The textile industry is undergoing a qualitative shift from 'having data' to 'using data.' On July 27, a product launch in Keqiao, Shaoxing, sent a key signal: Zhongke Wenge's 'Industrial Decision Brain' has entered core production processes of textile enterprises, demonstrating efficiency gains at the minute or even second level in order processing, production scheduling, and color matching. This is no longer a lab concept but a real force reshaping factory decision-making.
Background
Industry authorities including Yin Qiang, Director of the Informatization Department of the China National Textile and Apparel Council, and Zeng Dajun, Deputy Director of the Institute of Automation, Chinese Academy of Sciences, attended the AI-empowered Textile Industry High-Quality Development Conference. Zeng pointed out in his report that decision-making subjects are shifting from human-centered to 'human + agent' synergy, posing new challenges to existing industrial systems. Professor Liu Yuan from Zhejiang University emphasized that AI's technical capabilities do not naturally translate into industrial empowerment; they must be deeply integrated with specific industry scenarios. These insights highlight the core bottleneck of textile industry intelligence: data is piling up, but a 'decision brain' connecting data to execution is lacking.
Zhongke Wenge CTO Cao Jia stated bluntly at the launch that factories do not lack equipment, systems, or data; what they truly lack is a decision-intelligent brain. Over the years, ERP, MES, and PLC systems have been widely deployed, accumulating rich data, but most of this data remains at the recording and querying level, failing to transform into capabilities for production optimization and decision support. Industry public data shows that the utilization rate of data in most textile enterprises is less than 30%, making this the 'last mile' obstacle in intelligent transformation.
Industry Impact
The 'Industrial Decision Brain' is built on the DOMA technical architecture, connecting data, models, agents, and equipment across the entire chain. It does not require enterprises to dismantle existing systems but instead links order, equipment, material, and process data into a unified business language. This design lowers the implementation threshold, making it especially suitable for small and medium-sized textile enterprises with varying levels of informatization.
Results are significant. In trade order processing, an order that previously required 30-60 minutes of manual cross-checking is now completed in tens of seconds, with key element extraction accuracy exceeding 95%. Weaving scheduling has been compressed from hours to minutes, enabling quick response to ad-hoc orders and equipment anomalies. In printing and dyeing color matching, the number of sample rounds has dropped from 7-8 to 3-4, saving 30%-50% in costs while solidifying veteran expertise into reusable capabilities. Dyeing vat scheduling time has been cut from about 2 hours to about 2 minutes, and cross-departmental collaboration for completion feedback and order delivery has been reduced from 1.5-3 hours to about 5 minutes. These figures mean textile enterprises can not only dramatically improve efficiency but also significantly reduce reliance on manual experience, alleviating the industry's pain points of recruitment and retention difficulties.
Notably, this system also closes the loop from digital decision-making to embodied execution. Robots for intelligent inspection, yarn hanging, and cloth dropping, developed by Zhongke Juxi (a Zhongke Wenge subsidiary), are coordinated by the 'Industrial Decision Brain' to translate system decisions into production actions. The inspection robot can identify over 20 types of textile defects with an accuracy exceeding 95% and a single-charge endurance of 8 hours. The yarn-hanging robot achieves a repeat positioning accuracy of ±0.1 mm, and the cloth-dropping robot has a rated load of 300 kg. This means AI can not only analyze problems in the digital world but also directly drive physical equipment to execute tasks, forming a complete 'perception-decision-execution-feedback' loop.