OREANDA-NEWS. October 08, 2015. A new raw material is driving the economy. But unlike steel, gold, or plastic, it is invisible and intangible. This raw material is the constantly growing stream of data from connected factories, connected cars, and connected products. Used correctly, it offers huge potential for better customer service and optimized production processes ? and thus for greater competitiveness. The ability to generate new knowledge from big data is a key competence of the future, the computer scientist Dr. Lothar Baum says. At the new Bosch research campus in Renningen, he heads a team of experts that are involved in the computational process of discovering patterns in large data sets, or data mining, as it is known. Among other things, Baums research focuses on how data mining can be used to optimize connected industry.

Cents add up to millions of euros
Data is the new oil of the global economy, Baum says. To give one specific example: by evaluating manufacturing data, Bosch has managed to cut the time taken to inspect hydraulic valves by 17.4 percent. Such huge savings are a major advance in modern manufacturing, and this despite the many considerable improvements that have already often been made there. With some 40,000 valves manufactured every year, this represents a saving of 14 days for the company. In this particular case, an analysis of the production data relating to 30,000 manufactured hydraulic valves showed that certain subsequent testing steps in the inspection process are unnecessary, provided the results of several earlier steps are positive. The outcome of those subsequent steps can be reliably predicted by analyzing the earlier steps. Pinpointing such correlations which are generally much more complex than the example given here saves time and money. When the number of parts runs into the millions, even savings of just a few seconds can soon add up, turning a few cents into millions of euros, Baum says. Every second and cent saved strengthens the competitiveness, and thus the attractiveness, of the products manufactured.

Bosch operates several clusters
Technically, all this is extremely sophisticated. While the algorithms this requires have essentially been known for many decades, it has so far been impossible to gather data to the extent now made possible by the internet of things. And a lack of computing power has meant that it has not been possible to apply the algorithms to several billions of data points, Baum says. Now, however, thanks to clusters made up of many interconnected servers, these huge computational tasks can be performed on thousands of processors working at the same time. Around the world, Bosch operates several such clusters. Humans play an essential role in this process: they have to program the computers so that they can process billions of data points efficiently and in parallel, instead of sequentially.

The technician only rings once
These capabilities are also in evidence in another example that shows the benefits of data mining. The utility company British Gas provides its customers with heating and hot water. Many of the Bosch boilers installed by British Gas are now web-enabled and transmit a wide range of data from their day-to-day operation to the utility company: When is the boiler in use and for how long? How much time does it take for the flame to light? How hot is the water? When a boiler needs longer to ignite than it used to, analyzing this information can reveal the potential causes, Baum says. Now service technicians can take the appropriate replacement part with them right away when they call on a customer. They already know where the fault lies. Up to now, technicians have usually had to visit twice: once to find out whats wrong, and a second time to make the repairs. Data analysis thus saves British Gas money, and customers benefit from faster, better service.

34 projects with an international presence
Bosch operates hundreds of production lines at around 250 plants worldwide. Many of them are already connected to the web. Sensors on these lines transmit data, while algorithms use that information to detect potential wear and tear and provide information for timely maintenance work. This avoids unplanned downtime, increasing productivity. At its new location in Renningen, Bosch is helping its researchers and engineers communicate with each other even better than before, so that they can develop such solutions. In addition, 34 data mining-related projects have already been started. A team of 40 experts from around the world is working exclusively on such projects. In doing so, they are supporting manufacturing associates in putting such projects into practice. The data experts are based mainly in Palo Alto, California in the heart of Silicon Valley and in Bangalore, India. Boschs global alliance partners in this field include Stanford University and the University of Pittsburgh.

Advantages thanks to a transparent production process
The big data component of the Bosch IoT Suite already makes it possible to explore and evaluate large volumes of data. The IoT (internet of things) Suite is a comprehensive software solution that can be used to develop, provide, and operate IoT applications.

New jobs for new experts
As the use of data mining increases, so does Boschs need for qualified software experts. Data scientists must be familiar with software and be able to write it themselves for special purposes. They must have an understanding of math, statistics, and machine learning. Whats more, they have to have detailed knowledge of the products and how they are manufactured, so that they can correctly interpret the data generated, Baum says. This is another reason why Bosch plans to hire 12,000 university graduates this year. With software growing more important in all Bosch divisions, graduates with IT skills also have good prospects, as do engineers.

The next goal: faster service at the car workshop
Baum and his colleagues are currently working to provide a forecast stating which cars will be visiting a Bosch Car Service garage and when, and what will be wrong with them. Workshops can then prepare for the necessary repairs by ordering the required spare parts, for example. This will also allow them to optimize their inventories. Because the spare parts are already in stock at the workshop and do not have to be ordered first, drivers will get a faster and better service. We already have the methods needed to evaluate the data. Now all we have to do is amass enough data to be able to use those methods to full effect, Baum says.