基于变精度粗糙集与量子贝叶斯网络的变压器故障诊断研究

RESEARCH ON TRANSFORMER FAULT DIAGNOSIS BASED ON VARIABLE PRECISION ROUGH SET AND QUANTUM BAYESIAN NETWORK

及时准确的变压器故障诊断对电力部门正常运转而言意义重大。针对粗糙集与贝叶斯网络模型在变压器故障诊断中出现受噪声数据影响大、存在完全搜索NP困难等问题,提出基于变精度粗糙集与量子贝叶斯网络的变压器故障诊断模型。通过Grover量子搜索算法快速搜索变压器故障、征兆类型等目标数据,运用层次分析法删减对诊断故障影响较小的指标,并分析确定变精度粗糙集的错误分类率β,获得最小故障决策表,从而构建贝叶斯网络故障推理模型,实现对变压器故障的诊断研究。实例分析表明,与粗糙集、量子贝叶斯网络等模型相比,该模型更适合变压器故障的诊断且诊断精确。

It is significance for electric power department to diagnose transformer fault timely and accurately. Aiming at the problem that the rough set and the Bayesian network model have great influence on the transformer fault diagnosis and the complete search NP difficulty,a transformer fault diagnosis model based on variable precision rough set and quantum Bayesian network is proposed. By using Grover quantum search algorithm,the target data such as transformer fault and symptom type can be searched quickly. The analytic hierarchy process is used to delete the index which has less influence on the diagnosis fault and to analyze the misclassification rate of the variable precision rough set. The minimum fault decision table is obtained,and the fault reasoning model of Bayesian network is constructed to diagnose the transformer fault. Example analysis shows that compared with rough sets and quantum Bayesian network model,the proposed model is more suitable for the transformer fault diagnosis and more accurate diagnosis.