Real-time Control Strategy For HVAC System Based On Fast Prediction Of Non-uniform Distribution Of Indoor Environment

Currently,buildings account for 40%of the global final energy consumption,with heating,ventilation and air conditioning（HVAC）systems accounting for more than 50%of total building energy consumption,especially in non-residential buildings,where the proportion can be as high as 60%.With the rapid development of urbanisation and the rising living standards of the people,the energy consumption of HVAC systems is also increasing.However,traditional HVAC systems do not operate according to the real-time characteristics of the dynamic,non-uniformly distributed changes in the indoor environment,causing unstable comfort and energy wastage.From the point of view of the dynamic requirements of the indoor environment and the energy efficiency of the building,it seems that the traditional HVAC system control methods are no longer stable and reliable.From the perspective of dynamic indoor environmental requirements and building energy efficiency,traditional HVAC system control methods no longer seem to be stable and reliable.This research proposes a real-time control method and strategy for HVAC systems based on the prediction of dynamic,non-uniform distribution of indoor environments.Through real-time monitoring and rapid prediction of the indoor environment and real-time assessment of the indoor environment using self-define evaluation indicators,the HVAC system can then be regulated in real time.The HVAC control strategy proposed in this research helps to create a"healthy,comfortable and energy efficient"indoor environment.This research focuses on the rapid prediction of the distribution of indoor environmental parameters（CO₂ concentration,temperature and humidity）by means of limited indoor environmental monitoring and"low-dimensional linear"fast prediction models.The“low-dimensional linear”fast prediction models include neural network-based low-dimensional linear ventilation models（LLVM-based ANN）,contribute ratio of indoor temperature-based low-dimensional linear temperature models(LLTM-based CRI_（T）)and the contribute ratio of indoor humidity-based low-dimensional linear humidity models(LLHM-based CRI_（H）),which used for rapid prediction of the spatial distribution of indoor pollutants,temperature and humidity respectively.This is followed by a comprehensive evaluation of the health and comfort of people and the energy consumption of the HVAC system through self-designed evaluation indicators for pollutants,temperature and humidity（E_V,E_T,E_H）,which are further used for the real-time control of the HVAC system.At the same time,an intelligent ventilation real-time monitoring platform based on microcontroller design is built,the control logic is effectively utilised and an intelligent regulation and control strategy for the HVAC system that integrates multiple factors is proposed.The research was conducted in two phases:（Ⅰ）Feasibility study for combining fast prediction models and ventilation systemsIn this phase,a real-time monitoring ventilation control system was developed using a"low-dimensional linear"fast prediction model and microcontroller technology to verify the feasibility of the"low-dimensional linear linear"fast prediction model and the control logic of the ventilation real-time monitoring system.Validation using a small ventilated environmental chamber system（1.8 m³）which consists of:an indoor dry ice contamination source,a microcontroller-based design monitoring device and a ventilation system.Firstly,the LLVM-based ANN model was used to quickly predict the pollutant concentration distribution in indoor environments,and the predictions were experimentally validated.The information of the pollution source（i.e.,CO₂ from dry ice）is then transmitted via a Zigbee wireless module and the ventilation is evaluated according to a self-designed evaluation indicator（E_V）to determine the optimal ventilation mode.The ventilation control system then controls the indoor fans in real time,saving energy in the system while ensuring the health of the personnel.In particular,the energy saved by the ventilation system（especially by the fans）can reach up to 47%.（Ⅱ）Research on optimal control strategies for HVAC systems based on fast predictive modelsIn a previous study,we achieved optimal selection of ACH and supply air temperature by means of LLVM-based ANN and LLTM-based CRI_（T）fast prediction models.In this part of the study,we further developed a rapid prediction model for indoor humidity using the low-dimensional linear humidity model（LLHM）and the contribution ratio of indoor humidity(CRI_（H）),as well as the thermal susceptibility index（TS）.Firstly,the prediction database of CO₂,temperature and humidity was constructed using computational fluid dynamics（CFD）methods,and secondly,the database was expanded（reducing data storage）using fast prediction models such as LLVM-based ANN,LLTM-based CRI_（T）and LLHM-based CRI_（H）to achieve indoor CO₂ concentration,temperature and humidity Rapid prediction of spatial distribution.Finally,the optimal balance between indoor environmental quality and energy consumption was evaluated in a comprehensive manner based on self-designed evaluation indicators（E_V,E_T,E_H）,resulting in an optimal HVAC system control strategy.In particular,the total energy consumption of the HVAC system can be reduced by up to 35%.