针对动态系统的多故障诊断模型与优化算法  

Research on Dynamic Multiple Fault Diagnosis Model and Diagnosis Algorithm in Dynamic System

作　　者：刘晓芹[1,2] 黄考利[1] 连光耀[1] 王宝龙[1] 杨桦 

机构地区：[1]军械工程学院,河北石家庄050003 [2]陆军航空兵学院,北京101123 [3]南昌陆军学院,江西南昌330103

出　　处：《计算机测量与控制》2010年第2期249-251,266,共4页Computer Measurement &Control

基　　金：国家自然科学基金(60771063)

摘　　要：一般的静态故障诊断模型无法描述状态随时间变化的系统,为此用隐马尔可夫模型对动态变化的系统状态进行建模,给出了动态多故障诊断问题的形式化描述;该问题的目标函数是典型的集合覆盖问题,属于NP难解问题。通过将原始目标函数转化为若干个独立的子问题,并分别用动态规划算法进行求解,有效实现了动态系统的多故障诊断;与现有的方法相比,该方法在不影响检测率和隔离率的前提下,执行时间短,计算复杂度低,优化效果明显。Static multiple fault diagnosis models are unable to detect the changes of states in the system over time, so a Markov model and a formal define of DMFD （dynamic multiple fault diagnosis） are given to overcome the invalidation of static multiple fault diagnosis model in some cases. The optimal solution of objective function is a traditional set covering problem, which belongs to NP completeness problem. The paper decomposed original DMFD problem into several separable subproblems, and solve each of them by dynamic programming algorithm. Its optimal speed is faster than existing methods, and the overall computational complexity is reduced, so the optimal results are also better.

关 键 词：控制理论 动态多故障诊断 隐马尔可夫模型 动态规划 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]