Vibration detection and feature extraction for bolt loosening of transmission tower

Bolt connection failure of transmission towers poses threats to power grid operation safety. Aiming at the limitations of high cost and long cycle for full-scale tower tests, this paper takes a 330 kV dry-type terminal tower as the prototype and develops a 1:25 geometric scale-down test model with a height of 2.2 m according to similarity theory. Comparative tests on bolt loosening are carried out by active excitation via hammering the tower foot. Acceleration amplitude spectrum, sub-band energy ratio and spectral entropy are adopted to analyze vibration signals, so as to explore the variation rules of vibration response before and after bolt loosening. Different from acoustic-based methods that identify loose bolts through high-frequency radiation features and sound-source localization, this paper focuses on low-frequency modal-band vibration signals and realizes quantitative evaluation of the bolt-loosening state. The test results show that after bolt loosening, the spectral pattern within the 40-200 Hz band changes from “single-peak prominent type” to “multi-peak dispersed type”. The vibration amplitude of mid-low frequency rises, and vibration energy migrates from high-order modes to low-order modes. Both sub-band energy ratio and spectral entropy exhibit favorable channel consistency and damage discrimination capability, and can complement each other as quantitative detection indicators for tower bolt loosening. This research can provide experimental references for vibration-based detection of bolt loosening in transmission towers.