Abstract
A large amount of energy consumed globally is done by buildings, also, buildings are responsible for a great portion of greenhouse gas emissions. With progress in smart sensors and devices, a new generation of smarter and more context-aware building controllers can be developed. Consequently, zone-level surrogate artificial neural networks are used herein, where indoor temperature, occupancy, and weather data are the inputs. A new metaheuristic optimization algorithm, called Chaotic Satin Bowerbird Optimization Algorithm (CSBOA) is employed for the minimization of energy consumption. 24-hour schedules of the heating setpoint of each zone are created for an office building located in Edinburgh, Scotland. Two modes of optimization including day-ahead and model predictive control are applied for each hour. The consumption of energy decreased by 26% during a test week in Feb in comparison to the base case approach of heating. By definition of a time-of-use tariff, the cost of energy consumption is decreased by around 28%.