Up to one user account can be used per person or representative of a team. If different prediction methods are used, however, multiple entries may be permissible with revised pre-proposal.
[Handling of the information]
Participants are prohibited from sharing the data/information relating to prediction for this Contest with those who belong to other teams.
[Use of the data]
Participants can freely select, and use the following training data sets provided in advance by the Contest Hosts.
(1) Actual power output data in the past
- Actual power output data from targeted solar power plants.
- The period of the provided data is 2 years from January 1, 2016 to December 31, 2017.
- Based on variations of weather in Hokkaido, the contest area is divided into two major areas and the total power output is provided for each.
(2) Rating output, installation sites, azimuth angle, and inclination angle
- Separated rating output (published values) from each targeted solar power plant
- Installation sites (latitude, longitude)
- Azimuth angle (mean value) and inclination angle (mean value)
- (3) Independently measured temperature and global solar radiation values, and the measurement locations
- Total solar radiation (global solar radiation) and temperature data measured by the Contest Hosts
- Same submission period stated in (1) above.
- The measurement locations (latitude, longitude)*
* Global solar radiation was not measured at the locations of the targeted solar power plants.
(4) Basic information on the climate of Hokkaido
- Hokkaido, the Contest venue, is the northernmost island of the four major islands of the Japanese archipelago. The climate is cold with low humidity, and most of the area has snow/accumulated snow in winter (from December to March).
- Visit the following sites for overviews of the weather and seasonal features in Hokkaido.
◯ ”(1) Actual power output data in the past” will be provided by dividing the sites into two areas, “S1”and ”S2”.
Sum of S1：267.6MW（8 sites)
Sum of S2：139.7MW（7 sites)
The target area of prediction is sum of S1 and S2.
◯ Below image shows the location of ”(3)Independently measured amount of global solar radiation, and the measurement locations”.
- Other 7 locations in Hokkaido. Please see Japan Meteorological Agency Website.
- Please note the following limitations in the external data that participants can obtain.
- The data can be obtained in any way determined by the Contest Hosts.
- The data must be obtainable by 6:00 pm on the day before the targeted prediction day.
(Example) If the power output is predicted on January 11, 2019, the information must be published before 6:00 pm on January 10, 2019.
- The information should not prejudice the rights of any third party and must be public domain (anyone can obtain and reproduce the information for free or with charge.)
- If the forecasted values for the data to be predicted can be obtained by 6:00 pm on the previous day, the actual data on the targeted date can be used for prediction.
- Participants can freely consider and select a prediction method of their choice, provided that they follow the conditions below.
- The same method must be used for the prediction throughout the whole prediction period. (Do not use any method that cannot be automated in the future. The method must not include any process to change parameters during the period based on human judgment).
- Information on any tools to be used for prediction can be provided to the Contest Hosts.
- When referred to existing studies, references should be clearly provided.
- The power output unit used in the prediction results is MW. Values are to be calculated to the first decimal place.
- If submitted values are calculated to the second decimal place or a higher degree of accuracy, they will be rounded to the first decimal place before they are evaluated.
- If the column for any period is left blank (without data), a prediction result of "9999" will be filled in and evaluated accordingly.
[Reports explaining the prediction method]
(Report explaining the prediction method (summary): required from all participants)
- The following contents must be included.
- Outlines of the prediction model (logic, etc.)
- Type of data used
- Environment of performance, computation time
- Originality of the prediction model (points for accuracy improvement, specially designed points, etc.)
(Report explaining the prediction method (detailed): required from participants selected for the final review)
- In addition to explaining the prediction method in full detail, add the following.
- Discussion on the gaps between the prediction results and the actual data (actual power output values during the Contest period) to be separately provided.
[Presentation video: required from participants selected for the final review]
- Prepare a video explaining the prediction method within a run time of 5 minutes.
[Other common conditions]
- The time zone used for the Contest overview/rules is to be Japan time (JST).
- All submitted materials must written in Japanese or English.
- If there are periods when the power output of solar power plants is obviously unpredictable due to natural disasters or the like, those periods may be excluded from the targeted period for evaluation at the discretion of the Contest Hosts.
- After submission, the Contest Hosts (operation secretariat) may inquire separately about the prediction method or data used. If answers cannot be provided by the specified date and time, the entry will be removed from the selection review.
◆Points of prediction for large area
This page is not for entry. Users who want to enter the contest, please find the submit button on Application Guidelines.
→ Back to Application Guidelines