Wu et al. (2026) Phenology-Aware Collaborative Decision-Making and AG-PSTC Algorithm for Precision Irrigation in Smart Tea Gardens

Identification

• Journal: Electronics
• Year: 2026
• Date: 2026-03-30
• Authors: Luofa Wu, Helai Liu, Shifu Shu, Chun Ye
• DOI: 10.3390/electronics15071429

Research Groups

• Agricultural Engineering Institute, Jiangxi Academy of Agricultural Sciences, Nanchang, China

Short Summary

This study develops a precision irrigation system for smart tea gardens, integrating a Phenology-Aware Collaborative Decision-Making (PACD) model and an Adaptive Gain Predictive Super-Twisting Sliding Mode Control (AG-PSTC) algorithm. The system effectively mitigates time delays and nonlinear interference while dynamically adjusting irrigation based on actual plant needs, significantly improving precision and suppressing false irrigation commands caused by biomass fluctuations from plucking.

Objective

• To design a precision irrigation system for smart tea gardens that overcomes challenges such as significant time delays, nonlinear interference, and phenological biomass fluctuations (e.g., from plucking), which cause traditional irrigation models to fail in precise water supply.

Study Configuration

• Spatial Scale: Experimental base located in the tea plantation of the Jiangxi Academy of Agricultural Sciences, Gao’an City, Jiangxi Province, China (28°25′18″ N, 115°13′07.71″ E), with an elevation ranging from 85 meters to 122 meters. The core experimental area covers 533,300 square meters, with standardized planting zones accounting for 78% of the total area. The soil type is predominantly red soil, and slopes range from 0° to 15°. Soil moisture sensors were deployed at a depth of 20 centimeters.
• Temporal Scale: Simulation experiments were conducted over a 60-day cycle. Field verification was performed in June 2025 during the tea picking season. Model parameters (α and β) were calibrated using historical data collected from 20 fixed monitoring points spanning three growth cycles (2023 to 2025).

Methodology and Data

• Models used:
  • Phenology-Aware Collaborative Decision-Making (PACD) model.
  • Adaptive Gain Predictive Super-Twisting Sliding Mode Control (AG-PSTC) algorithm.
  • "Temperature–time–water" phenological reference growth model.
  • Crop Water Stress Index (CWSI) diagnostic operator.
  • Improved Smith predictor for phase compensation.
  • Integral sliding mode surface.
  • Super-twisting control law.
  • Barrier function-based adaptive gain law.
  • Dynamic model of the irrigation process: First-order inertial link with pure time delay (G(s) = K / (Ts + 1) * e^(-τs)).
  • Comparison algorithms: Traditional Proportional–Integral–Derivative (PID), Fuzzy PID, Adaptive PID, Model Predictive Control (MPC), traditional Sliding Mode Control (SMC), and Smith-SMC.
• Data sources:
  • Multi-modal field sensors (Weihai JXCT Electronic Technology Co., Ltd.) for soil moisture (0–100% with ±3% accuracy), electrical conductivity (0–10,000 µS/cm with 10 µS/cm accuracy), pH value (3–9 with ±0.3 accuracy), temperature (-40–125 °C with ±0.2 °C accuracy), and humidity (0–100% with ±3% accuracy).
  • Real-time video surveillance and IoT environmental data monitoring.
  • Historical canopy shoot density and root-zone soil moisture data (2023–2025).
  • Real-time multi-dimensional meteorological data (air temperature, light intensity, air humidity, atmospheric pressure).
  • Satellite maps of the tea plantation for IoT node deployment.

Main Results

• AG-PSTC Algorithm Performance:
  • Reduced rise time by 78% compared to traditional PID controllers.
  • Achieved a steady-state Mean Absolute Error (MAE) of 6.94 × 10⁻⁷, reducing the error by four orders of magnitude compared to traditional PID and improving accuracy by approximately 500 times compared to traditional SMC.
  • Demonstrated robust performance under ±40% time-delay perturbation (MAE stable at ~6.9 × 10⁻⁷) and ±33% inertial parameter perturbation (adjustment time fluctuation less than 2 seconds).
  • Effectively suppressed chattering, achieving smooth convergence with a maximum overshoot of 0.28%.
• Phenology-Aware Collaborative Decision-Making:
  • Successfully decoupled phenology-driven biomass changes from actual water stress, preventing misjudgment during plucking operations.
  • Suppressed false irrigation commands, maintaining the target soil moisture (Wtarget) near the physiological benchmark (0.736) even with a 60% drop in observed shoot density.
  • Achieved a decision variance of 5.77 × 10⁻⁶ during picking interference, representing a two-order-of-magnitude improvement in stability compared to traditional phenological models.
  • Controlled the maximum decision deviation within 0.024.
• Field Verification and System Stability:
  • Under high-temperature (up to 37 °C) and disturbed environments, the system maintained soil moisture with a standard deviation within 1.51%.
  • Soil physicochemical indicators in the root zone remained stable: pH value fluctuated within 8.83–8.97, and electrical conductivity was maintained between 23 and 29 µS/cm.
  • The system triggered zero erroneous irrigation events during continuous plucking operations.

Contributions

• Developed the first phenology-aware collaborative decision-making model for tea, which constructs an ideal "temperature–time–water" phenological baseline and utilizes a Crop Water Stress Index (CWSI) to decouple shoot density changes, enabling dynamic target soil moisture setting and eliminating false irrigation signals caused by plucking.
• Proposed the Adaptive Gain Predictive Super-Twisting Sliding Mode Control (AG-PSTC) algorithm, incorporating an improved Smith predictor for time-delay compensation, a second-order super-twisting sliding mode for chattering elimination, and a barrier function-based adaptive gain law for robust, finite-time convergence without requiring prior knowledge of disturbance upper bounds.
• Integrated and validated a comprehensive smart tea garden management system based on a closed-loop hierarchical architecture, demonstrating a vertically integrated framework from crop physiological modeling to low-level precision control in complex, real-world agricultural environments.

Funding

• Construction Project of Scientific Research Base for Tea Full Process Mechanization of Ministry of Agriculture and Rural Affairs of China (2103-00000-20-01-763233).
• Jiangxi Province Pilot Project for Integrated R&D, Manufacturing, and Promotion of Agricultural Machinery Equipment (YCTY202508).

Citation

@article{Wu2026PhenologyAware,
  author = {Wu, Luofa and Liu, Helai and Shu, Shifu and Ye, Chun},
  title = {Phenology-Aware Collaborative Decision-Making and AG-PSTC Algorithm for Precision Irrigation in Smart Tea Gardens},
  journal = {Electronics},
  year = {2026},
  doi = {10.3390/electronics15071429},
  url = {https://doi.org/10.3390/electronics15071429}
}