Machine Learning-Based Online Tuning Approach for One-Input Type-1 Fuzzy PID Controllers

Öz

In this study, a machine learning-based one-input type-1 fuzzy PID (ML-SIT1-FPID) controller design has been developed. The one-input type-1 fuzzy PID controller (SIT1-FPID) consists of a one-input one-output type-1 fuzzy controller (T1-FC) and a PID controller. The T1-FC acts like the non-linear characteristic of a traditional PID controller. In the design of the SIT1-FPID controller, five triangular type-1 membership functions are used for the input variable, while five singular membership functions are used for the output variable. The structure of the ML-SIT1-FPID controller is identical to that of the SIT1-FPID controller. The fundamental difference between them stems from the output membership function. Simulation studies were conducted on a non-linear system under various conditions using the SIT1-FPID controller, and data was collected accordingly. A machine learning approximation was used based on the collected data to derive an error-dependent dynamic equation. The output subscription functions of the ML-SIT1-FPID controller were adjusted using online tuning based on the error-dependent dynamic equation. The performance of the proposed controller was discussed with SIT1-FPID and PID controllers on a non-linear system under different conditions. The ML-SIT1-FPID controller demonstrated more robust performance compared to the other controllers.

Anahtar Kelimeler

• Machine Learning
• Type-1 Fuzzy Controller
• SIT1-PID
• PID
• Non-Linear System

Teşekkür

The authors state that ChatGPT (GPT-5, OpenAI) and DeepL tools were utilized during the preparation of this article to assist with language refinement, the enhancement of visual quality, and text editing. These tools were not used for data analysis, interpretation, or drawing scientific conclusions; all conceptual and methodological contributions are solely attributable to the authors.

Makine Öğrenimi Tabanlı Çevrimiçi Ayarlama Tek Girişli Tip-1 Bulanık PID Kontrolör Yaklaşımı

Öz

Bu çalışmada, makine öğrenimi tabanlı tek girişli tip-1 bulanık PID (ML-SIT1-FPID) kontrolör tasarımı geliştirilmiştir. Tek girişli tip-1 bulanık PID kontrolör (SIT1-FPID), tek girişli tek çıkışlı tip-1 bulanık kontrolör (T1-FC) ve PID kontrolöründen oluşur. T1-FC, geleneksel PID kontrolörünün doğrusal olmayan karakteristiği gibi davranır. SIT1-FPID kontrolörünün tasarımında, giriş değişkeni için beş üçgen tip-1 üyelik fonksiyonu kullanılırken, çıkış değişkeni için beş tekil üyelik fonksiyonu kullanılır. ML-SIT1-FPID kontrolörünün yapısı, SIT1-FPID kontrolörünün yapısı ile aynıdır. Aralarındaki temel fark, çıkış üyelik fonksiyonundan kaynaklanmaktadır. SIT1-FPID kontrolörü kullanılarak çeşitli koşullar altında doğrusal olmayan bir sistem üzerinde simülasyon çalışmaları yapılmış ve buna göre veriler toplanmıştır. Toplanan verilere dayalı bir makine öğrenimi yaklaşımı kullanılarak hataya bağlı dinamik bir denklem türetilmiştir. ML-SIT1-FPID kontrolörünün çıkış üyelik fonksiyonları, hataya dayalı dinamik denklem kullanılarak çevrimiçi ayarlama ile ayarlanmıştır. Önerilen kontrolörün performansı, farklı koşullar altında doğrusal olmayan bir sistemde SIT1-FPID ve PID kontrolörleri ile karşılaştırılmıştır. ML-SIT1-FPID kontrolörü, diğer kontrolörlere kıyasla daha sağlam bir performans sergilemiştir.

Anahtar Kelimeler

• Machine Learning
• Type-1 Fuzzy Controller
• SIT1-PID
• PID
• Non-Linear System
• Robustness.

Teşekkür

Yazarlar, bu makalenin hazırlanması sürecinde dilin iyileştirilmesi, görsel kalitesinin artırılması ve metin üzerinde düzenlemeler yapılmasına yardımcı olması amacıyla ChatGPT (GPT-5, OpenAI), Deepl araçlarından yararlanıldığını belirtmektedir. Söz konusu araçlar veri analizi, yorumlama veya bilimsel sonuçlara varma amacıyla kullanılmamış olup, tüm kavramsal ve yöntemsel katkılar yalnızca yazarlara aittir.

Kaynakça

1. [1] L. A. Zadeh., “Fuzzy sets, Information and Control”, 1965; vol. 8: no. 3: pp. 338 – 353, DOI:https://doi.org/10.1016/S0019-9958(65)90241-X
2. [2] L. A. Zadeh, “The concept of a linguistic variable and its application to approximate reasoning-I”, Information Sciences, 1975; vol. 8: no. 3: pp. 199– 249 DOI: https://doi.org/10.1016/0020-0255(75)90036-5
3. [3] Şahin İ., and Kumbasar T., “Catch me if you can: A pursuit-evasion game with intelligent agents in the Unity 3D game environment”, International Conference on Electrical Engineering (ICEE) Istanbul, Turkey. 2020; pp. 1-6, DOI: https://doi.org/10.1109/ICEE49691.2020.9249828
4. [4] Candan F., Beke A., and Kumbasar T., “Design and Deployment of Fuzzy PID Controllers to the nano quadcopter Crazyflie 2.0.” 2018 Innovations in Intelligent Systems and Applications (INISTA), Thessaloniki, Greece. 2018; pp. 1-6, DOI: https://doi.org/10.1109/INISTA.2018.8466284
5. [5] Kaplan M. R., Eraslan A., Beke A. and Kumbasar T., “Altitude and Position Control of Parrot Mambo Minidrone with PID and Fuzzy PID Controllers”, 2019 11th International Conference on Electrical and Electronics Engineering (ELECO), Bursa, Turkey. 2019; pp. 785-789. DOI: https://doi.org/10.23919/ELECO47770.2019.8990445
6. [6] Meng F. and Qi J., “Application of SSA-Fuzzy-PID in Liquid Level Control Systems”, 2023 IEEE 11th Joint International Information Technology and Artificial Intelligence Conference (ITAIC), Chongqing, China, 2023; pp. 667-671, DOI: https://doi.org/10.1109/ITAIC58329.2023.10409068
7. [7] Unni A. C., Junghare A. S., Mohan V. and Ongsakul W. , “PID, fuzzy and LQR controllers for magnetic levitation system”, 2016 International Conference on Cogeneration, Small Power Plants and District Energy (ICUE), Bangkok, Thailand, 2016; pp. 1-5 DOI: https://doi.org/10.1109/COGEN.2016.7728977
8. [8] García-Martínez, J. R., Cruz-Miguel, E. E., Carrillo-Serrano, R. V., Mendoza-Mondragón, F., Toledano-Ayala, M., and Rodríguez-Reséndiz, J. A, “PID-Type Fuzzy Logic Controller-Based Approach for Motion Control Applications”, Sensors. 2020; 20(18), 5323, DOI:https://doi.org/10.3390/s20185323

9. [9] Ahmed, T., “Analysis and design of a fuzzy controller and performance comparison between the PID controller and fuzzy controller”, International Journal of Scientific & Technology Research.2020; 9(10), 271–275.
10. [10] Chao, C.-T., Sutarna, N., Chiou, J.-S., and Wang, C.-J. “An optimal fuzzy PID controller design based on conventional PID control and nonlinear factors”, Applied Sciences. 2019; 9(6), 1224. https://doi.org/10.3390/app9061224
11. [11] Reznik, L., Ghanayem, O., and Bourmistrov, A., “PID plus fuzzy controller structures as a design base for industrial applications.” Engineering Applications of Artificial Intelligence.2000; 13(4), 419–430, DOI:https://doi.org/10.1016/S0952-1976(00)00013-0
12. [12] Nguyen, V.-K., Tran, V.-K., Pham, H., Nguyen, H.-D., and Nguyen, C.-N. “Design and implementation of fuzzy-based fine-tuning PID controller for programmable logic controller”, International Journal of Integrated Engineering.2024; 16(5), 359–372, DOI: https://doi.org/10.30880/ijie.2024.16.05.027
13. [13] Sahin I., and Ulu C., “Control of Inverted Pendulum System with Single Input Interval Type-2 Fuzzy PD Controllers”, Innovations in Intelligent Systems and Applications Conference (ASYU), Sivas, Türkiye, 2023; pp. 1-5.DOI: https://doi.org/10.1109/ASYU58738.2023.10296815
14. [14] Sahin I., and Ulu C., “Single Input General Type-2 Fuzzy PD Controller for Ball and Beam System”, Innovations in Intelligent Systems and Applications Conference (ASYU), Ankara, Türkiye. 2024; pp.1-5. DOI: https://doi.org/10.1109/ASYU62119.2024.10757128
15. [15] Sahin I., and Ulu C., “Single Input Interval Type-3 Fuzzy PID Controller Approach”, 2024 8th International Artificial Intelligence and Data Processing Symposium (IDAP), Malatya, Türkiye. 2024; pp. 1-6. DOI: https://doi.org/10.1109/IDAP64064.2024.10711013
16. [16] M. A. Alkan, “What is Machine Learning?” Access address: https://www.siskon.com.tr/dosya/PDF/Makale/Makine_Ogrenimi_Machine_Learning.pdf
17. [17] Aksoy, G., “Machine Learning Part 1: Basic Concepts.” Medium. 2023; Accessible address: https://medium.com/@grkemaksoy/makine-%C3%B6%C4%9Frenmesi-b%C3%B6l%C3%BCm-1-temel-kavramlar-78cae4d1d608
18. [18] Karakuş, C., “Introduction to Machine Learning”, Access address: https://ckk.com.tr/Kitap/M%C3%BChendislik/ML%2000%20Makine%20%C3%96%C4%9Frenmesi%20Giri%C5%9F.pdf
19. [19] Liu Q., Li P., Huang G, Zhang X., Liu S., and Yang Z. “Algorithm Research Based on an Elliptical Arc Fitting Curve”, IEEE ACCESS.2024; vol. 12, 113113- 113125, DOI: https://doi.org/10.1109/ACCESS.2024.3416465
20. [20] Yaohua J., “Pedestrian segmentation and shadow removal based on Bezier curve fitting for video image”, 2023 IEEE 3rd International Conference on Electronic Technology, Communication and Information (ICETCI).2023; May 26-28,54-57, DOI: https://doi.org/10.1109/ICETCI57876.2023.10176590
21. [21] Chen Z., Jin S., Shen Y., Wang M., “Minimum Loss Control Strategy of SPMSM Based on Curve Fitting”, Proceedings of the 2024 8th CAA International Conference on Vehicular Control and Intelligence (CVCI), Chongqing, China. 2024; 25-27, DOI: https://doi.org/10.1109/CVCI63518.2024.10830175
22. [22] Rauf M., “Machine Learning Research”, Karabük University, TOBB Vocational School, Computer Programming Department, Karabük. Accessible adress: https://www.scribd.com/document/805521491/MakaleMakinerenimiAratrmas-TOBBMYO
23. [23] Erbay Mermer Ş., “Digital Transformation in Education: Machine Learning”, Bilecik Şeyh Edebali University, DOI: https://doi.org/10.58830/ozgur.pub303.c1379
24. [24] Sahin I., and Ulu C., “Altitude control of a quadcopter using interval type-2 fuzzy controller with dynamic footprint of uncertainty”, ISA Trans. 2023;134, 86–94, DOI: https://doi.org/10.1016/j.isatra.2022.08.020