基于粒子群优化极限学习机的断路器故障诊断方法研究

Research on Fault Diagnosis of Circuit Breaker Based on Particle Swarm Optimization Extreme Learning Machine

断路器动作时,分合闸线圈电流特征能够反映断路器操动机构或者二次回路的运行状态,文中基于分合闸线圈电流特征值创建了样本库,并提出使用粒子群优化学习机进行故障诊断的方法。该方法首先通过粒子群优化算法寻找最优解,即极限学习机模型中输入层与隐含层间的权值以及隐含层的偏置,然后利用最优值进行极限学习机网络训练,最后使用训练好的网络对测试样本进行诊断并验证该方法的有效性。同时搭建了未优化的极限学习机模型和遗传算法优化的极限学习机模型,仿真结果表明,经过粒子群算法优化后的极限学习机能100%识别出样本中不同的断路器故障状态,相比另外两种算法具有更好的稳定性和更高的精确度。

When the circuit breaker is operated,the opening and closing coil current can reflect the operating state of the circuit breaker operating mechanism or the secondary circuit.In this paper,a sample library is created based on the current characteristic values of the opening and closing coils,and a particle swarm optimization learning ma⁃chine is proposed for fault diagnosis.Firstly,the method searches for the optimal solution through the particle swarm optimization algorithm,that is,the weight between the input layer and the hidden layer in the extreme learning ma⁃chine model as well as the bias of the hidden layer,then uses the optimal value to train the extreme learning machine network.Finally,the trained network is used to diagnose the test samples and verify the effectiveness of the method.At the same time,the unoptimized extreme learning machine model and the genetic algorithm optimized extreme learning machine model are built.The simulation results show that the extreme learning function optimized by the particle swarm optimization algorithm can identify 100%of the different circuit breaker fault states in the sample,which has better stability and higher accuracy than the other two algorithms.