跳转至

用于家禽养殖福利约束控制的混合边缘云数字孪生系统

文章背景与核心概要

商业化家禽养殖涉及环境与生物动力学之间紧密耦合的复杂系统,然而传统的环境控制手段大多依赖于启发式规则。本文提出了一种混合边缘云数字孪生框架,旨在实现家禽养殖设施中基于福利约束的实时环境控制。

该系统通过集成分布式传感、设备端状态估计、混合物理-数据模型以及模型预测控制(MPC),在严格的农场约束条件下实现了前瞻性与自适应管理。实验结果表明,该框架在提高温度预测精度、降低氨气浓度超标率以及提升通信效率方面表现优异,为智慧农业的精准化管理提供了新的技术路径。


执行摘要

商业化家禽生产管理着紧密耦合的环境和生物动力学,但传统的气候控制在很大程度上依赖于启发式、基于规则的方法。本文介绍了一种混合边缘云数字孪生框架,旨在用于家禽设施中实时、受福利约束的环境控制。

Commercial poultry production manages tightly coupled environmental and biological dynamics, yet traditional climate control relies heavily on heuristic, rule-based methods. This paper introduces a hybrid edge-cloud digital twin framework designed for real-time, welfare-constrained environmental control in poultry facilities.

通过集成分布式传感、设备端状态估计、混合物理-数据模型和模型预测控制(MPC),该系统能够在严格的农场约束下实现前瞻性和自适应管理。

By integrating distributed sensing, on-device state estimation, a hybrid physics-data model, and Model Predictive Control (MPC), the system enables anticipatory and adaptive management under strict farm constraints.


关键框架组件

  • 混合建模方法: 将灰盒热力学和质量平衡公式与学习到的残差模型相结合。这有效地捕捉了未建模的生物变异性,包括与活动相关的代谢热。
  • Hybrid Modeling Approach: Combines a grey-box thermodynamic and mass-balance formulation with a learned residual model. This effectively captures unmodeled biological variability, including activity-dependent metabolic heat.
  • 边缘优先架构: 将混合模型嵌入到状态空间表示中,以便直接在边缘进行实时估计和控制,而云端协调则负责跨农场学习和长周期优化。
  • Edge-First Architecture: Embeds the hybrid model into a state-space representation for real-time estimation and control directly at the edge, while cloud coordination handles cross-farm learning and long-horizon optimization.
  • 带宽感知处理: 利用异步同步技术,确保在连接受限的农业环境中实现无缝部署。
  • Bandwidth-Aware Processing: Utilizes asynchronous synchronization to ensure seamless deployment in connectivity-limited agricultural environments.

性能与实验结果

在保真度极高的肉鸡生产试验台中,所提出的数字孪生框架相较于传统的基于规则的控制和纯物理建模,表现出了显著的改进:

Evaluated within a high-fidelity broiler production testbed, the proposed digital twin framework demonstrated significant improvements over traditional rule-based control and physics-only modeling:

  • 温度精度: 将温度预测误差从 \(1.8^\circ\text{C}\) 降低至 \(0.4^\circ\text{C}\)
  • Temperature Precision: Reduced temperature prediction error from \(1.8^\circ\text{C}\) down to \(0.4^\circ\text{C}\).
  • 福利合规性: 氨气约束违规减少了 90%。
  • Welfare Compliance: Decreased ammonia constraint violations by \(90\%\).
  • 通信效率: 通过边缘优先处理,将网络和通信需求降低了约 30 倍。
  • Communication Efficiency: Lowered network and communication requirements by approximately 30-fold through edge-first processing.
  • 鲁棒性: 实现了 0.92 的域迁移分数(Domain Transfer Score),证明了在不同设施条件下均具有稳健的性能。
  • Robustness: Achieved a Domain Transfer Score of 0.92, demonstrating robust performance across varying facility conditions.

其他元数据

  • 文档规格: 14 页,4 张图,2 张表
  • Document Specs: 14 pages, 4 figures, 2 tables