Real-time Monitor And Assessment For AT Shifting Process

In order to improve the repeatability of AT shifting process control effect, studyreal-time monitor and assessment method of AT shifting process. Compared to traditionaloff-line shift quality assessment, real-time monitor and assessment of shifting processstudied in this paper is operated in MCU, not only assessing shift quality, as itsultimate purpose is to obtain correction direction of shifting process control parameterswhen shift quality is poor, to improve shift quality of next shift.First, dynamic model of shifting process was built, and key control parameters ofshifting process and its adaptive requirements were analyzed, providing theoretical basis forthe raise of assessment indexes, the establishment of evaluation model, andthe implementation of real-time monitor and assessment.Target reference trajectory of shifting process was designed, based on analysis of thegeneration mechanism of shift impact. Overall assessment indexes and phase assessmentindexes were put forward, based on shift quality requirements and target referencetrajectory of shifting process respectively. Duration, speed, changing rate of speed weremonitored during shifting process, and assessment indexes were calculated when shiftingprocess ended, implementing real-time monitor of shifting process.Assessment model of shifting process was established, based on the principle of errorback propagation neural network and neural network pattern recognition system. Neuralnetwork in assessment model was designed and trained, using Matlab neural networktoolbox. The trained assessment model was transplanted into MCU, through the piecewiseapproximation of sigmoid function and the conversion from floating point to fixedpoint, implementing real-time assessment of shifting process.Finally, experimental studies were carried out. The results show that, the proposedreal-time monitor and assessment method of AT shifting process can satisfy therequirement of real-time property with high accuracy rate, and shift quality is improvedwhen using the output results as the revised guidance of control parameters.