Akademik Veri Yönetim Sistemi
PROCEEDINGS OF THE INSTITUTION OF MECHANICAL ENGINEERS PART D-JOURNAL OF AUTOMOBILE ENGINEERING, 2025 (SCI-Expanded)
Road irregularities and bumps not only pose a threat to vehicle safety but can also lead to material damages. There are many studies on the acceleration and vibration analysis on the vehicle in the literature, but there are no studies on a vehicle speed classification system with neural network based on the acceleration signals. In this study, an accelerometer (ADXL345) was placed at two different points on the vehicle were used to measure and record the acceleration conditions during passage over the bumps that seven different sizes while the vehicle was traveling at five different speed levels (10, 20, 30, 40, and 50 km/h). A test road was created by placing speed bumps of different sizes at 5-m intervals on a total of 135 m of road. Resulting in the creation of a dataset consisting of 3.150 data points collected from a total of 35 test drives. The recorded acceleration data was used to establish the acceleration-velocity relationship and predict vehicle speed based on acceleration information, enabling the regulation of speed. The multilayer perceptron feed-forward neural network (MLPFFNN) was developed as the specific ANN algorithm to estimate vehicle speed from acceleration information. The proposed model was trained and tested using an 80%-20% split of training and testing data, with 2.250 data points used for training and 900 data points for testing. A total of 35 test drives were conducted, and measurements were recorded from two points on the vehicle. Vehicle speed was predicted from the accelerations on the vehicle with an accuracy of 93.74%, and vehicle speeds were successfully classified. A confusion matrix was established to assess the efficacy of the classifier utilized in the investigation. In this study, a system has been developed to predict vehicle speed based on the acceleration conditions that will occur on the vehicle. It is believed that this study will serve future studies on autonomous intervention to vehicle speed according to acceleration conditions.