论文部分内容阅读
提出一种复杂环境下自主控制的动态优化新方法.首先,利用动态贝叶斯网络作为进化算法 t 代到 t+1代的转移网络.将贝叶斯优化及概率模型进化算法的静态优化机制推广到动态系统.通过感知环境变化.转移网络可以适时改变优化的基本条件和重新确立优化方向.指导自主智能体在无人干预下顺利完成一系列复杂任务.仿真结果表明基本思路正确.其次.为提高优化速度.满足实时性要求,提出“约束函数”及“置换”的概念,通过减少进化过程中不必要的网络节点及继承上一代部分优良解的方式,使得进化优化不必每次都重头开始,提高算法效率.
A new dynamic optimization method for autonomous control in complex environment is proposed.Firstly, the dynamic Bayesian network is used as the transfer network from t to generation t + 1, and the static optimization mechanism of Bayesian optimization and probabilistic model evolutionary algorithm Which can be changed to dynamic system.Considering the changes of the environment, the transfer network can change the basic conditions of optimization and re-establish the optimization direction in time.It instructs the autonomous agent to successfully complete a series of complex tasks under the condition of no human intervention.The simulation results show that the basic idea is correct. In order to improve the speed of optimization, to meet the real-time requirements, the notion of “Constraint Function ” and “Replacement ” is proposed to make evolutionary optimization unnecessary by reducing unnecessary network nodes in the evolutionary process and inheriting some of the good solutions from the previous generation Every time you start again, improve the efficiency of the algorithm.