Building management system with dynamic energy prediction model updates
A building management system including building equipment operable to affect a variable state or condition of a building. The building management system includes a controller including a processing circuit. The processing circuit is configured to obtain an energy prediction model (EPM) for predicting energy requirements over time. The processing circuit is configured to monitor one or more triggering events to determine if the EPM should be retrained. The processing circuit is configured to, in response to detecting that a triggering event has occurred, identify updated values of one or more hyper-parameters of the EPM. The processing circuit is configured to operate the building equipment based on the EPM.
1. A building management system comprising:
building equipment operable to affect a variable state or condition of a building;
a controller comprising a processing circuit configured to:
obtain an energy prediction model (EPM) for predicting energy requirements over time;
monitor one or more triggering events to determine if the EPM should be retrained;
in response to detecting that a triggering event has occurred, identify updated values of one or more hyper-parameters of the EPM; and
operate the building equipment based on the EPM.
2. The building management system of claim 1 , wherein the processing circuit is further configured to:
calculate a performance metric of the EPM;
wherein the triggering event comprises the performance metric exceeding a threshold value.
3. The building management system of claim 1 , wherein the processing circuit is further configured to:
compare the EPM associated with the updated values of the one or more hyper-parameters to an existing EPM to determine what EPM generates more accurate predictions; and
deploy the EPM that generates more accurate predictions.
4. The building management system of claim 1 , wherein the processing circuit is further configured to pre-process operational data of the building equipment, wherein the updated values of the one or more hyper-parameters are determined based on the pre-processed operational data.
5. The building management system of claim 1 , wherein the processing circuit is further configured to retrain the EPM based on data describing operation of the building equipment, wherein the EPM is retrained respective to a proportion of old data and new data.
6. The building management system of claim 1 , wherein the one or more hyper-parameters include at least one of:
a transfer learning flag; or
a training window.
7. The building management system of claim 1 , wherein the processing circuit is further configured to switch to an older EPM in response to determining that the older EPM is more accurate than the EPM.
8. A method for operating building equipment of a building, the method comprising:
obtaining an energy prediction model (EPM) for predicting energy requirements over time;
monitoring one or more triggering events to determine if the EPM should be retrained;
in response to detecting that a triggering event has occurred, identifying updated values of one or more hyper-parameters of the EPM; and
operating the building equipment based on the EPM to affect a variable state or condition of the building.
9. The method of claim 8 , further comprising:
calculating a performance metric of the EPM;
wherein the triggering event comprises the performance metric exceeding a threshold value.
10. The method of claim 8 , further comprising:
comparing the EPM associated with the updated values of the one or more hyper-parameters to an existing EPM to determine what EPM generates more accurate predictions; and
deploying the EPM that generates more accurate predictions.
11. The method of claim 8 , further comprising pre-processing operational data of the building equipment, wherein the updated values of the one or more hyper-parameters are determined based on the pre-processed operational data.
12. The method of claim 8 , further comprising retraining the EPM based on data describing operation of the building equipment, wherein the EPM is retrained respective to a proportion of old data and new data.
13. The method of claim 8 , wherein the one or more hyper-parameters include at least one of:
a transfer learning flag; or
a training window.
14. The method of claim 8 , further comprising switching to an older EPM in response to determining that the older EPM is more accurate than the EPM.
15. A method for deploying an energy prediction model (EPM) for equipment, the method comprising:
obtaining the EPM for predicting energy requirements over time;
monitoring one or more triggering events to determine if the EPM should be retrained;
in response to detecting that a triggering event has occurred, identifying updated values of one or more hyper-parameters of the EPM; and
deploying the EPM for the equipment.
16. The method of claim 15 , further comprising:
calculating a performance metric of the EPM;
wherein the triggering event comprises the performance metric exceeding a threshold value.
17. The method of claim 15 , further comprising:
comparing the EPM associated with the updated values of the one or more hyper-parameters to an existing EPM to determine what EPM generates more accurate predictions; and
deploying the EPM that generates more accurate predictions.
18. The method of claim 15 , further comprising pre-processing operational data of the equipment, wherein the updated values of the one or more hyper-parameters are determined based on the pre-processed operational data.
19. The method of claim 15 , further comprising retraining the EPM based on data describing operation of the equipment, wherein the EPM is retrained respective to a proportion of old data and new data.
20. The method of claim 15 , wherein the one or more hyper-parameters include at least one of:
a transfer learning flag; or
a training window.