Building system with model training to handle selective forecast data

1. A building system for training a prediction model with augmented training data, the building system comprising one or more memory devices configured to store instructions thereon that, when executed by one or more processors, cause the one or more processors to:

obtain a first training data set comprising historical prediction data values associated with a data point of the building system and with a plurality of time-steps and historical actual energy values associated with consumption of the building system at each of the plurality of time-steps;

identify one or more time-steps of the plurality of time steps at which the historical prediction data values are inaccurate relative to the historical actual energy values;

replace one or more values of a set of current prediction data values with one or more predetermined values at one or more time-steps which correspond to the identified time-steps at which the historical prediction data values are inaccurate;

generate an augmented training data set, the augmented training data set comprising a second training data set, the second training data set comprising the current prediction data values after replacing the one or more values of the set of current prediction data values with the one or more predetermined values;

generate a prediction model by training the prediction model using the augmented training data set; and

operate one or more pieces of building equipment using a prediction generated by the prediction model.

2. The building system of claim 1 , wherein the augmented training data set comprises a third training data set, the third training data set comprising the historical actual energy values and the historical prediction data values with a historical prediction data value replaced with a predetermined value at a time-step of the plurality of time-steps.

3. The building system of claim 2 , wherein the augmented training data set comprises a fourth training data set, the fourth training data set comprising the historical actual energy values and the predetermined value at each of the plurality of time-steps.

4. The building system of claim 1 , wherein the instructions cause the one or more processors to:

receive a current prediction data set comprising a plurality of current prediction data values, the plurality of current prediction data values associated with the data point and with the plurality of time-steps; and

predict, using the prediction model and based on the prediction data set, for one or more time-steps into the future, an expected energy output indicating an expected amount of energy that the building system will utilize at each of the plurality of time-steps.

5. The building system of claim 4 , wherein the instructions cause the one or more processors to perform at least one of:

operate the one or more pieces of building equipment based on the expected energy output; or

generate one or more user interfaces including interface elements, the interface elements based on the expected energy output.

6. The building system of claim 1 , wherein the first training data set comprises second historical prediction data values associated with a second data point of the building system, wherein the second historical prediction data values are associated with historical actual values of the second data point at each of the plurality of time-steps.

7. The building system of claim 1 , wherein the data point is associated with outside air temperature, plug load schedule, light load schedule, occupancy, or scheduling information.

8. The building system of claim 1 , wherein the augmented training data set comprises an indicator training data set, the indicator training data set indicating time-steps of the first training data set that are associated with the predetermined value.

9. The building system of claim 8 , wherein the augmented training data set comprises a second indicator training data set, the second indicator training data set indicating second time-steps of the second training data set that are associated with the predetermined value.

10. The building system of claim 1 , wherein the historical prediction data values are associated with historical predictions of a historical prediction value of the data point at each of the plurality of time-steps.

11. A method for training a prediction model with augmented training data, the method comprising:

obtaining, by a processing circuit, a first training data set comprising historical prediction data values associated with a data point of the building system and with a plurality of time-steps and historical actual energy values associated with consumption of a building system at each of the plurality of time-steps;

identifying, by the processing circuit, one or more time-steps of the plurality of time steps at which the historical prediction data values are inaccurate relative to the historical actual energy values;

replacing, by the processing circuit, one or more values of a set of current prediction data values with one or more predetermined values at one or more time-steps which correspond to the identified time-steps at which the historical prediction data values are inaccurate;

generating, by the processing circuit, an augmented training data set, the augmented training data set comprising a second training data set, the second training data set comprising the current prediction data values after replacing the one or more values of the set of current prediction data values with the one or more predetermined values;

generating, by the processing circuit, a prediction model by training the prediction model using the augmented training data set; and

operating, by the processing circuit, one or more pieces of building equipment using a prediction generated by the prediction model.

12. The method of claim 11 , wherein the augmented training data set comprises a third training data set, the third training data set comprising the historical actual energy values and the historical prediction data values with a historical prediction data value replaced with a predetermined value at a time-step of the plurality of time-steps.

13. The method of claim 12 , wherein the augmented training data set comprises a fourth training data set, the fourth training data set comprising the historical actual energy values and the predetermined value at each of plurality of time-steps.

14. The method of claim 11 , comprising:

receiving, by the processing circuit, a current prediction data set comprising a plurality of current prediction data values, the plurality of current prediction data values associated with the data point and with the plurality of time-steps; and

predicting, by the processing circuit, using the prediction model and based on the prediction data set, for one or more time-steps into the future, an expected energy output indicating an expected amount of energy that the building system will utilize at each of the plurality of time-steps.

15. The method of claim 14 , comprising:

operating, by the processing circuit, the one or more pieces of building equipment based on the expected energy output; or

generating, by the processing circuit, one or more user interfaces including interface elements, the interface elements based on the expected energy output.

16. The method of claim 11 , wherein the first training data set comprises second historical prediction data values associated with a second data point of the building system, wherein the second historical prediction data values are associated with historical actual values of the second data point at each of the plurality of time-steps.

17. The method of claim 11 , wherein the data point is associated with outside air temperature, plug load schedule, light load schedule, occupancy, or scheduling information.

18. A non-transitory computer-readable storage medium having instructions stored thereon that, upon execution by a processor, cause the processor to perform operations to train a prediction model with augmented data, the operations comprising:

obtaining a first training data set comprising historical prediction data values associated with a data point of a building system and with a plurality of time-steps and historical actual energy values associated with consumption of the building system at each of the plurality of time-steps;

identifying one or more time-steps of the plurality of time steps at which the historical prediction data values are inaccurate relative to the historical actual energy values;

replacing one or more values of a set of current prediction data values with one or more predetermined values at one or more time-steps which correspond to the identified time-steps at which the historical prediction data values are inaccurate;

generating an augmented training data set, the augmented training data set comprising a second training data set, the second training data set comprising the current prediction data values after replacing the one or more values of the set of current prediction data values with the one or more predetermined values;

generating a prediction model by training the prediction model using the augmented training data set; and

operating one or more pieces of building equipment using a prediction generated by the prediction model.