Distributed Complex Event Processing Technology In Smart Grid Monitoring

Smart grid will gradually promote Distributed Energy Resources(DERs)with high permeability,electric vehicles and two-way interactive technologies,active entities emerging in large numbers and its intelligent monitoring system will involve massive measurement streaming data,real-time data exchange communication and high system complexity.To establish a big data stream computing management platform with real-time online monitoring and analytics,high reactivity and scalability is particularly critical.This paper analyzed distributed real-time monitoring characteristics,and data intensive monitoring big data real-time online analysis application requrements in smart grid,applied data centric real-time Publish/Subscribe(P/S)communication middleware,to study highly distributed,heterogeneous,dynamic and self-organizing distributed big data management application architecture for real-time online data collection and active entity's monitoring data streams sharing,and made research on the application of Complex Event Processing(CEP)with high performance streaming computing,pattern matching and active responsive decision making in smart grid.The thesis' s main works are as follows:(1)To build specialized distributed real-time big data management system in smart grid monitoring was proposed,adapting real-time online monitoring computation and reactive control requirements in energy Internet of Things with highly physically distributed,heterogeneous,dynamic and self-organized autonomous entities as the core computing components.Based on the analysis of the latest general big data stream computing architecture pattern,an event driven distributed stream data processing application technology framework oriented to the smart grid monitoring system was proposed.Based on hierarchical independent design pattern,in that CEP for multi-sources event streams real-time online query and pattern matching technology as the processing layer and P/S based communication service DDS as message layer,a distributed CEP stream computing system was constructed,and key techniques and case study analysis were given.(2)In ICT,smart grid on-line monitoring applications belong to distributed real-time system integration,in that real-time communication technology becomes the research focus.This paper proposed an architecture model and information security deployment design in building real-time P/S communication service in smart grid based on DDS.Smart grid real-time communication service based on DDS needs to follow IEC 61850 standard,in view of the difficulty problem of inconsistency and complexity of IEC 61850 Abstract Communication Service Interface(ACSI)specific mapping implementation in smart grid,this paper applied DDS real-time middleware to support the universal P/S based IEC 61850 communication implementation.An unified mapping implementation method of IEC 61850 ACSI to DDS was proposed,supporting Qo S policy dynamic configuration and enforcement.As a case study and experimental verification,DDS based IEC 61850 GOOSE multicast communication implementation was given,the DDS based mapping implementation in accordance with IEC 61850 meta model is simple and practical,easy to realize general unified dynamic configuration management for ACSI communication based on self-description SCL configuration language.Further work will be done to develop DDS based user configuration utility software.(3)This paper investigated distributed CEP event model and its application system.A semantic CEP event ontology model and semantic query processing flow in smart grid monitoring were given,also a distributed CEP multi-agent service application deployment method based on DDS in layered Active Distribution Network(ADN)monitoring was proposed.Event model standardization is a fundamantal problem of providing events collection,communication,storage and analysis,and improving real-time system reactivity.There are large amounts of data and events exchange in power system dynamic interaction,currently CIM data exchange standard has proposed,however,there is no standard for events exchange that expresses system dynamic characteristics.This paper investigated standardized event ontology modeling for smart grid semantic CEP system.As a case study,an AMI monitoring event ontology knowledge base system was designed,implementing the semantic knowledge representation of monitoring event streams in theme,time and space,and its relationship,improving the semantic CEP pattern matching and query ability based on knowledge base.CEP is an advanced stream computing model,supporting multiple sources of event streams pattern matching and active responses,meeting specific needs of smart grid monitoring system for system runtime status and behaviers online analysis.This paper investigated CEP key event model,pattern matching query language and rules,and distributed CEP service architecture model,identifying its communication interface to support system scalability,the data centric DDS communication standard interface was proposed to support distributed CEP application deployment.Aiming at the hierarchical distributed control architecture in ADN with high DER permeability,deploying customizable hierarchical CEP Agent nodes to realize active real-time monitoring and management,with DDS data bus supportting asynchronous,concurrent and self-organizing characteristics of CEP agent deployment,a flexible and scalable distributed CEP multi-agent monitoring application was developed,and the Globle Data Space(GDS)oriented registration management for CEP multi-type event topics,and case study were given.(4)Making research on smart grid customers demand side two-way interactive key application technology,this paper presented the distributed monitoring service integration architecture based on DDS/CEP in Automatic Demand Response(ADR)system,and two-way interactive customers Demand Response Performance(DRP)real-time online monitoring and analysis method,and also developed distributed CEP customers DRP computation simulation experiment and verification in ADR.ADR is the best application to embody bidirectional interaction between smart grid and customers,and adapt to DER high permeability.Based on Open ADR universal event model and IEC PC118 smart grid customer interface standard,the bidirectional interaction technology application in ADR system was studied.As a case study of real-time online monitoring and analysis of Customer Baseline Load(CBL)and DRP,the CEP service node function definition,ADR event driven demand side event model design and query algorithm were given;A hybrid query algorithm for real-time and historical data streams based on extended CEP engine was proposed and its performance was experimentally tested;A distributed CEP customer DRP simulation experiment in ADR was developed.Further work will explore the practical potential of distributed CEP monitoring service integration in large-scale fast ADR system.(5)A novel data driven piecewise short-term forecasting algorithm for nonlinear load series was proposed.At present,the nonparametric model based on Kernel Method(KM)is a promising Short-Term Load Forecasting(STLF)method following the common used artificial neural network method.In this paper,a combined forecasting algorithm based on Fuzzy C-Means(FCM)clustering and KM based nonparametric model was proposed,a detailed experiment and error analysis of actual load data in a region shows that the combined model has high prediction accuracy.The forecasting model is simple,with fewer training parameters,high perfomence,and strong self-adaptability,having practicabilty in smart grid monitoring time series big data short-term prediction analytics.