A Novel Bolt Looseness Detection Method Based on Dynamic Mode Decomposition and Multi-Sensor Information Fusion

For most currently looseness detection methods of bolts, the signal feature extraction is thoughtlessness of the impact of structural dynamics and signal coupling relationships. To overcome this, a novel method based on dynamic mode decomposition and multi-sensor vibration information fusion is proposed for bolt looseness detection. Firstly, data enhancement methods are adopted to increase the data sample, and the dataset is segmented. Moreover, dynamic mode decomposition is harnessed to obtain the corresponding mode matrix. Subsequently, the mode matrix data samples are used as the inputs of the convolutional neural network, and a Softmax classifier is utilized for damage scenario detection. Finally, a thermal-mechanical-vibration coupling experiment system is built to conduct experimental verifications. The experimental results show that the average detection accuracy of the proposed method can reach more than 98.7<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>, which is much higher than other methods, possessing of obvious clustering effects, high detection accuracy and rapid convergence.