多目标遗传算法改进的加权多点灰色模型

发布时间：2018-04-28 13:14

本文选题：多目标约束 + 遗传算法　；参考：《测绘科学》2017年04期

【摘要】：针对传统多点灰色预测模型MGM(1,n)白化背景值构造方法不合理性导致模型往往不符合变形体实际情况的问题,该文提出了一种基于遗传算法的加权MGM(1,n)模型。引入白化背景值最佳生成权值矩阵替换传统模型背景值构造公式中的紧邻均值生成权阵,较好地顾及变形区域内多监测点变形趋势的突变性与不规则性,弥补了线性系统MGM(1,n)模型在非线性动力学系统变形预测分析应用中的不足;建立多目标优化实数编码遗传算法,实现背景值最优构造权阵的迭代搜索。基于仿真和工程实例数据的建模结果表明:改进模型较传统MGM(1,n)模型预测精度提高,抗噪声干扰能力增强。
[Abstract]:Aiming at the problem that the traditional multi-point grey prediction model MGM1 / 1) whitening the background value is not reasonable, the model often does not accord with the actual situation of deformable body, this paper presents a weighted MGM1n) model based on genetic algorithm. By introducing the whitening background value optimal generating weight matrix to replace the adjacent mean generating weight matrix in the traditional model background value construction formula, the mutation and irregularity of the deformation trend of multiple monitoring points in the deformation region are well taken into account. It makes up for the deficiency of the linear system MGM1) model in the application of nonlinear dynamic system deformation prediction and establishes a multi-objective optimization real-coded genetic algorithm to realize the iterative search of the background value optimal construction weight matrix. The modeling results based on simulation and engineering example data show that the improved model has higher prediction accuracy and stronger anti-noise ability than the traditional MGM-1) model.
【作者单位】：西南交通大学地球科学与环境工程学院;西南交通大学高速铁路运营安全空间信息技术国家地方联合工程实验室;成都铁路局成都高铁工务段;
【基金】：长江学者和创新团队发展计划基金项目(IRT13092) 高速铁路运营安全空间信息技术国家地方联合工程实验室资助项目
【分类号】：TP18

【相似文献】