1st Competition
The set problem was to predict daily peak demands for the month of January 1999 based on 2 previous years and some additional information related to the average temperature data as well as status of the day were provided. Both maximum and average errors were of concern of the month January 1999 and the data is related to the Eastern Slovakia region load.
Additionally some other data from previous years was provided to support the procedure of prediction and get as precise information (prediction) as possible. There was no interaction with the distribution electricity center in Eastern Slovakia permitted so the expected results were assumed to be the worse possible results without any enhancement of the costumer which is possible during a real project. The available data consisted of:
|
|
Average Daily Temperatures |
Half-Hourly Loads |
|
1995 |
Y |
No |
|
1996 |
Y |
Y |
|
1997 |
Y |
Y |
|
1998 |
Y |
Y |
|
1999 |
No |
No |
In addition, a list of public holidays was provided. For the Jan-Feb period, the only holidays were date-fixed: Jan 1 and Jan 6. The units of the Load data were not specified.
Note that no temperature data was given for 1999. The implication of this was that if we were to develop a temperature-dependent load model, we would have to produce our own estimate of the January 1999 temperatures. So in basic form over 700x(48 data per day) were provided. The character of the data is shown on the following graph.
The character of the data within the day having 48 data information shows the types of periodicities which were analyzed and some of the technologies were based on depicting characters of the data and using some additional information including temperature to predict the max. values for January. In principle the possible model was to use the prediction of temperature and based on this prediction predict the load. The character of this data were as follows:
The character of the temperature in the studied region was as follows :
The predicted data for January 1999 was confidential and non-public and there was no publication of this data on any means of communication and that meant that the desired data was absolutely not accessible and possible to obtain. The character of the desired data to predict was as follows :
