Occupancy-based Adaptive HVAC Control for Energy Efficient Residential Buildings

根据加拿大统计局的数据，家庭仍然是最大的能源使用者，占全国能源使用量的24.0%。供暖、通风和空调(HVAC)已被列为气体排放的最主要贡献者。该项目旨在增强住宅暖通空调系统，以基于智能占有率检测/识别来优化能耗。虽然已经提出了几种技术/传感器来应对这一挑战，但成本、精度以及与现有建筑自动化基础设施的无缝集成仍然是主要挑战，特别是对于成本和复杂性是关键因素的家庭来说。最近，现代家庭中负担得起的传感器的数量和类型都有了显著的增加。例如，太阳辐射传感器被用来检测房间的照明状况。红外和超声波传感器被用作接近传感器和运动传感器。重量传感器在几个家庭中变得普遍。一些先进的暖通空调系统提供二氧化碳水平。这些传感器的可用性创造了一个机会，可以开发具有成本效益的先进入住率检测和识别技术，以实现优化的适应性住宅暖通空调系统。拟议的项目将采用传感器融合的方法来解决这个问题。在一种可能的情况下，基于雷达和硅MEMS的麦克风传感器可以用来绘制入住率地图。雷达和硅MEMS麦克风传感器通过音频波束形成和雷达目标存在检测相结合的方式提供远场声波捕获。随着其他传感器数量的增加，人工智能和机器学习可以应用于传感器融合。拟议的系统将分为两个阶段：培训阶段和识别阶段。在培训阶段，将收集入住率和身份数据以及可用的传感器测量数据。数据将使用隐马尔可夫模型进行建模。在识别阶段，首先使用可用的传感器测量来检测入住率。然后，基于雷达/MEMS的麦克风将被用于通过语音识别来识别乘客。

Per Statistics Canada, households continued to be the largest users of energy accounting for 24.0% of nationalenergy use. Heating, ventilation, and air-conditioning (HVAC) has been singled out as the most importantcontributor to gas emissions. This project aims at enhancing residential HVAC systems to optimize energyconsumption based on intelligent occupancy detection/identification. While several technologies/sensors havebeen proposed to address this challenge, the cost, accuracy, and the seamless integration with existing buildingautomation infrastructures remain major challenges especially for homes where cost and complexity are keyfactors. Recently, there is a noticeable increase in number and types of affordable sensors in modern homes.For example, solar radiation sensors are used to detect room lightening condition. Infrared and ultrasoundsensors are used as proximity and motion sensors. Weight sensors are becoming common in several houses.Some advanced HVAC systems provides level of carbon dioxide. The availability of these sensors creates anopportunity to develop cost-effective advanced occupancy detection and identification technology to enableoptimized adaptive residential HVAC systems. The proposed project will adopt a sensor fusion approach totackle the problem. In one possible scenario, radar and silicon MEMS-based microphone sensors may be usedto develop occupancy maps. Radar and silicon MEMS-based microphone sensors provide far field soundwaves capture by audio beam-forming combined with radar target presence detection. With the increasingnumber of other available sensors, artificial intelligence and machine learning can be used to apply sensorfusion. The proposed system will have two phases; training phase and identification phase. In training phase,occupancy and identity data will be collected along with available sensor measurements. Data will be modeledusing a Hidden Markov Model. In identification phase, first, the available sensor measurements will be used todetect occupancy. Then, the radar/MEMS-based microphone can be will be used to identify the occupantsusing voice recognition.