Informedia Electronic Co.

Dalian, China

Informedia Electronic Co.

Dalian, China
SEARCH FILTERS
Time filter
Source Type

Wang D.-G.,Dalian University of Technology | Wang D.-G.,Informedia Electronic Co. | Chen C.L.P.,University of Macau | Song W.-Y.,Dongbei University of Finance and Economics | Li H.-X.,Dalian University of Technology
IEEE Transactions on Fuzzy Systems | Year: 2015

In this paper, a novel Error-compensated MArginal LINEarization (EMALINE) fuzzy modeling method is proposed. This method models a group of data information to a piecewise linear fuzzy system with high accuracy within a given error bound. It is proved that the fuzzy system generalized by the EMALINE method possesses universal approximation capability for a class of nonlinear systems. In addition, the theoretical approximation error bounds of the fuzzy system generalized by the EMALINE method are established and proved. Theoretical and practical results indicate that theEMALINEhas better approximation accuracy than those of previous approaches. Numerical examples are shown to illustrate the validity of the proposed approach. © 2014 IEEE.


Wang D.-G.,Dalian University of Technology | Wang D.-G.,Informedia Electronic Co. | Song W.-Y.,Dongbei University of Finance and Economics | Shi P.,Victoria University of Melbourne | And 2 more authors.
Information Sciences | Year: 2013

In this paper, a dynamic fuzzy inference marginal linearization (DFIML) method is proposed for modeling nonlinear dynamic systems. This method can transfer a group of input-output data into a time-variant fuzzy system with variable coefficients. It is shown that solutions of time-variant fuzzy systems generalized by DFIML method are universal approximators to solutions of a class of non-autonomous systems. Also the analytical solutions of these time-variant fuzzy systems can be obtained. Finally, a simulation example is provided to illustrate the validity and potential of the developed techniques. © 2012 Elsevier Inc. All rights reserved.


Zhao G.,Heilongjiang University | Zhao C.,Heilongjiang University | Wang D.,Dalian University of Technology | Wang D.,Informedia Electronic Co.
Chinese Control Conference, CCC | Year: 2014

This paper presents a new chattering elimination method, and an optimal adaptive integral sliding mode controller design based on reinforcement learning for translational oscillations by a rotational actuator (TORA) system is demonstrated. At first, we introduce the tensor product model transformation based adaptive integral sliding mode controller. Next, we utilize an adaptive boundary layer width saturation function to get better performance. Reinforcement learning algorithm is employed to find the instantaneous optimal value for the boundary layer width of saturation function appeared in the adaptive integral sliding mode controller. The proposed tensor product model transformation based adaptive integral sliding mode controller with reinforcement learning strategy is verified by TORA system whereas the agent is rewarded for lower chattering, and punished for higher chattering. Simulation results show that chattering can be reduced effectively by incorporating reinforcement learning strategy into the adaptive integral sliding mode controller. © 2014 TCCT, CAA.


Zou L.,Liaoning Normal University | Zou L.,Informedia Electronic Co. | Liu X.,Liaoning Normal University | Pei Z.,Xihua University | Huang D.,Dalian University of Technology
International Journal of Machine Learning and Cybernetics | Year: 2013

We construct a kind of linguistic truth-valued intuitionistic fuzzy lattice based on linguistic truth-valued lattice implication algebras to deal with linguistic truth values. We get some properties of implication operators on the set of ∨-irreducible elements. And furthermore the implication operators on the linguistic truth-valued intuitionistic fuzzy lattice are discussed. The proposed system can better express both comparable and incomparable information. Also it can deal with both positive and negative evidences which are represented by linguistic truth values at the same time during the information processing system. © 2012 Springer-Verlag.


Wang D.-G.,Dalian University of Technology | Wang D.-G.,Informedia Electronic Co. | Song W.-Y.,Dongbei University of Finance and Economics | Li H.-X.,Dalian University of Technology
Neurocomputing | Year: 2015

In this paper, we utilize generalized Bernstein polynomials to construct fuzzy system. Different from traditional Bernstein polynomials, partition of interval on input variable can be chosen as non-equidistant division. We prove that generalized Bernstein fuzzy systems are universal approximators to a given continuous function and its high-order derivatives. Further, ELM method is used to tune the parameters of generalized Bernstein fuzzy system and Spline fuzzy system. It is proved that ELM-Spline fuzzy system can approximate a function and its derivative. Simulation examples show that the proposed ELM-Bernstein fuzzy system and ELM-spline fuzzy system can achieve high approximation capability for nonlinear model. © 2014 Elsevier B.V.


Fang P.,Dalian University of Technology | Wang D.,Dalian University of Technology | Wang D.,Informedia Electronic Co. | Song W.,Dongbei University of Finance and Economics
ICIC Express Letters, Part B: Applications | Year: 2015

In this paper, a novel learning method is proposed for training the parameters of fuzzy wavelet neural network (FWNN). Extreme learning machine (ELM) is utilized to train the linear parameters of FWNN. And the gradient descent (GD) algorithm is used to update the nonlinear parameters. Some numerical examples show that the proposed learning algorithm can achieve high accuracy with fewer epochs. © 2015, ICIC Express Letters Office. All rights reserved.


Yin M.,Liaoning Normal University | Zou L.,Informedia Electronic Co. | Zou L.,Liaoning Normal University | Liu X.,Liaoning Normal University
ICIC Express Letters, Part B: Applications | Year: 2013

In classical logic the truth value of a proposition is true or false. Since there are many fuzzy concepts in the real world, the truth value of a fuzzy proposition is a real number in the interval [0, 1]. Is it a singleton true value for a given proposition? We will give a fuzzy proposition different truth values because of different people or different circumstances. A kind of qualitative fuzzy propositional logic system that can reflect the "elastic" of a fuzzy proposition is introduced. The truth value of fuzzy proposition is not singleton that depends on the context in the real world. Considering a fuzzy proposition one will choose the equivalence relation and get different class. Then based on an equivalence class, a qualitative fuzzy proposition set can be held. Then the resolution method of qualitative first-order logic is discussed. © 2013 ISSN 2185-2766.


Song W.,Dongbei University of Finance and Economics | Wang D.,Dalian University of Technology | Wang D.,Informedia Electronic Co. | Liu Y.,Dalian University of Technology
ICIC Express Letters, Part B: Applications | Year: 2014

In this paper, a fuzzy neural network (FNN) which utilizes Taylor expansion as the consequent of fuzzy rules is proposed for modeling and predicting nonlinear systems. Fuzzy c-means (FCM) method is used to determine the number of fuzzy rules. Extreme learning machine (ELM) is adopted to determine the parameters of fuzzy neural network. Further, a structure-learning algorithm is obtained to determine the number of fuzzy rules and the order of the Taylor expansion. Simulation results show that the proposed model can achieve good approximation capability for some nonlinear systems with simpler structure. © 2014 ICIC International.


Tan R.,Liaoning Normal University | Zou L.,Informedia Electronic Co. | Zou L.,Dalian University of Technology | Yan D.,Liaoning Normal University
ICIC Express Letters, Part B: Applications | Year: 2013

The main task of discretization is to find significant discrete representation of continuous value with the information loss as less as possible. In this paper, based on class-attribute interdependency criterion, a discretization algorithm (called CAICD) is proposed, which considers data distribution and the interdependency between all the classes and attributes, and exploits rough sets technique to get better criterion for discretization. In CAICD algorithm, the class-attribute mutual information is adopted which can automatically control and adjust the extent of the discretization of continuous value, ensure discretization of real value attributes reasonable. © 2013.


Zou L.,Informedia Electronic Co. | Zou L.,Liaoning Normal University | Liu X.,Liaoning Normal University | Huang D.,Dalian University of Technology
World Scientific Proc. Series on Computer Engineering and Information Science 7; Uncertainty Modeling in Knowledge Engineering and Decision Making - Proceedings of the 10th International FLINS Conf. | Year: 2012

A kind of intuitionistic fuzzy logic system LP(S) based on linguistic truth-valued intuitionistic fuzzy algebra is proposed in this paper. The linguistic truth-valued intuitionisitic fuzzy algebra comes from the linguistic truth-valued implication algebra which is fit to express both comparable and incomparable information. The method can deal with the uncertain problem which has both positive evidence and negative evidence at the same time. Some axioms and reasoning rules of linguistic truth-valued intuitionistic fuzzy propositional logic are discussed. The proofs and theorems are also obtained.

Loading Informedia Electronic Co. collaborators
Loading Informedia Electronic Co. collaborators