on 28 June 18
Till recently big data analytics was confined to verticals which used to generate big data online and through telecommunication satellites, GPS and other such devices. This is now changing as stream analytics is being employed for predicative maintenance analytics. To put it straight, stream analytics is now making maintenance of the machinery easy and thus finding a place in manufacturing industry as well.
Till recently the maintenance could be divided into two types – corrective maintenance and preventive maintenance. The corrective maintenance, of course, meant when the equipment failed and stopped functioning or showed sign of failing. This indeed meant repairing or changing the equipment after stopping the production which naturally meant the loss in production and wastage of man-hours.
The second type of maintenance the preventive maintenance is of course much better than the corrective maintenance as it is done on the routine basis or on the assumption that a particular equipment needs to be maintained periodically before it actually breaks down or performs below its optimal level. But both these maintenances take place when either the equipment is failing or is about to fail.
This limitation led to the need to explore third type maintenance called condition based maintenance (CBM) or as it is now called predictive maintenance (PdM). This maintenance is based on predictive maintenance analytics based on the data obtained through various sensors, monitoring devices which are based on various parameters that affect the performance of a particular part of the equipment.
Predictive maintenance is far advanced than the other two types of maintenance, as its name suggests predicts the behavior of the equipment using the real-time data to attend to the equipment that would fail in near future. This is better than preventive maintenance as it does away with the routine maintenance of the equipment which is working fine and would not falter in near future; only that equipment whose real time data shows even a slight variation of performance is picked for maintenance.
This is like optimizing the whole process of maintenance by smart decision making based on stream analytics. The data that is being continuously being monitored gives hints of the problem in different sections or in particular piece of equipment. Stream analytics thus gives timely input to the administrator to act smartly.
The predictive maintenance analytics is based on mainly three things. Telemetric enabled sensor data is taken into account to predict the equipment degeneration or prediction is based on the history of its performance or reading the particular pattern or circumstances in which equipment malfunctioned. To achieve this different data mining methods are employed to understand various failure modes.
The Trend analysis is employed to reviews the down trend of data coming from a particular equipment thus giving ample time and scope to act and maintain the problem area. Similarly, Pattern recognition is employed to study the correlation between different events and their impact on equipment failure. The predictive maintenance analytics thus enable the administrator to do what is required, minimizing spare parts cost and system downtime.
This blog is listed under Data & Information Management Community