Systems and methods for statistical control and fault detection in a building management system
A building management strategy includes using exponentially weighted moving averages with statistical models to detect changes in the behavior of the building management system. Detected changes in the behavior of the system may indicate a detected fault, a change in a predicted behavior, or a need for the statistical models to be updated.
1. A controller for a building management system comprising:
a processing circuit configured to at least one of receive or calculate an updated performance value for the building management system;
wherein the processing circuit is further configured to maintain at least one threshold parameter relative to a history of performance values, to determine an exponentially weighted moving average of the performance values, and to compare the moving average to the at least one threshold parameter, wherein the threshold parameter is maintained by multiplying an estimator of scale of the history by a constant value and at least one of adding or subtracting a result of the multiplication to or from a median of the history; and
wherein the processing circuit is further configured to generate an output that indicates a result of the comparison.
2. The controller of claim 1 , wherein the processing circuit is further configured to update the threshold parameter automatically and without user input.
3. The controller of claim 1 , wherein the threshold parameter is a threshold for a predicted power consumption and the performance values comprise measured power consumptions.
4. The controller of claim 1 , wherein the processing circuit is further configured to cause a graphical user interface to be displayed on an electronic display device, wherein the graphical user interface comprises indicia that a fault has occurred.
5. A method of detecting faults in a building management system, the method comprising:
at a computer of the building management system, receiving first performance values for the building management system;
calculating an exponentially weighted moving average using the first performance values;
generating a threshold parameter by multiplying an estimator of scale of the first performance values by a constant value and at least one of adding or subtracting a result of the multiplication to or from a median of the first performance values;
comparing the moving average to the threshold parameter;
receiving new performance values for the building management system; and
updating the threshold parameter using the new performance values.
6. The method of claim 5 , wherein updating the threshold parameter comprises:
generating a target parameter, wherein the target parameter is a median of a history of performance values;
generating an estimator of scale of the history, wherein the estimator of scale reduces an effect of outliers in the history;
using the estimator of scale and the target parameter to generate the threshold parameter; and
storing the threshold parameter in a memory of the computer.
7. The method of claim 6 , wherein using the computer to generate at least one threshold parameter comprises:
determining if the history is autocorrelated;
applying an autoregressive model to the history; and
using the autoregressive model to generate the threshold parameter.
8. The method of claim 5 , further comprising:
updating the threshold parameter automatically and without user input in response to a result of the comparison.
9. The method of claim 5 , wherein updating the threshold parameter using the new performance values further comprises:
applying an anti-spike filter to the new performance values.
10. The method of claim 5 , wherein the new performance values are measured power consumptions and the at least one threshold parameter is selected based on a model of a predicted power consumption.