2004 Volume 124 Issue 6 Pages 824-834
We proposed the technology which estimates an electric current value at the time each being positive for each power distribution line. Generally, as for the reported demand forecast (for example, are demand forecast by the electricity supply place level and so on), to consider that the load which becomes a base moves isn’t necessary. However, as for the power distribution system, the change of the composition which is due to the system changing, and so on, and the load curve shape, too, have changed mainly. So, the section of the power distribution line was classified in four kinds of clusters by the composition ratio of every contract classification, and created in the estimate model by the multiple regression analysis. Furthermore, to improve the precision of the estimate model, we calculated an error coefficient of each power distribution line and established the error revision method which distributes an error coefficient to each section. As a result, the absolute average error with estimate electric current value by this technique was 7.7A and it was possible to confirm that 95% was stored within 30A.
This paper describes a method and validity of the load forecasting technology in the power distribution system.