Comparison of the Artificial Data Generation Methods for Deep Training of Monitoring System
MAGAZINE №3(86) June 2018
AUTHORS
SOBOLEVSKIY V.A. - Graduate student, Russian Academy of Sciences, Laboratory of Information Technologies in System Analysis and Modeling, St. Petersburg Institute of Informatics and Automation (St. Petersburg, Russia)
CATEGORY Analysis in logistics and SCM Simulation modelling
ABSTRACT
This article deals with the problem of input data generated for the creation and training of an artificial neural network which is the basis of the classification module of a dynamic monitoring system of the manufacture performance indexes. The input data that was used to train the neural network was divided into the following categories: real data, generated data for a given distribution, and data obtained using the simulation approach. The simulation model was created using the apparatus of Petri nets. Further, for the data used in the work, classification rules were set, after which the artificial neural network was trained on each data set. At the next stage, real data was submitted to the monitoring system which are previously did not appear in the training and validation of neural networks. The final step of this study was to compare the results of the classification of the described approaches of artificial generation of values of enterprise input parameters with respect to the control data set.
Keywords:
