The Studies Based On Gray RBF Neural Network Integration In The Express Valuation Of Construction

Determining the construction cost is an important aspect of building work, especially the pre-construction estimate is particularly important for the project bidding and cost control.Because it is the starting point and the foundation for cost control. Setting up the prediction model in line with the actuality,estimating accurately and rapidly the construction cost has important theoretical significance and practical value.Previous project cost estimation model only used the gray systems or the neural network mostly,because each existed the defects,its usefulness was limited.This paper combining the advantages of the gray system with the neural network,sets up the Express valuation model based on the gray RBF neural network integration.That is,for the same problem,setting up their respective neural network model separately from different angles,through establishing ties a gray accumulation of the input layer and a gray minus of the output layer with RBF neural network to decrease its randomness, getting out the different predictive values.Then through the optimal neural network integration,these predicted values eventually are unified an output.To verify the proposed model's validity and reliability,this paper combines with the project instance of multi-storey frame structure for the study.The 40 project samples are collected,30 samples are selected as training samples,and the remaining 10 samples as test samples.First the eight major factors affecting the construction cost are analyzed from two perspectives,the former seven are classified as some scale factors,and the last as external factor.After through quantizing the engineering characteristics vectors and a gray cumulating and normalizing,applying the MATLAB program to predict from two perspectives,anti-normalizing and a gray tired minus,then after optimal integration,the final result is got out.This paper also uses the overall gray RBF,RBF neural network, BP neural network model to predict the construction cost,through a comparative analysis,we can see that RBF neural network model error is less than BP neural network,gray RBF error is less than RBF neural network model,the new model has the smallest error.Therefore,the method can significantly improve the prediction accuracy, and has a strong practicality and promotional value.