一种作业车间调度问题的Hopfield神经网络优化方法

结合Hopfield神经网络结构和作业车间调度问题(JSSP)的约束特点,给出了适合于Hopfield神经网络求解的作业车间调度问题的矩阵表达数学模型。借用神经网络中能量函数的概念和含义确定网络的连接权,并将模拟退火算法应用于Hopfield神经网络求解,避免了系统输出陷入局部极值。将优化作业车间调度方案问题转换成求解网络系统的平衡点,即吸引子,该网络不仅能输出可行最优解,且优化速度快、实时性强。并通过计算机仿真表明了该方法的有效性。

An Optimization for Job-shop Scheduling Problem Based on Hopfield Neural Network

Combining the optimization function of Hopfield neural network and characteristics of jop-shop scheduling problem(JSSP),matrix expression mathematic model which fits to solve the JSSP with Hopfield neural network was put forward.The parameters of neural network circuit ascertained and the stability of the systems based on energy function concept of neural network was proved.The optimization about JSSP was turned into seeking attracting point or stable point of neural network system,so solving speed is very quick,and the result of Hopfield neural network is feasible and optimization.The performance of optimization method proposed was validated via computer simulations using an actual JSSP.