Research Of Equipment Fault Diagnosis Method Based On Dynamic Weighting Double Model

In recent years,with the advance of science and the development of modern production,the technological revolution taking manufacturing as the core has become the key to the competition of great powers.Therefore,combining the development of large data,artificial intelligence and other fields,making use of massive data to build intelligent manufacturing industry with international competitiveness has become the key to enhance national comprehensive national strength,build a world power.In the process of production,as the state data of large equipment is increasing,how to extract fault features quickly and efficiently based on data and use data mining to diagnose fault types has become a hot topic in the field of intelligent manufacturing.The equipment data is divided into operation data and manual detection data.Among them,the equipment operation data include on-line monitoring,live detection,and preventive test data,etc.the manual detection data is the text information recorded by the workers in the process of equipment maintenance,including the operation condition,maintenance,failure and so on.The data driven fault diagnosis method can transform the related,unanalyzed and high dimensional data into unrelated,easy to understand,and low dimensional knowledge by analyzing,mining and feature extraction of equipment running data.However,this method can not integrate the empirical knowledge,the data imbalance and the long training set training convergence time is too long,which reduces the effect of the model.For manual detection data,text data mining is often used to analyze and extract features,thus establishing the relationship between fault characteristics and failure modes.However,it is difficult to achieve the ideal accuracy by relying only on the artificial detection data because of the inadequacy of the results of the text data mining,such as the unstable results,the sensitivity to the data and the easy to overfit the data.In view of the above problems,this paper combines equipment operation data with manual detection data,and proposes a dual model equipment fault diagnosis method based on dynamic weighting.First,a DBN fault diagnosis model based on fast convergence optimization(Fault Diagnosis Model Based on Fast Convergence DBN,for short,FCDBN)is proposed to solve the problem of non balanced processing of equipment running data and the problem of fast extraction and classification of fault modes.Secondly,a fault diagnosis model based on text data mining is proposed.Fault Diagnosis Model Based on Text Data Mining,called TDM),is used to solve the problem of text preprocessing,topic model feature extraction and classification ofartificial detection data.Finally,a weighted combination algorithm is proposed and a fault diagnosis method for dual model equipment based on dynamic weighting(Equipment Fault Diagnosis)is proposed.Based on Dynamic Weighting Double Model,referred to as DWD).This method overcomes the characteristics of non-equilibrium and high dimensionality of mass operation data in equipment fault diagnosis,and achieves good results.Further,the Matlab2015 b software is used to model the equipment data provided by a company.It is proved that the DWD method can not only unbalance the equipment operation data,extract and classify the valuable information,but also effectively combine the empirical knowledge of artificial detection data for text data mining.This method can significantly improve the accuracy of fault diagnosis and get better performance evaluation indicators,which has better theoretical and practical value.