A structured methodology for intelligent system selection in nuclear power industry applications

As the number of successful implementations of intelligent systems (ISs) to perform nonsafety related operations increases, so the desire has grown to assess the possibility for use of ISs to support safety related functions. This interest has been recognised by the nuclear industry who wish to determine if ISs can be successfully and reliably used for performing condition monitoring of critical data. Therefore, in order to test the applicability of ISs in the nuclear domain, it is necessary to create a structured development methodology that explains why and how development decisions have been made. At the top level of this IS development methodology, it has consequently been necessary to develop a generic IS selection methodology which can be applied to any relevant problem. This paper has presented a methodology, to aid a decision maker in choosing an appropriate intelligent system as the solution to a given problem. This methodology provides a formal framework for justification of the decisions made and provides the facility for traceability and subsequent justification of these decisions. This is therefore useful for nuclear power software applications where it is desirable to make well-planned and justifiable decisions when the IS is to be used in safety-related applications.