Research On Glare Control Over Smart Window Blinds In Open-plan Offices Based On Machine Learning

The demand for improving the indoor daylight comfort performance of office buildings has prompted the academic community to pay attention to and reflect on the problem of indoor sunlight glare.The smart blinds with automatic adjustment function can prohibit the indoor glare,while the existing smart blinds glare control has a bottleneck in the face of complex glare scenes.Today,in the context of the era of artificial intelligence,the self-adaptive regulation of smart blinds assisted by machine learning has the potential to rapidly predict information and coordinate optimization of multi-objectives in the face of complex glare scenes in indoor offices,and can effecti vely reduce the risk of indoor glare in building office spaces.Based on the above background,this research takes "machine learning-driven adaptive glare control of smart blinds in office space" as the research proposition,and conducts theoretical sortin g,problem analysis,experimental analysis and strategy summary for the key content of its control process.Provide scientific basis and technical support for process formulation,control development and design application.According to the research route above,the research firstly discusses the three key process in building environmental control based on the theory of system science.Taking this as a theoretical framework,the three sub-processes in the glare control of smart shading blinds are clarified,and the research topic of each process are pointed out,namely: the confirmation of environmental dynamic information that affects indoor glare,the selection of mapping technology for predictive model,and the optimization mechanism for considering the b alance between the performance objectives.Combining machine learning technology with daylighting simulation technology to build an experimental platform for scientific research on these three key research topics.Secondly,the study analyzes the influence of different environmental dynamic information on the indoor glare of open office space.With the feature selection technique in machine learning,the correlation of different dynamic info against glare indicators,the of information collinearity,and the experimental analysis of the compressed variables are carried out,respectively.According to the experimental results,the selection strategies for environmental dynamic information are proposed.T hen,the glare data prediction ability of feasible mappin g technologies based on machine learning are evaluated.With the help glare mapping data of open office space,the experimental analysis is conducted in terms of prediction accuracy and modeling timeliness of different machine learning mapping technologies.The strategies for mapping technology selection are summarized according to the experimental results.Afterwards,a target optimization mechanism that can coordinate glare and other targets to balance and solve quickly is established.Based on the basic control objectives,the synergistic mechanism of the composite performance objectives is designed using the parameterization method.Using the target optimization task in the smart shading control,the optimization performance comparison is carried out among the optimization technology available for the target optimization mechanism,in terms of extreme value exploration depth,extreme value solution rate,and search stability is carried out.According to the experimental results,the strategies for establi shing the target optimization mechanism are summarized.Finally,an application test is conducted,attempting to verify research results and strategies of the above three key research topics,a feasible glare control is formulated for the smart shading bli nds in an open office space of the architectural design practice project.It turns out that the control strategies in the study outperformed other commonly used control strategies through parallel comparison in simulation experiments.In terms of the innovation points,the research proposed screening method for environmental dynamic information,revealing the quantitative relationship between different environmental dynamic information and indoor glare indicators.The research discovers a machine learning mapping technology suitable for solving complex indoor glare prediction in office spaces,assisting in the effective construction of glare prediction models,empowering the smart blinds application.A multi-objective optimization control mechanism for office space is proposed,which solves the problem of tradeoff between office space glare optimization and other daylighting-related targets.Its performance in complex glare control tasks can improve the control efficiency,as well as increasing the contr ol decisionmaking accuracy.The research can improve the intelligent level of blinds control from the perspective of building science.This research integrates the building environmental control theory,performance-driven design thinking and machine learning related algorithms,realizes the integration of multi-disciplinary in multiknowledge field,and expands the application scope of smart shading and glare control in architecture.