【作者】 徐盛友；

【导师】 陈民铀；

【作者基本信息】 重庆大学 ， 电气工程， 2013， 博士

【摘要】 变流器作为新能源转换系统中极其重要的组成部分，其运行安全性和可靠性越来越受到科研人员、电力系统运行和管理人员的关注。随着我国新能源发电系统装机容量的不断增长，对功率变流器的容量、效率和可靠性的要求也越来越高，尤其当变流器中IGBT功率模块出现隐形故障的时候，维护、检修等运行管理的难度也越来越大，所以研究变流器IGBT功率模块的可靠性和寻求一种状态监测与评估的方法对于新能源发电及其转换系统的安全可靠运行显得非常重要。由于电力电子器件大多封装在模块中，模块内部故障不易直接测量，状态难于监测与评估，可靠性难于保证。论文从这几个关键问题入手，分析了功率模块IGBT的失效原因及可靠性对策、研究了功率模块IGBT的运行驱动可靠性、IGBT的故障诊断与运行状态监测与评估。还研究了功率模块IGBT热模型、动态热阻和结温的提取、寿命预测等方面，以实现对功率模块的状态监测与有效评估，提高其工作可靠性。概括起来，论文主要研究了以下几个方面的内容和创新点：①功率模块IGBT的失效机理研究。分析了由IGBT可靠性降低引发的失效机理及失效模式并提出了失效评判标准，根据失效模式，把功率模块失效方式分为两大类，即与封装有关的失效以及与芯片相关的失效,总结出变流器IGBT失效的根本原因，提出其失效是内部疲劳逐渐积累并与外部运行环境等多种因素相互作用的结果,当其承受的电应力、热应力、化学应力、辐射应力和机械应力以及其他因素使所经受的应力条件超过最大额定值时，便会造成器件失效。②功率模块IGBT的驱动控制可靠性研究。对于目前常用的推挽结构的功率模块驱动电路,虽然减小驱动电阻将导致开通、关断时间的缩短,但同时也将导致开关过程中IGBT功率器件电热应力的增加，从而使功率模块可靠性降低。针对这一特点，首次提出了一种基于dSPACE的分段控制算法，通过软件和硬件结合分段控制功率模块开、关断速度,揭示了功率模块的表面温度变化规律，拟合出了表面温度变化的动态曲线，实验证明该控制方法具有良好的驱动和保护功能,在缩短开关时间的同时可以有效抑制电压、电流应力的增加，并能有效地减少新能源转换系统中输出电压和电流的谐波、提高电压利用率和波形质量，使新能源转换系统可靠性得到提高。③功率模块IGBT的状态监测与评估方法研究。针对风力发电等可再生能源转换系统中大功率变流器等电力电子设备的故障率较高、缺乏有效的状态监测问题，本文致力于研究新能源系统中功率模块的故障诊断与状态监测和评估方法研究，通过分析目前常用的状态监测方法的困难和弱点，首次提出了IGBT的外特性监测方法，建立了IGBT外特性瞬态模型与稳态模型及其评估标准，并首次提出了基于神经网络的黑箱算法，从外特性角度建立了基于功率模块多个运行状态参数(如工作电压、电流、调制比、表面温度等)的状态监测与评估系统，并运用智能计算方法和人工神经网络建立了IGBT表面温度预测模型，实验证明该方法方便可行，为研究和提高可再生能源发电系统中变流器功率模块的可靠性开辟了新的途径。④功率模块IGBT的热模型和寿命预测研究。本文分析和研究了功率模块IGBT的功率计算方法以及热模型及求解算法、热阻提取，寿命预测。IGBT的寿命和IGBT结温的变化有关，而结温源于它的发热，发热又源于功率损耗且和其热阻有关。因此，IGBT结温及热阻是影响其寿命、评估其可靠性的重要参数。本文提出了一种IGBT动态热阻的提取方法，给出了实验步骤，拟合出了热阻变化的动态曲线和热阻公式，该方法克服了IGBT内部参数难于测量和不易操作的特点，通过检测外部集射电压参数间接地获取IGBT结温，建立了外部参数和内部参数之间的联系，在动态曲线和结温公式基础上，提出了一种新的IGBT寿命预测理论模型，并对该理论模型进行了归一化，使该模型简单适用且适用范围得到拓宽，为提高IGBT的可靠性具有重要的意义。更多还原

【Abstract】 The convertor plays the extremely important role in the new energy conversionsystem, its safety and reliability in operation is more and more focused on by theelectrical power system operators, the managements and scientific researchers. With theincrease of our new generators total volume and the transmission capacity, therequirement of reliability and volume, efficiency is becoming more and more important,so it becomes more difficult to manage the convertors, especially when the IGBT powermodule presents the potential failure, therefore it seems important to seek one kind ofIGBT power module condition online monitor and the failure diagnosis methodregarding the IGBT power module safe reliability serviceFirstly, the failure mechanism is analysized and the protection action is presentedby means of introducing the common failure problems and normal condition duringconvertor operation. Then, the driving circuit and thermal model and lifetime predictionis proposed. After that, a new method, which is based on the algorithm of black box forthe IGBT power module, is proposed to monitor and forecast IGBT module’s condition.The following researching work has been finished in this dissertation①Mechanisms and failure mode resulted from the reliability decreasing of IGBTis analyzed. According to the failure Mechanism, the failure, which comprised of thefailure patterns of mosfet device and component hybrid pattern failure, is classified intotwo types whose failure mode is related with package and chips of IGBT respectively.The failure of IGBT power module is caused by the interacting multi-factors. Such as itsinternal fatigue accumulation, external operating environment etc.②In the present push-pull driving circuit for power module, reducing the drivingresistance will lead to make the time shorter, but this process will result in increase ofcurrent and voltage thermal stress. After the failure mechanism and thermal model ofpower module is analyzed, a novel driving methodology based on dSPACE has beenintroduced and validated in this paper. The ascending theory of case-temperature isinvestigated and then an exponential curve is achieved for power module, experimentsprove that this methodology has a good driving and the protective performance，andeffectively reduce harmonics of output current and voltage in the new energyconversion system, enhance the voltage utilization rate and waveforms quality, thus thereliability of power module has been greatly improved. ③Packaging-related internal failure such as solder fatigue or wire crack has beenacknowledged as one of the principal root causes of IGBT power module failures.Condition monitoring is needed by power electronic system operator as a cost-effectivemeans of improving IGBT reliability. In this paper, an approach is presented to monitorinternal failure inside a IGBT power module by identifying the increase of external casetemperature due to internal failure, In this method, an IGBT power module is regardedas a black box system without considering any internal factors or models associatedwith the heat sink as well as the power loss module, all electrical factors at certainworking points are classified as the inputs, meanwhile the case temperature isconsidered as the output of the black box system respectively, in which any devicepower loss model or heat sink model is not established to estimate the internal thermalresistance. Firstly, case temperature is predicted based on the GA-BP artificial neuralnetwork, after that, a monitoring method are presented based on the predicted look-uptable or model and a threshold for the difference between the measured and thepredicted case temperature is set reasonably. It is because that the total loss in themodule increases as junction temperature rises generated from internal failure, causingan increase in case temperature rise, the external characteristic of IGBT is the reflectionof internal condition, once the measured case temperature is higher than the predictedtemperature at a certain working point, taking account of the masking effect ofmeasuring accuracy and errors, then a conclusion can be get that the IGBT condition isabnormal. At last, experiment has validated and demonstrated the concept andcharacterized the proposed method.④The lifetime of IGBT power module is related with the change amplitude ofjunction temperature, the junction temperature is from the dissipating heat of IGBT, theheat comes from the power loss of IGBT, the thermal model, and power losscalculations are investigated. Furthermore, the prediction model of lifetime for IGBT isproposed.更多还原