[论文解读] Integrating machine learning paradigms and mixed-integer model predictive control for irrigation scheduling

Bernard T. Agyeman, Mohamed Naouri+3 more|arXiv (Cornell University)|Jun 14, 2023

Irrigation Practices and Water Management被引用 10

一句话总结

一种预测性灌溉调度器在混合整数模型预测控制框架内结合无监督分区（k-means）、基于 LSTM 的土壤水分模型，以及 PPO 训练的策略，旨在在多区田地中提高水资源利用效率和产量。

ABSTRACT

The agricultural sector currently faces significant challenges in water resource conservation and crop yield optimization, primarily due to concerns over freshwater scarcity. Traditional irrigation scheduling methods often prove inadequate in meeting the needs of large-scale irrigation systems. To address this issue, this paper proposes a predictive irrigation scheduler that leverages the three paradigms of machine learning to optimize irrigation schedules. The proposed scheduler employs the k-means clustering approach to divide the field into distinct irrigation management zones based on soil hydraulic parameters and topology information. Furthermore, a long short-term memory network is employed to develop dynamic models for each management zone, enabling accurate predictions of soil moisture dynamics. Formulated as a mixed-integer model predictive control problem, the scheduler aims to maximize water uptake while minimizing overall water consumption and irrigation costs. To tackle the mixed-integer optimization challenge, the proximal policy optimization algorithm is utilized to train a reinforcement learning agent responsible for making daily irrigation decisions. To evaluate the performance of the proposed scheduler, a 26.4-hectare field in Lethbridge, Canada, was chosen as a case study for the 2015 and 2022 growing seasons. The results demonstrate the superiority of the proposed scheduler compared to a traditional irrigation scheduling method in terms of water use efficiency and crop yield improvement for both growing seasons. Notably, the proposed scheduler achieved water savings ranging from 6.4% to 22.8%, along with yield increases ranging from 2.3% to 4.3%.

研究动机与目标

• 使用无监督学习来划分灌溉管理区以捕捉田间变异性。
• 在各区内用数据驱动的 LSTM 模型对土壤水分动态进行建模。
• 将 LSTM 模型整合到基于分区的混合整数 MPC 中以调度灌溉。
• 通过强化学习并引入受限管理区概念来消除每日显式的二进制灌溉决策。
• 在真实田地对该调度器进行两季生长和多個 MADs 的评估，以与传统方法进行比较。

提出的方法

• 使用作物类型、海拔和土壤水力参数，通过 k-means 聚类进行三阶段管理区划分。
• 基于一维里查茨方程的土壤水分建模，按区参数化，并数值求解以得到状态空间表示。
• 在开放环路的里查茨基数据上训练的 LSTM 网络，用于预测未来一天的根区水分。
• 带有区域目标的混合整数 MPC，旨在在最小化灌溉成本的同时将土壤水分保持在目标区间，带软约束。
• 使用 Proximal Policy Optimization (PPO) 推导每日灌溉决策序列，从而在每个区实现单一每日灌溉速率。
• 通过派生的绑定灌溉序列扩展到多区田块调度，而不是求解完整的多区 MI-MPC。

Figure 2 : An illustration of Stage 2 of the proposed delineation process.

实验结果

研究问题

• RQ1 如何有效地划分能捕捉田间变异性的灌溉管理区？
• RQ2 数据驱动的 LSTM 模型能否在每个区内准确预测根区土壤水分动力学？
• RQ3 将 LSTM 模型整合到基于分区的 MI-MPC 是否能相比传统调度提高水资源利用效率和产量？
• RQ4 强化学习（PPO）是否能在混合整数 MPC 框架中消除每日显式的二进制灌溉决策？
• RQ5 在真实田地的多季和多 MAD 设置下，所提出的调度器表现如何？

主要发现

• 在生长季节实现了 6.4% 至 22.8% 的用水节省。
• 在评估的季节中产量提高了 2.3% 至 4.3%。
• 相对于传统灌溉调度方法，调度器在水资源利用效率和作物产量方面显示出改进。
• 在加拿大 Lethbridge 的一个 26.4 公顷田地上对 2015 和 2022 季节进行了评估。
• 在两个管理允许枯竭（MADs）下显示出鲁棒性。

Figure 3 : A schematic diagram of the proposed LSTM model for each management zone.