The Study Of Vibration Signal Analysis System Based On Wavelet Fractal Theory

With the rapid development of modern industry and science technology, a lot of devices are being large-scale, integration, high-speed, automation and intelligentialize. The status of equipments play an important role in theirs work. Vibration signal is a carrier of state information and it is important symbol of its states. In this paper, the wavelet method and the fractal method are studied. With the above two methods, the theory and technology on vibration signal detection, analysis and processing is discussed. It provides a method of calculation wavelet fractal box counting dimension for vibration signal feature extraction. The result is of important realistic significance to vibration detection fault diagnosis.Signal analysis, which is a process that obtains useful information from signals. Feature extraction method is a research direction followed with great interest duo to the complexity of vibration signals and the combinability of multidisciplinary knowledge for feature extraction. Through analyzing the current research situation both at home and abroad, this paper carried out research work as follows:1. The basic theories about wavelet and fractal are introduced. Compare wavelet analysis method to the traditional Fourier method, because of more superior time-frequency characteristics, wavelet analysis method is more effective in vibration signal analysis. Vibration signal is non-stationary signal and has characteristic of fractal. Fractal theory which is a useful tool to describe the uncertainty and complexity of nonlinear system is effective to analysis vibration signal.2. How to select the suitable wavelet base function is a difficult problem to wavelet analysis, so the use of wavelet transform is much randomness and uncertainty. The results of singular signal analysis have different effect by different wavelet base. Singular value of vibration signal reflects singularity and is calculated. Based on singular value, wavelet base is selected to extract vibration characteristic effectively.3. Connection fractal and wavelet theory, it raises high-frequency coefficients box counting dimension algorithm about vibration signal wavelet transform. According to the changing regularity of the box counting dimension, transformer state feature is extracted, running condition is evaluated, faults are diagnosed more effectively.4. Vibration signal analysis system was designed. System design includes the data acquisition module, serial communication and upper computer module design. CPLD are used to control ADC chip. A/D convert and data storage is controlled by hardware; the speed of conversion is faster without restriction by CPU. The function of high-speed data acquisition and bulk capacity storage is implemented. Using MATLAB graphical user interface (GUI) functions a friend and visible human-machine interface is designed. Collected signal is analyzed and processed by Calling Function.