| Abstract [eng] |
The aim of the project is to create a universal measurement system for the main parameters of the electrical equivalent diagram of the electric machine, which allows real-time monitoring of the ongoing experiment remotely, sharing experiment data and results with interested parties, regardless of their geographic location, to detect machine torque correlation with stator voltage using artificial neural networks, Combine different data gathering devices that communicate on different protocols using the Siemens cloud MindSphere and MindConnect Nano device, which collects all data from the “Schneider Electric” analyzer PM 8000 and gateway Com'x 510. In this work, Online monitoring, Internet of things, Cloud Computing, are reviewed. Also, a review of electric motors faults and the causes of failure and their types. During the experiment first, permanent magnet synchronous machine was used as a generator and the nominal voltage which was 302V average between phases was measured. Second, using that voltage in permanent magnet synchronous machine as a motor and there were seven experiments done at nominal voltages also lower and higher side with different (20V) voltage steps. From that experiment there were calculations of efficiency done with the 92.08% — maximum among the all experiment, while using nominal 302V, maximum torque of 6.8 Nm was also achieved before loosing synchronism. During the experiment data using two measuring devices data was gathered by Schneider (Com‘X) and lorenz platform. After that, data were acquired from the cloud platform which accessed by siemens cloud gateway device. Using this data and artificial neural network with Levenberg Marquart (trainlm) training function, mean squared error of 0.0482 was achieved. After training the neural network torque meter was successfully simulated with 99.1% accuracy. |
