In order to realize aircraft engine sensor fault monitoring and diagnosis more better on embedded ARM11 platform

using extreme learning machine（ ELM） instead of BP network algorithm

only need to select node numbers and activation function of hidden layer

greatly reduce large numbers of parameters

slow training process and continuously adjust network parameters and easy to fall into local optimal solution of BP algorithm.Through simulation comparison of two algorithms

verify the superiority of ELM algorithm. The ELM algorithm is then converted into aircraft engine sensor fault diagnosis training and learning software and diagnosis software using C + + programming language

the final test show that test precision of the algorithm software is good

and meet requirement of diagnosis.