Verification and validation of artificial intelligent systems with applications to power systems

Software maintenance is a prominent issue in the practical environment of knowledge-based system applications. In the case of rule-based systems in power systems, both the rules and data have to be modified and updated to the continuously changing power system environment. In order to ensure the accuracy of the expert system, testing of the software is required. Validation and verification are well known expert systems maintenance tasks. Validation is a task to ensure that specifications of the expert system are met. Verification, on the other hand, is to ensure that the expert system is built correctly. Both tasks are important parts in establishing the functionality, reliability and flexibility of any software. Other type of artificial intelligent systems face similar challenges. This paper provides a summary of relation-checking, equivalence class and contextual representation methods that have been applied to the verification and validation of rule-based and logic-based systems. These methods have been motivated by power system applications, but they are generally applicable to similar classes of rule-based and logic-based systems.