论文部分内容阅读
为了提高变压器故障诊断精确度,提出量子粒子群算法(QPSO)优化相关向量机(RVM)的变压器故障诊断方法。采用4个二分类RVM来实现变压器故障诊断的多分类问题。相关向量机的组合核函数可融合变压器运行状态的多种特征信息,为非线性、有限样本数据的变压器故障诊断建模问题提供有效的方法。利用量子粒子群算法对RVM诊断模型参数快速寻优,并结合CV原理设置适应度函数可有效提高诊断模型的泛化能力。实例分析表明,该耦合算法诊断正确率为91.1%,优于三比值法、BPNN、PSO-SVM方法,可有效提高变压器故障诊断精度。
In order to improve the accuracy of transformer fault diagnosis, a fault diagnosis method based on quantum particle swarm optimization (QPSO) and optimal correlation vector machine (RVM) is proposed. Adopting four dichotomous RVM to solve the multi-classification problem of transformer fault diagnosis. The combined kernel function of the correlation vector machine can fuse many kinds of characteristic information of the transformer operating status, and provide an effective method for modeling the fault diagnosis of the transformer with nonlinear and finite sample data. Using quantum particle swarm optimization algorithm to rapidly optimize the parameters of RVM diagnosis model, and combining the CV principle to set the fitness function can effectively improve the generalization ability of the diagnostic model. The case study shows that the diagnostic accuracy of the coupling algorithm is 91.1%, which is better than the triple ratio method, BPNN and PSO-SVM methods, which can effectively improve the fault diagnosis accuracy of the transformer.