Data intelligence at work

The ultrasonic welding quality inspection project is built on the TACO intelligent application development platform and focuses on weld-process monitoring and quality inspection. TACO provides unified data access, device communication, time-series data management, historical query and analysis, and algorithm-service integration. On top of this, the project adds real-time acquisition of welding parameters, process-curve monitoring, intelligent quality judgment, SPC analysis, historical traceability and early warning of production anomalies.

By closing the loop between equipment data, process analysis and quality management, the project makes welding production more visible, more stable and fully traceable — showing how quickly TACO can be put to work in industrial quality inspection and smart manufacturing.

Each weld generates about 30,000 high-frequency data points. The model judges every weld in a split second, and NG parts are diverted by the PLC. The system runs on SVOLT and CATL production lines.

Key capabilities

• Unified Data Access
• Device Communication
• Time-Series Database
• Machine Learning
• SPC Analysis
• Historical Traceability
• Anomaly Alerts
• PLC Integration

This project addresses outgoing quality management and digital collaboration on Certificates of Conformance (COC). Building on TACO’s unified data access, workflow orchestration, file storage and intelligent service integration, it delivers customer rule configuration, template management, quality data organization, COC generation, approval routing and result export — a complete loop from data to certificate delivery.

The project shows how quickly applications for manufacturing quality management can be developed on TACO, giving enterprises more efficient, standardized and flexible digital delivery.

Key capabilities

• Unified Data Access
• Workflow Orchestration
• File Storage
• Customer Rules
• Template Management
• COC Generation
• Approval Workflow
• Result Export

The FDC application is built on the TACO platform. It collects sensor data from every tool in the semiconductor manufacturing process, stores and analyses large volumes of production data in real time, and monitors the whole production process. By comparing real-time data with historical data and applying statistical rules, it gives early warning of potential production anomalies.

FDC means many sensors, very large data volumes and strict real-time requirements. TACO’s time-series database and Lambda functions solve the problems of writing massive data instantly and analysing it for real-time alerts.

Key capabilities

• Sensor Data Acquisition
• Time-Series Database
• Lambda Functions
• Real-Time Analysis
• Historical Comparison
• Statistical Rules
• Early Warning

Parsing engine. TACO works as the parsing engine, handling many heterogeneous wafer image formats and converting them into one standard data specification so that data from multiple sources flows in seamlessly.

Intelligent root-cause attribution. Recognition algorithms designed on TACO perform deep pattern recognition and defect attribution on wafer maps, pinpointing the root cause of production anomalies.

Image review. A wafer image database stores historical wafer maps and inspection results in one place, ready to recall at any time as visual evidence for process optimization.

Key capabilities

• Multi-Format Parsing
• Data Standardization
• Wafer Map Pattern Recognition
• Defect Attribution
• Image Database
• Process Optimization