The technical approach is the development of a trainable, artificial intelligence based inspection system. The system will be introduced into the assembly line for the purpose of realtime in-line quality control. Output of the system is a fault diagnosis, received by the evaluation of complex acoustic signals. Environmental conditions like aging of test bench mechanics, the influence of spare parts to the classification result, and an optimized sensor concept will be considered.
Moulinex, a French domestic machine manufacturer, and Siemens, a German electric motor producer, anticipate cost savings in a volume of several hundred kECU per year by automatic test procedures, and well defined fault diagnosis. In the case of ZF, a German manufacturer of gearboxes, an objective and reliable test is necessary with a reproductive accuracy of 3 dB compared to present 12.15 dB.
The inspection system will detect and classify minimum 95% of the most frequent faults like missing parts, faulty assembled components, faulty gear mesh noise. These faults must be detected in an incipient and well defined stage. The inspection system must work in real time. A fault classification rate of at least 97% is envisaged.
The consortium consists of three end users (Moulinex, Siemens, and ZF), two SMEs (Automation Service and MEDAV) and three research institutes resp. universities (Austrian Research Institute for Artificial Intelligence, Imperial College of Science, and Université de Technologie de Compiegne).
The Project Team
The project is a joint cooperation of the following partners within the
Brite-Euram III RTD programme of the European Union managed by DG XII:
