Research On Energy Efficiency Prediction Technology Of Surface Water Source Heat Pump Based On The Neural Network

In recent years, with the ministry of finance, housing and urban-rural development vigorously promote application demonstration of the renewable energy construction across the country, as a kind of using shallow geothermal renewable energy technologies, the surface water source heat pump is in the large-scale application in the field of building energy efficiency. However, as a new energy-saving technology, the surface water source heat pump, with the short-term extension and application in the field of construction, bad applied basic conditions, weak technical reserves and deficient professional talent team, no matter from design, construction or management level is still in the stage of exploration. So far, guidelines of surface water source heat pump that suitable for Hunan province has been in lack. Design and construction personnel often design and construct on the basis of projects in other regions, after the partial completion of the project, resulting in a considerable difference between the actual energy efficiency and design in the heat pump, which seriously hindered the applications of surface water source heat pump to the field of building energy efficiency.In order to solve above problems, in this article, first of all, the energy efficiency for a surface water source heat pump system is tested under the summer operating conditions. Through the relevant testing instruments and equipment, parameter of the lake’s temperature, supply and return water temperature in lake side, flow, supply and return water temperature in user-side and flow are collected. The above parameters on the operating condition of service units and the impact of energy efficiency are analyzed to obtain the actual evaluation method of energy efficiency. Secondly, based on practical testing data, with the aid of BP neural network and genetic algorithm to construct a predictive model of energy efficiency whose input layer node is 6, hidden layer nodes is also 6 and output layer node is 1. After validation, the determination coefficient of reflecting the predictive accuracy is 0.98327. Finally, this article introduces the application of the model in the surface water source heat pump operating and the design stage.Through these studies, the following conclusions are made:(1) Surface water source heat pump system COP is in cyclical change with supply and return water temperature in lake side and that in user-side; (2) A BP neural network predictive model of energy efficiency whose input layer node is 6, hidden layer nodes also is 6 and output layer node is 1 is constructed; (3) In operation stage, the application of the model is found that the evaporator fouling, poor heat transfer is the main reasons in the lack of the unit refrigerating capacity, (4) In the design stage, with using the above model, the best match for supply and return water temperature and flow can be found.The construction of the above model can be supported technically from two aspects of design and operation for the application of the surface water source heat pump in our province, to promote ground source heat pump technology in the development of industrialization in our province.