相控随机优化地震反演技术及应用

Facies Constrained Stochastic Optimization Seismic Inversion and Application

摘要: 海上河流相储层横向变化快，井网稀疏，常规测井约束反演方法对井间储层预测的不确定性较大。为了提高井网稀疏条件下测井约束反演的精度，研究形成了相控随机优化地震反演方法。初始模型综合沉积认识、地震属性和测井信息来建立，不再仅依赖于测井信息来建立。反演过程中采用改进的随机爬山算法，先从多个随机的解开始，形成最优解的集合，再从中找出全局最优解，使合成地震记录与观测地震记录实现最大程度地相似，实现反演预测全局最优的目标。通过对实际资料的应用表明，相控随机优化地震反演方法建立的初始模型降低了因井少、井分布不均衡带来的储层预测误差，使模型正演地震与观测地震达到最佳匹配，提高了海上河流相储层的预测精度。

Abstract: The lateral variation of fluvial reservoir is fast, the well pattern is sparse in offshore oilfield, and the uncertainty of conventional logging constrained inversion methods on the prediction of inter-well reservoir is large. In order to improve the precision of logging constrained inversion under the condition of sparse well pattern, a sedimentary facies controlled stochastic optimal seismic inversion method was developed. The initial model integrates sedimentary data, seismic attributes and logging information, and no longer depends on logging information only. In the process of inversion, we adopt the improved stochastic hill-climbing algorithm. First we start with multiple random solutions, and form the set of optimal solutions. Then we find out the global optimal solution, making the synthetic seismograms and seismic records to achieve the maximum degree of similarity, and achieve the goal of global optimal inversion. The application of the real data shows that the initial model established by the facies-controlled stochastic optimization seismic inversion method reduces the reservoir prediction error caused by the lack of wells and uneven well distribution. The model forward seismic data and observational seismic data are matched optimally, and the prediction accuracy of offshore fluvial reservoir is improved.