The Fault Diagnosis Method Of Power Transmission System Using Weight Fuzzy Petri Net

With the accelerating of electrification progress and the development of the market reform of electric power industry, people more and more depend on the electric energy. Transmission network is a important part of power system , it undertakes the important task of large power and energy transfer. So the safety of the transmission network is very important for the whole electric network. When the transmission network faults occurred, especially when faults are complicated or unascertained information exists, locating faults and diagnosing fault element fast seem more important.Petri Network is a graphics modeling tool and it has perfect mathematic theory basis, so it is very suitable for the modeling and analysis of discrete events and dynamic system. In recent years, researchers have been trying to introduce the Petri Network and Fuzzy Petri Network into the fault diagnosis of power system, but each proposed method has its own drawbacks. For example, it cannot adapt to the changes in the network, it has difficulties in modeling and maintenance, it has bad fault tolerance and it would fail if the fault information is not enough.To settle this problem, this paper introduces weight fuzzy petri nets into power transmission system to improve the generality and fault tolerance. Firstly, this paper made a modification on the reasoning matrix method for improvement of the diagnosis speed. Secondly, this paper builds up modeles for line, bus and transformer using WFPN method. Probability and Statistic method were introduced to calculate the initial marking. Than this initial marking can be used to show the difference of the power network and the uncertainties of protection and circuit breakers actions.Thirdly,use a representative power transmission network as an example, to simulation different kinds of fault, such as complex, multiple fault and the condition of key information loss. WFPN shows its advantages and practicability in the simulation result.Finally, considering the shortcomings of strong subjectivity and weak explaining ability of each input weight from experts experience, learning ability of ANNS is introduced into Weight Fuzzy Petri Net in this thesis tentatively and parameters are trained by using stylebook, so regulate weight adaptive. Examples are taken to explain the thought.