An efficient production line and regular preventive maintenance is a key of success for any manufacturing industry as it avoids the costly breakdowns and is principal factor for increase in revenue and net profits. One of the major constraints in maintenance is failing to understand the maintenance cycle of internal moving parts like bearings which are the key parts in any machinery. The maintenance cycle of such parts requires comprehensive technical understanding and experience which may not be available in all type of manufacturing units. Recent developments in AI techniques - Machine learning, Deep learning and Random Forests can help us predict the maintenance cycle by taking all the factors in consideration. We hereby present a detailed comprehensive survey of the work done in predictive modeling for bearing failure. This paper provides a systematic literature review of state of art techniques proposed by various researchers for Predictive Maintenance and determination of Remaining Useful Life (RUL) of a component. This paper presents information of certain Industrial functions and failures respected to maintenance, review of various models based on Ensemble Learning based algorithms, Neural Network based algorithms and some other miscellaneous algorithms also various types of sensors used for condition monitoring, types of datasets etc.