Market Participant Self-Forecasting Technologies Workshop outcomes and summary.
Presentation extract
1.1 Summary of updates and further communication
After considering the feedback received, AEMO has made the following updates to the self-forecasting requirements:
- AEMO won’t perform a frozen value check in its validations;
- AEMO range check will allow small negative values in its validations;
- AEMO intends to perform a rolling Mean Absolute Error (MAE) check against an error threshold in its validations;
- AEMO won’t use Possible Power in its on-line validations. However, if available, Possible Power will form part of a suite of analysis in weekly offline checks.
- In addition to the variables proposed in the workshop, the confidential 5-minute reporting to participants will also include:
- Persistence forecast (MW) (= SCADA Initial MW)
- MP5F Validation Fail reason flag (Range Fail, MAE Fail, MP5F Quality Flag = “Not use”)
- Which forecast was used in Dispatch (MP5F, ANEMOS, PERSISTENCE, TARGET)
- MP5F Forecast Rolling 15-minute MAE (MW)
- ANEMOS Forecast Rolling 15-minute MAE (MW)
- Persistence Forecast Rolling 15-minute MAE (MW)
- All data produced in the real time reporting will be published publicly the next day.
- Participants will be able to re-submit MP5F for the same Next DI Ending as often as required within the MP5F submission acceptance window; and
- Participants will receive a time-stamped submission success notice, partial notice (for an incomplete 5MPD submission) or rejection notice with error message from AEMO via Web API.
AEMO also considers that further information needs to be provided on the following:
- Web API authentication and security information;
- Web API technical requirements and submission process; and
- MAE calculation and a specific list of suggested participant validations.