No abstract

No abstract

In recent years, Deep Learning (DL) and Internet of Things (IoT) technologies have been used and deployed jointly to solve a wide range of modern technical challenges in different areas. With the continuous advancement of IoT connectivity solutions, the range of applications that can benefit from such an increase is limitless. One area that can benefit significantly from the combined strength of DL and IoT technologies is Machine Health Monitoring (MHM) Systems. MHM utilizes different analytical approaches and tools to determine the state and health of different components in running machinery. The traditional MHM system uses control limits from predetermining values that determine if a component has failed depending on the preset limits of the machinery. The main disadvantage of using such technique us the unpredictable nature of the timing and component failure. This type of failure causes unplanned production time loss and increases the cost of maintenance due to the unpredictability of the failure events. With DL and low-cost sensors that use different IoT connectivity solutions, MHM systems can utilize the learning capabilities of the DL network to perform end-to-end prognosis. One crucial fact is that features learned by Deep Neural Networks (DNN) are part of a large black box, and there are valuable underlying physical meanings embedded within the features. Hence, there is an exciting research area to explore underlying mechanisms and interpret physical meanings within DNN. In this paper, DNN learning mechanisms are evaluated using three different models: stacked autoencoders (SAE), denoising autoencoders (DAE), and convolutional neural networks (CNN). Initial results indicate that the input layer behaves similarly to a band-pass filter.  However, deep layers require optimal input design to maximize neuron activation, which leads to an extensive understanding of deep layer learning consequently (In progress).