Open AccessPARAMETER CLASSIFICATION SOFTWARE BASED ON CHARACTERIZERS AND KNOWLEDGE BASE FOR ELECTRONIC ENGINE CONTROL UNIT
Author(s) -
Svitlana Kuznichenko,
Т. М. Терещенко,
Iryna Buchynska,
Viktoriia Klepatska
Publication year - 2021
Publication title -
kìberbezpeka. osvìta, nauka, tehnìka
Language(s) - English
Resource type - Journals
ISSN - 2663-4023
DOI - 10.28925/2663-4023.2021.11.110123
Subject(s) - computer science , software , electronic control unit , calibration , data mining , process (computing) , control unit , artificial intelligence , pattern recognition (psychology) , machine learning , mathematics , programming language , engineering , operating system , statistics , aerospace engineering
The article discusses the issues of increasing the efficiency of the classification process of cards of electronic control units of a car engine. The analysis of the existing software for editing calibration tables in electronic engine control unit, which has tools for determining calibrations and data recognition, was carried out. The limits of use of such software products are conditioned by a small number of specified classes of calibration tables and low data processing speed. The analysis of testing results of classification methods using spectral decomposition demonstrated that a system based on this method requires complex transformations of the results of spectral decomposition. The use of spectral decomposition as a solution of the classification problem is possible if some characteristics of the input data are determined and used as data for classification. It was developed a data classification algorithm that uses characterizers to compute a clearly identified characteristic of the input matrix. The software package for the implementation of the developed algorithm was carried out by using the .NET Framework and the C # programming language. The testing of the classification system performance performed by using the developed software system on a small sample of maps. The results of preliminary testing showed that the system determines correctly the class of the provided card after training. Further testing on the Mercedes-Benz Bosch EDC16C31 / EDC16CP31 car block family showed that in cases of a large number of training images, the result meets the requirements. The performed tests allowed us to determine the optimal number of images for training and the time required for this.