Causal AI
Understanding cause-and-effect relationships with causal AI
In SMD manufacturing, millions of data points are generated in every shift and stored for traceability purposes. But there is more that can be gained from this data:
- Analyses of error rates in production and how to reduce them
- The influence of PCB design on production and possible optimizations
- Correlations and dependencies between individual process parameters and countermeasures
However, data-driven problem analysis along the production line often requires a high level of manual data engineering effort because many different data sources, e.g., from printers, SPI, placement machines, oven profilers, and AOI, must first be merged.
Since millions of data points are distributed across various systems, conventional tools quickly reach their limits. But only when the bulk of production information is consolidated can artificial intelligence derive causal relationships from it and thus uncover optimization potential.
This is exactly where Xplain Data's causal AI comes in: Xplain Data's CausalDiscoverer, an AI tool optimized for processing large amounts of data, analyzes all data from the SMD line and often reveals unexpected cause-and-effect relationships. Gain a new perspective on your manufacturing and discover new, data-driven potential for optimizing your processes.
The advantages of causal AI
Comprehensive analysis
All factors along the SMD line that influence the process in terms of quality and design are included in the analysis
Greater understanding of processes
The analysis of causal AI provides a new perspective on the process and its cause-and-effect relationships, thereby creating a new, deeper understanding of the SMD process.
Reduction in errors and costs
The cause-and-effect relationships discovered by AI enable the process to be optimized even more precisely and sustainably, thereby reducing production errors and costs.
Causal quality control in 3 steps
Xplain Data's patented ObjectAnalytics technology follows an object-centric approach, in which all relevant data is consolidated in a central object model. A technologist or SMD manager can now create graphical evaluations with just a few mouse clicks, enabling them to identify the root causes of any problems that arise.
1. Integration of production data
In the first step, all production data from all systems is connected to the ObjectAnalytics database.
2. Evaluate causal analyses
The technology analyzes the SMT data and presents causal patterns that have led to quality deviations.
3. Implementation of a bot
In the future, a bot will monitor in the background and check for potential causes of errors across all production steps.
Complex evaluation, simple integration
Whether via Python, JavaScript, or WebAPI, Xplain Data's flexible interfaces enable dashboards to be queried in almost all existing analysis workflows, quality systems, or web applications.
What problems could comprehensive data analysis
reveal in SMD manufacturing?
Here are a few examples of causal relationships that AI could uncover: Slow wear and tear, lack of maintenance, problems in processes, or environmental influences are reflected in the data long before they are detected by a human.
- More errors on hot days: Correlations between environmental conditions (temperature, humidity) and error frequency
- High downtime due to feeder changes or material replenishment: Detectable through OEE analyses and machine log data
- Pick-and-place errors due to vacuum problems: Patterns in machine data (vacuum time series, deviations in placement times)
- Wear and tear on the stencil: Increasing error frequency after certain production cycles
About Xplain Data
Founded in 2015, Xplain Data specializes in causal analysis of complex data. Originally used in healthcare to identify the causal causes of health problems by examining all patient data, the company now applies its methodology to industrial production. The company is headquartered in Zorneding near Munich.
