有转运运输问题的Hopfield神经网络优化方法

Hopfield神经网络是一种递归神经网络,可以用于联想记忆和优化,运输问题是一类特殊的线性规划问题,结合Hopfield神经网络优化功能和有转运运输问题的特点,并根据运输问题的实际情况,将约束边界模糊化,设计针对有转运运输问题的连续Hopfield神经网络电路。借用神经网络中能量函数的概念和含义确定网络电路的参数并证明系统的稳定性。将优化有转运运输问题转换成求解网络系统的平衡点,即吸引子。在优化过程中,为了防止网络收敛到局部极值,采用模拟退火算法,使网络能够达到全局最小值,不仅优化速度快、实时性强,也给其它线性、非线性规划问题的优化提供了一个新的途径。并通过计算机仿真研究验证了系统的有效性、可行性。

Optimization of Transportation Problems with Transshipment by the Hopfield Neural Network

The Hopfield neural network is a recurrent neural network used to associate ideas and recollections and optimize computation.Transportation problems are a special type of linear programming problems.Combing the optimization function of the Hopfield neural network and the characteristics of transshipment problems under actual situations,constraint boundaries are made fuzzy.Continuous Hopfield neural network circuits are designed,aiming at transportation problems with transshipment.The energy function concept of the neural network is made use of to ascertain the parameters of the neural network circuit and to prove the stability of the system.Transportation optimization is turned into seeking the balancing point of the neural network,i.e.,the attracting point.The simulated anneal algorithm is adopted to help the system reach global optimal solutions and avoid system convergence to local sub-optimal solutions.The solving speed is very quick and of good real-time nature.A novel optimization approach to linear and nonlinear programming is put forward.Computer simulations prove the effectiveness and practicability of the Hopfield neural network.