Machine data is one of the fastest-growing areas of so-called “big data” in companies — and OEE can be calculated from some of this machine data. The most valuable part of the data is created when it does not meet expectations, i.e. when “target” is not equal to “actual.” Management intervention is then expected in a well-managed organization. So how nice would it be if these anomalies in the large amounts of data were automatically detected and presented to humans for review? Artificial intelligence, or more specifically machine learning, is ready to crack this nut.
An anomaly is something special
If data is well prepared for human processing, deviations from one's own expectations can be easily identified. Data in a tabular format, for example, is poorly prepared for humans. They are well presented in a — cleverly selected — graphic, which is why no spreadsheet program dispenses with graphics functionalities. In a graph, humans can easily identify changes in the slope or a deviation from the target value.
Machines deliver their data to a database. This is the worst possible option for humans. Not only does he often have no access to sensor data. It is also a table format that is difficult to understand. It is impossible for humans to recognize an anomaly, i.e. a deviation from an expected behavior, pattern or structure.
Even if the data is processed graphically and the anomaly becomes apparent to humans — with their understanding of the expected behavior of the time series — we have better uses of human intelligence than for monitoring temporal data series around the clock.
Anomaly detection in OEE
Anomaly detection of the OEE data stream can be carried out on two levels: At the OEE level itself and at the level of the individual data streams for availability, performance and quality.
The following measurement can serve as an example of anomalies at the OEE flow level.
OEE decreases at irregular intervals. There is not a loss of availability, but there is a loss of performance. There is therefore no data point with a fault cause. Without visualization, production management would therefore not even notice the loss. However, people recognize this anomaly in a graph even without it being formally described. It is only through visualization that it becomes visible and tangible. However, the cause is not yet known. If you now compare many of the product's data streams over a longer period of time, the anomaly may be assigned a pattern and thus a presumption of cause.
During ongoing operation of a factory, there are an endless number of such possible incidents:
- The OEE of one product is regularly lower than that of the other products
- The night shift has more short stops than the day shifts
- The OEE is lower than usual from 13:00 to 14:00 regardless of the product
- The starting curve of a product up to the ridge line takes longer than for other products
- ...
As a result of the usual reporting system based on quantities per shift or hour or the OEE of the shift or order, the vast majority of these opportunities for OEE optimization are lost.
But how do you get these productivity enhancement potentials without a person having to continuously monitor the data flows?
OEE anomaly detection through machine learning
Developments in the field of artificial intelligence, or more specifically machine learning, are suitable for automatically detecting anomalies in large amounts of data and data streams.
Machine learning is a generic term for the “artificial” generation of knowledge from experience: An artificial system learns from examples and can generalize them after the learning phase has ended. This means that the examples are not simply learned by heart, but it alone “recognizes” patterns and laws in the learning data.
In the field of machine learning, there are a variety of algorithms that must be selected and trained for the respective purpose.
So when the OEE data is now available in machine-readable form, a correctly trained and suitable algorithm evaluates it and processes the detected anomalies for humans, the technology opens up completely new opportunities for OEE optimization. We're working on that. Stay tuned for our extended product oee.ai.