索春鳳,王貴君
(天津師范大學數(shù)學科學學院,天津 300387)
基于最小推理機的模糊系統(tǒng)一階逼近性
索春鳳,王貴君
(天津師范大學數(shù)學科學學院,天津 300387)
推理機在模糊系統(tǒng)的前件模糊集運算中起著關(guān)鍵性作用.通過引入最小推理機(最小算子),重新建立一種模糊系統(tǒng),利用多元函數(shù)微分中值定理和最小運算的性質(zhì)證明了該模糊系統(tǒng)對連續(xù)可微函數(shù)具有一階逼近性.
最小推理機;模糊系統(tǒng);微分中值定理;最大模;一階逼近性
自日本學者Takagi-Sugeno[1]首次提出T-S模糊系統(tǒng)以來,人們對該模糊系統(tǒng)的研究興趣與日俱增.文獻[2-3]利用基函數(shù)和分層思想討論了模糊系統(tǒng)的泛逼近性,得到了一些有益的結(jié)果.文獻[4]提出了分片線性函數(shù)的概念,并以此為橋梁討論了廣義T-S模糊系統(tǒng)對P-可積函數(shù)的逼近性.文獻[5]借助分片線性函數(shù)研究了廣義Mamdani模糊系統(tǒng)在K-積分模下的逼近性.然而,文獻[4-5]雖然借助分片線性函數(shù)從理論上證明了一些模糊系統(tǒng)的逼近性能,但并沒有給出分片線性函數(shù)的具體解析式.文獻[6]利用超平面思想構(gòu)造了一個分片線性函數(shù)的具體解析表達式,從而為繼續(xù)探究模糊系統(tǒng)的逼近性提供了一個方便工具.本研究基于最小推理機(最小算子)、單點模糊化和中心平均解模糊化構(gòu)造了一般模糊系統(tǒng)模型,并利用多元函數(shù)微分中值定理證明了該系統(tǒng)對連續(xù)可微函數(shù)具有一階逼近性.
本節(jié)首先給出一致完備標準前件模糊集的定義;其次,基于最小推理機、單點模糊化和中心平均解模糊化重新構(gòu)造一種模糊系統(tǒng).
圖1 xi軸上一組一致完備標準的模糊集Fig.1 A set of uniform complete fuzzy sets on xiaxis
事實上,該模糊系統(tǒng)共有M=N1×N2×…×Nd條IF-THEN模糊規(guī)則,每條規(guī)則形式為
其中:i1=1,2,…,N1;i2=1,2,…,N2;…;id=1,2,…,Nd;而為第i1i2…id條規(guī)則輸出模糊集,不妨用表示模糊集的中心.
下面,根據(jù)式(1)中N1×N2×…×Nd條規(guī)則,采用最小推理機、單點模糊化和中心平均解模糊化構(gòu)造一般模糊系統(tǒng)f(x)如下:
其中x=(x1,x2,…,xd)∈U?Rd,且按定義的完備性知式(2)分母部分恒不為0.
下面針對式(2)所確定的數(shù)學模型,利用多元微分中值定理證明一般模糊系統(tǒng)f(x)在最大模意義下對連續(xù)可微函數(shù)具有一階逼近性.
引理設(shè)多元函數(shù)f(x1,x2,…,xd)在凸區(qū)域D?Rd內(nèi)的點上均可微,則對區(qū)域D內(nèi)任意2點P(a1,a2,…,ad)、Q(a1+h1,a2+h2,…,ad+hd),必存在θ∈(0,1),使得
[1]TAKAGI T,SUGENO M.Fuzzy identification of system and its applications to modeling and control[J].IEEE Transactions on Systems,Man,Cybern,1985,15(1):116-132.
[2]WANG L X,MENDEL J.Fuzzy basis functions,universal approximation,and orthogonal least-squares learning[J].IEEE Transanctions Neural Networks,1992,3(5):807-814.
[3]WANG L X.Universal approximation by hierarchical fuzzy systems[J].Fuzzy Set and Systems,1998,93(1):223-230.
[4]劉普寅,李洪興.廣義模糊系統(tǒng)對于可積函數(shù)的逼近性[J].中國科學:E輯,2000,30(5):413-423.LIU P Y,LI H X.Approximation of a generalized fuzzy system to integrable functions[J].Science in China:Series E,2000,30(5):413-423(in Chinese).
[5]王貴君,李曉萍,隋曉琳.廣義Mamdani模糊系統(tǒng)依K-積分模的泛逼近性及其實現(xiàn)過程[J].自動化學報,2014,40(1):143-148.WANG G J,LI X P,SUI X L.Universal approximation and its realization of generalized Mamdani fuzzy system based on K-integral norms[J].Acta Automatica Sinica,2014,40(1):143-148(in Chinese).
[6]彭維玲.基于剖分模糊系統(tǒng)輸入空間的多維分片線性函數(shù)的構(gòu)造及逼近[J].系統(tǒng)科學與數(shù)學,2014,34(3):340-351.PENG W L.Structure and approximation of a multi-dimensional piecewise linear function based on the input space of subdivision fuzzy systems[J].Journal of Systems Science and Mathematical Sciences,2014,34(3):340-351(in Chinese).
[7]王立新.模糊系統(tǒng)與模糊控制教程[M].北京:清華大學出版社,2003.WANG L X.Courses of Fuzzy System and Fuzzy Control[M].Beijing:Tsinghua University Press,2003(in Chinese).
(責任編校馬新光)
First-order approximation of fuzzy system based on minimum inference engine
SUO Chunfeng,WANG Guijun
(College of Mathematical Science,Tianjin Normal University,Tianjin 300387,China)
Inference engine plays a key role in the antecedent fuzzy sets of fuzzy system.A new fuzzy system is established by introducing minimum inference engine(minimum operator),and through the differential mean value theorem for multivariate function and some properties of the minimum operation,the first-order approximation of the fuzzy system to a continuous differentiable function is proved
minimum inference engine;fuzzy system;differential mean value theorem;maximum norm;the first-order approximation
O159;TP183
A
1671-1114(2016)02-0010-03
2015-06-20
國家自然科學基金資助項目(61374009).
索春鳳(1990—),女,碩士研究生.
王貴君(1962—),男,教授,主要從事模糊神經(jīng)網(wǎng)絡(luò)、模糊系統(tǒng)、模糊測度與積分方面的研究.