Efficient training of neural nets for nonlinear adaptive filtering using a recursive Levenberg-Marquardt algorithm

Ngia, L.S.H.; Sjöberg, Jonas

doi:10.1109/78.847778

Cited by 162 publications

(76 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The second derivative of the cost function in Equation (4) with respect to , known as the Hessian matrix , can be approximated classically [17], [18] by (7) represents the output sensitivity function defined by partial derivative of the predicted output with respect to the parameters vector as:…”

Section: Proposed Identification Methodsmentioning

confidence: 99%

“…Therefore, we introduce the proposed RLM algorithm that optimizes the following least square cost function , using the forgetting-factor mechanism [17], [18] as:…”

Section: Proposed Identification Methodsmentioning

confidence: 99%

“…The RLM method was proposed and successfully applied to various fields [17], [18], [19]. In [17], RLM algorithm has been presented for on-line training of neural nets for nonlinear adaptive filtering.…”

Section: Introductionmentioning

confidence: 99%

“…In [17], RLM algorithm has been presented for on-line training of neural nets for nonlinear adaptive filtering. Leading to improve performance, the adaptive RLM algorithm has been proposed in [18] for per-tone equalisation in discrete multitone systems.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

On -Line Stator Winding Inter -Turn Short -Circuits Detection in Induction Motors Using Recursive Levenberg -Marquardt Algorithm

Kouadri¹

2017

ijeei

View full text Add to dashboard Cite

Within the framework of the diagnosis of the stator windings faults, the authors propose in this paper a recursive diagnosis method for on-line detection and location of an inter-turn short-circuit by parameters identification. This approach based on the Recursive Levenberg-Marquardt (RLM) algorithm is used for the minimization of the objective function represented by the quadratic criterion obtained by the difference between the real outputs and their estimations. Tests and validations of failure detection by parameter identification require a model suited for fault modelling. For this purpose, a faulty induction motor model is proposed. Electric parameters of this model as well as fault parameters are estimated by RLM adaptive algorithm through an output-error technique. Because the parameters values of this model present magnitude order very different, the normalization of these parameters is proposed in order to obtain the sensitivity functions with the same magnitude order. The estimation results, which used simulated data, are presented to show the effectiveness and the advantage of the proposed approach for use in real-time stator faults diagnosis.

show abstract

Section: Proposed Identification Methodsmentioning

confidence: 99%

“…Therefore, we introduce the proposed RLM algorithm that optimizes the following least square cost function , using the forgetting-factor mechanism [17], [18] as:…”

Section: Proposed Identification Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

On -Line Stator Winding Inter -Turn Short -Circuits Detection in Induction Motors Using Recursive Levenberg -Marquardt Algorithm

Kouadri¹

2017

ijeei

View full text Add to dashboard Cite

show abstract

“…Here a recursive Levenberg Marquardt algorithm proposed by Ngia and Sjöberg [10] is used. This is essentially a regularised implementation of the Recursive Prediction Error algorithm and is defined as follows:…”

Section: Neural Network Implementationmentioning

confidence: 99%

Sequential Learning for Adaptive Critic Design: An Industrial Control Application

Govindhasamy

McLoone²,

Irwin³

2005 IEEE Workshop on Machine Learning for Signal Processing

View full text Add to dashboard Cite

This paper investigates the feasibility of applying reinforcement learning (RL) concepts to industrial process optimisation. A model-free action-dependent adaptive critic design (ADAC), coupled with sequential learning neural network training, is proposed as an online RL strategy suitable for both modelling and controller optimisation. The proposed strategy is evaluated on data from an industrial grinding process used in the manufacture of disk drives.Comparison with a proprietary control system shows that the proposed RL technique is able to achieve comparable performance without any manual intervention.

show abstract

Photofading of an Extended BOPHY Chromophore Dispersed in Poly(methyl methacrylate) as a Chemical Actinometer

2018

View full text Add to dashboard Cite

The target compound is a member of the symmetric pyrrole-BF 2 family, known as BOPHY, which has been equipped with 3,8styryl groups that push the absorption maximum to lower energy. To assist solubility and facile dispersion in plastic films, poly(ethylene glycol) chains are added to the styryl groups. The photophysical properties are somewhat sensitive to changes in solvent polarity and temperature due to the imposition of weak, intramolecular charge-transfer character. The chromophore bleaches under white-light illumination to form an intermediate species that absorbs and emits at higher energy. This species is believed to be a BOPHY unit with one styryl arm removed from the conjugation path. The intermediate is quite resistant to photobleaching but does fade on prolonged illumination. The kinetics associated with photobleaching in poly(methyl methacrylate) films allow development of a simple model that can be used to establish a robust chemical actinometer for recording exposure to sunlight. Similar photofading behaviour is observed in polar solution, where self-catalysis is observed, with the overall kinetics depending on the nature of the solvent.[a] O.

show abstract

Efficient training of neural nets for nonlinear adaptive filtering using a recursive Levenberg-Marquardt algorithm

Cited by 162 publications

References 18 publications

On -Line Stator Winding Inter -Turn Short -Circuits Detection in Induction Motors Using Recursive Levenberg -Marquardt Algorithm

On -Line Stator Winding Inter -Turn Short -Circuits Detection in Induction Motors Using Recursive Levenberg -Marquardt Algorithm

Sequential Learning for Adaptive Critic Design: An Industrial Control Application

Photofading of an Extended BOPHY Chromophore Dispersed in Poly(methyl methacrylate) as a Chemical Actinometer

Contact Info

Product

Resources

About