基于十进制遗传算法的含水层参数的反演

含水层参的反演是一个复杂的非线性优化问题,针对传统二进制遗传算法收敛性能差的缺陷,提出了反演含水层参数的十进制遗传算法。以直线隔水边界附近的井流模型为例,讨论了十进制遗传算法在含水层参数反演中的应用,并与二进制遗传算法的进行比较。结果表明,该方法在含水层参数的反演中不仅是可行的,而且具有较好的确定性和较高的精度;与二进制遗传算法相比,十进制遗传算法的收敛性较好,省时高效,且表示较为自然,容易引入相关领域知识。同时,结合实例的分析结果得出种群的规模对算法的收敛性没有明显的影响。

Identifying aquifer parameters based on the decimal strings genetic algorithm

The problem of aquifer parameter identification is actually a complex nonlinear program. With the limit of the ability of convergence of the binary strings genetic algorithm, the decimal strings genetic algorithm is proposed for identifying the aquifer parameters. An application of decimal strings genetic algorithm for aquifer parameter identification is discussed as an example of the well model near the linear impervious boundary, and compared with the binary strings algorithm. Application of the method to the published data shows the proposed method is not only applicable, but has higher accuracy and better determinacy. It is also shown from this paper that compared to the binary strings genetic algorithm, the decimal strings genetic algorithm can be natural to represent the problem to be studied, easy to introduce relative area knowledge, and has high efficiency, especially for multi-dimensional optimization problem.