摘要
提出了一种故障特征化的方法和基于距离分类的故障模式识别方法，既利用参数估计拓展了模式识别的特征空间，又避免了基于模型故障诊断方法求解非线性特征方程Ｐ＝ｆ１（θ）的困难，提高了故障诊断定位能力。
In order to diagnose faults inore cffectively, lsermann in several papers made use of parameter estimation to enlarge characteristic space[3,4]. In this paper, the authors also make use of pBLrarncter estimation in a way that is different from and, it is bclicvcd, easier than that of Isermann.Isermann's method requires the solution of a complicated non-linear equation: cqu.ation relating model parameters to physical coefficients. Such a difficulty is avoided in the authors' moth(3d. To make the essentials of the authors' method understood more castly, a specific six physical element example is given. There arc two steps in the authors' method: fault characteristics ex traction and fault diagnosis.Fault characteristics extraction. The scnsitivities S of model parameters θj(in the example,θs are α0, α1, α2, α3, b1, b2) with respect to physical element coefficients Pi(in the example, Pis are L,RO, R1, C1. CD and RH) are determilled by either eq.(1) or eq.(2). With S determined,, the tolerant thresholds of model parameters △θij are determined with eq.(3) from given tolerant thresholds of physical coefficients △Pi. For the example, numerical values of △Oij are given in Table 2. Then, without Tnuch difficulty, the characteristic vectors Ci of physical clements can be obtained frogal eq.(9) and eq.(4); the computed characteristic vectors arc shown in eq .(10).Fault diagnosis. Fault diagnosis can be carried out witll eq .(6) and eq.(7). In the exam pie, fault is assumed to exist in a ce rtain physical element R0, and eq(6) and eq.(7) is employed to see: if the fault docs exist in that physical element. In the example, calculated results shows that Ro is re2llly the faulty physical clement.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
1995年第1期61-64,共4页
Journal of Northwestern Polytechnical University
