Abstract: Aiming at the diagnosis of debonding damage of honeycomb sandwich structure, the sensor network was first constructed by integrating piezoelectric ceramic sensors, and the ultrasonic guided wave weighted distribution diagnostic imaging method was used to locate and diagnose the damage in the plane. The sensitivity difference of the debonding layer was used to extract the damage characteristics; finally, a large number of guided wave propagation simulations were carried out through the finite element model of the honeycomb sandwich structure to form a training database, and then a stable support vector machine (SVM) debonding layer classification machine learning model was formed. Diagnosis of intra-section debonding layer. The verification test results show that the method can effectively diagnose the debonding damage of the honeycomb sandwich structure, the positioning error in the plane is less than 2 cm, and the diagnostic accuracy of the debonding layer in the section was 100%.