Research On The Acquisition,Management And Application Of High Frequency Mass Instruction-domain Electronic Data

The processing of a massive amount of data as well as efficient acquisition and storage are important for the introduction of data science into manufacturing.In this paper,a design method of instruction-domain electronic data acquisition and management system based on cloud computing and big data technology is proposed,which can effectively deal with the characteristics of high sampling frequency and large amount of instruction-domain electronic data.This paper also proposes a method of NC machining quality analysis based on instruction-domain electronic data.By improving the traditional classification algorithm,the accuracy of following error sequence classification is improved effectively,and the quality of machining is evaluated according to the following error.The instruction-domain electronic data of CNC machine tools is one of the main sources of big data in manufacturing industry.It is necessary to acquire and store these data because the manufacturing task can be optimised by these data,and the health status of the machine can be diagnosed by these data.However,because of the massive amount,continuity and strict sequence timing of data,a data acquisition and storage system that considers scalability and throughput performance is necessary.This paper presents a method for instruction-domain electronic data acquisition and parallel storage for practical industrial production using the distributed message-oriented middleware Redis/Kafka and distributed file system HDFS and data parallel processing open-source framework Spark.In order to integrate the existing IT resources of manufacturing enterprises and realize the allocation of resources according to their needs,this paper proposes a private cloud building strategy based on Open Stack for manufacturing enterprises,and designs a private cloud management software for manufacturing enterprises.In order to make more reasonable use of limited storage resources,according to the periodicity and timeliness of manufacturing data,this paper puts forward the management strategy of big data in manufacturing industry,to make full use of limited storage space.Traditional NC workpiece quality discrimination depends on manual detection.This paper presents an instruction domain electronic-controlled datadriven machining quality evaluation method,which combines DTW distance with K-NN classification algorithm,so that the accuracy of classification of high dimensional following error sequence is greatly improved.This paper realized the acquisition,storage,management and application of instruction-domain electronic data in manufacturing industry,and has been deployed in several manufacturing enterprises.The application performance shows that the system designed in this paper can meet the demand of high throughput in the manufacturing field,and the storage system can be expanded dynamically.The system can provide steady support for instruction-domain electronic data collection and management.