Solvers chaining in the IDOS server for dynamic optimization

2014

Self Cite

This paper presents a newly deployed server, IDOS, an online-accessible environment providing the service of solving optimal control problems. Development and deployment of the Interactive Dynamic Optimization Server is a result of projects funded by NCBiR (National Center for Research and Development). One of the outcomes of the project was a modeling language (Dynamic Optimization Modeling Language, DOML) providing a uniform format for defining dynamic optimization problems. DOML is an extension of Modelica language and hence, not only a user can specify his problem in the way he does in Modelica but also (more importantly, for the purpose of this paper) models created in Modelica for simulation purposes can be easily transferred to DOML for solving their related optimization problems. In particular, Modelica models can be calibrated with the help of our server. The paper tries to illustrate the point in depth. It presents the workings of the server and reviews the scope of solvers implemented, focusing especially on those that can be used for calibrating Modelica models. Special attention is devoted to an algorithm using adjoint equations for evaluating sensitivities of model equations with respect to parameters and to calibrating models described by higher index DAEs.

Section: The Idos Servermentioning

confidence: 99%

“…then the gradient of it can be calculated by referring to adjoint equations for the functional (25) and the system (21). The adjoint equations for the functional (25) and the system (21) are considered, for example, in [17]:…”

Section: Estimation Of Parameters Of Higher Index Daesmentioning

confidence: 99%

IDOS - (also) a Web Based Tool for Calibrating Modelica Models

2014

Self Cite

“…The paper presents recent development of solver functionality implemented within DOML (Dynamic Optimization Modeling Language) environment and deployed as part of the IDOS (Interactive Dynamic Optimization Server, described in , see also (Pytlak et al, 2013)) infrastructure. It uses a numerical procedure based on control vector parameterization and RADAU5 together with event (discrete state transition) handling and is capable of solving optimal control problems for hybrid, highindex DAEs.…”

Section: Introductionmentioning

confidence: 99%

“…During the efforts of adapting Optimica for the purpose of deploying it within IDOS environment a conclusion was reached to redesign some of its optimization-related constructs, as its original design brought in some troublesome limitations (details can be found e.g. in (Pytlak et al, 2013) and ). To avoid confusion with Optimica, we then chose to refer to the language as DOML.…”

Section: Introductionmentioning

confidence: 99%

Defining and Solving Hybrid Optimal Control Problems with Higher Index DAEs

Suski

et al. 2017

Self Cite

The paper deals with optimal control problems defined for hybrid systems described by higher index DAEs. We present a prototype solution that supports the whole process from defining such problem to solving it and presenting results. Problem's definition is done with Dynamic Optimization Modeling Language (DOML) which is based directly on Modelica. The proposed numerical procedure for solving the problems of interest has the following features: 1) it is based on the appropriately defined adjoint equations formulated for the discretized equations being the result of the numerical integration of system equations by an implicit Runge-Kutta method; 2) initialization for higher index DAEs is performed with the help of Pantelides' algorithm; 3) it does not require the system to be transformed to ODEs (through differentiation of some algebraic equations). The paper presents numerical examples related to hybrid systems described by index three DAEs, showing the validity of the proposed approach. All software components needed to carry out the computations, i.e. the code editor, compiler, numerical libraries and GUI for presenting results are prepared as parts of a combined platform: Interactive Dynamic Optimization Server (IDOS).

“…The elements and functions of the IDOS server, including exemplary computational results, have already been described in a few publications (e.g. in [22], [23] and [24]) and, in particular, in a parallel paper to be presented on the very same 10 th Modelica Conference ( [25]). Hence, here the functioning of IDOS itself will not be discussed any further, the core focus goes to the DOML format and its implemented compiler (as installed on IDOS).…”

Section: Introductionmentioning

confidence: 99%

DOML -- a Compiler Environment for Dynamic Optimization Supporting Multiple Solvers

2014

Self Cite

The Modelica language may serve well as a base for defining optimal control problems, given a few relatively minor syntax extensions. One example proving that point is Optimica and another one is DOML (Dynamic Optimization Modeling Language) -installed on IDOS (Interactive Dynamic Optimization Language) and described in this paper. The DOML implementation is, actually, heavily based on the (open source) compiler of Optimica but it provides a number of important features absent in its precursor. One main extension of the compiler lies in its built-in mechanism supporting the use of many different optimization solvers (selected on the fly, depending on the content of the problem definition) and to seamlessly add new, external, solvers. In result, the range of problems that may be specified with DOML and solved in the IDOS environment is quite wide and keeps growing. The scope of problems ranges from some static optimization problems through regular ODE, parametric optimization, minimum time problems, up to DAE with higher index. DOML language extensions also provide preliminary support for multi-objective optimization and PDE problems.