Knowledge-Based System Identification - Components of the Expertise

Abstract The system identification problem is analysed from the perspective of the required (expert) knowledge. Recent years witnessed efforts to introduce artificial intelligence techniques to this field. Considering the problem on the whole, however, the developed expert systems covered important but limited issues, with similarly limited expert knowledge. The software system supporting various identification activities can aim at different levels of automation, related to the increased volume and complexity of knowledge. System identification problem turns out upon the analysis to be a knowledge-intensive, hard AI problem, where many essential developments are still to come. Accordingly an expert system conforming with a number of the introduced ideas is presented.