摘要
针对常规神经网络板形识别方法中存在的不足,提出了以PCA替代欧氏距离作为提取特征的手段,并将所设计的PCA-RBF板形识别模型以FPGA为硬件实现载体进行了仿真研究。仿真结果表明,设计的PCA-RBF板形识别模型能够正确识别出板形缺陷,网络结构比常规RBF板形识别模型相对简化,同时识别精度提升了59%,抗干扰能力提升了82%。FPGA仿真结果在精度和实时性上可以满足实际工程需要。
Aiming at the deficiency in the feature extraction by Euclidean distance for the conventional flatness recognition based on neural network, the method of PCA was proposed to extract the features. In addition, FPGA was chosen as the hardware carrier to implement simulation test. The results showed that the model based on PCA.RBF network could identify the flatness defects correctly, with network structure simpler compared to the model based on the conventional RBF neural network. Furthermore, its recognition accuracy was improved by 59% and the ability of anti.interference was increased by 82%. It is concluded that the simulation results of FPGA can meet the engineering requirement in terms of accuracy and real.time performance.
作者
张秀玲
代景欢
李家欢
张逞逞
ZHANG Xiu-ling;DAI Jing-huan;LI Jia-huan;ZHANG Cheng-cheng(Key Laboratory of Industrial Computer Control Engineering of Hebei Province, Yanshan University, Qinhuangdao066004, Hebei, China;National Engineering Research Center for Equipment and Technology of Cold Strip Rolling, Yanshan University, Qinhuangdao 066004, Hebei China)
出处
《矿冶工程》
CAS
CSCD
北大核心
2019年第1期109-113,共5页
Mining and Metallurgical Engineering
基金
河北省自然科学基金-钢铁联合研究基金项目(E2015203354)
河北省高校创新团队领军人才培育计划项目(LJRC013)
河北省教育厅科学研究计划河北省高等学校自然科学研究重点项目(ZD201610)
2016年燕山大学基础研究专项培育课题(16LGY015)
