Prevention of cyber attacks in smart manufacturing applying modern neural network methods

Abstract

Digital transformation is a driver of a modern approach to providing actual cyber security. In the context of the globalization of the economy, digital technologies are actively being introduced into production, factories and plants. Due to the increased mobility of topology and the growing amount of data undergoing the processing, traditional methods of protection becomes ineffective, so the researchers are faced with the task of creating new methods for ensuring cyber security that meet new challenges. The paper analyses new artificial neural network (ANN) architectures corresponding security tasks in cyber physical systems, identifies their major advantages, and evaluates the possibility of their application for solving the problem of attacks prevention in case of machine-to-machine smart manufacturing. A meta-neural system of comprehensive protection against cyber attacks on dynamic routing in intelligent production has been developed. Using the resources of a supercomputer, an experimental study of the developed neuroframework was carried out. Test results showed that the proposed solution provides high accuracy in detecting cyber attacks.