Defects Diagnostics Method Based on Statistical Big Data Analisis of Measured Parameters

Abstract

In this work, the proposed method effectively diagnoses defects by narrowing down suspected units, although it often underestimates the defect's numerical value. Reliability improves with more experimental data, aiding in identifying malfunctions, though false results can occur. Further research is needed on reliability's dependence on measured parameters and discretization degree. Diagnostics should consider different engine modes and note the linear model's error of up to 0.3% against an acceptable variation of +/- 2.5%. Reducing error requires narrowing the variation range, and expanding linearization applicability involves considering the Taylor expansion's second term. These results suggest limitations in other diagnostic methods, as parameter combinations can closely match experimental data.