Application Research On The Life Cycle Management Of Measuring Assets Based On Big Data

With the structural reform of state grid and construction of big marketing system, intensive management of metering assets in provincial electric company has become the core business of metering center. Entire life-cycle management is the best method to achieve this goal. Since measuring assets have the characteristics of large quantity, low unit price, widely used and easily replaced and have close relationship with national economy and people’s livelihood, quality management becomes the core part of entire life-cycle management. Production scheduling platform of provincial metering center has the function of whole process scheduling (including metering equipment procurement, inspection, verification, warehousing, distribution) and coordinated control of monitoring automatic test lines and intelligent warehousing. This has created favorable conditions for construction of management and control platform and the entire life-cycle management of metering assets in the headquarters of State Grid. However, with the analysis of production capacity and current data situation in provincial metering center, it can be concluded that data of semi-structured, audio, video, image and so on are mostly stored by documents, the retrieval of documents, audio and video and other unstructured documents as well as analysis of mixed data types are not realized. Conversion of unstructured data to structured data and panorama analysis of multiple types data are not realized. System can only determine the current status information of metering equipment and cannot achieve the function of large-scale, long cycle equipment status analysis and statistics. In the relational database, the device information data base and rule match mode are adopted for data analysis. Existed conventional technology cannot meet the analysis equipment quality data. The application of Big Data technology for quality analysis of metering equipment has become the key problems to be solved.This paper gives deep research on quality analysis of entire life-cycle management of metering assets. Life-cycle quality analysis model based on triangular fuzzy number analytic hierarchy method is established. Based on MDS big data analysis platform and data extraction program, the on-line and off-line data extraction are conducted. Thereby, research on comprehensive quality assessment, the potential correlation analysis, trend forecasting, real-time alarm test, the meter quality prediction can also be conducted. Big data technology is used for deepening and refining marketing measurement in all aspects of life-cycle test and fault data. Practical applications and data law are researched, which can help to the formation of multidimensional and multiple data processing models and algorithms, so multidimensional analysis scene including time, space and various business processes is established.The main contents of this paper are as follows:(1) Life-cycle quality analysis model based on triangular fuzzy number analytic hierarchy method is established. The quality analysis system model of every stage is divided into three levels:The quality target layer, quality analysis criteria layer and quality characteristics indicators layer. To determine the hierarchy mode of quality analysis system. Integrating error data of energy meter before full performance testing, full-performance test before delivery, as well as sampling, full inspection upon arrival links. Multidimensional analysis of distribution of products from suppliers operating region, and other aspects of product model data error and law; Calibration data error detection study various aspects of the smart meter, retroactive test failed projects and program selection, the main components of the relationship, and to avoid the forecast quantities due to problems caused by equipment failure verification test data fail to meet, explore meter gene defect design, quality risk aversion。(2) MDS big data analysis platform is constructed. Time stamps way is used to extract MDS platform error data, fault detection data batch information, data and failed projects SG186 marketing system, scrapped fault detection data, etc. Data integration technology for data conversion and processing is achieved by ETL. The extracted data are used for offline data mining analysis and real-time data mining analysis. Research is conducted through mathematical statistics and artificial intelligence algorithms for massive data based on a comprehensive evaluation of the quality of clustering, the meter calibration and data potentially relevant anomalies scrapped analysis, trend forecasting, real-time alarm verification, meter quality prediction based CCA method and data test failed and fault data dependence analysis。(3) Based on the error data verification, automatic test lines (or test device) information, batch information, vendor information and other data provided by MDS platform, multidimensional and deep data mining can be conducted using big data analysis techniques. The results of data mining is graphically displayed and then data sample is formed. Based on the analysis and comparison of data samples, the existence of differences between the sample and the error outliers can be found and achieve estimation of test line (or lines on the test device) potential hidden faults and potentia failure risks meter.(4) Through multidimensional and deep mining of substandard project data, the formation of various dimensional data samples is achieved. Then through the analysis and comparison of data sample, the relevance relationship of failed project categories and its subclass is deeply dug. To establish links with failed projects by manufacturer, batch, and other dimensions of the same kind of equipment, under the same external conditions, to detect the distribution of the reasons for failure, to give prompts warning of appropriate items, to form a special warning library of frequent reasons for failure, to estimate hidden fault storage device that may be present within the period of time, to estimate potential hidden faults and potential failure risks meter on test line (or lines on the test device).(5) Multidimensional and deep mining of data is conducted using big data analysis techniques. The results of data mining is graphically displayed and then data sample is formed. Thus a vendor in a batch operation of the equipment that may have hidden faults is estimated. Through scrapped mark detection data mining analysis, data state is displayed by a variety of graphical forms, and thus the manufacturers of various types of devices that can run the existence of hidden faults are estimated.