基于重构的传感器故障诊断贡献分析

基于重构的贡献值分析方法可用于传感器在线故障诊断。但是现有方法存在2个问题:1)在2个不同方向的贡献值计算中采用相同的交叉项;2)在每个故障方向的贡献值计算中包含与该方向无关的项。为了解决上述问题,该文提出2种改进的基于重构的贡献值分析方法。在改进的方法中,通过引入权重因子来区分交叉项在不同方向的贡献值中所起的不同作用。仿真结果以正确诊断率的形式给出。仿真结果表明:与传统的基于重构的贡献值分析法相比,改进的方法能有效提高传感器在线故障诊断的正确诊断率。

Reconstruction-based contribution analysis for sensor fault diagnostics

Reconstruction-based contribution analysis can be used in sensor fault diagnosis.But existing methods have the following two drawbacks,1) using the same cross-terms when calculating contribution values of two different directions;2) when calculating contribution value of a direction,items unrelated to it is included.In order to solve these problems,this paper proposes two improved algorithms for reconstruction-based contribution analysis.Weighting factors are introduced in the improved algorithm to discriminate the cross-terms of two different directions.The correct diagnosis rate is given in the simulation part.Simulation results show that the proposed algorithm can improve the correct diagnosis rate in sensor fault diagnosis.