Research on Radiator Area Design Based on BP Neural Network
Reducing the energy consumption of heating systems is highly needed to achieve energy saving and carbon reduction in the building sector,and the proper design of the radiator area is significant for the improvement of energy efficiency in heating systems,higher thermal comfort in heated rooms,and reduction of initial investments in systems.However,the traditional radiator design model based on steady-state heat balance usually neglects the thermal inertia of building envelopes,internal heat gains and others,which causes the oversized radiator.In this regard,a simulation database for different room types and heating operation conditions was built by means of building thermal simulation in this study.A new radiator design model based on BP neural network was developed,and its accuracy and applicability were comprehensively analyzed.Furthermore,the design effect of the model proposed was compared with that of the traditional radiator design model.Results show that the R2 of this neural network model is higher than 0.724,and the MAE and RMSE are less than 0.533 m2 and 0.680 m2,respectively.For different climate zones,different types of radiators and different supply water temperatures,the proposed model can control the minimum indoor heating temperature within the range of 18～21 ℃.For traditional radiator design models,the room temperature is far higher,and experience overheating even in some rooms.In this sense,the radiator design model proposed in this study can alleviate the indoor overheating in winter,effectively reduce energy consumption,and minimize the investment cost of the heating system.
万方数据
维普
