Machine Learning Decomposition Onset Temperature of Lubricant Additives

The thermal stability of lubricant additives is a fundamental parameter in practical applications, which is determined by the molecular structure. The ability to predict thermal properties, particularly lubricant additives' decomposition onset temperature, is of ultimate importance. We develop the Gaussian process regression model to present the relationship between molecular descriptors and onset temperature of decomposition of lubricant additives. This model is highly stable and accurate, which is promising as a fast, robust, and low-cost tool for estimating various types of lubricant additives' decomposition onset temperature.