中文摘要：

为解决一些对精度和实时性要求较高的调度问题，设计一个基于分枝定界算法和人工神经网络的实时调度算法。策略先使用分枝定界算法来找到m个作业的最佳排序。在生成足够多的排序以后，将排序作为训练样本来训练一个m维人工神经网络，从而得到一个m维的人工神经网络主矩阵。在实际的生产环境中，先对实际到达的n（n〉m）个作业进行分组，再利用离线生成的人工神经网络主矩阵对每个分组进行初始排序。最后将每个分组看作一个整体，根据Palmer算法得到n个作业的最终排序。仿真表明亥策略具有较好的实时性，同时也能达到较高的精确性。

英文摘要：

A scheduling method based on branch - bound algorithm and artificial neural network is designed to solve the scheduling problem with strict requirement for real time and accuracy. In this method, branch bound algorithm is used to find a sequence for m jobs. After enough sequences are obtained, those sequences are utilized as training examples to train neural network. Then, a matrix called neural network matrix is obtained. In the real production environment, n （n 〉 m） jobs are divided into several small groups. As to each small group, neural network master matrix is used to find m jobs＇ sequence. Finally, each group is regarded as a whole and Palmer algorithm is used to find the final sequence of n jobs. Stimulation demonstrates that this method costs fewer time than branch - bound algorithm and performs well in accuracy.