Accurate load forecasting plays a very important role in the operation of power selling companies in the power market. However

the load of enterprise users is unstable under the influence of various factors. Therefore

a short-term load forecasting method based on the seasonal Holt-Winters model of discrete wavelet decomposition and particle swarm optimization is proposed. In view of the unsteady periodicity of the original load sequence

the discrete wavelet transform is used to decompose the original load sequence

and the seasonal Holt-Winters model is adopted for prediction. Meanwhile

wavelet denoising and particle swarm optimization algorithm are used to further improve the accuracy of the prediction model. Wavelet denoising not only eliminates the potential noise in the original data

but also smoothes the data

and particle swarm optimization allows Holt-Winters model to find the optimal parameters during training. The model is used for short-term prediction of load data with different variation characteristics

and the experimental results show that the model has good prediction accuracy and can be applied to short-term load prediction of users with different power consumption types.