基于ANN的电力工程造价预测

ANN Based Power Engineering Cost Forecast

为了对新的电力工程进行造价预测与估算,采用人工神经网络(ANN)的机器学习与预测能力,通过分析电力工程造价的影响因素,并将这些影响数据归一化预处理后作为ANN的输入,用历史电力工程造价数据对ANN进行训练,并采用训练后的ANN对新的电力工程进行造价预测。多个电力工程造价预测结果表明,ANN的造价预测结果与实际造价的误差小于5%,符合电力工程造价人员的经验许可。

In order to forecast the cost of new power engineering,an ANN based power engineering cost forecast method is proposed.All the facts that are influential to power engineering cost are analyzed,and the influential data after normalization processed are used as the input of ANN.The historical power engineering cost data are used to train the ANN,and the trained ANN is used to forecast other new power engineering＇s cost.It was shown from several cost forecast experiment that the error between the forecasted cost and the actual cost is less than 5%,which is acceptable from the engineers′view.