Online Energy Management for a Sustainable Smart Home with an HVAC Load and Random Occupancy

In this paper, we investigate the problem of mini-mizing the summation of energy cost and thermal discomfort costin a long-term time horizon for a sustainable smart home withan HVAC (Heating, Ventilation, and Air Conditioning) load con-sidering uncertainties in electricity price, outdoor temperature,renewable generation output, electrical demand, most comfort-able temperature level, and home occupancy state. Specifically, wefirst formulate a stochastic program to minimize the time-averageexpected total cost (i.e., the summation of energy cost and thermaldiscomfort cost) by capturing constraints related to renewablegeneration, energy storage management, HVAC control, delayguarantee for charging an electric vehicle, power transactionsbetween the smart home and utility grid, and real-time powerbalancing. Typically, Lyapunov optimization techniques could beadopted to design an online energy management algorithm forthe above time-average optimization problem. The key idea of aLyapunov-based energy management algorithm is to buffer thedemand requests of devices in energy queues when electricityprices are high and serve requests when electricity prices arelow. However, different from loads with specific energy/powerrequirements (e.g., electric vehicle), estimating the power demandof an HVAC load in every time slot is very difficult since thedemand is related to many factors, such as the most comfortabletemperature level decided by occupants, the lower and upperbounds of indoor temperature, home occupancy state, and out-door temperature. Thus, Lyapunov optimization techniques couldnot be applied to our problem directly. Different from existingLyapunov-based energy management algorithms, we constructfour virtual queues associated with indoor temperature, EVcharging demand, EV charging delay, and energy storage control.By stabilizing such queues, we design an online energy man-agement algorithm without predicting any system parametersand submitting an HVAC power demand to an energy queue.Simulations based on real-world traces show that the proposedalgorithm could reduce energy cost effectively with small sacrificein thermal comfort.