论文部分内容阅读
提出了一种基于径向基函数网络与证据推理的模拟电路融合诊断方法,以解决模拟电路诊断中由于故障信息缺乏所致的诊断准确性问题,并提高其训练速度。采集多类电路信息,对应于每类特征参量构造一个径向基函数网络,由这多个彼此独立的径向基函数网络来完成故障的初级诊断。再用初级诊断中各子网络的输出结果构造证据体,通过证据融合推理分析,得出最终的故障定位结果。模拟实验结果表明,所提方法对于电路的硬故障与元件参数偏移较小的软故障诊断均有效,其充分挖掘了多类测试信号中的故障信息,提高了诊断结果的准确率。
An analog circuit fusion diagnosis method based on radial basis function network and evidence reasoning is proposed to solve the problem of diagnosis accuracy caused by the lack of fault information in analog circuit diagnosis and to improve its training speed. Collecting a variety of circuit information, corresponding to each type of characteristic parameters to construct a radial basis function network, from the multiple radial basis function network independent of each other to complete the fault diagnosis. Then the output of each sub-network in the primary diagnosis is used to construct the evidence body, and the final fault localization result is obtained through evidence fusion inference analysis. The simulation results show that the proposed method is effective for hard fault diagnosis and soft fault diagnosis with small component parameter offset, and it can fully mine fault information in many kinds of test signals and improve the accuracy of diagnosis results.