In traditional mind evolutionary algorithm

the searching radius is lack in target and supplement of temporary groups is lack in direction

neural network had slow training speed and weak generation ability

traditional extreme learning machine has many nerve cells in hidden layer. In order to solve these problems

multi-group self-adaptive mind evolutionary algorithm optimized extreme learning machine（ MSMEA—ELM）algorithm is proposed to solve these problems. Aiming at optimizing the adding of weighted mean square error and weighted 2 norm of weights

multi-group self-adaptive mind evolutionary algorithm is introduced to optimize ELM.The demonstration test is conducted on a turbofan engine and MSMEA—ELM algorithm is applied in aero-engine dynamic identification. The result shows that the introduced algorithm is efficient.