基于特征参数退化的DC/DC变换器故障预测

Fault prediction method of DC/DC converter based on characteristic parameter degradation

针对DC/DC变换器电路级故障预测问题,研究能反映变换器整体性能退化状况的故障特征参数,并提出基于DC/DC变换器特征参数退化的故障预测方法。首先,分析变换器中关键元器件的失效机理,确定反映元器件退化规律的失效敏感参数;然后,结合元器件退化对变换器整体性能的影响,将输出电压变化率ω作为变换器的故障特征参数,该参数能同时反映变换器和元器件的退化状况;最后,基于最小二乘(LS)、灰色系统及最小二乘支持向量机(LS-SVM)三种算法实现DC/DC变换器故障特征参数时间序列预测,并对结果进行比较分析。以Boost电路为例进行仿真及物理实验,验证了该方法的有效性和准确性。

Aiming at the problem of fault prediction of DC/DC converters, the failure characteristic parameter that can reflect the performance degradation status of the converter is studied, and a failure prediction method of DC/DC con- verter based on the degradation of characteristic parameters is proposed. Firstly, the failure mechanisms of the key components in the converter are analyzed and the failure sensitive parameters that reflect the degradation rules of the components are determined. Then, considering the effect of component degradation on the overall performance of the converter, the change rate of the output voltage tois taken as the failure characteristic parameter of the converter, which can reflect the failure states of both converter and components. Lastly, Least Square （LS） method, Grey System and Least Square Support Vector Machine （LS-SVM） algorithms are used separately to predict the time series of DC- DC converter fault characteristic parameters and realize the failure prediction of DC/DC converter, and the prediction results are compared and analyzed. The Boost circuit was taken as an example to carry out simulation and experi- ment, which validates the effectiveness and accuracy of the proposed method.