PSI - Issue 59

Victor Aulin et al. / Procedia Structural Integrity 59 (2024) 436–443

439

Victor Aulin et al. / Structural Integrity Procedia 00 (2019) 000 – 000

Determination of indicators characterizing the technical condition of the resource of the determining elements of the engine

Usage of Tytien Mur's criterion(E k criterion) for screening out gross errors

Exclusion of insignificant factors

Selection of the most significant for the formation of a set of controlled parameters

Construction of the mathematical apparatus models of ANN for the solution of the task of the recognition of defects in aggregates

Formation of recognition matrices of the defects combinations

Graphic display of convergence of defect recognition during training and testing of ANN

Construction of graphs of ANN models for defect recognition of the cylinder piston group, the slider-crank linkage, the valvetrain

Evaluation of results and conclusions

Fig. 1. Block diagram of mathematical processing of diagnostic information for building ANN models of resource-determining elements of car power units. To improve the quality of training and the accuracy of defect recognition, they are grouped by common diagnostic features. The grouping of detected defects by resource-determining elements of engines is carried out as follows: – the cylinder-piston group: piston grooves under the compression ring, compression rings, piston skirt, cylinder liners ; – the slider-crank linkage: connecting rod necks of the crankshaft, root necks of the crankshaft, posterior root neck, upper head of the connecting rod; – the valvetrain: valve stem, camshaft cams, camshaft gears, valve seats, valve stem guide bushings, camshaft supports. We determined the distributions between data sets for training and testing of the main resource elements of the engine: the cylinder-piston group (CPG), the slider-crank linkage (SCL), the valvetrain(V). Training datasets of R CPG , R SCL , RV were formed by random combinations of basic sets , which were composed on the basis of the relationship of controlled parameters. The test sets were formed on the basis of the obtained observations (in real production conditions at the stage of diagnosing YMZ-238 engines) of the values of these parameters. For the analyzed resource-determining elements of the engine, the percentage value of the test set is more than 50% of the total number of samples, which is more than