Controle inteligente de força em atuadores eletro-hidráulicos
Electro-hydraulic actuators are devices commonly applied in several services within the industry. Hence, for such processes to be done satisfactorily, it is desired that the system presents the best accuracy related to its position and force applied. From this direction comes the theme of the presen...
Na minha lista:
Autor principal: | |
---|---|
Outros Autores: | |
Formato: | bachelorThesis |
Idioma: | pt_BR |
Publicado em: |
Universidade Federal do Rio Grande do Norte
|
Assuntos: | |
Endereço do item: | https://repositorio.ufrn.br/handle/123456789/48509 |
Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
Resumo: | Electro-hydraulic actuators are devices commonly applied in several services within the industry. Hence, for such processes to be done satisfactorily, it is desired that the system presents the best accuracy related to its position and force applied. From this direction comes the theme of the present work, whose goal is to evaluate through numerical simulations the performance of an intelligent control technique which combines a non-linear approach of Feedback Linearization with an Artificial Neural Network on the model of an electro-hydraulic actuator to control its applied force. This network is the chosen compensation strategy that is responsible for estimating the non-linearity of a dead zone included in the model of the plant dynamics. In order to present one more control alternative for comparison purposes, the non-linear method combined with a purely adaptive compensating algorithm is also developed. The three control strategies are simulated through two force trajectories tracking, sinusoidal and step. According to a best performance scale related to the mean absolute error (MAE), the intelligent controller was in first place, with MAE_sin = 0,0035 N and MAE_step = 0,0816 N followed by, respectively, the adaptive method, with MAE_sin = 0,0103 N and MAE_step = 0,0936 N, and the execution of the Feedback Linearization strategy with no compensation, with MAE_sin = 0,3837 N and MAE_step = 0,3953 N. With a significant performance improvement and a similar control effort, the insertion of the network to the non-linear technique was able to estimate not only the dead zone, but also all the dynamics of the system which was initially set as unknown to the controller, then, the results indicated the feasibility and efficiency of the method. |
---|