Development of an Aircraft Engine Health Monitoring System Using DBNN (Deep Belief Neural Network)

The fact that aviation could remain reliable, remain efficient and cost-saving is closely related to aircraft engine reliability yet the outdated system of relying on a particular set of thresholds and choosing a time to conduct a check-up is no longer applicable. There is the likelihood of the new aircrafts crashing in a mess and non-detection by simple rules which would have identified the aircraft as the right one. This was what led us to the creation of a brilliant health monitoring system, the system was developed on the basis of a deep belief neural network. It is fundamentally structured to uncover the faults at an early stage and will do the maintenance requirement prediction in a more superior manner. The network was trained to retrieve the crucial information of all types of sensors placed in the aircraft such as temperature, pressure, vibration and speed and all the others. The network has been constructed utilizing the unsupervised learning utilizing the restricted Boltzmann machines, the network has been refined utilizing the fine tuning of the model utilizing the restricted Boltzmann machines. The network has been in a position to detect the wear on the engine and the abnormal patterns in the data and has also been able to work with the noisy data. The network can forecast the sickness and engine issues before it is too late and it notifies the engine issues that are peculiar and peculiar patterns. In this manner, it can save a significant amount of time, resources, maintenance costs and save a significant amount of money. Here we also got the improvement in the performance of the neural network over the traditional machine learning models and through this we are able to see the performance improvement.