Inverter fault prediction and diagnosis method based on Spark and PSO-FCM algorithm

In order to improve the anti-interference ability and diagnosis effect, this paper proposes a PV inverter fault prediction method based on Spark and PSO-FCM clustering algorithm. The method firstly adopts the t-SNE dimensionality reduction algorithm to extract the main feature vectors of PV inverter clusters and normalize them as the input of the clustering algorithm, and then adopts the PSO optimization algorithm to select the optimal initial clustering centers, and then establishes a PV inverter fault prediction model based on the FCM clustering algorithm; and finally adopts the memory-based parallel computation framework Spark to parallelize the PSO-FCM clustering algorithm Finally, the PSO-FCM clustering algorithm is parallelized using Spark, a memory-based parallel computing framework, to improve the fault prediction efficiency. The results show that the proposed PV inverter fault prediction technique can accurately predict PV inverter faults in advance, which helps to ensure the healthy and smooth operation of the equipment.