Abstract: Due to some defects of BP neural network,the power size and the threshold value of the network structure are optimized by combining genetic algorithm with neural network.The estimation model of bolt bearing capacity is accordingly built based on the improved optimization algorithm.In this model,five low strain variables from dynamic testing are used to estimate the bolt bearing capacity.The calculated results are compared with those of BP neural network.The presented example shows that the genetic neural network is of both higher computing efficiency and higher identification accuracy in estimating the bolt bearing capacity.