Research On Evaluation Criterion System Based On BP Neural Network For Traction Control System

Traction Control System (TCS) as a device of achieving optimum traction, better stability, controllability and drivability when accelerating on icy, snowy, wet and slippery road or sand has been more than 20 years abroad. This technology is also being actively developed in our country. Currently, the device is also taken for the optional or required equipment of their production as major automobile manufacturers in world. But as a very important step of this technique, a comprehensive performance evaluation of this device do not come into being a unified way. For the importance and urgency of the TCS evaluation problem, this paper research on the system evaluation of TCS as a professional problem.Based on the design principles and the operation process view, this paper systemic analyze what effect the TCS performance bring to all aspects of vehicle. Then the paper summarize the research achievement of TCS evaluation home and oversea, and put forward a evaluation system based on BP neural net which provide a systemic method for TCS evaluation. In additional, based on the evaluation system, the paper research evaluation means and test methods in which we can obtain the evaluation indicator. Finally, according to the test data which is the achievement of our research group's TCS development for many years, this paper prove the validity for the evaluation system which are put forward in the text. This paper main contain the follow 4 parts:1. This paper introduces the basic theory and principle of TCS device. It includes control manner and achieving approach for the control of driven slid ratio. In order to realize the purposes of TCS, it analyzes the improving of TCS device that reinforces the acceleration ability and stability for vehicle on the soft and weak adhesion coefficient road. It discusses theoretic contradiction about improving two kinds of performance at the same time. And it analyses the effect of TCS to improve vehicle fuel economics and driving stability and other espects. It also discusses the relationship of these performance improvements from the view of TCS system work mechanics.2. Analyze the main factor in design which affects TCS'control quality. PID control and the logic threshold control are the common controlling means of TCS, and the setting of their parameters affects TCS'performance, this paper made a specific analysis; Brake system is major implementation components of TCS, which control the wheel brake. The hydraulic responsive characteristics of the brake system have an important impact on the vehicle ride and dynamic characteristics of TCS control. In this paper, a systematic analysis of the intrinsic link between them was made, and concluded the affection that the parameters of hydraulic upon the performance of TCS control; in addition, the paper discusses the affection that the anti-interference of components and the TCS control strategy upon the quality of TCS control and the response of vehicle.3. In order to obtain the evaluation indicator, the paper summarized all the possible working condition of TCS from the design view of TCS system. These conditions include the normal condition and other non-normal condition, and on basis of these conditions, the paper put forward evaluation indicators and the way to obtain the indicators. Then the paper sorted the evaluation indicators according to the vehicle performance, and established the evaluation system, and studied the evaluation methods of every layer of evaluation system; finally, as it is very difficult to identify the weightiness value between the decentralized evaluation indicators, the paper put forward the evaluation system based on BP neural net.4. This paper take our research group's HIL(Hardware-in-Loop) test bed and systemic rapid development on vehicle as the evaluation object and data origin. The paper identify a number of evaluation indicators using the evaluation method on the basis of the TCS test bench, and experts made comprehensive subjective evaluation on groups of TCS evaluation indicators, so the paper obtain the comprehensive evaluation value of TCS; Using the comprehensive evaluation value as the samples of neural net training; The paper build the BP neural net model under the Matlab/simulink, and then trained the BP neural net using the comprehensive evaluation, and finally completed the BP neural net tests. The paper prove the validity of the evaluation system, and put forward a new method for the evaluation of TCS.