In this paper,a method that combines the salp swarm optimization algorithm with extreme learning machine and adaptive differential evolution algorithm is proposed to optimize the parameter configuration of the multi pump Raman fiber amplifier.The nonlinear mapping between pump parameters and Raman gain is constructed by extreme learning machine,and the optimal model is obtained by optimizing the parameters of extreme learning machine by using salp group optimization algorithm.The differences between the above model and BP neural network and the traditional extreme learning machine model in terms of evaluation indicators are compared and analyzed,and the results show that the model proposed in this paper has better prediction performance.In order to improve the gain flatness,an adaptive differential evolution algorithm is used to optimize the pump parameters to obtain the best parameter configuration.The simulation results show that the Raman amplifier designed by this method achieves the expected effect,and the maximum error between the target gain and the predicted gain does not exceed 0.5 dB.This method provides a new way of thinking for the design of Raman fiber amplifiers in the future.