Expectations for artificial intelligence as one of the Industry 4.0 technologies are very high. Everyone in the industry has already encountered and likely uses other technologies such as robotics, production automation, blockchain, or the Industrial Internet of Things. They are already visible in production. However, AI itself is only slowly finding its application, primarily in data analytics and production optimization systems. How can artificial intelligence support production, industry, and logistics? – comments Tomasz Haiduk, lecturer at the Warsaw University of Technology Business School and President of the Polish Forum of Automation and Robotics.
Data analytics is crucial for obtaining a proper basis for decision-making in a rapidly changing world. We must create the foundations to forecast and make the right decisions faster. Currently, this process is very long, especially in the production area. We are in a situation where the data we receive for analysis is purely historical. This is due to many things, primarily the fact that production lines are not properly metered. Also, utility monitoring in most companies in Poland is not sufficient.
Data consolidation is key
Only recently, after the wave of energy price hikes, has proper importance been attached to metering production and utilities to answer the question – is it profitable for me to produce at all, and if so – at what price?
In my opinion, the production area is still very neglected, and the first step to take is, of course, obtaining reliable data. We encounter the first barrier here – some plants already have sensors installed in production areas, but currently, each SCADA, MES, or ERP system is an isolated system, working on its own metering and its own database. As island solutions, they rarely cooperate with each other, which leads to reports from different systems being inconsistent. The first step to optimization is consolidating data in one place.
Data lake
Here, the Industrial Internet of Things comes to the rescue as one of the Industry 4.0 technologies. Due to infrastructure limitations and the time required for installation, intelligent wireless measuring sensors are increasingly used for production metering. They send pre-processed data (edge computing) to one place, one database. It is called a data lake and constitutes a shared information base for all production systems.
If we create a common data lake, a place where we have all data from the production area at our disposal, each system will download it for its own needs. This will eliminate the risk of discrepancies for SCADA visualization systems, MES production control systems, or ERP production management systems. The entry of artificial intelligence algorithms into production takes place mainly through industrial IoT.
From data to information
We already have the data – but what can we do with it next? How to process data into information faster? How to ensure that reports do not refer to historical situations but support management in a shorter time horizon?
Unfortunately, building classic Business Intelligence algorithms hits the barrier of finding the right correlation between data, its preparation, filtration, and consolidation. The transition from data to useful information is a key problem facing business support IT systems. Here, algorithms based on artificial intelligence come to the rescue. The transition from classic Business Intelligence to AI-based models will provide a good foundation for modern business analytics, optimization of production processes, and real-time decision-making. Correlations between individual data will be analyzed online in new algorithms supported by artificial intelligence. Industry 4.0 assumes interactions between people, machines, and IT systems based on data updated online. Optimization algorithms find or indicate the optimal method of conducting the production process, focused on the assumed target path and specific external situation scenarios. For example, they allow for modeling production in a rapidly changing environment with dynamic changes in energy prices or with limitations in the supply chain, access to personnel, or components necessary for production. Thanks to this, they answer the question – how, when, and what should be produced so that it is optimal for the enterprise under specific boundary assumptions.
Condemned to one’s own experience and education
Of course, currently, such questions are mainly the concern of management. For now, they are condemned to their own experience and education. AI-based production decision support and optimization systems will allow for conducting the process online to choose the optimal solution for the plant.
The next step is the transition from decision support systems to fully automated production control. In such a model, a management system based on artificial intelligence, having data from the process and expected business assumptions, controls the technological process by communicating directly with machines and changing their settings online. This is the true essence of Industry 4.0, in which cyber-physical systems communicate with each other and people in real-time and control production in the digital world.
This is the future and the main concept of applying AI algorithms in the industry. It can be compared to advanced regulatory algorithms that, based on data obtained from production, can answer the question – what, when, and how to produce, how to control the technological line, how to adjust its load, and how to dynamically change the production layout under specific initial assumptions.







