The interface debonding of the double-base propellant rubber coatings reduces the working performance of the propellant and makes a potential safety hazard to its operation. To solve this problem， a set of tap detection systems composed of signal processing components， automated control components and intelligent diagnostic components was designed and developed， aiming to obtain a more comprehensive acquisition of devices under test through the combination of automated control and intelligent tap diagnosis technology， and it improves the accuracy and reliability of detection. The results show the tap detection system can identify the debonding defects of the rubber coating， and when the tap detection resolution is between 3 to 10 mm， the detection accuracy of debonding defects is more than 87.5%. When the number of neurons in the hidden layer of the back propagation neural network is setted to 6 or 7， the fault recognition effect is good， and the fault diagnosis rate of K-means clustering algorithm to the tap detection data is more than 90%. In summary， the tap detection system has high detection resolution and accuracy， so it can realize the objective evaluation to the bonding quality of the double-base propellant coatings.