In order to improve the prediction accuracy,stability and generalization ability of the residential engineering assessment model,a BP neural network(GWO-BPNN)prediction model based on the Gray Wolf Optimization algorithm is proposed.The GWO algorithm is used to optimize the BPNN model network implicit layer weights w and node bias b,and the GWO-BPNN optimal prediction model is constructed.The prediction performance of the GWO-BPNN model is verified by random test samples,and the results show that:the R2 of the goodness of fit between the predicted value of the single-side project and the actual value of the samples of the GWO-BPNN model reaches 0.967,and the range of the absolute error value of the single-side cost is[-42.765,18.281],and the range of the relative error value is[-2.42％,0.92％];the MAE reaches 14.536 yuan/m2 and MBE reaches 6.601 yuan/m2,the GWO-BPNN model shows good prediction accuracy and robustness;meanwhile,the comparative analysis based on the prediction performance of the unilateral cost of the GWO-BPNN and BPNN models shows that the prediction accuracy and prediction stability of the GWO-BPNN model performs better.The GWO-BPNN prediction model proposed in this study can predict the cost of residential projects stably and efficiently,and has the feasibility of engineering application.
关键词
住宅工程/造价预测/灰狼优化算法(GWO)/BP神经网络(BPNN)
Key words
Residential project/Cost prediction/Grey Wolf Optimization(GWO)/Back Propagation Neural Network(BPNN)
维普
万方数据