US 12,613,050 B2 · Machine teaching with method of moments for mechanical systems

Abstract

The techniques disclosed herein enable utilizing a full range of setpoint values to control a mechanical system. A machine learning model is trained with states collected from the mechanical system. Some of the states may have little to no variation, limiting exploration of possible setpoint values when training the model. To enable a more thorough exploration of possible setpoint values, the states are augmented with a fluctuating delta value that is derived from a fixed setpoint value. For example, a delta outside air temperature may be computed by subtracting outside air temperature, which fluctuates, from a fixed chilled water setpoint. A method of moments computation converts delta values inferred by the model back into absolute values. The absolute values are used to compute a regression equation that is usable by the mechanical system to compute a setpoint action for a given set of input states.

Claims20

Dependent claims

A method comprising:

setting a chilled water temperature setpoint of a cooling system based on an absolute action, wherein the absolute action is obtained from a function that maps individual states of the cooling system to individual absolute actions, wherein the function is generated based on a regression analysis of a plurality of sets of cooling system state data and a corresponding plurality of absolute actions, wherein the plurality of absolute actions is obtained by converting a plurality of delta actions, wherein the plurality of delta actions are inferred by a machine learning model from the plurality of sets of cooling system state data, and wherein the machine learning model is trained by:
computing a delta outside air temperature based on a chilled water temperature setpoint of the cooling system and an outside air temperature,
computing a delta return water temperature based on the chilled water temperature setpoint and a return water temperature of the cooling system, and
training the machine learning model based on the delta outside air temperature, the delta return water temperature, and a state of the cooling system.

1 more element

A computer-readable storage medium comprising instructions that, when executed by a processor, cause the processor to:

capture a set of states of a cooling system;
invoke a function with a subset of the set of states of the cooling system, wherein the function returns an absolute action of the cooling system, wherein the function is generated by:
computing a delta outside air temperature based on a chilled water temperature setpoint of a cooling system and an outside air temperature;
computing a delta return water temperature based on the chilled water temperature setpoint and a return water temperature of the cooling system;
training a machine learning model based on the delta outside air temperature, the delta return water temperature, and a state of one of a plurality of sets of states collected during operation of the cooling system;
using the machine learning model to infer a plurality of delta actions from a plurality of sets of cooling system state data;
converting the plurality of delta actions to a plurality of absolute actions;
and
performing a regression analysis on the plurality of sets of cooling system state data and the corresponding plurality of absolute actions to generate the function that maps individual states of the cooling system to individual absolute actions;
and
set the setpoint of the cooling system based on the absolute action returned by the function.

8 more elements

Dependencies are inferred from the claim text.

Machine teaching with method of moments for mechanical systems

Life of this patent

Machine teaching with method of moments for mechanical systems

Life of this patent

Abstract

Claims20