Research On Big Data Modeling And Distributed Storage Of Equipment Operation State

Equipment "zero fault" operation is an important part of intelligent manufacturing.How to extract the key information from the data includes equipment states information,to learn a operation states of equipment,predict possible failures and take corresponding measures in time,which have become the key problem to solve the "zero fault" operation of the equipment.At present,the data of the equipment operation states are increasing explosively.Because of the various sources and different types of the data,it causes the data can't be shared and taken use of efficiently.In the process of data collection and transmission,there may be outliers due to various reasons,which affect the evaluation results of the operation states of the equipment.Thus,it is of great significance to study the unified description of the big data of the operation state of the equipment,the distributed storage of massive meta data and the large data cleaning of the equipment operation states.In order to achieve the efficient sharing and utilization of big data of the equipment operation states,based on analyzing the description techniques of existing resources,on the basis of the RDF data resource description method of equipment operation states was proposed,the meta-data model was established and the equipment operation states data resource instance validation was combined to verfied generality and feasibility of the model.In view of the traditional storage mode cannot meet the demand of mass of RDF data storage,a RDF data storage model based on the distributed database HBase was proposed.Using the HBase column storage and extensibility features meet the demand of mass of RDF data storage,through the des.ign of the storage model improve the efficiency of data management.In view of the problem of substanard data quality,the big data cleaning method for equipment operation states based on time sequence analysis was proposed.The outliers in the big data of equipment operation states were classfied,the effect of different types of outliers on the modeling was analyzed.The outliers have been detected and repaired by using the method of iteration.Finally,through the equipment operation states measured data to verified the correctness of the model.In order to improve the efficiency of data cleaning,the data cleaning model and MapReduce technology were combined.The big data cleaning model for equipment operation states based on MapReduce was established and the implementation process was designed detailedly.Finally,cleaning the big data of equipment operation states to verfied the high efficiency of the model.The RDF data storage model and data cleaning experiment were evaluated in stand-alone and Hadoop cluster,RDF data sets of different sizes were parsed and loaded,the big data for equipment operation states of different sizes were cleaned.The feasibility of storage model were verified and the performance of the hadoop cluster were analyzed,for the future using Hadoop to mass data storage and analysis to provide reference and basis.