基于核直接分解的过热蒸汽系统故障诊断方法

Kernel direct decomposition based fault diagnosis method for superheated steam system

中文关键词  过热蒸汽系统  质量相关  故障诊断  核直接分解  贡献图

英文关键词  superheated steam system  quality-related  fault diagnosis  kernel direct decomposition  contribution plot

中文摘要

      为提高过热蒸汽系统的运行效率并减少非紧要故障的报警率, 本文提出一种质量相关的非线性故障检测与诊断方法. 首先, 利用核函数将过程变量映射到高维特征空间以消除原始变量之间的非线性耦合. 然后, 在特征空间进行核直接分解得到两个正交子空间, 并在两个子空间中分别设计统计量指标进行质量相关的故障检测. 在此基础上, 利用偏微分贡献图提取每个变量对联合统计量指标的贡献率, 并根据贡献率大小最终确定故障变量. 仿真结果表明, 所提出的方法能够准确区分影响过热蒸汽温度和不影响过热蒸汽温度的故障, 有效降低了非紧要故障的报警率, 提高了过热蒸汽系统的运行效率.

英文摘要

      In order to improve the operating efficiency of the superheated steam system and reduce the alarm rate of non-critical faults, a quality-related nonlinear fault detection and diagnosis method is proposed. First, a kernel function is used to map the process variables to the high-dimensional feature space to eliminate the nonlinear coupling between the original variables. Then, the feature space is directly decomposed into two orthogonal subspaces, in which statistical indicators are designed separately for the quality-related fault detection. On this basis, the partial differential contribution plot is used to extract the contribution rate of each variable to a joint statistical index to achieve accurate diagnosis of the faulty variables. Simulation results show that the proposed method can accurately distinguish the faults that affect the temperature of superheated steam and those that do not affect the temperature of superheated steam, which effectively reduces the alarm rate of non-critical faults, and improves the operating efficiency of the superheated steam system.