Research On Modeling Of Data Flows In Power Systems And Applications

The operation efficiency and management level of smart grid are promoted with the development of information and communication technologies(ICTs), which also provides guarantees for realizing the important application of wide area situational awareness, substation and distribution automation, demand response of consumer side, etc. However, with the wide application of ICTs in power system, the real-time, reliability, and security of electrical information systems has also caused greater impacts on the normal operation of power system. Thus, research on the comprehensive performance of electrical informat ion systems has great theoretical and practical significance on promoting the security and stability of power grid.This thesis studies the modeling theory of power systems and performance evaluation methods of electrical information systems, and the relevant countermeasures for improving the system features are proposed too. Meanwhile, the application of ICTs in load active response as well as the impacts of communication delay on load response effect are discussed. The innovative achievement s in this thesis are summarized as follows:(1) The mathematical models of typical data flows in power systems are established.Based on the theory of stochastic process, the typical data flows in power systems are classified into three types according to their generati on characteristics, which are cyclic data flow, stochastic data flow, and burst data flow. The mathematical models of the three data flows are established respectively. Particularly, cyclic data flow can be described by the parameters of packet generation rate, packet size, and transmission delay, while stochastic data flow is modeled by Poisson distribution, and its transmission characteristics are mainly affected by packet size and packet arriving rate. Moreover, burst data flow has the feature of self-similarity and can be simulated by ON/OFF model. The generation of ON and OFF state obeys Pareto and Poisson distribution respectively. Further, the delay characteristics for the proposed three data flows are simulated and analyzed by using OPNET Modeler, an d the impacts of model parameters on network real-time performances are evaluated. The studies in this chapter are considered to provide theoretical support for evaluating the real-time performance of electrical information systems.(2) The real-time performances for VLAN-based substation communication systems are studied deeply.Based on the proposed three data flow models in chapter 2, the data flows in substations are analyzed and classified first. The typical cyclic data, stochastic data, and burst data in substation communication systems are summarized, and the influences of VLAN schemes on the transmission of sampled value(SAV) packets are investigated qualitatively. Furthermore, a comprehensive study on the real-time performance of substation communication systems is carried out by OPNET Modeler. Multiple simulation scenarios are considered to investigate the impacts of VLAN, system faults, star and ring topologies, as well as ring breaks on substation communication systems. The relevant measures for improving the comprehensive properties of substation communication systems have also been proposed. The results in this chapter are helpful for engineering personnel with the design and maintenance of substation communication systems.(3) A method for compensating the delayed and lost SAV packets in substation system is proposed.Further, based on the theory of Lagrange interpolation, a self-adaptive SAV estimation algorithm has been proposed to resolve the unreliable transmission of SAV packets. First, the realization process of the proposed algorithm is introd uced. The determination of synchronous interpolation time and processing method of discontinuous packet loss scenario are investigated particularly. Then, the original SAV packets are obtained by the simulation waveform of single-phase grounding fault in PSCAD/EMTDC. The accuracy of linear, quadratic, and cubic estimation algorithm are evaluated and analyzed based on the obtained SAV data. Finally, the error causes of the algorithm are discussed. The hidden danger caused by the delayed and lost SAV packets can be eliminated by embedding the proposed algorithm into conventional relay protection algorithm.(4) A robust algorithm for substation intrusion detection is presented.Based on the idea of multi-agent, a distributed intrusion detection model as well as related detection algorithm have been proposed to discover the intrusion and attack behaviors in substations. Three types of agents are included in the proposed model, which are intrusion awareness agent, communication service agent, and state monitoring agent. By mutual cooperation of these agents, the abnormal packets in substation communication systems can be detected, and the attack behavior can be discovered then. Moreover, a synthetical index for power system vulnerability evaluation is presented by considering the possibility of intrusion behavior in substation communication systems and influence on power grid with the attack of substation. The proposed algorithm and index are verified through I EEE 30 node systems finally. The studies in this chapter are helpful to promote the security defense ability of electrical information systems.(5) The impacts of communication systems delay on load active response are investigated.To realize the friendly interaction between electric load and power grid, the active response policies for grid friendly air conditioning loads as well as the implementation scheme of communication syst ems are studied. Three types of active response policies for air conditioning loads are presented, which are centralized response policy, distributed response policy, and hierarchical distributed response policy. According to the characteristics of different policies, the implementation schemes of active response policy are discussed based on advanced ICTs. Finally, the effects of proposed policies on improving voltage sags are evaluated by setting up a typical 15 nodes distribution network. The impacts of communication systems delay on the load response effects are investigated esp ecially, which provides useful references for the actual application of load active response.