Skip to Main Content
This paper proposes an accurate method that combines a hybrid genetic algorithm (GA) with a geometric compensation technique applied to an analytical Raman amplifier model to obtain the optimal design of multipump distributed Raman amplifiers. The geometric compensation enables partial determination of the GA initial population and works as a refinement in the search of the best possible solutions. We develop a self-contained algorithm that is capable of meeting on-off gain and ripple specifications to broadband Raman amplifiers, without the need for a previous study of the search space. As a result, we determine wavelengths and powers using the minimum number of pump lasers necessary to meet the given specifications. Our method has shown to be robust in the simultaneous analysis of multiple parameters and multiple objective problem. The processing time required to design large bandwidth Raman amplifier is minimized when compared with that obtained by a standard GA.