»By focusing on the plausibility of the BOM, errors could be identified at an early stage and rectified before system integration. This saved unnecessary processes, accelerated the previously manual detection process and enabled quick clarification with the customer.«
R. Knezevic (Member of the board)
GRÜNINGER ELECTRONICS GmbH
Contact at the AI Innovation Center
Christof Nitsche
Potential for the use of AI in electronic manufacturing services (EMS)
Initial situation
Grüninger Electronics is a service provider for Electronic Manufacturing Services (EMS) from Weinstadt in Baden-Württemberg. The company offers solutions for complex electronic projects and supports its customers with flexible and competent services from the planning stage to the finished and tested assembly. As part of order processing, Grüninger Electronics receives the corresponding bills of material (BOM) from the electronics developers, which can include up to 400 items depending on the scope of the project. When creating these BOMs, careless mistakes or classic copy-paste problems can occur. Each individual item must therefore be carefully compared with the technical specifications of the respective manufacturer. These activities are associated with high quality requirements and currently involve considerable manual effort in order to ensure 1001TP3 data quality. This is done through multiple full checks and dual control checks to ensure that all information is transferred to production without errors. In future, the use of artificial intelligence will speed up the customer order process and make it more efficient.
Solution idea
In order to identify the potential of AI in the customer order process, a structured series of workshops was held as part of the AI Explorer. Based on the current tasks and challenges, possible use cases were developed, which were then jointly evaluated and prioritized. A detailed analysis of the cost-benefit ratio and the technical and organizational requirements for implementation was carried out for the three most promising use cases. The aim of the project was to develop a sound basis for decision-making and a project plan for the implementation of the top 3 use cases.
Implementation of the AI application
As part of the collaboration, a feasibility analysis was carried out for this use case in the form of an initial AI prototype to automatically identify inconsistencies in BOMs. Excel files can be loaded via a user interface and relevant rows and columns of the BOM can be selected. The data is transferred line by line to a large language model, which recognizes and justifies similar deviations based on sample errors. The results are color-coded and commented on in a copy of the file. The results are promising, but not yet reliable due to the limited training data.