The Design Of Vibration Analyzer Based On FPGA

Any rotating equipment in the daily life will produce vibration in the operation. Normally vibration can be harmful, because it will cause damage to equipment and even cause serious accidents. As the carrier of fault characteristics, the vibration signal contains abundant information of mechanical running condition. The accuracy of the vibration signal analysis and processing have very important significance for condition monitoring, evaluation and fault diagnosis of rotating machinery.As the vibration signals of the device are usually non-linear and non-stationary, the traditional analysis methods have many limitations in the analysis of vibration signals. The Hilbert-Huang transform which was proposed by Huang et al. in 1998 stand out with its strong advantage in processing non-stationary random signal. For decades, HHT has been applied to the Marine data, seismic signals and device status detection and diagnosis. Field Programmable Gate Array(FPGA) has the powerful capacity in the digital signal processing and computing. With its internal high-speed parallel mechanism and rich logic resources, FPGA can accomplish hardware implementation of advanced algorithms such as FFT, HHT. It can also perform high speed, high efficiency, real-time analysis and processing on the vibration signal.In the view of this, in this dissertation based on the Hilbert-Huang transform, we focuse on the design and implementation of vibration signal analyzer on the FPGA platform. The main research contents as follows:(1) We discuss the pros and cons of the current popular various vibration analysis methods, and researches the basic principle of the Hilbert- Huang transform in depth. It selects the appropriate algorithm for every step after having a deep understanding of the process of Empirical Mode Decomposition(EMD).Furthermore,we optimize the structure by grasp the nature of the Hilbert transform.(2) According to the design requirements of system, taking into account computational complexity of algorithm, the data storage in the process of collecting data and the amount of data transmission, we reasonably select the FPGA chip, carry out the allocation of pins and resources, together with the component selection and task allocation.(3) We realize the envelope fitting method based on the cubic spline interpolation method, and the suppression method of endpoint effect based on the polynomial fitting. Finally we also realize the EMD method successfully in a FPGA chip. After having a sound grip of the essence of the Hilbert transform, based on the distributed algorithm we implement the Hilbert filter, and optimize the structure by using the linear phase characteristic of FIR filter and the segmentation of the lookup table method.(4) We verify the whole design by different simulation signals. The comparisons with the MATLAB simulation results show correctness and effectiveness of the design. At the same time, we apply the design to the monitor industrial motor vibration to give preliminary diagnosis of the motor working condition by using the spectrum analysis of the vibration signal.