Rapid Analysis Diagnosis And Application Of Non-sationary Signal

A lot of non-stationary signals can be found in nature, such as once mechanical equipment fault appeared it would lead to the dynamic phenomenon of non-stationary signal. Traditional time-frequency analysis methods work in different side to decomposition this signal, but their respective defects are hard to realize adaptive decomposition according to the characteristic, and it is difficult to achieve precision requirement. Empirical mode decomposition solution is a new way in the signal processing area, through the screening standard to select different scales signal, and then cooperate with Hilbert transform to extract the time and frequency information. But in the process of using mode decomposition method there are four disadvantage problems, although a variety of improve ways are put by many scholars, but it still can’t solve the four disadvantages at the same time. For example, EEMD solution can solve the modal aliasing effects, but it will cost a long time.Mechanical equipment of line condition monitoring systems has developed for a long time, and become to the intelligent expert. As large equipment becomes complicated, continuous work, high speed, has puts forward new requirements on the line condition monitoring system. How to extract the fault feature information quickly and effectively in the complex signal environment, is a main content during this whole paper.On the basis of traditional EMD algorithm and according to tow-dimensional fast adaptive empirical mode decomposition method, this paper has proposed a progressive solution, The main contents include:First, this paper has proposed a progressive one-dimensional fast adaptive EMD solution. First of all, the signal is acquired by means of the criterion of butterfly window of pick values of statistical properties; Then, by way of order statistics filter to get up and down envelope; Finally, put forward the shifting stop criterion, which is used for stopping decomposition. The biggest advantage of this algorithm is to avoid the traditional four major disadvantages of the EMD algorithm, along with the ability of adaptive processing.Second, the one-dimensional fast adaptive EMD solution can be applied to the fault diagnosis of gearbox, bearing fault diagnosis and rotor fault diagnosis. Compared with traditional EMD solution, the results show the superior performance of one-dimensional fast adaptive EMD method. At the same time, one-dimensional fast adaptive EMD method combined with shock pulse method is proposed, which is used in bearing fault diagnosis.Finally, the paper has studied the rotating machinery on-line condition system. Through the analysis of cold rolling units and ship shafting vibration characteristics, the paper has designed two sets of no-line monitoring system.