多领域统一建模初始化变量选择分析

Analysis of Initialization Variables Selection in Multi-Domain Modeling

多领域统一建模中,为求解DAEs(Differential Algebraic Equations)需要解决初始相容性问题。结构化分析可以验证初始相容性,求解部分初值并确定方程自由度,但无法确定需要初始化的变量。错误地选择初始化变量可能极大地降低求解与仿真的效率,甚至导致失效。分析了DAEs初始相容性问题,提出了基于DM分解(Dulmage and Mendelsohn Decomposition)的初始化变量选择策略,通过DM分解将结构化分析得到的系统分解为两部分,对欠约束部分进行可达性分析,确定需要初始化的变量。该方法可以有效地确定初始化变量,避免错误的选择。

In multi-domain modeling, the problem of initial consistency must be settled to solve differential algebraic equations(DAEs). Structural analysis is capable of verifying the initial consistency,solving some initial values and determining the degree of freedom, but not capable of affirming the variables that need to be initialized. If improper selection is made for the initiation variables, it will become low efficient and even fail to solve and simulate the model. After analyzing the initial consistency of DAEs, a strategy based on DM decomposition is proposed for selecting the initialization variables. The method decomposes the system into two parts with the use of DM decomposition after the system is structurally analyzed. By analyzing the reachability of under constraint part, the initialization variables can be affirmed. This method is efficient to affirm the initialization variables and avoids wrong selections.