| The large-scaled rotating machinery is widely used in petrochemical engineering, aeronautics and astronautics, metallurgy, and power generation industry. To avoid the disastrous accident and the huge property losses brought by the sudden failure of the rotating machinery, it is very common, and also efficient to conduct the fault diagnosis and early warning for the vibrating signal of the running status of the rotating machinery. The current researches indicate that it is relatively mature and complete to obtain and analyze the acquired signals when rotating machinery is running stably. However, there is not many complete researches on the unstable signals in the starting and stopping period of the rotating machinery. The quality of signals acquired in these period is commonly low, which can negatively affected the systems to make accurate fault diagnosis in these periods when the accidents are prone. To guarantee the fault diagnosis system can achieve the high-quality signals in the starting and stopping periods, it is necessary to develop a filter which can dynamically deal with the signals acquired in the starting and stopping periods.To deal with the problems above, this paper put forward a automatic tracking filter based on FPGA, which can automatically filter out the noise existing in the acquired vibrating signals dynamically. This filter realize the real-time process of the vibrating signals in the starting and stopping periods.And the modification of the filter structure can make the output signal switch without disturbance, which greatly retain the useful information in the vibrating signals. The main contents are as follows:1. Firstly, This thesis briefly introduced some basic knowledge about fault diagnosis, and pointed out the drawbacks in the starting and stopping period.2. This paper also analyzed the general requirement of the system, and put forward the general design of the system.3. This paper introduced the process of the hardware design, and pay more attention to the design of the measurement circuits of the key-phase and the vibration signal. And it also pointed out the selection of some vital devices. Through the test, we verify the validation of the hardware.4. Correcting the rotating speed when the pulse of rotating speed is measured. The first-order numerical differentiation five-point interpolation is employed to compute the real-time speed. The method realized by FPGA has a small calculated amount which can efficiently increase the computing speed and accuracy. This method can satisfy the requirement of the real-time. The module of real-time filter coefficients calculation has been developed based on the soft core--Nios â…¡. This module can calculate out the real-time filter coefficients based on the acquired rotating speed. The filter module whose coefficient can be modified in real-time has been developed. This module can adjust to its coefficients base on the rotating speed. And the employment of the state machine help to realize the switch of the output signal without disturbance.5.Finally, all modules are integrated to form an automatic tracking filter system. Through being tested by the experiments and simulations, these results have indicated the viability and validity in the real running. |
