基于LS-SVM模型的无级变速器夹紧力控制系统故障诊断

Fault Diagnosis of CVT Clamping Force Control System Based on Least Squares Support Vector Machine Model

在不增加硬件成本和过大的运算量的前提下,研究了基于模型的无级变速器夹紧力控制系统故障诊断方法。提出了根据测得的信号以及全局和部分元件诊断模型的输出结果来完成故障隔离与定位的诊断策略;针对系统的非线性特征,采用最小二乘支持向量机建立反映其输入输出关系的非线性预测模型。测试结果表明所建立的故障自诊断系统具有较高的精度和反应速度,可以粗略地实现故障的隔离和定位。

On the premise of without additional hardware cost and excessive computing effort, a modelbased fault diagnosis method for the clamping force control system in continuously variable transmission （CVT） is studied. A diagnosis strategy is proposed, which accomplishes fault isolation and location based on the measured signal and the outputs of the diagnosis models for both global system and some components. In view of the nonlinear feature of the system, least squares support vector machine （LS-SVM） is used to establish the nonlinear predictive models, reflecting the input/output relationship. The test results indicate that the proposed on-board diagnosis system has high accuracy and response speed, and can roughly realize the fault isolation and location of CVT clamping force control system.