ABSTRACT
Abnormality of a group of compressors is monitored by a robot applying an automated diagnosis technique based on acoustic signal processing. Frequently changing ambient noise conditions due to control action of the group are automatically learned and the monitoring system adapts itself to prevent from fault detecting. A practical model of the system has been successfully tested to monitor abnormal conditions of a compressors group in a compressor station of a car making factory. Its operational principle and possible applications are presented.
