给出了求解ELSP问题(Economic Lot Scheduling Problem)的可行域的特征、启发式规则和演化神经网络设计问题.经济批量问题采用基本时段方法表示,该方法产生2类决策变量:表示基本时间段的连续变量和表示时间倍数的整数变量.在求解ELSP问题的算法设计中,可行域是判定启发式规则有效性的基础.为了给出可行域的特征,利用神经网络的演化计算,给出了求ELSP问题的初值算法,设计演化参数函数、网络结构、演化函数、演化规则,并依此获得可行域的约束条件.对在可行域约束条件和启发式规则下设计的算法进行测试,并与用HGA和一般GA方法求解ELSP问题进行比较,求解效率明显提高,使得在满足可行性的前提下总费用减小.
The paper deals with oscillation of Runge-Kutta methods for equation x'(t) = ax(t) + a_0x([t]).The conditions of oscillation for the numerical methods are presented by considering the characteristic equation of the corresponding discrete scheme.It is proved that any nodes have the same oscillatory property as the integer nodes.Furthermore,the conditions under which the oscillation of the analytic solution is inherited by the numerical solution are obtained.The relationships between stability and oscillation are considered.Finally,some numerical experiments are given.
