单相脉冲整流器故障在线智能诊断实验设计

Design of Online Intelligent Fault Diagnosis Experiment for Single-phase Pulse Rectifiers

针对高速列车电力牵引系统中单相脉冲整流器的功率器件开路故障和传感器故障,设计了基于数据驱动的故障在线智能诊断实验。首先,对单相脉冲整流器系统进行数学建模,分析功率器件开路故障的故障拓扑和传感器故障的故障机理;然后,基于极限学习机算法和非线性自回归结构设计智能诊断模型,并将故障在线智能诊断实验分为离线训练和在线验证2个阶段;最后,基于RT Box半实物仿真器搭建快速原型控制实验平台,对故障模型和诊断模型进行实验验证。实验结果表明,构建的单相脉冲整流器故障在线智能诊断模型可以在1/4个基波周期内检测到故障的发生,并在3/4个基波周期内识别出具体故障类型。

A data-driven intelligent diagnostic method was proposed for power device open-circuit faults and sensor faults in the single-phase pulse rectifier of the high-speed train power traction system.Firstly,a mathematical model of the single-phase pulse rectifier system was performed,and was used to analyse the topology of power device open-circuit faults and the mechanism of sensor faults.Secondly,an intelligent diagnostic model was designed based on the extreme learning machine algorithm and nonlinear autoregressive structure.Besides,the experiment for fault online intelligent diagnosis was divided into offline training and online verification phases.Finally,a rapid prototype control experiment platform was built based on the RT BOX semi-physical simulator to experimentally validate the fault models and diagnostic models.Experimental results indicate that the constructed intelligent diagnostic model for single-phase pulse rectifier faults can detect faults within 1/4 of the fundamental wave period and identify the specific fault type within 3/4 of the fundamental wave period.