en.Wedoany.com Reported - The Ministry of Industry and Information Technology (MIIT) and the National Data Administration (NDA) recently jointly issued the "Notice on the Joint Implementation of the 2026 'Model-Data Resonance' Action," establishing a national-level roadmap for the large-scale deployment of industrial AI. This initiative aims to close the "data-model-scenario" loop, driving intelligent manufacturing from scattered pilot projects to full-scale, nationwide implementation.
The core logic of "Model-Data Resonance" is to create a bidirectional cycle between data from the entire industrial production process and industrial AI large models: data feeds the models, models empower scenarios, and scenarios generate new data that, in turn, upgrades the models. Currently, the industrial AI transformation faces bottlenecks such as general models lacking understanding of industrial mechanisms, fragmented data across manufacturing enterprises, and a lack of industry standards. Insufficient supply of high-quality industrial data and the inability to deeply couple data with models are key sticking points. The policy innovation proposes a dual-wheel system of "industry-general + specialized knowledge datasets" to address these industrial shortcomings.
Based on over 40 years of manufacturing experience, COSMOPlat has taken early steps in the full-process governance of industrial data. Its Tianzhi industrial large model is entirely built on real production line data. Through comprehensive data cleaning, labeling, and structured governance, it transforms scattered factory data into standardized industrial assets. IDC has recognized COSMOPlat as a "Leader" in the industrial large model space, ranking it first in strategic dimensions within the industry.
COSMOPlat adopts a "general-purpose foundation + industry-specific model" approach, deeply adapting to the process mechanisms and business logic of different industries. In the energy and chemical sector, COSMOPlat partnered with Shaanxi Yanchang Petroleum to build the Yanchang Cloud Enjoy Industrial Internet Platform, creating the world's first petrochemical industry large model covering the entire "oil, gas, coal, chemical, and electricity" value chain. This has led to the development of 38 intelligent agents across 6 categories, including a dynamometer card intelligent assistant and an aldehyde separation tower process optimization agent. This solution has increased the timely fault detection rate by 90%, doubled quality control accuracy, improved the one-time success rate of gasoline blending by 60%, and reduced labor and energy costs by 10%.
In the home appliance sector, COSMOPlat empowered Haier's Shanghai Washing Machine Interconnected Factory to be recognized as a "Lighthouse Factory." To address the pain points of high energy consumption and slow machine adjustment in the injection molding process, the two parties jointly developed an injection molding process intelligent agent. Technicians can now retrieve hundreds of process parameters and command equipment adjustments using voice commands, reducing adjustment time from 120 minutes to 10 minutes. In green development, COSMOPlat created the Tianzhi Energy-Carbon Large Model, helping the steel enterprise under Xinjin Group complete an energy-saving renovation of its air compressor station, achieving an energy saving rate of over 30% and saving nearly 7 million yuan in annual electricity costs. Its AI intelligent control energy-saving strategy algorithm platform further improves the air compressor station's energy saving rate by 4% compared to traditional methods.
COSMOPlat's intelligent agents possess proactive sensing and autonomous decision-making capabilities. A key representative, the Quality 8D intelligent agent, can fully automatically monitor production quality anomalies, generating an analysis report in 1 minute, initiating a response in 1 hour, and completing problem rectification and containment within 1 day. It also automatically accumulates process experience and solutions, creating a positive cycle of knowledge reuse. In the standards domain, COSMOPlat, in collaboration with IEEE, initiated the world's first international standard for intelligent interaction engines and has led or participated in the development of several international standards for industrial large models and intelligent assembly.
Currently, COSMOPlat's Tianzhi industrial large model has been deployed across 9 major industries and 45 high-value scenarios, helping to build 17 global Lighthouse Factories. Economist Pan Helin pointed out that "Model-Data Resonance" still faces three challenges: enterprise data integration, industry model adaptation, and intelligent talent. In its cooperation with Yanchang Petroleum and others, COSMOPlat builds industry-specific large models by going to the site to sort out business processes and aggregate and clean data. Its self-developed intelligent agent development platform lowers the barrier to industrial AI applications, enabling business personnel to interact with intelligent agents using natural language.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
