论文部分内容阅读
以参变量变分原理为基础，建立并证明了二类变量（ｄｕｉ，ｄλｐ）变分原理．针对最小势能原理，给出了有限元列式，并把非线性约束项划归到目标函数中，构造出适合于神经网络处理的变分原理形式．证明了非线性约束处理前后的等价性，讨论了非线性项系数的确定方法．
Based on the variational principle of parameter variation, the variational principle of two types of variables (dui, dλp) is established and proved. For the principle of minimum potential energy, finite element equations are given, and the nonlinear constraints are classified into the objective function to construct the variational principle form suitable for neural network processing. The equivalence before and after the nonlinear constraint is proved, and the method to determine the nonlinear coefficient is discussed.