In order to accurately

fastly and high efficiently detect mass fraction of each gas in automobile exhaust

a kind of automobile exhaust detection system is designed by combining sensor array and neural network technology. In order to improve fitting and prediction ability of neural network model

a pseudo-inverse-BP integrated neural network model is put forward. Pseudo-inverse method is used to get the optimal weight coefficient integrated different BP neural network

BP neural network is integrated by weighted average method. Pseudoinverse-BP integrated neural network model is used for regression analysis of sensor array signal. The results show that relative error predicted by pseudo-inverse-BP integrated neural network model is less than 5 %

compared with traditional Adaboost-BP integrated neural network model

this model is more simple and has faster convergence speed and higher precision of convergence and prediction.