OEE improvement through human or artificial intelligence?

In order to increase plant productivity, humans and machines will share work in the form of human and artificial intelligence and capabilities. Where are the tasks of humans, where are the tasks of technology to achieve the best results?

In order to increase plant productivity, data such as the number of pieces produced per unit of time and causes of availability, performance and quality losses are recorded as time series data. Time series means that this data is collected and documented at regular intervals, e.g. once a minute. As a result of this, large amounts of data are quickly created, which must be analyzed for patterns, anomalies and trends.

The chain of activities for this optimization process consists of 6 steps, which are carried out and can basically be carried out by both humans and technology. However, if you have the additional requirement that the processes should be carried out precisely, quickly, permanently and at low costs, the point comes when people and technology should focus on their respective strengths. People can do a lot, but technology can do a lot better. The following figure distinguishes between who/what performs the activity when no technology is available and when it is available.

At the first three levels of the process, there is no doubt that taking over the activity through technology makes sense. On the top two levels, humans are unrivalled. No currently available technology has ideas or can implement physical measures. That is the domain of man.

At level 4, data analysis, rules, statistics and artificial intelligence can be used in the form of machine learning. For example, it is possible to determine how the OEE or downtime trend is progressing in terms of duration and frequency, the plant operation can be clustered into stable and unstable, the quality of conversion processes can be measured, downtimes are grouped or classified, or faulty measured values can be identified.

This rapid and precise evaluation of large amounts of data is a domain of technology. With the availability of these options, the analysis of plant losses over the past (!) Day in the morning shop floor management of the past. Instead, unusual data flows are identified online and made available to the employee live on an Andon board or an industrial smartwatch. In this way, the attention of the plant team can be drawn to the problem even during the loss.

This reaction speed can only be achieved with a balanced mix of automation, analytics and people. The future belongs to this combination.