【摘要】：為了兼顧數(shù)據(jù)建模的準確性和診斷的實時性,提出一種K近鄰診斷證據(jù)融合新方法.利用故障特征的歷史樣本構(gòu)建隨機模糊變量(RFV)形式的故障樣板模式,由KNN算法獲取測試樣本的K個近鄰歷史樣本,并定義它們的RFV待檢模式;經(jīng)樣板和待檢模式的匹配獲取K個診斷證據(jù),再將各特征的K個診斷證據(jù)融合,并作出故障決策;使用RFV實現(xiàn)對故障數(shù)據(jù)的精準建模,利用K個歷史樣本豐富診斷信息,并增加診斷的時效性.診斷效果在電機轉(zhuǎn)子試驗臺上得到了驗證.
[Abstract]:In order to give consideration to the accuracy of data modeling and real-time diagnosis, a novel K-nearest neighbor diagnostic evidence fusion method is proposed. Using the history samples of fault features to construct the fault pattern in the form of (RFV) with random fuzzy variables, the KNN algorithm is used to obtain the K nearest neighbor historical samples of the test samples, and their RFV mode is defined. K diagnostic evidence is obtained by matching sample and untested pattern, then K diagnostic evidence of each feature is fused, and fault decision is made. The accurate modeling of fault data is realized by using RFV, and the diagnosis information is enriched by K historical samples. And to increase the timeliness of diagnosis. The diagnosis effect is verified on the motor rotor test rig.
【作者單位】： 杭州電子科技大學(xué)自動化學(xué)院;重慶交通大學(xué)信息科學(xué)與工程學(xué)院;
【基金】：國家自然科學(xué)基金項目(61433001,61374123,61573076,61573275) 浙江省公益性技術(shù)應(yīng)用研究計劃項目(2016C31071) 重慶市高等學(xué)校優(yōu)秀人才支持計劃項目(2014-18)
【分類號】：TP277
，

本文編號：2180270