Automated generation of feedforward control using feedback linearization of local model networks

In this paper, an actor critic neural network-based adaptive control scheme for micro-electro-mechanical system (MEMS) gyroscopes suffering from multiresource disturbances is proposed. Faced with multiresource interferences consisting of parametric uncertainties, strong couplings between axes, Coriolis forces, and variable external disturbances, an actor critic neural network is introduced, where the actor neural network is employed to estimate the packaged disturbances and the critic neural network is utilized to supervise the system performance. Hence, strong robustness against uncertainties and better tracking properties can be derived for MEMS gyroscopes. Aiming at handling the nonlinearities inherent in gyroscopes without analytically differentiating the virtual control signals, dynamic surface control (DSC) rather than backstepping control method is employed to divide the 2nd order system into two 1st order systems and design the actual control policy. Moreover, theoretical analyses along with simulation experiments are conducted with a view to validate the effectiveness of the proposed control approach.

show abstract

“…Kinetic Model of MEMS Gyroscopes. By resorting to [19][20][21][22], the kinetic model of MEMS gyroscopes is typically expressed as…”

Section: Problem Statementsmentioning

confidence: 99%

“…with u mi being the modification control term, which can be derived from (19) and (20). σ ai , σ ci , κ ai , and κ ci are commonly positive constants.…”

Section: Controller Designmentioning

confidence: 99%

Actor Critic Neural Network-Based Adaptive Control for MEMS Gyroscopes Suffering from Multiresource Disturbances

Jing

Qiao

2021

Mathematical Problems in Engineering

View full text Add to dashboard Cite

show abstract

“…The other sub-methods of the ANN refer to other research, such as ANN training, adjustment of the criterion weights, the activation function, the flow of information in a network, and reduction in error. The attributes are, respectively, supervised learning (Fachrurrazi et al, 2017a), the back propagation algorithm (Taghavifar et al, 2014;Zuna et al, 2016), sigmoid (Kusumoputro et al, 2016), feed forward (Euler-Rolle et al, 2016), and gradient descent (Kim et al, 2004). Figure 1 The supervised learning of ANN (Fachrurrazi et al, 2017a) The architecture of the ANN has been built using a series of inputs, output layers, hidden layers, and number of nodes.…”

Section: Framework Of the Ann Modelmentioning

confidence: 99%

The Subcontractor Selection Practice using ANN-Multilayer

Fachrurrazi¹,

Husin²,

Munirwansyah³

et al. 2017

IJTech

View full text Add to dashboard Cite

The practice of subcontracting selection emphasizes two important goals: the company's strategic goal to maximize profits by partnering with subcontractors and the project's operational goal for obtaining qualified subcontractors. Both goals are achieved by formulating the best multi-criteria weights. This is not easy to implement due to differences in subjectivity, viewpoint, and other consideration of assessors, but prioritizing the criterion weights can reduce these differences. This study presents an ANN (Artificial Neural Network) with the ability to generalize data. The purpose of the study is to develop an ANN model for subcontracting selection and to identify significant criteria related to the company's strategic goal. The initial training of the proposed ANN model utilized 40 subcontractor selection datasets containing data in the form of a subcontractor selection scheme consisting of 20 criteria and 5 major groups. Training of ANN model was successful with MSE learning at 1.37269e -7 , MSE validation at 0.07985, and epoch 600 to 800. The quotation price is the significant criterion of the selection, and it has a great outcome for the contractor strategic goal. The interaction between the subcontractor selection practice and the ANN model shows that the ANN has an important role in the subcontractor selection practice.

show abstract

“…No trabalho de (EULER-ROLLE et al, 2016) um modelo simples é introduzido para criar automaticamente uma lei de controle em um sistema dinâmico não linear utilizando um modelo de rede neural local de tempo discreto -Discrete-time Local Model…”

Section: Revisão Bibliográficaunclassified

Sistemas linearizados por relimentação fuzzy evolutiva

Silva¹

View full text Add to dashboard Cite

de Souza, que tanto apoio me deram durante esta longa jornada e à quem eu os dedico este trabalho. Aos colegas e amigos do Laboratório de Sinais e Sistemas do CEFET-MG campus Divinópolis, em especial o Professor Msc. Lucas Silva Oliveira e ao Professor Dr Valter Júnior de Souza Leite pela valiosa contribuição na parte experimental desta pesquisa. Aos colegas do LCA/FEEC pelos momentos de alegria compartilhados. Aos colegas da Universidade Federal de Roraima pela confiança, respeito e profissionalismo em especial à Professora Dra. Gioconda Santos e Souza Martinez. Um agradecimento especial à TODOS os professores do DCA/FEEC, em particular, ao Professor Dr Fernando Antônio Campos Gomide pelo compromisso e confiança na orientação desta pesquisa. À Universidade Estadual de Campinas, pela oportunidade. ResumoA linearização por realimentação é uma poderosa abordagem de controle, porém pode apresentar fragilidades diante de erros de modelagem. A falta de robustez desta abordagem de controle pode levar a um desempenho abaixo do esperado ou até mesmo à instabilidade. Esta dissertação sugere uma estratégia para melhorar a robustez em sistemas linearizados por realimentação usando um mecanismo do tipo modelo de referência e um algoritmo de aprendizagem participativa evolutiva granular fuzzy. A linearização por realimentação granular evolutiva robusta fuzzy é uma maneira de controlar sistemas não lineares com incertezas e assegurar a estabilidade da malha fechada. O resultado é uma abordagem mais robusta de controle em malha fechada em que a aprendizagem participativa evolutiva fuzzy é empregada para estimar as incertezas em tempo real e mitigar seus efeitos no sistema. A dissertação também aborda aspectos teóricos e práticos de um sistema de controle de nível em um tanque não linear, e mostra que o desempenho do controlador granular evolutivo robusto é superior à abordagens alternativas propostas na literatura como Classic Feedback Linearization, Fuzzy Model Reference Learning Control e Indirect Adaptive Fuzzy Control. Além disso, a dissertação discute os aspectos práticos e testes experimentais da linearização por realimentação granular evolutiva robusta fuzzy implementado em um controlador de nível de um tanque real de laboratório.Palavras-chaves: Controle robusto; Sistemas evolutivos, Sistemas de controle por realimentação.

show abstract

Automated generation of feedforward control using feedback linearization of local model networks

Cited by 7 publications

References 44 publications

Actor Critic Neural Network-Based Adaptive Control for MEMS Gyroscopes Suffering from Multiresource Disturbances

Actor Critic Neural Network-Based Adaptive Control for MEMS Gyroscopes Suffering from Multiresource Disturbances

The Subcontractor Selection Practice using ANN-Multilayer

Sistemas linearizados por relimentação fuzzy evolutiva

Contact Info

Product

Resources

About