Intelligent Planning Approaches for Electricity Generation and Distribution

To operate power generation and distribution industries efficiently and economically, their management must deal with a number of challenging problems. Of them, dynamic economic dispatch (DED) and bidding problems are two important topics. The purpose of a DED problem is to schedule the available generators to satisfy the daily load demands at minimum cost while that of a bidding one is to maximize the individual profit of an energy market by determining the optimal action of each participant. Over the last few decades, although these problems have been extensively studied, they mainly dealt with the thermal power plants while ignoring the renewable sources and their uncertainties. This thesis considers the mix of different thermal, hydro, solar and wind generators with their uncertainties. For solving these problems, although many solution approaches have been developed, the evolutionary algorithms (EAs) achieve the best results. However, no single EA performs consistently over a wide range of these problems. Also, because of their dimensionality, non-convexity, multi-modality and large number of equality constraints, current EAs are inefficient for solving them. Moreover, most existing methods for solving a bidding problem aim to find a single solution whereas detecting multiple ones is more practical and challenging. In addition, the uncertainties of renewable sources pose a new challenge for the electricity generation and distribution sectors.In this thesis, a general EA framework based on two EA variants, a self-adaptive differential evolution and real-coded genetic algorithm, is proposed to solve DED and bidding problems. To enhance the convergence rates of the proposed algorithms, a heuristic technique for repairing infeasible individuals while solving a DED problem is developed. For bidding problem, a co-evolutionary approach that detects multiple solutions in a single run is implemented. The effectiveness of the proposed approaches is evaluated on a number of bidding and DED problems considering the uncertainties of renewable generators. Comparisons of the simulation results with each other and those from state-of-the-art algorithms reveal that the proposed methods have merit in terms of solution quality and efficiency