Lokesh Anbalagan

Corrective maintenance occurs after the failure occurred. It is a reactive approach, leading to unscheduled downtime, loss of productivity and higher cost to cover up for the loss of productive hours. Preventive maintenance, relatively common method of checking the health of the system on a regular basis, i.e., once a quarter or once a year, is used for early identification and correction of problems. Though it is a proactive approach, regular maintenance costs time and money; and in many instances no defects identified. If failure happens between the preventive cycle, it leads to productivity loss and requires corrective maintenance. Predictive Maintenance (PdM) on the other hand identifies a problem before it occurs and alerts when maintenance is required. Also called as Condition Based Maintenance, Predictive maintenance is system of monitoring key parameters for smooth operation, identifying early warning signals and prompting for intervention. This reduces the probability of unexpected failure and improves predictability and reliability of the system. Predictive maintenance software uses a large data source to predict potential failure in advance. Continuous monitoring and use of analytics are essential components of predictive maintenance. Various tools like IoT, big data analytics, machine learning, artificial intelligence are used to accurately predict the need for maintenance. This is performed continuously or periodically, depends on the impact of failure. Since this requires a large data collection, regular monitoring, Predictive maintenance is more expensive than other types of maintenance. In addition to higher investment, it also requires a longer duration to assess and implement the solution.