Title: Development of K-nearest neighbours model for diagnosing vehicle automatic transmission failures

Authors: Anh Tuan Pham; Van Tra Nguyen; Ngoc Tuan Vu; Van Tu Nguyen

Addresses: Department of Automotive Engineering, Institute of Vehicle and Energy Engineering, Le Quy Don Technical University, Hanoi, 11355, Vietnam ' Department of Automotive Engineering, Institute of Vehicle and Energy Engineering, Le Quy Don Technical University, Hanoi, 11355, Vietnam ' Department of Automotive Engineering, Institute of Vehicle and Energy Engineering, Le Quy Don Technical University, Hanoi, 11355, Vietnam ' Department of Automotive Engineering, Institute of Vehicle and Energy Engineering, Le Quy Don Technical University, Hanoi, 11355, Vietnam

Abstract: This study investigates the operational conditions of automobiles in Vietnam and the standard automatic transmission (AT) failures that occur during their operation. Various types of vehicle automatic transmission failures, alongside normal operating conditions, are simulated in Simulation-X. The research involves data pre-processing and exploratory data analysis to identify appropriate models for classification. A comprehensive review of machine learning classification algorithms and hyperparameter tuning uses simulation datasets. The KNN model was trained and evaluated, achieving 92.2% accuracy on the test dataset. Permutation importance was evaluated using the open-source library scikit-learn. Potential improvements of the model classifiers are discussed, and recommendations are provided based on the findings. The results demonstrate that the proposed approach can effectively classify AT failures, supporting the development of software modules for real-time technical state supervision and the design of a test bench for assessing AT reliability.

Keywords: diagnose automatic transmission; failures; acceleration time; vehicle speed; vehicle power; Simulation-X; machine learning; classification; k-nearest neighbours; KNNs; hyperparameter tuning.

DOI: 10.1504/IJPT.2026.152001

International Journal of Powertrains, 2026 Vol.15 No.1, pp.86 - 106

Received: 23 Mar 2025
Accepted: 21 Nov 2025
Published online: 02 Mar 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article