Forecasting Crack Formation Using Artificial Neural Network and Internet of Things

This project proposes a pipeline monitoring system that uses a time based Artificial Neural Network method i.e., long short-term memory (LSTM) to predict the pressure measurements and to send an alert mail to the respective higher authorities, so as to take necessary steps in order to prevent a catastrophic situation. The network is trained and tested using the sensor data that is obtained using the experimental setup of a pipeline with and without cracks. The LSTM is trained using the data from the pressure sensor which was collected under normal working conditions of the system. Adding to this, the system is automated using IoT. The platform for IoT is the ThingSpeak. It is to this cloud that we connect our sensor, the ANN system and the system of the higher authority. The data is exchanged and collected here using NodeMCU as the Wi-Fi module. Finally, when trouble arises the IoT sends an alert alarm and mail to the higher authorities.