基于模型熵处理的泥浆脉冲信号去噪方法研究

Research on Denoising Method of Mud Pulse Signal Based on Model Entropy Processing

泥浆正脉冲传输技术在石油钻探中应用非常广泛,但泥浆正脉冲信号包含大量噪声,其中泵噪声为主要噪声来源,不仅时域上完全覆盖有用信号,而且频域上存在频谱混叠。在泵噪声特性的基础上,建立了一种泵噪声状态空间模型(线性时不变模型),引入最大相关熵准则去除泵噪声的野值,利用基于熵处理的标准卡尔曼滤波对泵噪声进行重构和滤波。此外,为实时处理信号的需求,采用随机共振原则滤除泵噪声后的剩余随机噪声。同时,模拟了稳态/非稳态泵噪声状态下几种不同信噪比的含噪泥浆正脉冲信号,分析评价方案的噪声抑制能力。结果表明,该方案能够抑制甚至去除原始信号中的稳态/非稳态泵噪声,可以将信号的信噪比提高约5~24 dB。

Mud positive pulse transmission technology is widely used in oil drilling,but positive pulse mud signal contains a lot of noise,among which pump noise is the main source of noise.Not only does the pump noise completely cover useful signal in time domain,but also causes spectrum aliasing in frequency domain.Based on the characteristics of strong pump noise,a state space model of pump noise(linear time-invariant model) was established,the outliers of pump noise is removed by introducing the maximum correlation entropy criterion,and the pump noise is reconstructed and filtered using the standard Kalman filter based on entropy processing.In addition,in order to process the signals in real time,the residual random noise after removing pump noise is filtered by stochastic resonance principle.At the same time,the positive pulse mud signals with different SNR at steady/unsteady pump noise state are simulated,and the noise suppression ability of the scheme is analyzed and evaluated.The results show that the proposed scheme can suppress or even remove the steady/unsteady pump noise in the original signals,and the signal-to-noise ratio of the signals can be increased by about 5~24 dB.