基于模拟退火算法的地质统计学反演方法

 地质统计学反演结合了地震反演储层预测方法和随机建模储层预测方法的优势，充分利用随机建模技术综合不同尺度的数据(如模拟过程可综合构造和地层信息)，反演时从井点出发，在井点处忠实于测井数据，井间符合地震数据，通过从地震道中抽取反射系数与子波褶积生成合成地震记录，利用合成地震记录与原始地震道的残差修改反射系数，得到新的反射系数序列，通过反复迭代，从而得到一个最佳逼近地震道的反射系数序列，通过反射系数序列可以求得多个等概率的相对波阻抗数据体。随机模拟方法包容了地震资料中的噪声，并考虑到数据在一定空间内的相关性，拓宽了地震数据频带，使随机模拟结果符合地震数据，使最终反演结果同时符合井数据、地震数据以及已知信息的地质统计学特征，结果的多个实现可用于不确定性评价，可以更加精确地描述储层的变化。

Inversion of geostatistics based on simulated annealing algorithm

The inversion of geostatistics combines superiorities of reservoir prediction of seismic inversion with random model build,and fully uses random model-building technique to integrate the data with different scale (for example, simulation procedure can integrates structural with stratigraphic information). Starting from well point, the inversion is faithful to logging data at well point and conforms to seismic data in cross-hole area. The synthetic seismograph is yielded by deconvolution of wavelet with reflection coefficients sorted from seismic traces. Using the residues of synthetic seismograph with raw seismic traces to modify the reflection coefficients and obtain new reflection coefficients sequence, the reflection coefficients sequence optimally approaching the seismic traces is obtained after repeated iteration. The relative wave impedance data volume can be computed by reflection coefficients sequence. The random simulation method holds the noises in seismic data and considers the correlation of data within a certain space, broadening the band of seismic data and making the random-simulated results conform to seismic data and final inverted results can conform to well data, seismic data and the feature of geostatistics of known information simultaneously. The multiple fulfillments of results can be used for uncertainty appreciation and more preciously describe the change in reservoir.