Adaptive thermal comfort learning for optimized HVAC control
A facility providing systems and methods for managing and optimizing thermal comfort is provided. The facility is a software algorithm that intelligently detects the thermal comfort preferences of residential smart thermostat users, where a user's “comfort preference” is defined to refer to an estimate of a measurement of the user's comfort across varying values of some set of exogenous factors, including but not limited to indoor temperature, the time of day, the day of the week, and weather conditions. This facility encompasses the use of this “comfort preference” for the creation of an optimal schedule of setpoints for a residential thermostat which is configured to optimize some objective, including potentially user comfort, energy efficiency, load shift, or cost.
1. A method for controlling a heating, ventilation, and air conditioning (HVAC) system, the method comprising:
constructing a comfort surface data structure, wherein the comfort surface data structure is defined by a plurality of comfort surface points;
for each of a plurality of time periods,
combining historical data and exogenous data, and
for each of one or more historical data points in the combined data,
mapping the historical data point in the combined data to a corresponding comfort surface point in the comfort surface data structure, and
for each of one or more comfort surface points in the comfort surface data structure,
calculating a proximity between the mapped historical data point and the one or more comfort surface points, and
updating at least one value for the one or more comfort surface points based on the calculated proximity to provide an updated comfort surface data structure;
generating a plan for controlling a thermostat based on the updated comfort surface data structure; and
controlling the thermostat for operation of the HVAC system in accordance with the generated plan.
2. The method of claim 1 wherein the comfort surface data structure is configured to determine a user's comfort at a plurality of combinations of internal temperature data and the exogenous data.
3. The method of claim 1 wherein updating the at least one value for the comfort surface point in the comfort surface data structure is based on a comfort measurement recorded at the historical data point.
4. The method of claim 1 wherein generating the plan for controlling the thermostat includes defining a comfort risk tolerance based on the updated comfort surface, wherein the comfort risk tolerance determines a likelihood one or more users will be comfortable in a temperature range.
5. The method of claim 4 wherein generating the plan for controlling the thermostat further includes:
defining one or more comfort constraints based on the comfort risk tolerance, wherein the comfort constraints include the temperature range and one or more exogenous constraints; and
generating a schedule at least partially based on the defined one or more comfort constraints.
6. The method of claim 1 wherein generating the plan for controlling the thermostat includes:
identifying one or more objectives, wherein the objectives are related to one or more of user comfort, cost savings, energy efficiency, and load shift; and
generating a schedule at least partially based on the identified one or more objectives.
7. The method of claim 1 wherein generating the plan for controlling the thermostat includes:
identifying a first comfort preference range based on the updated comfort surface, wherein the first comfort preference range is associated with a temperature range;
identifying a second comfort preference range to test comfort limits outside of the temperature range; and
generating a schedule at least partially based on the identified second comfort preference range.
8. The method of claim 1 wherein each of the plurality of comfort surface points is associated with at least one entry comprising one or more time attributes, one or more thermostat settings, one or more environmental attributes, and one or more comfort attributes.
9. The method of claim 1 , further comprising actively prompting user interactions with the thermostat through control of the HVAC system to cause temperature changes.
10. The method of claim 9 wherein actively prompting the user interactions includes causing changes in accordance with cost savings, energy efficiency, load shift, or other defined objectives.
11. The method of claim 1 , further comprising:
receiving one or more new comfort data points and new exogenous data related to one or more time periods in the generated plan;
combining the one or more new comfort data points and the new exogenous data; and
for each of one or more new comfort data points,
mapping the new comfort data point to a corresponding comfort surface point in the updated comfort surface data structure, and
for each of one or more comfort surface points in the updated comfort surface data structure,
calculating a second proximity between the mapped new comfort data point and the one or more comfort surface points, and
updating at least one value for the one or more comfort surface points in the updated comfort surface data structure based on the calculated proximity to provide further updates to the updated comfort surface data structure.
12. A method, performed by a computing system, for controlling a heating, ventilation, and air conditioning (HVAC) system, the method comprising:
constructing, by the computing system, a comfort surface data structure defined by a plurality of comfort surface points;
for each of a plurality of time periods,
combining historical data and exogenous data, and
for each of one or more historical data points in the combined data,
mapping the historical data point in the combined data to a corresponding comfort surface point in the comfort surface data structure, and
for each of one or more comfort surface points in the comfort surface data structure,
calculating a proximity between the mapped historical data point and the one or more comfort surface points, and
updating at least one value for the one or more comfort surface points based on the calculated proximity to generate an updated comfort surface data structure;
generating, by the computing system, a schedule for controlling a thermostat based on the updated comfort surface data structure; and
controlling the thermostat for operation of the HVAC system in accordance with the generated schedule.
13. The method of claim 12 wherein the schedule includes a plurality of control periods, and wherein the method further comprises:
for each control period in the schedule, receiving a comfort data point associated with comfort levels during the control period; and
for each comfort data point received, further updating the updated comfort surface data structure.
14. The method of claim 13 , wherein further updating the updated comfort surface data structure includes:
combining the received comfort data point with exogenous data based on timestamps associated with the control period;
mapping the received comfort data point to a corresponding comfort surface point in the updated comfort surface data structure, and
for each of one or more comfort surface points in the updated comfort surface data structure,
calculating a proximity between the mapped received comfort data point and the one or more comfort surface points in the updated comfort surface data structure, and
updating at least one value for the one or more comfort surface points based on the calculated proximity.
15. The method of claim 12 wherein the comfort surface data structure is a continuous data structure or a discrete data structure.
16. The method of claim 12 wherein generating the schedule includes:
identifying one or more objectives;
defining, by the computing system, one or more comfort constraints based on the updated comfort surface data structure, wherein the comfort constraints comprise a determined comfort preference range between a first temperature and a second temperature; and
generating, by the computing system, a plan for each time period in the schedule based on the defined one or more comfort constraints and the identified one or more objectives.
17. The method of claim 12 , further comprising actively prompting, by the computing system, user interactions with the thermostat through control of the HVAC system to cause temperature changes during the schedule, wherein the changes are associated with one or more objectives.
18. A non-transitory computer-readable storage medium storing instructions that, if executed by a computing system, cause the computing system to perform a method for controlling a heating, ventilation, and air conditioning (HVAC) system, the method comprising:
for each of a plurality of time periods,
retrieving historical data and exogenous data,
quantifying historical comfort levels,
combining the historical data and the exogenous data based on timestamps, and
for each of one or more historical data points in the combined data,
mapping the historical data point in the combined data to a corresponding comfort surface point in a comfort surface data structure, wherein the comfort surface data structure, is defined by a plurality of comfort surface points, and
for each of one or more comfort surface points in the comfort surface data structure,
calculating a proximity between the mapped historical data point and the one or more comfort surface points, and
updating at least one value for the one or more comfort surface points based on the calculated proximity to generate an updated comfort surface data structure;
generating a schedule for controlling a thermostat based on the updated comfort surface data structure; and
controlling the thermostat for operation of the HVAC system in accordance with the generated schedule.
19. The non-transitory computer-readable storage medium of claim 18 wherein the method further comprises actively prompting user interactions with the thermostat through control of the HVAC system to cause temperature changes as a function of cost savings, energy efficiency, load shift, or other user-defined objectives.