Industrial data analysis

This year, it should be so far: then machines exchange more data for the first time among people, who communicate with humans. This raises important questions: How to derive information that can be exploited from mere raw data, which is worthwhile at all?


Looking forward, not back


And what is needed to intelligently control machine-to-machine communication (M2M)? Answers provide a Forsa survey from the beginning of the year. On behalf of the software producer SAS, the market research institute determined the status quo for the evaluation of machine data in American industrial companies. The study shows that analytical procedures already play an important role here. At 75 percent, a clear majority of respondents surveyed machine and sensor data to identify sources of error and to reduce response times.


In-Memory makes tempo


Recognizing - or even better - avoiding errors in the products or production processes is also mentioned as a motivation for the analysis of data. In addition, many companies are concerned about securing and improving product quality and customer satisfaction. Therefore, awareness of the need to analyze machine data is available in large parts of the industry.


New Business Models


However, most companies do not yet use the data available to them. The view in practice shows that analytical solutions that "search" beyond the obvious for hidden contexts in the operating and production data are still not widely used.


M2M needs Analytics


Where present-looking reporting systems predominate, summarizing the facts that can no longer be shaken, however, analytical solutions should be used to anticipate and play through future developments. With regard to the penetration of analytical solutions in the industry, it should not be overlooked that 25% of American companies still do not deal with the issue at all.


As a reason for this, more than half means a lack of added value. Costs and technical feasibility also play an important role. However, even if these companies are currently ignoring the issue, they are well aware of the fact that more transparency is required in terms of the possibilities and requirements of Analytics in the industry.


If some companies are still skeptical about the analysis of machine data or do not exploit potential, then this has a good reason: It is not long ago that certain evaluations were technically not possible. Here, the company's current technical developments play a positive role in the
cards

Today, data can be analyzed in a quantity and speed that could not be expected a year ago. Big Data - and, in the end, nothing but the M2M data - is no longer a problem, but an area in which pioneers can currently gain real competitive advantage because they have more and better information than Others.


One solution for such analyzes is, for example, SAS Visual Analytics: the software is easy to use with visual elements. In addition, it can analyze a billion records in just nine seconds. As a result, evaluations that took up one to two days in the past are now completed within a few minutes. The reason for this enormous acceleration is the so-called in-memory technology and the price decline in the memory segment.


In-Memory means that all relevant data is loaded into the main memory and analyzed directly there. Advantage: During analysis, it is no longer necessary to move data between database and Analytics software.


This increase in speed finally makes possible an aspect that is particularly important for analytical processes in production: real-time. Modern analytics solutions are able to continuously monitor the production processes, not only looking back, but also predicting future events. And that is exactly the point of data analysis in production. It is not about individual measured values ​​in reports To be transferred. Instead, all data can be correlated with one another or examined for correlations.


The more data are collected, stored and analyzed, the more detailed the image of how the smallest elements within a production process - from the machine part to QM processes - correlate with each other, where there are sources of error, and which parameters are used to indicate errors at an early stage.


A classic application scenario for the analysis of machine data is therefore, for example, an early warning system which points to imminent quality losses in production and in the products, and at the same time shows what is to be done at which point in order to bring the process back into order. Keyword: Predictive Asset Maintenance.


From these technical possibilities directly new business models for the manufacturers of machines, which can expand their portfolio by intelligent maintenance and services: after all, such a Predictive asset maintenance system does not have to be installed at the production company, but can also be in a service center Of the machine manufacturer and calculate from the incoming data when the next maintenance is due in order to avoid a failure or quality degradation


And if this system is available for many machines with many customers, then it is even possible to calculate an optimum of maintenance cycles and utilization of the service technicians. This helps reduce downtime and downtime.


It is also exciting when machines produce such large quantities of data that it is simply impossible to store them completely. This is, for example, the case with wind turbines that produce two terabytes of usable information each hour. This means that considerable amounts of data are collected during the day and in a small wind farm. With the right software, however, it is possible to analyze the emerging data while it is being generated, and only when the software detects conspicuous patterns will it start recording the data.


Anyone wishing to initiate data exchange between machines must ensure that these data are also relevant, substantial and comparable. Only Analytics prepares the production, machine and operating data in such a way that it is worthwhile to exchange them.


Analytical solutions ensure that the data is given meaning, which gives information about the current status of the production - and about future developments that are hidden in the data. M2M is therefore reliant on reliable and high-performance analytics.