一种基于排序的LDF改进遗传算法

An Improved Genetic Algorithm for LDF Based on Sorting

针对高维数据建模过程中常见的非线性复杂问题,使用遗传算法进行全局寻优。考虑到标准遗传算法在搜索过程中存在的多种收敛性问题,提出基于排序的拉普拉斯分布函数(LDF)改进遗传算法。该算法针对性地改进了遗传算法的多种收敛性问题,解决了遗传算法过快收敛至局部最优解以及后期收敛速度过慢等问题。根据实际生产数据,对轧钢精轧机组进行了仿真模拟,仿真实验结果表明了该算法切实可行,在保证了优化效果的同时,也保证了算法的总体收敛速度和稳定性,具有良好的应用前景。

In high dimensional data modeling process,the complex nonlinear problem is commonly seen,thus global optimization is conducted by using genetic algorithm. Considering the multiple convergence problems existing in standard genetic algorithm,the improved genetic algorithm for Laplace distribution function（ LDF） based on sorting is proposed. The algorithm improves the multiple convergence problems,solves the problems of premature convergence of local optimum solution,and the slow convergence in late phase. The simulation based on practical productive data from the steel rolling finishing mill group is conducted,the results of simulation indicate that this algorithm is feasible,it ensures the optimization effect,overall convergence speed and stability,and possesses excellent applicable prospects.