In order to achieve accurate prediction of the stress-strain relationship of frozen soil,researchers have utilized genetic algorithm(GA),mind evolutionary algorithm(MEA),and sparrow search algorithm(SSA)to optimize the initial weights and thresholds of the backpropagation(BP)neural network.Taking the temperature,confining pressure,and axial strain as the main input parameters in the temperature-controlled triaxial test,and the corresponding deviator stress of the axial strain as the output,a BP neural network prediction model optimized by these three algorithms was established.The research results indicate that MEA achieves the best optimization performance for the BP neural network model.MEA-BP has the smallest root mean square error and a high degree of fit(R2)close to 1 between predicted values and actual val-ues,effectively predicting the stress-strain relationship of frozen soil.
NETL
NSTL
万方数据
