CyberSEES: Type 2: Human-centered systems for cyber-enabled sustainable buildings

在美国，建筑业约占一次能源使用量的40%，电力的71%和二氧化碳排放量的38%。因此，开发有效的解决方案以减少建筑物的能源消耗和环境影响具有至关重要的社会重要性。居住者在办公建筑的能源使用中起着重要作用，影响高达30%的能源使用。为了管理由于居住者的存在和行为造成的居住者的能量影响，环境控制系统（例如，HVAC、遮阳、照明）已经基于使用“广泛接受的”视觉和热舒适度指标实现了自动化。然而，居住者对定制的室内气候有强烈的偏好，并且居住者对其环境的控制感与生产力、健康和幸福感之间有很强的关系。为了应对环境定制控制的挑战，该项目的目标是实现以人为本的可持续建筑的新范式，通过概率方法和机器学习中的计算创新进行研究，与可持续性相关，并在多个科学和工程领域产生更广泛的影响。更广泛的影响是：（1）有关概率分类、推断及最佳控制的新计算方法及算法，可能会影响多个科学界，包括建筑、机械、电机、计算机及工业/人为因素工程、计算机及心理科学。潜在应用领域包括基因组学、交通流量预测、包括电力、交通等在内的基础设施系统（2）将项目的建模、仿真和实验平台整合到新的教学模块和体验式学习活动中，支持四所工程学院的课程和劳动力发展（土木，机械，电气和计算机，工业工程）和心理科学系。(3)通过出版物、讲习班、会议和定制的外部评价程序，向学术界和工业界传播研究成果。(4)为K-12教师和学生在STEM学习和研究中创建外展和参与计划。该项目采用多学科方法，基于（1）自动识别建筑物相关人类感知属性的新算法;以及（2）智能和自调整舒适度交付系统的新概念，用于建筑物中的定制热和视觉环境。研究内容包括：（1）实验室和现场研究，以人类测试对象绘制室内环境条件，热和视觉感知和舒适度，居住者与建筑物的相互作用和控制行动，以及周边建筑区域的相应空间性能。(2)典型人群的人类感知、舒适度和满意度概况的概率分类。(3)用于在线学习个人和群体水平人类偏好的计算高效推理算法。(4)最佳控制算法和模拟工具，用于建筑物管理系统的实施。研究成果将被整合到一个新的网络技术解决方案中，用于自调整舒适度传递设备（恒温器，遮阳和照明执行器）。实验原型和现场演示将通过量化的建筑能源使用减少和居住者满意度来实现更好的性能，以及由于覆盖频率降低而对不确定性的鲁棒性。

In the U.S., the building sector accounts for about 40% of primary energy usage, 71% of electricity and 38% of carbon dioxide emissions. For this reason, development of efficient solutions to reduce energy consumption and environmental impact of buildings is of critical societal importance. Occupants play a significant role in energy use of office buildings, affecting up to 30% of the energy use. To manage occupants' energy impact due to their presence and behavior, environmental control systems (e.g., HVAC, shading, lighting) have been automated based on the use of "widely acceptable" visual and thermal comfort metrics. However, occupants have a strong preference for customized indoor climate, and there is a strong relationship between occupants' perception of control over their environment and productivity, health and well-being. To address the challenge of customized control of the environment, the objective of this project is to realize a new paradigm for human-centered sustainable buildings, enabled by conducting research with computing innovations in probabilistic methods and machine learning, linked to sustainability, and with broader impacts in multiple domains of science and engineering. Broader impacts are: (1) New computing methods and algorithms on probabilistic classification, inference and optimal control that may impact a number of scientific communities, including Architectural, Mechanical, Electrical, Computer and Industrial/Human Factors Engineering, Computer and Psychological Sciences. Potential application areas include genomics, traffic flow prediction, infrastructure systems including power, transportation, etc. (2) Integration of the project's modeling, simulation, and experimental platforms into new teaching modules and experiential learning activities that support the curriculum and workforce development in four engineering schools (Civil, Mechanical, Electrical and Computer, Industrial Engineering) and the Department of Psychological Sciences. (3) Dissemination of research outcomes to the academic community and to the industry through publications, workshops, conferences and a customized external evaluation process. (4) Creation of outreach and engagement initiatives for K-12 teachers and students in STEM learning and research. This project takes a multidisciplinary approach that is grounded in (1) new algorithms for automated identification of the relevant human perception-attributes of buildings; and (2) new concepts for intelligent and self-tuned comfort delivery systems for customized thermal and visual environments in buildings. The research includes: (1) Laboratory and field studies with human test-subjects that map indoor environment conditions, thermal and visual perception and comfort, occupant-building interactions and control actions, as well as corresponding space performance for perimeter building zones. (2) Probabilistic classification of human perception, comfort, and satisfaction profiles for a typical population. (3) Computationally-efficient inference algorithms for online learning of individual and population-level human preferences. (4) Optimal control algorithms and simulation tools for implementation in building management systems. The research outcomes will be integrated into a new cyber-enabled technological solution for self-tuned comfort delivery devices (thermostats, shading and lighting actuators). The experimental prototypes and field demonstrations will achieve improved performance with quantified building energy use reduction and occupant satisfaction, as well as robustness to uncertainty due to the reduction in the frequency of overrides.