Object Oriented Nonlinear Modelling and Simulation of Thermo-fluid Systems with Evolutionary Optimisation of Parameters

B. Šulc and J. Alam Jan (Czech Republic)


modelling, simulation, object-oriented, training simulator, ptimisation, evolutionary computing.


This paper presents object oriented non-linear modelling and simulation of thermo-fluid dynamic systems. Object oriented modelling is an excellent means for building model libraries. Model libraries can easily be reused by users who are not expert in modelling. Object oriented modelling describes each part of the model as an object with a certain behaviour. Each object is formulated from basic physical laws, and it needs few parameters determinable mostly from construction data. Nonlinear modelling in the state space form has been per formed by dividing the process into control volumes called modules. The goal is to develop moderately complex non linear models that capture the key dynamic properties over a wide operation range. Thermo-fluid dynamic systems are well suited for object oriented modelling because of the extensive use of standard components like pumps, valves, heat exchangers, boilers etc. This makes it possible to capture a large number of complex process configurations with a limited number of model blocks. The model library has been utilised to build a model of a training simulator [1] of a coal fired thermal power plant. As electric energy generation processes cannot be interrupted, there is no opportunity to perform any special experiments for parameter optimisation; for this reason evolutionary computing methods such as "Adaptive Genetic Algorithm (AGA)" and "Simulated Annealing (SA)" have been used to optimise the parameter from normal recorded operation data.

Important Links:

Go Back