Article Preview
Top1. Introduction
Ever increasing energy needs across the world resulted in vast expansion of electric grids and utilities (IEA 2013). Thus, effective management of resources across the network is of vital importance. Consequently, unit commitment problem (UCP) hold the key for effective management of the technical and economic aspects of the power system planning and operation (Carlos Batlle et.al. 2010). The UCP can be visualized as a cost minimization problem accompanied by appropriate ON/OFF states and effective power sharing among the generating units to arrive at minimum cost of operation yet satisfying all system and generator constraints (D. Bertsimas et.al. 2013). Thus, the UCP problem has evolved into a complex optimization problem whose diversity was expanded into various research fields (B. Saravanan et.al. 2013).
In literature, the UCP problem is solved using various optimization algorithms from past decades. Techniques like dynamic programming (DP), mixed integer linear programming (MILP), branch and bound method (BBM), priority list (PL) and Langrangian relaxation (LR), algorithms are proposed in earlier stages of research as per UCP is concerned. However, every algorithm has its strengths and limitations. Among these algorithms, PL offers simplicity and least computational time, but a compromise has to be made with the quality of final solution (E. Delarue et.al. 2013). In case of DP, lack of flexibility and increased computation time at higher dimension turns out to be the disadvantageous in executing UCP for wide range, large number of generating units (C. Wang and S.M. Shahidehpour 1993). While, exponentially compounding computational time with respect to system size hampers the application of MILP and BBM algorithms for UCP in large networks (Ana Viana and Joao Pedro Pedroso 2013). Lastly, when executed with LR, UCP being a non-convex optimization problem, leads to difficulties in selection of feasible solutions (Quanyuan Jiang et.al. 2013).
In recent times, heuristic optimization approaches became very popular due to their simplicity and efficiency in finding global optimal solution. In earlier studies, genetic algorithm (GA) mimicking the biological mechanisms like natural selection, crossover and mutations were used to solve UCP problems (S. Kazarlis and A. Bakirtzis 1996). Also, hybrid version of hybrid approaches like LRGA are also proposed and applied to reduce the search space of UCP (Chuan-Ping Cheng et.al. 2000). Thereafter, inspired by swarm intelligence like social behavior and coordination principles, particle swarm optimization (PSO) is applied to enhance the quality of UCP (Zhao B et.al. 2006). Further, to allocate ON/OFF status of generation units different variations of PSO like Lagrangian relaxation particle swarm optimization (BPSO), quadratic binary particle swarm optimization (QBPSO), improved binary particle swarm optimization(IBPSO) are introduced into UCP (H. Balci and J. Valenzuela 2004; Lokesh Kumar Panwar et.al. 2014: Xiaohui Yuan et.al. 2009). Thereupon, techniques likes ant colony optimization (ACO) derived from natural behavior of ants in finding the shortest path in search of food, are applied in effective commitment and scheduling of generators at minimum possible cost (K. Vaisakh and L.R. Srinivas 2011).