混合BP网络与ARMA算法的房屋造价预测方法研究

Research on the House Cost Forecasting Method Based on Hybrid BP Network and ARMA Algorithm

研究为解决房屋造价预测难以适应外界变化并形成精准预测结果的问题,将BP神经网络算法进行改进,将输入层、隐含层、输出层之间的权值区间设计为灰数区间,并与ARMA模型相结合,利用BP神经网络解决造价预测中的非线性问题,利用ARMA模型解决造价预测中的线性问题,形成ARMA-BP模型,并对其进行性能检验和实证分析。结果显示ARMA-BP模型的单方造价误差百分比为0.0001%,单工日人工消耗误差百分比为1.3995%,钢材消耗量误差百分比为0.5238%,水泥消耗量误差百分比为0.0623%,预测准确率最高。由此可见研究设计的ARMA-BP模型能有效对房屋工程造价进行预测,并且形成更为高效精准的预测结果,具有一定实际应用意义。

In order to solve the problem that the house cost forecast can′t adapt to the external changes and form the accurate forecast result,the BP neural network algorithm is improved,the weight interval between input layer,hidden layer and output layer is designed as grey number interval,and combined with ARMA model,the nonlinear problem of cost prediction is solved by BP neural network.The ARMA model is used to solve the linear problem in cost forecasting,and the ARMA-BP model is formed.The results show that the error percentage of the ARMA-BP model for unit cost is 0.0001%,the error percentage of labor consumption per working day is 1.3995%,the error percentage of steel consumption is 0.5238%,and the error percentage of cement consumption is 0.0623%,indicating the highest prediction accuracy.From this,it can be seen that the ARMA-BP model designed in the research can effectively predict the cost of housing engineering,and form more efficient and accurate prediction results,which has certain practical application significance.