Font Size: a A A

Research Of Method Of Fault Diagnosis Of Power Network

Posted on:2011-01-16Degree:DoctorType:Dissertation
Country:ChinaCandidate:S H LiuFull Text:PDF
GTID:1102330332979993Subject:Power system and its automation
Abstract/Summary:PDF Full Text Request
With the development of national economy, high quality, reliable, stable power resource supply is needed widely. The power system faults are inevitable because of the natural or artificial reasons, causing interruption of power supply or the quality descend, even cause the equipments to be destroyed. On the other hand, the scale and complexity of power system are increasing, the level of automation is improving together with more and more automatic devices put into application. When the fault occurs,the automatic devices of the power system would produce a mass of alarm information and bring much difficulty to the operators for fault diagnosis, especially when the circuit breakers and relay protections have unwanted operation or fail to operate,the fault diagnosis becomes complex extraordinary. The paper analyzes the application of AI in fault diagnosis on power system and researches on the application of Petri net theory, Optimization Algorithms, mult-Agent system.The methods of fault diagnosis of power network using Petri net theory are most based on simple Petri net, whose simulation capability is not be satisfied. The models are based on the division of protection into main protection, back-up protection and second back-up protection commonly, which can not describe the complex protection configuration and the protection coordination of different levels in modern complex power system exactly. It faced the trouble of High-Dimensional data and the change of the power network topology and the configuration information of protection. In view of this method, the diagnosis idea thinking of Space-time two-dimensional parameter is proposed. The concept of connection-relation diagram of fault elements is proposed. It can give much help when the diagnosis model is constructed for the complex power network:search the connection relation of elements which was within the scope of the tripped relay protection and form the connection-relation diagram, then construct the diagnosis model based on the network configurations and priority level of different relay protection. The construction of diagnosis model based on predicate Petri net is introduced in detail. It has the advantages:As the substitute of simple Petri net, predicate Petri net can contain much more information such as information of circuit breakers and relay protections, so it enhance the simulation capability, the diagnosis process and result are more intuitionistic; we construct the diagnosis model for every direction dynamically when the information of power network topology and diagnosis data has changed, which reduce the scale and dimension of operation. The method can reduce processing time because the interrelated matrixes are sparse matrixes; The method is based on the actual relay protection type (pilot protection, distance protection, et al.), which provided the clear scopes of protections.When we diagnose the power system fault, the factor of time is seldom taken into account, causing the diagnosis model to be rough and multi-result.The SOE(Sequence Of Event) based on GPS(Global Positioning System) accurate time setting have the Temporal Information of Alarm Messages.If we make good use of the time character and the cause-effect relations of circuit breakers and relay protections,the accuracy and efficiency will be improved. In this paper, we enrich the concept of relevancy route and construct an improved optimization model for power network fault diagnosis with the temporal information of alarm messages taken into account.We use the time-t represent the state of the element, the state of the protection and breakers are the functions of the t.The state of elements, relays and circuit breakers change with t, the temporal relationship of the actions of protection relays and circuit breakers in modern power systems with multiple kinds of protection configurations can be rationally represented by the model developed.The increasing scale of power system and the complexity of equipments bring great trouble for the accurate and quick fault diagnosis. The multi-Agent system technology can solve the trouble brought by large scale and complex system taking place of traditional theory. It can divide the lager scale and complex system into small sub-system which are easy of management and maintenance. MAS solve the large scale and complex process problem through the cooperation of Agents. In this paper, an approach of power system fault diagnosis based on Multi-Agent System is presented. The fault diagnosis model is constructed according to the characteristic of power network management, operation, and framework. Agents adopt their own fault diagnosis method according to their different position. These intelligent bodies can achieve the common goals to make full use of the spatial distributed and qualitative layered fault information. The global framework of the model is built, and the believe-knowledge warehouse, capacity for action, schedule flow of agent are described in detail. The structure of core part-ADA (Area Diagnosis Agent) is built. The function and the process of every modules of ADA, the style and content of information interaction between agents are described in detail. The process of fault diagnosis based on MAS is given.JADE (Java Agent Development Framework) is a platform for Agent development which is comply with FIPA criterion. JADE provides a series of API which help designers to develop agent programme.In this paper, the JADE is introduced, including the creation of agents, the implementation process of agents, the communication mechanism of agents which is compliant to FIPA criterion. We researched the realization of communication between agents, and main agents in the fault diagnosis system based on JADE using Java language.
Keywords/Search Tags:power network, fault diagnosis, predicate petri net, power network topology, alarm messages, time character, large scale power system, Agent, multi-Agent system, cooperation, believe-knowledge warehouse, capacity for action, schedule flow
PDF Full Text Request
Related items