Study On Load Forecast Of Air Conditioning System Of Office Building Based On RC Network And Data Driven Method

With energy problems becoming increasingly tense today,it is particularly important to accurately predict the dynamic loads of buildings to control indoor thermal environment and reduce system energy consumption.The RC model can better reflect the thermophysical properties of the building,and the data-driven method based on computer technology can monitor and analyze the actual operation status of the HVAC system of the building at any time.This project mainly studies the dynamic load forecasting model of building based on the analysis of the sensitivities of various meteorological factors and indoor staff factors to the HVAC system loads of the building.The energy simulation software,Design Builder,is used to build the case building model to simulate the energy consumption of the HVAC system,with the human mode and the unmanned mode.Then,the simulation results are used to establish the database of the building envelope loads and the database of the staff,lighting,and equipment loads caused by staff disturbance.Based on the sensitivity analysis method,the significance of various meteorological and the staff factors on the load is deeply studied.Using the standard regression coefficient as the evaluation index and sorting them,the factors with greater influence are selected to simplify and optimize the input parameters of the load forecasting model.Based on the RC thermal network theory,a simplified RC model of the case building is constructed by using the original database of the envelope structure load and the genetic algorithm to identify the structure parameters of the RC model.Then based on the Bayesian network and the original database of staff disturbance load,the model for predicting staff disturbance load is established.Subsequently,this paper further uses fuzzy neural network theory combined with the above research results to establish a comprehensive forecasting model for the dynamic load of the HVAC system of the building.The prediction method comprehensively utilizes the strong ability of the RC thermal network to reflect the thermal physical properties of the building envelope,and the nonlinear processing capability of the data-driven method for the uncertainty of the staff factor.After comparison,the comprehensive load forecasting model established in this paper has higher prediction accuracy.