PREDICTIVE MAINTENANCE USING RNN AND LSTM MODELS

The art of farming is the oldest and challenging factor in human life. In this fast phased environment and
with the increase in the destruction of atmosphere and other natural resources, it is very questionable to
acquire quality crops. This paper focuses to predict options which control and track the natural factors
which are involved in the agriculture system. This work focuses on analyzing different features of crop and
initiate predictive maintenance activities for all the sensors associated with that farm land. This activity
facilitates the farmer with sensor failure reduction and helps in effective monitoring of the crop. Different
factors like humidity, soil temperature and the luminosity of the crops are considered for effective
maintenance activity. This work is implemented using the forward and backward propagation algorithms
using certain attributes of dataset. This paper facilitates an effective prediction ecosystem after investigating
the numeric data collected from different sensors attached to plants which are meant for earlier
failure prediction of those devicesdepending on the trained data. Using the forecast model and analyzing
time-series data, LSTM model has obtained good accuracy with almost 97% accuracy.