Intelligent Analysis and Data Processing of Parts Grinding Quality Using Neuronet Technologies

Abstract

Рассмотрен интеллектуальный анализ и об-работка массивов данных амплитудных и фазовых каналов вихретокового контроля шлифовальной обработки деталей с применением нейронных се-тей структуры многослойного персептрона. Прове-дено моделирование нейронной сети структуры многослойного персептрона для решения задачи автоматического распознавания дефектов на осно-вании массивов данных амплитудных и фазовых каналов вихретокового контроля. Представлена программная реализация информационно-измерительного канала вихретокового контроля.Ключевые слова: интеллектуальный анализ данных, нейронная сеть, распознавание дефектов, персептрон, поверхность качения, подшипник, вих-ретоковый контроль. E.M. Samoylova INTELLIGENT ANALYSIS AND DATA PROCESSING OF PARTS GRINDING QUALITY USING NEURONET TECHNOLOGIESAn intelligent analysis and processing of data files of amplitude and phase channels of eddy current testing in parts grinding with the use of neural networks of the structure of a multilayer perceptron was considered. The modeling of a neural network of the structure of a multilayer perceptron is carried out which is a solution of the problem of large data files processing regarding quality of parts allowing the creation of a more efficient automated system of monitoring and the connection in a single information area of the subsystem of parts control, control and management of a technological system, the decrease of cost price of quality control, the coverage of a larger sampling volume of parts under control and the increase the efficiency of a technological system control in manufacturing high-accuracy bearing parts.A program realization of the informationmeasuring channel of eddy current testing as a part of an expert system of decision-making support in the environment of SCADA-system TRACE MODE 6 is presented. The special-purpose developed software Image is designed for the visualization in an on-line mode, an automated collection, and the analysis and information storage of eddy current testing.