Machine learning for residential building HVAC analytics platform

住宅建筑空间条件占加拿大家庭平均能源消耗的64%。这些系统的优化和改进与商业应用中的建筑物和日益互联的家庭相比进展甚微。因此，有很大的机会来改善这些系统的运行，并为房主提供节省能源和金钱的机会。目前的项目将使用相对较新的、未被触及的数据，随着联网恒温器的出现，这些数据已经成为主流。现在，第一次真正的，长期的操作，互动和性能数据可以收集在个人家庭层面。这些数据将允许为单个家庭提供更先进的模型，控制，监控和建议。 为了继续进行拟议中的项目，两名具有互补技能和经验的领先研究人员与加拿大的ecobee公司合作。任期五年该行业合作伙伴是住宅连接恒温器领域的市场领导者。他们有能力和兴趣学习和整合研究成果，并将其开发成可销售的产品，为客户提供更高的价值。Ecobee公司承诺为该项目提供大量资源，包括：技术咨询/支持、工程服务、硬件、计算协助和资金。从生态蜂种群现场操作单元收集的数据以及高质量建模沿着，为设计的控制和机器学习产品提供了测试平台。研究的结果是系统能够从其环境（系统和占用者）中学习，并提供改进的控制，对效率低下的建议和诊断性能异常。这些都将提高能源效率和潜在的舒适度的居住者。除了帮助加拿大实现其温室气体排放目标和减少对额外能源基础设施的需求外，开发的建筑控制将有助于行业合作伙伴在加拿大和国外保持竞争力。

Residential building space conditions makes up 64% of the energy consumption of the average Canadian home. The optimization and improvement of these systems has progressed little in comparison to the buildings in commercial applications and the ever more connected home. As such there is a substantial opportunity to improve these systems in their operation and present homeowners opportunity to save both energy and money. The current project will use the relatively new, and untouched, data that has became more mainstream with advent of the connected thermostat. Now, for the first time really, long term operation, interaction and performance data can be collected at the individual home level. This data will allow for more advanced models, controls, monitoring and recommendations to be made for an individual home. To proceed with the proposed project, two leading researchers with complementary skills and experiences with collaborate with Canada-based ecobee inc. for a five-year term. The industry partner is a market leader in the residential connected thermostat space. They have the capacity and interest to learn from and potentially integrate the research results and develop them into marketable products and provide their customers increased value. Ecobee inc has committed to contributing significant resources to the project, including: technical advice/support, engineering services, hardware, computational assistance, and monetary funding. Data collected from operational units in the field across the ecobee population along with high quality modelling with provide the testbed for designed controls and machine learning products. The outcome of the research is systems able to learn from its environment (both system and occupants) and provide improved controls, recommendations on inefficiencies and diagnose anomalies in the performance. Each of these will increase the energy efficiency and potentially comfort of the occupants. Aside from helping Canada achieve its objectives in greenhouse gas emissions and reducing the need for additional energy infrastructure, the developed building controls will help the industry partner remain competitive within Canada and abroad.