Full metadata record
DC FieldValueLanguage
dc.contributor.author何智南en_US
dc.contributor.authorHo, Jhy-Lanen_US
dc.contributor.author陳福川en_US
dc.contributor.authorChen Fu-Chuangen_US
dc.date.accessioned2014-12-12T02:17:07Z-
dc.date.available2014-12-12T02:17:07Z-
dc.date.issued1996en_US
dc.identifier.urihttp://140.113.39.130/cdrfb3/record/nctu/#NT850327005en_US
dc.identifier.urihttp://hdl.handle.net/11536/61656-
dc.description.abstract由於小腦模組關節控制器(CMAC)能夠非常快速地學習非線性函數,並 進行引申 (generalization),因此在非線性系統的即時(real time) 控制上,它是一項有力且實際的工具。 在本論文中,我們將提出利用FPGA晶片實現之數位並行運算的CMAC類神經 網路。所製作的CMAC晶片是固定點(fixed point)系統,採用SIMD結構實 現CMAC並行演算法則。完成後的CMAC類神經網路包含三顆Xilinx XC4000 FPGA晶片,其中兩顆晶片執行記憶體位址的映射,另一顆晶片計算連結量 的累加及更新。這三顆FPGA晶片配合DSP處理器、SRAM及一些週邊電路構 成一完整的測試系統。我們測試了CMAC的學習特性及其在非線性控制上的 效能。 The Cerebellar Model Articulation Controller(CMAC) is capable of learning nonlinear functions extremely quickly due to its generalizing capability,so itis a powerful and practical tool for real time control.In this studywe present a realization of the CMAC neural network by FPGA chips.We employ the fixed point system and adopt the SIMD architecture to implement the CMAC parallel algorithm.Hardware design is accomplished by three Xilinx XC4000 FPGA chips.Two FPGA chips carry out address mapping and the other chip does weight accumulation and weight modification.The test system includes FPGA chips,DSP processor, SRAM and some peripheral circuits.We show the performance of the CMAC chips in learning and in nonlinear control.zh_TW
dc.language.isozh_TWen_US
dc.subject小腦模組關節控制器zh_TW
dc.subjectCMACen_US
dc.title以FPGA晶片實現CMAC類神經網路控制器zh_TW
dc.titleRealization of CMAC Neural Network Controller by FPGA Chipsen_US
dc.typeThesisen_US
dc.contributor.department電控工程研究所zh_TW
Appears in Collections:Thesis