论文部分内容阅读
非线性、非凸、不连续的数学模型的使用,使得过程优化问题难以求解。虽然确定性方法已经取得了重大的进步,但随机方法,特别是遗传算法提供了一种更有优势的方法。然而,遗传算法的性质决定了其不适合求解带有高约束的问题。本文提出了一个适用于高度约束问题的目标遗传算法,算法中的算子：交叉和变异,是在数据分析步骤得到的关于可行区域和目标函数行为信息的基础上定义。数据分析是以平行坐标系中的可视化描述为基础,一种模式匹配算法,扫描园算法,通过学习向量量化的使用被扩展来自动地确定目标函数和搜索空间的关键特征,这些特征被用于确定遗传算子。对石油稳定问题应用新的目标遗传算法,其结果证明了方法的有用、高效和健壮性。作为数据分析的核心,可视化技术的使用也可以用于解释优化过程得到的结果。
The use of non-linear, non-convex and discontinuous mathematical models makes the process optimization problem difficult to solve. Although significant advances have been made in deterministic methods, stochastic methods, especially genetic algorithms, provide a more advantageous approach. However, the nature of the genetic algorithm determines that it is not suitable for solving problems with high constraints. In this paper, we propose a genetic algorithm that is suitable for highly constrained problems. The operators in the algorithm: crossover and mutation are defined on the basis of the information about the feasible region and the objective function obtained in the data analysis step. Data analysis is based on a visual description in a parallel coordinate system, a pattern matching algorithm, a scanning park algorithm, which is extended by the use of learning vector quantization to automatically determine the key features of the objective function and search space, which are used in Identify genetic operators. Applying a new target genetic algorithm to the oil stability problem, the results prove that the method is useful, efficient and robust. As the core of data analysis, the use of visualization techniques can also be used to explain the results of the optimization process.