EN 15085 rail-vehicle welding: how AI process recognition makes weld-by-weld traceability real

EN 15085 is the standard dedicated to welding rail vehicles and their components. Compared with general structural work, it sets a higher bar for safety-relevant welds: graded control by safety level, with weld-by-weld traceability, along with the corresponding personnel qualifications and quality system.

It is built on the quality requirements of ISO 3834, and the difficulty again centers on records and traceability — every safety weld must be traceable to its procedure, operator and parameters, and remain retrievable over the entire life of the vehicle. Relying on manual work and paper makes completeness hard to guarantee, and retrieval years later harder still.

A workable approach is to build a born-digital weld-by-weld digital record and traceability for each weld, organized and searchable by safety level, to match EN 15085’s demand for high traceability. It should be clear that these records support traceability itself; they do not replace welder qualification or third-party certification.

In a setting like rail vehicles — multiple procedures, multiple safety levels, and an out-of-order welding sequence — making weld-by-weld traceability catch every single weld is all but impossible with manual classification. This is exactly where AI comes into its own. Today’s welding AI can already combine scheduling information with the real-time execution data of each weld (current, voltage and gas-flow waveforms) to perform process recognition and classification automatically: determining which class of procedure each weld belongs to, which safety level and WPS it corresponds to, and then filing and checking it accordingly.

As a result, quality-compliance records and traceability can be covered systematically by AI: each safety weld is correctly classified, logged and checked as it is being made — weld by weld, level by level, rather than pieced together from spot checks after the fact. The high traceability EN 15085 requires thus shifts from “assured by people” to “assured by data and AI” — every weld auditable and provable.

For the rail supply chain, digital records that are weld-by-weld and retrievable over the long term are both a precondition for passing audits and a basis for reducing later liability and recall risk.

Suxin has long focused on smart welding, with 70+ invention patents in welding process recognition and weld traceability and a role in drafting two national standards. If your company welds rail vehicles and must meet EN 15085’s weld-by-weld traceability, you can book a traceability demo.