Determining the Main Characteristics of Motor Lubricants Using Color Indices and Prediction Models

Abstract

This paper proposes tools for predicting the main characteristics of motor lubricating oils that can be used to assess their quality. A comparative analysis of four measuring devices, operating on optical principle, was performed to predict the main characteristics of motor lubricating oils. It has been found that the combined use of seven color indices leads to increased accuracy in predicting the value characteristics of motor oils. The two main components are needed to reduce the data volume of the vector containing these features, which leads to a reduction in the volume of data used. It was found that according to colorimeter data, regression models have higher predictive properties than those obtained with a video camera and a digital camera. The proposed research tools can be used with single-board microcomputer systems, which do not require complex computational procedures. The use of the proposed in this article techniques and tools in practice would reduce the effect of the measuring instrument experience.