摘要
研究了结构优化问题求解的随机神经网络方法 ,可较好地克服 Hopfield网络方法容易陷入局部解的缺点。针对实例 ,对随机神经网络和 Hopfield网络进行了动态仿真 ,直观地给出了两种网络的动态并行运行、能量函数的动态变化过程及两种网络的全局性的差异 ,同时也仿真出了网络并行运行至稳定优化解的时间。
The method of stochastic neural Network to solve structural optimization problem is presented, which can overcome the shortcoming of local solution of Hopfield model. For two example problem, computer simulation, which is made for stochastic Neural Network and Hopfield model, shows the two kinds of model's parallel moving behavior, computation energy's dynamic changing and the two kinds of model's difference of globability. Simulation also obtains the time, which is taken to move to the stable solution by the two kinds of model.
出处
《计算力学学报》
CAS
CSCD
北大核心
2001年第2期242-245,共4页
Chinese Journal of Computational Mechanics
基金
国家自然基金!( 596850 0 3 )
四川省跨世纪杰出青年学科带头人培养基金资助项目
