Forecast The Car Failure Based On BP Neural Network

In general, the automobile has the breakdown appearance in the use process. Regarding the automobile producer, it would reduce the corresponding time cost and inventory cost after knowing the automobile failure rate under certain travel condition.As to this, firstly, through the automobile overall performance analysis, the article finds the possible causes to reduce the automobile breakdown, such as the automobile each installment buckle, the components quality differences, the automobile service frequency, the automobile fuel oil, the automobile travel environment, the driving technology and the driving method, the travel course and so on and etc. All of these have developed the automobile breakdown model in the new way, and analysis the weigh of the different breakdown factors influence, then the automobile failure rate control model has been got.Next, this article has used the GA-BP neural network to solve the question. On the foundation of the traditional BP network basic model, it introduces the several common kinds of improvement algorithm in the BP network: Additional momentum and learning rate auto-adapted adjustment improvement BP algorithm (BPX), Levenberg-Marquardt optimized algorithm (LM), Bayes standardized BP algorithm, and carries on the optimization on the neural network weight and bias using the genetic algorithm. It has separately carried on the training model by MATLAB on the basis of these algorithms, and forecast the data which we need through regression ability using the neural network.