| Heat exchanger network synthesis (HENS) problem is very complex and involves combinatorial problems in the "matching" between hot and cold streams. Both qualitative and quantitative considerations are needed in network structure optimization. Here an intelligent system integrating knowledge-based system (KBS) and artificial neural network (ANN) is developed to solve the HENS problem.Firstly, the dissertation gives a review on the study history of the HENS. An evaluation of various methods is performed from both theory and industry point of view. The idea of establishing an intelligent system is proposed, with knowledge-based system to describe process rules and artificial neural network to do minimum searching. The feasibility and features of this system are briefly mentioned.The Hopfield Neural Network(HNN) , which has a patentiality to solve combinatorial problems, is selected to be the minimum searching model. According to the stability theory of Lyapunov, the working mechanism and character of HNN, which are the mathematic basis of HNN applying in HENS, is analysed. It is shown that almost all of the engineering optimiztion problems can be reduced to minimum search problems of real-valued quadratic function with two-valued variables excluding linear terms. From this, A set of neurons is physically defined and the establishment of energy function corresponding to the objective function of the problem is detailed from which the dynamic equation of the neuron-state is derived.The implementing technique of the dynamic system of HNN is investigated. The paper explores the reason of non-robust of original dynamic system in term of mathematics and suggests some modifications. A modified neuron-state model is imported intending to improve the behavior of the system. The value space of operating parameters is explored by a lot of experiment. After examining the influence of the initiation of the neuron-state and...
|
PDF Full Text Request