基于变学习率BP模型的城市燃气短期负荷预测

建立了基于变学习率BP模型的城市燃气短期负荷神经网络预测模型。输入元为城市燃气短期负荷的五个影响因素：日期类型、天气类型、日最高气温、日最低气温、日平均气温。网络结构为5-6-1。输出元为燃气短期负荷。用VC＋＋编程，变学习率为0．3和0．7，经过19086次迭代，模型收敛，全局误差为0．00049999。数据对比分析发现相对误差在5％之内，说明该模型是准确有效的。

Forecasting city gas short-period load based on BP model with varied learning rate

The forecasting model has been established based on the BP model with varied learning rate.The input cell includes the five factors influencing the city gas short-period load,i.e.date style,climate style,daily maximum temperature,minimum temperature and averaged temperature.The configuration of the BP network is 5-6-1.The output cell is gas short-period load.The programming is carried out with VC++ and the varied learning rate is 0.3 and 0.7,respectively.The model converges after 19 086 iterations,and the overall error is 0.00049999.The relative error is less than 5%,which indicates that the model is accurate and efficient.