Abstract: In work is described the monitoring system of a technical condition of the electromechanical equipment with the block of decision-making on the basis of a neural network. As an object of control was used an asynchronous gear drive. Decision making was carried out on the basis of a comprehensive analysis of the vibration data (from the gear train) and the current consumption of the induction motor. As diagnostic parameters are selected the vibration velocity, acceleration and current in the phases of the stator winding of the driving motor. From the selected diagnostic parameters are calculated discrete wavelet transform coefficients (using the Daubechies mother wavelet, 8 levels of decomposition). After that are distinguished diagnostic features: rms and peak (maximum) values of each of the wavelet coefficients and the entire signal (total level) for each diagnostic parameter. The resulting set of wavelet coefficients is fed to the input of the decision block. The paper presents the development of the architecture and software of the decision block based on the neural network, as well as its training and testing. The neural network is a 3-layer perceptron with a nonlinear activation function and a learning algorithm based on backpropagation of error. Each neuron of the output layer corresponds to a certain technical state of the monitored object and indicates the probability of this state. The paper shows the possibility of increasing the efficiency of monitoring the technical condition of electromechanical equipment through the use of complex analysis using an intelligent decision block.
Index terms: non-destructive testing, neural network, diagnostics, electric drive.