This paper contains research results of the diagnostics of induction motor bearings based on measurement of the supply current with usage of artificial neural networks. Bearing failure amount is greater than 40% of all engine failures, which makes their damage-free operation crucial. Tests were performed on motors with intentionally made bearings defects. Chapter 2 introduces the concept of artificial neural networks. It presents the general structure of a multilayer neural network (Fig.1) and the model of a single neuron (Fig. 2) which explains how to create an output signal (1,2,3). As learning method for created network back-propagation algorithm was chosen. It uses equation (4) for calculating the errors in the k-th layer. As the model data for network learning, DREAM vibration diagnostics system results were used. Chapter 3 describes how the network input data were created. The essence of the algorithm is to choose the right set of weights for each rotor speed. This is innovative solution of this diagnostic problem. Results of this study are shown in Table 1. Equations (6) - (12) describe how each error was counted. Method shown in this paper, after development, can be useful for the industry.
- Aktywność konferencyjna
- publikacja w wydawnictwie zbiorowym recenzowanym (także w materiałach konferencyjnych)
- Rok wydania:
- Opis bibliograficzny:
- Ciszewski T., Swędrowski L.: Supply current signal and artificial neural networks in the induction motor bearings diagnostics// The Tenth Conference on Condition Monitoring and Machinery Failure Prevention Technologies, 17-20 June 2013/ : Coxmoor Publishing Co., 2013,
- Politechnika Gdańska
wyświetlono 10 razy