Monitoring and abnormal diagnosis of flood discharge vibration state of radial gate based on improved permutation entropy

ABSTRACT This study proposes an improved multivariate multiscale weighted permutation entropy (IMMWPE) combined with correlation variance dedication rate to monitor radial gate vibrations under discharge excitation, ensuring safe gate operation. A non-parametric test based on kernel density estimation is used to determine alarm thresholds for abnormal vibrations. The dung beetle optimization (DBO) algorithm optimizes the penalty coefficient and kernel parameters, constructing a least squares support vector machine (LSSVM) model to diagnose gate vibration states. Results show that IMMWPE offers greater accuracy and anti-interference capabilities than existing methods. This approach enables efficient data-level dynamic fusion of multi-sensor vibration signals for real-time monitoring of gate vibrations induced by flood discharge, providing early warnings of abnormal vibrations. It also diagnoses fault types in gate discharge vibrations through IMMWPE–DBO–LSSVM, improving diagnostic accuracy. This method offers a new solution for vibration monitoring during gate operation.