Fault Diagnosis Of CVT Based On TMS320LF2407A

The electric control of automobile improves its safety, dynamic and economy, but it also brings about new problems. One hand, the complexity of electric control unit (ECU) makes the car maintenance more and more difficult. On the other hand, the faults of ECU often result in malfunction and severe failure of cars. Aiming at these problems, it's essential to monitor and diagnose the faults of cars. On account of fault diagnosis, the operation becomes safe and the maintenance becomes simple. As the ideal transmission mode for vehicle, continuously variable transmission (CVT) has the feature of simple configuration, convenient operation and good performance. In this dissertation, the fault diagnosis system of CVT has been put forward with the study of advanced fault diagnosis methods, and the main contents are as follows:The structure and work principle of CVT are analyzed. And the category and characteristics of auto faults are studied. The primary theories and methods of fault diagnosis are also expatiated.The hardware structure of CVT fault diagnosis system is established with TMS320LF2407A as the core chip. Through analyzing the signals of CVT system, proper sensors are chosen to implement data acquisition. The hardware circuits are devised, including signal processing circuits, serial communication interface circuit, CAN controller drive interface circuit and power supply circuit.In the lower computer, the design of software main frame and each functional module is presented. DSP software is compiled and debugged under integrated development environment CCS2.0, and data acquisition is achieved. The program of serial communication is build as well, so data exchange between lower computer and upper computer is realized.On the basis of studying the basic theories of neural network, the structure, learning process and design factors of BP neural network are analyzed from the view of application. Finally, the structure and algorithm of BP neural network are developed to meet the need of auto fault diagnosis; the simulation of BP neural network that proposed is accomplished. The simulation results indicate that the fault diagnosis system based on BP neural network with momentum and self-adaptive learning rate algorithm is effective.