Research on the Tightness Diagnosisand Network Security Risk Protection of Slot Wedge Tightness for Generator Wall-Climbing Robot Based on Multi-Scale Frequency Band Energy Entropyand CNN-LSTM

Abstract

During its operation, the generator stator is prone to vibration as a result of the joint effect of powerful centrifugal force and significant electromagnetic stress. This phenomenon may potentially give rise to significant incidents when the slot wedge is found to be loose. For an extended period, the detection of a loose slot wedge in a generator has necessitated the removal of the end cover and extraction of the rotor, followed by tapping and inspection by a designated professional. This particular process has been noted to demand a substantial quantity of human and material resources, thereby leading to a rise in operational expenditures. The present study showcases the creation of a space-efficient inspection robot, engineered to access the air gap of a sizable generator stator without the necessity of rotor pumping. The robot is furnished with a percussive head, allowing for the assessment of slot wedge looseness. Additionally, this study puts forward a technique for extracting the characteristics of slot wedge tightness using multi-scale frequency band energy entropy,which proficiently dissects the energy dispersion of the sound signal. Subsequently, a CNN-LSTM algorithm is employed to identify the faults of the extracted features. It is contrasted with a CNN and LSTM algorithms under identical experimental circumstances. The outcomes signify that the method proposed in this study exhibits greater precision.