Computer based diagnostic system for the final control of compressors is described. The system is based on a measurement of vibrations with an accelerometer located on the top of the compressor. Signal preprocessing based on psychoacoustic features is applied to estimate the signal power for each of the psychoacoustic filters. Diagnostic strategy is designed as an adaptive one and during the operation constantly adapts to the current population of compressors. With this strategy, compressors that emit unusual vibrations caused by specific faults, can be eliminated.
F.15 Development of a new information system/databases
COBISS.SI-ID: 10944283By the system a circular laser beam distribution is transformed into a ring distribution that is focused onto the wire circumference. By means of this a uniform heating of the wire and preheating of the spot of droplet deposition can be achieved simultaneously by using a single laser beam only. Using the system droplets of diameter from 0.5 to 1.5 mm can be generated from a metal wire. By selective deposition of generated metal droplets various application like joining of temperature sensitive parts, joining of different incompatible materials, 3D structuring, etc, are possible.
F.08 Development and manufacture of a prototype
COBISS.SI-ID: 000000Mechatronic diagnostic system for the monitoring of compressor’s crankshaft rotation in company Danfoss Compressors, d.o.o., was developed and made out. The system operates on the principle of force measuring on the sensor tip that rotates the crankshaft by pneumatic manipulators. Mechatronic diagnostic system was developed and manufactured in the Laboratory of Synergetics at the Faculty of Mechanical Engineering in Ljubljana. During the several months of pilot operation, the system reliably detected crankshaft rotation faults.
F.08 Development and manufacture of a prototype
COBISS.SI-ID: 000000The possibility of system adaptation to novel data without loosing connection to old data, was studied. Suitable adaptability of the system was achieved by a method of weighting forecasting errors, where the model errors on novel data are weighted as more important. Consequently, after the optimization the system is more suitably adapted according to novel conditions. Without loosing the robustness of the system we achieved the improved forecasting accuracy for new data which is important for practical applicability of the system.
F.16 Improvements to an existing information system/databases
COBISS.SI-ID: 000000In this work we have successfully implemented closed-loop control of rubber mixer by using soft computing methods in combination with information theory. Experiments on industrial mixer showed that the process models, which include neural networks in combination with information theory, allow us to efficiently follow the optimal trajectory. We have developed an on-line predictive control system on an industrial rubber mixer of a local rubber producer. Tests in production showed improved compound quality.
F.14 Improvements to existing production methods and tools or processes
COBISS.SI-ID: 7219796