원문정보
초록
영어
Based on researching parameter optimization method of neural network deeply, a new parameter optimization method is presented and applied to surface defect online inspection system of cold rolled strips. The method takes advantages of small-samples fully, and can get a group of neural network parameters which can mostly express the neural network under a certain specific condition. The method is advantageous for its simplicity, easy to maintain and fast, it can be applied to many fields too, such as iron-steel industry, medicine. Experiments showed that a best recognition effect by using the parameters for neural network which are achieved by the new parameter optimization method can be got among all the parameters optimized randomly for surface defect of cold rolled strips.
목차
1. Introduction
2. Theory for parameter optimization
2.1 Acquirement of small-samples
2.2 Parameter segmentation and encoding method
2.3 Training and visual method in getting result
3. Instance for processing parameter optimization
4. Experiment results
5. Conclusion
6. References