Characteristic Analysis and Test of Mixed Pulley Sound Signal Based on Hidden Markov Model

The sliding door pulley produces noise in the process of movement and forms mixed noise after combining with the environmental background noise. The noise produced by pulley movement is related to its structure and material. This article discusses how to use sound signals to analyze pulley characteristics. Noise source is separated by simulating the mixing of a given sound source through the Clustering Hidden Markov Model (C-HMM), and the mechanism of the noise source generated by the pulley is studied, which can provide data support for judging the pulley mass. This paper designs a mixed pulley sound signal perception test platform based on C-HMM. The experimental results show that the system can analyze and judge the quality of the pulley under certain conditions by entering the noise characteristics of sheave motion. The characteristic recognition rate of different combinations is different, and the comprehensive recognition accuracy of the Mel Frequency Cepstrum Coefficient (MFCC) + Short Time Energy (STE) combination can reach 91.76%.