In modern manufacturing, artificial intelligence (AI) is fundamentally changing data analysis. While the acquisition and analysis of production and machine data used to focus on finding remedial measures for existing problems, AI now makes it possible to proactively prevent defects and optimize processes.
COSMINO AG has been a solution partner for the digital factory for 36 years and has been involved in acquiring and analyzing data from production right from the start. Over the years, the requirements and possibilities for data analysis have changed again and again. Initially, the best possible presentation of historical evaluations was an issue. This was always linked to how the data could be acquired as time-saving as possible. A machine connection only became common later and was initially nowhere near as simple and standardized as it is today.
With the increase in real-time data, it was also used for factory visualization and to trigger automatic alarms in the event of certain data constellations. The expected gain in knowledge through the analysis of big data required the use of the most modern technologies and software components. Our experience in data acquisition and analysis has grown over these 36 years in a similar way as the amount of data to be collected from our customers has increased.
In contrast to the past, we no longer think about which data we need and therefore want to store in projects but assume that certain data will be needed for a big data analysis later anyway. In projects with data acquisition, we prepare these for the later use of AI. This is generally guaranteed by the Cosmino system architecture and therefore does not require any additional effort. As big data has no place in the relational MES database, a time series database and an analysis tool are used here, which are also already prepared for processing big data and using AI.
System architecture with integration of AI solutions. Data is made available to any system for analysis in near real time. Identified anomalies and the like are then reported to the Cosmino software, also in near real time.
The manual analysis options have long since reached their limits given the amount of information available from production today. This is where artificial intelligence (AI) comes into play to recognize complex patterns in huge amounts of data. AI not only promises automation and increased efficiency in data analysis, but also a completely new dimension in knowledge acquisition.
These new possibilities will determine the success of a company. Those who rely exclusively on conventional analyses without AI will no longer have the same competitive opportunities in the medium term.
This means that both the software partners, but above all the industrial companies, must work together to explore the possibilities of AI in order not to lose out in the competition for more cost-effective and efficient production processes.
Through various projects and together with partners and institutes, COSMINO AG has built up expertise in the use of AI. We have learned what potential can be extracted from mass data (big data) using artificial intelligence (AI).
Artificial intelligence is not only interested in production and quality data. Process data, which is often available in large quantities, forms the basis for finding dependencies between values and therefore anomalies.
The more data is acquired for a process, the greater the possibilities for analysis by artificial intelligence. Where humans can only manually access mass data on a situation-specific basis, neural networks are formed instead, which artificial intelligence can use to identify connections and relationships between the data.
This enables a contextual and semantic analysis of problems and causes. And, of course, early detection, which means reacting to problems before they affect the process. And thus preventing downtimes, tool breakages, poor quality, increased energy consumption and the like from occurring in the first place!
The best way to build up expertise in the sensible use of artificial intelligence is with the support of experts. This is why we are involved in various research projects on the use of AI and have expanded our network to include meaningful AI partnerships that will support us in our AI analysis.
The following example shows the enormous potential of artificial intelligence in data analysis:
The question was investigated as to whether the process parameters indicate whether an increased reject rate can be expected on this machine in the near future.
Fig.: Graphical listing of the data of machines A-E by scrap rate over a period of time
In the next step, a neural network was created. The AI was trained to detect an increase in reject rate 5 days in advance.
Fig.: Illustration of the early warning enabled by the AI (red), i.e. before the reject rate increases.
Fig.: Prediction rate of good and poor quality by AI
The above example shows that AI-supported analyses can predict an increase in the reject rate with a high degree of probability.
In general, the more data, the greater the probability of success for data analysis with AI. In a joint discussion with our AI experts and the customer’s process specialists, we find out what data is available and whether an AI analysis makes sense.
According to a business survey conducted by the info-Institute in 2023, 17.3% of the manufacturing industry is already using AI. 51.6% are considering the idea.
It is therefore time for every production company if it does not want to fall behind its competitors.
Contact us if you are interested in an initial consultation. Regardless of whether we can help you get started with AI-supported data analysis or whether you have already gained your own experience and now need an additional boost.