Plastic Processing Machine Selection Using Simulation and Federated Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems lack an efficient method to determine the most suitable machine for processing plastics and other plasticisable materials based on specific manufacturing criteria, leading to suboptimal selection and operation of machines.
Innovation Solution
A computer-implemented method using a reference database with machine data, user data, and simulation data to identify suitable machines through federated learning, allowing for the determination of machines that best meet manufacturing criteria by comparing partial data sets and forming clusters based on machine configurations, processes, and materials.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a comprehensive reference database with machine data and simulation data is used to determine suitable machines, then the accuracy of machine selection is improved, but the complexity of the system increases
Solution Approach 1:
The patent segments the machine determination process into distinct functional modules: a reference database module storing machine data and simulation data, a data processing module that performs federated learning and cluster analysis, and a machine recommendation module. This segmentation allows each module to specialize in specific tasks, improving overall selection accuracy while managing system complexity through modular architecture.
Solution Approach 2:
The patent introduces a computer-implemented method as an intermediary layer between user requirements and machine selection. This intermediary processes user data, performs simulations, conducts federated learning analyses, and generates machine recommendations, thereby improving selection accuracy without directly increasing the complexity of physical machine systems.
2Reliability
If federated learning and cluster analysis are performed to optimize machine selection, then the quality of machine determination is improved, but the computational time and resources increase
Solution Approach 1:
The patent performs preliminary actions by pre-storing comprehensive machine data and simulation data in the reference database before actual machine selection is needed. Federated learning models are pre-trained and cluster analyses are pre-computed based on historical data, enabling faster real-time machine determination while maintaining high quality through advance preparation of analytical frameworks.
Solution Approach 2:
The patent uses simulation data as copies of actual machine operation data to perform federated learning and cluster analysis. These simulated datasets replicate real-world scenarios without requiring actual machine operation time, allowing comprehensive analysis to be performed computationally rather than through physical experimentation, thereby reducing time loss while maintaining determination quality.
3Adaptability or versatility
If detailed criteria including material characteristics, target machine characteristics, and target process data are considered, then the suitability of machine selection is improved, but the difficulty of data collection and processing increases
Solution Approach 1:
The patent creates a universal reference database that stores multiple types of data (machine data, simulation data, material characteristics, process data) in a unified format. This universal database structure allows the system to handle diverse criteria for different machine types and manufacturing scenarios through a single integrated platform, improving machine suitability across various applications while reducing the difficulty of data collection through standardized interfaces.
Solution Approach 2:
The system performs self-service by automatically collecting, processing, and analyzing the detailed criteria data through federated learning algorithms. The computer-implemented method autonomously gathers material characteristics, target machine characteristics, and target process data from the reference database, performs the necessary analyses, and generates machine recommendations without requiring manual data collection efforts, thereby improving versatility while reducing data collection difficulty.
Data Source
AI summary
The invention relates to a computer-implemented method and system for determining at least one machine for processing plastics on the basis of criteria determined for producing a component (2.14). In a reference database (RD), reference machine data (2.4) are provided, comprising reference partial datasets (RTDS) required for determining machine parameters. User data (2.3) are provided as determined criteria in a data store (2.2), which user data (2.3) comprise data required for producing at least one component (2.14). A simulation for producing the component (2.14) takes place on the basis of the user data (2.3) on a machine by generating simulation datasets. The simulation datasets are segmented in order to extract partial datasets, wherein the partial datasets relate to machine-related machine parameters with which a machine should be operated in order to produce the component (2.14). The partial datasets are compared with reference partial datasets to identify partial matches, in order to output a minimum requirement of a machine when producing the component (2.14). The minimum requirement is compared, using federated learning, with available machines in order to determine at least one machine suitable for producing the at least one component (2.14), and this machine is then operated accordingly. As a result, at least one machine can be provided to a user for use thereof, which machine is the best-suited for this purpose, according to the circumstances.