Research And Application Of Equipment Health And Failure Analysis Based On Industrial Big Data

PHM system is a systematic project that enables the detection,prediction and management of the health of complex industrial systems.It can do signal processing and data analysis for Various types of data generated in industrial systems.PHM technology transforms the health management of equipment from traditional fault management to recession management which can enable zero downtime and reliable operation with predictive maintenance.In the architecture design of current PHM system,The underlying sensor is directly connected to the data analysis tool.For each type of equipment it has a dedicated analysis suite with a strong degree of specialization,but it is lack of reconfigurability and algorithm reusability.On the other hand,the analysis equipment based on the data scale is limited by the stand-alone mode,and the analysis equipment is often connected with the monitoring equipment due to the design pattern.For large-scale data analysis of large numbers of devices,the processing power of a single device is often insufficient at low response time requirements.Based on the above requirements,a reconfigurable and generalized PHM system process is proposed.At the same time,the industrial large data analysis platform under distributed cluster is realized,which improves the applicability and efficiency of the system.In this paper,the industrial equipment health management and fault diagnosis analysis method are the main research content.Firstly,the background of the research and the development of PHM system are introduced,and its design architecture and platform characteristics are analyzed.Then the paper introduces the data mining algorithm model and Hadoop,Spark,Django and other platform technologies.After the study of the typical industrial equipment,including bearing and industrial capacitor prognosis algorithm,we understand the basic analysis processthe of general PHM system.Then the top-level design of the common PHM platform,including data acquisition mode,the six-tier architecture of the algorithm platform and the evaluation of test methods and other basic modules,is proposed.Then,the paper presents a PHM system implementation scheme based on distributed platform,including the construction,operation and maintenance,tuning and distributed transplantation of the cluster.