*2.5. Other Faults*

This refers to any fault that cannot be categorized under any of the previously discussed faults. These types of faults include MPPT and inverter faults that mostly occur due to inverter components failure, such as IGBTs, capacitors, and converter switch failure [51–53]; bypass diode faults resulting from a massive reverse current flow during faults, which leads to short-circuits [54,55]; blocking diode faults, also as a result of a reverse current flow [39]; open-circuit faults caused by items falling on PV panels, physical failure of panel-panel cables or joints, and sloppy termination of cables, plugging, and unplugging connectors at junction boxes [56]; faulty connections damage of connecting cables or a wrong connection of panels [57]; battery bank failures due to abnormal charging conditions [39]; and blackouts caused by natural disasters, such as a storm and lightning [58]. Most of these mentioned faults are usually due to an after effect of the other faults [39].

#### **3. Artificial Intelligence and Machine Learning**

Machine learning (ML) is a subsection of AI, which, in turn, is one of the most recent branches of science and engineering that emerged shortly after the Second World War, in order to attempt to understand and build intelligent entities [59]. ML thrives on extracting meaning from a large amount of data. Therefore, ML refers to the tools and technology that can be used to answer questions using the currently available data. Philosophy, statistics, biology, computational complexity, information theory, artificial intelligence, cognitive science, and control theory are some of the fields that machine learning draws on concepts and results [60]. In recent years, numerous machine learning algorithms and applications have been successfully created and utilized for different applications from autonomous vehicles, where these algorithms learn to drive on public highways, detect fraudulent transactions using data-mining programs, and are applied to information-filtering systems. This has led to critical advancements in the foundations of this field, which are theory and algorithms.

According to [61–63], machine learning is a branch of artificial intelligence (AI) that allows a system to learn from experience and improve without the need to be explicitly programmed. Its goal is to create computer programs that can access data and learn on their own [61]. Basically, the concept of ML is the use of fed-in data and algorithm to generate artificial knowledge, which is guided by sets of pre-defined analytical rules for pattern recognition in collected data [60]. In simple clear terms, ML is the use of quantitative and qualitative data to answer questions with ease and precision. Using ML, computers and IT systems are able to perform tasks, such as discovering, extracting, and summarizing relevant data, making predictions based on analyzed data, independently adapting to certain developments, calculating probabilities for specific results, and optimizing processes based on recognized patterns [61].

AI and ML are found in our everyday activities from transportation in the form of selfdriving and self-parking cars, image recognition, google searches, fraud detection in the banking and finance sector, diagnosis and prescription in medical fields, recommendation systems, as well as social media applications [59]. In PV system applications, ML and AI are used in predicting solar radiation, sunshine duration, and clearness index; mean temperature; and insolation and diffusion fractions [64]. In addition, they are used for the sizing, configuring, modeling, simulation, and control of PV systems, detection and diagnosis of faults, and the forecasting of the output electricity from standalone and grid-connected PV systems [22,64–72].
