Hybrid expert systems for monitoring and fault diagnosis

The integration of connectionist and symbolic techniques within the one system is considered. The authors discuss the advantages of hybrid connectionist/symbolic knowledge-based systems. Integrating the two approaches can take many forms. A problem could be represented as a mixed connectionist and symbolic knowledge base with modes of reasoning selected as appropriate to the current task, or as one mode of reasoning embedded in a system where the overall flow of control is determined by the other mode. Both of these allow reasoning tasks to be undertaken using the appropriate mode, but also require a knowledge representation accessible to both modes. Two telecommunications problems that were solved using hybrid connectionist and rule-based reasoning are presented. The first application for hybrid reasoning is in the area of performance monitoring and fault diagnosis in telecommunication network management. The second application is a system that predicts and diagnoses faults in a telecommunication cable distribution network. Hybrids can extend the role played by knowledge-based systems by allowing different modes of knowledge representation and reasoning to be used when strict rule-based techniques do not work