建模者觀點模擬方法比較-以防空作戰模擬器為例

Comparative Analysis of Simulation Methods from the Modeler's Perspective: A Case Study of an Air Defense Operations Simulator

劉達生、賴智明、吳欣諾
D. S. Liu, C. M. Lai and H. N. Wu

國防大學管理學院資源管理及決策研究所

摘要

在作業研究領域中，模擬技術被視為解析複雜動態問題的重要工具，當中代理人模擬適合探討具有差異性的個體及其行為與彼此的互動；而離散事件模擬則適合程序性議題，尤其是物件進入系統後的一連串歷程。當有建模需求的時候，建模者需要根據問題特性選擇最適合的方法來建立模型描述對象。防空議題在國防領域中複雜且重要，一個防空系統中有許多個子系統，接戰流程中每個子系統必須互相合作，並且需考慮多種威脅和防禦策略的互動影響，牽涉到流程也牽涉到互動，單一模擬方法可能面臨難以涵蓋，或是建模過程複雜度高的問題。
近期研究顯示，將兩種或更多的模擬方法綜合運用的混合模擬方法逐漸引起廣泛關注。然而，混合模擬與單一模擬方法之間的比較性研究相對不足。因此，本研究以防空問題為對象，分別採用離散事件模擬、代理人模擬與混合模擬，以AnyLogic模擬平台建立模擬模型。從建模者的角度紀錄建模過程中所遇到的困難、感受與啟發。研究目的旨在從建模者角度比較三種方法的差異，彌補模擬方法比較的研究空白。
本研究從建模者的角度深入探討不同建模方法的差異，並就三種模擬方法的應用進行評估。結果發現：對建模者而言，在單一模擬平台上針對模擬對象的不同細節，選擇合適的方法來描述，建立起混合模擬模型是侷限性較小，也較具有彈性的建模方式。相對的，混合模擬也需要建模者熟悉多種建模方法與建模工具，而有較高的門檻。

關鍵字：混合模擬、離散事件模擬、代理人模擬、模擬方法比較、建模者。

ABSTRACT

In the field of operations research, simulation techniques are considered important tools for analyzing complex dynamic problems. Agent-based simulation is suitable for exploring diverse individuals and their interactions, while discrete event simulation is suitable for procedural issues, particularly the sequence of events after objects enter the system. When there is a need for modeling, modelers need to choose the most appropriate method to build models that describe the objects based on the problem characteristics. Air defense issues are complex and critical in the defense domain, with many subsystems within an air defense system needing to cooperate during engagement processes. Various threats and defense strategies must be considered, involving processes and interactions. A single simulation method may face challenges in coverage or complexity during the modeling process.
Recent research indicates that the hybrid simulation method, which integrates two or more simulation methods, is gaining widespread attention. However, there is a relative lack of comparative studies between hybrid simulation and single simulation methods. Therefore, this study focuses on air defense issues and adopts discrete event simulation, agent-based simulation, and hybrid simulation, using the AnyLogic simulation platform to build simulation models. The study aims to compare the differences between the three methods from the perspective of modelers, addressing the research gap in simulation method comparison.
This study delves into the differences between different modeling methods from the perspective of modelers and evaluates the applications of three simulation methods. The results show that, for modelers, building hybrid simulation models with appropriate methods to describe different details of the simulation object on a single simulation platform has fewer limitations and is a more flexible modeling approach. However, hybrid simulation also requires modelers to be familiar with various modeling methods and tools, presenting a higher threshold.

KEYWORDS: Hybrid Simulation; Discrete Event Simulation; Agent-Based Simulation; Simulation Method Comparison; Modelers