Inverter open circuit fault diagnosis strategy based on KNN algorithm

Inverters play a crucial role in renewable energy generation system. Once a fault occurs, it will cause damage to equipment and even threaten personal safety. Therefore, accurate inverter fault diagnosis algorithm is crucial. This paper proposes a new inverter open circuit fault diagnosis strategy based on k-Nearest Neighbor (KNN) algorithm to address the problems of complexity, high computational resource consumption, and inapplicability to embedded control system in existing fault diagnosis algorithms. Firstly, the inverter sampling data is preprocessed, and the moving average method is used to process the data, reducing the amount of data and improving the calculation speed. Pearson thermal maps are used to reduce the dimension of the data and extract fault features. Then, the sampled data obtained from preprocessing is used to train the fault diagnosis strategy based on KNN algorithm. Finally, the trained model was validated using a test set, and the results showed that the proposed algorithm occupies less computational resources and has high fault diagnosis accuracy, making it suitable for embedded control system.

I. Introduction

With the promotion of the "dual carbon" goal, renewable energy sources such as wind power and photovoltaic have rapidly developed. Inverters convert direct current generated by renewable energy sources such as photovoltaic into alternating current that meets grid connection requirements, so that playing a crucial role in renewable energy generation systems. Once the inverter malfunctions, it may cause damage to other equipment in the renewable energy generation system and even threaten personal safety. Statistical data shows that inverter faults account for 60% of common faults in the operation of photovoltaic power plants, and it is very important to quickly and accurately diagnose inverter faults [1]-[2].