The Cost Audit Method Of Bp Neural Network-Based Residential Building Design Stage

The Bill of Quantities A paradigm shift in the field of project cost management, the country’s ongoing reform, the progressive realization of fixed pricing to the government’s macroeconomic regulation and control, corporate self-quotations, the formation of prices in the market ". The greatest impact on the design phase of the project,investment in residential projects in the residential construction process. Residential construction parties increasing emphasis on cost management in the design phase.Therefore, to explore a workable, practical residential architectural design phase of the project cost audit approach has become the urgent needs of the construction industry.This paper analyzes the main factors affecting the project cost in the residential construction process. The architectural features of the house and consumed in the process of building artificial materials machinery costs affect the determination of residential project cost. The architectural features of the house、based on the characteristics of the residential construction to ensure that similar projects. And Draw Search by characteristics similar to a flow chart of the project. Based on the analysis of the residential construction cost factors.the establishment of the cost of the correction factor. To reduce the cost correction factor due to the time factor, labor, cost of materials price fluctuations, the impact of the project cost. Validation of the prediction model, with a construction-in-progress as a pending project, and as a similar project with eight characteristics, subject to audit, similar project has been built residential projects. Residential project quantities audit model, the characteristics of a similar project to quantify the value of BP neural network input vector, to similar residential construction engineering major projects as the output vector. First eight teachers group training, pending approving project architectural features quantized values predicted by the model as an input vector obtained model predicts the pending works predictive value occurs. And then to predict the actual value of the audit value, subject to audit for comparison. In the residential construction project cost audit model, the first major projects of a similar project as the input vector, similar residential project cost of the project as the output vector. Training.teacher group of eight major projects subject to audit projects as input vector, to draw pending engineering prediction audit value predicted by the model. Then predicted audit value, subject to audit, engineering project cost for comparison.The establishment of standards of residential project cost data acquisition cost audit model data. Similar work is based on the model, to improve the prediction accuracy. Static historical cost data, residential building construction, decoration engineering and installation works cost of correction.Reducing the time factor on the accuracy of the model. MATLAB toolbox has been calculated,the audit results and the actual value model is feasible through the use of actual cases. Compared using MATLAB toolbox calculated results with actual cases through actual case,the error is within5%. Validate the model calculation, efficient, and effective high accuracy of the calculation.