Potential Application of Neural Networks to Verification and Validation of Complex Systems

Abstract

In parallel with technological developments, large scale, complex and integrated human-machine systems are in operation to meet the increasing demands in various areas such as energy, traffic, production, and so on. Concerning such operating systems, high reliability in operation is required as the consequences might be very severe otherwise. Reliable operation concerns human operator behavior, as well as the system’s reliability based on individual system components’ reliabilities. Due to some serious accidents in nuclear industry in the last decades, the reliability and risk assessment of such complex and integrated human-machine systems has gained more importance than ever. The nuclear industry is only an example that human error plays the most essential role causing severe consequences, and there are similar errors in other areas like aviation, for example. Therefore, in recent years there has been a growing concern over utilization of the computer and information processing technology in online applications introducing advanced surveillance, early fault detection, and diagnosis techniques into human-machine interfaces present in the complex systems. The important consideration in such designs is the high degree of automation on one hand, and the proper integration of such an automated system with other existing human-machine interface systems forming more reliable operation, on the other. Specifically, the integration can be designed with various mode of validation such as consistency checking, redundancy, and so that reliability in system operation is improved. The purpose of this paper is to present conceptual methodology for the verification and validation of complex and integrated human-machine systems, and accordingly introduce related potential applications of an artificial intelligence (AI) based information processing paradigm which is known as artificial neural networks.