Abstract: In view of the problem that voltagesag has a great impact on the frequency conversion speed regulation system in engineering,the BP neural network is used to predict the DC side voltage after voltagesag of low-voltage convertor, and the nonlinear mapping relationship of load power, DC side capacitor, depth of voltage dip, duration and the DC side voltage of the convertor is established. First, the simulation model was built based on MATLAB/Simulink, four parameters were adjusted for mass simulation, sufficient data samples were obtained for different types of voltagesag. Then, the BP neural network was established for prediction, the voltagesag tolerance of low-voltage convertor was evaluated by comparing the DC side voltage predicted and protecteed value. The results show that the BP neural network model has high prediction accuracy and can accurately predict the DC side voltage value, so as to judge the protection action of low-voltage convertor after voltage sag.

    Key words: voltagesag; BP neural network; low-voltage convertor; tolerance