I. Introduction
Application of capacitors on distribution systems produces benefits like reduction of power and energy losses, release of generation, feeder and sub-station capacity, and improves voltage regulation of the system. Benefits can be maximize by optimizing the location, number, and sizes of shunt capacitors to be place

Numerous researches were done on optimal capacitor allocation. Such researches differ from each other by their problem formulation, methodology and assumptions employed. Ng, et al.[1] classify the optimum capacitor allocation methods into: (1) analytical methods, (2) numerical programming methods, (3) heuristics methods, and (4) AI-based methods. Each methodology has certain advantages and drawbacks. It was concluded on their works that a switched capacitor is indeed needed for distribution system where the load level varies with time.

On the other hand, the recent popularity of AI has led many researchers to investigate its use for power engineering applications. AI seeks, as its main goal, to create artificial system which mimic aspects of human behavior, such as perception, evolution, memory, learning, adaptability and reasoning. Moreover, AI techniques are simpler to implement. One of these AI methods that are commonly employed for combinatorial optimization problems is Genetic Algorithm (GA). GA model emulates biological evolutionary theories: genetic inheritances and Darwinian strife for survival. These models require four basic elements; Initial population, evaluation of the fitness, selection, and genetic operators which includes crossover and mutation.

In this paper, a method of optimal capacitor allocation using AI-based approach is proposed. There are many AI- based methods like Genetic Algorithms (GAs), Simulated Annealing (SA), Fuzzy Logic, Tabu Search and Artificial Neural Network (ANN). The project has focused only on Genetic Algorithms (GAs) as the AI-based method used in our optimization problem. The method was only applied on radial distribution networks characterized by unbalanced line-to-neutral loads, varying loads over time, contains three-phase, two-phase and single-phase lateral feeders, and involves mutual coupling between conductors. For practical reason, only balanced three-phase capacitor bank was applied on buses with all of the three-phase wires present and a single-phase shunt capacitors for single-phase or two-phase laterals. Since the primary objective of the project is to maximize the savings through optimal capacitor allocation, we did not consider the lower voltage limit as a constraint. However, during the search process, GA considers the upper voltage limit as a constraint that causes penalty to the fitness function.

Introduction
Problem Formulation and Solution
Implementation, Testing and Results
Testing and Results (continued1)
Testing and Results (continued2)
Conclusion and References