Software system for regression analysis and visualization of multidimensional data from bench tests of internal combustion engines

Abstract

A Python-based software module is presented that implements complex processing of multidimensional experimental data typical for testing technical systems (using the example of internal combustion piston engines with ceramic coatings). The module includes algorithms for two-stage hierarchical averaging, automatic determination of the dimensionality of physical quantities, paired t-test, calculation of the reproducibility index, as well as second-order polynomial regression and Gaussian regression with an RBF kernel for constructing response surfaces and interval forecasts. Implemented visualization tools (3D surfaces, contour plots, trends, and correlation diagrams) and automatic generation of output reports. The module's architecture, basic algorithms, and verification scenarios are described. The results of applying the module to real data are presented, confirming its correctness and effectiveness. The software module is registered with Rospatent (certificate No. 2026661664). doi: 10.54708/19926502_2026_30311370