核辐射探测器模拟电路混合故障智能诊断与定位

Intelligent Diagnosis and Location of Hybrid Fault in Analog Circuit of Nuclear Radiation Detector

故障的智能诊断技术在机械、电力等领域得到了很好的研究,但核辐射探测器模拟电路的故障智能诊断研究工作仍欠缺。本课题选用了一种辐射能谱仪的模拟电路,除了正常状态外,设置了26种1~3个器件参数偏移30%的多种混合故障,应用支持向量机(SVM)直接对电路输出的波形信号进行识别分类。测试结果表明,故障诊断与定位的准确度较为满意,故障漏报率为0.2%;故障器件定位准确,定位准确度最低为76%。应用SVM直接对输出波形进行识别分类是一种值得继续研究和尝试投人应用故障智能诊断与定位的方法。

The intelligent fault diagnosis technology has been well studied in the fields of machinery and electric pow er,but the research on the intelligent fault diagnosis of the analog circuit of nuclear radiation detector is still lacking.In this project,an analog circuit of a radiation spectrometer is selected.In addition to the normal sta te,a variety of mixed faults with parameter deviation of 30%of 26 kinds of 1〜3 devices are set up.Support vector machine(SVM)is used to identify and classify the waveform signals output by the circuit directly.The test results show that the fault diagnosis and location accuracy is satisfactory and the fault failure rate is 0.2%.The location of the fault device is accurate,with the minimum accuracy of 76%.It is worth to continue researching and trying to apply fault intelligent diagnosis and location method to recognize and classify output waveform by SVM.