Model Predictive Control for Energy Optimization in Generators/Motors as Well as Converters and Inverters for Futuristic Integrated Power Networks

Predicting, controlling and distributing energy in an efficient way represents and will represent one of the most critical points for the future of our industrial societies. In general, prosperity and processes to achieve peace are also connected to an equal distribution of common resources, such as energy, among humanity. In this context, the integration of energy control and the optimization of its prediction seems necessary to be realized. The control and optimization of power engineering systems are crucial and pose essential demands. The highly non-linear system behavior and the need to consider inherent dead times and the limitations of the manipulated output variables requires new methods and algorithms. Conventional control concepts cannot meet the high control quality requirements necessary for the existing strict regulations. For example, with regard to emissions or the integration of components such as generators, motors, converters and inverters for futuristic power networks, many specifications and issues arise. Model predictive controls (MPC) offer the possibility to take these requirements into account in a systematic way. They can ensure a high level of control quality over the entire operating range while also taking into account suitable prediction and control horizons needed to optimize energetic cost functions. Roughly speaking, MPC are based on a repetition of the real-time optimization of a mathematical system model, which guarantees the predictions based on a time prediction horizon. Based on this system model, the MPC predict the future system’s behavior, considering it in an optimization problem in which a cost function is defined. This cost function typically consists of a squared predicted error between the predicted variables (goals) and the measured or observed ones.

Moreover, a part consisting of manipulated squared inputs is added to guarantee its stability. In this way, an optimal trajectory of the manipulated input variable can be calculated at each step of a receding horizon, see [1]. The theory of MPC was developed at the end of the 1970s in [2] and at the beginning of the 1980s in [3,4]. From the beginning of the 1980s until the beginning of the 1990s, the founders of MPC in terms of their applications [5,6], stressed the fact that to achieve the normal control requirements, 90% of the existing control strategies can be used to reach the requested goals. Only the remaining 10% of the advanced process controls must be controlled by advanced applied methods. In fact, advanced processes require control properties such as robustness and optimization, which very often are difficult to be obtained together because optimization requests a model which is affected by explicit and implicit uncertainties. Nowadays, the complexity of the process that is to be controlled, in particular in the context of interconnected power energy networks, calls for new advancements and new methods in the area of MPC. In this sense, optimization algorithms, and in particular MPC-based algorithms, still represent future quantitative methods to be used to control the dispatching of energy in an optimal way in micro- and macro-grids. New ideas for efficient calculation methods are requested to overcome the non-convexity of the optimization problems, which are the consequence of complex integrated systems with different energetic cost functions.

MPC represents a viable futuristic strategy of control to overcome the limitations of conventional control concepts. This is very suitable in areas of application such as emerging energy optimization, in which prediction and control horizons play a crucial role. The following strictly selected papers indicate possible solutions to control and integrate power components in modern power networks, optimally and efficiently, considering the energetic requirements needed by users.

The article [7] presented a technique based on a finite control-set model predictive control (FCS-MPC). In the MPC that was presented, the objectives of controlling the current and the switching frequency were used to pick the best inverter switching state by minimizing a cost function. This allows for the most efficient use of the inverter. The weighting system combined two distinct control goals into a single structure. The weighting factor was determined by striking a balance between the current harmonic distortion and the switching frequency. Both models were compared to each other. To verify the viability of the proposal, as well as the MATLAB/Simulink simulation, a prototype of the hardware, in a scaled-down form, was created. This was performed even though there was a 0.25% increase in current THD. The incorporation of a switching frequency component into the cost function allowed for the achievement of a gratifying empirical result for an on-grid photovoltaic (PV) inverter. This was accomplished by reducing the instantaneous overall power loss by 2%.

Direct model predictive control, abbreviated as DMPC, is a method that in [8] is suggested to be used for power converters that are connected to a utility. This method is not only reliable but also advantageous from a computational standpoint. The DMPC that was presented made use of virtual voltage vectors (VVs), in addition to actual ones, to achieve improved steady-state responsiveness. This was performed to achieve the desired goal. A deadbeat (DB) function was applied to ease the strain that the suggested control method placed on the computer and made it simpler to comprehend. This was done so that the reference VV could be determined more accurately. This DB function had an additional discrete-time integral term added to it to improve the DMPC technique’s robustness in the face of varying model parameter values. The goal of this modification was to make the DMPC technique more accurate. A quality function, as the last step but certainly not the least, selected the best possible virtual or actual VV for the subsequent sample power converter. Comparisons were made between the performance of the recommended method and that of the DMPC and voltage-oriented control (VOC).

The authors of the contribution [9] proposed an MPC-VSG method as a means of automatically controlling the output power of a converter in response to variations in the grid’s frequency and voltage. To improve the precision of the reconstruction and guarantee continuous operation for the three-phase converters that had errors in their current sensors, MPC-VSG made use of an improved voltage vector selection and a reconstructed current. The strategy that has been proposed was resistant to errors, adapted to shifts in the frequency and voltage of the power grid, and did not call for the use of a PWM or a PI controller. The experiments have demonstrated that the control approach that was proposed was effective.

The article [10] presented a predictive control model with a grid-linked inverter modulator that used a stationary reference frame and a grid-distorted voltage. This model was created by using an inverter that was linked to the grid. After obtaining the model of the system’s stationary reference frame at its fundamental frequency, the predictive model technique was applied to predict the actions of the system utilizing the model without taking into account any harmonics. This was done to gain an understanding of how the system would behave in the future. The process of minimizing the cost function included calculating the voltage vector of the inverter as part of the process. Therefore, the proposal demonstrated, through tests, that the moderate impact of a distorted grid voltage can be demonstrated even if the model does not make use of harmonics. Experiments and comparisons have been done to back up the claim that this work should be conducted. The use of parallel UPS systems increased both the capacity of the system’s power supply and the level of resilience that it possessed. When uninterruptible power supply (UPS) systems were connected in parallel to each other, they had the potential to generate a circulating current which made it more difficult to use the system. For paralleled uninterruptible power supply systems to function properly, controlled load power distribution is required.

In [11] the authors proposed a novel Finite Control Set Model Predictive Control (FCS-MPC) technique for paralleled uninterruptible power supply (UPS) systems that made use of an additional inverter leg for connection to the load neutral point. In addition, it was recommended to use a parallel-connected UPS system that can supply both single-phase and three-phase significant loads. The experiments showed that the proposed control strategies were not only effective but also robust when applied to a wide variety of loads. This is demonstrated the effectiveness of the proposed control strategies.

In [12] an MVFCS-MPC method with fuzzy logic for PMSMs is proposed. This method were used in electric drive systems. This method was intended for use with fuzzy logic. This method employed discrete space-vector modulation as its modality of choice (DSVM). Real voltage vectors from the converter were combined with new, virtual voltage vectors and this was done so that the converter’s performance could be improved when it was in a steady state. A deadbeat function, also known as DB and abbreviated as such, had been incorporated into the proposed MV-FCS-MPC technique to facilitate the reduction of the number of necessary calculations. When determining which real or virtual voltage vector would provide the most accurate real or virtual value at the next sampling instant, a cost function was applied to the vectors to compare them. An external fuzzy logic controller was responsible for regulating the speed at which the rotor spun. As a direct consequence of this, the dynamic speed response was enhanced, and the proportional-integral (PI) controller tuning was made easier. The conclusions from the simulation draw parallels between the method that was suggested to be used and the conventional FCS-MPC.

The previous strictly selected papers show a trend of research in the area of energy control and optimization, proposing MPC as a futuristic topic of investigation of new algorithms and methods. Generators, motors, converters and inverters are considered, considering crucial emerging problems related to their optimal integration in modern power networks. These modern power networks, with their different multifunctional components for the generation, distribution, accumulation and transformation of energy, are seen within a holistic optimization predictive structure to guarantee their maximal integration with the maximal optimal energy for the whole network.