Bolt Anchorage Quality Levels Classification Method Based on HO-VMD-CNN-BiLSTM

Abstract

At present, ultrasonic guided wave nondestructive testing technology is widely used in the detection of bolt anchorage defects. Traditional detection methods are confronted with problems such as serious signal noise interference, low detection accuracy, and poor real-time performance. In this paper, a new model named HO-VMD-CNN-BiLSTM is proposed to optimize the accuracy of signal decomposition and quality classification. The model accurately identifies internal defects in bolts through precise signal decomposition and feature extraction, achieving a high classification accuracy of 96.78% even in a complex noise environment. The model incorporates the Logistic-Tent chaotic mapping optimization algorithm, which enhances global search capability, improves feature extraction, and increases detection efficiency and accuracy. The HO-VMD-CNN-BiLSTM model offers an innovative and efficient solution for nondestructive testing of the bolt anchorage quality, enabling high-precision structural assessments while addressing the issues of signal noise interference and feature extraction in traditional inspection methods. By overcoming challenges related to signal noise interference and feature extraction, the model provides technical support for real-time monitoring and intelligent assessment of bolt anchorage quality.