Deep Learning Based Phase Current Prediction Model in Three-Phase IGBT Two-Level Inverter for Electrical Drives

Power electronics circuits are useful for changing the waveform of a power supply from DC to AC, and these electronic circuits, of which the power inverter is one of them, play a crucial role when it comes to the management of renewable energy sources and the conversion of power supply waveforms. Research for artificial intelligence applications to predict the non-linear behavior of power inverters has garnered prominence recently. This study proposes a novel bidirectional long-short-term memory (BiLSTM) based deep learning model to predict the current value of the inverter based on the duty cycle, voltage values, motor rotation speed, and mean phase voltage of the inverter for the first time. The proposed model was compared to mainly used recurrent neural network (RNN) variations such as long-short term memory (LSTM) and gated recurrent unit (GRU). According to the experimental findings, it was observed that the proposed model yielded lower mean absolute error (MAE), mean absolute percentage error (MAPE), root mean square error (RMSE), and higher r-squared (R2) results for predicting the current values of three phases. Therefore, it can be inferred that the proposed model is effective for current prediction.