新一代「文鳐」海洋原生多模态大模型在厦门发布,AI 轮机长完成实船验证Next-Generation Wenyao Ocean-Native Multimodal Model Launched in Xiamen; AI Chief Engineer Passes Shipboard Validation


AI free token/AI免费积分

新一代「文鳐」海洋原生多模态大模型在厦门发布,AI 轮机长完成实船验证

新一代“文鳐”海洋原生多模态大模型在厦门发布赋能海洋装备自主智能化发展

2026-11-22 16:08 发布于 北京 人民网资讯精选官方账号

11月22日,新一代“文鳐”大模型在厦门正式发布。该模型由众数文鳐联合哈尔滨工程大学、厦门理工学院等科研力量协同攻关,是海洋原生 多模态大模型 领域的一项重要成果,标志着我国在海洋装备智能化领域迈出关键一步,为海洋经济向更远、更深海域拓展提供了坚实的技术底座。

当前,海上风电、深远海养殖、海底巡检等产业向远海拓展,离岸装备常面临海况多变、故障突发、通信受限等难题,岸基指挥难以快速响应现场变化,行业亟需装备具备自主分析、任务调整、风险处置能力。

深远海装备实现高水平自主作业,必须打通感知、决策、执行全链条。新一代“文鳐”大模型正是适配复杂海洋任务的核心大脑。相较于上一代版本,模型融入海洋环境与装备运行规律,可综合解析多类装备数据,动态规划、调整作业任务,能力由辅助人工研判升级至支撑装备自主决策,相关功能已进入场景验证阶段。

项目核心技术攻关由众数文鳐联合哈尔滨工程大学智能科学与工程学院教授蔡成涛科研团队、厦门理工学院共同开展,将海洋装备运行机理用于模型对专业信号的理解。依托 哈尔滨工程大学 在海洋装备领域的科研积淀,模型能够结合工况条件与工程知识,输出可溯源的研判结论,为装备自主感知、智能任务规划夯实基础。

研发团队直击远海通信弱、工况复杂、人工处置滞后等现实难题,推动海洋装备行业由传统远程监视模式,迈向现场自主研判、辅助处置。目前,基于“文鳐”大模型开发的船舶设备健康管理应用已在营运船舶上开展实船验证,诞生了7×24小时在线值守的“AI轮机长”。该智能体可持续监测已接入的核心设备状态,发现异常趋势后,联查相关设备和测点,查阅工程图纸与设备说明书,并将研判依据和优先排查建议推送给轮机长。在船舶运维管理中,系统可结合维护计划、工单、备件库存给出运维建议。岸基管理人员通过统一平台可集中查看船队设备状态和待处理事项。船员实操经验和现场处置结果经授权、脱敏及专家审核后,可纳入行业知识库,为后续模型评测和迭代提供参考。

立足船舶场景已经跑通的实战成果,研发团队将快速把“文鳐”大模型能力拓展至海上风电、海洋牧场等核心深远海赛道,落地设备监测、风险研判、智能运维系列能力。三方将持续释放海洋原生多模态大模型的底座价值,面向各类海洋作业装备输出标准化自主智能能力。在严守安全底线、以人为监督核心的前提下,推动一大批海洋装备解锁自主作业新范式,打破深远海智能化发展技术桎梏,为我国经略海洋、发展深远海蓝色经济,提供国产自主可控的大模型核心力量。(文/图 林林 吴碧琳)

本文来源:https://news.qq.com/rain/a/20261122A08NC500
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。

Next-Generation Wenyao Ocean-Native Multimodal Model Launched in Xiamen; AI Chief Engineer Passes Shipboard Validation

The new-generation "Wenyao" ocean-native multimodal large model is released in Xiamen, empowering the autonomous and intelligent development of marine equipment.

2026-11-22 16:08, posted in Beijing by the People's Daily Online Information Selection official account.

On November 22, the new-generation "Wenyao" large model was officially released in Xiamen. Developed through joint research by Zhongshu Wenyao together with Harbin Engineering University, Xiamen University of Technology, and other research forces, the model is an important achievement in the field of ocean-native multimodal large models, marking a key step forward for China in the intelligentization of marine equipment and providing a solid technological foundation for expanding the marine economy into farther and deeper waters.

Currently, as industries such as offshore wind power, deep-sea aquaculture, and subsea inspection expand into far seas, offshore equipment often faces challenges such as changeable sea conditions, sudden failures, and limited communications, making it hard for shore-based command to respond quickly to on-site changes. The industry urgently needs equipment with capabilities for autonomous analysis, task adjustment, and risk handling.

To achieve a high level of autonomous operation, deep-sea equipment must connect the entire chain of perception, decision-making, and execution. The new-generation "Wenyao" large model is precisely the core brain suited to complex marine tasks. Compared with the previous version, the model incorporates the laws of the marine environment and equipment operation, can comprehensively analyze multiple types of equipment data, and dynamically plan and adjust work tasks. Its capability has been upgraded from assisting human judgment to supporting autonomous decision-making by equipment, and the relevant functions have entered the stage of scenario verification.

The project's core technical research was jointly carried out by Zhongshu Wenyao together with the research team of Professor Cai Chengtao of the School of Intelligent Science and Engineering at Harbin Engineering University, and Xiamen University of Technology, applying the operating mechanisms of marine equipment to the model's understanding of professional signals. Drawing on Harbin Engineering University's research accumulation in the field of marine equipment, the model can combine working conditions with engineering knowledge to output traceable judgments and conclusions, laying a solid foundation for autonomous perception and intelligent task planning of equipment.

The R&D team directly tackles real-world challenges such as weak communications in far seas, complex working conditions, and delayed manual handling, pushing the marine equipment industry from the traditional remote-monitoring model toward on-site autonomous judgment and assisted handling. At present, a ship equipment health-management application developed on the basis of the "Wenyao" large model has been verified on operating vessels, giving birth to an "AI Chief Engineer" on duty 7×24 hours online. This agent can continuously monitor the status of connected core equipment; upon detecting abnormal trends, it cross-checks related equipment and measurement points, consults engineering drawings and equipment manuals, and pushes the basis for its judgment and prioritized troubleshooting suggestions to the chief engineer. In ship operation and maintenance management, the system can give O&M recommendations based on maintenance plans, work orders, and spare-parts inventory. Shore-based managers can centrally view the fleet's equipment status and pending items through a unified platform. Crew members' hands-on experience and on-site handling results, after authorization, desensitization, and expert review, can be incorporated into the industry knowledge base to provide references for subsequent model evaluation and iteration.

Building on the combat-tested results already achieved in ship scenarios, the R&D team will rapidly extend the "Wenyao" large model's capabilities to core deep-sea tracks such as offshore wind power and marine ranches, deploying a series of capabilities in equipment monitoring, risk judgment, and intelligent O&M. The three parties will continue to unleash the foundational value of the ocean-native multimodal large model, outputting standardized autonomous intelligence capabilities for all types of marine operating equipment. Under the premise of strictly guarding the safety bottom line and keeping human supervision at the core, they will help a large number of marine equipment unlock new paradigms of autonomous operation, break the technological shackles of deep-sea intelligent development, and provide a domestically developed, autonomous, and controllable core large-model force for China's maritime governance and the development of the deep-sea blue economy. (Text/photos by Lin Lin and Wu Bilin)

本文来源:https://news.qq.com/rain/a/20261122A08NC500
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry