scispace - formally typeset
Search or ask a question

Showing papers by "Yang Gao published in 2001"


Journal ArticleDOI
TL;DR: Comprehensive comparisons with other latest approaches show that the proposed approach is superior in terms of learning efficiency and performance.
Abstract: A fast approach for automatically generating fuzzy rules from sample patterns using generalized dynamic fuzzy neural networks (GD-FNNs) is presented. The GD-FNN is built based on ellipsoidal basis functions and functionally is equivalent to a Takagi-Sugeno-Kang fuzzy system. The salient characteristics of the GD-FNN are: (1) structure identification and parameters estimation are performed automatically and simultaneously without partitioning input space and selecting initial parameters a priori; (2) fuzzy rules can be recruited or deleted dynamically; (3) fuzzy rules can be generated quickly without resorting to the backpropagation (BP) iteration learning, a common approach adopted by many existing methods. The GD-FNN is employed in a wide range of applications ranging from static function approximation and nonlinear system identification to time-varying drug delivery system and multilink robot control. Simulation results demonstrate that a compact and high-performance fuzzy rule-base can be constructed. Comprehensive comparisons with other latest approaches show that the proposed approach is superior in terms of learning efficiency and performance.

403 citations




Journal ArticleDOI
TL;DR: An adaptive fuzzy neural controller suitable for multilink manipulators motion control with self-organizing fuzzy neural structure, fast convergence of tracking error, and adaptive control is presented.
Abstract: This paper presents an adaptive fuzzy neural controller suitable for multilink manipulators motion control. The proposed controller has the following salient features: (1) self-organizing fuzzy neural structure; (2) online learning of the robot dynamics; (3) fast convergence of tracking error; and (4) adaptive control. Computer simulation results of a two-link manipulator demonstrate that excellent tracking performance can be achieved under external disturbances.

34 citations



Proceedings ArticleDOI
01 Jan 2001
TL;DR: A new approach towards feature representation for speech recognition, named state transition matrix (STM), is proposed to address temporal varying problem in speech recognition using only a single-layer perceptron neural network.
Abstract: A high performance neural-network-based speech recognition system is presented. A new approach towards feature representation for speech recognition, named state transition matrix (STM), is proposed to address temporal varying problem in speech recognition. Using STM, we need only a single-layer perceptron neural network to perform speech recognition. Experimental results show that an overall accuracy of 95% and 87% was achieved for speaker-dependent isolated word recognition and multi-speaker-dependent isolated word recognition, respectively.

15 citations




Proceedings ArticleDOI
15 Jul 2001
TL;DR: This paper presents a robust adaptive fuzzy neural controller suitable for trajectory control of robot manipulators and asymptotic stability of the control system is established using Lyapunov theorem.
Abstract: This paper presents a robust adaptive fuzzy neural controller suitable for trajectory control of robot manipulators. The proposed controller has the following salient features: 1) self-organizing fuzzy neural structure; 2) online learning; 3) fast learning speed; 4) fast convergence of tracking error; 5) adaptive control; and 6) robust control, i.e. asymptotic stability of the control system is established using Lyapunov theorem. Computer simulation studies were carried out and comparison of simulation results with some existing controllers demonstrate the flexibility, adaptability and good tracking performance of the proposed controller.

8 citations


Proceedings ArticleDOI
27 Jun 2001
TL;DR: This paper presents a robust adaptive fuzzy neural controller suitable for motion control of a multi-link robot manipulator that has the dynamic fuzzy neural networks structure, i.e. fuzzy control rules, can be generated or deleted automatically.
Abstract: This paper presents a robust adaptive fuzzy neural controller suitable for motion control of a multi-link robot manipulator. The proposed controller has the following salient features: (1) the dynamic fuzzy neural networks structure, i.e. fuzzy control rules, can be generated or deleted automatically; (2) adaptive learning; (3) online learning of the robot dynamics; (4) fast learning speed; and (5) fast convergence of tracking error. The global stability of the system is established using the Lyapunov approach. Computer simulation studies of a two-link robot manipulator demonstrate that an excellent tracking performance can be achieved under external disturbances.

7 citations


01 Dec 2001
TL;DR: The developed AFNC consists of a combination of a fuzzy neural network (FNN) controller and a supervisory PD controller that outperforms some of the existing adaptive fuzzy and neural controllers in terms of tracking speed and accuracy.
Abstract: This article presents the design, development, and implementation of a new adaptive fuzzy neural controller (AFNC) suitable for real-time industrial applications. The developed AFNC consists of a combination of a fuzzy neural network (FNN) controller and a supervisory PD controller. The salient features of the AFNC are: (1) dynamic fuzzy neural structure, that is, fuzzy control rules can be generated or deleted automatically; (2) fast on-line learning ability; (3) fast convergence of tracking error; (4) adaptive control; and (5) robust control, where global stability of the system is established using Lyapunov approach. Experimental evaluation conducted on a SEIKO TT-3000 SCARA robot demonstrates that excellent tracking performance can be achieved under time-varying conditions. The proposed controller also outperforms some of the existing adaptive fuzzy and neural controllers in terms of tracking speed and accuracy.

Proceedings ArticleDOI
01 Oct 2001
TL;DR: The unique feature of the HAFC is that no mathematical model of the plant is required and the proposed controller is able to adaptively estimate the bound functions on-line, which are required for determination of the supervisory controller.
Abstract: This paper presents the design, development and implementation of a Hybrid Adaptive Fuzzy Controller (HAFC) suitable for real-time industrial applications. The developed HAFC consists of a weighted combination of the Direct Adaptive Fuzzy Controller (DAFC) and Indirect Adaptive Fuzzy Controller (IAFC) and a gradually activated supervisory controller. The unique feature of the HAFC is that no mathematical model of the plant is required and the proposed controller is able to adaptively estimate the bound functions on-line, which are required for determination of the supervisory controller. The supervisor controller guarantees global stability of the closed-loop system in the sense that all signals are bounded. Simulink, an interactive graphical software for simulating dynamic systems, is used to model, simulate and analyse the dynamic system. The HAFC is implemented in real-time through Real-Time Workshop (RTW). The performance of the HAFC was found to be superior and it matches favourably the simulation results.


Proceedings ArticleDOI
01 Oct 2001
TL;DR: A hierarchical on-line self-organizing learning paradigm is employed so that not only parameters can be adjusted, but also the determination of structure can be self-adaptive without partitioning the input space a priori.
Abstract: This paper presents a new approach, which exploits the recently developed Dynamic Fuzzy Neural Networks (DFNN) learning algorithm The DFNN is based on extended Radial Basis Function (RBF) neural networks, which are functionally equivalent to Takagi-Sugeno-Kang (TSK) fuzzy systems The algorithm comprises 4 parts: (1) Criteria of rules generation; (2) Allocation of premise parameters; (3) Determination of consequent parameters and (4) Pruning technology The salient characteristics of the approach are: (1) A hierarchical on-line self-organizing learning paradigm is employed so that not only parameters can be adjusted, but also the determination of structure can be self-adaptive without partitioning the input space a priori; (2) Fast learning speed can be achieved so that the system can be implemented in real time The application of the proposed approach is demonstrated in application to a demanding, highly nonlinear, missile control design task Scheduling on instantaneous incidence (a rapidly varying quantity) is well known to lead to considerable difficulties with classical gain-scheduling methods It is shown that the methods proposed here can, however, be used to successfully design an effective intelligent controller