【摘要】 以弹性力学为例，依靠最小势能原理，将弹性力学计算问题转化为等式约束或无约束的二次优化问题。利用人工种经网络的并行计算能力和动态稳定特点寻优，建立弹性力学人工神经计算模型；给出改进的Hop－field和TH网络的结构和参数对应情况；对一个简单结构的弹性力学人工神经计算进行数值模拟。更多还原

【Abstract】 How to increase the speed ofstructural analysis when considering complexmechanical behaviors is always a remarkableproblem for engineers and researchers. In thispaper,taking the elastic mechanics problem as anexample, based on the Minimal Potential Princi-ple, the computation was transmit to theQuadratic Programming (QP) with constraints.The optimal point can be sought by using an ar-tificial neural network parallel and dynamic sta-bilization property. The elastic neurocomputingmodel was built up, and the neural net- -Mod-if ied Hopfield and Modified TH net, and its cor-responding parameters were given. The neuro-computation of a simple elastic structure wassimulated on digital computer. The results showthat the neurocomputed conclusion is in agreewith one from traditional FEM and the computa-tion is in real-time manner.更多还原