基于Duffing振子混沌系统的地震速度分析方法

强随机噪声干扰是导致地震勘探资料低信噪比的主要原因,如何在强随机噪声干扰下获取有效的信息是值得关注的问题.Duffing振子混沌系统是一个非线性的动力学系统,其对强随机噪声具有免疫能力,而对特定的周期性信号具有敏感性.本文提出一种基于Duffing振子混沌系统的速度分析方法.对CMP道集按照时距曲线关系进行移动窗口截取,将所截取的信号构建为待测信号加入Duffing振子混沌系统,通过相图网格分割方法(GPM)判断系统状态的改变,从而在强随机噪声背景下获得高分辨率的速度谱.理论模型和实际资料的处理结果表明,与传统的水平叠加速度分析方法相比,本方法能够在强随机噪声背景下获得更准确的速度分析结果.

Seismic velocity analysis based on the Duffing oscillator chaotic system

The strong random noise is the main reason for low signal-to-noise ratio of seismic data. It is a much concerned issue how to extract useful information from data under a strong random noise background. The Duffing oscillator chaotic system is a nonlinear dynamic one that is immune to noise, but very sensitive to particular periodic signals. This paper presents a seismic velocity analysis method based on the Duffing oscillator chaotic system. In a chaotic system, one needs to make the system reach a critical state before it is used to analyze signal. We propose a new method called the grid partition method (GPM) to estimate the chaotic system state. The principle of GPM is dividing the phase plane diagram into many small square grids. If the phase trajectories pass any small grid, this grid will be assigned to 1; otherwise, the value of the small grid is 0. Summing all the values of small grids will obtain the grid judgment parameter, which provides a stable criterion for seismic velocity analysis. A moving window intercept method following the time-distance curve is used to reconstruct the signal from the CMP gathers. Then we add the reconstructed signal to the Duffing oscillator chaotic system and use the grid partition method to judge the state of system. Therefore, the high resolution velocity spectrum can be obtained by the proposed method from the data with a strong random noise background. We apply the proposed method to the synthetic model which is a CMP gather contains two events. Adding strong random noise to the synthetic model and two events are buried in noise, which cannot be identified. The signal-to-noise ratio (SNR) is -16.14 dB. For comparison, we use standard velocity scan and chaotic system with GPM to calculate velocity spectra of noisy data, respectively. Velocity analysis with stacking criterion fails in providing correct velocity trends. The results of the chaotic system with the GPM velocity analysis method shows higher resolution and random noise has less influence on the low velocity zone, which may cause the incorrect velocity picking. We use a marine field dataset with swell noise to further evaluate the proposed method. For standard velocity scan, the swell noise causes low quality of velocity panels, which leads to inaccurate velocity picking. The Duffing chaotic system is less affected by the strong swell noise and shows more accurate velocity picking results. The velocity analysis method using the Duffing chaotic system produces a stacking result with high SNR. Because the Duffing chaotic system is less sensitive to strong noise, so suitable for detecting weak seismic events. A new approach called grid partition method (GPM) can judge state of the Duffing oscillator system, which provides a stable criterion to detect accurate RMS velocities. To reconstruct the periodic seismic signal, we use an intercept window to select seismic wavelets from the CMP gathers. Where after, the Duffing chaotic system provides a better way for acquiring velocity information in the case strong random noise exists. Results of applying the proposed method to a synthetic example and field data show that this new method, compared with traditional velocity analysis, can obtain more accurate velocities in a strong random noise environment.