Influence learning for managing physical conditions of an environmentally controlled space by utilizing a calibration override which constrains an actuator to a trajectory

1. A method of updating an influence model of an environmental maintenance system that includes a plurality of environmental maintenance modules and a plurality of sensors, each sensor measuring a value of a corresponding physical condition of an environment, each module including one or more actuators, the method comprising:

running the environmental maintenance system in a first production mode that uses the influence model to determine operation levels of the actuators by:

measuring sensor values of the sensors; and

determining first operation levels of the actuators that optimize a first cost function such that predicted sensor values do not exceed a threshold, wherein the predicted sensor values are determined using the influence model based on the first operation levels of the actuators;

identifying an event indicating a second production mode in which the influence model is updated, wherein an operation level of at least one of the actuators is varied as part of an update procedure of the second production mode; and

running the environmental maintenance system in the second production mode by:

selecting a first actuator among the actuators for calibration override, wherein the calibration override includes:

constraining an operation level of the first actuator to vary according to a first predetermined trajectory, wherein the first predetermined trajectory overrides an operation level determined for the first actuator using the influence model, constraining including:

assigning a first predetermined value to the first actuator at a first predetermined time, the first predetermined trajectory including the first predetermined value at the first predetermined time; and

assigning a second predetermined value to the first actuator at a second predetermined time, the first predetermined trajectory including the first predetermined value at the first predetermined time and the second predetermined value at the second predetermined time;

determining second operation levels of a subset of the actuators that optimize a second cost function based on measured sensor values, the subset not including the first actuator; and

updating the influence model based on the first predetermined trajectory, the second operation levels, and measured sensor values.

2. The method of claim 1 , wherein identifying the event includes:

determining a first confidence score of the influence model associated with the first actuator;

selecting the first actuator among the actuators for the calibration override based on the first confidence score being lower than a first threshold.

3. The method of claim 2 , wherein the first confidence score is based on a length of time that a variation in the operation level of the first actuator is less than a threshold.

4. The method of claim 1 , wherein the first predetermined trajectory changes linearly over time.

5. The method of claim 1 , wherein the event indicating the second production mode is periodically scheduled or is an unchanged operation level in one of the actuators for more than a specified length of time.

6. The method of claim 1 , wherein the second operation levels of the subset of the actuators are selected to maintain physical conditions within acceptable ranges of values.

7. The method of claim 1 , wherein updating the influence model comprises:

receiving a time sequence of data comprising values of the sensors and the actuators at a plurality of first time instants;

using the values of the sensors and actuators at the plurality of first time instants to predict a value of a selected sensor at a second time instant that is subsequent to the plurality of first time instants;

comparing the predicted value of the selected sensor at the second time instant to an actual value of the selected sensor at the second time instant; and

updating model parameters based on the comparing.

8. The method of claim 7 , wherein updating the model parameters uses a least squares technique.

9. The method of claim 1 , further comprising:

exiting the second production mode in response to a sensor value exceeding a corresponding threshold.

10. The method of claim 1 , wherein the first predetermined value is equal to the second predetermined value.

11. A computer product comprising a non-transitory computer readable medium storing a plurality of instructions that when executed cause a computer system to update an influence model of an environmental maintenance system that includes a plurality of environmental maintenance modules and a plurality of sensors, each sensor measuring a value of a corresponding physical condition of an environment, each module including one or more actuators, the instructions comprising:

running the environmental maintenance system in a first production mode that uses the influence model to determine operation levels of the actuators by:

measuring sensor values of the sensors; and

determining first operation levels of the actuators that optimize a first cost function such that predicted sensor values do not exceed a threshold, wherein the predicted sensor values are determined using the influence model based on the first operation levels of the actuators;

identifying an event indicating a second production mode in which the influence model is updated, wherein an operation level of at least one of the actuators is varied as part of an update procedure of the second production mode; and

running the environmental maintenance system in the second production mode by:

selecting a first actuator among the actuators for calibration override, wherein the calibration override includes:

constraining an operation level of the first actuator to vary according to a first predetermined trajectory, wherein the first predetermined trajectory overrides an operation level determined for the first actuator using the influence model, constraining including:

assigning a first predetermined value to the first actuator at a first predetermined time, the first predetermined trajectory including the first predetermined value at the first predetermined time and

assigning a second predetermined value to the first actuator at a second predetermined time, the first predetermined trajectory including the first predetermined value at the first predetermined time and the second predetermined value at the second predetermined time;

determining second operation levels of a subset of the actuators that optimize a second cost function based on measured sensor values, the subset not including the first actuator; and

updating the influence model based on the first predetermined trajectory, the second operation levels, and measured sensor values.

12. An environmental maintenance system comprising:

a plurality of sensors, each sensor measuring a value of a corresponding physical condition of an environment,

a plurality of environmental maintenance modules, each module including one or more actuators; and

a supervisory controller that is communicatively linked with the modules, sensors, and the computer product of claim 11 , and that is configured to execute the plurality of instructions of the computer product of claim 11 .

13. The computer product of claim 11 , wherein identifying the event includes:

determining a first confidence score of the influence model associated with the first actuator;

selecting the first actuator among the actuators for the calibration override based on the first confidence score is being lower than a first threshold.

14. The computer product of claim 11 , wherein updating the influence model comprises:

receiving data of values of the sensors and the actuators at a plurality of first time instants;

using the values of the sensors and actuators at the plurality of first time instants to predict a value of a selected sensor at a second time instant that is subsequent to the plurality of first time instants;

comparing the predicted value of the selected sensor at the second time instant to an actual value of the selected sensor at the second time instant; and

updating model parameters based on the comparing.

15. A method of updating an influence model of an environmental maintenance system that includes a plurality of environmental maintenance modules and a plurality of sensors that measure a value of a physical condition of an environment, each module including one or more actuators, the method comprising:

running the environmental maintenance system in a first production mode that uses the influence model to determine operation levels of the actuators by:

measuring sensor values of the sensors; and

determining first operation levels of the actuators that optimize a first cost function such that predicted sensor values do not exceed a threshold, wherein the predicted sensor values are determined using the influence model based on input levels of the actuators;

identifying an event indicating a second production mode in which the influence model is updated, wherein identifying the event includes:

determining a first confidence score of the influence model associated with the one or more actuators;

selecting a first actuator among the actuators for calibration override based on the first confidence score for the first actuator being lower than a first threshold, and

varying an operation level of the first actuator according to a predetermined trajectory as part of the calibration override, wherein the predetermined trajectory overrides an operation level determined for the first actuator using the influence model, varying including:

assigning a first predetermined value to the first actuator at a first predetermined time, the predetermined trajectory including the first predetermined value at the first predetermined time; and

assigning a second predetermined value to the first actuator at a second predetermined time, the predetermined trajectory including the first predetermined value at the first predetermined time and the second predetermined value at the second predetermined time.

16. The method of claim 15 , wherein updating the influence model comprises:

receiving a time sequence of data comprising values of the sensors and the actuators at a plurality of first time instants;

using the values of the sensors and actuators at the plurality of first time instants to predict a value of a selected sensor at a second time instant that is subsequent to the plurality of first time instants;

comparing the predicted value of the selected sensor at the second time instant to an actual value of the selected sensor at the second time instant; and

updating model parameters based on the comparing.

17. The method of claim 15 wherein, in the case that during the varying of the operation level of the first actuator according to the predetermined trajectory a sensor value exceeds a threshold, exiting the second production mode.

18. A computer product comprising a non-transitory computer readable medium storing a plurality of instructions that when executed cause a computer system to update an influence model of an environmental maintenance system that includes a plurality of environmental maintenance modules and a plurality of sensors, each sensor measuring a value of a corresponding physical condition of an environment, each module including one or more actuators, the instructions comprising the method of claim 15 .

19. An environmental maintenance system comprising:

a plurality of sensors, each sensor measuring a value of a corresponding physical condition of an environment,

a plurality of environmental maintenance modules, each module including one or more actuators, and

a supervisory controller that is communicatively linked with the modules, sensors, and the computer product of claim 18 , and that is configured to execute the plurality of instructions of the computer product of claim 18 .

20. The environmental system of claim 19 , wherein, in the case that during the varying of the operation level of the first actuator according to the predetermined trajectory a sensor value exceeds a threshold, the supervisory controller causes the environmental maintenance system to exit the second production mode.