TY - GEN
T1 - Energy demand prediction through novel random neural network predictor for large non-domestic buildings
AU - Ahmad, Jawad
AU - Larijani, Hadi
AU - Emmanuel, Rohinton
AU - Mannion, Mike
AU - Javed, Abbas
AU - Phillipson, Mark
N1 - Acceptance in SAN
Exception form sent 14/11/19 ET
PY - 2017/5/29
Y1 - 2017/5/29
N2 - Buildings are among the largest consumers of energy in the world. In developed countries, buildings currently consumes 40% of the total energy and 51% of total electricity consumption. Energy prediction is a key factor in reducing energy wastage. This paper presents and evaluates a novel RNN technique which is capable to predict energy utilization for a non-domestic large building comprising of 562 rooms. Initially, a model for the 562 rooms is developed using Integrated Environment Solutions Virtual Environment (IES-VE) software. The IES-VE model is simulated for one year and 10 essential data inputs i.e., air temperature, dry resultant temperature, internal gain, heating set point, cooling set point, plant profile, relative humidity, moisture content, heating plant sensible load, internal gain and number of people are measured. Datasets are generated from the measured data. RNN model is trained with this datasets for the energy demand prediction. Experiments are used to identify the accuracy of prediction. The results show that the proposed RNN based energy model achieves 0.00001 Mean Square Error (MSE) in just 86 epochs via Gradient Decent (GD) algorithm.
AB - Buildings are among the largest consumers of energy in the world. In developed countries, buildings currently consumes 40% of the total energy and 51% of total electricity consumption. Energy prediction is a key factor in reducing energy wastage. This paper presents and evaluates a novel RNN technique which is capable to predict energy utilization for a non-domestic large building comprising of 562 rooms. Initially, a model for the 562 rooms is developed using Integrated Environment Solutions Virtual Environment (IES-VE) software. The IES-VE model is simulated for one year and 10 essential data inputs i.e., air temperature, dry resultant temperature, internal gain, heating set point, cooling set point, plant profile, relative humidity, moisture content, heating plant sensible load, internal gain and number of people are measured. Datasets are generated from the measured data. RNN model is trained with this datasets for the energy demand prediction. Experiments are used to identify the accuracy of prediction. The results show that the proposed RNN based energy model achieves 0.00001 Mean Square Error (MSE) in just 86 epochs via Gradient Decent (GD) algorithm.
KW - IES-VE and building simulation
KW - non-domestic building
KW - random neural network
KW - energy demand prediction
KW - optimizations
U2 - 10.1109/SYSCON.2017.7934803
DO - 10.1109/SYSCON.2017.7934803
M3 - Conference contribution
SN - 9781509046225
BT - 2017 Annual IEEE International Systems Conference (SysCon)
PB - IEEE
CY - Montreal, QC, Canada
ER -