Industrial-AI "Huashan Sword": WOD mDX Transformation Summit gathers PTC, Baosight, Phoenix Contact, Delta, Siemens, SUPCON, Emerson, HollySys & more

On June 3, under the theme "Industrial-AI Deep Scenario Landing — Path & Practice," the mDX Transformation Summit was held at SNIEC. CAICT, PTC, Baosight, Phoenix Contact, Delta, Siemens, SUPCON, Emerson, HollySys, LingCore, Nordrive and CIIC-Interlink — a dozen leading companies and authorities — gave a full-stack, replicable, scalable panorama of industrial-AI landing across data governance, agents, platform architecture, scenario practice, hardware base and value loop.

1. CAICT —全域 data governance builds the engineering foundation for AI

He Jun (CAICT) proposed an engineering method of domain data governance: a three-layer architecture (semantic, kinetic, dynamic ontologies) progressively turning unstructured data into machine-explainable, executable, even self-evolving knowledge assets, with an XBS multi-dimensional decomposition structure for rigorous, traceable modeling.

2. PTC — build an AI-native data system to empower manufacturing innovation

Dr. Shi Zhanbei (PTC Greater China CTO) noted China's AI adoption leads the world. PTC's intelligent PLM solution supports end-to-end AI via data graphing and ontological modeling — generative design, AI assistants and batch feature editing in CAD; 3D model de-duplication and cross-PLM/ERP change-impact analysis in PLM.

3. Baosight — "Skywalker" covers the full automation lifecycle

Xiao Luo introduced Baosight's industrial-AI agent "Skywalker (Tianxingzhe)": auto-generating technical schemes, then PLC code and automated tests from industry knowledge, with parameter optimization and a "cloud training – edge inference – device control" architecture pushing AI to the equipment.

4. Phoenix Contact — vPLCnext opens a real-time, open, intelligent industrial-AI paradigm

Fang Yiran showed vPLCnext: a Linux-container virtual-control platform compatible with IEC 61131-3 yet open to C++, Python and Matlab, fusing AI inference with real-time control on one hardware, with a built-in AI assistant generating engineering code and supporting full-chain predictive maintenance.

5. Delta — AI-driven smart-factory full-process practice

Dai Hongchun shared Delta's 55 years of manufacturing fused with AI, following "value-driven improvement" along "lean → automation → digitalization → intelligence," with digital-twin and big-data-AI across virtual workshops, parameter optimization and behavior monitoring.

6. OCTAVE (Intergraph) — new-quality smart manufacturing drives the smart workshop

Zhao Maolin argued future direction is intelligent control of workflows; the key is linking structured data from CAD/CAPP to MES and building a workshop-level decision platform for dynamic scheduling and smart decisions.

7. SynexiaAI — an ontology-native OS for industrial AI

Chen Kangping presented "SynexiaAI (Lingxi Shixing)," built on a 7-year IEC 61499 "Flex Control" system, proposing a five-dimension / 14-semantic-root kernel so AI understands business from the bottom rather than from massive training data.

8. Siemens — from automation to intelligence, reshaping industry

Luo Zihan presented Siemens' 50+ years in industrial AI: Flamingo (analytics & parameter optimization), the "AI Engineering Agent" generating PLC code and HMI via natural language, and a dedicated industrial foundation model for soft-sensing.

9. SUPCON — a time-series foundation model for the autonomous-run plant

Chen Yongchun introduced SUPCON's Time-series Pre-trained Transformer (TPT), a MoE model learning from industrial time-series to deliver anomaly warning, equipment health, parameter optimization and soft measurement, moving toward autonomous operation like autopilot.

10. Emerson — a cool look at industrial AI, aiming for closed-loop autonomous operation

Huang Xuefeng urged clear boundaries and "inheritable" behavior (same input → same output). He defined "autonomous operation" and called for rational views of "black-screen" and "lights-out" factories — the core is a "perceive–decide–control" autonomous loop, not full无人.

11. HollySys — XMagital platform, AI-enabled autonomous smart-factory practice

Ma Jun shared HollySys' XMagital: deep IT/OT fusion and a unified control-management platform; its electronics plant runs partial "lights-out" lines, with AI auto-programming spray equipment in 10 minutes vs. 1–2 days.

12. LingCore — AI digital employees fill discrete-manufacturing gaps

Xiao Mingyang presented execution-type AI agents that directly operate business systems — auto-handling email orders, syncing design changes and smart scheduling, sharply cutting error and receivable cycles.

13. Nordrive — AI industrial-PC hardware base for edge inspection

Liu Tao showed Nordrive's AI industrial-PC platform on Intel Core Ultra and Rockchip, with strong real-time and protocol compatibility for local edge inference in visual inspection and knowledge bases.

14. CIIC-Interlink — an industrial-agent OS accelerating full-scene landing

Liu Qing presented an industrial-agent OS integrating all elements from edge PLC to cloud ERP, supporting hybrid large/small-model scheduling and five agents (design, manage, control, coordinate, decide) — e.g. cutting new-product design from weeks to minutes.

From domain data governance to industrial agents, the June 3 mDX Summit pooled top industry wisdom, offering both inspiration for breaking data silos and a clear evolutionary direction for AI-enabled smart manufacturing.

2026 WOD Manufacturing Digitalization Expo (Shenzhen) · December 3–6, 2026