The Research And Application Of Mechanical Vibration Signals Analysis Algorithm Based On Wavelet Transform

With the rapid improvement of science and technology, machinery equipment is becoming high-speed,integrated, automated and complex. And the health status of mechanical equipment is becoming more and more important. The mechanical vibration signal can directly reflect the running status of the system, because it carries important information of the mechanical equipment’s work state. Therefore, people begin to analyze and research the vibration signal deeply. Subsequently, useful imformation is extracted from the said signal. These results can provide a strong evidence for health monitoring of system.There are a lot of methods to process the signals, including the wavelet transform which is used widely. Wavelet transform is a processing method with the characteristic of multi-resolution analysis, and has the ability of localizing information in time domain and frequency domain. The non-stationary signals and the aperiodic vibration signals can be effectively processed by this method. In addition, Wavelet Transform can highlight the fault feature. So it has the reputation of“Mathematical Microscope”. Accordinglly, wavelet transform is used to process and analyse the mechanical vibration signals in this thesis.As a part of the subject“Data Acquisition and Signal Processing System”, wavelet transform is adopted to process the mechanical vibration signals based on analysis of the existing fault diagnosis method and mechanical signal feature in this dissertation. And then, the characteristic information is extracted, which provides an important judgment for fault diagnosis. When the signals are processed with wavelet transform, it is important to determine the wavelet function. So, the dissertation implement comprehensive research, and makes selection principles. Then, DSP based algorithm of wavelet transform and the optimization measures are deeply studied. And the optimization results are contrasted. Finally, rate of code performing can reach 195k sampling points per second. The experimental results indicate that the optimized algorithm can accurately and real-timely process vibrational time-frequency signals to meet the requirements of the project.