Scheduled thermal control system

1. A cold storage facility comprising:

a cold storage enclosure defining an enclosed space for storing inventory;

a refrigeration system configured to cool the enclosed space;

a plurality of sensors configured to sense, at least, air temperatures at locations within the enclosed space, and detect parameters of the refrigeration system; and

a processor configured to perform operations comprising:

determining a thermal model of the enclosure and the refrigeration system, the thermal model modeling one or more thermal properties of the enclosure and the refrigeration system under varied use and environmental conditions, wherein the thermal properties include a thermal capacity of the inventory and a thermal resistance between the air temperatures and inventory temperatures;

obtaining an energy demand model, the energy demand model including a schedule of an energy demand for a predetermined period of future time;

determining an operational schedule to control the refrigeration system over the predetermined period of future time by:

generating a plurality of candidate schedules for controlling the refrigeration system using target air temperatures at the locations for the predetermined period of future time, the plurality of candidate schedules estimating target inventory temperatures that are determined based on, at least, the target air temperatures, the thermal capacity, and the thermal resistance, wherein the plurality of candidate schedules are determined based on the thermal model and the energy demand model, wherein the target inventory temperatures have a dampened thermal response relative to the target air temperatures over the predetermined period of future time;

generating a multi-dimensional graph providing costs for cooling the enclosure according to the plurality of candidate schedules;

randomly selecting a seed schedule from the plurality of candidate schedules;

evaluating the seed schedule in the multi-dimensional graph using an iterative optimization algorithm; and

selecting the operational schedule that provides an optimal cost from among the plurality of candidate schedules, the optimal cost corresponding to a local minimum of the costs identified when starting with the seed schedule in the multi-dimensional graph; and

a controller configured to control the refrigeration system over the predetermined period of future time according to the determined operational schedule.

2. The cold storage facility of claim 1 , where evaluating the seed schedule includes:

comparing a cost of the seed schedule to costs of a portion of the plurality of candidate schedules.

3. The cold storage facility of claim 1 , wherein the costs represent efficiency of the plurality of candidate schedules in controlling the refrigeration system.

4. The cold storage facility of claim 1 , wherein the costs represent an energy demand.

5. The cold storage facility of claim 1 , wherein the costs represent an energy cost, an energy consumption, or a combination of the energy cost and the energy consumption.

6. The cold storage facility of claim 1 , wherein the operational schedule is determined for one or more points in time over the predetermined period of future time.

7. The cold storage facility of claim 1 , wherein the iterative optimization algorithm includes gradient descent.

8. The cold storage facility of claim 1 , wherein the plurality of candidate schedules provides different levels of cooling of the enclosure at different points in time, the different levels of cooling including different levels of electric power for operating the refrigeration system.

9. A method for determining an operational schedule to control a refrigeration system for an enclosure for storing inventory, the method comprising:

determining a thermal model of the enclosure and the refrigeration system, the thermal model modeling one or more thermal properties of the enclosure and the refrigeration system under varied use and environmental conditions, wherein the thermal properties include a thermal capacity of the inventory and a thermal resistance between the air temperatures and inventory temperatures;

obtaining an energy demand model, the energy demand model including a schedule of an energy demand for a predetermined period of future time;

determining an operational schedule to control the refrigeration system over the predetermined period of future time by:

generating a plurality of candidate schedules for controlling the refrigeration system using target air temperatures at locations within the enclosure for the predetermined period of future time, the plurality of candidate schedules estimating target inventory temperatures that are determined based on, at least, the target air temperatures, the thermal capacity, and the thermal resistance, wherein the plurality of candidate schedules are determined based on the thermal model and the energy demand model, wherein the target inventory temperatures have a dampened thermal response relative to the target air temperatures over the predetermined period of future time, and wherein each of the plurality of candidate schedules provides a different schedule of, at least, operation levels for the refrigeration system over the predetermined period of future time;

generating a multi-dimensional graph providing costs for cooling the enclosure according to the plurality of candidate schedules;

randomly selecting a seed schedule from the plurality of candidate schedules;

evaluating the seed schedule in the multi-dimensional graph using an iterative optimization algorithm; and

selecting the operational schedule that provides an optimal cost from among the plurality of candidate schedules, the optimal cost corresponding to a local minimum of the costs identified when starting with the seed schedule in the multi-dimensional graph; and

controlling the refrigeration system over the predetermined period of future time according to the determined operational schedule.

10. The method of claim 9 , where the operational schedule is determined further by:

evaluating the costs for cooling the enclosure according to the plurality of candidate schedules, based on the thermal model and the energy demand model.

11. The method of claim 9 , wherein evaluating the seed schedule includes:

comparing a cost of the seed schedule to costs of a portion of the plurality of candidate schedules.

12. The method of claim 9 , wherein the costs represent efficiency of the plurality of candidate schedules in controlling the refrigeration system.

13. The method of claim 9 , wherein the costs represent an energy demand.

14. The method of claim 9 , wherein the costs represent an energy cost, an energy consumption, or a combination of the energy cost and the energy consumption.

15. The method of claim 9 , wherein the operational schedule is determined for one or more points in time over the predetermined period of future time.

16. The method of claim 9 , further comprising:

calibrating the multi-dimensional graph over time.

17. The method of claim 9 , wherein the iterative optimization algorithm includes gradient descent.

18. The method of claim 9 , wherein the thermal properties include at least one of a thermal capacity of content within the enclosure and a thermal resistance of the enclosure.

19. The method of claim 9 , wherein the plurality of candidate schedules provides different levels of cooling of the enclosure at different points in time, the different levels of cooling including different levels of electric power for operating the refrigeration system.

20. A cold storage management computer system for controlling a refrigeration system for an enclosure for storing inventory, the cold storage management computer system comprising:

one or more processors; and

memory storing instructions that when executed by the one or more processors cause the one or more processors to perform operations comprising:

receiving, from a control system, a request for an operational schedule for the refrigeration system;

determining a thermal model of the enclosure and the refrigeration system, the thermal model modeling one or more thermal properties of the enclosure and the refrigeration system under varied use and environmental conditions, wherein the thermal properties include a thermal capacity of the inventory and a thermal resistance between the air temperatures and inventory temperatures;

obtaining an energy demand model, the energy demand model including a schedule of an energy demand for a predetermined period of future time; and

determining the operational schedule to control the refrigeration system over the predetermined period of future time by:

generating a plurality of candidate schedules for controlling the refrigeration system using target air temperatures at locations within the enclosure for the predetermined period of future time, the plurality of candidate schedules estimating target inventory temperatures that are determined based on, at least, the target air temperatures, the thermal capacity, and the thermal resistance, wherein the plurality of candidate schedules are determined based on the thermal model and the energy demand model, wherein the target inventory temperatures have a dampened thermal response relative to the target air temperatures over the predetermined period of future time;

generating a multi-dimensional graph providing costs for cooling the enclosure according to the plurality of candidate schedules;

randomly selecting a seed schedule from the plurality of candidate schedules;

evaluating the seed schedule in the multi-dimensional graph using an iterative optimization algorithm; and

selecting the operational schedule that provides an optimal cost from among the plurality of candidate schedules, the optimal cost corresponding to a local minimum of the costs identified when starting with the seed schedule in the multi-dimensional graph.