Identify components in the construction thanks to AI
When parts are designed, they are often designed using CAD software. Computer Aided Design (CAD) software assists users with design tasks such as creating, modifying and optimizing a design for a part or assembling parts. In some companies, the design of parts is done in a distributed manner. As a result, similar or even identical parts may be created for different products, but with the same functional performance. This not only creates more effort in design, but also causes a higher variety of parts, which leads to additional costs in service. In this context, the identification of components can solve the problem thanks to Artificial Intelligence.
The focus of component recognition is on common parts. Identical parts are components that can be used unchanged in different products but are not standard parts. Standard parts such as screws or pipes are usually not produced in-house, but are purchased from manufacturers who specialize in this area.
Identifying common parts on a database is done on the one hand to reduce the number of parts for service and maintenance, and on the other hand to suggest similar parts during design and thus reduce the effort and time of developing the common part.
Compare components: This algorithm can help
To be able to achieve this, the CAD equal parts must first be converted into a histogram or spectrogram. In these representations, the equal parts can be compared more easily. This is done using statistical classification methods such as the k-Nearest-Neighbor Algorithm (KNN) and the Euclidean Distance.
The k-Nearest-Neighbor Algorithm is considered the simplest and most proven algorithm when it comes to classifying data. The algorithm can be used to get a first overview of the data. Unlike other algorithms, it does not require extensive training, since all data is used in the classification. The algorithm finds similar data points (neighbors) for a given data point. This allows to determine to which class the point belongs. To compare the distance of the given data point with the similar data points the Euclidean Distance is used. This method is a distance function and can determine the distance between two points both on a plane and in a three-dimensional space, as in the comparison of equal parts. Thanks to the k-Nearest-Neighbor Algorithm and the Euclidean Distance, documents can be determined automatically, manufacturing processes can be predicted, or equal parts can be compared.
CLAAS relies on the identification of components
The internationally active agricultural machinery manufacturer CLAAS GmbH & Co. KGaA is testing a special use case for the integration of AI in computer-aided design in the AI Marketplace. Against the background of constantly increasing product complexity and number of variants, the reuse of components offers an important opportunity to save manufacturing, development and storage costs.
Identifying similar components should result in a shorter time-to-market by accelerating the design and release of components. In addition, lower costs can be achieved through larger order quantities. Identifying similar parts in a CAD database can also lead to the removal of existing parts to clean up the database.