基于人工神经网络原理，设计了一个三层的ＢＰ网络模型。充分利用了神经网络高度非线性建模能力，实现电力系统的短期负荷预测。文中对样本数据进行了预处理，以及在算法中引入附加冲量项，以提高训练速度。预测仿真结果证明使用人工神经网络方法进行短期负荷预测是可行的。 Based on the principle of artificial neural network, a three-layer BP network model is designed. Make full use of neural network highly nonlinear modeling ability to achieve short-term power load forecasting. The paper preprocesses the sample data and introduces additional impulse in the algorithm to improve the training speed. Predictive simulation results show that using artificial neural network method for short-term load forecasting is feasible.