降维处理的多指标人机系统方案评价方法

Scheme Evaluation for Man-machine System of Multiple Indices Based on Dimension Reduction

为解决人机系统方案评价中指标众多、权重难于确定等问题,提出一种基于降维处理的多指标人机系统方案评价方法。该方法利用投影寻踪法构造投影函数,在无需确定指标权重的情况下,按照最优投影方向将高维指标数据投影到一维,通过比较一维投影值可获得最优设计方案,实现了高维指标数据的降维处理。同时,针对最优投影方向搜寻速度较慢、全局收敛性难以保证等问题,引入实数编码加速遗传算法,以投影值的标准差与局部密度的乘积为目标函数,投影向量为优化变量,搜寻最优投影方向,可快速获得评价结果。最后,以钻机的方案评价为例,对提出的方法进行了验证。

In order to solve the problems such as mass indices and uncertain weights, a new scheme evaluating method for man-machine system based on dimension reduction was presented. Without determining index weights, a projection function was constructed by use of PP （projection prosuit）, it can project high-dimension data to one-dimension, and optimal design scheme can be achieved through comparing one-dimension projection values. At the same time, aiming at the problems that seart：hing speed was too slow and global convergence property was hard to be guaranteed, RAGA （real coding based accelerating genetic algorithm） was introduced to search optimal projection direction and accelerate searching speed, prnduct of standard deviation and local density was used as objective function, projection vector was used as searching variable, and evaluation result can be obtained quickly. In the end, this method was validated by example of drilling machine＇s scheme evaluation.