Conference Paper

Proceedings of Building Simulation 2019: 16th Conference of IBPSA

     

Artificial Neural Network Based Model Predictive Control Vis-à-Vis Simple On-Off Control Of Windows Opening Position For Mixed-Mode-Operated Building

Brijesh Pandey 1, Rajat Pungaliya 2, Rangan Banerjee 1
1 Department of Energy Science and Engineering, IIT Bombay
2 VIT University, Vellore

DOI: https://doi.org/10.26868/25222708.2019.211065
Abstract: This paper presents an artificial neural network (ANN) based model predictive control to implement the mixedmode strategy in buildings. Simple heuristics are unable to incorporate the occupant interaction with the built environment whereas; significant efforts are required to configure and calibrate the model used in the model predictive controller. To address the above issues an ANN-based predictive controller is developed, and used within EnergyPlus-Matlab co-simulation framework with particle swarm optimization (PSO) as an optimizer. Simulations are performed for the five consecutive days for January with mixed-mode control strategy. The results show significant cooling energy savings compared to the rule-based heuristic.
Keywords: Mixed-mode, Thermal Comfort, Model Predictive Controller, Artificial Neural Network, Co-simulation
Pages: 4918 - 4925
Paper:

BS2019_211065