As a new type of neural network, Extreme Learning Machine (ELM) has the advantages of fast learning speed and good generalization ability.However, due to the random generation of its input weight matrix and hidden layer bias, it is easy to cause network instability and large deviation.An Improved Differential Evolution (IDE) multi-objective optimization algorithm is introduced to obtain the input weight matrix and hidden layer bias of the ELM with the smallest training error, so as to improve the performance of it.Finally, the optimized extreme learning machine is applied to fault diagnosis of a rudder system.The results show that the improved ELM has high learning speed and diagnostic accuracy.