摘要
针对传统人工设计实验方案的模式难以适应智能设备快速迭代需求的现状,提出了一种基于知识图谱与反馈的实验方案智能生成方法。该方法通过构建“知识建模-需求解析-方案生成-反馈优化”的闭环系统,可提升实验方案设计效率。该方法具体实现包含四个关键环节:首先基于实验领域的语料构建实验领域知识图谱,实现知识的结构化表征;接着,通过提取模型获取需求文档的关键要素,并激活图谱中相关知识形成动态背景知识空间;随后在知识约束下生成候选实验方案;最终引入人机协同反馈机制,通过优化策略实现背景知识的动态更新与方案持续优化。通过构建的实验领域数据集进行相关实验,实验结果证明了融合知识与反馈信息有助于提升实验方案的质量。为验证方法可应用性,研发了原型系统平台并开展案例验证,结果表明该方法可有效地提升实验方案设计的智能化水平和迭代效率。
To overcome the limitations of traditional manual trial design methods in adapting to the rapid iteration demands of intelligent equipment,this paper proposes a knowledge graph and feed-back-enhanced framework for automated test scheme generation.The framework establishes a closed-loop system comprising"knowledge modeling,requirement parsing,scheme generation,and feedback optimization"to enhance design efficiency.Its implementation involves four key phases:First,a domain-specific knowledge graph is constructed using trial-related corpora to achieve structured representation of expertise.Second,key elements from requirement documents are extracted through NLP-based extraction models,activating relevant knowledge nodes to form a dynamic contextual knowledge space.Third,candidate test schemes are generated under knowledge constraints through intelligent algorithms.Finally,a human-machine collaborative feedback mechanism is introduced,where optimization strategies dynamically update background knowledge and refine schemes iteratively.Relevant experiments were conducted on the test scheme domain dataset,and the results demonstrate that the integration of knowledge and feedback information contributes significantly to enhancing the quality of the test scheme.To validate the applicability of the proposed framework,a prototype system was developed and evaluated through industrial case studies.Results demonstrate that the framework significantly improves the intelligence level and iteration efficiency of test scheme design.
作者
李文璋
郭海辰
李学恩
Wenzhang Li;Haichen Guo;Xueen Li(Unit 63891 of PLA,LuoYang,Henan 471000;School of Computer And Artificial Intelligence,Beijing Technology and Business University,Beijing 100048;Institute of Automation,Chinese Academy of Sciences,Beijing 100190)
关键词
实验领域知识图谱
知识图谱增强生成
反馈信息收集
优化实验方案生成
原型系统平台
domain-specific knowledge graph
knowledge graph-enhanced generation
feedback information collection
refine test scheme generation
prototype system
