Market simulation in electric power system

In this dissertation, we contribute to current debates in the context of electricity restructuring. First, we study the constrained economic dispatch (CED) problem considering a nonsmooth cost function. Nonsmooth characteristics of generator cost functions, representing prohibited operating zones, multi-fuel, and valve-point effects are considered in CED. Mixed-integer programming (MIP) is applied to the CED formulation and the MIP solution is compared with a heuristic approach for smoothing the cost function.; In the second part of the research, a new approach based on artificial neural network (ANN) is presented here to forecast the next 30-minute load in 5-minute intervals. The approach includes three ANN modules in which each module captures various trends in load data. The construction of forecasters is based on a multilayer perceptron trained with backpropagation learning rule. In part three, we introduce a hybrid model based on ANN and fuzzy system to forecast monthly peak load for next several years. Results can be used to generation and transmission planning.; In last part, a fuzzy-neural network (FNN) model is applied to predict hourly electricity prices based on historical prices, demands, and reserves. We consider a few alternatives for improving training and forecasting tasks. The hourly price forecasting results of the proposed FNN model are presented and compared with those of ANN and conventional techniques.