Abstract: Aiming to solve the problem of selecting echo state network parameters, a method for optimizing echo state network (ESN) parameters based on the behavior space is proposed. The essence is to construct ESN behavior space through generalization rank, kernel rank, and memory capacity. The optimization algorithm adopted the novel search genetic algorithm (NSGA), which combines the K-nearest neighbor individual distance and normalized mean squared error, limits the genetic direction of the genetic algorithm by establishing the minimum configuration of the behavior space to screen genes, therefore improves the optimization efficiency. Next, the factors that affect network performance were determined. This method overcomes traditional ESN parameter selection difficulties, the long optimization time of the genetic algorithm, and no suitable theory to clarify the impact of reservoir performance on tasks. As a result, it improved the optimization efficiency and network learning performance. The experimental results showed that the proposed NSGA-ESN method optimizes ESN parameters close to the best network structure, with a better learning performance compared to the growing echo state network. Furthermore, the factors that affect the performance of the ESN network can be explained through the behavior space.