# ANN-Based Modeling of Engine Performance from Dynamometer Sensor Data

## TL;DR

This paper introduces an artificial neural network model that accurately predicts engine performance using sensor data from a dynamometer and hardware-in-the-loop setup.

## Contribution

The novelty lies in the high-accuracy ANN model for engine load prediction using real sensor data and its robust validation through cross-validation.

## Key findings

- The ANN model achieved 99% accuracy in multiclass classification of engine load.
- Regression performance showed an R2 of approximately 0.98, indicating strong predictive capability.
- Validation results confirmed robustness with an average accuracy of 0.988±0.011 and macro-F1 of 0.984±0.011.

## Abstract

Accurate prediction of the performance of an internal combustion engine is an essential step towards achieving efficiency and complying with emission standards. This study presents an artificial neural network (ANN) model that uses sensor-derived parameters, such as design power, wheel power, torque, and rotational speed, to predict engine load. Data were collected from a dynamometer and a hardware-in-the-loop (HiL) setup to ensure realistic, sensor-based measurements. The proposed ANN architecture achieved high accuracy (99%) in multiclass classification and strong regression performance (R2≈0.98), demonstrating its ability to model complex engine load relationships under normal operating conditions. Performance was validated using 5-fold stratified cross-validation, achieving an average accuracy of 0.988±0.011, macro-F1 of 0.984±0.011, and regression R2 of 0.962±0.052, confirming strong generalization and robustness. The model can be extended to include additional sensor inputs and adapted for use with other powertrain systems, allowing it to be used in a range of automotive and industrial applications.