DC power flow optimization with a parallel evolutionary algorithm

A full power flow model has been used to analyze, operate, and plan power systems in the steady state. This model enables the active and reactive power flow to be analyzed and planned. With a DC power flow model, an approximated solution can be reached which minimizes the power flow in transmission lines. A method that has been used in recent years for optimization problems is the evolutionary algorithm. It has the disadvantage of great computational effort being needed when it is used to optimize large and complex systems. This paper presents a method which uses parallel computing in order to implement an evolutionary algorithm with multi-populations. In this approach, each computer can receive a population, which uses a different parameter control strategy. Thus, problems inherent in the evolutionary algorithm, such as premature convergence can be reduced, and, thus, its performance enhanced. The proposed method was validated in an active power flow exchange problem in power systems. The efficiency of the method was tested and analyzed in experiments using the IEEE 14-bus system.