智能预测在三峡永久船闸中隔墩时效变形分析中的应用

Application of intelligent prediction to time-effect deformation analysis for the partition frusta of Three Gorges Permanent Shiplock

人工神经网络通过调整网络中的连接权值、网络结构,实现输入输出非线性映射关系,对三峡永久船闸中隔墩的变形进行预测。具有全局最优化遗传算法可使连接权值、网络结构得到最优解。将遗传算法与神经网络结合起来进行变形演化特征识别,对三峡永久船闸中隔墩时效变形进行智能预测,应用结果表明该方法具有较高的预测精度。

The artificial neural network carries out the input and output nonlinear mapping relationship to predict the deformation of the partition frusta of the Three Gorge＇s Permanent Shiplock by adjusting the connected weighted value and the network structure. The global optimal solution to the weighted value and the network structure is obtained by the optimized genetic algorithm. Combined the genetic algorithm with neural network, the characteristics of the deformation evolvement is identified. And then the time-effect deformation for the Three Gorges Permanent Shiplock is intelligently predicted; and it is shown that the proposed approach has higher predicting precision.