Scaled Conjugate Gradient Based Adaptive ANN Control for SVM-DTC Induction Motor Drive

Abstract : In this work, an Artificial Neural Network (ANN) is developed to improve the performance of Space Vector Modulation (SVM) based Direct Torque Controlled (DTC) Induction Motor (IM) drive. The ANN control algorithm based on Scaled Conjugate Gradient (SCG) method is developed. The algorithm is tested on MATLAB Simulink platform. Results show smooth steady state operation as well as fast and dynamic transient performance. This is due to the SCG training algorithm of ANN which has the benchmarked performance against the standard Back-propagation (BP) algorithm. BP uses gradient descent optimization theory which has user selected parameters; learning rate and momentum constant. The network is trained offline and has fixed parameters. This leads to extra control effort and demands for online tuning of the parameters. SCG algorithm tunes these parameters with the use of second order approximation. Additionally, it takes less learning iterations and hence results in faster learning. Robustness to parameter variations and disturbances is the basic advantage of ANN, thus effectively controlling inherently non linear IM.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [19 references]  Display  Hide  Download

https://hal.inria.fr/hal-01557626
Contributor : Hal Ifip <>
Submitted on : Thursday, July 6, 2017 - 1:55:25 PM
Last modification on : Friday, December 1, 2017 - 1:16:26 AM
Document(s) archivé(s) le : Wednesday, January 24, 2018 - 2:41:50 AM

File

430537_1_En_33_Chapter.pdf
Files produced by the author(s)

Licence


Distributed under a Creative Commons Attribution 4.0 International License

Identifiers

Citation

Lochan Babani, Sadhana Jadhav, Bhalchandra Chaudhari. Scaled Conjugate Gradient Based Adaptive ANN Control for SVM-DTC Induction Motor Drive. 12th IFIP International Conference on Artificial Intelligence Applications and Innovations (AIAI), Sep 2016, Thessaloniki, Greece. pp.384-395, ⟨10.1007/978-3-319-44944-9_33⟩. ⟨hal-01557626⟩

Share

Metrics

Record views

72

Files downloads

7