Publication: ANFIS ve ARMA modelleri ile elektrik enerjisi yük tahmini
Abstract
Elektrik enerjisi, insanın refahı ve çağdaş bir hayat sürdürebilmesi için en önemli ihtiyaçlardan biridir. Böylesine önemli bir ihtiyacın yeterli ve kaliteli bir şekilde temin edilebilmesi için elektrik enenerjisine olan talebin önceden tahmin edilerek buna göre üretime yönelik yatırımların yapılması gerekir. Bu çalışmada tahminleme yöntemlerinden Adaptif Ağ Tabanlı Bulanık Çıkarım Sistemi (Adaptive Network Based Fuzzy Inference Systems-ANFIS), Otoregresif Hareketli Ortalamalar (Autoregressive Moving Average-ARMA) teknikleri ile elektrik enerjisi talep tahminleri yapılmış ve hangi yöntemin talep tahmininde daha başarılı olduğu ortaya konulmuştur. ANFIS modelin için Matlab 7.04, ARMA model için SPSS15 paket programı kullanılmıştır. Elde edilen sonuçlar ilgili kurumların öngörüleri ile karşılaştırılmıştır. Bu çalışmada, 1970-2007 arası Gayri Safi Milli Hasıla (GSMH), Üretilen Enerji (ÜE), Tüketilen Enerji (TE), Nüfus (N) ve Kurulu Güç (KG) verileri kullanılarak ANFIS ve ARMA modelleri oluşturulmuştur. Elde edilen model yardımıyla 2006-2010 yıllarına ait elektrik enerjisi talep tahminleri elde edilmiştir. ANFIS ve ARMA ile bulunan sonuçların karşılaştırmalı yorumları yapılarak önerilerde bulunulmuştur.
Electricity power is one of the needs of people to be able to live a contemporary life and for his well-being. In order for this need to be met, it is vital that electricity is produced enough and in good quality. It is necessary to predict the need of electricity beforehand, and thereby deciding on the generation of it. In this study, of the prediction methods, Regression, ANFIS, and ARMA have been used to assess the results obtained and the most successful method in the prediction of electricity demand has been determined. Fuuzy Logic and Neural Networks toolboxes of Matlab 7.04 for ANFIS medel and SPSS 15 for ARMA model were used respectively. In this study, Gross National Product, Produced Energy, Consumed Energy, Population and İnstalled Capacity data have been used in the prediction of the consumed electricity between the years 1970–2007. ANFIS and ARMA models have been used and thus the energy demands between 2006–2010 have been predicted. The results obtained by ANFIS and ARMA models were compared and some suggestions were presented.
Electricity power is one of the needs of people to be able to live a contemporary life and for his well-being. In order for this need to be met, it is vital that electricity is produced enough and in good quality. It is necessary to predict the need of electricity beforehand, and thereby deciding on the generation of it. In this study, of the prediction methods, Regression, ANFIS, and ARMA have been used to assess the results obtained and the most successful method in the prediction of electricity demand has been determined. Fuuzy Logic and Neural Networks toolboxes of Matlab 7.04 for ANFIS medel and SPSS 15 for ARMA model were used respectively. In this study, Gross National Product, Produced Energy, Consumed Energy, Population and İnstalled Capacity data have been used in the prediction of the consumed electricity between the years 1970–2007. ANFIS and ARMA models have been used and thus the energy demands between 2006–2010 have been predicted. The results obtained by ANFIS and ARMA models were compared and some suggestions were presented.