These can measure almost anything, from temperature, to the number of times a particular function has been performed, to levels of consumables, to key performance metrics – and that data can then be used to inform more proactive and intelligent approaches to maintenance than ever before. In turn, organisations are fixing small problems before they escalate, planning their maintenance at the most efficient times, and extending the useful life of their most complex and costly physical assets.
However, there is more than one approach to this data-driven maintenance. In this blog we’re taking a look at the differences between two key types – predictive and prescriptive.
Both forms of maintenance are proactive rather than reactive. That is, rather than carrying out maintenance only after a fault has been discovered, they focus on anticipating potential faults and failures, and carrying out preventative work accordingly.
What is predictive maintenance?
Predictive maintenance is becoming increasingly commonplace in industrial and manufacturing settings. It focuses on determining when physical assets actually require maintenance – perhaps because performance is beginning to slightly deteriorate, or because a certain number of tasks have been carried out. As outlined above, this is achieved through IoT-enabled devices which work in real-time, giving an up-to-date, centralised picture of the machine in question’s condition.
Predictive maintenance, then, relies on monitoring those conditions either continuously or at very regular intervals, and transmitting that information to a centralised analytics platform. In turn, the platform suggests the optimum time for maintenance to be carried out. As mentioned previously, this means that maintenance can be scheduled for the optimum time both for the machine and for wider operations, leading to dramatic
cost savings.
What is prescriptive maintenance?
Prescriptive maintenance takes predictive maintenance one step further by introducing an additional layer of automation. Rather than simply monitoring the machine’s condition and providing recommendations on when and how maintenance should take place – which humans then need to implement – prescriptive maintenance aims to enable the machine itself to make its own decisions regarding maintenance steps.
This, then, requires artificial intelligence (AI) and machine learning (ML) techniques, which continually learn from and improve on maintenance recommendations. For example, a machine might suggest that current productivity is reduced in order to extend the time until the next required maintenance session. The machine learns to adjust its operating conditions for optimal outcomes, and intelligently plans maintenance precisely when it is needed. As the machine gathers more data over time, so its recommendations improve.
As with predictive maintenance, this requires real-time data collection, but it also requires the ability to compare different machines throughout the environment. Because prescriptive maintenance improve by analysing data over time, the more data you can ‘feed’ the system with, the better.
In practice, all this means that prescriptive maintenance offers an even more proactive approach than predictive maintenance. Predictive maintenance predicts when a failure is likely to occur so that you can schedule maintenance ahead of time; prescriptive maintenance aims to help you avoid whole categories of
failure altogether.
As IoT, AI and ML technology continue improving in speed and sophistication, so the opportunities for intelligence, automated and forward-thinking maintenance are improving.