FMS-AGV 传输系统自学习调度方法

Algorithm of Self Learning Scheduling in FMS AGV Transportation System

针对ＦＭＳ－ＡＧＶ传输系统调度的不确定性因素，对ＡＧＶ在线运行状态参数实时记录并学习，运用动态规划算法和模糊数学的方法，提出了模糊动态规划（ＦＤＰ）算法．给出了适于计算的迭代算式，结合人工智能（ＡＩ）编制了ＡＧＶ传输系统调度软件包．本算法利用在线记录的ＡＧＶ运行数据进行模糊处理和学习，有效地调节用于描述ＦＤＰ算法的模糊模型参数．这种自学习ＦＤＰ算法在ＡＧＶ的路径规划中有较好的适应性．对于一个新的ＦＭＳ－ＡＧＶ系统，经几次规划自学习之后，该算法就能很好地用于系统的实时调度中．

This paper focuses on the approaches to deal with the problem of uncertainties in FMS AGV transportation system. By recording and studying AGV moving status parameters in real time, a fuzzy dynamic programming which combines the methods of dynamic programming and fuzzy mathematics is proposed. The calculating recursively expressions are given too. The scheduling and simulation software in AGV transportation system is programmed with AI. By fuzzy processing and self learning the new recorded AGV moving data in real time operations, the parameters of fuzzy model for describing FDP algorithm are adjusted effectively. The algorithm of self learning FDP has good adaptability to scheduling and planning in FMS AGV system. After planning and self learning several times, this algorithm may provide a way to schedule AGV plan in real time for a new FMS AGV system.