Load Forecasting Analysis of Overhead Transmission Lines Based on EMD-LSTM Algorithm

The power system needs to maintain the immediate balance between the generating power and the load, and the load has the characteristics of nonlinear, time-varying and uncertain. In order to solve this problem, a predictive model is developed by integrating empirical mode decomposition (EMD) and long short-term memory (LSTM) neural network. This model takes into account the impact of weather meteorological characteristics, date characteristics and historical load data to accurately forecast short-term power load. Firstly, feature extraction and signal denoising are carried out by empirical mode decomposition to eliminate the fluctuation of data. Secondly, the pre-processed data is trained by LSTM, and the output results are reconstructed. Finally, load forecasting is carried out, and the forecasting effect of EMD-LSTM combined forecasting model is compared with back propagation (BP) neural network forecasting model, LSTM neural network forecasting model and EMD-SVM forecasting model respectively. The results show that the combined prediction model of EMD-LSTM neural network has the best prediction effect and improves the prediction accuracy effectively.