Local fisher formation identification method based on shield tunneling parameters

Abstract

Neural network technology and statistical classification are often used to identify the stratum in the traditional shield construction. However, the linear relationship between the parameters of shield tunneling is not ideal after analyzing the parameters of shield tunneling by curve fitting and statistical regression. Is adopted in this paper, in order to improve the accuracy of formation to identify local Fisher discriminant analysis method, through the data in the construction process of shield machine, the real-time recognition: formation of randomly selected sample of 228 set of training samples to train the model, 68 groups of data, which can identify using KNN classifier to classify, soil layer identification accuracy of 80-100%. The experimental results show that the local Fisher formation identification model based on shield tunneling parameters can realize the online stratification discrimination in the process of shield tunneling, improve the construction efficiency of shield tunneling machine, and reduce the construction risk and cost.