Measurement compensation method for an industrial asphalt penetration index detector

Bitumen penetration index is a crucial performance metric for the asphalt, which is defined at the standard 25°C. The industrial detecting machines are, however, often operated with temperature errors, which undermine the measurement accuracy. This paper proposes a compensation method for the measurement of asphalt penetration index considering the operating temperature errors. The relationship between temperature and bitumen penetration measurement error is investigated, within the range of 25±2°C. For this purpose, the polynomial fitting, least squares support vector machines (LSSVM) and BP neural networks are incorporated, and to predict the bitumen penetration errors. The results show that when there are only a few experimental samples, the LSSVM model performs best, with approximately 2% accuracy improvement over the traditional polynomial fitting; while the BP neural network model works best when there are sufficient experimental samples. In this case, the BP neural network model improves the accuracy by 2% compared with the LSSVM model. Overall, this paper provides an in-depth analysis of the impact of temperature on bitumen penetration and proposes effective methods for predicting bitumen penetration errors, which are useful for industrial detecting machines.