基于多信息融合的设备管理策略模型研究

Research on equipment management strategy model based on multi-information fusion

近年来,随着物联网、大数据、人工智能、云计算等新技术的日新月异,电力生产设备也呈现自动化、智能化、环保型等发展趋势。海量的工业数据正在向云端迁移,大数据概念日渐深入人心,数据挖掘与大数据分析贯穿于设备生产运行过程中,这对于电力企业设备的智能化管控,提出了更高的要求。然而,传统电力生产企业在追求设备智能化的进程中,对于设备的智能化管理依然存在着某些短板。比如各种设备运检类业务系统存在信息孤岛问题,系统间信息和数据融合度、互动性差,设备状态评估依赖人工经验,脱离现场运检数据,严重影响了设备管理决策的科学性。文章引入文本挖掘技术,实现以设备多源结构化和非结构化数据的融合;通过计算相对隶属度,对设备故障描述进行对比分析并采用对比词云进行可视化建模;基于Bathtub曲线和灰色关联分析,识别设备缺陷与时间、厂商以及故障之间的关系;通过建立设备缺陷事件贝叶斯网络,对事故缺陷的演化发生进行预测和推断。通过深入的数据挖掘分析,构建设备管理策略模型,为设备管理从业务驱动向数据驱动转变提供了基础,由依赖人工经验向让数据说话转变提供了可能,进而为设备运行维护、检修技改、设备选型提供决策支持。

In recent years,with the rapid development of new technologies such as Internet of things,big data,artificial intelligence and cloud computing,power production equipment also presents development trends such as automation,intelligence and environmental protection.Massive industrial data is migrating to the cloud,and the concept of big data is gaining popularity.Data mining and big data analysis run through the process of equipment production and operation,which puts forward higher requirements for the intelligent management and control of equipment in power enterprises.However,traditional power production enterprises still have some shortcomings in the intelligent management of equipment in the process of pursuing intelligent equipment.For example,various equipment operation inspection business systems have the problem of information island,the degree of integration and interaction of information and data between systems is poor,and the evaluation of equipment status relies on manual experience and is separated from on-site operation and inspection data,which seriously affects the scientific nature of equipment management decisions.This paper introduces text mining technology to realize the fusion of multi-source structured and unstructured data;by calculating the relative membership degree,the equipment fault description is compared and analyzed and the comparison word cloud is used for visual modeling;based on Bathtub curve and gray correlation analyze,the relationship between equipment defects and time,manufacturer and fault is identified;through the establishment of the Bayesian network of equipment defect events,the evolution of accident defects can be predicted and inferred.Through in-depth data mining and analysis,the establishment of equipment management strategy models provides a foundation for the transformation of equipment management from business-driven to data-driven,and provides the possibility for the transformation from relying on manual experience to allowing data to speak,so as to provide decision support for equipment operation,maintenance,overhaul,technical transformation and equipment type selection.