The term “assistance system” refers to electrical, information technology or mechatronic systems (e.g. a software, a car) that provide the user with assistance in a current real situation or even actively intervene in the operating process and carry out actions in the background.
As virtual assistants (e.g. in the PC area) already simple visualized information, for example labels or tooltips (hints that appear automatically when the mouse pointer is moved over them), can be displayed to the user exactly at the moment when he draws his attention to this point. By moving the mouse, the assistance system registers which question may be of interest to the user and reacts promptly with precise, targeted information.
In the automotive sector, such assistance systems already perform a wide variety of safety functions, so that software can register, for example, on the basis of the rotational speed of the individual wheels, the steering angle and the information from various acceleration sensors, in which driving dynamic state the vehicle is currently in and can restore driving stability through targeted actuator intervention.
Both systems have the basic functionality that they register the initial situation of the system. The assistance system recognizes what the situation is. This knowledge of its own situation is the first step necessary to initiate all further follow-up actions.
What are extrusion lines already capable of today?
Today’s extrusion lines are generally equipped with modern machine control systems (PLC), which are equipped with a large number of sensors and actuators. The information from the sensors is graphically visualized for the machine operator and the storage of historical data is usually also possible without problems (with more modern systems).
The various actuators of the system can be parameterized quickly and easily by the machine operator, so that operating point changes or other changes to the process can be efficiently implemented from a central point (the machine control system). Changes to the operating points can be stored and retrieved as a recipe in more modern systems or stored in network databases in the event of changes.
At various points, the use of controls is also state of the art, so that, for example, the heating capacity of a heating zone does not have to be “controlled” by specifying a switch-on time for the heating, but is automatically “controlled” by the machine controller by specifying a set temperature.
In other places peripheral systems are used for control, which take over sub-functions as an autonomous solution (e.g. temperature control units, haul-off, saw, IR systems). These systems have their own machine control and internal control, which adapts system-specific setting variables. In modern systems, the target parameters are often communicated from the main controller to these subsystems and sensor information from the subsystems are returned to the main controller so that it can be visualized there (bidirectional communication). Often, however, the target values still have to be set directly on the peripheral system.
A current extrusion line thus already has:
- Sensors to detect the situation
- Actuators to influence the situation
- integrated controllers in the main controller
- additional, decentralized controllers in peripheral components
- Communication of setpoints to the periphery, partly bidirectional
What can extrusion lines not (yet) do today?
Although extrusion lines today already have a large number of sensors and actuators, it has not yet become standard practice for extrusion lines to recognize and control the quality of the product or to provide the machine operator with information on how to operate the lines. A virtual assistance is therefore not available.
Even if the process parameters of the plant (e.g. screw speed, melt pressure, melt temperature, cylinder zone temperatures, melt pump speed, material throughput, metre speed, cooling parameters, etc.) are as well known in the machine control as important quality parameters (e.g. ovality, wall thickness distribution, surface quality, etc.), a correlation between process parameters and the direct effects on the product (quality parameters) is not available today.
One of the reasons why such links are not yet state-of-the-art is the complexity of the plastics processing process. As can be seen schematically in the graph above, almost every process parameter that can be set on an extrusion line somehow has an influence on other process variables as well as on various quality parameters.
It is therefore not possible to identify simple (one-dimensional) relationships of style: “IF “poor quality”, THEN “reduce process parameter X by Y%”. Due to the interactions that prevail during extrusion, the interrelationships are much more complex and influence each other.
Virtual assistance systems in extrusion
1. Sensor data from the process
As shown above, the presence of sensor information to evaluate the condition of a production system is the first step in building a virtual assistance system. The system must be able to detect the actual condition of the system. The current trend towards digitization / Industry 4.0 means that more and more sensors are being installed in extrusion lines and that the various information is increasingly converging in networked systems. This fact is one of the factors why virtual assistance systems for extrusion can only be created today.
However, the increased equipment of machines with sensor technology is not yet sufficient to be able to use the full advantages of virtual assistants. Unfortunately, many important information from the process are not known or not economically measurable (e.g. due to the very high price of measuring systems). However, the availability of digital information about the process status is one of the essential factors without which an assistance system cannot function, so this point must be further advanced.
2. Live computer simulation as sensor
The possibilities of simulating processes with the help of numerical simulation systems have gained enormous importance in recent years. Today it is possible to calculate extruder screws, extrusion tools or cooling lines with the help of different computer simulation systems, so that additional information about the process are available that would be impossible to capture with physical measuring principles available today.
A simple example of this is the measurement of the core temperature of an extruded profile, a sheet or even a solid bar. While the introduction of a sensor into this area is practically impossible, the calculation of the core temperature (e.g. with the chillWARE cooling simulation) is simple and quick to realize. Therefore, computer simulation systems that automatically parameterize themselves live on the basis of process parameters can provide countless additional meaningful information about the process without the need for real measurements.
3. Process know-how / interactions
If a process as well as the product quality resulting from the process is completely available as digital information, the fundamental step is done to actually start building a virtual assistance system. Based on this information, it is now possible to identify relationships between input parameters and output parameters. A “model” or a “mathematical formula” is required, which is able to predict a quality parameter on the basis of the input information.
4. Data-driven models vs. physical models
There are different approaches to create such a model, whereby essentially the data-driven models have to be distinguished from the physical models. A data-driven model learns correlations based on the available data, regardless of whether the connection is physically logical or illogical. A physical model always has a real, physical background that must be comprehensible and correct.
At first glance, the physical model appears superior, since only physiaclically plausible information is considered in a physical model. Unfortunately, however, the physical models are partly inferior to the data-driven models due to their enormous complexity. Ultimately, both approaches have advantages and disadvantages and a decision should be made on a case-by-case basis as to whether a physical modeling approach, a data-driven approach or even a hybrid approach should be chosen.
5. Assistance functions
If points 1-4 have been successfully processed, an assistance system can be provided on this basis, which either provides information or alternatively actively intervenes in the process. For example, it is conceivable that an extrusion line could provide the machine operator with the following information:
By changing the process parameters to the recipe XYZ:
- an increase in product quality of X
- a reduction of the residual stresses in X
- a reduction of the deformation of the profile (curvature) by X
- the mechanical properties of the product can be increased by X
Alternatively, direct intervention in the process parameters by the assistance system would of course also be possible.
Virtual assistance systems for plastics processing hold enormous potential for increasing robustness, productivity and product quality. The current trend towards digitized systems and processes is an important first step for the realization of such systems. The continuing trend and the ever improving technical possibilities for computer simulation (computing power) also contribute positively to making additional information from live simulation possible. Developments in the field of data-driven modelling but also in the field of fuzzy logic and approaches such as those of neural networks enable the targeted evaluation of data and the identification of correlations. Thus, the prerequisites for the development of virtual assistance systems are excellent.
If you are interested in further information about virtual assistance systems for extrusion, please register here for our free newsletter and download area. You will receive a notification when new articles are published and will have access to many interesting additional information, checklists and calculation tools.
(SHS plus GmbH (operator of this site) is currently working actively with various cooperation partners, customers and various institutes and universities on the development of virtual assistance systems for plastics processing and is currently involved in various publicly funded projects. If you have specific questions regarding content, we are at your disposal for an informal discussion.