中文摘要：

遗传算法是一种通过模拟自然进化过程来搜索最优解的非线性优化算法．它模拟达尔文的进化论，即生物的进化总是遵循适者生存、优生劣汰的规则。遗传算法用于地球物理反演的基本思想是从模型群体开始搜索，把模型参数用二进制进行编码，将模型空间的点映射到染色体空间的染色体，然后通过选择、交换和变异等遗传操作对模型群体进行繁殖，逐次迭代，在模型参数空间进行群体搜索，最后求取非线性反演问题整体极值所对应的最优解或近似最优解。遗传反演算法利用了生物进化过程和地球物理反演问题求解过程的相似性，开辟了地球物理反演的新途径，是非线性反演算法中一种最常用的算法。遗传算法是一种非线性的全局优化算法，它可避免目标函数梯度类方法的缺点，同时也可明显提高模型参数空间随机搜索的效率。本讲座概要地介绍了遗传算法的基本原理，遗传操作的基本步骤及实现方法，并给出了遗传算法在地球物理资料反演中的实例，最后总结和归纳了遗传算法的特点以及该方法的局限性。

英文摘要：

Genetic algorithm is originated from the Darwinian theory of evolution by natural selection, which simulates the progress of biological evolution from lower to higher animal. The idea of this biological process can be used to develop a new nonlinear optimization method. The model parameter can be encoded as chromosome representation. The genetic operations such as selection, crossover and mutation of gene are used for the evolution of initial models. The initial models are updated by the new models of next generation. The essence of genetic algorithm is a heuristic Monte Carlo method with higher efficiency and effectiveness. This paper not only introduced the principle, classification, diagram, application, advantages and disadvantages of the genetic algorithm method, but also pointed out the necessity of the research of the improved genetic algorithm method.