故障诊断中几种征兆自动获取技术研究

On Automatic Acquisition of Symptoms in Fault Diagnosis

研究了故障诊断系统中几种诊断征兆的自动获取机制，给出了对不同类型的征兆实现自动获取的有效方法．着重研究了轴心轨迹图形的自动识别问题，提出了利用图形的不变性矩特征的识别技术，构造了基于不变性特征矩的特征向量，定义了特征矩向量之间的距离，提出了置信度的度量方法．通过对轴心轨迹的仿真和实际识别。

The automatic symptom acquisition in the fault diagnosis system for large turbine generators is discussed. The methods of automatic symptom acquisition are given. Automatic recognition of the locus of axis using the recognizing technique for the specific features of the invariant moment is studied at length and the method of constructing the characteristic vector based on the invariant characteristic moment is proposed. The distance between the characteristic vectors is defined, and the measure for the degree of confidence of the symptoms acquired is suggested. Simulation and practical recognition of the axis locus show that the method is feasible.