Research On Fault Diagnosis For AGC Hydraulic System Of Rolling Mill

With the improvement of strip quality requirements and the increasing intellectualization and informatization of equipment,the fault diagnosis of rolling equipment has become an important means to ensure the safe and reliable operation of the system.The HAGC system of the rolling mill is an indispensable key link for modern plate and strip mill to realize high-precision rolling.The system has complex failure mechanism and high failure rate,which is the key and difficult point in maintaining the rolling mill and also the main reason for the decline of product quality.In this paper,a rolling mill hydraulic AGC system as the research object.A case-based reasoning as the main body coordinate system was set by using case-based reasoning,deep belief networks,clustering analysis and rough sets built.It can provide decision support for the rolling mill HAGC system of fault diagnosis.Then a remote fault diagnosis system was built.In order to achieve the expected goal,the following studies were mainly carried out:(1)Aiming at the problem of few failure samples of hydraulic system in the rolling mill and difficulty in on-line failures simulation of complex equipment,AMESim simulation software is used to model the system,so as to obtain the equipment operation data for comprehensive analysis.Through modeling and simulation,a variety of faults can be simulated,a knowledge base of fault cases is constructed to provide effective and complete data support for fault diagnosis of HAGC system of rolling mill.(2)Fault feature extraction is the primary key task of fault diagnosis.Aiming at the uncertainty,complexity and empirical dependence of traditional fault feature extraction methods,an adaptive fault feature extraction based on deep confidence network is proposed.Deep learning is applied into the field of hydraulic fault diagnosis of rolling mill.Compared with the traditional methods,it has shown good effects in feature classification and self-adaptability,and is very suitable for dealing with the difficult problem of industrial "big data" fault diagnosis in the new era.(3)By using case reasoning based on knowledge learning,the hierarchical case organization form is proposed.Representative cases are obtained through clustering,so as to improve the retrieval efficiency.Aiming at the lack of objectivity due to the excessive reliance on expert experience for the existing method of weight assignment of subjective indexes,this paper introduces rough set theory into the determination of index weight,and describes incomplete and uncertain information.The method of the nearest neighbor index is used for case retrieval to obtain the fault solution.Finally the simulation platform data is used to verify the practicability and effectiveness of the case-based reasoning method.(4)In view of the uncertainty characteristic of the hydraulic system failure,and at the same time,the fault diagnosis needs to consume a lot of manpower,financial resources and material resources,etc.By combining with Internet technology,the database is built and the HAGC remote fault diagnosis system of the rolling mill is designed and developed.By using the system,the expert in this field can be centralized to provide technical support for equipment fault diagnosis in a short time.