Machine Learning on the rise The pioneer for Industrial Intelligence in the industrial environment

Anyone who follows the media coverage of artificial intelligence (AI) and machine learning (ML) a little will find that it requires a great deal of mental stamina: anything seems possible, and that is precisely the crux of the matter. Potential users from the industrial environment, for example, are asking themselves what concrete possibilities there are for using AI approaches profitably - here and now? If the assessment of leading experts proves to be correct, there is an urgent need for action... The rapid increase in the spread of AI in the next 1.5 years will also play a not inconsiderable role. There is no time for exploratory expeditions to distant galaxies, instead it is time to act and exploit the potential today. With this in mind, we have summarized everything you need to know about the significance of Machine Learning for production and logistics.

Artificial Intelligence versus Machine Learning

In many cases, "artificial intelligence" (AI) and "machine learning" (ML) are used synonymously. But this is not entirely correct. AI stands for applications in which IT systems and machines, such as robots, perform with the help of trained intelligence. Machine learning, on the other hand, is practically a subdiscipline of AI, albeit a crucial one. The focus here is on the (industrial) Internet of Things ( IoT) and the ambition to teach systems (put simply) in such a way that they begin to learn on their own and also incorporate the exorbitant data volume of the IoT.

Deep learning, on the other hand, is about constantly developing acquired skills. The idea is that machines that have already reached a certain level of intelligence will use available information and algorithms to recognize repetitive patterns on their own, perform analyses and derive conclusions. Ultimately, devices are capable not only of finding suitable solutions in the shortest possible time, but also of learning from mistakes made and continuously optimizing themselves in the process. In order for this to succeed, humans are called upon in advance to provide, among other things, appropriate algorithms as well as rules for pattern identification.

What technological leaps are driving further development?

With predictive maintenance , many companies in the manufacturing and processing industry have already taken the first steps toward the transformation to a smart factory . But this is practically just another logical step, considering that pick-by-voice, voice-assisted picking, for example, is already an AI-based application. Speech and text recognition capabilities have also been used for other digital assistants for some time. In addition, there are gesture or facial recognition (pick-by-motion) or automatically generated purchase recommendations, as we know them from Amazon & Co. In addition, self-learning control loops can already be integrated into the Manufacturing Execution System (MES) by means of machine learning-based action rules.

Machine learning algorithms have thus been used intensively for quite some time. Their use is recommended wherever existing business processes are to be further automated . In view of existing Big Data technologies, even enormous amounts of information can be processed and fed accordingly. The high-performance in-memory database SAP HANA, for example, provides unprecedented support here.

Among other things, it is possible to implement image analysis and monitoring based on deep learning algorithms at the operational process level. In addition, texts and graphics contained in electronic documents can be extracted and evaluated. De facto learning machines are much more powerful than humans when it comes to deriving reliable predictions from large amounts of data. For users of such technologies, this means that they can draw on unprecedented solution quality and continuously build up new knowledge.

Typical applications in industrial environments

In industry, the potential of Artificial Intelligence (AI) and Machine Learning (ML) can be used especially in the areas of monitoring, quality assurance, logistics and energy management. Using ML, for example, managers are always informed about the behavior and status (down to the smallest detail) of their machines. In the event of deviations from the regular status, appropriate warnings are triggered. In addition, forecasts can be made regarding the quality of a product - even at a very early stage of production.

ML also functions as a key technology for automated intralogistics. Capable of "deep learning," driverless transport systems (AGVs), for example, take over the just-in-time supply of assembly lines with the required components. During the journey or transport, they localize their surroundings and navigate their course safely as a self-organizing fleet, if necessary parallel to humans. Last but not least, machine learning can be used to optimize procurement on the energy market: past consumption patterns can be used to level out future fluctuations in demand and prices.

How disruptive is machine learning (ML)?

"Machine Learning is revolutionizing business models": This was the conclusion of a study conducted by SAP and the Economist Intelligence Unit in the summer of 2018, which stated that machine learning has the characteristics of a disruptive technology. For example, machines could constantly learn on the basis of acquired data and in conjunction with intelligent algorithms and subsequently draw conclusions from this, similar to the ability of humans. In addition, Machine Learning is said to increase the efficiency of business processes "exponentially" without any increased effort.

Added value through machine learning on the rise

According to the study, automation frees up both human and monetary resources. Last but not least, artificial intelligence (AI) and machine learning (ML), which are on the rise, create the conditions for multiplying knowledge and making forecasts that previously seemed unthinkable in terms of their hit rate. On this basis, the chances of being able to develop new viable business models that offer customers clear added value are increasing.

SAP Leonardo - Enabler for Industrial Intelligence Applications

Big Data, Blockchain, Internet of Things (IoT), Analytics, Machine Learning, Data Intelligence - these fields of innovation are served by SAP Leonardo, the central "Digital Innovation System" that the Walldorf-based company named after the Italian polymath Leonardo Da Vinci. However, the integrated functions and services in SAP Leonardo, which are geared to machine learning, are not as new as one might assume: Rather, they have been continuously adapted by SAP to meet growing requirements. At their core, the ML components include the SAP Predictive Analytics, HANA Predictive Analytics Library and Leonardo Machine Learning Foundation applications.

SAP Predictive Analytics is offered either as a cloud application or as an on-premise version. The latter can be licensed and used as a standalone software solution. In parallel, it is possible to connect the SAP S/4HANA in-memory database. SAP Predictive Analytics is characterized by a graphically oriented user interface that can be used to access SAP HANA Library functions. These, in turn, simplify the training of ML pilots to then start directly with the deployment in practice.

Are you curious about how artificial intelligence and machine learning in SAP can also benefit your company? We would be happy to evaluate together with you which production processes promise positive effects through the implementation of Industrial Intelligence and under which conditions a quick and secure implementation based on the "SAP Leonardo Machine Learning Foundation - Functional Services" is possible. We will also be happy to inform you about the potential of a "Digital Twin" in the cloud with the SAP Manufacturing Execution System (SAP MES).