A short-term power load forecasting model based on the beluga whale optimization (BWO) algorithm and long short-term memory network (LSTM) is proposed to address the high demand for accuracy in power load forecasting in smart grids, while also considering the instability of traditional optimized LSTM models in short-term power load forecasting. The model first uses the hyperparameters of LSTM as the location of the beluga whale, historical load, day type, and weather factors as the dataset, and the prediction error of LSTM on the training set as the fitness of BWO. Then, BWO is used to optimize the parameters and hyperparameters of the LSTM model that is most suitable for the training set, and predict and analyze the power load data of a certain region based on the optimal LSTM model. The research results indicate that the average absolute error percentage and root mean square error of BWO-LSTM prediction results are smaller, the prediction accuracy is higher, and the prediction results are more stable. It can be used as a reliable tool for short-term power load forecasting and provide strong support for the safe and stable operation of the power system.