Shield tunneling parameter prediction method based on geological parameter quantification

Abstract

The invention relates to a shield tunneling parameter prediction method based on geological parameter quantification, and the method comprises the following steps: carrying out the unitization processing of a stratum along the burial depth direction, building a geological condition quantification matrix based on the geological parameters and tunnel parameters of each unit segment, generating training data through the combination of the corresponding tunneling parameters, and carrying out the prediction of the shield tunneling parameters. Establishing a convolutional neural network model based on the training data; and obtaining a geological condition quantization matrix of a to-be-constructed section, and inputting the convolutional neural network model to obtain a predicted value of the corresponding tunneling parameter. Compared with the prior art, the method has the advantages of fully reflecting the complexity of the soil layer, being high in prediction precision and the like.

Description

Technical field [0001] The present invention relates to the field of intelligent control of tunnel boring machines, in particular to a shield boring parameter prediction method based on the quantification of geological parameters. Background [0002] When using a shield machine for construction, the setting of tunneling parameters under different formations relies heavily on manual experience, but it is difficult to set reasonable tunneling parameters through manual experience when the ground layer is more complex, especially in composite formations. [0003] In order to solve the above problems, the prior art discloses some technical means. For example, patent application CN112163316A discloses a method for predicting the boring parameters of a hard rock tunnel TBM based on deep learning, dividing the tunneling section of the shield machine into an empty thrust section, an ascending section, and a stationary segment, and using the ascending section boring parameter data continuo...

AI Technical Summary

Problems solved by technology

[0004] The current research mainly focuses on two aspects. On the one hand, establish the relationship between the tunneling parameters and the tunneling parameters, that is, divide the tunneling parameters of the shield machine into input parameters and output parameters, and use machine learning algorithms to realize the prediction of output parameters, or according to the initial The tunneling parameters of the tunneling section realize the prediction of the parameters of the stable tunneling section. These predictions based only on the tunneling parameters do not fully take into account the change of geological conditions, and are often only suitable for a certain geological situation, and the application range is not wide

On the other hand, the prediction of tunneling parameters is realized based on geological parameters. At present, the quantitative processing of geological parameters is too simple. First, the weighted average is used to calculate the corresponding geological parameters for composite strata, which cannot reflect the complex stratum conditions of the tunnel face. The stratum cannot be reflected in the whole due to its small weight; or when there is a single stratum with the four parameters of natural weight, soil deformation modulus, soil cohesion, and internal friction angle and the composite stratum When the above four parameters are the same after weighting, the existing method will predict the same design value of tunneling parameters. However, there are often large differences in setting tunneling parameters between single stratum and compound stratum; secondly, the existing method only considers the tunnel face The average geological situation does not take into account the geological conditions above and below the working face and the distribution relationship between the upper and lower layers of the working face, and the upper soft and lower hard strata and the upper hard lower soft strata will be different in the setting of tunneling parameters. There are great differences; finally, the existing methods do not consider the connection of geological parameters at different depths when predicting tunneling parameters, that is, the characteristics of geophysical parameters changing with depth cannot be reflected; the quantitative processing of geological parameters adopts simple domain knowledge. Several geological parameter features are accurately extracted, which cannot reflect complex geological conditions, so it is impossible to establish a convolutional neural network model with high precision

Images