Issue link: https://iconnect007.uberflip.com/i/1539960
64 SMT007 MAGAZINE I OCTOBER 2025 AI Use Cases Adopted by Industry Use Case#1: Incorrect components (or change in supplier) leading to failures Often, due to changes in the component supplier/ vendor, the component measurements show signif- icant deviations, sometimes leading to failures. This might also happen due to component quality varia- tions between production batches. These types of failures are very difficult for production engineers to identify and are a common reason behind failures. Sometimes, even zeroing in on the issue to compo- nent behavior takes hours of analysis from the test logs. This involves expensive manpower and critical production time. An AI-based anomaly detection model was devel- oped to identify anomalous measurement values from the normal point and send a real-time alert. The model can precisely point to the equipment, fixture, project, and down to the test name show- ing this abnormal component behavior. The model uses an unsupervised learning algorithm consist- ing of Time Series and Clustering models. It runs on real-time streaming data received from live produc- tion and compares it with learnings from historical data. One critical consideration in the model is to continuously learn from the measurement behavior. The next part of the model is the Alert Scoring Module, which scores each component anomaly and provides severity levels: Low, Medium, and High. The Alert Scoring Module uses a supervised neural network algorithm that runs on the compo- nent anomalies and assigns a score to the real-time data, considering past behavior and various other C h a r t s h ow i n g te st n a m e s h av i n g h i g h fa i l u re s . ▼ Fixture with issue C h a r t s h ow i n g m e a s u re m e nt b re a kd ow n by e q u i p m e nt a n d f i x t u re. A n o m a l i e s a n d fa i l u re s a re v i s i b l e o n t h e p ro b l e m at i c f i x t u re. ▼