A daylight-linked shading strategy for automated blinds based on model-based control and Radial Basis Function (RBF) optimization. (15th June 2020)

• Record Type: Journal Article
• Authors: Luo, Zhaoyang; Sun, Cheng; Dong, Qi
• Is Part Of: Building and environment. Volume 177(2020)
• Publication Date: 2020-06-15
• Subjects: Automated shading -- Model-based control -- Daylight simulation -- Radial basis Function -- Surrogate model -- Supervised learning; Buildings -- Environmental engineering -- Periodicals; Building -- Research -- Periodicals; Constructions -- Technique de l'environnement -- Périodiques; Electronic journals; 696
• DOI: 10.1016/j.buildenv.2020.106854
• Languages: English
• ISSNs: 0360-1323

Abstract

Abstract: Addressing both daylight maximization and glare control over the entire workplace is always challenging for developing the automated shading control system. For the sake of cost and space usage, it is impractical to mount multiple sensors or cameras for real-time daylight environment monitoring to guarantee the control precision. Cut-off control is popular while it cannot attenuate the glare caused by excessive diffuse daylight. This paper introduces a model-based shading control for predetermining shading positions at each time step. A Useful Daylight Illuminance paradigm modality called rUDI is proposed as a variable criterion added to assist the cut-off strategy for further eliminating glare. The controller could be developed through real-time daylight simulations and an optimizer based on the surrogate model. This method was implemented in a full-scale office in Harbin, China. The surrogate model grounded on the Radial Basis Function Neural Network (RBF) was trained, validated and test with the experimental data sets. The control strategy was further incorporated with an adaptive light-switch model. The comparative simulations were conducted, and their corresponding results were generated for evaluating their performance in visual comfort, daylighting and electrical energy savings, demonstrating the advantages of the proposed control approach in terms of its adequate performance. Highlights: A simplified controller with better daylighting performance based on simulation modelling is proposed for automated blinds. A new metric concerning the dynamic illuminance distribution is proposed for automated blinds controller. An optimizer derived from a machine learning algorithm is proposed for achieving predictive control. Optimizing the indoor horizontal illuminance can effectively alleviate the glare caused by the intense diffuse daylight.