Inverse estimation of power system load parameters based on random sampling strategy Title

In order to realize the accurate inversion of power system integrated load model parameters using actual operation data, a power system integrated load parameter inversion method based on random sampling strategy is proposed, which utilizes the random forest method to screen the time series of the power simulation model output to obtain the amount of time series changes that are sensitive to the integrated load parameter; using the simulation data, the time series changes of the integrated load parameter with the RF-LSTM are established. The relationship between the integrated load parameter changes and the simulation output is established by RF-LSTM using the simulation data. Using this method to estimate the inversion of the integrated load parameters of the power system in the actual network of a power supply company, the results show that the variation of the inverted fault curves of the predicted parameters of the hybrid model from the real curves is only 0.27%, which is more desirable than the traditional single Random Forest, LSTM and other machine learning models that are not filtered by features.