基于并行模拟退火算法的微地震速度模型校正方法研究

Study on the micro-seismic velocity model correction based on the parallel simulated annealing algorithm

在水力压裂微地震监测中,速度模型的误差对微震定位结果的偏差影响尤为显著,因此如何获得较为准确的速度模型是微震监测过程中的重要一步.一般使用测井数据和射孔监测数据来反演计算微地震速度模型.该计算过程耗时较长,这不利于微震监测的实时化.针对这个问题,本文从模拟退火法入手,提出采用多马尔科夫链的并行方案,将其与速度模型的校正过程相结合.通过拆分马尔科夫链并分别在多线程中并行计算,可以大大缩短单条线程中的计算时间以达到提升速度模型校正效率的目的.经过合成数据及实际监测数据的验证,证明了本文方法相对串行模拟退火法,能有效提升速度模型校正的效率,并且精度上也能得到保证,能充分满足野外实时监测的需求.

During the hydraulic fracturing micro-seismic monitoring,velocity model is crucial factor for micro-seismic source location accuracy.Therefore,getting reasonable velocity model is an importance step of micro-seismic monitoring.Micro-seismic velocity model is usually constructed using well logging and perforation record with inverse methods.However,this calculation processing is time-consuming,which greatly limits real-time monitoring of hydraulic fracturing.In order to solve this problem,this paper proposes a parallel method using multi-Markov chain.By splitting the Markov chain and computing it in parallel in multiple threads,the computation time in a single thread can be greatly shortened to improve the efficiency of the speed model correction.Using synthetic data and actual monitoring data for method verification,the proposed method can not only effectively improves the speed of velocity correction but also guarantee accuracy,it can satisfy the need of real-time monitoring.