Systems and methods for predictive building energy monitoring
A system and method for predictive modeling of building energy consumption provides predicted building energy load values which are determined using kernel smoothing of historical building energy load values for a building using defined scaling factors for scaling predictor variables associated with building energy consumption. Predictor variables may include temperature, humidity, windspeed or direction, occupancy, time, day, date, and solar radiation. Scaling factor values may be defined by optimization training using historical building energy load values and measured predictor variable values for a building. Predicted and measured building energy load values are compared to determine if a preset difference threshold has been exceeded, in which case an alert signal or message is generated and transmitted to electronically and/or physically signal a user. The building energy monitoring system may be integrated with a building automation system, or may be operated as a separate system receiving building energy and predictor variable values.
1. A non-transitory computer readable medium, comprising executable instructions to:
receive historical building energy load values and corresponding historical predictor variable data comprising values for one or more predictor variables;
define a scaling factor for each said predictor variable by cross-validation error minimization for said historical building energy load values using said corresponding historical predictor variable values scaled by said scaling factors;
receive measured building energy load values measured by at least one building energy meter;
receive measured predictor variable values corresponding to the measured building energy load values from a predictor variable datasource;
determine a predicted building energy load value by kernel smoothing of one or more nearest neighbor historical building energy load values using said historical predictor variable values each scaled by said defined scaling factors;
compare said predicted building energy load value with said measured building energy load values to determine if an alert threshold level is exceeded; and
transmit an alert signal to a user when said alert threshold level is exceeded.
2. The non-transitory computer readable medium according to claim 1 , wherein said one or more predictor variables comprise at least one of: time, day, temperature, humidity, wind speed, wind direction, solar radiation and occupancy.
3. The non-transitory computer readable medium according to claim 1 , wherein said historical building energy load values and said measured building energy load values comprise consumption load values of one or more of: electrical energy; thermal energy; natural gas; steam and water.
4. The non-transitory computer readable medium according to claim 1 , additionally comprising executable instructions to:
transmit a profile of measured building energy load values and a profile of predicted building energy load values to a user.
5. The non-transitory computer readable medium according to claim 1 wherein said kernel smoothing comprises a unit Gaussian kernel smoothing of said one or more nearest neighbor historical building energy load values using said predictor variable values each scaled by said respective defined scaling factor.
6. The non-transitory computer readable medium according to claim 1 additionally comprising executable instructions to:
determine said nearest neighbor historical building energy load values using a k-dimensional tree.
7. The non-transitory computer readable medium according to claim 1 wherein said executable instruction to define a scaling factor for each said predictor variable further comprises an executable instruction to:
define a single scaling factor for each said predictor variable by cross-validation error minimization for said historical building energy load values using said corresponding historical predictor variable values using a Levenberg-Marquardt optimization to minimize said cross-validation error between said historical building energy load values and said one or more predicted historical building energy load values.
8. The non-transitory computer readable medium according to claim 1 , wherein said alert signal comprises one or more of: an electrical impulse, an audible sound, a mechanical force, a movement of an object, and a visible display.
9. A computer implemented method of building energy monitoring, said method comprising:
receiving historical building energy load values and corresponding historical predictor variable data comprising values for one or more predictor variables;
defining a scaling factor for each said predictor variable by cross-validation error minimization for said historical building energy load values using said corresponding historical predictor variable values;
receiving measured building energy load values measured by at least one building energy meter;
receiving measured predictor variable values corresponding to the measured building energy load values from a predictor variable datasource;
determining a predicted building energy load value by kernel smoothing of one or more nearest neighbor historical building energy load values using said historical predictor variable values each scaled by said respective defined scaling factors;
comparing said predicted building energy load value with said measured building energy load values to determine if an alert threshold level is exceeded; and
transmitting an alert signal to a user when said alert threshold level is exceeded.
10. The computer implemented method according to claim 9 , wherein said one or more predictor variables comprise at least one of: time, day, temperature, humidity, wind speed, wind direction, solar radiation and occupancy.
11. The computer implemented method according to claim 9 , wherein said historical building energy load values and said measured building energy load values comprise consumption load values of at least one of: electrical energy; thermal energy; natural gas; steam and water.
12. The computer implemented method according to claim 9 , additionally comprising transmitting a profile of measured building energy load values and a profile of predicted building energy load values to a user.
13. The computer implemented method according to claim 9 wherein said kernel smoothing comprises a unit Gaussian kernel smoothing of said one or more nearest neighbor historical building energy load values using said predictor variable values each scaled by said respective defined scaling factors.
14. The computer implemented method according to claim 9 , additionally comprising computing said one or more nearest neighbor historical building energy load values using a k-dimensional tree.
15. The computer implemented method according to claim 9 wherein said measured building energy load values are measured by at least one building energy meter from at least one of: a building management computer; a building management database; a building energy smart meter; and a utility database.
16. The computer implemented method according to claim 9 wherein defining a scaling factor for each said predictor variable further comprises:
defining a single scaling factor for each said predictor variable by cross-validation error minimization for said historical building energy load values and said corresponding historical predictor variable values using a Levenberg-Marquardt optimization to minimize said cross-validation error between said historical building energy load values and predicted historical building energy load values.
17. The computer implemented method according to claim 9 wherein said predictor variable datasource comprises at least one of: a weather service database; a weather station computer; an offline weather data archive; and a building occupant monitoring system.
18. The computer implemented method according to claim 9 wherein said alert signal comprises one or more of: an electrical impulse, an audible sound, a mechanical force, a movement of an object, and a visible display.
19. A computer implemented method of building energy monitoring, said method comprising:
receiving historical building energy load values and corresponding historical predictor variable data comprising values for one or more predictor variables;
receiving scaling factor values for each said predictor variable;
receiving measured building energy load values measured by at least one building energy meter;
receiving measured predictor variable values corresponding to the measured building energy load values from a predictor variable datasource;
determining a predicted building energy load value by kernel smoothing of one or more nearest neighbor historical building energy load values using said historical predictor variable values each scaled by said respective scaling factor values;
comparing said predicted building energy load value with said measured building energy load values to determine if an alert threshold level is exceeded; and
transmitting an alert signal to a user when said alert threshold level is exceeded.
20. The computer implemented method according to claim 19 , wherein said one or more predictor variables comprise at least one of: time, day, temperature, humidity, wind speed, wind direction, solar radiation and occupancy.
21. The computer implemented method according to claim 19 , wherein said one or more nearest neighbor historical building energy load values are determined using a k-dimensional tree.
22. The computer implemented method according to claim 19 , additionally comprising transmitting a profile of said measured building energy load values and a profile of said predicted building energy load values to a user.
23. The computer implemented method according to claim 19 wherein said alert signal comprises one or more of: an electrical impulse, an audible sound, a mechanical force, a movement of an object, and a visible display.
24. A computer, comprising:
a processor that executes instructions; and
a computer-readable memory that stores instructions that cause the processor to determine parameters for building energy monitoring by:
receiving historical building energy load values and corresponding historical predictor variable data comprising values one or more predictor variables;
receiving scaling factor values for each of said predictor variables;
receiving measured building energy load values measured by at least one building energy meter;
receiving measured predictor variable values corresponding to the measured building energy load values from a predictor variable datasource;
determining a predicted building energy load value by kernel smoothing of one or more nearest neighbor historical building energy load values using said historical predictor variable values each scaled by said respective scaling factor values;
comparing said predicted building energy load value with said measured building energy load values to determine if an alert threshold level is exceeded; and
transmitting an alert signal to a user when said alert threshold level is exceeded.
25. The computer according to claim 24 wherein said one or more predictor variables comprise at least one of: time, day, temperature, humidity, wind speed, wind direction, solar radiation and occupancy.
26. The computer according to claim 24 wherein said one or more nearest neighbor historical building energy load values are determined using a k-dimensional tree.
27. The computer according to claim 24 wherein said measured building energy load values are measured by at least one building energy meter from at least one of: a building management computer, a building management database, a building energy smart meter, and a utility database.
28. The computer according to claim 24 wherein said predictor variable datasource includes at least one of: a weather service database; a weather station computer; an offline weather data archive; and a building occupant monitoring system.
29. The computer according to claim 24 wherein said alert signal comprises one or more of: an electrical impulse, an audible sound, a mechanical force, a movement of an object, and a visible display.
30. The computer according to claim 24 wherein said computer comprises one or more distributed computers communicatively connected by a computer network.