Applications of AI techniques to generation planning and investment

• Published in: 2004 IEEE Power Engineering Society General Meeting pp. 936 - 940 Vol.1
• Main Authors: Wu, F., Zheng Yen, Yunhe Hou, Yixin Ni
• Format: Conference Proceeding
• Language: English
• Published: IEEE 2004
• Subjects: Artificial intelligence; Dynamic programming; Electricity supply industry; Investments; Large-scale systems; Load forecasting; Mathematical programming; Power generation; Power system planning; Power system reliability
• ISBN: 9780780384651; 0780384652
• DOI: 10.1109/PES.2004.1372969

The generation planning and investment problem in restructured industry is to determine what, when, where and how to install generating units to supply electricity to the power system, while satisfying various constraints imposed by load forecast, reliability and other operating conditions, in order to maximize investors' profits and minimize the investing risks. Mathematically, a GP problem can be expressed as a large-scale, nonlinear, mix-integer stochastic optimization problem with the objective of maximizing the profit and minimizing the risk, subject to a set of complicated constraints of load demand and supplying reliability. It is a challenging problem due to the combination of non-linearity, combinatorial and randomness. Traditional approaches are based on mathematical programming methods, such as dynamic programming, mix-integer programming, etc. In most cases, mathematical formulations have to be simplified to get the solutions, due to the extremely limited capability of available mathematical methods for real-world large-scale generation planning problems. The other type of approaches is based on artificial intelligence (AI) techniques. The major advantage of this second type of approaches is that they are relatively versatile for handling various qualitative constraints that are prevalent in generation planning problem in the restructured power industry. This panel paper is devoted to a review of the state-of-the-art of AI techniques to generation planning and investment problems. Several AI-based methods have been applied to the problem: simulated annealing, genetic algorithms, ant colony optimization method, particle swarm optimization method. The convergence issue of such methods are discussed and their applicability to the generation planning and investment problem is analyzed.