Energy Collection Network Modeling And Prediction Of Indoor Temperature And Humidity Environment

With the further development of industrialization and urbanization in China,the proportion of building energy consumption is rising,and the green and low-carbon sustainable development of buildings has become the current focus.Air-conditioning energy consumption accounts for more than one third of the total energy consumption of buildings in the construction industry.Reducing energy consumption of HVAC systems by advanced control and optimal operation have been regarded as an important measure to alleviate energy crisis and improve environmental sustainability.Accurate building environment information is the key to ensuring the successful application of these strategies,and accurate modeling of the building environment is the basis for obtaining this information.Due to the characteristics of nonlinearity,high coupling,large time delay and large inertia of the building air conditioning system,in order to achieve the accurate control of the indoor environment of the building,it is necessary to predict the building environment information on the basis of obtaining accurate information.Aiming at the problem of poor real-time performance of CFD simulation,this paper proposes a real-time accurate energy connection network modeling method for building indoor temperature and humidity field,and establishes an improved BP neural network prediction model to achieve simultaneous prediction of indoor temperature and humidity,which realizes simultaneous prediction of indoor temperature and humidity and fast and accurate reconstruction of temperature and humidity field,and provides a good basis for the design and optimization of subsequent controllers.Although CFD can achieve high-precision modeling,but the calculation speed is slow and can not be used for controller design and analysis.The regional model and multi-region model can improve the modeling speed,but the accuracy is low.In this paper,the real-time accurate energy connectivity network modeling method and the selection rules of indoor temperature and humidity energy nodes are proposed.By introducing POD technology,the indoor temperature and humidity field system is mapped into a linear system within ten steps only related to the POD mode coefficient,and a linear stochastic estimation and Kalman filter model is established to estimate the optimal POD mode coefficient.Rapid reconstruction of indoor temperature and humidity field is realized by real-time temperature and relative humidity values collected at energy nodes.This model is compared with the method proposed by Meyer R D,the distribution of the error of the reconstruction field is analyzed.The results show that the our method has good noise suppression capability,and the modeling accuracy is higher,and can achieve high-precision reconstruction of transient and steady temperature and humidity fields.The forward modeling based on the principle of building heat transfer has high modeling accuracy and good prediction effect,but it requires a lot of detailed building and environmental parameters and requires expert knowledge,which is limited in engineering application.Since the neural network has strong nonlinear approximation ability and model independent characteristics,we propose an improved BP neural network prediction model to achieve continuous prediction of indoor temperature and relative humidity.Based on the historical data collected by the temperature and humidity detection system of the Qianjiang Cigarette Factory,the influence of different input variables on the prediction accuracy was studied with 10 minutes as the prediction interval.he short-term and long-term prediction ability of the model was tested and Regression analysis was carried out on the predicted values and actual measured values.Experiments show that the improved BP neural network prediction model can simultaneously predict the indoor temperature and relative humidity for 6 hours and the indoor temperature for 72 hours.The comparison coefficient is compared with the results of similar research work.The results show that the improved BP neural network prediction model has better prediction effect on indoor temperature,and the prediction effect on indoor relative humidity needs to be further improved.