一种虚拟人作业行为的自主优化模型

Automatic Optimization Model for Virtual Human's Working Behaviors

人体自适应行为仿真是实现人机工程学评估的前提条件.针对已有技术存在的不足,提出了一种基于多Agent合作式博弈的虚拟人作业行为自主优化模型.该模型将工作环境中人体自适应行为定义为一个多目标优化问题,提出了人体工作状态空间和人体行为元素的概念,以实现人体行为的离散化.设计了人体行为仿真算法以求解上述模型.算法采用梯度上升的策略来搜索满足模糊多目标Nash谈判条件的人体作业姿态的Pareto最优解.仿真实验表明,该方法可以在缺少相关数据的情况下推导出舒适的人体工作姿态,在工程领域中表现出较好的适用性.

Simulation of humanoid adaptive behaviors is a prerequisite of ergonomic evaluation. To overcome the shortages of existing technologies, an autonomic optimization model of virtual human＇s working behaviors is presented based on multi-agent cooperative games. In this model, the humanoid adaptive behaviors in working environment is defined as multi-objective optimization problem, and the definitions of mannequin workspace and behavior elements are proposed for discretization of human behaviors. A simulation algorithm based on fuzzy multi-objective games theory is presented to solve this model. With fuzzy multi-objective Nash negotiation theory, a grads-up method is adopted to search the Pareto optimizing solutions of human＇s working posture. Case studies show that the human＇s comfortable working posture can be derived by the new strategy even in cases with only insufficient data and better adaptability of the method in engineering domain can be reached.