Short-term wind power load forecasting based on ISSA-CNN-BiLSTM

Accurate wind power prediction is of great significance in the balance of electricity supply and demand. Aiming at the problems of poor convergence and easy to fall into local extremes of the power prediction model using the traditional sparrow algorithm (SSA) to optimise the parameters of the bidirectional long- and short-term memory neural network (CNN-BiLSTM), this paper adopts the refractive inverse learning strategy, the positive cosine strategy and the Cauchy variation strategy to improve the sparrow search algorithm, and a better convergence is obtained. An improved sparrow search algorithm (ISSA) with stronger convergence is obtained. The ISSA algorithm is used to optimise the bi-directional long and short-term memory neural network, and the ISSA-CNN-BiLSTM neural network prediction model is established. This paper takes the data of a wind farm in Inner Mongolia as the research object, and uses SSA-CNN-BiLSTM algorithm,CNN-BiLSTM algorithm and ISSA-CNN-BiLSTM algorithm to simulate and test the measured historical data of this wind farm, respectively. The experimental simulation comparison shows that the combined model proposed in this paper has a 7.6% improvement in prediction accuracy over its base model.