Selection of control diagnostic symptoms and technical state features of marine engine conventional injectors using multiple regression analysis

This article presents the results of a study of multivariate application statistics to approximate experimental test results. The influence of explanatory variables on the approximation using multiple regression was assessed using various quantities that affected the test conditions and technical state. Measurements of key injector performance features were made using various geometric and fuel dose measurement instruments. The diagnostic symptoms of the injector nozzle vibration displacement signals in the frequency domain, which are most correlated with relevant properties of the technical state and test conditions, were selected. The relevance of individual state features was tested by best fitting the diagnostic parameter model to the dependencies of the variables, including features, and the control quantities. The most useful symptom was the first component of the envelope amplitudes in the nozzle needle displacement spectra, and the most relevant properties of the injection apparatus's technical state were the injected fuel dose at the exact injection frequency.