TOPIC 1: Automation and optimisation of crimping tool designs

INITIAL SITUATION

A crimping process is used to connect cable ends with cable lugs. Currently, the design of crimping tools (matrices) for new cable lug geometries is done manually, based on documented guidelines and empirical values from in-house experts. After setting up sample tools and carrying out tests, a favourable design is confirmed or an iterative need for improvement of the tool geometries becomes apparent.

THE CHALLENGE

We are looking for a solution for the optimised and automated design of crimping tools. Existing guidelines are to be mapped and the previously documented die designs and associated results are to be taken into account, for example through machine learning approaches. The aim is to determine the optimal matrices design for a combination of cable type and cable lug.

THE TARGETED RESULT

During the runtime, an (AI-based) prototype algorithm for the design of crimping tools should be developed.

THE PERSPECTIVES

We support by providing hundreds of design data of existing tools of the past years, the related data and evaluation results of crimping connections, as well as (if required) access to MS Azure machine learning tools. Our process experts are also available for meetings and workshops. For the final test of the prototype algorithm, matrices can be made so that crimping tests and assessments of the pressings can take place.

Topic 2: Automated evaluation of crimping micrograph images

INITIAL SITUATION

In order to evaluate the quality of a crimp between cable lug and stripped cable end, cross-sectional and longitudinal micrographs are made. Currently, the micrographs are evaluated manually based on documented guidelines. The extracted quality parameters are manually compared with target values and a manual report is generated.

THE CHALLENGE

We are looking for a solution for the automated evaluation of micrographs of different crimping types. Existing documented guidelines are to be taken into account and an automated report is to be generated afterwards. The aim is to reduce the manual steps to a maximum. If necessary, certain quality features can also be recognised or derived directly from photos of the test specimens, even without a micrograph.

THE TARGETED RESULT

During the runtime, an (AI-based) image recognition and evaluation algorithm for extracting the relevant features including automatic reporting should be developed.

THE PERSPECTIVES

We support you by providing you with micrographs, the associated evaluation results and guidelines, as well as access to the MS Azure Machine Learning Tools (if required). Our process experts are also available for meetings and workshops.