System and method of modeling and monitoring an energy load

• US 8,239,178 B2
• Filed: 09/16/2009
• Issued: 08/07/2012
• Est. Priority Date: 09/16/2009
• Status: Active Grant

• Current Assignee: Schneider Electric USA, Inc. (Schneider Electric SE)
• Original Assignee: Schneider Electric USA, Inc. (Schneider Electric SE)
• Inventors: Gray, Anthony R., Hope, Shaun
• Primary Examiner(s): Craig, Dwin M
• Assistant Examiner(s): Chad, Aniss
• Application Number:
  US12/561,024
• Publication Number:
  US 20110066299A1
• Time in Patent Office: 1,056 Days
• Field of Search: 703/2, 700/291
• US Class Current: 703/2
• 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 11/46 Improving electric energy efficiency or saving
• 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: 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 50/06 Energy or water supply

Abstract

A system, method, and computer program product for predicting operation for physical systems with distinct operating modes uses observable qualities of the system to predict other qualities of the system. Independent variables including temperature or production volume are observed to determine the degree to which a dependent modeled variable, including energy load, is influenced. Partition variables representing operating conditions of the dependent variables are defined as discrete values. Reference datasets with coincident values of the dependent variable, independent variable, and partition variables are received, and models are created for each discrete value of the partition variables in the reference dataset. Each model is populated with the values of the dependent variable and the independent variable. The dependent variable is modeled as a function of the independent variable. Model accuracy is evaluated by processing new input data to generate output data that includes values of the coincident dependent variable, the independent variable, and the partition variable from the input dataset.

Claims

1. A computer-implemented method of modeling and monitoring an energy load, the method comprising:

defining a dependent variable with a load monitoring server, the dependent variable representing an operation of the energy load;

defining at least one independent variable with the load monitoring server, the at least one independent variable representing at least one influencing driver of the operation of the energy load;

defining at least one partition variable with the load monitoring server, the at least one partition variable representing an operating condition of the energy load as a set of two or more discrete values;

receiving a reference dataset at the load monitoring server, the reference dataset including coincident values of the dependent variable, at least one independent variable, and at least one partition variable;

analyzing the reference dataset with the load monitoring server to arrange the reference dataset into interdependent data based upon the discrete values of the at least one partition variable;

creating a model for each discrete value of the at least one partition variable in the analyzed reference dataset with the load monitoring server in which the dependent variable is modeled as a function of the at least one independent variable, the model representing operation of the energy load;

receiving an input dataset at the load monitoring server, the input dataset including additional coincident values of the at least one independent variable and the at least one partition variable;

processing the additional coincident values of the at least one independent variable and the at least one partition variable with the created models; and

generating an output dataset with the load monitoring server from the created models, the output dataset including predicted dependent variable values from the at least one independent variable and the at least one partition variable from the input dataset.

2. The method of claim 1, wherein each created model is developed using a linear regression method.

3. The method of claim 2, wherein the linear regression method is a piece-wise linear regression method.

4. The method of claim 2, wherein the linear regression method computes a nonlinear transform of a predicted quantity as a linear combination of at least one scaled input quantity.

5. The method of claim 1, wherein the partition variables include a set of discrete values derived from a range of continuous values.

6. The method of claim 1, wherein the coincident values in the reference dataset are measurements taken in the past.

7. The method of claim 6, wherein the measurements are performed with an external instrument.

8. The method of claim 1, wherein the coincident values in the reference dataset are hypothetical measurements.

9. The method of claim 1, wherein the additional coincident values in the input dataset are hypothetical measurements.

10. The method of claim 1, wherein the coincident values of the dependent variable are measurements of a utility service quantity.

11. The method of claim 10, wherein the measurements of a utility service quantity are an electrical utility service and include at least one of current, voltage, power, and energy.

12. The method of claim 10, wherein the utility service is a gas utility service, a water utility service, an air utility service, or a steam utility service.

13. The method of claim 1, wherein the coincident values of the independent variable is at least one measurement of temperature or production volume.

14. The method of claim 1, wherein the coincident values of the partition variables include at least one of building occupancy, building management system mode, and type of manufactured product.

15. The method of claim 1, wherein the influencing drivers include at least one of the physical properties of outdoor temperature, barometric pressure, humidity, cloud cover characteristics, length of day, building occupancy, production units, and man-hours worked.

16. The method of claim 1, wherein the reference dataset is provided to the load monitoring server by an intelligent electronic device (IED).

17. A system for modeling and monitoring an energy load, the system comprising:

a load monitoring server configured to;

define a dependent variable, the dependent variable representing an operation of the energy load;

define at least one independent variable, the at least one independent variable representing at least one influencing driver of the operation of the energy load;

define at least one partition variable, the at least one partition variable representing an operating condition of the energy load as a set of two or more discrete values;

receive a reference dataset, the reference dataset including coincident values of the dependent variable, at least one independent variable, and at least one partition variable;

analyze the reference dataset to arrange the reference dataset into interdependent data based upon the discrete values of the at least one partition variable;

create a model for each discrete value of the at least one partition variable in the analyzed reference dataset in which the dependent variable is modeled as a function of the at least one independent variable, the model representing operation of the energy load;

receive an input dataset, the input dataset including additional coincident values of the at least one independent variable and the at least one partition variable;

process the additional coincident values of the at least one independent variable and the at least one partition variable with the created models; and

generate an output dataset from the created models, the output dataset including predicted dependent variable values from the at least one independent variable and the at least one partition variable from the input dataset.

18. The system of claim 17 further comprising:

an intelligent electronic device (IED) communicatively coupled to the load monitoring server via a communications network, the TED storing monitored characteristic values of the reference data set.

19. The system of claim 18, wherein the stored monitored characteristic values of the reference data set include the coincident values of the dependent variable, independent variable, and partition variables.

20. A non-transitory computer-readable storage media for modeling and monitoring an energy load, the computer-readable storage media comprising one or more computer-readable instructions configured to cause one or more computer processors to execute the operations comprising:

defining a dependent variable with a load monitoring server, the dependent variable representing an operation of the energy load;

defining at least one independent variable with the load monitoring server, the at least one independent variable representing at least one influencing driver of the operation of the energy load;

defining at least one partition variable with the load monitoring server, the at least one partition variable representing an operating condition of the energy load as a set of two or more discrete values;

receiving a reference dataset at the load monitoring server, the reference dataset including coincident values of the dependent variable, at least one independent variable, and at least one partition variable;

analyzing the reference dataset with the load monitoring server to arrange the reference dataset into interdependent data based upon the discrete values of the at least one partition variable;

creating a model for each discrete value of the at least one partition variable in the analyzed reference dataset with the load monitoring server in which the dependent variable is modeled as a function of the at least one independent variable, the model representing operation of the energy load;

receiving an input dataset at the load monitoring server, the input dataset including additional coincident values of the independent variable and the partition variable;

processing the additional coincident values of the at least one independent variable and the at least one partition variable with the created models; and

generating an output dataset with the load monitoring server from the created models, the output dataset including predicted dependent variable values from the at least one independent variable and the at least one partition variable from the input dataset.