An enhanced discriminability recurrent fuzzy neural network for temporal classification problems

Wu, Gin-Der; Zhu, Zhitao

doi:10.1016/j.fss.2013.05.007

Cited by 32 publications

(9 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, the future research of this study are listed as follows: 1) the structure determination of this novel control scheme can be improved by using self-constructing [44][45][46][47][48][49][50][51][52]; 2) the learning rate can be chosen in an optimal way; 3) the solution to RFCMAC's memory requirement will be further studied and improved using the hierarchical technique; 4) since the proposed scheme is considering two input variables only, further investigation can be underway to expand it to a more generalized case of multi-variables.…”

Section: Discussionmentioning

confidence: 99%

Recurrent Fuzzy Neural Cerebellar Model Articulation Network Fault-Tolerant Control of Six-Phase Permanent Magnet Synchronous Motor Position Servo Drive

Lin

Sun

Yang

et al. 2016

IEEE Trans. Fuzzy Syst.

101

View full text Add to dashboard Cite

A recurrent fuzzy neural cerebellar model articulation network (RFNCMAN) fault-tolerant control of a six-phase permanent magnet synchronous motor (PMSM) position servo drive is proposed in this study. First, the fault detection and operating decision method of the six-phase PMSM position servo drive is developed. Then, an ideal computed torque controller is designed for the tracking of the rotor position reference command. In general, it is impossible to design an ideal computed control law owing to the uncertainties of the six-phase PMSM position servo drive, which are difficult to know in advance for practical applications. Therefore, the RFNCMAN, which combined the merits of a recurrent fuzzy cerebellar model articulation network (RFCMAN) and a recurrent fuzzy neural network (RFNN), is proposed to estimate a nonlinear equation included in the ideal computed control law with a robust compensator designed to compensate the minimum reconstructed error. Furthermore, the adaptive learning algorithm for the online training of the RFNCMAN is derived using the Lyapunov stability to guarantee the closed-loop stability. Finally, the proposed RFNCMAN fault-tolerant control system is implemented in a 32-bit floating-point DSP. The effectiveness of the six-phase PMSM position servo drive using the proposed intelligent fault-tolerant control system is verified by some experimental results. Index Terms-Recurrent fuzzy neural cerebellar model articulation network (RFNCMAN), fault-tolerant control, six-phase permanent magnet synchronous motor (PMSM), Lyapunov stability, Taylor series expansion.

show abstract

Section: Discussionmentioning

confidence: 99%

Recurrent Fuzzy Neural Cerebellar Model Articulation Network Fault-Tolerant Control of Six-Phase Permanent Magnet Synchronous Motor Position Servo Drive

Lin

Sun

Yang

et al. 2016

IEEE Trans. Fuzzy Syst.

101

View full text Add to dashboard Cite

show abstract

“…If the trend of the above events can be predicted accurately, the probability of risk occurrence can be eliminated or reduced. GM(1,1): For the purpose of reducing or eliminating the bias that existed in the conventional GM(1,1) model [10,11], the unbiased grey model is introduced to fit those statistical data.…”

Section: Node Trust Predictionmentioning

confidence: 99%

Node Trust Assessment and Prediction in Mobile Ad Hoc Networks

Zhou¹,

Xia²

2015

IJSIA

View full text Add to dashboard Cite

show abstract

“…In principle, an RNN is a powerful tool that has been successfully employed in many applications such as system identification and control [3][4], time series prediction [5][6], and some other engineering areas [7][8]. However, RNNs have been proved to be difficult to choose a suitable structure for a particular application [9-10].…”

Section: Introductionmentioning

confidence: 99%

“…The input neurons are set by the environment and the output neurons are computed using the connection weights and the hidden neurons [1-2]. The states of hidden neurons can store information through time.In principle, an RNN is a powerful tool that has been successfully employed in many applications such as system identification and control [3][4], time series prediction [5][6], and some other engineering areas [7][8]. However, RNNs have been proved to be difficult to choose a suitable structure for a particular application [9-10].In fact, if the structures of RNNs are too large, the networks tend to overfitting and poor generalization.…”

mentioning

confidence: 99%

An adaptive growing and pruning algorithm for designing recurrent neural network

2017

View full text Add to dashboard Cite

The training of recurrent neural networks (RNNs) concerns the selection of their structures and the connection weights. To efficiently enhance generalization capabilities of RNNs, a recurrent self-organizing neural networks (RSONN), using an adaptive growing and pruning algorithm (AGPA), is proposed for improving their performance in this paper. This AGPA can self-organize the structures of RNNs based on the information processing ability and competitiveness of hidden neurons in the learning process. Then, the hidden neurons of RSONN can be added or pruned to improve the generalization performance. Furthermore, an adaptive second-order algorithm with adaptive learning rate is employed to adjust the parameters of RSONN. And the convergence of RSONN is given to show the computational efficiency. To demonstrate the merits of RSONN for data modeling, several benchmark datasets and a real world application associated with nonlinear systems modeling problems are examined with comparisons against other existing methods. Experimental results show that the proposed RSONN effectively simplifies the network structure and performs better than some exiting methods.Index Terms-Recurrent self-organizing neural network; adaptive growing and pruning algorithm; information processing ability; competitiveness; convergence. I. INTRODUCTIONN the past decades, recurrent neural networks (RNNs) have received considerable attention. An RNN has a set of neurons and connection weights between the neurons. The input neurons are set by the environment and the output neurons are computed using the connection weights and the hidden neurons [1-2]. The states of hidden neurons can store information through time.In principle, an RNN is a powerful tool that has been successfully employed in many applications such as system identification and control [3][4], time series prediction [5][6], and some other engineering areas [7][8]. However, RNNs have been proved to be difficult to choose a suitable structure for a particular application [9-10].In fact, if the structures of RNNs are too large, the networks tend to overfitting and poor generalization. While if the structures of RNNs are too small, the networks may not be good enough to achieve the desired performance [11][12]. In general, to design the structures of RNNs for improving their generalization performance, there are three ways: constructive algorithm, Recently, the constructive-and-pruning algorithms, executing a constructive phase first and then a pruning phase, have been used to solve some of the above problems [27][28]. Hsu et al. developed a self-constructing RNN (SCRNN) to mimic an ideal computation controller [29]. The results demonstrated that this SCRNN owned high learning ability and low computation complexity. Juang et al. introduced a recurrent self-evolving fuzzy neural network with local feedbacks (RSEFNN-LF) for modeling nonlinear systems. This RSEFNN-LF was constructed via simultaneous structure and parameter learning phases.The simulations results validated the advantages of ...

show abstract

An enhanced discriminability recurrent fuzzy neural network for temporal classification problems

Cited by 32 publications

References 34 publications

Recurrent Fuzzy Neural Cerebellar Model Articulation Network Fault-Tolerant Control of Six-Phase Permanent Magnet Synchronous Motor Position Servo Drive

Recurrent Fuzzy Neural Cerebellar Model Articulation Network Fault-Tolerant Control of Six-Phase Permanent Magnet Synchronous Motor Position Servo Drive

Node Trust Assessment and Prediction in Mobile Ad Hoc Networks

An adaptive growing and pruning algorithm for designing recurrent neural network

Contact Info

Product

Resources

About