The Prediction Of Residential Engineering Budget Based On BP Neural Network

Under the condition of market economy,the pricing principle of engineering cost is "the enterprise makes its own quotation for its price which formed by competitions".The budget estimation of residential engineering play an important role in the process of project feasibility study and the comparison and selection part for designs,and its accuracy or not will directly affect the investment decision of the project.In today’s prosperous market economy,the government has strengthened the reform of engineering cost valuation,and issued a series of relevant policies,especially the issuance of "valuation standard of bill of quantities of construction projects",which makes the construction enterprises move towards the high requirements of rapid valuation.The competition in the construction bidding market comes fierce.Every enterprise is making every minute and second count,and the time limit of bidding is becoming shorter and shorter.Therefore,it is very important for the construction enterprises to obtain the priority by the accurate estimation method.On the basis of reading a large number of relevant materials,combined with specific project examples,the paper makes a rapid prediction of the residential project budget by selecting a suitable mathematical model.In this paper,firstly,the fuzzy comprehensive evaluation method is used to analyze the factors that affect the housing project budget estimate,to determine the characteristic factors of the housing project,and according to its value domain,the factors can be divided into qualitative and quantitative categories.The fuzzy theory is used to solve the fuzzy membership degree of the fuzzy characteristic elements;finally the housing project budget estimate index system is selected.In this paper,BP neural network,as the main model,the optimization of model algorithm is carried out.Firstly,the genetic algorithm with global search ability is used to improve BP neural network,and the rapid estimation model of housing project budget is established,and Hooke Jeeves algorithm is introduced to optimize GA-BP neural network.Taking the influencing factors determined by fuzzy comprehensive evaluation method as the input variables and the cost per square meter as the output variables,the non-linear function relationship between the main influencing factors and the cost per square meter is established.Through sample training,the model has better approximation ability,and finally the rapid estimation of residential project budget is realized.In this paper,the program of residential engineering estimate is compiled by the MATLAB software.The index system for BP neural network is calculated by the fuzzy comprehensive evaluation method,and then BP neural network and two improved BP neural network models are established by MATLAB software.Finally,the feasibility of the three models is verified by engineering examples,and the most suitable prediction model is obtained through comparative analysis.It provides a theoretical reference for the rapid quotation of construction enterprises.Through the empirical analysis,it can be concluded that HJGA-BP neural network estimation model could meet the same accuracy conditions with the fastest training speed.The error between the predicted value and the actual value of a specific residential project is narrow,and the satisfactory effect is obtained.