改进快速模拟退火方法进行AVO岩性参数反演

IMPROVEMENT OF FAST SIMULATED ANNEALING ALGORITHM AND ITS APPLICATION TO INVERSION OF AVO LITHOLOGICAL PARAMETERS

以广义波尔兹曼的吉布斯（ Ｂｏｌｔｚｍａｎｎ＿ Ｇｉｂｂｓ） 统计理论为基础，将数论（ｎｕｍｅｒａｌ） 和依赖于温度的似柯西（ Ｃａｕｃｈｙ） 分布相结合建立一种新的扰动模型，改进接受概率，提出了一种非常快速模拟退火法（ｖｅｒｙ ｆａｓｔ ｓｉｍｕｌａｔｅｄ ａｎｎｅａｌｉｎｇ） ．通过多参数的振幅随炮检距变化（ Ａ Ｖ Ｏ）的合成数据和实际资料试算结果表明，该方法能在高温下进行大范围内的搜寻，解的稳定性好和抗随机干扰能力强，反演结果精度较高。

A very fast simulated annealing (VFSA) algorithm is proposed in this paper, on the basis of the turbulent model established on the generalized Boltzmann_Gibbs statistical theory and derived from numeral and temperature_dependent pseudo_Cauchy distribution for the purpose of the improvement of the acceptable probability. The results of the synthetic and actual data of amplitude versus offset (AVO) inversion show that VFSA can be used to search for the global optimal target within a large range and at a high temperature, to secure a stable solution, a strong resistance to random noise, a high precision of inversion, and especially a fast_speed VFSA convergence.