Research On Faults Of Truck Air Pressure System Based On Classification Algorithm

The truck air pressure system is closely related to the safe driving of the truck,so it is important to monitor and predict the faults in the truck air pressure system in real time.In this paper,six empirical analyses are carried out on the truck air pressure system.The four algorithms used are standard machine learning algorithms,including logistic regression model,AdaBoost algorithm,BP neural network and XGBoost algorithm.At the same time,this paper improves two traditional machine learning algorithms,which are improved logistic regression model based on genetic algorithm and improved AdaBoost algorithm.Among them,the improved logistic regression model based on genetic algorithm mainly modified the loss function of the standard logistic regression model,aggravated the positive class judgment weight,and then solved the parameters with genetic algorithm.Compared with the standard logistic regression model,the average loss of the improved logistic regression model based on genetic algorithm is reduced by 61.31%;The improved AdaBoost algorithm adjusts the threshold of the discriminant function to make the discriminant function more biased to determine the positive class on the test data set.Compared with the standard AdaBoost algorithm,the average loss of the improved AdaBoost algorithm is reduced by 39.19%.A comprehensive comparison of the above six empirical analysis algorithms shows that the improved AdaBoost algorithm works best,with an average loss of 0.491.