In this work, a line-up competition algorithm (LCA) is applied to solve the dynamic optimization problems derived from unsteady chemical systems. The problems are first converted ito nonlinear programming problems using the concept of control vector parametrization. The parameters embedded in the converted problems then are selected by LCA. To improve numerical accuracy, the normal (Gaussian) sampling policy is introduced to replace the uniform sampling policy used in basic LCA. Variable step input (VSI) and variable ramp input (VRI) are respectively considered to rebuild the control policy in solutions. Some typical examples are provided to demonstrate the robustness and efficiency of this modification.

No abstract

2. Reach the specifi ed yield for the desired product(s) in the minimum operating period under some constraints. The operating constraints in the above expressions usually arise from the physical bounds on the control variable(s), safety limitations, and environmental regulations. In optimal control studies, the fi rst expression can be formulated as a maximum yield (or conversion) problem, while the second one as a minimum operating time problem. By comparing with the solutions from the above optimal control problems (OCPs), the operators can easily assess whether or not the real production qualities are satisfactory and further consider how to modify the practical operating policies to make the processes more better.Basically, to solve the above OCPs is a challenging task due to the highly non-linear, non-convex and multimodal nature derived from chemical systems. In the past decades, most developed solution techniques are based on one of the following theories:How to apply the global optimization technique, simulated annealing, and to explore the operation of batch reactors is addressed in this study. Based on the operating purposes and the imposed constraints, the batch reactor operations are fi rst formulated as two optimal control problems: the maximal yield (or conversion) problem and the minimal operating time problem. The problems are then converted into nonlinear programming problems by the concept of control vector parameterization. The converted problems are solved by the algorithm derived from simulated annealing to determine the optimal operating policy and the performance index. These results are useful in assessing design and operation of batch reactors. In this article, the CSTR model is used to demonstrate the convenience and robustness of the proposed algorithm. Two typical reaction models are used to discuss the operations based on the optimal solutions.On s'intéresse dans cette étude à la façon d'appliquer la technique d'optimisation globale, dite de recuit simulé, pour explorer le fonctionnement de réacteurs discontinus. À partir des objectifs de fonctionnement et des contraintes imposées, le fonctionnement des réacteurs discontinus est d'abord formulé sous forme de deux problèmes de contrôle optimums: le problème du rendement (ou la conversion) maximal et le problème du temps de fonctionnement minimal. Ces problèmes sont ensuite convertis en problèmes de programmation non linéaires par le concept de paramé-trisation des vecteurs de contrôle. Les problèmes convertis sont résolus par l'algorithme établi à partir du recuit circulé afi n de déterminer la politique de fonctionnement optimale des réacteurs et l'indice de performance. Ces résultats sont utiles dans l'évaluation de la conception et du fonctionnement des réacteurs discontinus. Dans cet article, le modèle CSTR permet de démontrer le côté pratique et la robustesse de l'algorithme proposé. Deux modèles de réaction typique sont utilisés pour analyser le fonctionnement des réacteurs d'après des solutions optimales.