Application Research Of Generalized S Transform Time-frequency Analysis

Time-frequency analysis is a powerful tool for time-varying non-stationary signal analysis, and a hotspot in modern signal processing research. Most signals in nature world and real life are non-stationary. The one-dimensional time-domain signals can be mapped into the two-dimensional time-frequency domain. Meanwhile, time-frequency analysis can provide the joint distribution information of both time-domain and frequency-domain, then express joint characteristic of non-stationary signals entirely.This paper mainly researches the application of generalized S transform time-frequency analysis and applies it to time-frequency filtering and discontinuity detection. Firstly, we introduce the development of non-stationary signal processing based on methods of time-frequency analysis at home and abroad. And the concepts of non-stationary signal and time-frequency analysis are also mentioned. Then fundamental principles of traditional time-frequency analysis methods and S transform are described and then the comparison of them is provided.In the signal processing, time-frequency filtering method is common to reduce the interference and improve the signal to noise ratio. This paper presents a new time-frequency filtering method based on generalized S transform and applies it to seismic signal processing. First of all, through the frequency spectrum received by generalized S transform, we can determine the suppression function of the surface wave in order to suppress the surface wave in real seismic record. Afterwards, we apply the filtering method combining the threshold filtering and lowpass filtering to seismic signal processing. It will achieve the purpose that both improving the noise and protecting the effective signal.Simulation results show that this method is much better than other time-frequency analysis methods in improving the signal to noise ratio. More significantly, it is an adaptive algorithm. Finally, the application of this method in real seismic data processing can extract the effective signal, and the vertex graph can be gained by the method of steepest descent. The combination of them plays a part in finding the reservoir of seismic data.The break points in the detection signal usually means damage of structure, therefore, it is very important to find the break points in the signal. In this paper, generalized S transform is applied to discontinuity detection. Take the railway signal as an example; we process the signal by the filtering method above for the purpose of removing the noise and extracting the effective signal. Then, whether including wheel-flat signal or not will be known through short-time energy method. At last, the wheel-flat signal will be detected and located by combination of the extracted effective signal and the amplitude envelope of generalized S transform. Simulation results indicate that this method can detect the wheel-flat signal correctly and avoid the detection missing.