A Medium Term Prediction on Feeder Trip Profile in Power Systems Network using Artificial Intelligence

Abstract

The prediction on Feeder Trip Profile (FTP) on feeders radiating from TCN Benin was carried out in this study. Data on reliability were collected from literature between the years 2010–2015 for GRA, Guinness, Koko, Switch-Station & Ikpoba-Dam Feeders (FDRs) respectively. The prediction on the feeders' reliability was carried out using Artificial Neural Network (ANN) to determine the FTP and observable trend of the FDRs, while the statistical tools; Root Mean Square Error RMSE and Mean Absolute Percentage Error MAPE were utilized to determine if the error from the forecast falls within acceptable limits. Results obtained indicated reliability values of 0.9786, 0.8306, 0.7707, 0.9467 and 0.9467 for GRA, Guinness, Koko, Switchstation, and Ikpobadam FDRs by the year 2020, and when compared with FTP of 2015, it was observed that GRA FDR showed a fairly constant FTP while the reverse was the case for the duo of observed for Koko and Guinness FDRs. Results from the statistical and ANN tool showed that the error margin falls within acceptable standard of a maximum of 0.5. The results from this study could therefore be useful for system's planning and operations.