《電子技術(shù)應(yīng)用》
您所在的位置:首頁 > 人工智能 > 設(shè)計應(yīng)用 > 工業(yè)大模型賦能制造業(yè)數(shù)字化轉(zhuǎn)型的路徑與對策
工業(yè)大模型賦能制造業(yè)數(shù)字化轉(zhuǎn)型的路徑與對策
網(wǎng)絡(luò)安全與數(shù)據(jù)治理
秦崢1,李育濤2,郭淑芳1
1.國家工業(yè)信息安全發(fā)展研究中心;2.中國科學(xué)院大學(xué)
摘要: 在全球制造業(yè)加速邁向數(shù)字化、智能化的背景下,工業(yè)大模型作為新一代智能技術(shù),正成為推動制造業(yè)數(shù)字化轉(zhuǎn)型的重要引擎。通過系統(tǒng)梳理工業(yè)大模型的概念、發(fā)展脈絡(luò)和發(fā)展現(xiàn)狀等基礎(chǔ)理論,提出工業(yè)大模型賦能制造業(yè)數(shù)字化轉(zhuǎn)型的理論框架,并詳細(xì)闡述工業(yè)大模型在研發(fā)設(shè)計、生產(chǎn)制造、運(yùn)維服務(wù)、經(jīng)營管理和供應(yīng)鏈管理等制造業(yè)典型應(yīng)用場景的賦能作用。針對工業(yè)大模型在深度應(yīng)用過程中所面臨的高質(zhì)量訓(xùn)練數(shù)據(jù)匱乏、工業(yè)場景分布碎片化、工業(yè)應(yīng)用魯棒性欠缺、關(guān)鍵場景風(fēng)險需警惕和計算與系統(tǒng)能力不足等挑戰(zhàn),進(jìn)一步探討其賦能制造業(yè)數(shù)字化轉(zhuǎn)型的方法路徑,并從政策機(jī)制、示范引領(lǐng)、標(biāo)準(zhǔn)體系、自主創(chuàng)新、安全韌性和人才培養(yǎng)等多個維度提出對策建議,以期為工業(yè)大模型驅(qū)動制造業(yè)高質(zhì)量發(fā)展提供有價值的參考和啟示。
中圖分類號:TP391.9文獻(xiàn)標(biāo)識碼:ADOI:10.19358/j.issn.2097-1788.2025.07.006
引用格式:秦崢,李育濤,郭淑芳. 工業(yè)大模型賦能制造業(yè)數(shù)字化轉(zhuǎn)型的路徑與對策[J].網(wǎng)絡(luò)安全與數(shù)據(jù)治理,2025,44(7):36-42.
The path and countermeasures of empowering manufacturing digital transformation with industrial large models
Qin Zheng1, Li Yutao2, Guo Shufang1
1.China Industrial Control Systems Cyber Emergency Response Team;2.University of Chinese Academy of Sciences
Abstract: Against the backdrop of the global manufacturing industry accelerating towards digitization and intelligence, industrial large models, as a new generation of intelligent technology, are becoming an important engine for promoting the digital transformation of the manufacturing industry. The concept, development context, and current status of the industrial large model are systematically reviewed, and a theoretical framework for empowering the digital transformation of the manufacturing industry with the industrial large model is proposed. The empowering role of the industrial large model in typical application scenarios of the manufacturing industry, such as research and development design, production and manufacturing, operation and maintenance services, business management, and supply chain management, is elaborated in detail. In response to the challenges faced by industrial large models in the process of deep application, such as the lack of high-quality training data, fragmented distribution of industrial scenarios, lack of robustness in industrial applications, vigilance against key scenario risks, and insufficient computing and system capabilities, this paper further explores the methods and paths to empower the digital transformation of the manufacturing industry, and proposes countermeasures and suggestions from multiple dimensions such as policy mechanisms, demonstration guidance, standard systems, independent innovation, safety resilience, and talent cultivation, in order to provide valuable reference and inspiration for industrial large models to drive the high-quality development of the manufacturing industry.
Key words : industrial large models; manufacturing; digital transformation; artificial intelligence

引言

當(dāng)前,全球制造業(yè)正經(jīng)歷深刻的數(shù)字化與智能化變革[1],工業(yè)大模型作為人工智能技術(shù)與制造業(yè)深度融合的核心技術(shù),正在以前所未有的深度和廣度重塑產(chǎn)業(yè)鏈、供應(yīng)鏈及價值鏈體系[2]。依托深度學(xué)習(xí)、自然語言處理與多模態(tài)感知等前沿技術(shù),工業(yè)大模型具備強(qiáng)大的數(shù)據(jù)處理、知識推理與智能決策能力,能夠打破數(shù)據(jù)孤島,構(gòu)建跨企業(yè)、跨環(huán)節(jié)、跨場景的智能協(xié)同體系。工業(yè)大模型強(qiáng)大的泛化能力與自主學(xué)習(xí)特性,顯著提升制造企業(yè)的生產(chǎn)效率與資源配置效率,助力其實現(xiàn)高效化、柔性化轉(zhuǎn)型,增強(qiáng)整體競爭力[3]。

盡管工業(yè)大模型在制造業(yè)的應(yīng)用前景廣闊,但其發(fā)展仍面臨諸多挑戰(zhàn)[4-5]。制造業(yè)數(shù)據(jù)高度異構(gòu)、復(fù)雜多樣,實現(xiàn)高效整合與智能應(yīng)用仍存在技術(shù)瓶頸。模型的可解釋性、安全性和適用性亦亟待提升,以確保其在實際生產(chǎn)環(huán)境中的穩(wěn)定性與可控性。同時,從政策和產(chǎn)業(yè)生態(tài)的角度來看,當(dāng)前工業(yè)大模型的標(biāo)準(zhǔn)體系不完善、產(chǎn)業(yè)鏈協(xié)同機(jī)制不健全,也制約了其規(guī)?;茝V[6-8]。

基于此,本文聚焦工業(yè)大模型賦能制造業(yè)數(shù)字化轉(zhuǎn)型的路徑,構(gòu)建理論分析框架,系統(tǒng)梳理其典型應(yīng)用場景與關(guān)鍵痛點(diǎn)問題,深入探討賦能機(jī)制與落地路徑,提出推動其規(guī)?;瘧?yīng)用的政策與技術(shù)舉措。旨在為工業(yè)大模型在制造業(yè)中的有效落地提供理論支撐與實踐參考,助力制造業(yè)向智能化、綠色化、高質(zhì)量方向發(fā)展。


本文詳細(xì)內(nèi)容請下載:

http://ihrv.cn/resource/share/2000006612


作者信息:

秦崢1,李育濤2,郭淑芳1

(1.國家工業(yè)信息安全發(fā)展研究中心,北京100040;2.中國科學(xué)院大學(xué),北京100049)


Magazine.Subscription.jpg

此內(nèi)容為AET網(wǎng)站原創(chuàng),未經(jīng)授權(quán)禁止轉(zhuǎn)載。