节点文献

智能变电站通信系统的实时性和可靠性研究

Real-time Performance and Reliability of the Communication System of Smart Substation

【作者】 赵建利

【导师】 刘教民;

【作者基本信息】 河北工业大学 , 电气工程, 2012, 博士

【摘要】 智能变电站为智能电网提供标准的、可靠的节点支撑,作为智能变电站“神经系统”的通信系统是连接变电站内各种智能电子设备的纽带,其实时性、可靠性直接决定变电站的可用性。采用不同智能模型对其各项性能进行建模分析可有效提高整个系统的可靠性。本文通过理论分析和仿真实验,取得了以下主要成果:建立了智能变电站内数据流的数学模型,分析变电站内报文传播时延构成及不确定特性,采用OPNET定量分析了不同因素下报文传播时延的不确定行为,探讨了VLAN技术对智能变电站通信网络实时性能的影响,结果表明,VLAN的设置在一定程度上降低了智能变电站通信网络的时延、控制了广播风暴的蔓延、提高了系统的安全性。采用多服务台串联排队网络建立变电站层主控单元报文处理时延的数学模型,研究由紧急情况下缓冲区出现排队现象引起的时延不确定性,推导了主控单元缓冲区大小与丢包率、时延等性能指标间的解析关系,提出了满足报文丢失率和时延要求的缓冲区大小确定方法,并通过仿真实验验证了理论推导的正确性。提出了基于服务分类机制和分形漏桶的智能变电站综合信息传输方法,利用分形漏桶对具有自相似特性的突发数据流进行流量控制,并应用网络演算相关理论计算端到端信息传输的确定时延上界及统计时延分布,对时延上确界进行了客观评价,为构建满足时延要求的通信系统提供了更好的理论工具。建立了智能变电站通信网络过程层单间隔系统的模糊可靠性数学模型,引入T-S模糊门构造通信系统的T-S模糊故障树,通过基本事件的模糊失效率推导出系统失效的模糊可能性。该方法可用于解决零部件失效概率和系统失效机理的不确定性问题,丰富了传统的智能变电站通信系统可靠性评估方法。提出了基于树增强型朴素可能性网络分类器的智能变电站通信网络运行状态评估方法。该方法将影响网络运行状态各指标的劣化度作为分类器的属性集,将通信系统的四种运行状态作为类型集,结合指标权重和岭形函数刻画单一实例对四种状态的模糊隶属度。实验证明该方法有较好的分类精度,为解决可靠性评估方法中零部件失效概率和系统失效机理的不确定性问题提供了崭新思路。研究智能变电站通信系统的应用技术特点,确立通信系统的统一体系结构,分析其信息应用模式、通信网络的实时性、可靠性,采取合理的技术方案适应智能变电站试点工程的应用和发展,对于构建智能电网具有非常重要的意义。

【Abstract】 The smart substation (SS) has provided standard and reliable supports for smart grid. As theneural system of the SS, the communication network is the links that bonds all kinds of IEDstogether, and its real-time performance and reliability determine the availability of the smartsubstation. Modeling and analyzing the performances of the network system by differentintelligent models can effectively improve the reliability of the whole system. The followingachievements have been obtained through theory analysis and simulation experiments.This paper has constructed the mathematical models for inner data flow in SS to analyze thecomposition of the transmission delay non-determinism of messages, and made simulationexperiments by using OPNET. The paper also discussed the influence of VLAN to the real-timeperformance of the communication network of SS, and the results have showed that theapplication of VLAN in network of SS can decrease the time delay of messages, control thespread of the broadcast storm, and improve the security of the system.This paper has researched the delay-non-determinism caused by queuing in the buffers ofSMU on emergency based on multi-server queuing theory, and deduced the analytic relationsbetween the buffer size, loss rate of package, and delay, by which, we can compute the buffersize of the SMU in station-level under the requirements of message loss rate and time delay. Thesimulization expreriment has showed that the theoretical method is correct and effective.This paper has proposed the integrative information transmission method based on the fractalleaky bucket and classification of service mechanism, which uses fractal leaky bucket to controlthe data flows those have Self-Similar characteristics, and uses the network calculus theory tocalculate the upperbound of transmission delay and statistical delay distribution. The method cannot only determine wheacher the End-To-End transmission delay is within the time limits, butalso objectively assess the pessimism degrees of the upperbound by statistical delay distribution,and all of this can provide better theoretical analysis tool for the construction of communicationnetwork under the requirements of time limits.This paper has constructed the fuzzy reliability model for the single bay-layer in process-level and introduced T-S fuzzy gate to construct the fuzzy fault tree of communication system of SS. By TS-FTA, we can deduce the fuzzy possibility of system failure from the fuzzy failure rateof basic events. This can not only solve the non-determinism of failure probability of eachcomponent and failure mechanism of the whole system, but also enrich the reliability evaluationmethods on communication system of SS.This paper has put forward the running state evaluation method of communication networksof SS based on tree augmented na ve possibilistic network classifier. By using the deteriorativedegree of each index which influences the running state of network as the attributes set, fourrunning states as the classes set, the proposed method used the index weight and ridge-shapeddistribution to describe the fuzzy memberships of a real instance to a class. Experiments hasshown that the method has better classification accuracy and can effectively solve the non-determinism in the state evaluation of network of SS.Researching the characheristics of application and technology of communication system ofSS, establishing the unite architecture of the communication system, analyzing it’s applicationmechanism of information and real-time performance and reliability, taking reasonable technicalmeasures to adapt to the application and development of the pilot project of SS, are veryimportant for the construction of smart grid.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络