State identifying method for rolling tire in lab test using acoustic signal

Tire bulges result from internal component damage in a lab endurance test. Considering the time delay and uncertainty of the traditional identification method for monitoring tire bulges, this paper proposes a state identifying method for the bulge question in tire endurance tests based on acoustic signals of rolling tires, which has the characteristics of non-contacting and early diagnosis. Firstly, the hardware selection was built, and the improved Spectral Subtraction (SS) method was proposed to effectively extract the pure acoustic signal of rolling tires from the background noise in the lab tire endurance tests. Secondly, it analyzes the limitations of MelFrequency Cepstral Coefficients (MFCC) and proposes the utilization of Short-Time Energy (STE) and improved Mel-Frequency Cepstral Coefficient (SMFCC) to represent acoustic signal features. Thirdly, based on the signal feature, the acoustic signal is divided into frames, and then the healthy date set and bulge date set are built to state identifying, and the Support Vector Machine (SVM) model employing STE and SMFCC is developed to identify rolling tire bulge, and the state identifying accuracy rate of the proposed method increases with the increasing of Signal-to-Noise Ratio (SNR); Finally, the application of identifying method is expanded to identify the tire fault in high-speed test condition, and the diagnosis accuracy rate reaches up to 98.6%. The results show that the method proposed using acoustic signals has a bright prospect for use in tire industrial applications for fault diagnosis.