In view of the complex characteristics such as nonlinearity and time-varying in modern industrial control

the traditional extreme learning machine（ELM） cannot deeply mine data information

which leads to the influence of model parameters on soft measurement prediction precision.The rate parameter and displacement parameter of the activation function in the extreme learning machine and the multiplication factor in the ridge regression have an important influence on the prediction precision of the modeling algorithm.In order to improve the prediction precision of the soft-measurement model

shuffled frog leaping algorithm（SFLA） is used to accurately search the parameter optimal solution in ELM.SFLA can effectively solve the instability of ELM prediction

thereby increasing the reliability and accuracy of model prediction.The extreme learning machine modeling method based on the hybrid frog leaping algorithm is used to predict the content of butane component in the bottom of the debutanizer column and the biological oxygen demand in the sewage treatment.The simulation results show that the SFLA-ELM method is feasible and effective.