A hybrid deep transfer learning strategy for thermal comfort prediction in buildings. (15th October 2021)

• Record Type: Journal Article
• Authors: Somu, Nivethitha; Sriram, Anirudh; Kowli, Anupama; Ramamritham, Krithi
• Is Part Of: Building and environment. Volume 204(2021)
• Publication Date: 2021-10-15
• Subjects: Thermal comfort -- HVACs -- Energy efficiency -- Buildings -- Transfer learning -- Deep learning; HVAC heating, ventilation and air conditioning -- PMV predicted mean vote -- TL transfer learning -- CNN convolutional neural networks -- LSTM long short term neural networks -- SMOTE synthetic minority over sampling technique -- MCC Matthew's correlation coefficient -- IEQ indoor environment quality -- ANN artificial neural networks -- SVM support vector machine -- RF random forest -- k−NN k nearest neighbor -- LR logistic regression -- GBM gradient boosting machine -- MLP multi layer perceptron -- TCP thermal comfort parameters -- AT air temperature -- ST skin temperature -- RH relative humidity -- PR pulse rate -- MR metabolism rate -- SC skin conductance -- CF clothing factor -- OS oxygen saturation -- CT clock time -- BA body surface area -- ID indoor duration -- E energy consumption -- WS wind speed -- AAT average AT -- OT outdoor temperature -- OAAT outdoor AAT -- AV air velocity -- OARH outdoor average relative humidity -- MRT mean radiant temperature -- AMR average MR -- TC thermal comfort -- AMRT average MRT -- OH outdoor humidity -- HSS HVAC status and set point -- AAS average air speed -- OI outdoor illuminance -- MST mean surface temperature -- IT indoor temperature -- TV temperature variance -- IH indoor humidity -- MSTG mean surface temperature gradient -- IMRT indoor mean radiant temperature -- MSTD mean surface temperature difference -- RT radiant temperature -- SS skin surface -- CO2 CO2 concentration -- CS clothing surface -- IL illuminance level -- IAT indoor AT -- FA floor area -- OAT outdoor AT -- We weight -- SR solar radiation -- He height -- TL thermal load -- Pos position -- TSV thermal sensation vote -- IL heating/cooling intensity and location -- AL activity level -- CO chair occupancy -- CT climate type -- GT global temperature -- AC adaptive control -- TA thermal acceptability -- BOM building operation mode -- TP thermal preference -- ART average radiant temperature -- VAV variable air volume control settings -- HR heart rate -- TR thermostat readings -- IAV indoor air velocity -- WD weather data -- IAH indoor air humidity -- ORH outdoor relative humidity -- OAV outdoor air velocity -- A Turb air turbulence; Buildings -- Environmental engineering -- Periodicals; Building -- Research -- Periodicals; Constructions -- Technique de l'environnement -- Périodiques; Electronic journals; 696
• DOI: 10.1016/j.buildenv.2021.108133
• Languages: English
• ISSNs: 0360-1323

Abstract

Abstract: Since the thermal condition of living spaces affects the occupants' productivity and their quality of life, it is important to design effective heating, ventilation and air conditioning (HVAC) control strategies for better energy efficiency and thermal comfort. An essential step in HVAC control and energy optimization is thermal comfort modeling. Recently, data-driven thermal comfort models have been preferred over the Fanger's Predicted Mean Vote (PMV) model due to higher accuracy and ease of use. However, the unavailability of comprehensive labelled thermal comfort data from the occupants poses a significant modeling challenge. This paper addresses data inadequacy issues by adopting 'transfer learning' to leverage well learned knowledge from source domain (same climate zones) to target domain (different climate zone) where modeling data is sparse. Specifically, a Transfer Learning based Convolutional Neural Networks-Long Short Term Memory neural networks (TL CNN-LSTM) is designed for effective thermal comfort modeling that exploits the spatio-temporal relations in the thermal comfort data. The significant modeling parameters for TL CNN-LSTM are identified using the Chi-squared test. Further, the lack of sufficient samples across all thermal conditions in the available thermal comfort datasets was handled by Synthetic Minority Oversampling Technique (SMOTE). Experiments with two source (ASHRAE RP-884 and Scales Project) and one target (Medium US office) datasets demonstrate the ability of TL CNN-LSTM in achieving an accuracy of >55% with limited data in target buildings. The limitation of TL CNN-LSTM is its continued dependence on intrusive parameters and the challenges in assessing its adaptability to different climate zones. Highlights: Transfer learning based CNN-LSTM model is presented for thermal comfort modeling. It provides accurate thermal comfort prediction for buildings with limited data. Significant thermal comfort parameters are identified using Chi-square statistics. SMOTE oversampling technique is used to address the class imbalance problem. Impact of different thermal comfort parameters on the accuracy of model is studied.