基于证据推理的旋转机械故障诊断模型

Diagnosis Model of Rotating Machinery Fault Based on Evidence Reasoning

针对实际工况运行下的旋转机械各故障对应的量纲一的指标的范围难以严格区分的问题,提出了一种基于量纲一的指标和证据推理(Evidence Reasoning,ER)的旋转机械融合故障诊断模型。该模型利用ER算法在处理概率不确定性、模糊不确定性及非线性融合等方面的优势,通过信息变换技术将输入信号变换成信度分布结构,应用解析ER算法对输入数据进行融合,最后通过一种简单的决策规则得到诊断结果。实证分析结果表明:该方法可以有效地提高旋转机械设备故障诊断的识别率。

Aimed at the overlapping problem of range of dimensionless parameter of each fault of rotating machinery under actual working conditions,a fusion fault diagnosis model for rotating machinery based on dimensionless parameters and evidence reasoning （ER）was proposed. The model was taken advantages of using ER algorithm in solving probability uncertainty,fuzzy uncertainty and non-linear fusion. Input signal was converted to the form of belief distribution by the information transformation technology,and then the analytical ER algorithm was adopted to fuse the input data. Finally,the diagnosis results were achieved through a simple decision rule. Experimental analytic result demonstrates that the method can remarkably improve recognition ratio of fault diagnose for rotating machinery equipment.