为有限资源均衡问题提供一个神经网络解决方法.首先提出增广置位矩阵,描述了资源均衡的神经网络表示,使得神经元的输出和问题的解彼此对应起来;然后在时间和资源约束下利用多种技巧构造网络的能量函数,使其能量最小值对应于资源最均衡的状态;并且提出基于“权值状态发生器”的离散Hopfield与模拟退火算法(DHNN-SA)融合的镶嵌式混合结构,从本质上提高了网络的优化质量;最后设计了资源优化神经网络的模拟程序. In order to provide a neural network solution to the problem of limited resource balance, an augmented setting matrix is ​​proposed to describe the neural network representation of resource equalization, which makes the output of neurons and the solution of the problem correspond to each other, and then utilizes the time and resource constraints A variety of techniques are used to construct the energy function of the network so that its energy minimum corresponds to the most balanced state of resources. A mosaic hybrid structure based on the discrete Hopfield-Simulated Annealing Algorithm (DHNN-SA) based on a “weight state generator” , Which essentially improves the quality of network optimization. Finally, a simulation program of resource-optimized neural network is designed.