Operating an HVAC system based on predicted indoor air temperature

1. A controller for a heating, ventilation, and air condition (“HVAC”) system, the controller comprising processing circuitry and a computer readable storage medium comprising instructions that, when executed by the processing circuitry, cause the controller to:

receive, at a current time, a setpoint and a time of interest indicating a time in the future when the setpoint is to be reached;

obtain a first data set comprising a plurality of lag values, the plurality of lag values associated with one or more variables related to the HVAC system, each lag value corresponding to a respective time within a lag time period, the lag time period beginning prior to the current time;

calculate a duration between the time of interest and the current time;

determine a lag value start time based on the calculated duration and the current time;

select a second data set comprising a subset of the lag values from the first data set, wherein the subset includes lag values for a time preceding the lag value start time and excludes each lag value corresponding to a variance inflation factor exceeding a variability threshold;

determine a predicted condition at the time of interest based at least in part on the lag values in the second data set;

determine a schedule for operating heating or cooling components of the HVAC system such that the setpoint is reached by the time of interest, the schedule determined at least in part based on the predicted condition at the time of interest; and

communicate one or more signals instructing the HVAC system to operate according to the schedule.

2. The controller of claim 1 , wherein the schedule includes a start time for switching from an unoccupied configuration to an occupied configuration, the start time determined at least in part based on the predicted condition at the time of interest.

3. The controller of claim 1 , wherein the subset of the lag values included in the second data set are dynamically selected at least in part based on the received time of interest.

4. The controller of claim 1 , wherein the controller is configured to perform an ordinary least squares (OLS) linear regression on the second data set.

5. The controller of claim 1 , wherein:

the controller is further configured to perform a 10-fold cross validation technique on the second data set.

6. The controller of claim 5 , wherein the 10-fold cross validation technique is an L1-Norm Regularized regression model.

7. The controller of claim 1 , wherein the lag time period is no more than 26 hours.

8. The controller of claim 1 , wherein the processing circuitry is deployed on a cloud platform.

9. The controller of claim 1 , wherein the one or variables associated with the HVAC system comprise one or more of:

an indoor temperature;

a cooling set point;

a heating set point;

a cooling demand;

a heating demand;

an indoor humidity;

a total capacity;

a sensible capacity;

a compressor frequency;

an airflow rate;

an input voltage;

an input current;

a simulated power; and

a measured power.

10. A method for a heating, ventilation, and air condition (“HVAC” system, the method comprising:

receiving, by a controller, a setpoint and a time of interest indicating a time in the future when the setpoint is to be reached;

obtaining, by the controller, a first data set comprising a plurality of lag values, the plurality of lag values associated with one or more variables related to the HVAC system, each lag value corresponding to a respective time within a lag time period, the lag time period beginning prior to the current time;

calculate a duration between the time of interest and the current time;

determine a lag value start time based on the calculated duration and the current time;

selecting, by the controller, a second data set comprising a subset of the lag values from the first data set, wherein the subset includes lag values for a time preceding the lag value start time and excludes each lag values corresponding to a variance inflation factor exceeding a variability threshold;

determining, by the controller, a predicted condition at the time of interest based at least in part on the lag values in the second data set;

determining, by the controller, a schedule for operating heating or cooling components of the HVAC system such that the setpoint is reached by the time of interest, the schedule determined at least in part based on the predicted condition at the time of interest; and

communicating, by the controller, one or more signals instructing the HVAC system to operate according to the schedule.

11. The method of claim 10 , wherein the schedule includes a start time for switching from an unoccupied configuration to an occupied configuration, the start time determined at least in part based on the predicted condition at the time of interest.

12. The method of claim 10 , wherein the subset of the lag values included in the second data set are dynamically selected at least in part based on the received time of interest.

13. The method of claim 10 , further comprising applying an ordinary least squares (OLS) linear regression to the second data set.

14. The method of claim 10 , further comprising applying a 10-fold cross validation technique to the second data set.

15. The method of claim 14 , wherein the 10-fold cross validation technique is an L1-Norm Regularized regression model.

16. The method of claim 10 , wherein the lag time period is no more than 26 hours.

17. The method of claim 10 , wherein the one or variables associated with the HVAC system comprise one or more of:

an indoor temperature;

a cooling set point;

a heating set point;

a cooling demand;

a heating demand;

an indoor humidity;

a total capacity;

a sensible capacity;

a compressor frequency;

an airflow rate;

an input voltage;

an input current;

a simulated power; and

a measured power.

18. A heating, ventilation, and air condition (“HVAC”) system operable to condition an enclosed space, the HVAC system comprising:

one or more heating components operable to heat the enclosed space when the HVAC system is in a heating mode;

one or more cooling components operable to cool the enclosed space when the HVAC system is in a cooling mode;

a controller comprising processing circuitry and a computer readable storage medium comprising instructions that, when executed by the processing circuitry, cause the controller to:

receive, at a current time, a setpoint and a time of interest indicating a time in the future when the setpoint is to be reached;

obtain a first data set comprising a plurality of lag values, the plurality of lag values associated with one or more variables related to the HVAC system, each lag value corresponding to a respective time within a lag time period, the lag time period beginning prior to the current time;

calculate a duration between the time of interest and the current time;

determine a lag value start time based on the calculated duration and the current time;

select a second data set comprising a subset of the lag values from the first data set, wherein the subset includes lag values for a time preceding the lag value start time and excludes each lag value corresponding to a variance inflation factor exceeding a variability threshold;

determine a predicted condition at the time of interest based at least in part on the lag values in the second data set;

determine a schedule for operating heating or cooling components of the HVAC system such that the setpoint is reached by the time of interest, the schedule determined at least in part based on the predicted condition at the time of interest; and

communicate one or more signals instructing the HVAC system to operate according to the schedule.

19. The HVAC system of claim 18 , wherein the schedule includes a start time for switching from an unoccupied configuration to an occupied configuration, the start time determined at least in part based on the predicted condition at the time of interest.

20. The HVAC system of claim 18 , wherein the subset of the lag values included in the second data set are dynamically selected at least in part based on the received time of interest.