基于神经网络的产品造型美学特征识别与修正

根据产品造型的要求,总结出若干美学特征并构造该特征与美学评价的对应关系,利用人工神经网络的非线性映射能力,建立各种特征组合与相应造型美度等级的映射关系,从而实现基于神经网络的产品造型美学特征识别与修正。网络训练完成后,可针对新特征对产品的造型美度等级进行评价,进而对不良特征进行修正。神经网络采用一种综合改进的BP网络,并进行了动量项自适应调整的改进。通过算例表明,该方法是有效的。

Aesthetics Feature Identification and Modification for Product Molding Based on Neutral Network

Some aesthetics features are summarized and the relationship between these features and their aesthetics evaluations according to the requirements of product molding is established. The map relation between diverse feature groups and corresponding degree of mold aesthetics is set up using nonlinear map ability of neural network. Therefore, aesthetics feature identification and modification for product molding based on neural network can be realized. When the network training is finished, the degree of molding aesthetics for products can be evaluated based on the new features, and the exact features to effect the degree of molding aesthetics can be found. A comprehensively improved BP neural network is adopted, and an improvement for momentum adaptive adjustment is carried out. A method is provided to analyze the problem of aesthetics of product molding at a new point of view, and it is proved to be feasible through an example.