摘要
以挖掘机动臂为例,采用正交设计安排试验,在ABAQUS软件中获取有限元试验数据,利用Matlab软件编程仿真,比较BP神经网络和RBF神经网络在挖掘机动臂应力预测应用性能,针对BP神经网络应力预测准确度不高和RBF神经网络在学习样本输入区域很大,样本繁多时,需要更多的径向基神经元的问题,扩展出BP-RBF的组合神经网络,通过仿真实验证明,该组合网络提高了BP神经网络预测精度,扩展了RBF神经网络的应用范围,为工程应用提供参考。
Taking the hydraulic excavator＇s boom for example, this paper uses the orthogenal design to arrange experiment and ac- quire the finite element experimental data in ABAQUS software, compares the application of the BP neural network with the applica- tion of RBF neural network in the hydraulic excavator＇s boom structural optimization and puts forward a BP-RBF hybrid neural net- work according to the existing problems of the BP neural network and RBF neural network. The simulation experiment proves that this new hybrid neural network can be used to improve the BP neural network prediction accuracy and the application range of the RBF neural network. It can provide a reference for similar projects.
出处
《机械制造与自动化》
2013年第2期122-125,共4页
Machine Building & Automation
