Formulation and application of an economic model predictive control scheme for connected thermostats.

An economic model predictive control (MPC) scheme, which is an MPC scheme equipped with an economicallyoriented objective function, is developed for a connected thermostat application. The economic MPC selects a setpoint from an occupant defined comfort range to minimize the HVAC power cost. Specifically, under a time-varying electric rate structure (e.g., time-of-use or real-time pricing), the economic MPC leverages the building mass as thermal energy storage to shift HVAC power consumption from high to low cost periods. The resulting economic MPC system includes a parameterized building thermal zone model, a parameter estimation procedure to identify the model parameters for a specific zone application, a state/disturbance estimator, a heat load disturbance forecaster, and an underlying optimal control problem formulation. Each of these features are tailored for broad application as a supervisory controller manipulating the zone temperature setpoint for a zone controlled by a thermostat. Given a lack of measurements available to estimate/measure HVAC power or load in a typical thermostat, the HVAC load is approximated via a filtered version of the thermostat equipment stage commands, which provides a normalized time-averaged approximation of the HVAC load. Simulation results are presented to demonstrate the effectiveness of the strategy.