CAJ | 학술논문

为了利用相对较少的故障数据样本对断路器主要故障类型进行较为准确的在线判断，提出了一种基于融合粒子群的模糊核聚类(particle swarm fused kernel fuzzy C-means， P-KFCM)与支持向量机(support vector machine，SVM)的故障诊断方法。通过对断路器分合闸电流信号的分析，找出与断路器主要故障类型相对应的特征量；据此对采样信号进行处理，建立故障特征样本空间；利用 P-KFCM 算法对故障训练样本进行预分类，并以此为基础建立多SVM故障预测模型。P-KFCM算法将粒子群(particle swarm optimization，PSO)的全局搜索能力融入KFCM中，有效的解决了局部最优问题，在一定程度上提升了诊断结果的可靠性。实验结果表明，该方法在诊断断路器主要机械故障方面能够取得较好的效果。
위료이용상대교소적고장수거양본대단로기주요고장류형진행교위준학적재선판단，제출료일충기우융합입자군적모호핵취류(particle swarm fused kernel fuzzy C-means， P-KFCM)여지지향량궤(support vector machine，SVM)적고장진단방법。통과대단로기분합갑전류신호적분석，조출여단로기주요고장류형상대응적특정량；거차대채양신호진행처리，건립고장특정양본공간；이용 P-KFCM 산법대고장훈련양본진행예분류，병이차위기출건립다SVM고장예측모형。P-KFCM산법장입자군(particle swarm optimization，PSO)적전국수색능력융입KFCM중，유효적해결료국부최우문제，재일정정도상제승료진단결과적가고성。실험결과표명，해방법재진단단로기주요궤계고장방면능구취득교호적효과。
To make accurate judgments of circuit breakers’ main fault types in on-line system using relatively small amounts of fault data, a fault diagnostic method was proposed in this paper. This method combined particle swarm fused kernel fuzzy C-means (P-KFCM) and support vector machine (SVM). Through analysis of the opening and closing current signals, characteristic values corresponding to main fault types could be found, based on which we could process sample signals and establish feature space of fault samples. P-KFCM was utilized to pre-classify fault training samples, on the basis of which multi-SVM fault prediction model could be established. At the same time, it integrated global search ability of particle swarm optimization (PSO) into KFCM to solve local optimum, which would effectively improve reliability of diagnostic results. Experiment results have proved that the proposed method achieves perfect results in diagnosing circuit breakers’ main mechanical faults.