基于SOM-VPRS的平衡机故障诊断

Application of Equilibrator Fault Diagnosis Based on SOM-VPRS

针对平衡机故障的特点,采集了振动信号进行故障诊断;设计了故障信息采集系统,解决了故障信息提取困难的问题,减少了噪声信号;融合自组织(SOM)网络和变精度粗糙集(VPRS)形成了SOM-VPRS算法,实现了平衡机的故障诊断。运用SOM网络进行了连续属性的离散化,采用变精度粗糙集的近似依赖模型进行属性约简,得到故障诊断决策规则,属性约简后,属性集由20个减少为7个,规则集由70个减少为34个,计算复杂度降低;对决策规则进行了验证,诊断正确率可以达到95%以上,且模型和算法具有普遍适用性。

According to the fault characteristics of equilibrator, vibration signals were collected. By designing a collecting system, prob lems of difficult collecting were dissolved and noise signals were reduced. The integration of SOM （Self--Organizing Map） and VPRS （Vari able Precision Rough Set）, coming to a new algorithms called the SOM--VPRS. The approach is realized by applying SOM to diseretive continuous attributes, using property of approximation dependency of VPRS to carry through attribute reduction and concluding decision--mak ing rules. The attribute set is reduced from 20 to 7, and the rule set is reduced from 70 to 34. So the computational complexity is decreased. Faults were given to make sure the decision--making rules good, the result said that the correct diagnosis rate can reach more than 95 %. The model and method are both of widely using.