HVAC controller with regression model to help reduce energy consumption

• US 9,519,874 B2
• Filed: 08/30/2012
• Issued: 12/13/2016
• Est. Priority Date: 08/30/2012
• Status: Active Grant

• Current Assignee: Honeywell International Inc.
• Original Assignee: Honeywell International Inc.
• Inventors: Macek, Karel, Marik, Karel, Rojicek, Jiri
• Primary Examiner(s): Lo, Kenneth M
• Assistant Examiner(s): Chu, Alan
• Application Number:
  US13/599,748
• Publication Number:
  US 20140067132A1
• Time in Patent Office: 1,566 Days
• Field of Search: 700/276, 700/277, 700/278, 236/1.B, 236/1.C
• US Class Current: 1/1
• CPC Class Codes: F24F 11/30 for purposes related to the operation of the system, e.g. for safety or monitoring
• CPC Class Codes: F24F 11/47 Responding to energy costs
• CPC Class Codes: F24F 11/58 using Internet communication
• CPC Class Codes: F24F 11/62 characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
• CPC Class Codes: F24F 11/63 Electronic processing
• CPC Class Codes: F24F 11/64 using pre-stored data
• CPC Class Codes: F24F 2130/00 Control inputs relating to environmental factors not covered by group F24F2110/00
• CPC Class Codes: F24F 2130/10 Weather information or forecasts
• CPC Class Codes: G05D 23/1917 using digital means
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: G06Q 30/0283 Price estimation or determination
• CPC Class Codes: G06Q 50/06 Energy or water supply

Abstract

A thermal control system for a building is disclosed, which includes a regression model: Given a forecast temperature outside the building, the regression model predicts how much an HVAC system will cost to run during a day, for a given set of time-varying target temperatures for all the thermostats in the thermal control system. The thermal control system may also include an optimizer, which invokes multiple applications of the regression model. Given a forecast temperature outside the building, the optimizer predicts an optimal set of time-varying target temperatures for all the thermostats in the thermal control system. Running the HVAC system with the optimal set of time-varying target temperatures should have a reduced or a minimized cost, or a reduced or minimized total energy usage. The optimizer works by running the regression model repeatedly, while adjusting the time-varying target temperature for each thermostat between runs of the model.

Claims

1. A controller for an HVAC system of a building, the controller comprising:

an input for receiving a forecast of an ambient temperature outside of the building;

a memory for storing measures of past operational costs of the HVAC system in conjunction with past operating conditions and defined comfort limits within the building that change with time, wherein the past operational costs include past monetary costs of operating the HVAC system to maintain temperature setpoints within the defined comfort limits, and wherein the defined comfort limits include an upper comfort limit and a lower comfort limit that each change with time based on when the building is expected to be occupied, such that a region between the upper comfort limit and the lower comfort limit is narrower when the building is expected to be occupied than when the building is expected to be unoccupied;

an output for setting a set-point of the HVAC system;

a control unit coupled to the memory, the input and the output, the control unit including an optimizer for a regression model that uses a risk acceptance profile based on the past operational costs of the HVAC system to maintain the temperature setpoints of the HVAC system within the defined comfort limits, the forecast of the ambient temperature, the defined comfort limits, and at least some of the measures of the past operational costs of the HVAC system in conjunction with the past operating conditions to produce a temperature profile that includes a plurality of future temperature set-points for the HVAC system that are within the defined comfort limits, wherein each respective future temperature set-point is set to occur at a different predefined future time for a different area of the building, and the optimizer calculates the temperature set-point values at each different predefined future time, but the optimizer does not itself calculate the different predefined future times;

wherein the risk acceptance profile is defined by;

f(r)=μ

(r,c)+α

·

σ

(r,c), wherein;

f(r) is the risk acceptance profile;

r is a generalized temperature profile;

c is the past operating conditions;

μ

(r,c) is a mean value of a predicted operational cost of the HVAC system;

σ

(r,c) is a standard deviation of the predicted operational cost of the HVAC system; and

α

is a risk avoidance parameter;

wherein the control unit;

forwards the plurality of future temperature set-points to the HVAC system via the output of the controller; and

decreases a risk acceptance level of the risk acceptance profile as the regression model is refined; and

wherein the risk acceptance level is based on the risk avoidance parameter.

2. The controller of claim 1, wherein at least some of the plurality of predefined future times of the temperature profile correspond to fixed times during a day.

3. The controller of claim 1, wherein at least some of the plurality of predefined future times of the temperature profile correspond to an event.

4. The controller of claim 3, wherein the event corresponds to sunrise.

5. The controller of claim 3, wherein the event corresponds to sunset.

6. The controller of claim 3, wherein the event corresponds to a utility price change.

7. A controller for an HVAC system of a building, the controller comprising:

an input for receiving a forecast of an ambient condition outside of the building;

a memory for storing past monetary costs of operating the HVAC system in conjunction with past operating conditions and defined comfort limits within the building that change with time, wherein the defined comfort limits include an upper comfort limit and a lower comfort limit that each change with time based on when the building is expected to be occupied, such that a region between the upper comfort limit and the lower comfort limit is narrower when the building is expected to be occupied than when the building is expected to be unoccupied;

an output for setting a set-point of the HVAC system;

a control unit coupled to the memory, the input and the output, the control unit executing an optimizer for a regression model that uses a risk acceptance profile based on the past operational costs of the HVAC system to maintain temperature setpoints of the HVAC system within the defined comfort limits, the forecast of the ambient condition, the defined comfort limits, in conjunction with the past monetary costs of operating the HVAC system, to produce a profile that includes a plurality of future set-points for the HVAC system that are within the defined comfort limits, wherein each respective future set-point is set to occur at a different pre-defined future time for a different area of the building;

wherein the risk acceptance profile is defined by;

f(r)=μ

(r,c)+α

·

σ

(r,c), wherein;

f(r) is the risk acceptance profile;

r is a generalized temperature profile;

c is the past operating conditions;

μ

(r,c) is a mean value of a predicted operational cost of the HVAC system;

σ

(r,c) is a standard deviation of the predicted operational cost of the HVAC system; and

α

is a risk avoidance parameter;

wherein the plurality of calculated set-points are forwarded to the HVAC system via the output of the controller; and

wherein a risk acceptance level of the risk acceptance profile is decreased as the regression model is refined; and

wherein the risk acceptance level is based on the risk avoidance parameter.

8. The controller of claim 7, wherein the optimizer for the regression model calculates set-point values for each different predefined future time but the optimizer for the regression model does not itself calculate each different predefined future time.

9. The controller of claim 7, wherein the defined comfort limits change at discrete times, and at least some of the different predefined future times of the profile correspond to at least some of the discrete times of the defined comfort limits.

10. The controller of claim 7, wherein at least some of the different predefined future times of the profile correspond to fixed times during a day.

11. The controller of claim 7, wherein at least some of the different predefined future times of the profile correspond to an event.

12. The controller of claim 11, wherein the event corresponds to sunrise.

13. The controller of claim 11, wherein the event corresponds to sunset.

14. The controller of claim 11, wherein the event corresponds to a utility price change.

15. The controller of claim 7, wherein each of the plurality of set-points of the profile correspond to a temperature set-point that includes a set-point time and a set-point temperature, and wherein the input of the controller receives a forecasted temperature for each of the set-point times of the plurality of temperature set-points.

16. The controller of claim 7, wherein the optimizer for the regression model receives a measure related to the operational cost associated with execution of the profile, and uses the operational cost associated with past profiles, along with the ambient conditions that correspond to past profiles, to help produce the profile in order to help minimize expected operational cost associated with the profile.

17. The controller of claim 16, wherein the measure related to the operational cost associated with execution of the profile includes a percent of time that the HVAC system is in an “

on”

state during execution of the profile.

18. The controller of claim 16, wherein the regression model calculates a measure related to an expected operation cost of the profile, along with an uncertainty factor of the expected operational cost.

19. A method for controlling an HVAC system of a building, the method comprising:

obtaining a forecast of an ambient temperature outside of the building;

obtaining a measure of past operational cost of the HVAC system in conjunction with past operating conditions and defined comfort limits within the building that change with time, wherein the past operational cost includes a past monetary cost of operating the HVAC system, and wherein the defined comfort limits include an upper comfort limit and a lower comfort limit that each change with time based on when the building is expected to be occupied, such that a region between the upper comfort limit and the lower comfort limit is narrower when the building is expected to be occupied than when the building is expected to be unoccupied;

inputting the forecasted ambient temperature, the defined comfort limits, and the measure of the past operational cost of the HVAC system in conjunction with the past operating conditions to a regression model that uses a risk acceptance profile based on the past operational costs of the HVAC system to maintain temperature setpoints of the HVAC system within the defined comfort limits to produce a temperature profile that includes a plurality of future temperature set-points for the HVAC system that are within the defined comfort limits, wherein each respective future temperature set-point is set to occur at a different predefined future time for a different area of the building;

wherein the risk acceptance profile is defined by;

f(r)=μ

(r,c)+α

·

σ

(r,c), wherein;

f(r) is the risk acceptance profile;

r is a generalized temperature profile;

c is the past operating conditions;

μ

(r,c) is a mean value of a predicted operational cost of the HVAC system;

σ

(r,c) is a standard deviation of the predicted operational cost of the HVAC system; and

α

is a risk avoidance parameter;

outputting the plurality of future temperature set-points to the HVAC system; and

decreasing a risk acceptance level of the risk acceptance profile as the regression model is refined;

wherein the risk acceptance level is based on the risk avoidance parameter.

20. The method of claim 19, further comprising refining the regression model over time.