随着电力能源系统的飞速发展，电力电子技术的应用日益广泛，已经高比例地渗透到智能电网、新能源发电、电力牵引、航空航天等国民经济关键领域。然而，为了确保系统的稳定运行，电力电子装备的可控性与可靠性亟待提升。其中，电力电子系统的建模与仿真成为关键支撑技术，这也是一个具有挑战性的瓶颈问题。由于电力电子系统是一个典型的混杂系统，包含工频（毫秒级）、开关频率（微秒级）、开关瞬态过程（纳秒级）等多个时间尺度的变化过程，传统的时间离散方法难以高效、准确完成仿真解算。离散状态事件驱动（Discrete State Event-driven，DSED）仿真框架利用电力电子系统的混杂特性，能够精准地捕捉开关事件发生的时刻，实现多时间尺度的高效仿真，开拓了电力电子系统仿真的新思路，具备解决上述瓶颈问题的潜力。然而，该仿真框架针对两类重要问题还没有解决方案：1）非线性模型建模及其解算；2）刚性问题解算。本论文以DSED仿真框架为基础，对该框架进行了非线性问题与刚性问题解算的研究，试图打破原有仿真框架的限制，从而能够使其面向更广泛的应用。在非线性问题的建模与解算方面，以电力牵引系统为典型研究对象，对多种电机模型进行建模，并提出了交互界面解耦解算方法，以使得能够将电机模型的非线性方程独立解算，并与外部电力电子部分进行无误差的数据交换。该建模解算方法打破了原有DSED仿真框架只能解算线性电路的限制，能够解决一类包含非线性的方程模型，将DSED仿真优势拓展到电力牵引、新能源等领域。所提的基于事件驱动的建模解算方法与原有方法相比，在相同精度下，具有解算效率的优越性。该方法通过实验以及与其他同类软件的对比进行了准确性与效率的验证。在刚性问题的解算方面，以后向微分公式（Backward Differential Formula，BDF）为数学基础，提出了与事件驱动机制相融合的变步长变阶数刚性算法，形成了一套后向离散状态事件驱动（Backward Discrete State Event-driven，BDSED）仿真机制。该仿真机制能够高效解算包含杂散参数的刚性系统，解决了原有DSED仿真方法计算极慢的问题，且仿真效率优于其他刚性算法，因而能够支持电磁干扰现象的相关研究，为设备可靠性的研究提供高效的分析工具。所提的刚性解算方法通过实验以及与其他同类软件的对比进行了准确性与效率的验证。

With the rapid development in energy field, the application of power electronics technology is increasingly widespread, and has penetrated into important fields, including smart grid, renewable energy, electric traction, aerospace and etc. However, in order to ensure the stability of the system, the controllability and reliability of power electronic equipment need to be improved. Therefore, modeling and simulation of power electronic system become indispensable, which is also a challenging bottleneck. Power electronic systems are typical hybrid systems, which include multi-time-scale processes, such as power grid frequency (millisecond level), switching frequency (microsecond level), switching transient process (nanosecond level), so the traditional time-discrete method is difficult to simulate efficiently and accurately. The discrete state event driven (DSED) simulation framework can accurately capture the time of switching events and realize multi-time-scale high-efficiency simulation by using hybrid characteristics of power electronics, which has the potential to solve bottleneck problems. However, there are two aspects of problems remaining unsolved under this simulation framework: 1) modeling and solving non-linear systems; 2) solving stiff systems. Based on DSED simulation framework, this paper enables the capabilitiy of solving non-linear and stiff problems, breaks the limitation of the original simulation framework, which helps DSED to be applied to a wider range of applications.As for the modeling and solving of non-linear problems, this paper takes the electric traction system as a typical research target to model a variety of motor models, and proposes a real-time interfacing decoupling solution method, which can solve the non-linear equations of the motor model independently and exchange the data with the external power electronic parts without theoretical error. The interfacing method breaks the limitation that the original DSED simulation framework can only solve linear circuits, and it can solve a class of models with non-linear equations, which extends the advantages of DSED simulation to electric traction and renewable energy fields. The proposed modeling method is verified by experiments and pairs of comparisons with other software.In terms of solving stiff systems, based on backward differential formula (BDF), a variable-step algorithm is proposed by combining event driven mechanism, called backward discrete state event driven (BDSED) simulation method. The simulation method can solve the stiff system with stray parameters efficiently, breaks the limitation of the original DSED simulation method, which help support the research of electromagnetic interference, and provide an effective tool for improving the reliability of equipment. The proposed method is verified by experiments and comparison with other software.