Construction Project Cost Estimation Based on Improved BP Neural Network

To determine investment and cost estimation scientifically and simplify the investment estimating preparation, an improved BP neural network estimation model with GA optimization is proposed, based on the learning process of standard BP neural network. Our scheme set initial weight and whitening positioning coefficient as genetic population. The coefficients are optimized according to the principle of GA optimization, to acquire optimized engineering cost estimation model. The training samples in the same group are used for training, to compare the computing results, error curve and training performance. The results verifies the reliability of improved BP neural network. The simulation based on practical case also indicates that optimized network model can realize rapid and accurate engineering estimation.