摘要
提出了求解地震反演问题的免疫粒子群优化算法(IPSO).定义了种群多样性指标,在此基础上提出了抗体动态选择机制,分别在进化初期和后期采用不同策略进行抗体选择,与适应度选择、q-选择以及确定性选择方法比较优势明显.提出了基于种群多样性的权重自适应调整策略,在进化初期进行大范围全局搜索,进化后期则在相对较小的空间进行局部精细搜索,缓解了全局搜索和局部搜索之间的矛盾.二十层理论模型计算表明,IPSO算法不依赖于初始模型选择,在无噪和加入20%噪声情况下反演精度均明显优于标准粒子群算法(PSO)和自适应粒子群算法(APSO);成庄煤矿四、五盘区实际地震资料反演结果表明,IPSO算法能够准确识别煤层及顶底板位置,明显提高了弱反射波的连续性和可检测性.
Immune particle swarm optimization algorithm（IPSO） was proposed for the seismic inversion problem.The dynamic selection mechanism of antibodies based on the definition of swarms diversity was advanced.The mechanism is proved to be superior to the fitness selection,q-selection and confirmation selection by using different strategies in different evolution stages.The self-adjustment of the weight based on swarms diversity was put forward to relieve the contradiction between the global convergence and the local convergence by global search in the early stage and local search in the late stage.The simulation results indicate that the proposed IPSO has better efficiency and higher accuracy than PSO and APSO.The practical inversion results show that the inversion resolution is obviously higher than the seismic resolution,the continuity and the detection capability of weak reflection are improved greatly and the roof and floor of coal seam can be correctly identified.
出处
《中国矿业大学学报》
EI
CAS
CSCD
北大核心
2010年第5期733-739,共7页
Journal of China University of Mining & Technology
基金
国家自然科学基金项目(50674086)
关键词
波阻抗反演
免疫粒子群
粒子群
非线性优化
wave impedance inversion
immune particle swarm optimization
particle swarm optimization
nonlinear optimization
