Short-Term Power Load Forecasting Study Based on IWOA Optimized CNN-BiLSTM

Abstract

In order to improve the prediction accuracy of short-term power load, a CNN-BiLSTM-Attention short-term power prediction model based on the hybrid strategy Improved Whale Optimization Algorithm (IWOA) is proposed. The model first combines convolutional neural network (CNN) and bidirectional long and short-term memory network (BiLSTM) to fully extract the spatio-temporal features of the load data itself. Then, the time-attention (TPA) mechanism is introduced to automatically assign corresponding weights to the BiLSTM hidden layer states to distinguish the importance of different temporal load sequences. Meanwhile, in order to solve the problems of the temporal model's parameter difficult to choose and the whale algorithm's poor global search ability, which is prone to fall into the local optimum, the Tent chaotic mapping is used to optimize the whale algorithm. “ to optimize the whale algorithm (IWOA), and better search for optimization of model parameters. The proposed method achieves 97.82% prediction accuracy and is compared with LSTM, BiLSTM and CNN-BiLSTM prediction models. The experimental results show that the proposed method has higher prediction accuracy and can provide a reliable basis for power system planning and stable operation.