Abstract:

In order to solve the problems of complex network structure parameters and slow operation speed in traditional intelligent algorithm, a new intelligent prediction model based on genetic algorithm and extreme learning machine is proposed. Pearson correlation coefficient is used to evaluate the correlation between different influencing factors and the deformation of foundation settlement, so as to determine the input layer of extreme learning machine. And then the optimal excitation function and the number of hidden nodes are determined by trial method; and the genetic algorithm and extreme learning machine are coupled. The genetic algorithm is used to optimize the initial weights and thresholds of extreme learning machine to improve the prediction accuracy. The results show that: (1) Excavation time, excavation depth, soil shear parameters and unit weight are significantly related to the foundation pit settlement deformation, which provides a basis for the construction of extreme learning machine input layer. (2) In the prediction process, excitation function and the number of hidden nodes have a certain influence on the predictive effect of the extreme learning machine; and the predictive effect of Sigmiod excitation function and 13 hidden layer nodes is optimal. (3) The optimization of algorithm can further improve the prediction accuracy, and verify the optimization ability and effectiveness of genetic algorithm. The prediction results of the model under different conditions are superior, so as to show that the prediction model has high stability and reliability.

Key words: foundation pit, Pearson correlation coefficient, extreme learning machine, genetic algorithm, deformation prediction