Predictive building control system with discomfort threshold adjustment

1. A method for controlling HVAC equipment for a building, comprising:

generating, based on historical building data, a discomfort tolerance defining an acceptable amount of occupant discomfort;

determining a first value of an environmental condition by predicting that the occupant discomfort would exceed the discomfort tolerance in a first direction at the first value of the environmental condition;

determining a second value of the environmental condition by predicting that the occupant discomfort would exceed the discomfort tolerance in a second direction opposite the first direction at the second value of the environmental condition; and

operating the HVAC equipment to maintain the environmental condition between the first value and the second value.

2. The method of claim 1 , further comprising widening or narrowing the discomfort tolerance based on a number of user overrides of a setpoint for the environmental condition during the controlling the HVAC equipment.

3. The method of claim 1 , wherein the environmental condition is indoor air temperature, the first value of the environmental condition is a first indoor air temperature, and the second value of the environmental condition is a second indoor air temperature, and wherein the determining steps are performed using a neural network trained to predict the occupant discomfort for a plurality of indoor air temperatures comprising the first indoor air temperature and the second indoor air temperature.

4. The method of claim 1 , wherein the occupant discomfort is quantified as an expected size of a setpoint change by an occupant.

5. The method of claim 1 , wherein generating the discomfort tolerance comprises determining that a preset percentage of setpoint changes occur outside of the discomfort tolerance.

6. The method of claim 1 , wherein generating the discomfort tolerance comprises using a neural network model trained using the historical building data to determine a maximum acceptable predicted setpoint increase and a maximum acceptable predicted setpoint decrease defining the discomfort tolerance.

7. The method of claim 1 , further comprising:

collecting new building data while controlling the HVAC equipment; and

updating the discomfort tolerance based on a combination of the new building data and the historical building data.

8. The method of claim 7 , comprising aggregating the combination of the new building data and the historical building data using a binning process.

9. The method of claim 8 , wherein the binning process comprises:

defining a plurality of bins configured to retain a fixed number of data sets;

sorting the historical building data into the plurality of bins;

sorting the new building data into the plurality of bins; and

discarding a subset of the historical building data from a bin of the plurality of bins in response to exceeding the fixed number of data sets for the bin.

10. The method of claim 8 , wherein the binning process comprises sorting data sets into a plurality of bins based on measurements of the environmental condition indicated in the data sets.

11. A building system, comprising:

HVAC equipment operable to affect one or more environmental conditions; and

circuitry configured to:

generate, based on historical building data, a discomfort tolerance defining an acceptable amount of occupant discomfort and satisfying or optimizing a predefined criterion;

determine a region of values of the one or more environmental conditions by predicting that the occupant discomfort would be within the discomfort tolerance when the one or more environmental conditions are within the region; and

control the HVAC equipment to maintain the one or more environmental conditions within the region.

12. The building system of claim 11 , wherein the circuitry is further configured to widen or narrow the discomfort tolerance based on a number of user overrides of one or more setpoint for the one or more environmental conditions during the control of the HVAC equipment.

13. The building system of claim 11 , wherein the circuitry is configured to determine the region using a neural network trained to quantify the occupant discomfort across a test region of values of the one or more environmental conditions, the test region comprising the region.

14. The building system of claim 11 , wherein the occupant discomfort is quantified as an expected size of a setpoint change by an occupant.

15. The building system of claim 11 , wherein the circuitry is configured to generate the discomfort tolerance by determining that a preset percentage of occupant overrides occur outside of the discomfort tolerance.

16. The building system of claim 11 , wherein the circuitry is configured to generate the discomfort tolerance using a neural network model trained using the historical building data to determine a maximum acceptable predicted setpoint increase and a maximum acceptable predicted setpoint decrease defining the discomfort tolerance.

17. The building system of claim 11 , wherein the circuitry is further configured to:

collect new building data during the control of the HVAC equipment; and

update the discomfort tolerance based on a combination of the new building data and the historical building data.

18. The building system of claim 17 , wherein the circuitry is further configured to aggregate the combination of the new building data and the historical building data using a binning process.

19. The building system of claim 18 , wherein the binning process comprises sorting data sets into a plurality of bins based on measurements of the one or more environmental conditions indicated in the data sets.

20. One or more non-transitory computer-readable media storing program instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

generating, based on historical building data, a discomfort tolerance defining an acceptable amount of occupant discomfort;

determining a first value of an environmental condition by predicting that the occupant discomfort would cross a first boundary of the discomfort tolerance at the first value of the environmental condition;

determining a second value of the environmental condition by predicting that the occupant discomfort would cross a second boundary of the discomfort tolerance at the second value of the environmental condition;

operating HVAC equipment to maintain the environmental condition between the first value and the second value.

21. The one or more non-transitory computer-readable media of claim 20 , wherein the environmental condition is indoor air temperature, the first value of the environmental condition is a first indoor air temperature, and the second value of the environmental condition is a second indoor air temperature, and the determining operations are performed using a neural network trained to predict the occupant discomfort across a plurality of indoor air temperatures comprising the first value of the indoor air temperature and the second value of the indoor air temperature.

22. The one or more non-transitory computer-readable media of claim 20 , wherein generating the discomfort tolerance comprises determining that a preset percentage of setpoint changes occur outside of the discomfort threshold.