地球物理学报
地毬物理學報
지구물이학보
2009年
12期
3116-3123
,共8页
岳碧波%彭真明%洪余刚%邹文
嶽碧波%彭真明%洪餘剛%鄒文
악벽파%팽진명%홍여강%추문
角道集子波%叠前反演%最优粒子群%速度均值滤波%全局优化
角道集子波%疊前反縯%最優粒子群%速度均值濾波%全跼優化
각도집자파%첩전반연%최우입자군%속도균치려파%전국우화
Angle-gather wavelet%Pre-stack inversion%Particle swarm optimization%Velocity mean filter%Global optimization
本文探讨了粒子群优化(PSO)算法在叠前地震角道集子波反演中的应用.在基本最优PSO算法的基础上,提出了对粒子更新速度进行平滑滤波的改进最优粒子群算法.由于代表子波的粒子的维数较大,如果粒子的各维元素相互独立,将导致粒子速度更新紊乱,影响搜索速度.通过对粒子速度进行三点均值滤波,加强了单个粒子各维元素的相互联系,并防止了粒子速度逃逸,使粒子更快地向有利于最优解的位置收敛.该方法应用于叠前角道集子波的反演中,取得了较好的子波反演效果,证明了本文方法的有效性.
本文探討瞭粒子群優化(PSO)算法在疊前地震角道集子波反縯中的應用.在基本最優PSO算法的基礎上,提齣瞭對粒子更新速度進行平滑濾波的改進最優粒子群算法.由于代錶子波的粒子的維數較大,如果粒子的各維元素相互獨立,將導緻粒子速度更新紊亂,影響搜索速度.通過對粒子速度進行三點均值濾波,加彊瞭單箇粒子各維元素的相互聯繫,併防止瞭粒子速度逃逸,使粒子更快地嚮有利于最優解的位置收斂.該方法應用于疊前角道集子波的反縯中,取得瞭較好的子波反縯效果,證明瞭本文方法的有效性.
본문탐토료입자군우화(PSO)산법재첩전지진각도집자파반연중적응용.재기본최우PSO산법적기출상,제출료대입자경신속도진행평활려파적개진최우입자군산법.유우대표자파적입자적유수교대,여과입자적각유원소상호독립,장도치입자속도경신문란,영향수색속도.통과대입자속도진행삼점균치려파,가강료단개입자각유원소적상호련계,병방지료입자속도도일,사입자경쾌지향유리우최우해적위치수렴.해방법응용우첩전각도집자파적반연중,취득료교호적자파반연효과,증명료본문방법적유효성.
A novel method for pre-stack seismic wavelet extraction based on particle swarm optimization(PSO)algorithm is presented in this paper.Under the framework of basic PSO algorithm,a mean filter is introduced for updating velocity of particles in order to obtain the inversion results of global optimal wavelet.Seismic wavelet vector iS described in the form of particle that is composed of many elements.If elements of a certain particle are independent of others,the particle velocity updating will be in disorder,which will have an adverse effect upon the process of searching for the optimal moving trajectory of particles.Three-point mean filter is used in the process of particle velocity updating and in that way the relationship among the elements of a certain particle is strengthened,which avoid particle velocity shifting and particles converge to the optimal solution quickly.The proposed algorithm in this paper was successfully performed for wavelet extraction in real pre-stack seismic data,which demonstrates that the improved PSO algorithm is efficient and feasible in seismic wavelet inversion.