The Paint Shop Can Now Rely on Dürr's Artificial Intelligence

Date: 17/04/2020

Dürr presents Advanced Analytics, the first market-ready AI application for paint shops. Part of the latest module in the DXQanalyze product series, this solution merges the latest IT technology and Dürr's experience in the mechanical engineering sector, identifies the sources of defects, defines the optimal maintenance programs, tracks previously unknown correlations and uses this knowledge to adapt the algorithm to the system using the self-learning principle.

Why do pieces frequently show the same defects? When is the latest that a mixer in the robot can be replaced without stopping the machine? Having accurate and precise answers to these questions is fundamental for sustainable economic success as it every defect or every unnecessary maintenance that can be avoided saves money or improves the product quality. "Before now, there were very few concrete solutions that would have allowed us to promptly identify quality defects or failures. And if there were, they were generally based on a scrupulous manual evaluation of the data or trial-and-error attempts. This process is now much more accurate and automatic thanks to Artificial Intelligence", explains Gerhard Alonso Garcia, Vice President of MES & Control Systems at Dürr .

Dürr's DXQanalyze digital product series, which already included Data Acquisition modules for acquiring production data, Visual Analytics for visualizing it, and Streaming Analytics, can now count on the new self-learning Advanced Analytics plant and the process monitoring system.

The AI application has its memory

The peculiarity of Advanced Analytics is that this module combines large amounts of data including historical data with machine learning. This means that the self-learning AI application has its own memory and that it can therefore use the information from the past to both recognize complex correlations in large quantities of data and predict an event in the future with a high degree of precision based on current conditions of a machine. There are lots of applications for this in paint shops, whether at component, process, or plant level.

Predictive maintenance reduces plant downtimes

When it comes to components, Advanced Analytics aims to reduce downtimes through predictive maintenance and repair information, for example by predicting the remaining service life of a mixer. If the component is replaced too early, the costs of the spare parts increase and consequently the general repair costs increase unnecessarily. On the other hand, if it is left running for too long, it can cause quality problems during the coating process and machine stoppages. Advanced Analytics starts by learning the wear indicators and the temporal pattern of the wear using high-frequency robot data. Since the data is continuously recorded and monitored, the machine learning module individually recognizes aging trends for the respective component based on actual use and in this way calculates the optimum replacement time.

Continuous temperature curves simulated by machine learning

Advanced Analytics improves quality at process level by identifying anomalies, for example by simulating a heat-up curve in the oven. Until now, manufacturers only had data determined by sensors during measurement runs. However, the heat-up curves which are of fundamental importance in terms of the surface quality of the car body vary since the oven ages, during the intervals between the measurement runs. This wear causes fluctuating ambient conditions, for example in the intensity of the air flow. "Up to now, thousands of bodies are produced without knowing the exact temperatures to which the individual bodies have been heated. Using machine learning, our Advanced Analytics module simulates how the temperature changes under different conditions. This offers our customers a permanent proof of quality for each individual part and allows them to identify anomalies", explain says Gerhard Alonso Garcia.

Higher first-run rate increases overall equipment effectiveness

As for the implant, the DXQplant.analytics software is used in combination with the Advanced Analytics module in order to increase the overall effectiveness of the equipment. The German manufacturer's intelligent solution tracks recurring quality defects in specific model types, specific colors or on individual body parts. This allow the costumer to understand which step in the production process is responsible for the deviations. Such defect and cause correlations will increase the first-run rate in the future by allowing intervention at a very early stage.

The combination between plant engineering and digital expertise

Developing AI-compatible data models is a very complex process. in fact, to produce an intelligent result with machine learning, it is not enough to insert unspecified amounts of data into a "smart" algorithm. Relevant signals must be collected, carefully selected and integrated with structured additional information from production. Dürr was able to design a software that supports different use scenarios, provides a runtime environment for machine learning model and initiates model training. "Developing this solution was a real challenge as there was no valid machine learning model and no suitable runtime environment that we could have used. In order to be able to use AI at the plant level, we have combined our knowledge of mechanical and plant engineering with those of our Digital Factory experts. This led to the first artificial intelligence solution for paint shops", says Gerhard Alonso Garcia.

Skills and knowledge combined to develop Advanced Analytics

An interdisciplinary team made up of data scientists, computer scientists and process experts developed this intelligent solution. Dürr has also entered into cooperation partnerships with several major automotive manufacturers. In this way, the developers had real-life production data and beta site environments in production for different application cases. First, the algorithms were trained in the laboratory using a large number of test cases. Subsequently, the algorithms continued on-site learning during real-life operation and adapted themselves to the environment and conditions of use. The beta phase was recently completed successfully and showed how much AI potential it has. First practical applications are showing that the software from Dürr optimizes plant availability and the surface quality of painted bodies.