节点文献

三相电机驱动系统中逆变器障诊断与容错控制策略研究

Diagnosis and Fault-Tolerant Control Strategies of Inverters in Three-Phase Motor Drive Systems

【作者】 安群涛

【导师】 孙力; 孙立志;

【作者基本信息】 哈尔滨工业大学 , 电机与电器, 2011, 博士

【摘要】 在现代航空航天领域,电机系统作为执行部件其可靠性已经成为保障航空航天器安全运行的一项关键技术。为此,近年来高容错电机驱动系统的概念被提出并得到了研究者的关注。高容错电机驱动系统通过对故障进行实时诊断和分析,在发生故障后主动重构系统的软硬件结构,从而确保整个系统在不损失性能指标或部分性能指标有所降低的情况下安全运行。研究表明,在电机驱动系统中,功率变换环节尤其是逆变器中的电力电子器件及其驱动电路是容易发生故障的薄弱环节。通过在线实时诊断逆变器功率器件的故障,采取及时和适当的容错控制策略,是提高电机驱动系统可靠性的有效途径。因此,对电机驱动系统中逆变器的故障诊断和容错控制策略开展研究具有重要的理论意义和工程应用价值。故障诊断是实施容错控制策略的前提,诊断的可靠性和实时性直接关系到容错的效果。尽管采用解析模型是实现快速故障诊断的有效方法,但传统电机驱动系统的建模方法无法将电机和逆变器统一在一个解析模型下,因此传统模型不能准确揭示系统的状态特性,不适合于逆变器内部故障的分析和诊断。电机驱动系统包含连续状态变量和开关离散事件,是一个典型的混杂系统。因此,本文将混杂系统理论引入到其分析过程中,把逆变器的离散电压模型嵌入到电机的连续状态方程中,建立了电机驱动系统的混合逻辑动态模型。在此基础上对系统的故障特征与行为进行描述和分析,为有效诊断方法的提出提供理论基础。在建立电机驱动系统混合逻辑动态模型的基础上,构建了系统的状态残差方程,通过在两相静止坐标系中的残差评价来诊断和定位逆变器功率管的故障。混杂系统状态残差法可以克服系统闭环控制算法对诊断效果的影响,并将诊断时间缩小到四分之一个基波周期以内。在诊断实施过程中,通过设置残差阈值来抑制参数误差、测量误差、扰动和噪声等影响。为进一步缩短诊断时间、消除负载对诊断算法的影响,考虑逆变器的混合逻辑动态模型,分别提出了基于误差电压和桥臂中点电位监测的逆变器故障诊断方法。误差电压诊断法通过桥臂中点对地电压实测值与估计值之间的误差来诊断故障。该方法对误差电压的幅值及时间宽度进行双重判断,消除了功率管开关延时和死区的影响,提高了诊断方法的可靠性和鲁棒性,并且诊断时间在几个开关周期以内。桥臂中点电位监测法采用光耦监测桥臂中点电位,将其转化为逻辑信号,并与开关信号进行逻辑运算,在一个开关周期内完成对逆变器功率管开路故障的诊断。该方法省去了电压检测与比较环节,克服了电压检测带来的测量误差、测量噪声等对诊断方法的影响。同时,采用对开关信号上升沿延时后再进行逻辑运算的办法,消除了功率管开通延时造成的误诊断,保障了诊断的可靠性。所提出的诊断方法得到了仿真和实验验证。逆变器的容错控制策略通过重构软硬件结构来确保系统在故障后具有一定的降额运行能力。本文对开关冗余逆变器和级联两电平逆变器两种容错拓扑及其控制策略开展了研究。开关冗余逆变器结构简单,单管或单桥臂故障后运行在四开关三相状态,维持了故障前的电流输出能力。对四开关三相逆变器的控制策略进行研究,提出了基于补偿电压的四开关SVPWM过调制方法,来提高四开关逆变器的母线电压利用率。然后,将四开关逆变器应用于三相永磁同步电机和无刷直流电机系统中,研究了四开关三相永磁同步电机的矢量控制策略和四开关三相无刷直流电机的磁链跟踪控制策略。为提高故障后系统的电压输出能力,满足系统对电机转速指标的要求,具有良好容错性能的级联两电平逆变器被应用于开放式绕组电机驱动系统中。单电源级联两电平的容错策略归纳为故障相短接和故障相开路两种重构拓扑,与传统三相半桥逆变器具有同等的电压输出能力。结合级联两电平逆变器容错拓扑,提出了永磁同步电机的两相矢量控制和两相准矢量控制策略,以及无刷直流电机的两相直接电流控制策略,实现了电机系统的容错运行。

【Abstract】 In modern aeronautic and astronautic fields, motor systems are used as actuators and their reliability has been a key technology for safe operation of aircraft and spacecraft systems. Therefore, highly fault-tolerant motor drive systems are recently proposed and concerned by researchers. Highly fault-tolerant motor drive systems can diagnose and analyze faults in real time, and then actively reconfigure hardwear and software structures so that the whole systems can operate safely with non-losing or partly derated performance. As research indicated, power semiconductors and their drives in power converters especially inverters are fragile parts of motor drive systems. By using on-line and real-time fault detection and tolerant control strategies, reliability of motor drive systems can be improved. Therefore, research on fault diagnostics and fault-tolerant control strategies of inverters in motor drive systems has great significance in theory and application.Fault diagnosis is the precondition of implementing fault-tolerant control strategies, and reliability as well as immediacy of fault diagnosis is directly related to effects of the fault-tolerant strategies. Analytic model-based diagnostics is a fast and effective fault diagnostic method, but conventional modeling methods of motor drive systems cannot be used for analyzing and detecting faults inside inverters. It is because these models are incapable of unifying both the motor and inverter into one model and revealing state characteristic of the systems. Motor drive systems are a typical hybrid system including continuous state variables and discrete switching events. In this thesis, hybrid system theory will be adopted to analyze motor drive systems and build their mixed logical dynamic(MLD) models. The behavior and characteristic of motor drive systems with some faults are analyzed based on the proposed model, which is the fundamental of fault diagnostic methods.Based on the presented MLD model of motor drive systems, system state residual functions are built to detect and locate inverter switch faults by means of residual evaluation in the static two-phase coordinate system. The MLD model-based inverter fault diagnostic method can get rid of effects on the diagnostic performance of close-loop control algorithms, and reduce the detection time to a quarter of fundamental-wave period. In this method, a threshold value is set in order to restrain effects of parameter errors, measurement errors, disturbances and noises. To reduce detection time further and eliminate effects on the diagnostic algorithm of loads, the MLD model of inverters is considered, and hereby error voltage based and leg neutral-point potential supervising based fault diagnostic methods are proposed, separately. Error voltage based diagnostic method uses errors between measured leg neutral-point-to-ground voltages and estimated ones to detect open-switch fault. In this method, amplitude and time width are adopted as dual criteria to evaluate error voltages in order to improve reliability and robustness by eliminating effects of swiching delay and dead time. Detection time of error voltage based method is reduced within a couple of switching periods. Photocouplers are used to supervise leg neutral-point potentials and transfer them to logical signals in leg neutral-point potential based fault diagnostic method. Then, these logical signals are used for Boolean calculation with switching signals to detect open-switch fault during one switching period. This method can get rid of effects of measurement errors and noises because it eliminates voltage sense and comparison. Meanwhile, this method uses rising edge delay of switching signals to eliminate effect of switch turn-on delay so as to obtain high reliability. These proposed methods are validated by simulations and experiments.Fault-tolerant control strategies of inverters can ensure post-fault systems have such derated operation capability by reconfiguring software and hardware structures. Switch-redundant inverters and cascaded two-level inverters as well as their post-fault configurations and control strategies are studied in this thesis. Switch-redundant inverters have simple configuration, and can operate as four-switch three-phase inverters after single switch fault or single leg fault. Four-switch three-phase inverters retain the same current output capability. Then, control strategies of four-switch three-phase inverters are studied, and compensated voltage-based over-modulation four-switch space vector PWM is proposed in order to improve DC-link voltage utility. Four-switch inverters are applied in permanent magnet synchronous motor(PMSM) and brushless DC motor(BLDCM) drive systems, and vector control scheme of four-switch three-phase PMSMs and flux linkage tracking control strategy of four-switch three-phase BLDCMs are researched, separately. In order to improve voltage output capability of post-fault systems and achieve motor speed index, cascaded two-level inverters(CTLIs) with good fault-tolerant performance are applied in open-end winding motor drive systems. Fault-tolerant strategies of single-source CTLIs have two reconfigured topologies: faulty phase short-circuit connected and open-circuit connected two-phase full-bridge inverters, which have the same voltage output capability with three-phase self-bridge inverters. They are used in PMSM and BLDCM drive systems, and two-phase vector control and quasi-vector control schemes of PMSMs as well as two-phase direct current control scheme of BLDCMs are proposed to maintain operation of motor drive systems.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络