论文部分内容阅读
提出一种基于神经元网络的轴类零件分类模型,采用基于反向传播算法的多层前馈式神经网络(BP)和自适应共振理论网络(ART1)实现基于特征的轴类零件家族的动态聚类与从聚类模板到每一事例的索引,完成轴类零件的实例分类三层模型。这种并行、分布式的神经网络分类处理过程大大提高了推理效率,为实例推理提供了崭新的思路。
A neural network-based classification model of shaft parts is proposed. Based on backpropagation algorithm, the multi-layer feedforward neural network (BP) and adaptive resonance theory network (ART1) are used to realize the feature- Clustering and indexing from the clustering template to each case complete a three-tier model of the instance classification of the shaft part. This kind of parallel, distributed neural network classification process greatly improves the reasoning efficiency and provides a brand new idea for case reasoning.