Smart buildings generally adopt a centralized optimal control for heating, ventilation and air-conditioning (HVAC) systems to improve the building system performance. As a crucial technology adopted in smart buildings, Internet of Things (IoT) based wireless sensor networks (WSNs) are promising platforms for implementing novel distributed optimal control to achieve distributed intelligence effectively. Such a future mist computing paradigm can be hindered by the limited energy resource of WSNs. The event-driven optimization method activates the optimization only when events occur. It can save the energy resource of WSNs, and thus pave the way for implementing the distributed optimal control architecture in IoT-based WSNs. This study therefore proposes an event-driven multi-agent based distributed optimal control strategy for HVAC systems for implementation in IoT-based battery-powered WSNs. The strategy consists of two novel schemes. First, an event determination scheme determines the event threshold by comprehensively considering the system performance and the total energy consumption of individual sensors implementing the distributed optimal control architecture. Second, an event-driven distributed optimization scheme solves the optimization problems with distributed optimization algorithms in IoT sensors of limited data processing capacity when an event occurs. Comparison and case studies are conducted to compare different strategies and validate the proposed strategy. Results show that different strategies require very different sensor energy consumption. The proposed strategy can provide satisfactory system performance while reducing the energy consumption of IoT sensors.

智能建筑通常对采暖、通风和空调（HVAC）系统采用集中优化控制，以提高建筑系统性能。作为智能建筑中采用的一项关键技术，基于物联网 (IoT) 的无线传感器网络 (WSN) 是实现新型分布式优化控制以有效实现分布式智能的有前途的平台。这种未来的薄雾计算范式可能会受到 WSN 有限能源的阻碍。事件驱动优化方法仅在事件发生时激活优化。它可以节省无线传感器网络的能源资源，从而为在基于物联网的无线传感器网络中实现分布式优化控制架构铺平道路。因此，本研究为 HVAC 系统提出了一种基于事件驱动的多代理分布式优化控制策略，以在基于物联网的电池供电 WSN 中实施。该策略由两个新颖的方案组成。首先，事件确定方案通过综合考虑系统性能和实现分布式优化控制架构的单个传感器的总能耗来确定事件阈值。其次，事件驱动的分布式优化方案解决了在事件发生时数据处理能力有限的物联网传感器中分布式优化算法的优化问题。进行比较和案例研究以比较不同的策略并验证所提出的策略。结果表明，不同的策略需要非常不同的传感器能​​耗。