Research On Prediction Of High-rise Residential Project Cost Based On PSO-BP Neural Network

The rapid and accurate prediction of project cost is the focus of industry professionals and many scholars.The prediction of high-rise residential project cost is the premise of project proposal and feasibility study,and also the basis of quota design.For the feedforward management of project cost,it is of great significance to accurately and effectively predict the project cost.In recent years,the research on project cost prediction based on artificial neural network is widel used.However,in the most representative feedforward neural network,that is BP neural network,the shortcomings of slow convergence and poor generalization performance stands out.Therefore,this study improves BP neural network by particle swarm optimization algorithm,and proposes a prediction method based on BP neural network of particle swarm optimization,and applies it to the high-rise residential project cost prediction.Firstly,on the basis of reading a large number of relevant literature and taking composition and influencing factors of high-rise residential project cost into consideration,a comprehensive and objective index system of high-rise residential project cost prediction is established.For these indexes,principal component analysis is used to reduce dimension,and independent comprehensive indexes are obtained,which are used as the input vector of the prediction model,thus improving the learning efficiency of the model.Secondly,BP neural network toolbox is used to build a prediction model of high-rise residential project cost.Aiming at overcoming the defects of BP neural network in gradient descent method to update weights and thresholds,PSO algorithm is used to optimize BP neural network by virtue of the advantages of particle swarm optimization algorithm in parameter optimization field.The weights and thresholds of each neuron in the BP neural network structure are encoded,and the most appropriate weights and thresholds are found through the intelligent search of each particle,so that the BP neural network converge faster,is more widely used and predicts more accurately.Finally,taking the high-rise residential project in Lanzhou in the past three years as an example,44 groups of sample data are collected and sorted out.which are then preprocessed by SPSS.14-19 groups of data are selected as test samples,and the remaining 38 groups of data are used as training samples.The results show that the prediction model of particle swarm optimization BP neural network is of higher accuracy,which verifies that the BP neural network model of particle swarm optimization can meet the requirements of project cost management and control,and has certain practical significance.