Fast self-learning multivariable fuzzy controllers constructed from a modified CPN network

Nie, Junhong; Linkens, D.A.

doi:10.1080/00207179408921470

Cited by 33 publications

(21 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several learning algorithms developed in the field of artificial neural networks have been applied to fuzzy system learning. Supervised learning based on the gradient method [15], unsupervised learning [16], and reinforcement learning [17] have proved to be very effective to improve the performance of the fuzzy systems. Furthermore, fuzzy systems that can learn from data will be much more objective and the knowledge they acquired is believed to be more profound.…”

Section: B Global Parameter Learningmentioning

confidence: 99%

Fuzzy modeling of high-dimensional systems: complexity reduction and interpretability improvement

Jin

2000

IEEE Trans. Fuzzy Syst.

431

187

View full text Add to dashboard Cite

Abstract-Fuzzy modeling of high-dimensional systems is a challenging topic. This paper proposes an effective approach to data-based fuzzy modeling of high-dimensional systems. An initial fuzzy rule system is generated based on the conclusion that optimal fuzzy rules cover extrema [8]. Redundant rules are removed based on a fuzzy similarity measure. Then, the structure and parameters of the fuzzy system are optimized using a genetic algorithm and the gradient method. During optimization, rules that have a very low firing strength are deleted. Finally, interpretability of the fuzzy system is improved by fine training the fuzzy rules with regularization. The resulting fuzzy system generated by this method has the following distinct features: 1) the fuzzy system is quite simplified; 2) the fuzzy system is interpretable; and 3) the dependencies between the inputs and the output are clearly shown. This method has successfully been applied to a system that has 11 inputs and one output with 20 000 training data and 80 000 test data.

show abstract

Section: B Global Parameter Learningmentioning

confidence: 99%

Fuzzy modeling of high-dimensional systems: complexity reduction and interpretability improvement

Jin

2000

IEEE Trans. Fuzzy Syst.

431

187

View full text Add to dashboard Cite

show abstract

“…CFNN, a combination of CPN and fuzzy arithmetic, was first introduced by Nie and Linkens (1994). The architecture of CFNN consists of an input layer, a Kohonen layer, and a Grossberg layer (see Fig.…”

Section: Counterpropagation Fuzzy Neural Network (Cfnn)mentioning

confidence: 99%

Auto-control of pumping operations in sewerage systems by rule-based fuzzy neural networks

Chiang

Chang

Tsai

et al. 2011

Hydrol. Earth Syst. Sci.

View full text Add to dashboard Cite

Abstract.Pumping stations play an important role in flood mitigation in metropolitan areas. The existing sewerage systems, however, are facing a great challenge of fast rising peak flow resulting from urbanization and climate change. It is imperative to construct an efficient and accurate operating prediction model for pumping stations to simulate the drainage mechanism for discharging the rainwater in advance. In this study, we propose two rule-based fuzzy neural networks, adaptive neuro-fuzzy inference system (ANFIS) and counterpropagation fuzzy neural network for on-line predicting of the number of open and closed pumps of a pivotal pumping station in Taipei city up to a lead time of 20 min. The performance of ANFIS outperforms that of CFNN in terms of model efficiency, accuracy, and correctness. Furthermore, the results not only show the predictive water levels do contribute to the successfully operating pumping stations but also demonstrate the applicability and reliability of ANFIS in automatically controlling the urban sewerage systems.

show abstract

“…Until now, a lot of efforts have been made to realize such features. To provide the fuzzy system with the first function abovementioned, Yamaguchi et al [11] and Nie et al [12] use BAM and CPN neural network, respectively, whilst Nakamori et al [13] select the clustering technique. As for the second function, backpropagation neural networks [14], pi-sigma neural network [15] and simulated annealing [16] are attempted for supervised parameter learning, while a neuron-like structure is used as reinforcement learning [17].…”

Section: Off-line Training and Optimization Of The Feedforward Fumentioning

confidence: 99%

“…Given the desired trajectory and provided the feedforward torques are acceptably accurate, (12) can be linearized along the nominal trajectory (13) where and are the gradients of evaluated at and , respectively, , and and are the nominal values of and . Let and , then we have the following perturbation equation for the robot system: (14) As a result, the control problem reduces to producing a proper so that converges to zero.…”

Section: A Decentralized Perturbation Model Of the Robotmentioning

confidence: 99%

“…The dynamics of a robot in (1) can be expressed in the following form [26]: (12) where are the position and velocity vectors, respectively, and is the number of the degree of the freedom. With this formulation, we intended to find a feedback control law such that the closedloop system is asymptotically stable and tracks the desired trajectory as close as possible.…”

Section: A Decentralized Perturbation Model Of the Robotmentioning

confidence: 99%

See 1 more Smart Citation

Decentralized adaptive fuzzy control of robot manipulators

Jin

1998

IEEE Trans. Syst., Man, Cybern. B

View full text Add to dashboard Cite

Abstract-This paper develops a decentralized adaptive fuzzy control scheme for robot manipulators via a combination of genetic algorithm and gradient method. The controller for each link consists of a feedforward fuzzy torque-computing system and a feedback fuzzy PD system. The feedforward fuzzy system is trained and optimized off-line by an improved genetic algorithm, that is to say, not only the parameters but also the structure of the fuzzy system are self-organized. Because genetic algorithm can operate successfully without the system model, no exact inverse dynamics of the robot system are required. The feedback fuzzy PD system, on the other hand, is tuned on-line using gradient method. In this way, the proportional and derivative gains are adjusted properly to keep the closed-loop system stable. The proposed controller has the following merits: 1) it needs no exact dynamics of the robot systems and the computation is time-saving because of the simple structure of the fuzzy systems; and 2) the controller is insensitive to various dynamics and payload uncertainties in robot systems. These are demonstrated by analyses of the computational complexity and various computer simulations.

show abstract

Fast self-learning multivariable fuzzy controllers constructed from a modified CPN network

Cited by 33 publications

References 14 publications

Fuzzy modeling of high-dimensional systems: complexity reduction and interpretability improvement

Fuzzy modeling of high-dimensional systems: complexity reduction and interpretability improvement

Auto-control of pumping operations in sewerage systems by rule-based fuzzy neural networks

Decentralized adaptive fuzzy control of robot manipulators

Contact Info

Product

Resources

About