Motor vehicle inspection and error associations prediction Motorlu araç muayene ve hata birliktelikleri tahmini


Creative Commons License

Yılmaz G. Ç., TANRIÖVER Ö. Ö.

Journal of the Faculty of Engineering and Architecture of Gazi University, cilt.40, sa.1, ss.455-465, 2024 (SCI-Expanded) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 40 Sayı: 1
  • Basım Tarihi: 2024
  • Doi Numarası: 10.17341/gazimmfd.1036562
  • Dergi Adı: Journal of the Faculty of Engineering and Architecture of Gazi University
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Art Source, Compendex, TR DİZİN (ULAKBİM)
  • Sayfa Sayıları: ss.455-465
  • Anahtar Kelimeler: association analysis, classification, machine learning, Vehicle inspection
  • Ankara Üniversitesi Adresli: Evet

Özet

Vehicle inspection is a system in which the technical qualifications of motorized or non-motorized vehicles in traffic are measured and whether passenger and traffic safety is provided or not. In Turkey, periodic inspections of approximately 6 million vehicles are carried out every year in order to carry out technical inspections of vehicles traveling on the road more effectively and healthily and to ensure road traffic safety. In this research study, vehicle inspection result prediction and defect association analysis were performed by using vehicle inspection data, machine learning and deep neural networks. With the apriori algorithm, which is one of the association rules extraction methods, the analysis of the defects seen as a result of the inspection of the vehicles was carried out and significant relationships were found between the vehicle defects. In addition, from machine learning prediction methods, Logistic Regression (LR), Naive Bayes (NB), Decision Trees (DT), Random Forest (RF), K-Nearest Neighbor (KNN), Gradient Boost (XGBoost), AdaBoost, Deep Neural Network (DNN) and Convolutional Neural Network (CNN) were used and each model was compared in terms of AUC, ROC curve, accuracy, precision, recall and F1 score values. LR %73, 69, KNN %70, 47, NB %73, 01, DT %69, 53, RF %70, 94, XGB %79, 73, ADA %78, 46, DNN %79, 66 and CNN %80 of the estimation of the examination result classified as slightly defective, severely defective and unsafe by machine learning methods. It has been concluded that with an accuracy of 80, certain defects can be seen together with the highest %90.