Regional Short-term Load Forecasting Based on PLE-residual Correction

With the rapid development of photovoltaic, the scale and number of photovoltaic power stations are increasing, and the photovoltaic output has an impact on the load forecasting. Ignoring the influence of photovoltaic output in regional short-term load forecasting will reduce the accuracy of prediction. Aiming at this problem, a regional short-term load forecasting method based on PLE (Progressive Layered Extraction)-residual correction is proposed. Firstly, the derivative features of load forecasting are constructed, and the historical load data are extracted with various features, including time features, seasonal features and holiday features, so as to fully mine the potential information in load data. Then, a regional power grid bus load forecasting model based on PLE is established. The hierarchical feature extraction method is used to gradually transform the input data into a higher-level abstract representation to capture the nonlinear relationship in the data. In order to further improve the accuracy of the prediction model, the LSTM model is used to correct the residual. The results of the example analysis show that the regional short-term load forecasting method based on PLE-residual correction has higher prediction accuracy and stronger stability than the traditional method.