基于VPMCD与DS证据理论的PT燃油系统故障诊断

A Fault Diagnosis Approach of PT Fuel System Based on VPMCD and DS Evidence Theory

根据PT燃油系统共轨低压信号的特点,提出一种基于VPMCD(Variable predictive model basedclass discriminate,简称VPMCD)和DS证据理论的故障诊断方法。对信号的时域和时频域指标采用VPMCD方法分别进行建模,按照预测误差最小的原则对故障模式进行判断。当故障的模式难以判断时,将VPM模型作为证据体,由故障训练样本中心的输出确定证据体的信任度,按预测误差的均方根计算基本概率分配,最后采用DS证据理论对各证据进行合成及决策。实践表明,VPMCD与DS证据理论的结合使用能够准确识别出PT燃油系统故障。

According to the characteristic of common rail oil pressure signal of PT fuel system, a fault diagnosis method based on variable predictive model based class discriminate （VPMCD） and DS evidence theory was proposed. Variable predictive models of time domain and frequency domain were established respectively by VPMCD, and fault category was determined on the principle of least forecast error. If the fault category was not easy to determine, the models of VPM were extracted to serve as the evidences, the trust degree was calculated by the output of the center of various faults training sample, the basic probability assignment for evidence was calculated by the RMS value of the forecast error, finally, the combination and decision-making were achieved by the DS evidence theory for the given evidences. Diagnosis in a PT fuel system gave a better result by combining VPMCD with DS evidence theory.