Prescriptive Maintenance Machine Learning Leads to More Productivity
Prescriptive Maintenance heralds a paradigm shift: Instead of merely predicting failures, recommendations can be drawn from it. Which conditions must be met? How does its implementation succeed?
With Prescriptive Maintenance, machines or devices take over an active role in their own maintenance. The prerequisite: Machine and sensor data are captured and evaluated in real time. Machine learning is used to identify specific patterns in data sets. In this way, the causes of problems can be identified and countered with precise, timely measures. Algorithms with natural speech processing capabilities take over the evaluation of the captured maintenance data at a speed that computers could not achieve before. With this data, neural networks can analyze millions of potential associations and determine exactly what caused a particular event and why.
Technical Requirements for Maintenance
From a technical point of view, Prescriptive Maintenance requires a combination of different systems and elements. This includes a Data Lake in which data from various sources - such as IoT sensors or automation systems - is consolidated. In addition, there are analysis tools based on machine learning approaches as well as the integration of software systems for close collaboration in the further processing steps. Historical data, such as information about transactions and business processes for analytics, usually originates from existing IT solutions, such as ERP systems. They make it possible to draw conclusions. Finally, a Data Scientist evaluates the simulation.
But how can you determine whether an investment in Prescriptive Maintenance really pays off? And what concrete benefits does it have? The first step is to determine which tasks and activities in the administrative environment can be performed by system solutions. For this purpose, it is necessary to evaluate current workloads and activities of employees and to identify regular, recurring tasks. The aim is to free up as much capacities as possible for smarter tasks. Furthermore, existing machine data has to be used for Prescriptive Maintenance. This includes information on the state of machines and systems such as vibration or telematics - information that is often already available in ERP systems and databases. With Prescriptive Maintenance, these data are no longer only used for simple evaluations, but for productive work processes as well.
Checklists Make Work Orders Transparent
As far as implementation is concerned, a certain procedure has proven to be successful whereby a few systems or machines, such as pumps, are first integrated with maintenance plans, operating and target data and historical data. Further machines can then be added step by step as the process progresses. In this context, checklists developed by Infor especially for Prescriptive Maintenance have made a great contribution: This is a method in which different text modules are merged and then assigned to a work order. While traditional work orders become long and confusing over time, the checklists make it possible to work through individual steps. The basic advantage: Employees become can complete tasks faster. Moreover, the checklists indicate when a task deviates from the tolerance range that must be met to initiate a subsequent work order. From a business perspective, checklists are used to create consistent documentation that transparently prepares related work orders.
Change management is essential for the successful implementation of Prescriptive Maintenance. Although the purpose of Prescriptive Maintenance is to free up manpower, this does not mean that manpower will become superfluous. On the contrary: With Prescriptive Maintenance, employees can be deployed in a more productive way.
* Michael Burkard is Account Manager EAM at Infor in 66299 Friedrichsthal, Tel. (05 11) 93 68 91-00, firstname.lastname@example.org, www.infor.de
This article was first published by MM MaschinenMarkt.
Original by Sebastian Hofmann / Translation by Alexander Stark