Penerapan Machine Learning Untuk Deteksi Kesalahan Pada Skema Speed Sensorless Fault Tolerant Control Di Motor Dc

Prawitaningrum, Alfiya (2022) Penerapan Machine Learning Untuk Deteksi Kesalahan Pada Skema Speed Sensorless Fault Tolerant Control Di Motor Dc. Other thesis, Institut Teknologi Sepuluh Nopember Surabaya.

Text 02311840000002-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2024. Download (2MB) | Request a copy

Abstract

Speed sensorless kontrol merupakan sistem pengendalian kecepatan tanpa sensor yang mengunakan observer untuk mendapatkan estimasi kecepatan. Sebagai sistem pengendalian, sistem ini dapat dipengaruhi oleh gangguan seperti perubahan torsi beban dan kesalahan sensor. Oleh karena itu, pada penelitian ini diterapkan fault tolerant control (FTC) pada sistem kendali tanpa sensor untuk mengatasi kedua kesalahan tersebut. Skema kontrol FTC memerlukan fault detection and isolation (FDI) untuk mengetahui waktu terjadinya kesalahan serta lokasinya. FDI dibangun dengan menerapkan dengan machine learning berupa classifier, yang bertujuan melakukan pembelajaran pintar sebagai sistem deteksi kesalahan. Pada sistem ini nilai estimasi gangguan diperoleh dari extended state observer. Estimasi gangguan tersebut diproses secara frame based untuk diambil ekstraksi fitur sebagai input untuk FDI. Dari hasil simulasi diperoleh kesimpulan bahwa penerapan classifier decision tree mampu mengidentifikasi secara baik kondisi motor dari dataset secara offline dengan akurasi 91%. Adapun integrasi classifier dengan sistem kontrol FTC juga mampu bekerja dengan cukup baik dimana kondisi perubahan torsi beban menghasilkan error steady state 0%, sedangkan untuk kesalahan sensor menghasilkan error steady state 4,27%
=====================================================================================================================================
Speed sensorless control is a sensorless speed control system that uses an observer to obtain an estimate speed. As a control system, it can be affected by disturbances such as changes in load torque and error in sensor. Therefore, in this study, fault tolerant control (FTC) is applied to the sensorless control system to overcome these errors. The FTC requires fault detection and isolation (FDI) to find out when the error occurred and its location. FDI is built by applying machine learning classifier, which aims to perform smart learning as an error detection system. Value of the disturbance was estimated from an extended state observer which then taken into a frame based process for feature extraction as input for FDI. This research tries to apply FDI with machine learning in the form of a classifier which is expected to be able to do smart learning in terms of fault detection. From the simulation results, it can be concluded that the application of the decision tree classifier is able to properly identify the condition of the motor from offline datasets with an accuracy of 91%. The integration of the classifier along with the FTC also able to work quite well where the load change conditions produce a steady state error of 0%, while for sensor errors it produces a steady state error of 4.27%.

Item Type:	Thesis (Other)
Additional Information:	RSF 629.8 Pra p-1 2022
Uncontrolled Keywords:	Classifier, Fault Tolerant Control, Motor DC, Speed sensorless, Sistem Deteksi Kesalahan
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ213 Automatic control. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK2785 Electric motors, Induction.
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	EKO BUDI RAHARJO
Date Deposited:	16 Jan 2023 02:08
Last Modified:	16 Jan 2023 02:08
URI:	http://repository.its.ac.id/id/eprint/95399

Actions (login required)

View Item