What is Predictive Maintenance? Fixing problems before they occur
For the reliable operation and maintenance of technical equipment, machines and elements, regular spot checks and inspections were carried out in analogue times. Replacement or repair in case of unforeseen downtime. In the age of data analysis, machines automatically report when control power decreases or when they need maintenance. Thanks to this practical foresight, problems can be solved before they even occur.
Why Predictive Maintenance?
Today, industry and IT are so closely interlinked that we are talking about the Industrial Internet of Things (IIoT). An essential component of this “fourth industrial revolution” is intelligent production management based on data storage and analysis. Thus the condition of a technical device or component can be transparently displayed and analyzed on the basis of data. Even the slightest deviations from the control data can be detected as signs of an error and addressed at an early stage.
How does Predictive Maintenance work?
Predictive Maintenance uses measurement and production data from machines and systems to increase the availability of these critical infrastructures and to minimize downtimes. In the long term, the aim is to extend the service life of a system and thus sustainably reduce costs for the operator.
In the first step, it is essential to capture and store large amounts of data. The more comprehensive and specific certain information about a machine and its influencing environmental factors is, the more concretely it can be analysed and measured. With the connection of sensors to server systems and cloud infrastructures, the IIoT provides the basis for later analysis and calculation of probabilities of occurrence.
Analyzing data, recognizing, and reacting to deviations
As a rule, modern industrial machines, airplanes or wind turbines generate huge amounts of data material. The machines equipped with sensors forward large amounts of data – for example on temperature, humidity or fuel consumption – to a central data center or cloud server, where they are stored in the form of a protocol. The analysis and subsequent evaluation of the information is then carried out using defined measurement variables or key performance indicators (KPI).
Measured variables must be clearly defined
For large amounts of data, quality and benefits are selected first after collection and storage. Not all data offer added value in the analysis and help to eliminate disturbing factors preventively. To analyze the probability of an error occurring in a particular case, production controllers or production managers need KPIs that match the analysis objective.
For example, when it comes to temperature measurement and data, both man and machine have to define the target value for the temperature indicator. During the following operation or production, the actual value has to be measured in real time and constantly compared with the target values. Only now the system can evaluate whether a particular area of the resource is in the green or red range. The possible deviations that have been made transparent create connections and the resulting recommendations for action for controllers.
An integral part of industrial digitization
Machines can learn from historical data and statistics in the long term and optimize their processes independently. Once certain patterns have been defined and standard ranges defined, a system learns, for example, that the temperature and speed of a motor are interdependent. If a variable is outside the normal range – i.e. a date deviates significantly from the general average of the data set or the data combination – it becomes apparent as an aberration and can be identified.
What happens next with Predictive Maintenance?
Predictive Maintenance and repair is a crucial potential field on the agenda of companies and one of the most important advantages of digitization. With their help, complete production processes can be optimized. However, this does not (yet) work without humans. In the age of digitization, it is much more a matter of generating knowledge and experience and securing reliable and data-based decision-making aids. We will continue to retain management and responsibility.