论文部分内容阅读
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。
The reconstruction based contribution value analysis method can be used to diagnose the sensor online fault. However, there are two problems with the existing method: 1) the same cross term is used in the calculation of the contribution values in two different directions; and 2) the calculation of the contribution value of each fault direction includes the term unrelated to the direction. In order to solve the above problems, this paper proposes two improved methods of contribution value analysis based on reconstruction. In the improved method, we introduce the weight factor to distinguish the different effects of cross-terms in the contribution of different directions. Simulation results are given in the form of correct diagnostic rates. The simulation results show that the improved method can effectively improve the correct diagnosis rate of on-line fault diagnosis of sensors compared with the traditional reconstruction based contribution value analysis.