Research On Markov Forecasting Model Of Power Transformer Fault Based On Association Rules Analysis

Power transformer,as the hub of the primary grid security defense system,plays a critical role in energy transformation and transmission.Once a transformer accident happened,it would bring huge loss would occur.If we can make timely and effective assessment of the future operation of power transformer with certain methods according to its current operating state,it would be of great theoretical and practical meaning to operation maintenance and state overhaul of the transformer,and could predict and reduce the occurrence rate of faulting.With the aim of predicting faulting of power transformers,this paper studied the transformer state analyzing method based on cloud-Petri net,constructed and corrected matrix of state transfer rate according to association rules,and thereby established transformer fault predicting model based on Markov model according to the result of transformer result analysis and the constructed matrix of state transfer rate.The results of the study include:(1)With the help of cloud theory,we could extract state parameters and the association rules of various states,expand general Petri network model according to the expression format of cloud production rules,establish cloud-Petri net state analyzing model applicable to the transformer,and thereby realize state reasoning based on cloudPetri net and state reasoning result assessment based on centroid.In a word,this model can obtain current state of the transform and effectively distinguish different states of the equipment.(2)When using Apriori algorithm to conduct data mining against the association rules existing among various states,matrix of transformer state transfer rate and a correction factor system containing such factors as family defects,operating environment,overhaul record and historical run have built up.Besides,matrix of transformer state transfer was corrected while relative impairment grade was adopted to calculate the value of indexes.(3)Transformer fault was predicted according to the transformer fault predicting model of Markov and the constructed matrix of transformer state transfer rate,with the results of transformer state analysis being the initial state vector.Practice has shown that corrected matrix of transfer rate has made full use of various state information about the transformer,and the predicting model established on this basis can predict the development trend of faults of the transform more effectively and accurately.