability to control nonlinear processes, a characteristic commonly observed in power systems, particularly in the context of power quality enhancement. Despite this, an updated and comprehensive literature review on the applications of fuzzy logic in the domain of power quality control has been lacking. To address this gap, this study critically examines published research on the effective and efficient use of fuzzy logic in resolving quality issues within power systems. Data sources included the Web of Science and academic journal databases, followed by an evaluation of target articles based on predefined criteria. The information was then classified into seven categories, including control system type, features of the fuzzy logic controller, fuzzy logic inference strategy, power quality issue, control device, implementation methodology (efficacy testing), and efficiency improvement. Our study revealed that fuzzy-based control systems have evolved from simple type-1 fuzzy controllers to advanced control systems (type-2 fuzzy and hybrid) capable of effectively addressing complex power quality issues. We believe that the insights gained from this study will be useful to both experienced and inexperienced researchers and industry engineers seeking to leverage fuzzy logic to enhance power quality control.

1. Introduction

More than six decades have passed since the introduction of the fuzzy sets by Lotfi A. Zadeh [1]. It has also been 50 years since M. Mamdani [2] proposed fuzzy control based on Zadeh’s papers [3,4]. The first fuzzy logic controller (FLC) appeared in [5], where the authors presented a laboratory application used to control a steam engine. These articles had a significant impact on fuzzy control research, leading to thousands of control systems based on fuzzy logic being proposed in the literature since the original FLC 49 years ago [6].

In the field of power systems, and particularly power quality (PQ) [7], one of the primary fuzzy logic applications used to control reactive power was produced in 1990. Later, more researchers presented FLCs in the literature dedicated to controlling PQ using filters and specialized devices [8,9,10].

Table 1. History of fuzzy logic applications dedicated to power quality control.

Year	Published Works
1991	The authors of [11] introduce frequency control in power systems using an FLC with two inputs and an output.
1993	The authors of [12] propose an FLC for reactive power compensation using automatically switched capacitor banks employing TCR/TSC 1.
1994	Hiyama and his colleagues propose a PID 2 fuzzy stabilizer for voltage control [13].
1998	The control of a UPQC with fuzzy logic control was achieved in the research presented in [14].
2004	Jurado and Valverde demonstrate that voltage sags can be effectively eliminated by using a dynamic voltage restorer controlled by an FLC [15]. Kirawanich determines that harmonic distortion and power factor issues are improved using an FLC that commands a line conditioner [16].
2006	The authors of [17] present a Takagi–Sugeno FLC dedicated to operating an active power filter to eliminate harmonics.
2008	Using simulations made in MATLAB, the authors of [18] demonstrate the efficacy of a fuzzy hysteresis controller in controlling a UPQC.
2011	The authors of [19] consider balanced and unbalanced conditions when analyzing the performance of a shunt active filter controlled by an FLC.
2014	To improve the operation of the control system, the authors of [20] employ a type-2 FLC that controls an active power filter.
2015	A hybrid FCS that combines artificial neural networks and Takagi–Sugeno–Kang probabilistic fuzzy control performs reactive power compensation with superior results compared to conventional techniques [21].
2018	Ghafouri and his colleagues propose an adaptive fuzzy controller to stabilize power system frequency in consideration of microgrids and other distributed generation units [22].
2019	The authors of [23] propose a fuel-cell-integrated UPQC controlled by a hybrid fuzzy system to compensate PQ problems. The frequency control of modern multi-area power systems is solved with fuzzy logic in [24].
2020	The authors of [25] control a 24-pulse GTO-based STATCOM 4 using fuzzy logic. Echalih and his colleagues describe the use of hybrid fuzzy control to command a shunt active filter [26].
2021	Vanaja and his colleagues are researching the use of an FLC-controlled STATCOM in association with an enhanced second-order generalized integrator [27]. The voltage dynamic problem is solved using fuzzy logic considering the Malaysian power system [28].
2022	The authors of [29] propose the use of fuzzy fault-tolerant control and four-switch voltage source inverters to increase the stability of power systems.
2023	The authors of [30] present the use of ANFIS 3 to improve power quality in a microgrid.

¹ TCR/TSC—thyristor-controlled reactor/thyristor-switched capacitor, ² PID—proportional–integral–derivative, ³ ANFIS—adaptive neuro-fuzzy inference system, ⁴ STATCOM—static synchronous compensator.

shows a historical sequence of 17 FLCs proposed for PQ improvement in the literature [11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30]. A review of the works published in the 1990s reveals that FLC research was initially focused on reactive power compensation, power factor rise, and voltage and frequency regulation, i.e., power system stability [31]. Basic information on the structure of fuzzy control systems (FCSs) in power systems was also provided during this period [32]. In the later years of the decade, new prospects of fuzzy logic were considered for power quality control, such as the use of UPQCs (Unified Power Quality Conditioners). In the next decade, many researchers proposed FLCs to control other PQ issues, including harmonic distortion and voltage sags. Additionally, the effectiveness of fuzzy logic control combined with other artificial intelligence techniques like genetic algorithms was demonstrated in several studies [33,34]. In the last 15 years, there have been many papers focused on PQ improvement considering UPQCs, distributed generation, and other AI technologies in combination with fuzzy logic control.

Fuzzy control systems are an important component of intelligent control systems. The most significant advantage of fuzzy logic that attracts power systems specialists is its ability to handle imprecise, incomplete, and vague data, as it broadens the set theory so that a set’s element belongs to it based on a membership function value, as an alternative to conventional binary methods. Therefore, FCSs provide a way to deal with uncertainties through linguistic values and logical inferential rules, allowing specialists to perform dynamic modeling and controller descriptions using simple linguistic statements [35]. FLCs have been criticized due to the absence of a systematic design and a stability analysis method, plus the financial aspects regarding the replacement of traditional PID controllers [36]. Then again, in many cases, they are the first option when working with non-linear, time-delayed, and high-order systems [37], and are overtly acknowledged as superior replacements for their PID counterparts [8].

Modern power systems are complex and often experience uncertainty due to the diverse range of electricity consumers and prosumers they serve, as well as the presence of distributed generation units and interconnected microgrids. The integration of different generation units that rely on renewable resources such as solar energy and wind can lead to PQ issues, including voltage sags, voltage swells, voltage transients, harmonics, voltage oscillations, unbalance, and low-power factors [38]. In addition, devices and equipment with nonlinear characteristics, as well as unbalanced and varying loads, can cause non-sinusoidal and unbalanced currents and voltages, voltage fluctuations, and flicker in utility electricity networks and consumer installations [39,40].

PQ disturbances in power systems can cause a range of problems, including supplementary power losses, premature aging of components, low-power factors, equipment malfunction, and even total failure [41,42,43]. To help prevent these issues, international standards have been developed by organizations such as the European Norms, American National Standard Institute, IEC, and IEEE. These standards provide guidelines for power-generating companies, consumers, manufacturers, and national power system organizations on the acceptable limits of PQ disturbances.

Table 2. Power quality limits.

Issue	Index	Limits		Standard/Observations
		LV	MV
Frequency variation	Δf% 1	±1%		EN 50160
Voltage sag	Δus% 2	10–95% of fundamental		IEC 61000-4-11/ 0.5 cycle—several seconds
Harmonics	THD 3	8%	5%	IEC 61000-4-7, IEEE 519 2022
Unbalance	VUF 4	2% (3%)	1%	IEC 61000-2-5/ IEC 61000-2-12
Voltage transients	ΔV% 5	5%	3%	IEC 61000
Voltage fluctuations/Flicker	Pst 6 Plt 7	1 0.65		IEC/EN61000-3-3
Reactive power/Power factor (PF)	PF	0.9
Voltage interruptions	Δui% 8	more than 95% of fundamental		IEC 61000-4-11/ 10 ms–60 s

¹ Frequency deviation (error); ² voltage deviation; ³ total harmonic distortion factor; ⁴ voltage unbalance factor; ⁵ voltage deviation; ⁶ short-term flicker perceptibility; ⁷ long-term flicker perceptibility; ⁸ voltage deviation.

shows the limits of PQ disturbances according to IEC 61000, EN 50160, and IEEE 519-2022 [44,45]. The thresholds of the PQ indices are the reference magnitudes against which the adjustment of voltages and currents are made.

After analyzing the information provided above, it is important to understand the significance of utilizing appropriate control systems, such as FCSs, to improve power quality. To achieve an effective and efficient controller, one must have a sound knowledge of the structure and limitations of FCSs, as well as explore their possibilities. Although FCSs have been studied and analyzed in various engineering fields, such as refrigeration, and hydraulic and pneumatic systems [46,47,48,49], only a few authors have focused exclusively and intensively on PQ improvement in power systems in recent years. These studies emphasize the usefulness of fuzzy logic and explore different aspects of control systems and power systems. In the field of microgrids, a study [50] was conducted to examine the effectiveness of fuzzy control. The study concludes that a simple FLC with only one input can provide comparable results to a traditional PI (proportional–integral) controller, but it is easier to adjust. Another study [10] classified the different types of UPQCs, along with their corresponding control strategies, to eliminate or reduce PQ issues. The authors of this study mention that fuzzy logic controllers in association with other artificial intelligence techniques have the capability to improve PQ. Similarly, in a review [51], voltage variation compensators are examined, and the authors conclude that employing fuzzy logic in the control of these compensators can improve PQ. Devices such as active power filters are also reviewed in [10,52,53], and different control methods are discussed. One significant conclusion from these studies is that a PI controller in association with fuzzy logic provides better results than a conventional PI controller in PQ improvement.

Upon reviewing the available literature, it is evident that there are still unanswered questions regarding the use of fuzzy control for improving power quality. These questions can only be addressed by systematically analyzing the literature. Thus, the main objective of this work is to provide a comprehensive review of the applications of fuzzy logic control in power quality issues, with a focus on FCSs’ effectiveness and efficiency. The review covers information on fuzzy-based control systems, input and output values, fuzzy numbers, inference rules, decisions, and defuzzification methods, as well as the implementation of fuzzy logic. The work focuses on control systems that contain fuzzy logic controllers in their simple (type-1 fuzzy) and advanced (type-2 fuzzy) forms, or in association with other conventional techniques, such as PI or hysteresis control, and novel techniques like genetic algorithm optimization. In the research, ANFIS was not considered; only control strategies that maintain the classic structure and logic-based functioning of an FLC were considered.

The second section of the work details the methodology employed to obtain the articles used in the review and the criteria to assess the literature. The third section presents the results of the research concerning bibliometric aspects and the seven criteria (technical characteristics): control system, FLC features, FLC inference strategy, PQ issue, control device, implementation, and assessing efficiency. This section also enlists the most relevant research. The next section of the work discusses the findings and the future work that can be completed in this area. Special attention is given to different types of PQ issues and the types of control systems used to solve them. The last section concludes with the main results of the review and underlines future directions of the present work.

2. Methodology

This literature review presents an organized analysis of relevant international studies on the topics of power quality improvement, fuzzy logic control, and devices for PQ control. To initiate the study, a comprehensive search was conducted for surveys, reviews, and state-of-the-art reviews that focused on the aforementioned subjects. These works were subsequently used as the foundation for the research and selection process of the review’s portfolio. Following the trends of literature reviews [10,54], a thorough examination of diverse databases was executed to obtain information on the content of the papers to make up the final set of papers. The Web of Science database was the primary source scrutinized to obtain the list of papers, considering the data found in the titles, keywords, and abstracts of papers. Only the research areas of “Engineering,” “Automation Control Systems,” “Mathematics,” and “Computer Science” were considered. The resulting unfiltered list was then meticulously evaluated, and papers that did not conform to the research scope were eliminated. This action led to the initial set of papers. The primary search keywords for the papers were as follows:

“power quality” OR “harmonics” OR “frequency” OR “voltage sag” OR “unbalance” OR “reactive power” OR “voltage variation” OR “power factor” OR “flicker”

AND

“fuzzy control” OR “fuzzy logic control” OR “fuzzy controller” OR “fuzzy logic controller” OR “FLC.”

Secondarily, the papers’ main sources were accessed, among which were IEEE Xplore, MDPI journals, Taylor & Francis journals, Elsevier journals, and Wiley journals. Further, the papers were again selected based on a more detailed content analysis. It must be emphasized that the selected papers that went in the final set and additionally were further analyzed and listed as references were the ones that respected the review’s eligibility criteria, i.e., presentation of the efficacy and effectiveness of the fuzzy logic control systems in the process of power quality improvement. Because of this, more than half of the initial list of papers were ignored in the next stage of the study.

To provide a timeframe for the study, the research focused on studies published in the last 30 years, with particular emphasis on papers from the last 15 years. However, eligible works were included in the final set regardless of publishing year. The articles were classified based on the criteria outlined in

Table 3. Assessment criteria of the article set—overview of the criteria used to classify and assess the articles in the final set.

Analysis Criteria	Details	Examples
Control system	Control systems may include not only a typical FLC, but also other elements from conventional methods or other AI techniques.	Typical FLC (Type-1 fuzzy)
		Fuzzy-PI Hysteresis fuzzy control
		Fuzzy-PID logic controller
FLC’s features	The number of inputs and outputs of the controller, the number of linguistic variables, the types of fuzzy membership functions, and defuzzification methods.	Triangular, trapezoidal, and gaussian fuzzy numbers Center of gravity defuzzification method
FLC’s inference strategy	The inference strategy refers to the way the output is obtained. It is implemented in the FLC to realize the aggregation of the rules and combine them.	Mamdani Takagi–Sugeno
PQ issue	The type of disturbance affecting the power quality in power systems [38].	Frequency variation
		Voltage variation, harmonics, unbalance
Control device	Devices used for PQ improvement depend on the type of PQ issue and its location in the power system [39].	SVC UPQC DVR STATCOM
Implementation (efficacy proof)	The control systems were implemented and tested using simulations, that is, employing appropriate environments like MATLAB and DIgSILENT, or experimental laboratory or infield tests.	Simulations Laboratory or on-site hardware implementation
Assessing efficiency and improvements	Was the control system assessed for improvement and/or was the fuzzy controller tested considering different fuzzy numbers, inferences, and/or defuzzification methods?	Yes/No Yes—testing diverse features of FLCs or enriching FLCs with additional elements

. Additionally, taking into account the methodology from reference [10], the table includes detailed information about the criteria and their characteristics.

3. Results

3.1. Bibliometric Analysis

The set of papers considered in this review includes publications from journals and conference proceedings. The results of the paper selection process from the Web of Science were as follows:

• Initial unfiltered set of papers—460 papers. This set contains the raw list of papers used in the review.
• Initial set of papers—278 papers. The initial set is a list of papers that correspond to the subject of power quality improvement using fuzzy control.
• Intermediary set of papers—122 papers. These articles were the most relevant of the initial set of papers that were published in the last 15 years (2009–2023).
• Final set of papers—135 papers. The final set of papers that was considered included the previous ones plus 13 more papers available before 2009. Going forward, all recommendations and observations we make are regarding this final set of papers.

The first noticeable applications of fuzzy control in PQ were in the 1990s. An assessment of the publications from the 1990s and 2000s showed that, yearly, less than ten articles were published in the research area of interest. Later, the number of articles grew substantially, with a maximum of 43 in 2016.

When focusing on the most relevant journal, IEEE Access was the most prolific, with 13 publications, followed by Electric Power Systems Research, IEEE Industrial Electronics Society, and IET Generation, Transmission & Distribution, with more than 8 articles each. Regarding conference proceedings, relevant papers from the initial set were distributed among many of them, but the majority were found in the proceedings published before 2020.

The most cited paper found through the selection was “Load Frequency Control of a Multi-Area Power System: An Adaptive Fuzzy Logic Approach” [55], with 207 citations in the Web of Science. The paper introduces fuzzy logic for the adaptive (type-2 FLC) control of load frequency and demonstrates through simulations that the proposed control system is more efficient than a PID system and a simple type-2 fuzzy controller.

3.2. Technical Analysis

The technical analysis aimed to evaluate the articles based on the seven specific criteria. To achieve this, a separate subsection is dedicated to each criterion. These subsections provide statistical results regarding the corresponding criterion’s characteristics and list the papers significant to each criterion.

3.2.1. Control System

Control systems based on fuzzy logic have a structure that may include not only the body of a typical FLC, but also other elements borrowed from conventional control (PI, PID, and hysteresis) or other AI techniques, such as neuronal networks and evolutive algorithms. The types of control systems that have extra elements are hybrid fuzzy controllers, which in the literature we can find as fuzzy-PI, fuzzy-PID, hysteresis fuzzy, self-tuning fuzzy, adaptive fuzzy, and neuro-fuzzy controllers.

The structure of a typical FLC (type-1 fuzzy) consists of three blocks: fuzzification, inference, and defuzzification blocks, as shown in [12]. The structure of a fuzzy-based hybrid system additionally contains blocks such as P, I, and D blocks [9], hysteresis-band blocks [18], and second-order generalized integrators [27]. A type-2 fuzzy controller has the same structure as a type-1 fuzzy controller, plus a type-reduction block positioned before the defuzzification component [55]. To increase the performance of the control process, authors use evolutionary algorithms such as genetic algorithms [33,34] and machine learning algorithms like artificial neuronal networks, which lead to ANFIS [30].

In the review and research process, we considered papers that proposed typical and hybrid fuzzy-based control systems that maintained the general structure of a fuzzy logic controller. As a result, this work does not include detailed information about adaptive neuro-fuzzy inference systems or other similar control systems. Consequently, we excluded more than 18% of the total number of papers from our initial filtered set of 278 papers due to the type of control system. The number of excluded papers was 53, with 26 of them being published within the last five years, of which [56,57,58,59,60,61,62,63,64,65] are the most significant.

The complete list of types of fuzzy-based control systems and their numbers is displayed in Figure 1. We ought to mention that we added genetic algorithm fuzzy control systems to the final list because the additional components do not change the structure of the fuzzy controller, and they are used to increase the efficiency of control systems. The chart indicates that the most popular type of control system is based on a type-1 fuzzy controller, with 49 papers proposing it. The other types of control systems are fuzzy adaptive with 19 papers, and fuzzy-PI with 17 papers. A time–distribution analysis of these control systems’ appearance shows that they have been proposed mostly in the last 15 years. The types of control systems characterized by small numbers of published papers were mainly proposed in the last five years.

Table 4. Relevant papers considering the type of fuzzy-based control system.

Reference	Control System	Description	Efficacy	Efficiency
[66]	Type-1 fuzzy	Supercapacitors are proposed and controlled to reduce THD and regulate reactive power for wind farms	Tested through simulations	Comparison between the system with and without supercapacitors
[67]	Fuzzy-PI	Multi-feeder UPQC fuzzy controlled to reduce voltage and current imperfections	Software implementation for efficacy evaluation	Results compared with classical PI and type-1 fuzzy controller
[68]	Fuzzy-PID	Frequency and power control in insolated distribution generation units, considering also superconducting magnetic energy storage	Efficacy assessment through software implementation	Results compared with only PID and PID plus superconducting magnetic energy storage
[69]	Hysteresis fuzzy	Hysteresis controller bands adapted by a fuzzy controller for harmonics elimination	Demonstrated through simulations	The proposed method compared hysteresis with fixed bands with zero fixed-band controllers
[70]	Adaptive fuzzy	Inverters controlled for power factor tracking changes and power fluctuation decrease	Software implementation for efficacy evaluation	The proposed controller contrasted with conventional PI and Takagi–Sugeno probabilistic fuzzy controller
[71]	Type-2 fuzzy	Load frequency control and frequency stability considering the high penetration of distributed generation	Tested through simulations	Comparison with PI and type-1 fuzzy controller
[72]	Self-tuned fuzzy-PI	Mitigation of voltage sags, voltage, and THD by using a dynamic voltage restorer and two FLCs	Efficacy assessment using simulations	Comparison between the system with and without dynamic voltage restorer
[73]	Fuzzy fractional-order PI	Single-phase active power filter controlled to increase power factor and limit harmonics	Simulations and experimental testing of the efficacy	Proposed methodology with the controller without fuzzy usage
[74]	EV-fuzzy-PID	Voltage and frequency control with automatic voltage regulator and particle swarm optimization	Software implementation for efficacy evaluation	Results show the comparison of the proposed method with another hybrid method from the literature
[75]	Tuned fuzzy-PID	Frequency regulation employing static synchronous series compensator (SSSC) and teaching learning optimization technique	Efficacy assessment through software implementation	The proposed method compared with the control without SSSC and with the genetic algorithm

enumerates 10 relevant papers which contain detailed information about the type and structure of the proposed control systems. The table presents the control systems’ type and description, plus data about the control systems’ efficacy testing and efficiency demonstrations. The articles from the table relate to the first 11 types of control system proposed in the literature: type-1 fuzzy, fuzzy-PI, fuzzy-PID, hysteresis fuzzy, adaptive fuzzy, tuned, type-2 fuzzy, evolutive algorithm fuzzy, fractional-order fuzzy, hybrid fuzzy, and self-tuned fuzzy controllers.

3.2.2. Fuzzy Logic Controller Features

During the development of an FLC, researchers must consider several critical aspects: the inputs and outputs, the number of linguistic variables, the type of fuzzy numbers used in the fuzzification block, and the defuzzification method. Additionally, the FLC’s rule aggregation strategy, or inference strategy, is a vital consideration. Analyzing the papers from the final set, we found the following about the features of the proposed fuzzy controllers by the researchers:

• Number of inputs and outputs: The most used configuration is two inputs and one output, with a percentage of more than 75%, of which [76,77,78,79] and [80,81,82,83,84,85,86,87] are typical examples. This aspect is seen clearly in Figure 2, where other configurations that appear in more than one article are one input–one output [88], two inputs–three outputs, two inputs–two outputs, and four inputs–two outputs.
• Number of linguistic variables: The most popular number of linguistic variables used in the fuzzification stage for the inputs and outputs was seven [78,79,82,88], followed by five and three [76,80,83]. Considering the differences between the inputs and outputs, the proposed FLCs had the same number of linguistic variables for both inputs and outputs [76], but there were also FLCs characterized by different numbers of linguistic variables [83,87]. This aspect can be seen in

  Table 5. Relevant papers considering the features of fuzzy logic controllers.

  Reference	Inputs and Outputs	Number of Linguistic Variables	Fuzzy Numbers	Defuzzification Method	Efficacy	Efficiency
  [76]	2->1	5	Triangle singleton	Weighted average	Processor-in-the-loop technique and simulations	Comparison with a simple repetitive controller
  [77]	2->1	5 7	Triangle and trapeze triangle	Weighted average	Simulations	Demonstrated by relation with conventional droop control
  [79]	2->1	7	Gaussian	Weighted average	Simulations	Comparison with PI controller
  [88]	1->1	7	Triangle and trapeze	Centroid	Simulations	Comparison with a hysteresis controller with fixed bands
  [80]	2->1	3 5 7	Gaussian triangle and trapeze	Centroid	Simulations	Results comparison without the control system
  [74]	2->3	4 3	Triangle singleton	Weighted average	Simulations	Proposed method compared with other methods from the literature
  [89]	2->2	3 5	Singleton triangle and trapeze	Weighted average	Simulations and hardware implementation	No additional methodology is tested, only the proposed method considering diverse situations
  [75]	2->3	5	Triangle and trapeze	Centroid	Simulations and hardware testing	The proposed method compared with the control without SSSC and with the genetic algorithm
  [85]	2->1	3 5	Triangle singleton	Weighted average	Experimental setup and simulations	Proposed adaptive fuzzy with type-1 fuzzy
  [87]	2->1	3 5	Triangle gaussian	Centroid	Simulations	Results comparison without the control system

  . Thus, we introduced both numbers of variables in the assessment. The chart from Figure 3 illustrates the number of linguistic variables.
• Type of fuzzy numbers (membership functions): The authors extensively used a combination of trapezoidal and triangular membership functions [78,81,83,88], as shown in Figure 4. Close behind are the utilization of the triangular function, the Gauss function [79], and singletons [76]. Other functions used in the fuzzification stage, but rarely, are the bell-shaped, trapeze, and sigmoid functions. Like the number of linguistic variables, we found FLCs that use the same type of function to describe the memberships of the inputs and outputs [86], but also FLCs with different membership functions for the input and output quantities [87].
• Defuzzification method: The center of gravity [78,81,86] is the defuzzification method normally used to obtain crisp outputs, with a percentage of 41 (55 papers). The other defuzzification methods are bisector, weighed average [76,77], and singleton, which together appear in 28 articles (20%). Unfortunately, the rest of the 51 papers did not contain clear information about the defuzzification method, even though there were data about other features of the FLC that implied the use of certain defuzzification methods. For example, when using the Mamdani style of inference from MATLAB’s Fuzzy Logic Toolbox, the predefined defuzzification method is the centroid—the center of gravity—or, when using the Takagi–Sugeno style of inference, the authors can define corresponding functions to determine the crisp outputs.

Table 5 presents 10 articles that contain whole data about all features of the proposed FLC. The table’s header is divided into the four FLC features, the reference number, and the fuzzy-based control systems’ efficacy testing and efficiency demonstrations.

3.2.3. Fuzzy Logic Controller’s Inference Strategy

There are three main types of fuzzy control system inference strategy that are classified by the way the logical rules between the inputs and outputs are built: Mamdani, Takagi–Sugeno, and the singleton strategy [6,90].

Evaluating the final set of papers to determine the utilized inference strategy, we found that the most employed inference strategy used by the inference engine was Mamdani’s. Figure 5 shows this aspect, as 41% of the authors of the analyzed articles mention the use of Mamdani’s inference strategy [91]. The next most used was Takagi–Sugeno [92], followed by the Mamdani’s “Min-Max” method [93] (declared by the authors, as 18 mentioned Mamdani’s “Min-Max” method, or simply the “Min-Max” method). The other strategy is the singleton methodology. Unfortunately, more than 27% of the papers did not contain clear information about the inference strategy. Consequently, we did not consider these studies further when selecting the most relevant papers, which are listed in

Table 6. Relevant papers considering the inference strategy of the FLCs.

Reference	Inference Strategy	Efficacy	Efficiency
[90]	Takagi–Sugeno	Simulations	Proposed method results are compared to results obtained with a PI controller or without a PQ improvement apparatus
[91]	Mamdani	Simulations and experimental tests	Comparison with a PI controller
[92]	Takagi–Sugeno	Simulations	Comparison with a conventional PI controller
[94]	Mamdani	Simulations	Comparison between different versions of the proposed fuzzy-based on-load tap changer control (OLTC) and a classic OLTC
[95]	Product inference Singleton	Simulations	Comparison with a PI controller
[96]	Explained but not classified	Simulations and experimental tests	Simulation results of the proposed method compared with results associated with a PI controller
[89]	Takagi–Sugeno	Simulations and experimental tests	No additional methodology is tested, only the proposed method considering diverse situations
[97]	Mamdani	Simulations	Comparison with a PID controller
[98]	Takagi–Sugeno	Simulations	Proposed adaptive fuzzy compared with conventional fuzzy controller
[99]	Min inference	Simulations	Results comparison without the control system

.

Table 6 introduces 10 significant articles that clearly describe their inference strategies; thus, it is easy for the readers to understand these aspects of fuzzy control. The table below gives data about the references, the inference style, the proposed method’s efficacy, and the efficiency assessment. One can observe that five papers from the table used Takagi–Sugeno fuzzy strategies, but this aspect does not follow the data from the chart in Figure 5. We made this choice as the authors of these articles gave greater descriptions of their methodologies.

3.2.4. Power Quality Issues

The power quality issues described in the analyzed papers were very varied, from frequency variations to high-voltage THD, from voltage sags/swells to low-power factors. Considering the number of PQ issues considered, 88 studies focused on one problem, whereas the other 46 were on multiple issues. Figure 6 illustrates these aspects, as one can see that the first seven categories describe only one disturbance, e.g., voltage sag/swell or harmonics/THD, and the last eight categories describe multiple disturbances, such as frequency variations, harmonics/THD, and reactive power (stability power systems).

Assessing the data from the chart in Figure 6, we deduced that diminishing harmonic distortions and decreasing THD was the goal of almost a quarter of the studies that used fuzzy logic control, i.e., 33 papers. Furthermore, the number of studies that focused only on one disturbance represented more than 65% of the total of 134 analyzed papers. The next most studied PQ issues were voltage variations (category 6) and the combination of frequency variations, harmonics/THD, and reactive power (category 8), which influence the stability of power systems. Closely behind, 12 studies focused on the power system’s frequency control and the combinations of harmonics with reactive power, frequency variations, voltage variations, harmonics/THD, and reactive power.

Table 7. Relevant papers considering power quality issues.

Reference	Power Quality Issues	Description	Efficacy	Efficiency
[100]	Voltage (sag/swell, imbalance), frequency, real/reactive power, and harmonics	The inverter is controlled using the fuzzy space vector pulse-width modulation (FSV-PWM) technique	Simulations	The proposed method compared with the conventional ST-PWM control
[101]	Harmonics reduction	Seven-level modular multilevel converter (SLMMC)-based shunt active filter controlled by type-1 fuzzy	Simulations	The proposed method’s results compared with PI—SLMMC control and without filter
[102]	Load frequency control (LFC)	Hybrid fuzzy control for inner loop control and genetic algorithm to optimize the control parameters	Simulations	The proposed method compared with an improved PI and adaptive fuzzy controller
[103]	Voltage sag and flicker	Superconducting magnetic energy storage (SMES) with hysteresis fuzzy controller	Simulations	Results comparison without the control system
[104]	Voltage sag, harmonics, sudden load change	Power electronic distribution transformers and adaptive PI fuzzy controller	Simulations	Comparison with a PI controller
[105]	Voltage control and reactive power	Capacitor banks and on-load tap changer in substations controlled by dynamic programming and type-1 fuzzy	Simulations	The proposed method’s results compared with results obtained only using dynamic programming
[106]	Harmonics and reactive power	Three-level shunt active power filter (APF) controlled by type-1 fuzzy	Simulations	Comparison with a PI and digital RST controller
[107]	Harmonics	Takagi–Sugeno fuzzy controller and shunt active power filter	Simulations and laboratory testing	Results comparison without the control system
[108]	LFC	Indirect adaptive fuzzy control for multi-area power system	Simulations	Comparison with a PID controller
[109]	Harmonics and reactive power under unbalance	Type-1 fuzzy controller used to optimize the energy storage of a DC capacitor voltage of a three-phase shunt active power filter	Simulations	Results comparison without the control system and APF

lists 10 papers that are relevant to the categories mentioned in Figure 6. The table provides information about the PQ issues addressed in each paper, a brief summary of the proposed method, and the results of the efficacy and efficiency testing of the fuzzy-based control systems.

3.2.5. Control Devices

Power quality control devices are introduced to eliminate and/or diminish the disturbances that negatively affect the quality of electrical energy. These devices are classified depending on their operation, physical structure, applications, and when they were first pioneered. The devices were comprehensively synthesized in the review articles [10,68,69]. In this review, we divided the control devices by their applications. Subsequently, the complete list of these devices that we found in the analyzed set of papers is summarized in Figure 7: APF (active power filter—47 papers), STATCOM, UPFC (Unified Power Flow Controller), UPQC (Unified Power Quality Conditioner), FACTS (Flexible Alternating Current Transmission System), SVC (Static Var Compensator), DVR (dynamic voltage restorer). In addition to this list, as shown in Figure 7, there is the “Controller” category, which includes the applications where the researchers combine diverse components of the power system, i.e., converters, generators, storage systems, transformers, power line conditioners, and capacitor banks, with the fuzzy-based controller to improve power quality issues. Of these devices, the most popular are the converters, with 25 appearances, followed by storage systems in 10 papers.

Table 8. Relevant papers concerning control devices.

Reference	Control Device	Description	Efficacy	Efficiency
[110]	Controller (asynchronous motor)	Type-1 fuzzy control asynchronous motor to increase power factor and regulate voltage	Simulations	Results comparison without the control system
[111]	APF	Hybrid automata–fuzzy control for THD and power factor decrease	Simulations	Comparison with a simple PI
[112]	STATCOM	Rise of STATCOM efficiency through sub-synchronous resonance using type-1 fuzzy	Simulations	Results comparison without the fuzzy control of STATCOM and active disturbance rejection control
[113]	UPQC	Mamdani and Takagi–Sugeno fuzzy controllers and phase-locked loop (PLL) control Harmonics, unbalance, and voltage variations	Laboratory experimental setup, hardware in loop real-time system	Comparison with conventional modified PLL and TS-PLL control strategy
[114]	DVR	Self-tuned type-2 fuzzy PI controller for voltage sag and THD alleviation	Simulations	Comparison with a type-1 fuzzy PI controller
[115]	UPFC	Type-1 fuzzy controls with active and reactive power variations	Simulations	Results comparison without the fuzzy control of UPFC and PID control
[116]	Controller (inverter)	Fuzzy supervisory with fuzzy PID control for voltage and current THD reduction	Simulations	Comparison with classic supervisory PID and classic supervisory FPID
[117]	Controller (inverter)	Direct power control (DPC) based on type-1 fuzzy and fuzzy PI for THD diminishing	Simulations and laboratory experiments	Results comparison with the basic DPC
[118]	APF	Review about the control strategies of series APF for decreasing THD, fuzzy hysteresis method	Simulations	Comparison with a simple fixed-band hysteresis, adaptive hysteresis band, and fixed hysteresis band
[16]	Power line conditioner (PLC)	Line current harmonic distortions and power factor Fuzzy PI controller	Software and hardware implementation	Comparison with a simple PI and a gained scheduled controller

is a collection of data that show information on 10 important articles. These articles used various control devices to improve the quality of power and employed FCSs. The table provides details about the characteristics of each study presented in the selected papers, which are from each category of the chart in Figure 7. Moreover, the table also gives information about the efficacy testing and efficiency demonstration of the proposed control methodology.

3.2.6. Implementation (Efficacy Proof)

To verify the functionality of the control systems proposed in their papers, the researchers employed two strategies. The first strategy is based on simulations and implies the use of a dedicated simulation environment to model the controlled process together with the control system using predefined blocks or blocks and functions designed and developed by the researchers. The most popular software used to perform simulations is MATLAB, which has specialized components like Simulink and Fuzzy Logic Toolbox [118]. Other software found rarely in the papers included DIgSILENT, dSPACE, C++, and PSCAD/EMTP. The second strategy used to show the efficacy and efficiency of the proposed control methodology is physical experimental implementation. When approaching the experimental assessment, the researchers used down-scaled and equivalent power system models, signal generators, qualified measurement apparatus, dedicated sensors and transducers, and specialized microcontrollers.

In the analyzed papers, we found studies that used simulations and papers that utilized both simulations and laboratory (real-time) experiments. Figure 8 graphically shows the yearly distribution of implementation strategies employed in the surveyed articles. The chart’s columns underline the preference for simulations over hardware laboratory experimentations, as software implementations represent more than 80% of the total number of 135 implementation methods used in the analyzed papers.

Table 9. Relevant papers considering the implementation of fuzzy-based systems.

Reference	Description	Implementation	Efficiency
[121]	Voltage stability, active and reactive power control in island microgrid Fuzzy adaptive impedance controller	RT-LAB experiments MATLAB/Simulink	Efficiency testing considering diverse operating states.
[122]	Adaptive virtual capacitor and rotational inertia control based on fuzzy logic (virtual synchronous generator VSG strategy) for frequency and voltage stability	RT-LAB experiment (experimental platform hardware in loop) MATLAB/Simulink	Demonstrated by comparing the proposed method with the traditional VSG control.
[123]	Modified adaptive fuzzy control with APF for power quality improvement	Experimental setup—APF prototype with the proposed controller MATLAB/Simulink	Simulation testing—Proposed method versus two improved sliding-mode control strategies. Hardware testing—Efficiency testing considering various conditions.
[124]	DPC is applied to parallel active filtering using type-1 fuzzy to decrease THD	Laboratory experimental setup MATLAB/Simulink	Simulation—Demonstrated by comparing the proposed method with the conventional DPC strategy. Hardware—Before and after filtering.
[125]	Harmonics mitigation using three-level inverter-based APF controlled by fuzzy-based dwell-time allocation algorithm (type-1 fuzzy)	Laboratory implementation using digital signal processor and experimental setup MATLAB/Simulink	Software—Results obtained when considering APF + fuzzy control, APF without fuzzy, and without APF.Hardware—Before and after connecting APF+ fuzzy control.
[126]	Selective harmonic compensation using microcontroller fuzzy-based control	Laboratory hardware testing	Proposed method vs. hysteresis, “predictive 1” and “predictive 2” control.
[127]	Harmonics current compensation using APF controlled by three type-1 fuzzy controllers	MATLAB/Simulink and SimPowerSystems toolbox	The proposed method’s results compared the obtained results simulating the system without the APF and fuzzy control.
[128]	Fuzzy-PI current control of DSTATCOM for power quality improvement	MATLAB/Simulink	Results obtained with the proposed method vs. results obtained using a simple PI controller.
[129]	Stability improvement in an island microgrid using a battery storage system controlled by a fuzzy controller	DIgSILENT Power Factory software	Proposed method vs. robust control.
[86]	Harmonic current compensation using hybrid power filter, P-Q theory, and fuzzy control	MATLAB/Simulink Hardware implementation using FPGA	Results obtained with the proposed method vs. results obtained using a simple PI controller.

enumerates and briefly describes the 10 chosen papers from the final set that clearly explain the implementation procedure and the obtained results. Another four papers had detailed information about the implementation process, but they are not included in the table below as they are already described in Table 8 [118], Table 7 [119], Table 6 [76], and Table 4 [120].

3.2.7. Assessing Efficiency and Improvements

When demonstrating the efficiency of the proposed methods described in their papers, researchers chose one or more of the following four strategies enumerated below to compare the results obtained using their control methods with the results obtained when (1) no control system was employed or no PQ improvement device was utilized [87], (2) a conventional PI, PID, or control method was used as the control system [111], (3) a simple (less efficient) version of the proposed method was employed [130], or (4) other methods proposed in similar studies from the literature were used [131].

The chart in Figure 9 illustrates the yearly distribution of the efficiency demonstrations of the four strategies enumerated above. As some papers contained more than one testing strategy [131], the number of methodologies is greater than the number of papers.

Fuzzy logic’s strength lies in its capability to work with the nonlinearities of complex systems and the fact that no mathematical description of the control process is necessary, so only the knowledge and experience of experts are needed to provide the data to build an operative FCS [131,132]. However, a classic (type-1) FLC proved deficient when working in the presence of uncertainty factors [130]. This is the reason why many researchers studied and proposed the use of enhanced hybrid fuzzy-based controllers to obtain superior results, for example, the combination of an FLC and a PID controller, where the FLC is used to tune the three constants of the PID controller [68,75,133], or the use of evolutionary algorithms, like particle swarm optimization [74,134], to improve the parameters of the control system.

Figure 10 presents the results (number of papers per year) of the analysis of the final set in relation to the efficiency-improving methodologies used by the researchers in their studies.

Another aspect that we considered in the analysis was the FLC’s efficiency testing and improvement concerning the fuzzy controller’s features (Section 3.2.2 and 3.2.3), as in [132], where the authors present the difference between the Mamdani and T-S strategies when controlling an SVC for power factor reduction. From the 135 papers of the final set, as underlined in Figure 10, only 12 papers contained this type of FLC efficiency-improving methodology.

Table 10. Relevant papers assessing and improving the efficiency of the proposed FLCs.

Reference	Description	Efficacy	Efficiency
[63]	Frequency control using optimized type-2 fuzzy	Simulations	Comparison with type-1 fuzzy and PI controller
[130]	LFC in power systems with wind-generation units using robust fuzzy controller	Simulations	Development and comparison of a simple type-1 fuzzy controller with the proposed hybrid fuzzy controller
[64]	Online frequency regulation using fractional-order type-2 fuzzy	Simulations	The proposed method vs. three methods proposed in the literature
[135]	APF controlled by hybrid (inverted error deviation) fuzzy controller for decrease in harmonics	Simulations and experiments	Development of both fuzzy controllers (type-1 fuzzy and hybrid) and comparison with a classic PI controller
[136]	Frequency control in a microgrid with multiple types of distributed generation using hybrid type-2 fuzzy PI	Simulations	The proposed method vs. conventional PI and classic type-1 fuzzy PI controller
[137]	Frequency deviations and flicker issues solved using a hybrid (genetic algorithm optimizer) FLC that controls the generators	Simulations	Results obtained with the proposed method compared with the results obtained without control
[138]	SVC is controlled using fuzzy-PI and grey theory to improve voltage fluctuation issues	Simulations	Comparison with the classic type-1 fuzzy and without the control system
[139]	D-STATCOM controlled by different topologies of FLC for decrease in THD	Simulations	Comparison between PI-like fuzzy controller, PI gain scheduled fuzzy controller, and hybrid fuzzy-PI controller
[140]	Fuzzy load frequency control with auto-tuned membership functions and fuzzy control rules	Simulations	Comparison with the classic PI controller and a hybrid fuzzy controller (FLC + particle swam optimization)
[55]	Direct adaptive fuzzy logic control for load frequency regulation in multi-area power system	Simulations	The proposed method’s results compared with a classic PID and a type-2 fuzzy controller

describes 10 studies that focused on the efficiency of the fuzzy control assessment and improvement. The information in the table gives a general description of the selected papers, information on the efficacy testing technique, and a detailed presentation of the efficiency testing and improvement method for each proposed FLC.

4. Discussion

The analysis of the literature in the field of PQ improvement using FCSs highlights the potential of fuzzy logic in solving complex problems of modern power systems, including frequency, voltage, reactive power, harmonic distortion, and voltage unbalance control issues. Next, we focus on relevant solutions to solve these issues and future directions.

Frequency control is an essential aspect of modern power systems and is closely related to connecting distributed generation units to the grid, operating isolated microgrids, and interconnecting different areas of power systems.

Isolated microgrids that contain fluctuating generation sources (such as PV, wind, and hydro) and energy storage systems (EESs) often experience frequency fluctuations. To address this issue, researchers have used fuzzy logic to control EESs and loads, as demonstrated in [68,129,141]. Simple multiple type-1 fuzzy controllers were applied in [129,141]. Indeed, the authors of [141] use four FLCs to control the battery, dump load, load, and whole system, while [129] proposes two FLCs for active power and reactive power control, respectively. Mukherjee in [68] improved the proposed fuzzy-PID controller by tuning the PID parameters using a harmony search algorithm. These studies prove the effectiveness of FCSs through simulations by employing MATLAB [129,141] and DIgSILENT PowerFactory [129]. Efficiency is demonstrated by comparisons with other literature solutions [141], including simple PID [68] and robust control [129]. The comparison with the other methods showed that the fuzzy approach does not give frequency ripples as the robust control [129], and the FOD performance analysis’s results are superior in the case of fuzzy PID (0.21 with 1% increase in load demand), contrary to PID (0.68 in the same conditions) [68].

Frequency stability using improved FCSs when interconnecting multi-area power systems is the aim of many researchers. Thus, [63,64,71,136] used simulations to demonstrate the efficacy of their hybrid type-2 fuzzy controllers and compared their proposed methods with type-1 fuzzy controllers. Aluko and his colleagues in [71] combine the UIO (Uknown Input Observer) with interval type-2 control to increase the stability and the unknown input aspect. They demonstrate in their work that the proposed type-2 FLC deals much better with undershoot frequency deviations than type-1 FLCs and PI controllers. The authors of [136] describe the combination of an interval type-2 FLC (two inputs and two outputs), a PI controller, and a dynamic selector switch to build a hybrid control system that decreases frequency ripples (maximum overshoot (MO) 0.1), contrary to type-1 fuzzy + PI (MO-0.51) and PI control (MO-0.56). The authors of [75,140] use optimization algorithms (TLBO—teaching learning-based optimization; and ABC—artificial bee colony) to adapt the parameters of the controllers. The researchers prove in [75] that their TLBO better tunes the proposed fuzzy-PID through a filter controller than a GA (ITAE = 2.74) and other methods from the literature (fuzzy PD-PI—IATE = 0.17), as the proposed method gave the lowest error (ITAE = 0.0976). The ABC algorithm tunes the two FLCs used in [140] better than PSO (particle swam optimization), as the ISE is 3.75 for ABC-FLCs and 4.25 for PSO-FLCs, and the PI is 30.13. Smart fuzzy control that solves uncertainty factors by dealing with the FLC’s outputs and values differently is proposed in [130]. Experimental validation using test setups is presented in [75,122], whereas [102] shows real-time efficacy justification. Indeed, a large-scale real-time laboratory simulation is used in [102] to compare the methods described in the paper. Thus, the BAAL (balancing authority ACE limit) for the proposed universe variable FLC was 94.84%, better than the adaptive FLC (BAAL = 92.36%) and the improved PI (BAAL = 89.06%) [102].

Maintaining voltage levels between the standard limits is a desiderate of electricity utility companies, as this aspect is required for power systems’ safe operation and the normal functioning of electricity consumers’ installations and apparatus. Thus, voltage control is necessary in cases of voltage sags or swells, voltage fluctuations, and variations. These scenarios appear in several situations, for example when connecting distributed generation units, as presented in [94,142,143]. A STATCOM regulated the voltage in a wind power system being controlled by a type-1 (two inputs–two outputs) fuzzy controller and gave superior results than a PI controller in a simulation study [143]. The authors of [142] demonstrated using simulations that a multi-mode fuzzy (three type-1 FLCs with three inputs and one output) controller that controlled an inverter in a power system with high-ratio PVs was better at suppressing voltage variations than a simple fuzzy controller and a simple multi-mode controller. The results showed that the multi-mode + fuzzy controller had a voltage offset equal to 12.23, and the simple fuzzy approach had a voltage offset of 34.46. Fuzzy-based reactive and active power control is proposed in [94] to regulate voltage considering an on-load tap changer and distributed generation units. Simulation and hardware in-loop tests demonstrated the superiority of the proposed method in comparison with a simple version of the control system. The authors of [119] present two FCSs (adaptive tap control and adaptive reference control) that command power transformers dedicated to voltage control, considering simulations with real data and real-time experimental tests. The adaptive reference control was the superior method. Voltage sag and flicker reduction using a hysteresis fuzzy control that drove the EES showed superior voltage stability in simulations than in a scenario without the control system [103]. The authors of [72] developed a self-tuned fuzzy-PI controller for a DVR and demonstrated through simulations that the proposed method better mitigated severe voltage sag.

Decreasing harmonic distortions of voltages and currents is solved using APFs, which in many cases are upgraded by FSCs to better serve their purpose. This situation is found in [88], where the authors designed a thyristor-controlled LC-coupling hybrid APF and used a hysteresis FLC to adjust the parameters of the filter for THD reduction in medium-voltage-level systems. The results showed a switch loss decrease of 38% compared to using hysteresis control with a fixed band. In [84], the authors proposed an auto-tuning scheme based on an FLC to automatically calibrate the APF’s control coefficient and maintain the voltage’s THD within the standard limits in wind farms. Furthermore, an adaptive fuzzy method with a supervisory compensator for a three-phase APF is described in detail in [95]; more specifically, the authors use a Sugeno FLC, an adaptive law and supervisory controller, to reduce the THD from 24.71% to 1.72%. The efficacy assessment was performed using both simulations and experiments from [73,76,96,123]. The researchers in [73] used a hybrid FCS that contained a Takagi–Sugeno type-1 fuzzy controller to automatically switch between a PI and fractional-order PI controller. This approach gave greater results than employing only a PI controller, as the THD reduced from 55.8% to 2.4% and 5%. An adaptive hysteresis FLC has proven to give superior outcomes than a PI controller under various scenarios [75]. The implementation of a modified FLC on a microcontroller to control a shunt APF is described in [96]. The authors of [76] upgraded a repetitive controller with fuzzy logic, obtaining an adaptive version that showed better performance than the fixed repetitive version. This was clear, as the harmonic distortion factor decreased from 35.1% to 3.1%. The authors of [123] describe a modified adaptive FLC—more specifically, a Lyapunov-based fuzzy control. The authors compare the proposed controller with two methods from the literature through simulations. Additionally, they built a prototype that was tested in the laboratory. All tests showed that the proposed method was better, reducing the THD from 26.04% to 4.14%.

Reactive power control plays a crucial role in maintaining voltage control and power system stability. Hence, researchers often associate reactive power regulation with voltage variations and develop control systems that solve both issues simultaneously. This is the situation in [83], where the authors used a type-1 FLC to control the reactive power produced by distributed generation units. Similarly, [144] describes the use of a type-1 fuzzy to adjust the SVC’s capacitance for better control of reactive power and voltage control. In [145], a fuzzy self-tuning PI control together with two other regulation methods performed a three-layer coordinated reactive power compensation, and the results showed the better stability of reactive power variations than a PI-based control strategy. The reactive power control in power systems with wind generators is reflected in [93,133]. A fuzzy-PID where two FLCs give the k parameters of the PID controller plus a modulated hysteresis method is proposed in [133], and the authors show that the proposed method gave a 50.23% reduction in reactive power ripples in comparison to the conventional approach. A type -1 fuzzy with three inputs and one output is described in [93] to reduce the ripples of reactive power. All these studies used simulations to demonstrate the efficacy of the proposed methods and compared the FLCs with conventional approaches, i.e., the sensitivity method [83], no control [144], PI control [73], and conventional control [146]. Indeed, in [146], the fuzzy-based approach reduces the reactive power ripples by 40.5% in comparison with the conventional method (direct power control). For electricity consumers, reactive power control is significant as they must maintain the power factor at the connection point within a certain limit. Thus, [132] presents an FLC that controls an SVC, and shows using MATLAB simulations the difference between the Mamdani and Takagi–Sugeno inferences strategies. In [147], the author describes an active power factor correction scheme that contains a boost converter controlled by an FLC with parameters optimized by a pattern search algorithm. According to simulations, the proposed method is effective and more efficient (PF = 0.99987, THD = 1.65%) than a PI controller (PF = 0.99945, THD = 2.35%).

Unbalanced three-phase voltages and currents is an issue that appears in three-phase power systems. In the literature, this problem is treated together with other PQ issues that often occur in single-phase power systems as well. For example, the authors of [99] demonstrated, through simulations, the usage of a three-phase shunt APF controlled by a self-tuning control system containing an FLC in decreasing the THD and balancing the three-phase systems of currents and voltages. The authors of [148] present an adaptive hysteresis and FLC that controls a three-phase APF. This PQ improvement system proved to give superior results in simulations and experimental laboratory tests than a PI controller.

Considering a holistic approach to PQ improvement, that is, a reduction in all PQ disturbances, the selection of a proper control apparatus is essential. The literature demonstrates that using UPQCs controlled by hybrid FCSs is an excellent choice. The authors of [87] present the use of three type-1 fuzzy controllers to obtain switching signals for PWM logic control that command the UPQC. Simulations showed the efficacy of the proposed method in reducing THD (current from 26.58% to 5.76, and voltage from 46.93% to 3.67%), voltage sags/swells, and unbalance. A multi-feeder UPQC controlled by two FCSs is described in [67], in which a type-1 fuzzy and a fuzzy-PI controller are designed and compared through simulations. The hybrid FCS performed better in decreasing THD, unbalance, and voltage variations. The authors of [113] propose a hybrid FCS that uses two FLCs (Mamdani and Takagi–Sugeno) and the feedback integral phase-locked-loop (PLL) strategy to control a UPQC. They showed the results of the experimental assessment which proved that their method is superior to a Takagi–Sugeno-modified PLL technique in improving harmonic distortions, voltage sags, and unbalance. Indeed, with the proposed method, the THD was 2.13%, and the Takagi–Sugeno method gave a THD of 3.56%.

The tendency in recent years has been to study and propose fuzzy control inverters that connect distributed generation units but also reduce PQ issues. For example, [100] describes a fuzzy space vector pulse-width modulation technique to control the inverter and improve PQ in a microgrid, without the need to depend on the utility grid. The proposed method efficacy was tested through simulations and plotting voltage, frequency, harmonics, and reactive power characteristics. The authors concluded that their method is superior to the conventional ST-PWM control, as the THD was 1.18% in comparison to 1.53%, and there were less ripples in reactive power. An FLC-based Improved Second-order Generalized Integrator (I-SOGI) scheme that controlled the assembly of a Z-source inverter that compensated a DVR solved PQ issues like balanced and unbalanced voltage sags, swells, and harmonics [131]. Additionally, the authors used the FLC (2 inputs–3 outputs) to tune the PID controller’s k parameters. The simulation showed that the proposed method gave better results than other approaches from the literature. The proposed method’s THD was 2%, and the results for the other methods were as follows: ANN—7.5%; RFA—5.5%; and ASO—4%.

Based on the literature review, it can be noted that type-2 and hybrid fuzzy controllers are more effective in improving PQ. Additionally, the use of membership functions that are tailored to a range of quantities, such as the type, width, and nucleus of membership functions, selected after careful examination, has also demonstrated better outcomes. Similarly, the defuzzification method and inference strategy have a significant impact on the results. For future directions in PQ improvement using FCSs, researchers should consider the use of UPQCs and other similar devices that address a large range of PQ issues.

5. Conclusions

This paper presents a comprehensive review of fuzzy-based control system applications for PQ improvement, highlighting the importance of efficacy demonstration, efficiency assessment, and improvement for FCSs. A proposed literature classification outlines seven criteria for evaluating papers that examine the characteristics of FCSs and PQ issues. The classification also presents statistics and tendencies in this area. Additionally, the most pertinent papers are selected and described in detail to provide a comprehensive overview of the subject matter.

Control systems that use hybrid fuzzy control and employ adaptive functions, optimization algorithms, and self-tuning techniques have been found to provide the best results for addressing specific issues. Such approaches have been proven to be superior to conventional and simpler fuzzy-based control methods.

To ensure the effectiveness of FCSs, it is crucial to conduct efficacy testing using a combined approach of simulations and experimental tests. This approach provides a more comprehensive understanding of the system’s performance, which is essential for ensuring the safety and reliability of FCSs. To enhance the efficiency of FLCs, it is important to explore all features and inference strategies and make sure they are the right choice for the application. To demonstrate the efficiency of the proposed methodologies, it is imperative to explore the various simulation methods mentioned in the literature and conduct experimental tests that cover a range of scenarios. By doing so, we can gain valuable insights that can help us optimize our processes and achieve better results.

This study has relevant implications not only for researchers, but also for participants. From an academic point of view, the main contributions of this paper are linked to the characteristics of fuzzy-based control systems, in particular, aspects regarding the structure, features, and inference strategies of FLCs, as well as combinations of fuzzy logic and other techniques that increase the performance of control systems. Additionally, the paper presents an analysis of the literature regarding efficiency testing, thus helping researchers in understanding new directions and tendencies. From a practitioner’s viewpoint, this work presents studies from a more practical point of view, focusing on PQ issues and control devices, and including studies that describe experimental implementations and prototypes.

In conclusion, maintaining PQ within standard limits is crucial for power systems. However, due to the unpredictable and nonlinear nature of power systems, it is challenging to achieve this. Fuzzy logic, along with optimization techniques and adaptive strategies, is an appropriate approach for improving PQ. This work aims to simplify the understanding and use of fuzzy logic control in power quality issues and power quality control devices.

Regarding future directions for research, we consider further assessing published research that focused on power quality control using ANFIS and similar methodologies (data-driven fuzzy-based control) that were omitted in this research.