Abstract
This paper illustrates a conceptual framework for the development of Monitoring and Control Systems (MCS), based on a four level agent-based architecture. Traditional MCS are designed according to a three-level architectural pattern, in which intelligent devices are usually devoted to evaluate if data acquired by a set of sensors could be interpreted as anomalous or not. Possible mistakes in the evaluation process, due to faulty sensors or external factors, can cause the generation of undesirable false alarms. To solve this problem, our framework introduces a fourth level to the traditional MCS architecture, named correlation level, where an intelligent module, usually a Knowledge-Based System, collects the local interpretations made by each evaluation device building a global view of the monitored field. In this way, possible local mistakes are identified by the comparison with other local interpretations. The framework has been adopted for the development of Automotive MCSs.