Research on MPPT control of photovoltaic power generation under complex lighting conditions based on Long Short-Term Memory deep neural network

In view of the disadvantages of the existing MPPT algorithm, which is difficult to adapt to the multi-peak characteristics of photovoltaic power generation module output under complex lighting environment and easy to fall into local optimization, In this paper, the MPPT control model of photovoltaic power generation module under complex lighting conditions is established by using Long Short-Term Memory deep neural network (LSTM DNN), and the actual photovoltaic power generation module is replaced by the established neural network model, and the MPPT controller is designed. The verification shows that it is feasible to use LSTM DNN neural network to establish a nonlinear model of photovoltaic power generation module under complex lighting conditions, and this model can be used to predict the MPPT response of photovoltaic power generation module under complex lighting conditions online. The designed LSTM DNN neural network MPPT controller is effective, and compared with the traditional perturbation observation control method, The MPPT tracking accuracy, tracking speed and output power improvement have shown significant advantages, which effectively solves the multi-peak problem of photovoltaic arrays under partial shade conditions, and improves the tracking accuracy and speed of MPPT control.