Study On Failure Evaluation Method Of IGBT Module Based On Temperature Gradient And Statistic Characteristic

With the development of renewable energy on a global scale, reliability of power modules in energy conversion system has more and more prominent influence on the safe and reliable operation of power grid. Further research on the reliability testing and relevant evaluation method of power module is required, which is important to the safety assessment of renewable electrical equipment and system. Temperature characteristic is the key index and content of product design and reliability evaluation, and the temperature distribution of IGBT module under different operation status is often used to determine its actual condition.According to the definition of vector analysis, the gradient of a scalar field can reflect itself variation rule more effectively. Thus, the temperature gradient and its statistical characteristic is proposed to present the thermal behavior of IGBT module. Based on the built electrical-thermal-mechanical coupling model, the variation of temperature and its gradient characteristic of IGBT module under different operation conditions are studied, then a method based on temperature gradient and its statistical characteristic are utilized to indicate the internal failure information; finally the corresponding healthy evaluation model of IGBT module is built. In the paper, the simulation model of typical IGBT modules and the corresponding temperature measuring experimental platform are built. The validation of the proposed method is verified through the three dimensional finite element method(FEM) and experiment results. The main creativity work of the paper is as following:(1) An electrical-thermal-mechanical coupling model is built, in which the temperature dependence of material and its mechanical viscoplasticity are taken into consideration. The main attention is focused on the calculation of multiple coupling field model and the equivalent method of heat sink. In addition, temperature measuring experiment is done to verify the accuracy of the built model. Meanwhile, the temperature distribution of IGBT module for different failure cases is analyzed, and temperature concentration and stress concentration caused by solder defect are analyzed combined with the power cycling experiment results, as well as the damage accumulation process of solder layer.(2) According to the continuous and statistical characteristic of temperature distribution of IGBT module, a method based on temperature gradient and its statistical characteristic is proposed to show the internal thermal behavior. Based on the built FEM model, the temperature gradient and the statistical characteristic of temperature image of IGBT module for different operation conditions and different failures are studied. Results indicate that the temperature gradient and its statistical characteristic can not only reflect the change of thermal behavior of IGBT module, but also be used to detect the internal defects.(3) The temperature measuring and power cycling accelerated aging platform are built at first, then the temperature gradient and temperature probabilistic density distribution of IGBT module under different operation conditions and aging conditions are carried out, to verify the accuracy of the proposed method by laboratory experiments. Experimental results show that the proposed method has higher sensitivity and can also present the overall temperature distribution difference by comparing with the single temperature analysis method.Besides, to verify the validation and the extension of the proposed method further, the temperature gradient and temperature probabilistic density distribution of three-phase full bridge IGBT module under different control schemes are also estimated.(4) Based on the experiment platform, the influence of electrical operating point on temperature gradient and temperature probabilistic density distribution is discussed. By studying the temperature characteristic of IGBT module under different operation parameters, the relationship between temperature characteristic parameters and conduction current, switching frequency and output frequency is obtained.(5) A healthy condition evaluation model of IGBT module based on multiple characteristic parameters is built up, in which the BP neural network is utilized to improve the accuracy of assessment model. Results indicate that the total recognition rate of the built model is more than 93% and it can help to provide reference to product detection. In addition, to consider the effect of aging process on thermal parameters, the failure rate function of IGBTs in double-feed wind turbine generator converter is also preliminarily discussed, based on which the multi-time sale lifetime consumption of IGBT module is also calculated.