Abstract
The iterative process of simulation optimization is a time-consuming task, as it involves executing the main simulation program in order to evaluate the optimal constraints and objective functions repeatedly according to the values of tuner parameters. Parameter optimization for a model of a multi-domain physical system based on Modelica is a typical simulation optimization problem. Traditionally, each simulation during each iterative step needs resolve all the variables in all the mass differential-algebraic equations (DAE) generated from the simulation model through constructing and traversing the solving dependency graph of the model. In order to improve the efficiency of the simulation optimization process, a new method named partial simulation resolving algorithm based on the set of input parameters and output variables for complex simulation model was proposed. By using this algorithm, a minimum solving graph (MSG) of the simulation model was built according to the set of parameters, constraints, and objective functions of the optimization model. The simulation during the optimization iterative process needs only to resolve the variables on the MSG, and therefore this method could decrease the simulation time greatly during every iterative step of the optimization process. As an example, the parameter optimization on economy of fuel for a heavy truck was realized to demonstrate the efficiency of this solving strategy. This method has been implemented in MWorks—a Modelica-based simulation platform.
Highlights
Processes 2019, 7, 334 structure and system architecture of Modelica, we made further research on the solving strategy for large scale continuous-discrete hybrid differential-algebraic equations (DAE) [9], and exerted ourselves to implement a hybrid modeling platform based on Modelica for multi-domain physical systems [10,11]
Parameter optimization for multi-domain physical systems under the simulation platform based on Modelica is a typical simulation optimization problem
Parameter optimization for a Modelica model needs to execute simulation repeatedly in order to evaluate design functions, and the efficiency of the process of design optimization depends on the efficiency of repetitious simulation to a great extent
Summary
A great deal of basic models from multi-domains, including mechanical, electrical, hydraulic, thermodynamic, and control system are defined in these libraries. There are several applicable modeling and simulation tools based on Modelica, such as Dymola [3,4], Math Modelica [5,6], and Open Modelica [7,8] Up until now, these are the three most influential and powerful software systems based on Modelica. Processes 2019, 7, 334 structure and system architecture of Modelica, we made further research on the solving strategy for large scale continuous-discrete hybrid differential-algebraic equations (DAE) [9], and exerted ourselves to implement a hybrid modeling platform based on Modelica for multi-domain physical systems [10,11]. Our work has come into being as a prototype software system—MWorks [12]—which provides six main modules: integrated modeler, compiler, analyzer, solver, optimizer, and postprocessor
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
