基于粗糙集和分层支持向量机的agv多分支路径识别-journalofnuaa.pdf

45 1 V ol . 4 5 N o . 1 2013 2 Journal of N anjing U niversit y of A ero nautics A st ronautics Feb . 20 13 AGV 1 2 1 2 1 2 1 2 喻　俊 　楼佩煌 　武　星 　杨　旭 ( 1. 210016; 2 . 2100 16) : 双向型视觉导引自动导引车多分支路径识别的关键问题是实时性和鲁棒性。基 智能信息融合的思想 将 粗糙集理论与多类支持向量机方法结合起来 提出一种基 知识获取实时性和类的相似性的分层多分支路径识 别新方法。采用粗糙集知识粒度理论和分层递阶约简算法获得最小的决策规则 有效降低分类识别的复杂性; 提 出分类决策安全区域的学习方法 提高识别的鲁棒性。最后 仿真实验和多种环境下的运行测试验证了该方法的 有效性和可靠性。 : 柔性制造系统; 自动导引车; 粗糙集; 支持向量机 : T P242 . 2 : A : 1005 -2615( 2013) 01-0062 -0 8 Automated Guided Vehicle Cross Path Recognition Based on Rough Set and Hierarchical Support Vector Machine 1 2 1 2 1 2 1 2 Yu Jun Lou Peihuang Wu X ing Yang X u ( 1. Co lleg e of M ech anical and Elect rical Engineer ing N anjing U niver sity of A er onautics A str onautics N anjing 2 10016 China ; 2. Jiangsu Key L abo r ator y o f Pr ecisio n and M icr o-M anuf actur ing T echno lo gy N anjing U niv ersit y of A er onautics A str onautics N anjing 2100 16 China) Abstract: R eal -t ime ability and r obustness are t he key pro blems o f bi-direct ional v ision -b ased aut om ated gu ided vehicle cros s p at h recog nit ion . A ccording t o co mpl em ent ary of r oug h set t heo ry and m ul ti-class support vect or m achine( SV M ) a new cro ss p ath hierarchical reco gnition method is present ed b ased on real-tim e abilit y of know ledg e and sim ilarit y of clas s. Granularit ies of know ledge and hierarchical redu ct ion r