Bus Driving Behavior Control Method Based on Improved BP Neural Network
In order to reduce the occurrence of urban intelligent public transport safety accidents,the driving behavior data is obtained through the driving assistant supervision system,the driving behavior characteristic values are extracted,and a control method of intelligent public transport bad driving behavior is proposed in combination with the influencing factors of driving behav-ior characteristics to solve the problem of urban public transport safety driving.BP(Back Propagation)neural network technology is used to learn and train the bad driving behavior data,model and analyze the historical driving behavior data,and realize the orderly supervision and intelligent assistance of the bad driving behavior of public transport.The system collects the data of the surrounding environment inside and outside the bus,combines the improved BP neural network algorithm with the principal component analysis method to identify and judge the driving behavior,learns the bad driving behavior training set data,updates the network weight and threshold,reduces the network error,expects the output results,reduces the defects of multiple collinear driving behaviors and da-ta redundancy,and realizes the parameter matching of the bus rear end collision alarm and anti-collision algorithm.The experimen-tal results show that the contribution rate of the first six principal components reaches 69.731%through preprocessing the data of the 16 principal components of bad driving behavior.When the 16th principal component is used,the contribution rate reaches 100%.The accuracy rate of the improved BP network is 85.33%,which is higher than the classification accuracy of BP network and DBN network.It meets the requirements of the prevention of bad driving behavior of urban public transport and provides a basis for rear end alarm and collision prevention of public transport.The example proves that the system and algorithm are practical and reliable.
intelligent public transportationdriving behaviorassistant supervision systemBP neural networkprincipal component analysis
万方数据
维普
