Research challenges in real-time classification of power quality disturbances applicable to microgrids: A systematic review

Microgrids with distributed renewable energy sources are especially sensitive to power quality disturbances. To mitigate the effects of distortions, they must first be detected and classified. Automatic classification of power quality disturbances has been extensively studied. However, real-time classification is yet to be investigated. Real-time classification is especially important in microgrids as they include a large number of subsystems. This paper presents a critical systematic review focused specifically on real-time applications. For this review, 809 papers were identified and the most cited papers of each year were analyzed in detail, i.e., a total of 134 papers were analyzed. Studies on all types of power systems were considered as their distortions can be observed in microgrids. These studies were categorized into three groups depending on their real-time abilities, and a comprehensive analysis to examine key items was performed. Subsequently, the research challenges in real-time operation were identified, i.e., extracting a reduced number of discriminant features with minimal processing, achieving a balance between classification accuracy and computational complexity, using datasets with more types of disturbances, including more types of combined disturbances, using real data to validate the classifiers, distributing public comprehensive datasets, embedding classifiers in dedicated hardware, improving the performance of real-time classification systems, conducting objective real-time analyses in physical devices, and setting a common evaluation framework to objectively compare real-time operations. These research challenges must be tackled to obtain a viable, accurate, fast, low-cost, and embeddable power quality classification system that facilitates the inclusion of distributed renewable energy sources in microgrids.