基于关联规则分析的电力变压器故障马尔科夫预测模型

Markov Forecasting Model of Power Transformer Fault Based on Association Rules Analysis

为了有效预测电力变压器故障发展的情况,提出了一种基于关联规则分析的电力变压器故障马尔科夫预测模型。运用云理论提取状态参量的云概念,使用Apriori算法挖掘状态参量与状态之间的关联规则以及各状态之间的关联规则;根据状态参量与状态之间的关联规则,建立基于云–Petri网的变压器状态分析模型,从而得到变压器在初始时各状态发生的可能性;利用各状态之间关联规则构建变压器状态转移矩阵,并建立修正因子体系对状态转移矩阵进行修正;将变压器初始时各状态的可能性结合修正后的状态转移矩阵对故障进行预测。实例计算表明,相比于IEC、BPNN与SVM,基于云–Petri网的分析模型具有更快的响应时间或更高的准确率,而对状态转移矩阵的修正可提高马尔科夫模型预测的准确率,能对变压器故障发展趋势进行更有效、合理的预测。

In order to forecast the development of power transformer fault effectively,we propose a Markov model to forecast the fault of power transformer which is based on association rules analysis.Firstly,the cloud concept of transformer state parameter is extracted by the cloud theory,and the Apriori algorithm is applied to mining the association rules between the transformer state parameter and the states of power transformer and the association rules between different states.Secondly,according to the association rules between the state parameter and the states,an analysis model of transformer state based on a cloud-petri network is established,which can get the possibility of each initial state.Thirdly,the state transition matrix of transformer states based on the association rules between different states is built.And a correction factor system is established,which can correct the state transition matrix.Finally,this Markov model uses the possibilities of the initial states of power transformer to forecast the fault with the corrected state transition matrix.The test result reveals that,compared with IEC,BPNN,SVM,the analysis model based on could-petri network has faster response time or higher accuracy.And the matrix correcting can improve the accuracy of Markov model prediction,which can forecast the power transformer fault trend more effectively and reasonably.