PyFMI: A Python Package for Simulation of Coupled
Dynamic Mo dels with the Functional Mo ck-up Interface
Christian Andersson
a,b
, Johan
˚
Akesson
b
, Claus F¨uhrer
a
a
Centre for Mathematical Sciences, Lund University, Lund, Sweden
b
Modelon AB, Lund, Sweden
Abstract
With the advent of the Functional Mock-up Interface (FMI) standard, ex-
changing dynamic models between modeling and simulation tools has been
greatly simplified. At the core of it, FMI is a standardized and unified model
execution interface for dynamic models. FMI has gained widespread adoption
among users and numerous commercial and open source tools implement sup-
port for the standard. In this article, the Python package PyFMI is introduced.
PyFMI supports loading and execution of models compliant with the FMI stan-
dard, called Functional Mock-up Units (FMUs). It includes a master algorithm
for simulation of coupled FMUs together with connections to both Assimulo,
for simulation of single FMUs, and to SciPy, for performing parameter estima-
tion. Accessing models compliant with FMI in Python, which is an open and
accessible scripting language, is intended to further spread the standard and
also promote and facilitate future development of the standard. This is due
to Python being a convenient language for experimentation and prototyping of
numerical algorithms. PyFMI is also demonstrated on a number of problems
that highlights its viability for solving industrial grade simulation problems with
FMUs.
Keywords: Functional Mock-up Interface; FMI; Python; Simulation;
Co-Simulation; Ordinary differential equations; Parameter Estimation;
1. Introduction
Different simulation and modeling tools often use their own definition of how
a model is represented and how model data is stored. Complications arise when
trying to model parts in one tool and importing the resulting model in another
tool, or when trying to verify a result by using a different simulation tool. The
Functional Mock-up Interface (FMI) [1] is a standard to provide a unified model
execution interface for exchanging dynamic system models between modeling
tools and simulation tools. The standard has been a great success as numerous tools,
both open source and commercial have adopted it as well as having gained widespread
adoption among users.
Email addresses: chria@maths.lth.se (Christian Andersson),
johan.akesson@modelon.com (Johan
˚
Akesson), claus@maths.lth.se (Claus F¨uhrer)
不同的仿真和建模工具通常使用它们自己对
模型如何表示以及如何存储模型数据的定义。
当尝试在一个工具中对零件进行建模并在另一
个工具中导入结果模型时，或者在尝试使用不
同的模拟工具验证结果时，会出现并发症。 功
能模拟接口（FMI）[1]是一种标准，用于提供
统一的模型执行接口，用于在建模工具和仿真
工具之间交换动态系统模型。 该标准取得了巨
大的成功，因为众多工具，无论是开源还是商
业，都采用了它，并且已经在用户中得到广泛
采用。
随着功能模拟接口（FMI）标准的出现，
在建模和仿真工具之间交换动态模型已经大大
简化了。 FMI的核心是动态模型的标准化和统
一模型执行界面。
FMI已在用户中得到广泛采
用，众多商业和开源工具实现了对该标准的支
持。在本文中，介绍了Python软件包PyFMI。
PyFMI支持加载和执行符合
FMI标准的模型，
称为功能模拟单元（FMU）。它包括用于仿真
耦合FMUs
的主算法并与Assimulo结合、用于仿
真单个FMU、结合SciPy用于执行参数估计。访
问Python中符合FMI的模型，这是一种开放且
易于访问的脚本语言，旨在进一步传播标准，
并促进和促进标准的未来发展。这是因为
Python
是一种方便的语言，可用于数值算法的
实验和原型设计。 PyFMI还演示了一些问题，
强调了它在用FMUs解决工业级模拟问题方面
的可行性。
