Multi-agent systems in industrial fault diagnosis

Industrial diagnosis systems aim to anticipate the occurrence of failures or, if failures have occurred, to detect them and identify their cause, based on observable symptoms captured by sensors from the process. In case of large and complex industrial systems, characterized by dynamic situations and extensive data, such diagnosis system must satisfy requirements for modularity, flexibility and adaptability. Thus, it is difficult to tackle the problem through the strength and capability of a single intelligent entity. This paper introduces a multi-agent cooperative architecture customized for computer-supported fault diagnosis, as a solution to better solve the variety of problems in modern, real-time applications.