Parameter estimation of a predator-prey model using a genetic algorithm

This paper presents the parameter estimation of a biological system with real data extracted from literature, and different model structures: Lotka-Volterra model (basic predator-prey model) and saturation predator-prey model. A third model is proposed and tested in simulation due to lack of appropriated real data; it includes inputs that excite the system and makes the estimation process more manageable. The model parameters were estimated using a genetic algorithm, which gives a combination of parameters used in simulation to compare outputs with real data and decide, using a cost function, which parameters are better. Comparing the models, the Lotka-Volterra model provides better adjustment but with unrealistic assumptions, while saturation model represents a system with real assumptions but the fit is not very high.