基于状态空间模型的最小二乘反褶积

Least Square Deconvolution Based on State Space Model

中文关键词  卡尔曼滤波及平滑  最小二乘  反褶积  模型参数及阶次估计

英文关键词  Kalman filter and smoother  least square  deconvolution  estimation of parameters and model orders

中文摘要

      反褶积是石油勘探中提高地震剖面分辨率的一种十分重要的方法。本文对地震信号用状态空间模型建模，提出了适用于地震信号处理的模型阶次及初始参数估计方法。为提高估计精度，依据卡尔曼滤波的新息序列建立了二次型目标函数，通过非线性寻优较精确地估计了模型参数。仿真及应用于实际均表明本文所提方法较之传统的最小二乘反褶积提高了地震剖面分辨率。

英文摘要

      The least square deconvolution is an important method in improving resolution of seismic profiles in oil exploration. Seismic signal is method as state space model in this paper. The method to estimate the orders and the initial parameters of the state space model is proposed and the quadratic object function is established according to the innovation sequence of Kalman filter to estimate parameters more accurately. The model parameters obtained by non-linear optimization is used to estimate reflectivity sequence in the fixed-interval optimal smoother. It`s shown from simulation and by applying our method to real seismic data that the resolution of seismic profiles is higher by our method than that by the traditional least square deconvolution method.