求解位场反演问题的混合编码遗传算法

在求解地球物理反演问题时，复杂的解析关系往往使计算变得十分困难. 对于这类问 题，遗传算法有其独特的优势. 然而传统的遗传算法在搜索最优解的过程中往往效率不高. 研究表明，遗传算法的编码机制在很大程度上决定了交换和变异操作的搜索能力. 二进制编 码的交换操作能产生更多的新样本数目而具有较强的搜索能力，十进制编码的变异操作因搜 索范围更大而具有较强的产生新样本的能力. 本文分析了二进制与十进制编码的搜索机制， 提出了混合编码遗传算法(Hybrid Encoding Genetic Algorithm,简称HEGA)，其原理是利用 十进制编码进行变异操作，其他操作采用二进制编码. 针对位场反演问题的特点，结合混合 编码、动态编码和大概率变异技术，有效地提高了搜索及产生“新”有效基因物质的能力. 理论模型及实际资料处理结果表明，该方法是有效的，尤其是在模型的分辨力方面有显著的 提高

The inversion of gravity data by using hybrid encoding ge netic algorithm

Genetic algorithm has some advantages in solving the inversion problems of complex non linear geophysical equation. Its multi point searching is able to find the global optimal solution, avoiding falling into a local optimum. The searching efficiency of genetic algorithm largely depends on encoding mechanism. Standard genetic algorithm (SGA) can not make search effective, because the crossover and mutation do not get most effectively searching in either binary or decimal encoding mechanism. The operation of crossover in binary encoding mechanism may produce more new individuals. On the other hand, decimal encoding mechanism makes the operation of mutation have larger searching range to find solutions. This paper gives the comparison of the searching capabilities of mutation operators in binary and decimal, and presents a hybrid encoding genetic algorithm (HEGA) mechanism. The method is based on hybrid encoding in genetic procedure, in which, the mutation operation is executed in decimal code and other operations in binary code. It actually introduces the two encoding mechanisms into genetic algorithm procession and has the mutation operated with high probability. HEGA may solve the inversions of complex non linear geophysical equations. In this paper, the inversions of synthetic 2D models and an observed gravity anomaly by using HEGA are shown.