{"id":42642,"date":"2024-04-17T14:50:24","date_gmt":"2024-04-17T12:50:24","guid":{"rendered":"https:\/\/xitaso.com\/projects\/mawis-ki-lebensdauerprognose-elektronik\/"},"modified":"2025-08-12T13:33:11","modified_gmt":"2025-08-12T11:33:11","slug":"mawis-ai","status":"publish","type":"page","link":"https:\/\/xitaso.com\/en\/projects\/mawis-ai\/","title":{"rendered":"MaWis-AI: Lifetime prediction of high-end electronics"},"content":{"rendered":"

[vc_row css_animation=”” row_type=”row” use_row_as_full_screen_section=”no” type=”full_width” angled_section=”no” text_align=”left” background_image_as_pattern=”without_pattern” z_index=”” el_class=”header_level2_container”][vc_column][vc_row_inner row_type=”row” type=”full_width” text_align=”left” css_animation=”” css=”.vc_custom_1720518350293{background-image: url(https:\/\/xitaso.com\/wp-content\/uploads\/Header-Lebensdauerprognose-Elektronik.jpg?id=40696) !important;}”][vc_column_inner offset=”vc_hidden-xs”][vc_empty_space height=”500″][\/vc_column_inner][\/vc_row_inner][vc_row_inner row_type=”row” type=”full_width” text_align=”left” css_animation=””][vc_column_inner offset=”vc_hidden-lg vc_hidden-md vc_hidden-sm”][vc_single_image image=”38830″ img_size=”full” css=”” qode_css_animation=””][\/vc_column_inner][\/vc_row_inner][vc_row_inner row_type=”row” type=”grid” text_align=”left” css_animation=””][vc_column_inner][vc_raw_html css=””]JTNDZGl2JTIwY2xhc3MlM0QlMjJoZWFkZXJfbGV2ZWwyJTIyJTNFJTBBJTNDZGl2JTIwY2xhc3MlM0QlMjJ0b3BwZXIlMjIlM0VBSSUyMCUyNiUyME9wdGltaXphdGlvbiUzQyUyRmRpdiUzRSUwQSUzQ2gxJTNFTGlmZXRpbWUlMjBwcmVkaWN0aW9uJTIwb2YlMjBoaWdoLWVuZCUyMGVsZWN0cm9uaWNzJTNDJTJGaDElM0UlMEElM0MlMkZkaXYlM0UlMEElMEE=[\/vc_raw_html][\/vc_column_inner][\/vc_row_inner][\/vc_column][\/vc_row][vc_row css_animation=”” row_type=”row” use_row_as_full_screen_section=”no” type=”full_width” angled_section=”no” text_align=”left” background_image_as_pattern=”without_pattern” css=”.vc_custom_1720167204161{background-color: #e8e0e0 !important;}” z_index=””][vc_column][vc_row_inner row_type=”row” type=”grid” text_align=”left” css_animation=””][vc_column_inner width=”1\/2″][vc_empty_space height=”60″][vc_column_text]<\/p>\n

The challenge<\/h2>\n

Modern cars are becoming increasingly autonomous and rely on a large number of electronic assemblies such as sensors and control units. These must function reliably for many years under extreme conditions such as cold, snow, rain or heat. The quality of the solder connections plays a decisive role here, as they connect the components mechanically and electrically. Under thermal-mechanical stress, however, there is a risk of failure of the solder connections and therefore of the entire assembly.[\/vc_column_text][vc_empty_space height=”30″][vc_empty_space height=”30″ el_class=”ausgeblendet”][\/vc_column_inner][vc_column_inner width=”1\/2″][vc_empty_space height=”30″ el_class=”not-on-mobile”][vc_empty_space height=”30″][vc_column_text]<\/p>\n

How we will help<\/h2>\n

As part of the MaWis-AI research project, XITASO and its project partners are researching and developing innovative AI and deep learning approaches that can efficiently evaluate the suitability of solder materials for the respective use case in the early development phase of electrical assemblies. Ultrasound microscopy, X-ray images, thermomechanical analyses and FEM simulations are available as data sources. The goal of the research is to create highly reliable electronic assemblies with minimal time investment and lower costs.[\/vc_column_text][vc_empty_space height=”30″][\/vc_column_inner][\/vc_row_inner][vc_row_inner row_type=”row” type=”grid” text_align=”left” css_animation=””][vc_column_inner][vc_column_text]<\/p>\n

Technologies<\/h2>\n

[\/vc_column_text][vc_empty_space height=”15″][vc_raw_html]JTNDc3BhbiUyMGNsYXNzJTNEJTIydGVjaG5vbG9naWVuJTIyJTNFTWFjaGluZSUyMExlYXJuaW5nJTNDJTJGc3BhbiUzRSUwQSUzQ3NwYW4lMjBjbGFzcyUzRCUyMnRlY2hub2xvZ2llbiUyMiUzRVJlcHJlc2VudGF0aW9uJTIwTGVhcm5pbmclM0MlMkZzcGFuJTNFJTBBJTNDc3BhbiUyMGNsYXNzJTNEJTIydGVjaG5vbG9naWVuJTIyJTNFTmV1cmFsJTIwT0RFcyUzQyUyRnNwYW4lM0UlMEElM0NzcGFuJTIwY2xhc3MlM0QlMjJ0ZWNobm9sb2dpZW4lMjIlM0VUaW1lJTIwU2VyaWVzJTNDJTJGc3BhbiUzRQ==[\/vc_raw_html][vc_empty_space height=”60″][\/vc_column_inner][\/vc_row_inner][\/vc_column][\/vc_row][vc_row css_animation=”” row_type=”row” use_row_as_full_screen_section=”no” type=”grid” angled_section=”no” text_align=”left” background_image_as_pattern=”without_pattern” z_index=””][vc_column][vc_empty_space height=”30″ el_class=”not-on-mobile”][vc_empty_space height=”50″][\/vc_column][\/vc_row][vc_row css_animation=”” row_type=”row” use_row_as_full_screen_section=”no” type=”full_width” angled_section=”no” text_align=”left” background_image_as_pattern=”without_pattern” el_class=”two-column-with-heading-on-top” z_index=””][vc_column][vc_row_inner row_type=”row” type=”grid” text_align=”left” css_animation=””][vc_column_inner width=”1\/2″][vc_column_text]<\/p>\n

Deep learning and the challenge of data shortage<\/h2>\n

[\/vc_column_text][\/vc_column_inner][vc_column_inner width=”1\/2″][\/vc_column_inner][\/vc_row_inner][vc_row_inner row_type=”row” type=”grid” text_align=”left” css_animation=””][vc_column_inner width=”1\/2″][vc_column_text]Deep learning makes it possible to learn the representation of data in multiple abstraction layers and use it for much more than classification or regression.
\nThese methods have redefined the state of the art in many areas, such as computer vision or natural language processing. Deep learning, which is based on neural networks, requires a large amount of data as a basis for learning the representations, as the result of the parameters that can be learned is highly dimensional. However, building large data sets in materials science is difficult, as it requires a large number of samples and complex experiments. Learning precise and robust neural networks on the basis of small amounts of data therefore proves to be challenging.[\/vc_column_text][\/vc_column_inner][vc_column_inner width=”1\/2″][vc_column_text]Therefore, the goal of the MaWis-AI project is to develop, implement and test novel machine learning approaches for the lifetime prediction of solders made of tin, silver, copper and possibly other components (SAC or SAC+ solders). However, due to the small amount of data, standard deep learning methods cannot be used. Instead, model- and knowledge-based AI methods such as Neural Ordinary Differential Equations (Neural ODE) are being researched and their potential evaluated as part of the research project.[\/vc_column_text][\/vc_column_inner][\/vc_row_inner][\/vc_column][\/vc_row][vc_row css_animation=”” row_type=”row” use_row_as_full_screen_section=”no” type=”grid” angled_section=”no” text_align=”left” background_image_as_pattern=”without_pattern” z_index=””][vc_column][vc_empty_space height=”30″ el_class=”not-on-mobile”][vc_empty_space height=”30″][\/vc_column][\/vc_row][vc_row css_animation=”” row_type=”row” use_row_as_full_screen_section=”no” type=”full_width” angled_section=”no” text_align=”left” background_image_as_pattern=”without_pattern” el_class=”two-column-with-heading-on-top” z_index=””][vc_column][vc_row_inner row_type=”row” type=”grid” text_align=”left” css_animation=””][vc_column_inner width=”1\/2″][vc_column_text]<\/p>\n

Contrastive learning for the learning of good representations<\/h3>\n

[\/vc_column_text][vc_empty_space height=”15px”][\/vc_column_inner][vc_column_inner width=”1\/2″][\/vc_column_inner][\/vc_row_inner][vc_row_inner row_type=”row” type=”grid” text_align=”left” css_animation=””][vc_column_inner width=”1\/2″][vc_column_text]Contrastive learning is a machine learning method that aims to create useful representations of data. It works by creating pairs of data points that are either similar or different. These pairs are placed in an embedding space, with similar data points close to each other and different data points away from each other. Through this process, the model learns how to extract relevant characteristics that cause similar data points to be close to each other in a high-dimensional space and allows semantic information to be captured from the data. As part of the project, contrastive learning has proven to be useful for the targeted detection of cracks and defects in ultrasound microscopy images of solder connections.[\/vc_column_text][vc_empty_space height=”30px”][\/vc_column_inner][vc_column_inner width=”1\/2″][vc_single_image image=”42139″ img_size=”large” add_caption=”yes” alignment=”center” css=”.vc_custom_1711016581099{margin-bottom: 5px !important;}” qode_css_animation=””][vc_empty_space height=”15px”][vc_column_text]While classic supervised learning involves clustering according to component and solder types in the embedding space, contrastive learning arranges the data points according to the degree of damage in the embedding space. This results in relevant characteristics being extracted by the model. (Illustration from the publication “Contrastive pretraining of regression tasks in reliability forecasting of automotive electronics”)[\/vc_column_text][vc_empty_space height=”30px”][\/vc_column_inner][\/vc_row_inner][vc_empty_space height=”30″ el_class=”not-on-mobile”][\/vc_column][\/vc_row][vc_row css_animation=”” row_type=”row” use_row_as_full_screen_section=”no” type=”grid” angled_section=”no” text_align=”left” background_image_as_pattern=”without_pattern” z_index=””][vc_column][vc_empty_space height=”30″][vc_column_text]<\/p>\n

Publications on the research project<\/h2>\n

[\/vc_column_text][vc_empty_space height=”15″][vc_accordion active_tab=”1″ collapsible=”yes” style=”toggle”][vc_accordion_tab title=”2023″][vc_column_text css=””]<\/p>\n