本文針對工業(yè)零件設(shè)計(jì)中曲面數(shù)據(jù)的破損問題，提出了一種基于徑向基函數(shù)(RBF)神
經(jīng)網(wǎng)絡(luò)的數(shù)據(jù)修補(bǔ)方法。闡述了數(shù)據(jù)修補(bǔ)方法的原理，設(shè)計(jì)了網(wǎng)絡(luò)結(jié)構(gòu)和訓(xùn)練算法，運(yùn)用K均值算法實(shí)現(xiàn)隱層節(jié)點(diǎn)的中心初始化。通過兩種曲面的數(shù)據(jù)進(jìn)行訓(xùn)練，并對此兩種曲面數(shù)據(jù)破損進(jìn)行修補(bǔ)，結(jié)果表明，本文提出的曲面數(shù)據(jù)破損修補(bǔ)方法是有效的。
關(guān)鍵詞: 曲面破損 數(shù)據(jù)修補(bǔ) 徑向基函數(shù)神經(jīng)網(wǎng)絡(luò)
Curved Surface Insufficient Data Mending Using RBF Neural Network Yan Juan1 Wu Yuzhi2
1. College of Computer and Information Technology, Henan Normal University 2. Department of Computer, Jiaozuo University Abstract. The paper proposes a curved surface data mending method aimed to the problem of insufficient data in the modern industry based on the radial basis function(RBF) neural network. Introduces the theory of data mending. Designs the configuration and train algorithm of the network. Uses the K-means to initialize the points input of hidden-lay. Trains the neural network and does mending examination through two groups of curved surface data. The results show that the method proposes in the paper is efficient.
Key Words insufficient curved surface, data mending, RBF neural network