基于IHPO-SVM的能源路由器交流端口在线故障诊断OA

Energy routers have the role of optimizing energy flow and improving grid stability and reliability in microgrids containing new energy accesses such as photovoltaic,wind power,etc.However,in practical applications,energy router AC port open-circuit faults are more difficult to identify,and real-time fault diagnosis is challenging.Aiming at the above problems,a real-time fault diagnosis method for AC port inverter based on the combination of multi-head attention mechanism,convolutional neural network,long and short-term memory network,improved hunter prey optimization algorithm and support vector machine is proposed.The method first collects the inverter three-phase current data and divides them by cycles,and extracts features using the proposed multi-head Self-Attention-CNN-LSTM method as the basis for subsequent fault diagnosis.Then,based on the above extracted features,the IHPO-optimized SVM is used to train the offline fault diagnosis model on the measured data;finally,the offline diagnosis model is imported into the online diagnosis program to realize the real-time fault diagnosis of the inverter and determine the fault location.The proposed method is compared with other diagnostic methods using measured data to verify the effectiveness and accuracy of the proposed method.The online test results show that the proposed method has faster diagnosis speed and accuracy.