PASTIC Dspace Repository

Using Denoising Autoencoders to Predict Behavior of an Inverted Pendulum on a Cart System

Show simple item record

dc.contributor.author Khalid, J
dc.contributor.author Nasir, A
dc.contributor.author Shami, U. T
dc.contributor.author Baig, A
dc.date.accessioned 2022-10-26T09:55:09Z
dc.date.available 2022-10-26T09:55:09Z
dc.date.issued 2017-01-04
dc.identifier.citation Khalid, J., Nasir, A., Shami, U., & Baig, A. (2017). Using Denoising Autoencoders to Predict Behavior of an Inverted Pendulum on a Cart System. University of Engineering and Technology Taxila. Technical Journal, 22(1), 30. en_US
dc.identifier.issn 2313-7770
dc.identifier.uri http://142.54.178.187:9060/xmlui/handle/123456789/13710
dc.description.abstract This paper presents a method for precise prediction of the behavior of an inverted pendulum on a cart system. We have improved the accuracy of prediction beyond what can be achieved through traditional model-based simulation. This improvement has been achieved through learning of the differences between simulation and experimental results. Specifically, a three layered neural network known in the literature as denoising autoencoder has been used for learning. The proposed method consists of three steps. First step is to design linear controller for the inverted pendulum using text book methods and perform simulations. Second step is to perform experiments on the actual hardware of the inverted pendulum in the laboratory using the same controller as in first step. Third step is to learn the difference between simulation results and the results from the experiments using neural networks. Now the learned neural network is used to predict lab experiment results based on simulations with different initial conditions and reference values than the ones used to train the network. We have designed Linear Quadratic Regulator for demonstration of the proposed method. Results from the autoencoder have been reported. It is found that the autoencoder can predict the actual behavior of the pendulum with reasonable accuracy. en_US
dc.language.iso en en_US
dc.publisher Taxila:University of Engineering and Technology(UET)Taxila, Pakistan en_US
dc.subject Denoising Autoencoder en_US
dc.subject LQR Controller en_US
dc.subject Controller Implementation en_US
dc.title Using Denoising Autoencoders to Predict Behavior of an Inverted Pendulum on a Cart System en_US
dc.type Article en_US


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


Advanced Search

Browse

My Account