Study On Air Supply Volume Prediction Of Air-Conditioned Room Based On GA-ANN

The market demand for air-condition system in hotels, stores and office buildings is increasing rapidly as the fast development of social economy and people's living standard, the supply-and-need of electric power has being intensified, decreasing the energy consumption of air-conditioning system is not only the urgent task to ease the tension but also the requirement of environmental protection. Therefore, some advanced air-condition systems, such as VAV , ice-storage air-conditioning system, are welcomed by consumers and applied to more projects after a long period of recession. However, some problems are also to be optimized in the control system for the reason of the system itself, therefore, some universities, enterprises and research institutes are undertaking their researches on the two air-condition forms. The prediction of air supply volume in air-conditioned rooms are playing a big role in the application of the total air volume control method in VAV system and accurate control of the amount of ice preservation in ice-storage air-conditioning system.The paper firstly analysised the disadvantages of some traditional design model methods in the prediction of air supply volume in air-conditioned rooms, then introduced the idea of artificial neural network. It combined the genetic algorithm with BP algorithm on the basis of clarifying the disadvantages of BP algorithm such as low speed converge and easily being subject to the partial minimum. By the way of genetic algorithm, it sought the overall advantages for the weights and bias of neural network, then the partial advantages by BP algorithm in order to get a high converging speed and accurate prediction.The paper imitated the relative load data of air-conditioning system in an office building in Nanjing by taking the advantage of the powerful function of DeST after analysizing the actual weather condition. We believe the dada available is reliable and can be set as the training sample and test sample of air supply volume prediction neural network mold for air-conditioned rooms in building.