A deep learning model combining Long Short-Term Memory networks (LSTM) and Parallel Residual Networks (ResNet) is proposed to enhance the accuracy of short-term load forecasting in power systems. This model leverages the time series analysis capabilities of LSTM with the deep feature extraction strengths of ResNet to effectively process complex electrical load data. The forecasting accuracy of the model is comprehensively assessed using three key performance indicators. Additionally, a detailed feature analysis further validates the appropriateness of the input features.