State Characterization of AC Contactor Based on Multi-feature Enhanced Fusion
Accurately grasping the operating state of AC contactors is the basis for realizing the intelligentization of AC contactors.To this end,a state representation method based on multi-feature enhanced fusion is proposed.Firstly,accord-ing to the characteristics of AC contactor's characteristic parameters of nonlinear,non-stationary,random change,and insignificant trend but in line with certain statistical characteristics,the Wasserstein probability distance is used to carry out feature transformation to obtain the initial state features with strong trend.Driven by the dynamic feature parameters and after the characteristics of strong coupling between parameters being taken into consideration,the auto-encoder neural network is used for feature compression and extraction,redundant information is eliminated,and useful information is re-tained.The neural network is used to realize the competitive fusion output of multi-dimensional features,to obtain the comprehensive health index of the operating state of the AC contactor,and to realize the quantitative representation of the state of the AC contactor.Finally,combined with the measured data,the effectiveness of the proposed method is veri-fied.Compared with the results from other two methods,the results show that the health index trend,monotonicity and robustness obtained by the proposed method are improved by at least 4%,24%and 5%,which can provide references for the next study on the precise control and intelligence of AC contactor.
AC contactormulti-feature fusionstate characterizationWasserstein distanceautoencoderself-organizing maps
万方数据
维普
