Research On Identification Technology Of Driver’s Braking Intention Based On Fuzzy Neural Network

In order to adapt to the accelerating pace of society, and to meet the higher requirement for human safety and comfort performance, more and more advanced technology, especially electronic technology, is more and more widely used in automotive products, has become the trend of the times. With the increase of the types of the driver’s brake assist system, the smooth transition of man- machine control mode has also been an important development. Brake by wire system(B-by-Wire) because of the abolition of the traditional system of mechanical connection, using sensor driver’s intent, and then by the central control unit sends a command to the corresponding actuator, to complete the driver related operations, has been widespread concern.Visible, to achieve the smooth transition of man-machine model, all need to identify the driver’s braking intention.In this paper, based on the analysis, summary and study of the domestic and foreign existing research results of driver intention recognition and prediction, the method of fuzzy neural network is selected to establish the model of driver’s braking intention identification.Using dSPACE and CarSim/Simulink to build passenger car driving simulator for the typical braking test and to get test data for training model, and application of off-line training method obtain driver’s braking intention recognition model. On this basis, with the help of driving simulator, do test to authentication the driver braking intention recognition model in this paper.Firstly, the advantages and disadvantages of common methods of artificial intelligence were comparative analysis, choose the fuzzy neural network method as the theoretical basis of intention recognition. There is an adaptive fuzzy control system based on neuron in T-S model, the matching degree is very high, so the T-S model is selected in the typical fuzzy neural network model in this paper. Secondly, by using the brake pedal and the dSPACE connected to build passenger car simulator, through the joint simulation of dSPACE and CarSim/Simulink, simulation test of normal braking, continuous braking and emergency braking. Finally, with the screening of test parameters, the paper chooses the brake pedal displacement and its rate, braking deceleration as the driver’s braking intention identification model input.Then, an identification model for driver’s braking intention identification is established based on the ANFIS editor in the Matlab. Using principle of maximum similarity to classify the data,we get quantitative braking intention.The braking intention identification model is verified by the way of off-line and on-line verification. Compared with the three typical braking conditions, the results of thebraking intention identification can prove the accuracy of the identification method and the validity of the model are verified.