Vibration signal prediction model for the miniature transducer using deep learning network

This work presents a deep artificial neural network (DNN) using input/output measurements of the miniature transducer. The DNN consists of a convolution network with one-dimensional convolution layers, pooling layers and full connection layers with ReLU activation function. In the training process, the coefficients of each layer are adapted to minimize the loss function as the least-square function. In the experiment tests, the numerous type of signals used for training and validation tests. As a result, the proposed model can fit the output response of the miniature transducer in high accuracy.