Short-term Electric Load Combination Forecasting Model Based on LSTM-LSSVM

Load forecasting is crucial for economic dispatch of power systems, with accuracy impacting grid operation. Due to rising energy demand and changing load characteristics, forecasting complexity has increased. Traditional methods struggle with nonlinear data, complicating load forecasting. This study proposes a novel approach using a hybrid long and short-term memory network with a least-squares support vector machine model. A hybrid seagull algorithm and an improved whale algorithm are employed to optimize the prediction model. Results show superior accuracy compared to individual models, promising advancement in power load forecasting.