Decision-makers use system dynamics models to understand how model structure causes problematic behaviors, and how the structure should be changed to improve performance. However, understanding problem behavior and designing policies can be complicated without analytical tools. Existing methods focus on feedback loops that drive dynamic behavior. We propose a new method, Analysis of Dynamic Models by Optimization (ADMO), which focuses on understanding the causes of well-defined problems and on guiding system and policy design. Unlike existing methods, ADMO can evaluate the influence of exogenous variables and nonlinearities, leading to new understandings of endogenous behavior and providing insights for policy design. ADMO can be employed with any model, including differential equation, agent-based, discrete event, or hybrid, requiring minimal effort. Two cases demonstrate the application of ADMO to analyze problems and improve both system design and policies. © 2023 The Authors. System Dynamics Review published by John Wiley & Sons Ltd on behalf of System Dynamics Society.

中文翻译：

---

通过优化分析动态模型

决策者使用系统动力学模型来了解模型结构如何导致有问题的行为，以及应如何更改结构以提高性能。然而，如果没有分析工具，理解问题行为和设计政策可能会变得复杂。现有方法侧重于驱动动态行为的反馈循环。我们提出了一种新方法，即优化动态模型分析（ADMO），其重点是理解明确问题的原因以及指导系统和政策设计。与现有方法不同，ADMO 可以评估外生变量和非线性的影响，从而对内生行为有新的理解，并为政策设计提供见解。ADMO 可以与任何模型一起使用，包括微分方程模型、基于代理的模型、离散事件模型或混合模型，只需最少的努力。两个案例展示了 ADMO 的应用来分析问题并改进系统设计和政策。© 2023 作者。系统动力学评论由 John Wiley & Sons Ltd 代表系统动力学协会出版。