智慧农业发展综合评价体系构建、实践格局与驱动机制
学术论文，构建了智慧农业综合评价体系，方法新颖，数据详实，但发表于2026-05-30，时效性较低。
房洪伟等构建智慧农业发展综合评价体系,分析智慧农业实践格局与驱动机制,为科学评估智慧农业发展水平提供方法论支撑。
欢迎您访问《智慧农业（中英文）》官方网站！English
Toggle navigation智慧农业
Smart Agriculture ›› 2026, Vol. 8 ›› Issue (3): 26-39.doi: 10.12133/j.smartag.SA202510009
智慧农业发展综合评价体系构建、实践格局与驱动机制
房洪伟, 胡然然, 热孜燕·瓦卡斯(
)
- 新疆农业大学经济管理学院，新疆 乌鲁木齐 830052，中国
收稿日期:2025-10-11出版日期:2026-05-30基金项目:新疆维吾尔自治区科技特派员重点项目(2023KZ016); 新疆维吾尔自治区研究生教育创新计划项目(XJ2025G129)作者简介:房洪伟，硕士研究生，研究方向为农林经济。E-mail：[email protected]通信作者:热孜燕·瓦卡斯，博士，教授，研究方向为农林经济。E-mail：[email protected]
Construction of A Comprehensive Evaluation System, Practice Pattern, and Driving Mechanism for Smart Agriculture Development
FANG Hongwei, HU Ranran, REZIYAN· Wakasi()
- College of Economics and Management, Xinjiang Agricultural University, Urumqi 830052, China
Received:2025-10-11Online:2026-05-30Foundation items:Key Project of the Science and Technology Commissioner of Xinjiang Uygur Autonomous Region(2023KZ016); Project of the Postgraduate Education Innovation Plan of Xinjiang Uygur Autonomous Region(XJ2025G129)About author:FANG Hongwei, E-mail: [email protected]Corresponding author:REZIYAN·Wakasi, E-mail: [email protected]
在线阅读 11
知网下载
本地下载 27
## 可视化 0摘要/Abstract
摘要：
【目的/意义】 发展智慧农业是构建农业新质生产力的核心路径。针对当前发展中存在的区域异质性与动力错配问题，本研究旨在构建一个融合地理与产业异质性的综合评价体系，实证分析数据要素与农业物质资本的协同驱动机制，并揭示其在省域空间关联中的作用，从而为优化农业空间布局与差异化政策制定提供理论依据。 【方法】 基于2015—2023年中国省级面板数据，首先采用层次分析法与熵权法相结合的组合赋权方式构建智慧农业发展综合评价指数，并进行地形与主导产业校正。进一步，通过空间自相关检验与空间杜宾模型识别智慧农业发展的空间关联特征与溢出模式；运用广义矩估计模型检验数据要素的驱动作用；并通过中介效应与调节效应模型，系统验证数据要素与物质资本的协同机制及其影响路径。 【结果和讨论】 （1）智慧农业呈现梯度发展格局，区域间增长动力与智慧化水平存在系统性错配。（2）核心驱动机制在于数据要素与农业物质资本的结构性互补，二者协同效应显著；技术资本发挥关键中介作用，且其促进作用具有非线性门槛特征。（3）空间上以竞争效应为主导：产业结构相似省份间呈现明显的“虹吸效应”；“数据-资本”协同模式具有正向空间溢出性，可形成可效仿的区域发展路径。异质性分析表明，上述机制的效果因农业功能区划与发展阶段不同而存在差异。 【结论】 本研究为理解智慧农业发展的要素协同机制与空间竞争逻辑提供了直接经验证据，有助于识别区域差异化驱动路径，从而为促进智慧农业均衡发展与政策精准实施提供参考。
关键词:智慧农业,数据要素,空间溢出,空间杜宾模型,区域异质性,评价体系
Abstract:
[Objective] This study addresses regional disparities and imbalanced drivers in developing smart agriculture, a core approach to fostering new quality agricultural productivity. It aims to: (1) construct a comprehensive evaluation system incorporating geographical and industrial heterogeneity; (2) empirically analyze the synergistic drive between data elements and agricultural physical capital; and (3) reveal their role in inter-provincial spatial linkages. The significance lies in its potential to inform strategic planning and policy-making, thereby contributing to the sustainable transformation of agriculture and balanced regional development within the context of rural revitalization. [Methods] A quantitative spatial econometric approach was employed using panel data from 30 Chinese provinces spanning from 2015 to 2023. The research was executed in three key stages. First, a comprehensive provincial-level smart agriculture development index was constructed. This index integrated multiple dimensions and was weighted by combining the Analytic Hierarchy Process and the Entropy method, with adjustments made for terrain and leading industry heterogeneity. Second, a series of econometric models were specified. Baseline fixed-effects and generalized method of moments models were used to examine the driving role of data elements, while mediating and moderating effect models were employed to systematically verify the synergistic mechanism between data and physical capital and its pathways. Third, spatial autocorrelation tests and spatial Durbin models were employed with three spatial weight matrices—geographical contiguity, agricultural resource zoning, and agricultural economic structure similarity—to identify spatial correlation characteristics and spillover patterns. Direct and indirect effects were decomposed to precisely quantify local impacts and spatial spillovers. [Results and Discussions] The analysis yielded four clusters of key findings that confirmed and refined the proposed core hypotheses. 1) Gradient development and regional mismatch: Smart agriculture development exhibited a pronounced "ladder-like" spatial pattern. The Huang-Huai-Hai region remained the persistent leader in absolute development level. The southwestern region also maintained a relatively high level, which appeared partly "forced" by its challenging terrain, necessitated efficiency-seeking technology. In contrast, the Northeast and Northwest regions lagged. A critical systemic mismatch was revealed: Regions with the highest growth momentum were not necessarily those with the highest current smartization levels, indicating divergent developmental pathways. 2) The core synergistic mechanism: A significant positive interaction was found between data inputs and agricultural physical capital. Crucially, the independent coefficients of each were often found to be insignificant or even negative in spatial models, which underscored that limited or even negative marginal returns were yielded by isolated, uncoordinated investment in either domain. Significant systemic gains were unlocked precisely by their structural complementarity. Furthermore, this synergy was found to operate partially through the channel of technological capital accumulation. The mediating effect of technological capital was confirmed to be significant, and its own impact on smart agriculture output was exhibited as a nonlinear threshold characteristic. This confirmed that a critical mass of technological capital had to be accumulated before its benefits could be fully realized. 3) Competition-dominated spatial interactions: The spatial analysis revealed that inter-provincial dynamics were characterized primarily by competition rather than cooperation. A significant negative spatial spillover was detected specifically under the economic structure similarity matrix. This indicated that resources, talent, and investment were competed for by provinces with similar agricultural economic profiles, potentially hindering each other's growth. However, a nuanced finding was observed: while raw data or capital might be siphoned away, significant positive spatial spillovers were generated by successful provincial models of "data-capital" synergy. This suggested that best practices and development paradigms could be diffused, offering a pathway to transcend pure competition. 4) The effectiveness of the core drivers was found to be highly context-dependent. From a zoning perspective, the Huang-Huai-Hai region was characterized by digital-drive but was found to lack deep synergy; the Northeast was constrained by traditional path dependency; the Southwest was shown to exhibit a paradox of high knowledge spillover but low local application; and the northwest was characterized by singular, weak drivers. From a developmental stage perspective, the role of RD investment was assessed as stable, while the payoff from digital infrastructure was seen to be contingent on an "efficiency threshold", and its spatial spillover effect was observed to diminish as regional total factor productivity increased. [Conclusions] It is demonstrated that development is not merely a function of increased inputs, but is critically determined by the structural coupling of data and physical capital. This coupling is facilitated by technological capital, which acts as a nonlinear mediator. At a practical level, a one-size-fits-all approach to investment policy is argued against. For leading regions such as the Huang-Huai-Hai, policy focus should be placed on deepening existing synergies. For regions like the northeast, breaking path dependence is seen to require policies that forcefully couple new digital tools with legacy physical assets. The pervasive "siphon effect" is identified as necessitating national-level coordination mechanisms among structurally similar provinces to mitigate destructive competition. Ultimately, the promotion of smart agriculture is concluded to require spatially differentiated policies that strategically foster local factor synergy while managing the competitive externalities inherent in regional linkages.
Key words:smart agriculture,data factor,spatial spillover,spatial Durbin model,regional heterogeneity,evaluation system
中图分类号:
引用本文
房洪伟, 胡然然, 热孜燕·瓦卡斯. 智慧农业发展综合评价体系构建、实践格局与驱动机制[J]. 智慧农业(中英文), 2026, 8(3): 26-39.
FANG Hongwei, HU Ranran, REZIYAN· Wakasi. Construction of A Comprehensive Evaluation System, Practice Pattern, and Driving Mechanism for Smart Agriculture Development[J]. Smart Agriculture, 2026, 8(3): 26-39.
使用本文
[收藏文章](javascript:;) /[推荐](javascript:void(null))
导出引用管理器EndNote|Ris|BibTeX
链接本文:https://www.smartag.net.cn/CN/10.12133/j.smartag.SA202510009
https://www.smartag.net.cn/CN/Y2026/V8/I3/26
[图/表 12](javascript:;)
表1
智慧农业核心变量数据来源与处理（2015—2023）
| 变量类别 | 变量名称 | 主要数据来源 | 处理与计算说明 |
|---|---|---|---|
| 被解释变量 | 智慧农业应用水平 | 综合各年鉴指标计算 | 基于AHP-熵权组合赋权法构建，经地形与主导产业校正，详见2.1节 |
| 核心解释变量 | 数据要素（移动基站密度） | 《中国高技术产业统计年鉴》 | 取对数处理 |
| 农业物质资本（农业投资强度） | 《中国农业机械工业年鉴》 | 以农林牧渔业固定资产投资额与总产值比值衡量，并经价格平减 | |
| 技术资本（RD投入强度） | EPS数据库 | 以RD内部经费支出与主营业务收入比值衡量 | |
| 控制变量 | 人均农业产值、受灾比例等 | 《中国人口与就业统计年鉴》EPS数据库 | 具体定义见1.2.3节，连续变量均进行标准化 |
| 工具变量 | 历史邮局数量交互项 | 《中国电子信息产业统计年鉴》 | 1984年地级市邮局数与上一年全国信息技术服务收入的交互项，取对数 |
| 地形因素 | 地形修正系数 | 《中华人民共和国地貌图集（1∶100万）》 | 通过量化各省主导地形的农业适宜性（1-10分），并经过非线性映射（α=0.7）将其转换为0.6至1.0的加权值，用于修正种植业相关的智慧农业指标 |
| 主导产业权重 | 各省“农林牧渔产值占比” | 《中国统计年鉴》 | 各省农业内部各产业（种植业、林业、畜牧业、渔业）的产值占总产值的比重 |
表1
表2
智慧农业变量描述性统计（2015—2023）
| 变量 | 观测值 | 均值 | 标准差 | 最小值 | 最大值 |
|---|---|---|---|---|---|
| 智慧农业应用水平 | 270 | 30.838 | 22.304 | 3.141 | 100.000 |
| 技术资本 | 270 | 0.074 | 0.965 | -1.558 | 2.468 |
| 数据要素 | 270 | 0.309 | 0.749 | -0.921 | 3.343 |
| 农业物质资本 | 270 | 8.084 | 9.030 | 1.353 | 10.937 |
| 数据要素×农业物质资本 | 270 | 0.020 | 1.043 | -0.390 | 6.155 |
| 受灾比例 | 270 | 5.460 | 1.714 | -1.050 | 8.349 |
| 人均农业产值 | 270 | 0.160 | 0.149 | 0.003 | 0.799 |
| 农业土地流转率 | 270 | 0.376 | 0.167 | 0.041 | 0.922 |
| 受教育年限 | 270 | 0.145 | 0.750 | -2.290 | 2.637 |
| 农林水事务支持度 | 270 | 0.000 | 0.000 | 0.000 | 0.001 |
| 工具变量 | 270 | 14.424 | 0.573 | 13.291 | 16.714 |
表2
表3
智慧农业发展评价指标体系
| 一级指标 | 指标阐释 | 数据来源 |
|---|---|---|
| 生产加工数字化 | 智能农机作业面积占比 | 《中国农业机械工业年鉴》 |
| 单位面积无人机作业强度 | 《中国农业机械工业年鉴》 | |
| 智能灌溉面积占比 | EPS数据库 | |
| 智能化播种作业占比 | 《中国林业和草原统计年鉴》 | |
| 智能化养殖场占比 | 《中国畜牧兽医年鉴》 | |
| 林业智能监测覆盖率 | 《中国林业和草原统计年鉴》 | |
| 智能化渔船占比 | 《中国渔业统计年鉴》 | |
| 流通营销数字化 | 电商企业占比 | 《中国高技术产业统计年鉴》 |
| 村级物流网点覆盖率 | 《中国高技术产业统计年鉴》 | |
| 农产品网络零售额占比 | EPS数据库 | |
| 数字技术服务业 | 智慧服务组织数量密度 | 《中国统计年鉴》 |
| 高新技术企业密度 | 《中国高技术产业统计年鉴》 | |
| 农业科研创新数字化 | 发明专利占比 | 国家知识产权局 |
| 数字农业研发投入强度 | 《中国统计年鉴》 | |
| 数字农业研发人员占比 | 《中国人口与就业统计年鉴》 |
表3
图1
智慧农业发展的理论分析框架
图1
表4
智慧农业发展的空间计量模型选择检验结果
| 空间权重矩阵 | LR统计量 | P值 | 模型选择 |
|---|---|---|---|
| 农业分区权重 | 15.090 | 0.002 | SDM |
| 经济权重 | 42.430 | 0.001 | SDM |
| 地理权重 | 2.630 | 0.452 | SAR |
表4
图2
区域智慧农业发展指数与增速表
图2
表5
省级智慧农业结构趋同动态研究的变异系数与莫兰指数
| 年份 | 全国结构指数CV | 空间邻接平均差异 | 经济组内平均差异 | 莫兰指数 | Z 值 |
|---|---|---|---|---|---|
| 2015 | 0.809 | 0.100 | 0.124 | 0.161* | 1.638 |
| 2016 | 0.858 | 0.112 | 0.122 | 0.159* | 1.585 |
| 2017 | 0.762 | 0.113 | 0.139 | 0.113 | 1.211 |
| 2018 | 0.816 | 0.128 | 0.165 | 0.204** | 2.015 |
| 2019 | 0.741 | 0.093 | 0.133 | 0.334*** | 3.031 |
| 2020 | 0.555 | 0.07 | 0.105 | 0.286*** | 2.632 |
| 2021 | 0.681 | 0.107 | 0.155 | 0.308*** | 2.846 |
| 2022 | 0.740 | 0.114 | 0.159 | 0.295*** | 2.720 |
| 2023 | 0.707 | 0.122 | 0.170 | 0.251** | 1.963 |
表5
图3
中国省域智慧农业指数空间分布格局及演变 a. 2019年智慧农业发展水平空间分布格局 b. 2023年智慧农业发展水平空间分布格局 注：该图基于自然资源部标准地图服务网站下载的审图号为GS（2024）0650号的标准地图制作，底图边界无修改；数据来源：AHP+熵权法综合指数评估，灰色区域表示数据缺失。
图3
表6
数据要素与农业物质资本协同驱动智慧农业发展的实证检验结果
| 变量 | OLS | 中介路径 | 直接效应 | 调节效应 | GMM | TOPSIS | 传感器企业 |
|---|---|---|---|---|---|---|---|
| （1） | （2） | （3） | （4） | （5） | （6） | （7） | |
| 数据要素 | 0.577*** | 0.017*** | 0.450** | 0.629*** | 0.300*** | 3.865 1*** | 74.907*** |
| （0.193） | （0.005） | （0.202） | （0.194） | （0.109） | （1.106） | （17.486） | |
| 农业物质资本 | 1.153* | 3.494*** | 0.001 1*** | 96.078* | |||
| （0.571） | （0.864） | （0.000） | （54.038） | ||||
| 数据要素×农业物质资本 | 1.763*** | 2.358** | 7.705 6*** | -70.306 | |||
| （0.710） | （1.045） | （2.386） | （41.580） | ||||
| 技术资本 | 7.442** | ||||||
| （3.538） | |||||||
| 滞后项 | 0.731*** | ||||||
| （0.137） | |||||||
| 常数项 | 12.173 | -0.260 | 14.110* | 12.813* | 5.595* | -10.087 | -1 247.619* |
| （7.537） | （0.255） | （8.009） | （7.336） | （3.131） | （7.200） | （657.436） | |
| AR（1）p值 | 0.014 | ||||||
| AR（2）p值 | 0.671 | ||||||
| Hansen检验p值 | 0.424 | ||||||
| 0.957 | |||||||
| 控制年份 | 是 | 是 | 是 | 是 | 是 | 是 | 是 |
| 控制省份 | 是 | 是 | 是 | 是 | 是 | 是 | 是 |
| R 2 | 0.657 | 0.555 | 0.670 | 0.668 | 0.576 | 0.746 | |
| 观测值 | 270 | 270 | 270 | 270 | 270 | 270 | 270 |
表6
表7
数据要素与物质资本协同效应的空间计量分析结果
| 变量 | 农业分区权重 | 经济权重 | 地理权重 |
|---|---|---|---|
| 数据要素 | -8.800*** | -3.997* | -8.230*** |
| （2.235） | （2.400） | （2.292） | |
| 农业物质资本 | -0.001** | -0.002*** | -0.001* |
| （0.001） | （0.001） | （0.000） | |
| 数据要素×农业物质资本 | 9.879*** | 18.314*** | 8.269** |
| （4.054） | （5.737） | （3.867） | |
| 技术资本 | 12.299*** | 10.857*** | 11.112*** |
| （1.986） | （1.957） | （2.014） | |
| 常数项 | 42.594*** | 202.173*** | 49.546*** |
| （7.384） | （48.680） | （7.439） | |
| Rho | 0.025 | -0.326* | 0.044 |
| （0.065） | （0.313） | （0.083） | |
| LR | 13.66 | 26.41 | 1.21 |
| P值 | 0.001 0 | 0.000 0 | 0.545 1 |
| 控制年份 | 是 | 是 | 是 |
| 控制省份 | 是 | 是 | 是 |
| R 2 | 0.490 | 0.474 | 0.483 |
| 观测值 | 240 | 240 | 240 |
表7
表8
智慧农业发展的空间溢出效应分解结果
| 变量 | 农业分区权重 | 经济权重 |
|---|---|---|
| 直接效应 | 间接效应 | 总效应 |
| 数据要素 | -8.832*** | 6.225*** |
| （2.158） | （2.005） | （2.864） |
| 农业物质资本 | -0.001** | 0.001 |
| （0.000） | （0.001） | （0.001） |
| 数据要素×农业物质资本 | 9.395*** | -6.419 |
| （3.795） | （5.274） | （5.168） |
| RD投入强度 | 12.396*** | 0.302 |
| （2.043） | （0.800） | （2.238） |