Research on short-term power load forecasting method based on multi-factor feature analysis and LSTM

Abstract

In order to improve the accuracy of short-term power load forecasting and fully consider the influence of weather factors on power load, a short-term power load forecasting model based on multi-factor analysis and Long-Short Term Memory (LSTM) neural network is proposed. Firstly, the correlation between different weather factors and load is analysed using the Spearman coefficient method to extract the weather features that have a greater impact on power load. Then the original time series data are reconstructed using the sliding window method. Finally, the forecasting model is established by using LSTM. The proposed model is validated by using the power load data from the 2016 Electrician’s Cup modelling competition, and compared with other models. The results show that the average absolute percentage error of the forecasting model proposed in this paper reaches 7.41% and the average absolute value error is 380.67 MW, which is better than the other models mentioned in the paper.

Insulation life prediction of stacking multi-model fusion motors based on Conv1d-LSTM

“…Firstly, the forgetting gate is utilized to read the last output value ht-1 and the current input value xt, which performs a nonlinear mapping and then outputs a vector ft. The output values are compressed by a Sigmoid function between 0 and 1, with closer to 0 meaning that the more it needs to be discarded [11] . Input gates are used to determine the new information that is being stored in the cellular state; the output gates determine the value of the next hidden state.…”