| 引用本文: | 李文博,王大轶,刘成瑞.有干扰的控制系统故障可诊断性量化评估[J].控制理论与应用,2015,32(6):744~752.[点击复制] |
| LI Wen-bo,WANG Da-yi,LIU Cheng-rui.Quantitative fault diagnosis ability evaluation for control systems with disturbances[J].Control Theory and Technology,2015,32(6):744~752.[点击复制] |
|
| 有干扰的控制系统故障可诊断性量化评估 |
| Quantitative fault diagnosis ability evaluation for control systems with disturbances |
| 摘要点击 2222 全文点击 951 投稿时间：2014-06-29 修订日期：2015-02-15 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2015.40605 |
| 2015,32(6):744-752 |
| 中文关键词 可诊断性量化评估 控制系统 随机干扰 等价 空间方法 巴氏系数 |
| 英文关键词 quantitative fault diagnosis ability evaluation control systems stochastic disturbance parity space methods Bhattacharyya coefficient |
| 基金项目 国家自然科学基金项目(61203093, 61004073), 国防重点实验室基金项目(9140C590304130C59016)资助. |
|
| 中文摘要 |
| 故障可诊断性的量化指标在控制系统设计过程中极为重要. 为此, 我们提出了一种适用于线性动态系统的故障可诊断性量化评估方法. 考虑到过程和观测噪声等干扰因素对评估结果正确性的影响, 我们采用等价空间方法获取系统输入/输出与故障之间的解析冗余关系, 将故障可诊断性评估问题转化为概率统计中多元分布的差异度判别问题. 引入巴氏系数(Bhattacharyya coefficient, BC)对多元分布之间的差异度进行量化, 通过严格的数学证明得到可诊断性量化指标, 并给出具体评估流程. 以卫星姿态控制系统为仿真算例, 将本文所提评估方法应用于该系统; 仿真结果表明: 该方法能够在不依赖于任何诊断算法的前提下, 定量分析故障诊断的难易程度. |
| 英文摘要 |
| Quantitative evaluation of fault diagnosis ability is important for a control system in its design phase. We propose a novel approach to evaluate quantitatively the diagnosis ability of a linear dynamic system. To avoid the influence of disturbances such as process noises and measurement noises on the precision of the evaluation results, we use the parity space method to convert the problem of evaluating fault diagnosis ability for a stochastic linear system to an issue of distinguishing the discrepancies of different multivariate distributions. Bhattacharyya coefficient is introduced to quantify the discrepancies between two different distributions, and then a novel quantitative diagnosis ability criterion is developed by using rigorous mathematical derivation. The effectiveness of the proposed approach is tested through a numerical simulation on the model of a satellite attitude control system. The simulation results show that this approach can quantify the difficulty level in diagnosing a fault, independent from the specific diagnosis algorithm in use. |