At the Niederrhein University of Applied Sciences, Krefeld/Germany, research is being conducted on an artificial system that predicts the necessary machine parameters on the basis of material parameters and predefined product properties. A system solution for correlating these parameters is based on an artificial neural network that is taught using previously generated training data sets. The aim is to make the setting process of textile production machines more efficient in order to meet today's challenges.
Representation of the improved process flow based on the machine parameter determination by ANN
The textile market is facing enormous challenges due to issues such as sustainability/eco-balance, epidemics and digitalization/functionalization. The use and development of new highly functional and at the same time sustainably produced materials is rapidly increasing. In addition, there is a trend that the intersection between digitalization and textiles is becoming increasingly larger as a result of functionalization. In order to cope with an increasing number of material variants and a growing complexity of manufacturing processes, e.g. through the integration of functional components, production processes must be designed to be significantly more effective as well as more variable and flexible. In this context, the adjustment process of textile production machines is of decisive importance. Up to now, this has been characterized by a large number of iteration loops. This is due to the high number of parameters, both on the input side and on the output side. Yarn materials/properties, process parameters as well as product properties contain an abundance of variables, which the machine operator has to adjust to each other through a high degree of know-how and adjustment trials by means of the correct machine setting. This offers enormous potential for saving costs, time and material and thus making the machine setting process much more efficient.
Artificial intelligence (AI), in particular artificial neural networks (ANN), offer great potential for improvement in the area of more efficient production processes. Such an artificial system can support the machine operator or take over parts of the work. For example, it is possible to use an artificial neuronal network to develop the interrelationships of a textile manufacturing process in order to generate setting parameters for new process constellations in a further step. As a result, the adjustment process of a textile manufacturing machine is not only simplified, but by eliminating the need for repetitive or adjustment tests, both time and material are saved, and costs are reduced. Such an efficient machine adjustment process also enables the manufacturer to react quickly and flexibly to changing conditions and specifications.
Artificial intelligence offers a large and complex field of implementation possibilities for such an approach, which can be divided into different levels. Each level represents a component that must be considered in order to implement the desired approach. Therefore, one of the essential tasks is to choose the right path. Since a strong artificial intelligence, with its own mind and will, does not yet exist, we are in the realm of weak artificial intelligence. Machine learning is a subfield of this and can be assigned to data analysis. Within machine learning there are different learning methods. The selection of the right method depends, among other things, on the input data. Concrete numerical values, which for example describe the product characteristics, lead to the "supervised learning". The next level describes which applications are possible with the help of the previously selected learning method. The goal is a prediction of the necessary machine parameters and is implemented by a regression. An artificial neural network can have a different structure. Due to the complex production parameter correlations, an artificial neural feedforward network is used. The ANN is trained with the backpropagation method. The described artificial system is able to predict the necessary and previously unknown machine parameters on the basis of material parameters and previously defined product properties.
Implementation approach in the field of artificial intelligence (Source: HS Niederrhein)
The solution approach is for the artificial neural network to use input parameters, such as material properties and textile/product properties, to provide the necessary output parameters for setting the machine. The ANN forms a complex mathematical algorithm that describes the interrelationships of all parameters. Similar to the human brain, the ANN must be trained for this purpose with training data sets before it can be used. These training data sets are generated in advance by large test series on the machine. The data sets contain all the information of the manufacturing process. This means that for a training process both the input and the output must be given. For the result quality of an ANN, a high variance of the data sets is crucial. This is achieved by test series with different process constellations. In addition to the variation of individual parameters, changes in the overall constellation are also taken into account. After the training process, the ANN is able to generate the output for other process constellations by using only the input.
Artificial neural network for configuring textile production machinery (Source: HS Niederrhein) Vision
The goal and vision of the research project is the development of an application for tablets that contains the trained neural network. With this application it is possible to calculate the necessary machine parameters on the one hand and to access and read out the calculated values on the other, directly at the machine. The application can also be continuously expanded with data, creating an expert database and thus a personalized company application.
“NeuroTEX” application for prediction of machine parameters (source: HS Niederrhein)
For the implementation of the research project, the HS Niederrhein is open for cooperation with interested partners from the industry.