Research On Data-driven Method Of Crane Digital Twin System For Intelligent Operation And Maintenance

Bridge crane(crane)is the core equipment in metallurgical production and other fields,its busy operation and high safety operation requirements.In traditional operation and maintenance,manual frequent spot inspection is still used to diagnose and analyze equipment.This method is time-consuming and laborious,with high work intensity and heavy dependence on experience.With the rise of digital twinning technology,it is a hot research topic to collect realtime operating data and analyze equipment diagnosis based on data drive.This paper presents an intelligent operation and maintenance method based on digital twinning and data drive for a large metallurgical enterprise.The main research contents are as follows:(1)In view of the lack of digital twin model in the field of crane operation and maintenance,especially in the field of data-driven,there is no complete digital twin modeling system,this paper proposes a model of digital twin system for intelligent operation and maintenance.Among them,the formal expression of the four-tuple of the operating digital twin is given,and the virtual entity of the operating virtual twin is constructed,which includes the dimensions of geometry,physics,rules and behavior.The data-driven model and operation and maintenance service model of the operating digital twin system are defined.Finally,various interaction ways of the operating digital twin are designed.(2)The paper proposes a multi-source heterogeneous data fusion method for crane operation and maintenance,including the fusion between time-series data and between time-series data and document object data,in response to the problems of multi-source heterogeneous crane operation and maintenance data,the intermingling of online and historical data,and the difficulties of data integration and fusion.Based on the detailed analysis of the feature engineering of multi-source heterogeneous data,the time alignment method of multi-source dynamic sequence data is studied,and the time registration and hierarchical fusion algorithm of heterogeneous data based on the novel cascade binary tagging framework is designed.The experimental results show that the fusion efficiency and accuracy meet the application requirements.(3)In view of the difficulties in analyzing the root causes of multi-factor hidden faults in crane operation,a data-driven multi-class support vector machine-fuzzy Bayesian method is proposed in this paper.The multi-factor hidden fault is defined,and the key influencing factors of the multi-factor hidden fault are classified by the proposed method.The discriminant network which combines the knowledge base of crane operation and maintenance with Bayesian network and fuzzy set theory is given.The experiment shows that the network can identify the root cause probability of the multi-factor hidden fault.At the same time,a question-answering model for running faults is established by combining multi-source heterogeneous data,and a question-answering service for running faults driven by temporal knowledge graph is provided.Finally,taking the intelligent operation and maintenance case of a large iron and steel enterprise as the object,the driving digital twin system is developed and put into operation.Through the field operation analysis,the multi-factor recessive fault root cause diagnosis and analysis method proposed in this paper has an obvious effect on reducing the downtime of driving equipment and improving the maintenance efficiency.