Corrected Local-Projection Method For Noise Reduction And Study On Its Application To Fault Diagnosis

Since the 20th century, with the development of industry production and science and technology, the reliability, usability and maintainability of any mechanical equipment has been more and more important. So the research on failure machinery and diagnosis technology to mechanical equipment is being advanced. In recent years, especial importance is attached to status monitoring, malfunction diagnosis and prediction of complex mechanical equipment.I have introduced the noise reduction algorithmic based on correction projection into gear's fault diagnosis. The result shows that this algorithmic has a great advantage on authenticity and availability of gear's fault diagnosis. Through the reconstruction of phases pace, a one-dimensional time-series of some single variable can be embedded in a high dimensional phase space, which shows the whole information of the system. In a certain condition, the space can keep its equivalence to the original space that ensures the reconstruction of the dynamical characteristics.Using technique of the nonlinearity time queue phase space reconfiguration, we projected one dimension time series including fault messages onto the high dimensions space and selected appropriate number of constrains and the minimal neighborhood to constitute the covariance matrix and calculated the corrections. By iterated some times, the clean dynamical time series was obtained. Some simulated experiments such as sine signal, Lorenz and Henon map including noise were made to prove that this algorithm had strong effects on noise reduction. Apply this algorithm in fault diagnosis of bias gear.