Optimising dimming systems for visual experience and energy consumption using genetic algorithms and neural networks

• An optimisation process that uses neural networks to rapidly simulate lighting performance is proposed. • This optimisation process extends traditional lighting control from horizontal planes to vertical surfaces. • It provides precise lighting control and has been validated in a realistic office scenario. • An innovative gaze-dependent lighting strategy is introduced and assessed, achieving a 45 % reduction in energy consumption. Occupancy-based lighting control systems have been widely used for decades to reduce energy consumption. With the increased use of machine learning, more techniques were proposed and developed to enhance lighting controls. However, certain limitations, such as the lack of consideration for vertical surfaces and uniformity, may negatively impact occupants’ visual experience. This research thus developed a lighting dimming optimisation method that integrates genetic algorithms and neural networks to enhance existing occupancy-based lighting control systems, while also supporting an innovative gaze-dependent lighting strategy. The proposed method can quickly provide optimised dimming solutions, minimising energy consumption while maintaining a satisfactory visual experience of the environment. The feasibility and accuracy of the optimisation process were tested in an office equipped with 147 luminaires and 85 target areas on horizontal and vertical surfaces. The gaze-dependent lighting strategy was further assessed in over 750 dimming scenarios to evaluate its potential to reduce energy consumption. Compared to the traditional non-dimming lighting strategy, the optimised dimming solutions reduced energy usage by more than 45 % on average.