离散状态事件驱动仿真方法及自适应预估校正算法

Discrete State Event Driven Method and Self-Adapted Predictor-Corrector Algorithm

在电力电子系统分析中,需要对带间断和刚性的常微分方程组进行仿真计算。然而,采用传统的时间离散算法来求解此类方程组时会遇到诸多困难。Kofman等基于离散事件系统规范(DEVS)提出了量化状态系统(QSS)算法,它不是对时间进行离散,而是将状态量进行离散。QSS算法可以有效求解带间断和刚性的常微分方程组。基于离散事件算法思想,提出适合于电力电子仿真的离散状态事件驱动(DSED)仿真方法;同时,为提高DSED方法的精度,提出基于DSED的预估校正算法;为大幅减少计算量,通过研究计算步数与状态量幅值和频率的关系,提出自适应方法。仿真算例证明了所提算法的有效性。

In the analysis of power electronics system, it is necessary to simulate ordinary differential equations （ODEs） with discontinuities and stiffness. However, there are many difficulties in using traditional discrete-time algorithms to solve such equations. Kofman and others presented the quantized state systems （QSS） method in the discrete event system specification （DEVS） formalism. The discretization is applied to the state variables instead of time range in QSS. QSS is very efficient to solve ODEs with discontinuities and stiffness. Based on the idea of discrete event, a discrete state event driven （DSED） method is presented in this paper. This method is fit for simulation of power electronics system. Furthermore, a predictor-corrector algorithm is presented based on DSED to improve the accuracy. Also calculation steps are studied with the amplitudes and frequencies of state variables. Accordingly, a self-adapted method is proposed, which can reduce the computation substantially. Numerical examples verified the effectiveness of the proposed algorithms.