基于双通道检索增强的农业病虫害Agent智能诊断系统

doi:10.14088/j.cnki.issn0439-8114.2026.06.033

摘要/Abstract

摘要： 针对现有通用大语言模型在农业病虫害诊断中存在的专业知识约束不足、长链条推理能力弱以及可能产生事实性幻觉等问题,提出一种融合知识图谱与大语言模型的Agent双通道智能诊断系统。在数据层,设计三元组感知高效去噪机制（TAED）,剔除低质量冗余数据,构建用于领域微调语料库。在模型层,提出阶段自适应思维链（PA-CoT）策略,通过指令微调使模型习得诊断逻辑,降低逻辑跳跃带来的幻觉。在架构层,构建基于意图识别的Agent双通道检索机制,实现非结构化语义信息与结构化知识的互补增强。结果表明,该系统在复杂农业诊断任务中的准确率达82.4%,幻觉率为10.8%,在推理完整性与知识准确性上优于传统检索增强（RAG）方法,可为农业智能专家系统的落地应用提供解决方案。

关键词: 农业病虫害诊断, 大语言模型, 知识图谱, 双通道检索

Abstract: To address the issues of insufficient professional knowledge constraints, limited long-chain reasoning capabilities, and possible factual hallucinations in existing general large language models for agricultural pest and disease diagnosis, this study proposed an Agent-based dual-channel intelligent diagnostic system integrating knowledge graphs and large language models. At the data level, a Triple-Aware Efficient Denoising (TAED) mechanism was designed to eliminate low-quality redundant data and construct a corpus for domain-specific fine-tuning. At the model level, a Phased-Adaptive Chain-of-Thought (PA-CoT) strategy was proposed. Through instruction fine-tuning, the model learned diagnostic logic and reduced the risk of hallucinations caused by logical jumps. At the architecture level, an intent recognition-based Agent dual-channel retrieval mechanism was constructed to achieve the complementary enhancement of unstructured semantic information and structured knowledge. Experimental results demonstrated that the proposed system achieved an accuracy of 82.4% and a hallucination rate of 10.8% in complex agricultural diagnostic tasks. It outperformed traditional Retrieval-Augmented Generation (RAG) methods in terms of reasoning completeness and knowledge accuracy, providing a solution for the practical application of agricultural intelligent expert systems.

Key words: agricultural pest and disease diagnosis, large language models, knowledge graph, dual-channel retrieval

中图分类号:

TP391
S43

刘疆泉, 李永可. 基于双通道检索增强的农业病虫害Agent智能诊断系统[J]. 湖北农业科学, 2026, 65(6): 213-220.

LIU Jiang-quan, LI Yong-ke. Dual-channel retrieval-augmented Agent system for intelligent diagnosis of agricultural pests and diseases[J]. HUBEI AGRICULTURAL SCIENCES, 2026, 65(6): 213-220.

参考文献

[1] 翟肇裕,曹益飞,徐焕良,等.农作物病虫害识别关键技术研究综述[J].农业机械学报,2021,52(7):1-18.
[2] BROWN T B, MANN B, RYDER N, et al. Language models are few-shot learners[EB/OL].2020: arXiv: 2005.14165. https://arxiv.org/abs/2005.14165.
[3] LIN S, HILTON J, EVANS O.TruthfulQA: Measuring how models mimic human falsehoods[A]. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics[C]. Dublin, Ireland: Association for Computational Linguistics,2022.3214-3252.
[4] LINDERS J, TOMCZAK J M.Knowledge graph-extended retrieval augmented generation for question answering[J]. Applied intelligence, 2025, 55(17): 1102.
[5] ZHU X R, XIE Y X, LIU Y, et al.Knowledge graph-guided retrieval augmented generation[A]. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies[C]. Albuquerque, New Mexico: Association for Computational Linguistics, 2025.8912-8924.
[6] ZHAO X Y, CHEN B Y, JI M X, et al.Implementation of large language models and agricultural knowledge graphs for efficient plant disease detection[J]. Agriculture, 2024, 14(8): 1359.
[7] 张航,盛志伟,张仕斌,等.Simhash算法在文本去重中的应用[J].计算机工程与应用,2020,56(11):246-251.
[8] 袁乐,刘绍华,王禹,等.大语言模型检索增强生成优化技术研究综述[J].计算机学报,2026,49(2):383-422.
[9] ZHAO Q F, WANG R B, CEN Y K, et al.LongRAG: A dual-perspective retrieval-augmented generation paradigm for long-context question answering[A]. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing[C]. Miami, Florida, USA: Association for Computational Linguistics, 2024.22600-22632.
[10] HUANG J, WANG M, CUI Y P, et al.Layered query retrieval: An adaptive framework for retrieval-augmented generation in complex question answering for large language models[J]. Applied sciences, 2024, 14(23): 11014.
[11] BOSMA M, CHI E, ICHTER B, et al.Chain-of-thought prompting elicits reasoning in large language models[A]. Advances in Neural Information Processing Systems 35[C]. New Orleans, Louisiana, USA: Neural Information Processing Systems Foundation, Inc.(NeurIPS), 2022.24824-24837.
[12] MAVROMATIS C, KARYPIS G. GNN-RAG: Graph neural retrieval for large language model reasoning[EB/OL].2024: arXiv: 2405.20139. https://arxiv.org/abs/2405.20139.
[13] 李玫,高庆,马森,等.面向代码相似性检测的相似哈希改进方法[J].软件学报,2021,32(7):2242-2259.
[14] QAISER S, ALI R.Text mining: Use of TF-IDF to examine the relevance of words to documents[J]. International journal of computer applications, 2018, 181(1): 25-29.
[15] HOGAN A, BLOMQVIST E, COCHEZ M, et al.Knowledge graphs[J]. ACM computing surveys, 2022, 54(4): 1-37.
[16] KAMILARIS A, KARTAKOULLIS A, PRENAFETA-BOLDÚ F X. A review on the practice of big data analysis in agriculture[J]. Computers and electronics in agriculture, 2017, 143: 23-37.
[17] YANG W X, YANG S, WANG G P, et al.Knowledge graph construction and representation method for potato diseases and pests[J]. Agronomy, 2024, 14(1): 90.
[18] REIMERS N, GUREVYCH I.Sentence-BERT: Sentence embeddings using Siamese BERT-networks[A]. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)[C]. Hong Kong, China Association for computational Linguistics: ACL, 2019.3980-3990.
[19] KARPUKHIN V, OGUZ B, MIN S, et al.Dense passage retrieval for open-domain question answering[A]. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)[C]. Online: Association for Computational Linguistics, 2020.6769-6781.
[20] LYU Q, HAVALDAR S, STEIN A, et al.Faithful chain-of-thought reasoning[A].Proceedings of the 13th International Joint Conference on Natural Language Processing and the 3rd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics[C].Nusa Dua, Bali: Association for Computational Linguistics, 2023.305-329.
[21] HU E J, SHEN Y L, WALLIS P, et al. LoRA: Low-rank adaptation of large language models[EB/OL].2021: arXiv: 2106.09685. https://arxiv.org/abs/2106.09685.
[22] PAPINENI K, ROUKOS S, WARD T, et al.Bleu: A method for automatic evaluation of machine translation[A]. Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics[C]. Philadelphia, Pennsylvania, USA: Association for Computational Linguistics, 2002.311-318.
[23] LIN C Y.ROUGE: A package for automatic evaluation of summaries[A]. Text Summarization Branches Out[C]. Barcelona, Spain: Association for Computational Linguistics, 2004.74-81.
[24] ZHANG T Y, KISHORE V, WU F, et al. BERTScore: Evaluating text generation with BERT[EB/OL].2019: arXiv: 1904. 09675. https://arxiv.org/abs/1904.09675.
[25] GUO Z R, XIA L H, YU Y H, et al. LightRAG: Simple and fast retrieval-augmented generation[EB/OL].2024:arXiv:2410.05 779. https://arxiv.org/abs/2410.05779.
[26] GUO D Y, YANG D J, ZHANG H W,et al. DeepSeek-R1: Incentivizing reasoning capability in LLMs via reinforcement learning[EB/OL].2025: arXiv: 2501.12948. https://arxiv.org/abs/2501.12948.