Elektrik piyasasının rastsal ve gürbüz optimizasyon kullanılarak modellenmesi.

---

Tezin Türü: Doktora

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Türkiye

Tezin Onay Tarihi: 2015

Tezin Dili: İngilizce

Öğrenci: Miray Hanım Yıldırım

Eş Danışman: Gerhard Weber, GERHARD WİLHELM WEBER

Özet:

Sustainable development relying on sustainable and renewable energy systems is becoming one of the major policies of many countries. This forces the policy makers to establish many reforms and revolutions, which evolve electricity markets into a more competitive form. The competitive environment results in surging electricity demand and supply that brings in a critical challenge: uncertainty. In this thesis, the uncertainties with respect to prices and demand in the market are explored by using stochastic portfolio optimization and robust optimization techniques. A stochastic optimization model is developed to maximize the overall expected profit in the electricity market by generating possible stochastic electricity supply and demand curves. Stochastic electricity supply curves of prices are generated by using Ornstein-Uhlenbeck mean-reverting process and running Monte-Carlo simulations. In order to overcome the drawbacks of this model, a second model is developed by using robust optimization techniques. This model handles uncertainties both in supply-demand balance of electricity and in renewable energy resources. The supplydemand balance of electricity is explored by using a novel hybrid approach: Wavelet-Multivariate Adaptive Regression Splines (in short: W~MARS). This method forecasts day-ahead electricity prices by considering the challenges such as high volatility, high frequency, nonstationarity and multiple seasonality. Then, we refine W~MARS by a novel robust optimization model, called Robust W~MARS (in short: R~W~MARS), which ensures sustainability and renewability by projecting ellipsoidal uncertainty. The models developed in the thesis are tested by using real electricity market data. Concluding remarks on the models and an outlook to future studies are presented at the end of the thesis.