基于复数信号状态空间的船舶电力推进系统逆变模块故障诊断研究

Switch open-circuit and short-circuit faults often occurred in the inverter with marine electric propulsion system (MEPS).These faults can result in normal drive function lost,even led to breakdown of the drive system,and the catastropic event may occur especially in severe cases. Aiming at the unbalanced voltage problem resulting from switch open or short-circuit fault, the research project putforwards a new complex signal state-space based fault detection and isolation method.The project builds a complex signal state space model with multiple positve sequence(PS) and negtive sequence(NS) harmonic symmetrical components using symmetrical components theory.Considering operating condition with varying load and frequency, the design and stability condition of the complex observer are analyzed, the fault detection and isolation filter is designed, each harmonic of PS or NS are estimated and a bank of residuals is achieved. Analyzing the residual expression and influnce of maximum variation of state matrix on residual, the threshold istablished with help of experiment and the fault decision is determined. Studying the theory and algorithm of complex support vector machine (CSVM), CSVM based multiple fault isolation method is built and the faults are isoalted. The fault detection and isolation algorithm is verified by simulation and experient method when MEPS operating in switch open circuit and short circuit failure mode.The research provide a new road to detect and isoalte faults occurred in the inverter with MEPS, it has an important scientific significance and practical application value on improving reliability of MEPS and other industrial drive system.

船舶电力推进系统中的逆变模块开关器件最易发生开路、短路故障。逆变器故障会导致系统丧失正常的运行能力，甚至导致系统崩溃，严重情况下，更会导致灾难性的后果。针对故障后逆变器输出电压不平衡特点，本项目提出基于复信号状态空间的故障检测与诊断方法。利用对称分量理论建立由各阶正序、负序谐波对称分量构成的复数信号状态空间模型；考虑变载及变频等工况，研究复数状态观测器设计方法及其稳定性条件，完成故障检测与滤波器设计，获得各阶正序、负序谐波对称分量估计及残差；建立残差解析表达式，分析系数矩阵变化对残差影响并通过实验确定阈值，形成故障决策逻辑；研究复数支持向量机理论及算法，构建基于复数支持向量机的多模故障分离算法；设置开关器件开路、短路故障模式，仿真、实验验证理论算法的有效性。此项研究为电力推进系统逆变模块故障检测与诊断提供一崭新途径，对提高推进系统及其它工业驱动系统可靠性具有重要的科学意义和实际应用价值。

船舶电力推进系统中的逆变模块承担着电能变换功能，主要用于将船上直流电转换成不同频率不同电压幅值的交流电，并驱动推进船舶电机及螺旋桨。逆变模块中的功率开关器件最容易发生短路、开路故障，故障会导致系统丧失正常的运行能力，甚至导致系统崩溃，严重情况下，更会导致灾难性的后果。为提高船舶电力推进系统可靠性，研究船舶电力推进系统故障诊断就显得十分必要。对称分量分析方法是分析电动机系统不对称状态下稳态性能的最基本、重要的方法。利用对称分量理论，建立由各阶正序、负序谐波对称分量构成电动机系统的正序对称瞬时分量表达式，通过定义适当状态变量，构建电力推进系统复数信号状态空间模型。研究基于复数扩展卡尔曼滤波的谐波对称分量估计方法，实现谐波对称分量估计。将估计获得的谐波对称分量与系统无故障状态谐波对称分量进行比较，获得各阶正序、负序谐波对称分量残差。通过仿真分析及实验，确定阈值，形成故障决策逻辑；研究支持向量机理论及算法，构建基于支持向量机的多模故障分离算法；设置开关器件开路、短路故障模式，利用仿真及实验数据验证理论算法的有效性。项目研究基本完成了项目研究内容，基本达到了项目研究目标。此项研究为电力推进系统逆变模块故障检测与诊断提供一崭新途径，对提高推进系统及其它工业驱动系统可靠性具有重要的科学意义和实际应用价值。