摘要
根据电力系统短期负荷预测的需要,用C++开发了单隐含层BP人工神经网络程序。程序用Levenberg-Marquardt训练算法实现神经网络训练,大大提高了训练速度。采用24个单输出人工神经网络模型分别预测每天的整点负荷。该预测模型可动态生成,提高了预测模型的自适应性。实际算例结果表明,采用该算法及其程序进行短期负荷预测,可得到令人满意的训练速度及预测精度。
According to the demand of short- term load forecast of power system, a one- hidden- layer artificial neural network program is developed with C ＋ ＋. The training process is accomplished by Levenberg- Mantuardt algorithm, and the training speed is much faster. The hourly load of the forecasting day is forecasted by 24 artificial neural network models reepectively. The adaptability of the model is enhanced by the dynamic training before forecasting. The actual calculation results indicate that the training speed and the forecasting accuracy are satisfactory.
出处
《四川电力技术》
2006年第3期29-31,共3页
Sichuan Electric Power Technology
