An Auxiliary Judgment Method for Nondestructive Testing of Tunnel Lining Combined with Machine Learning

Abstract

The invention discloses an auxiliary judgment method for tunnel lining non-destructive testing combined with machine learning: extracting characteristic parameters from the collected vibration signals; performing regression fitting on the reflection time in the characteristic parameters to obtain calibration values; The parameter represents the original signal, mark this group of eigenvalues, record its defects, and use it as a training set; repeat the above steps on different test objects to increase the number of training sets; use model training software for model training; pass The built model is used to analyze the data of unknown detection results. The invention reduces the adverse effects caused by changes in thickness and material, increases the reflection time on the back of the lining as a defect judgment parameter, can better reflect the defect characteristics, and solves the problem that the existing detection method is greatly affected by the subjective factors of the staff. The problem of poor detection accuracy has achieved the effect of improving detection accuracy, reducing subjective interference of personnel, and ensuring objective and accurate detection results.

Description

technical field [0001] The invention relates to the field of civil engineering inspection, in particular to an auxiliary judgment method for tunnel lining non-destructive inspection combined with machine learning. Background technique [0002] my country has a vast territory and complex and changeable geographical conditions. Most tunnels are constructed in alpine jungles, and are affected by many factors such as geological environment, construction environment, construction technology, design, management, etc. If the construction level is not enough or the construction technology is not Specifications or lax procedures and lack of management can easily lead to insufficient thickness of tunnel lining, loose contact between lining and surrounding rock, voids, deformation and cracking of lining, water leakage, and even block loss. These problems are difficult in the early stage of tunnel construction. These quality problems have left a huge hidden danger to the safety of the tun...

AI Technical Summary

Problems solved by technology

In the tunnel lining detection, the following problems will occur: (1) The pressure of the sensor also has a certain influence on the test results; (2) The surface flatness of the measured object is greatly affected. When there are other impurities such as pulp, the test error will become larger; (3) The natural vibration characteristics of the sensor and the vibration pickup system will affect the test signal;

[0008] Regardless of the elastic wave, ultrasonic wave, electromagnetic wave, etc., one or two target parameters are detected according to the detection requirements of the test object. The detection parameters are relatively single, and the analysis method is also single, and a large number of relevant parameters cannot be truly utilized;

[0009] Existing detection methods have high requirements for operators. Due to the uneven and unknown thickness of the tunnel lining, different technicians may obtain very different results by using the same equipment to detect the same object. data analysis experience

At the same time, many judgments do not have clear benchmarks and thresholds, and there are large randomness and individual differences;

[0010] At the same time, many tests require technicians to go back to the room and use the analysis software to follow the relevant steps step by step and analyze the data one by one. It is difficult to achieve rapid automatic data processing and real-time feedback of test results.

[0011] Since it is difficult to centralize the data of various units and personnel, and the verification results cannot be summarized, it is difficult to quantify the accuracy and confidence interval of the test results, and the objectivity is poor

Images