Akademik Veri Yönetim Sistemi
Measurement: Sensors, cilt.27, 2023 (Scopus)
This paper focuses on designing and implementing a system that would increasingly contribute towards the enhancement of society through safe driving. It includes providing alerts to the vehicle drivers beforehand regarding the bad climate and road conditions in real-time so that the drivers would ensure safety as per the information propagated through the wireless sensor network attached to the mobile application. As a result, by using the newly developed system, there will be more assurance of the safety of drivers and a reduction in the cases of accidents. Apart from this, the structure could be changed into a complete network that would help in predicting road conditions at certain waypoints in the route. In this study we used the Gray wolf optimization method to select the best features and use in the machine learning to detect the highway classes. Total 6 classes are used and tested over proposed dataset from Ankara city that include the icing, wet, dry, damp, salty wet and salty moist. We used the decision tree, boosting ensemble learning classification and support vector machine for the classification. When the feature number was selected as 11, we obtained the best accuracy result for the decision tree and SVM with 99.99% and 99.80% accuracy respectively. With the feature number as 14, the Ensemble and Discriminant obtain the best accuracy with 100% and 94.37% respectively. Only KNN classifier obtain the best accuracy with the 17 feature as 99.81%.