| Abstract [eng] |
Aim of this research – to create an autonomous water network and motor monitoring system, which can identify leaks in water lines and broken motor bearings, using artificial neural networks. In this work, water networks, water extraction and their types are reviewed. Also, a review of electric motor bearings, their causes of failure, types and features is done. During the experiment, a model that can detect water leaks in the network, using the data, imitating normal and emergency modes was made. This model, can detect the leak in the system with 5,4×10-12 mean squared error. After creating specific failures in bearing construction, vibration patterns were measured. Using this data and artificial neural networks, a model that can separate bearings in good condition from broken ones with 2,41 × 10-10 mean squared error was made. Using pattern recognition neural networks and data, obtained from previous experiment a model which can identify the broken part of bearing with 78,5 % accuracy was created. This model had a mean squared error of 0,0935. |
