Vibration feature extraction and fault detection method for transmission towers

This paper presents a novel bolt looseness detection method for power transmission towers based on vibration signal analysis. The proposed method utilizes pulse excitation to extract the vibration signal of the tower, which is then adaptively decomposed using the Variational Mode Decomposition of Spider Wasp optimizer (SWVMD). This overcomes limitations of traditional Variational Mode Decomposition methods by leveraging bio-inspired optimization to improve signal decomposition. Simulated signals processed with different optimization methods verify the superiority of the SWO approach. Field tests on a 110-kV transmission tower further demonstrate the effectiveness of the proposed SWVMD technique for analyzing on-site vibration data. A new improved intrinsic multiscale sample entropy feature is also introduced for bolt state characterization. A Spider Wasp Support Vector Machine classifier is developed to realize accurate bolt loosening monitoring using the extracted features. Dynamic response tests under varying bolt conditions show that the method can identify early loosening and reduce tower damage risks compared to conventional techniques. This novel vibration-based detection framework presents an innovative application of nature-inspired computing for power infrastructure health monitoring. This paper presents a detection method for tower bolt looseness based on vibration responses. In this method, the tower vibration signal is extracted by pulse excitation, and the collected signal is adaptively decomposed by Variational Mode Decomposition of Spider Wasp optimizer. The IIMSE is selected as a new feature index, and the SWSVM method is used as the classifier to extract the feature set to realize the accurate monitoring of the bolt state.image