The Automatic Transmission Control Method Study Based On Artificial Neural Network

Automatic mechanical transmission (AMT) is improved on the base of the dry clutch and the manual gearshift clutch. AMT reserves many benefits about the manual clutch such as higher transmission efficiency, low cost, reliable operation, and realizes the gearshift automation. Because AMT has the good reservation and easily rebuilt, it greatly meets China's situation at present and owns a wile development foreground.Vehicle automation variable-speed technology has already entered the intelligent age. People haven't satisfied the simple function, while people starts to pursue higher vehicle intelligence. ANN possesses the capacity of the distributed memory and parallel disposal communication and may simultaneously transfer and dispose the communication. Besides ANN has the redundancy character in the parallel disposal mechanism. So its permission error capacity is strong;its control precision is high. ANN self possesses nonlinear characteristic. So it can realize the nonlinear mapping. Because ANN controller has self-study and self-adaptation function, ANN control system can attract a great deal attention. So ANN control system can carry through the study control for the he sealed object according to the definite study rule. Through ANN modeling as long as MATLAB trains the process in-out data of locale measurement, ANN model can be acquired, and in-out characteristic and fact process are equivalent. ANN modeling needn't analyze the process or object interior characteristic, which is greatly convenient for the sealed system modeling.The paper firstly introduces AMT development summarization, control principle, control equipment composing, and hydraulic execution organization;secondly introduces ANN application field, characteristic, data unitary disposal and RBFN model, work characteristic, study algorithm. The paper emphatically selects the clutch jointing process as the research object and analyses the clutch jointing and separation process, influence factor, value index and the clutch jointing measurement and speed. At the same time the paper introduces the function and characteristic of neural cell chip (ZISC036) based on RBF algorithm and PIC SCM (Single Chip Micyoco), establishes the clutch jointing control strategy based on RBF network. Last the paper discusses AMT optimization gearshift rule. The handy data is trained and simulated by MATLAB. The simulink result shows RBF network discerning method can solve the gearshift-recognizing problem. So the paper researcher builds the theory foundation for AMT application and improved design.