机构地区: 北京航空航天大学能源与动力工程学院
出 处: 《铸造技术》 2005年第9期827-829,共3页
摘 要: 研究目的是通过神经网络方法反求铸造模具复杂曲面。利用径向基函数(RBF)神经网络优异的非线性逼近能力,将外形数据、加工余量、变形数据作为神经网络输入,在神经网络的输出上可以得到铸造模具曲面离散数据点。再通过输出数据,可以对模具曲面进行造型。模具曲面的重构精度高、速度快。通过在复杂曲面模具造型上的实际应用,证明该方法能够实现产品和工艺装备的并行设计,可以缩短产品研制周期,提高设计、生产速度和效率,具有实用推广价值。 The aim of the paper is to reconstruct the surface of casting mould by artificial neural network (ANN). Radial Basis Function (RBF) network applied in the fitting model has strong capability of nonlinear approximation. Firstly, surface data of casting component should be collected by reverse engineering technique. The data of processing deformation should be decided as well. Secondly, input these data into the RBF network designed to computer the discrete points of casting mould surface. Finally, the reconstructed surface modeling of casting mould can be gained. The precision of reconstructing is precise and the train rate of RBF network is fast. The example presented has shown that this method can realize parallel design and data share with high efficiency. The method is useful and reliable.