基于指数平滑和PSO-BP混合模型的建筑工程造价指数预测

Prediction of Construction Cost Index Based on Exponential Smoothing and PSO-BP Hybrid Model

建筑工程造价指数是进行工程造价管理的重要依据。为了提高建筑工程造价指数预测的准确性,首先,利用HP滤波将造价指数分解为趋势序列和波动序列,然后采用指数平滑模型对趋势造价指数序列进行预测;其次,利用粒子群算法(PSO)优化的BP神经网络对建筑工程造价指数的波动序列进行预测,PSO-BP神经网络模型的输入指标为引起造价指数变化的4种材料价格;最后,叠加二者预测值即为建筑工程造价指数的预测值。实验结果表明:该混合模型对6个月的造价指数预测的平均相对误差为0.55%,取得了很好的效果,为准确预测建筑工程造价指数提供了一定参考。

Construction cost index is an important basis for construction cost management.In order to improve the accuracy of construction cost index prediction,firstly,the HP filtering was used to decompose the cost index into a trend sequence and a fluctuation sequence,and then an exponential smoothing model was used to predict the trend sequence of the cost index.Secondly,the BP neural network optimized by particle swarm optimization(PSO)was used for prediction.The input indicators of the PSO-BP neural network model were the prices of four materials that cause the change of the cost index.Finally,the final superposition of the two predicted values was the predicted value of the construction cost index.The experimental results show that the average relative error of the mixed model for the six-month cost index prediction is 0.55%,which has achieved good results,and provides a certain reference for the accurate prediction of the construction engineering cost index.