Yapay sinir ağları destekli PID ile kuvvet altında robot kol kontrolü ve performans analizi
Yükleniyor...
Dosyalar
Tarih
2023
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Trakya Üniversitesi Fen Bilimleri Enstitüsü
Erişim Hakkı
info:eu-repo/semantics/openAccess
Özet
Robot kolları günümüzde birçok iş alanında kullanılmaktadır. Endüstride ve kişisel kullanımda robot kol performansları yapılan işin kalitesini etkilemektedir. Bu tez çalışmasında robot kollarının performansını etkileyen kontrol sistemleri üzerine bir araştırma yapılmıştır. Bu doğrultuda geleneksel bir PID denetleme sistemi ile yapay sinir ağları destekli PID denetleme sistemi karşılaştırılmıştır. Bu tez çalışmasında Solidworks çizim programında iki serbestlik derecesine sahip bir robot kol tasarlanmıştır. Tasarlanan robot kolun Matlab/Simulink ortamına aktarımı ile fiziksel modelin elde edilmesi sağlanmıştır. Matematiksel denklemlerle sistemin hareket denklemi oluşturulmuş ve Matlab/Simulink ortamında blok diyagramlar ile modellenmiştir. Matematiksel model ve fiziksel model karşılaştırılıp çıktıların eşleştiği kanıtlanmış yani üzerine çalışılan robot kol sisteminin doğruluğu ortaya konulmuştur. Ardından denetim sistemleri tasarlanmıştır. Tasarlanan denetim sistemleri ilk olarak normal şartlarda daha sonra robot uç noktasına dış kuvvet uygulanarak test edilmiştir. Bu testler ilk olarak referans değerler için ardından düz çizgi yörüngesi ve daire yörüngesi oluşturularak iki farklı durum için yapılmıştır. Çalışmanın sonucunda yapay sinir ağları destekli modern denetim sisteminin geleneksel PID denetleme sistemine göre daha iyi sonuç verdiği ortaya konmuştur. Ayrıca dışarıdan uygulanan kuvvetlere karşı yapay sinir ağları destekli sistemin cevap verme tepkisinin daha iyi olduğu görülmüştür.
Robot arms are used in many business areas today. In industry and personal use, robot arm performances affect the quality of the work done. In this thesis, a research on control systems affecting the performance of robot arms has been carried out. In this direction, a conventional PID control system and a PID control system supported by artificial neural networks are compared. In this thesis, a robot arm with two degrees of freedom is designed in Solidworks drawing program. The physical model was obtained by transferring the designed robot arm to Matlab/Simulink environment. The equation of motion of the system was created with mathematical equations and expressed with block diagrams in Matlab/Simulink environment. The mathematical model and the physical model were compared and it was proved that the outputs matched, that is, the accuracy of the robot arm system studied was demonstrated. Then the control systems were designed. The designed control systems were first tested under normal conditions and then by applying external force to the robot endpoint. These tests were performed first for reference values and then for two different situations by creating a straight line trajectory and a circle trajectory. As a result of the study, it was revealed that the modern control system supported by artificial neural networks gives better results than the conventional PID control system. In addition, it has been observed that the response of the system supported by artificial neural networks against external forces is better.
Robot arms are used in many business areas today. In industry and personal use, robot arm performances affect the quality of the work done. In this thesis, a research on control systems affecting the performance of robot arms has been carried out. In this direction, a conventional PID control system and a PID control system supported by artificial neural networks are compared. In this thesis, a robot arm with two degrees of freedom is designed in Solidworks drawing program. The physical model was obtained by transferring the designed robot arm to Matlab/Simulink environment. The equation of motion of the system was created with mathematical equations and expressed with block diagrams in Matlab/Simulink environment. The mathematical model and the physical model were compared and it was proved that the outputs matched, that is, the accuracy of the robot arm system studied was demonstrated. Then the control systems were designed. The designed control systems were first tested under normal conditions and then by applying external force to the robot endpoint. These tests were performed first for reference values and then for two different situations by creating a straight line trajectory and a circle trajectory. As a result of the study, it was revealed that the modern control system supported by artificial neural networks gives better results than the conventional PID control system. In addition, it has been observed that the response of the system supported by artificial neural networks against external forces is better.
Açıklama
Anahtar Kelimeler
Robot kol, PID, Yapay sinir ağları, Yapay zeka, Robot arm, Artificial neural networks, Artificial intelligence
