A study on the recognition algorithm for loose screws in converter valve saturation reactors based on continuous wavelet transforms

With the complexity of modern power systems, the application of converter valve systems in high-voltage DC transmission is becoming more and more widespread. As a key component of the converter valve, the operation status of the saturation reactor directly affects the stability and reliability of the system. Screw loosening is a common fault in saturation reactors, and traditional detection methods are difficult to accurately detect screw loosening faults in real time and are all contact monitoring. This study proposes a screw loosening recognition algorithm that combines wavelet transform and grayscale fusion techniques to analyze the acoustic noise of screw loosening in saturated reactors by wavelet transform, and then deepen the extraction of time frequency features by fusing grayscale maps, so as to realize the accurate recognition of screw loosening. Experimental results show that the combined method can effectively and accurately identify and localize various disturbances, which significantly improves the operational stability and reliability of the converter valve. Comparative analysis with traditional methods shows that the method has advantages in detection accuracy and computational efficiency, and is suitable for real-time online monitoring of the converter valve system.