What is happening in the painting cell and what is the condition of the technology? EcoScreen Equipment Analytics provides the answers. The software records and analyzes all relevant signals from the sensors and actuators integrated in Dürr painting robots. The application technology of the front robot arm, for example, permanently provides data on the pressure regulators, metering pumps and color valves. Main needles, turbine speed, shaping air and air heaters from the electrostatic rotary atomizers are also connected to the data recorder. The software also records the positions, torques, and temperatures of the painting and handling robots’ individual axes. Data from the conveyor technology is additionally collected for the positions of the bodies in the painting booth. This information is compared with data from the painting cells, such as start and end time of painting, or production data for individual bodies such as type and color code. All errors and warnings from the painting process are also collected.
Visual representation creates transparency
In addition to data recorder and database, the IIoT solution comprises a visualization for the graphical representation and analysis of the results. “With the help of the Visual Analytics module we can precisely trace the data from the previous weeks with millisecond accuracy,” explains Dr. Lars Friedrich, President and CEO of Dürr Systems AG. “The detailed information helps us to identify and eliminate faults quickly. The cross comparisons of different robots improve root cause analyses and help the plant operator to boost plant availability, increase the first-run rate, and improve the overall process sequence.” This creates transparency. For example, the signals collected from the metering pump together with flow rates, pressure regulators, and main needles of the atomizers can be displayed synchronously on the control computer and correlations can be defined. The software can display these process signals in a 3D view together with the trajectories traveled by the robots and the exact position of the body, and perform a comparison. By superimposing the signal curves, EcoScreen Equipment Analytics can also compare bodies by type and color.
Algorithms calculate the ideal plant condition and show anomalies
The additional module “Streaming Analytics” goes one step further: it analyzes the data in real time. Using algorithms, it calculates patterns and correlations that describe the ideal plant condition from the data collected. Even the smallest anomalies both within the plant and in the process are automatically identified. Employees can initiate corresponding service measures straight away and get to the bottom of the deviation before the body even leaves the paint booth. EcoScreen Equipment Analytics uses software from the IIoT platform ADAMOS for the streaming analytics function. Dürr operates ADAMOS together with Software AG and several mechanical engineering companies.
Model-based plant monitoring
“Streaming Analytics” provides the plant operator with easy-to-use graphical interfaces that do not require any programming expertise to create analysis models. For this, Dürr developed a Model Editor that provides the user with a library of analysis modules. These can also be individually combined into new models
(e.g. a model for monitoring the painting pressure during application). The data is displayed graphically on the screen. The operator can see that all values are within tolerance. With Streaming Analytics, the expert can create an analysis model that automatically determines anomalies such as air bubbles in the paint from the characteristics of the signal curve. These possible applications make the Streaming Analytics solution a powerful tool, since the software automates the plant operators’ process knowledge using online analyses.
Integration of machine learning in the future
“In parallel with Streaming Analytics, Dürr is developing further modules for batch analytics applications that work on the basis of artificial neural networks. The software learns the optimal process condition fully automatically and registers every deviation. This means that the software will solve problems itself in the future through machine learning,” explains Dr. Friedrich. “If the self-learning modules detect trends, problems in the equipment can even be identified in advance by specifying a forecast of the event time.”
A number of car manufacturers have been using the EcoScreen Equipment Analytics software since 2018. The additional module Streaming Analytics based on real-time data has been used for the first time in a complete production line since the fourth quarter of 2018. The Batch Analytics module, which uses artificial intelligence, is currently in the trial phase with a customer.
The EcoScreen Equipment Analytics software was developed in Dürr's Digital Factory. In this competence center, founded in early 2018, around 100 software experts are working on solutions for the digitization of production processes.