论文部分内容阅读
分析了12脉波可控整流电路的294种故障模式,根据整流电压畸变波形提出一种特殊的故障分类方法。通过对故障电压波形的逻辑预处理得出12维故障向量及相应的故障编码。为改善粒子群算法(particle swarm optimization,PSO)的性能提出一种改进的带扰动项粒子群算法。引入进化速度因子,当粒子进化速度低于一定值时在粒子速度更新方程中添加一扰动项,算法的搜索效率和全局优化性能显著提高。将改进算法用于神经网络故障诊断建模,实验结果表明该系统具有诊断速度快、精度高的特点。适用于复杂电力电子设备或系统的故障诊断场合。
A total of 294 fault modes of 12-pulse controllable rectification circuit are analyzed. A special fault classification method is proposed according to the rectified voltage distortion waveform. The 12-dimensional fault vector and the corresponding fault code are obtained by logic preprocessing of the fault voltage waveform. To improve the performance of particle swarm optimization (PSO), an improved particle swarm optimization algorithm with disturbances is proposed. When the particle evolution rate is lower than a certain value, a perturbation term is added to the particle velocity update equation. The search efficiency and global optimization performance of the algorithm are significantly improved. The improved algorithm is applied to the neural network fault diagnosis modeling. The experimental results show that the system has the characteristics of fast diagnosis and high precision. Suitable for complex power electronic equipment or system fault diagnosis occasions.