The Research Of Refrigeration System Simulation Based On Artificial Neural Network

As an application foundational science, System Simulation has gotten great development in recent years, which has been applied widely in various fields. The widely use of System Simulation has created a great deal of value to the economy and society. System Simulation will play an important role in science research as a means of discovering problems, analyzing problems and resolving problems.A branch of the refrigeration field comes into being after System Simulation is applied in the refrigeration system-----the refrigeration system simulation.The step, making samples to do tests, can be cancelled when the refrigeration system simulation being adopted in the design of air-conditionings, which will, not only save considerable money, but also shorten the design period. Those advantages brought by the refrigeration system simulation technology can make products more competitive. So it is very necessary for corresponding companies to develop and use refrigeration system simulation technology in a time filled with more and more competition. In addition, it will easily get the performance of an air-condition imposed certain control means on through using refrigeration system simulation technology, which can help to choose or modify the control means.In recent years, Artificial Neural Network (ANN) has received extensive concern from many fields because of its distinct characteristics such as associative memory and self-learning ability. Among all kinds of neural networks, BP neural network is the one that gets used most extensively in a great deal of fields such as pattern recognition, intelligent control, signal process and nonlinear optimization just for its excellent abilities of nonlinear mapping and generalization. The main reason of why ANN can be easily applied in refrigeration system simulation is its ability of fitting any nonlinear continuous function in any precise and its distinct number-based feature.The research object of this thesis is the steady-state simulation of home air-conditioning. Home air-conditionings generally adopt vapor-compressed refrigeration cycle, so they are made up of four main components: compressor,condenser, capillary and evaporator. The objective of this thesis is to build up ANN simulation models of those four components and the whole system.Based on Refrigeration System Simulation and Optimization, Refrigeration System Intelligent Simulation and other literature, this thesis, at first, builds up the thermo-parameter model of compressor with "dark-box" method, and the distributed model of condenser, capillary and evaporator with infinitesimal analysis method. After that, it gets the simplified models of condenser, capillary and evaporator by simplifying the corresponding distributed models with Lagrangian Middle-value Theorem. And then, through applying ANN in the thermo-parameter model and the simplified models of condenser, capillary and evaporator, the corresponding ANN models of the four components are obtained. The samples to train the ANN models of compressor are from Refrigeration System Intelligent Simulation, and those of the left three ANN models are produced by respective distributed models. At last, a simple simulation software of refrigeration system is developed with Microsoft Visual Basic 6.0.