第九章模糊与神经网络的比较 以倒车系统为例 丁瑶 李静
1 丁瑶 李静 第九章 模糊与神经网络的比较 ——以倒车系统为例
CONTENTS Introduction Backing up a Truck With a Fuzzy System With a Neural System Comparison Sensitivity Analysis With an Adaptive Fuzzy Neural System Parking a Truck with a Trailer Conclusion
2 CONTENTS Introduction Backing up a Truck With a Fuzzy System With a Neural System Comparison Sensitivity Analysis With an Adaptive Fuzzy Neural System Parking a Truck with a Trailer Conclusion
Introduction 模糊系统 描述和处理人的语言思维中存在的模糊性概念，模拟人 的智能 神经网络系统 根据人脑的生理结构和信息处理过程，来模拟人的智能 两者的结合 神经网络黑盒子结构难以确定，但神经网络具有自学习等 功能。 具体问题-：Backing up a truck 2:Parking a truck with a trailer
3 Introduction 模糊系统 描述和处理人的语言、思维中存在的模糊性概念，模拟人 的智能 神经网络系统 根据人脑的生理结构和信息处理过程，来模拟人的智能 两者的结合 神经网络黑盒子结构难以确定，但神经网络具有自学习等 功能。 具体问题-1:Backing up a truck 2: Parking a truck with a trailer
Backing up a Truck -Background loading dock (xr.Jr) (100.100) rear (y) 如 front (0.0) State variables o.x,y Determine the truck position 0 Steering angle Ignore variable y and discretize all values 01 One degree each
4 Backing up a Truck -Background State variables :φ,x, y θ 0.1 One degree each Determine the truck position Steering angle Ignore variable y and discretize all values
Backing up a Truck with a Fuzzy System 模糊系统的一般处理方法 模糊规则库 输入参数 模糊化 模猢推理 去模糊 模糊集 模糊集 输出参数
5 Backing up a Truck -with a Fuzzy System 模糊化 模 模糊推理 去模糊 糊 集 输 入 参 数 模 糊 集 输 出 参 数 模糊规则库 模糊系统的一般处理方法
Backing up a Truck 模糊系统的模糊集 -with a Fuzzy-System }} LE LC CE RC RI 08 0.6 Angle z-positionx Steering signal 0.2 LE:Left NB:Negative Big 0 RB:Right Below 10 20 40 50 60 80 90 100 RU:Right Upper LC:Left Center NM:Negative Me NB NM NS ZE PS PM RV:Right Vertical CE:Center NS:Negative Sma 0.8 VE:Vertical RC:Right Center ZE:Zero 6 LV:Left Vertical RI:Right PS:Positive Smal 0.4 LU:Left Upper PM:Positive Med 0.2 LB:Left Below PB:Positive Big RB RU RV VE LV LU LB 6 0.2 0 0 0 100 150 200 250 300
6 Backing up a Truck 模糊系统的模糊集 -with a Fuzzy System
Backing up a Truck -with a Fuzzy System FAM Rules and Control Surface LE LC CE RC RI RB PS PM PM PB PB RU NS PS PM PB PB 1510 RV NM NS PS PM PB VE NM NM 18ZE PM PM -40 LV NB NM NS PS PM 20 100 LU NB NB NM NS PS 500 -50 20 025020015010 LB NB NB NM NM NS input2 Both reflect the symmetry of the control system
7 Backing up a Truck -with a Fuzzy System FAM Rules and Control Surface LE LC CE RC RI RB PS PM PM PB PB RU NS PS PM PB PB RV NM NS PS PM PB VE NM NM 18ZE PM PM LV NB NM NS PS PM LU NB NB NM NS PS LB NB NB NM NM NS Both reflect the symmetry of the control system
Backing up a Truck -with a Fuzz④y System 相关最小FAM推理 >Each FAM rule emitted a fit-weighted output fuzzy set o, at each iteration >The total output O added these weighted outputs 0=∑0=∑mim(S,f0 f,denotes the anteacedent fitvalue and S,represents the sequent fuzzy set of steering- angle values in the ith FAM rule 以前的fuzzy system用pairwise maximum来合并输出序列。当FAM 规则数增多时，会产生一个平滑分布的输出序列。 8
8 Backing up a Truck -with a Fuzzy System 相关最小FAM推理 i o i f ➢Each FAM rule emitted a fit-weighted output fuzzy set at each iteration ➢The total output O added these weighted outputs denotes the anteacedent fit value and represents the sequent fuzzy set of steeringangle values in the ith FAM rule Si 以前的fuzzy system用pairwise maximum来合并输出序列。当FAM 规则数增多时，会产生一个平滑分布的输出序列
Backing up a Truck -with a Fuzzy System Centroid Defuzzification ∑0，mo(0,D FAM RULE 12: ◆RV and i CE 0 thes a PS. mo(0) PAM RULE 17: f◆VE and X is CB 边nghZ2 Input Input X 详细过程可以参见书中11章 Fuzry comtroller 例 P390,Fig11.4-11.5 eap■l8
9 Backing up a Truck -with a Fuzzy System Centroid Defuzzification = = = p j j p j j j mo mo 1 1 ( ) ( ) 例一 详细过程可以参见书中11章 P390,Fig11.4-11.5
Backing up a Truck -with a Fuzzy System Kosto.系统仿真结果 Project Run Project Run Run Truck Fuzzy Truck Backe Fuzzy Truck Backe Project Run Run Truck Restart Truck Run Truck Fuzzy Truck Backe Restart Truck Truck Ya Situation Restart Truck Truck Ya Situation Truck Ya! Situation x=20 30 X=30 y-20 y10 y=40
10 Backing up a Truck -with a Fuzzy System Kosto系 统 仿 真 结 果 x=20 y=20 x=30 y=10 x=30 y=40