Abstract:Rolling bearing is the main part of rotating machinery, prone to various faults. These faults will bring a series of potential safety hazards, and cause economic losses. Therefore in industrial production,research about rolling bearing fault diagnosis is very important, with great use.Taking the rolling bearing as the research object, the rolling bearing acoustic emission experiment platform was established,bearing acoustic emission signal under different defects and different rolling speed was collected.Based on HMM theory and algorithms, the acoustic emission signal was processed. MATLAB software was used to extract data,then they were framed, and the feature vector was extracted, the likelihood probability was gotten. The likelihood of rolling bearings with different defects was fitted, the fitting formula between likelihood and rotational speed was obtained.Two kinds of acoustic emission signal under different speed and known defects were tested,a kind of acoustic emission signal under some defects was diagnosed. The result shows that, HMM can be used to make fault diagnosis of the rolling bearing effectively.