基于Hopfield的转子热应力场的实现方法

Realizing Thermal Stress Field of Rotor on Real Time Basing Hopfield Neural Network

现在对汽轮机转子热应力的实时计算方法均是采用解析方法 ,由于解析方法是对转子轴对称二维热传导方程的简化 ,给热应力的计算带来了误差 ,应用有限元的中间计算结果 ,应用Hopfield神经网络的方法 ,实现有限元计算中热应力整体刚度矩阵代数方程组的快速求解。仿真结果表明 。

The simplified analytical model to calculate thermal stress of rotor is often used ,but its calculated error is generated by the simplification of the model of two-dimension heat conducting equation. The result of finite element is more precise than that of simplify method, so the Hopfield neural network(HNN) uses the middle result of finite element to calculate algebraic equation group of the whole stiffness matrix of thermal stress, and a simulative example is given to prove feasibility of HNN method.