【摘要】 根据人工神经网络的基本优化机理,研究了基于Lagrange乘子法神经网络求解弹塑性力学有限元问题.该神经网络对弹塑性力学有限元问题模型的不等式约束直接进行处理,无需添加松弛变量,降低了网络模拟和硬件实现的复杂程度.还分析了该神经网络的收敛性和稳定性.最后对一个简单弹塑性问题进行了数值仿真,计算结果表明了该神经网络求解弹塑性力学有限元问题的可行性.更多还原

【Abstract】 According to the basic optimization principle of artificial neural network,the neural network model based on Lagrange multiplier is used to solve the elasto-plastic mechanic finite element problems.It is no longer necessary to convert inequality constraints into equality ones by slack variables,which reduces the difficulty of the implementation of simulation and their circuits.The convergence and stability of these neural networks are analyzed.Simulation of a simple elasto-plasticity problem demonstrates that these neural networks are feasible and effective.更多还原