Incorporating Limited Monitoring Data And Low-dimensional Linear Models Into Coupled Prediction For Ventilation Online Control System

Nowadays with the rapid development of mechanical ventilation systems(provided with the large ventilation rate)in the man-made environments(e.g.,building and rail transportation environments),the total energy consumption of ventilation system has been increasing in excess.Facing with the serious problems about ventilation energy waste,the main reason was that mechanical ventilation systems cannot meet the non-uniform and dynamic demands of indoor environmental parameters.The aim of this study was to propose a coupled prediction model based on limited monitoring data and low-dimensional linear models(LLM),in order to solve 'faster-than-real-time' prediction issue of indoor environmental fields and construct ventilation online control system for the evaluation and regulation of optimal indoor environments.The ventilation online control system proposed in this work is able to contribute to the favorable development of healthy,comfortable and energy-efficient indoor environmentThis study mainly utilized a coupled prediction model based on limited monitoring data and low-dimensional linear models(LLM)to realize 'faster-than-real-time' prediction of indoor environmental fields(e.g.,pollutant concentration and temperature fields).Of this model,LLM mainly included low-dimensional linear ventilation model(LLVM)and low-dimensional linear temperature model(LLTM),aiming to reduce the simulation costs of indoor environmental fields.In order to further improve the prediction efficiency of pollutant concentration and temperature fields,this study incorporated two advanced rapid prediction models,i.e.,artificial neural network(ANN)and contribution ratio of indoor climate(CRI),into the coupled prediction model.By combining LLM and above rapid prediction models,LLVM-based ANN and LLTM-based CRI were firstly put forward to largely reduce the costs of fast prediction.This study mainly consisted of three stages.(1)The first stage was intelligent control of optimal ventilation rate(ACH)based on LLVM and ANN.By employing numerical simulation and LLVM methods,the database about CO2 concentration was firstly built.Then,after utlizing ANN method for database training,the fast prediction of pollutant concentration fields was also realized by taking ventilation mode,ACH and source position as inputs.Finally,according to prediction results and evaluation index Ev,the optimal control of ACH and ventilation mode was achieved.The results showed that LLVM and ANN could effectively predict the pollutant concentration fields with the maximum errors of 10%and 2%.The maximum prediction error of LLVM-based ANN model was 1%.Based on the optimal evaluation results of ACH,CO2 concentration and ventilation energy consumption can be largely reduced by 30%and 50%.(2)The second stage was intelligent control of optimal supply air temperature based on LLTM and CRI.By using numerical simulation and LLTM methods,the temperature database was firstly constructed.Then,we adopted CRI method to realize the rapid prediction of indoor temperature fields by taking the intensity of heating source(i.e.,supply air temperature,heating body,and window)as inputs.According to the prediction results and evaluation index ET,the optimal control of supply air temperature was realized.The results showed that LLTM and CRI methods can effectively achieve the rapid prediction of temperature fields with the maximum errors of 8%and 10%.The maximum prediction error of LLTM-based CRI model was about 10%.With the optimal evaluation results,indoor thermal comfort can be significantly improved(i.e.,PMV?0),and energy consumption of air-conditioning system was reduced by 32%.(3)The third stage was implementation and visualization of ventilation online control system by using the coupled prediction model based on limited monitoring data and LLVM.By adopting numerical simulation and LLVM methods,the database of CO2 concentration was also built.Then,we expanded the database by incorporating monitoring data into initial CO2 database,and the concentration fields were rapidly predicted by using ANN method based on the coupled prediction model.Combined with wireless technology,visualization device and evaluation index Ev,the optimal ACH control was realized and the actual ventilation online control system was further constructed.The results showed that the ANN method based on limited monitoring data and LLVM can rapidly predict the pollutant concentration fields with the error less than 10ppm.Based on the optimal ACH selection results,CO2 concentration and ventilation energy consumption can be largely decreased by 28%and 43.8%.Moreover,the ventilation online control system can show the accurate control effect and performance about inlet ACH.