EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Proximal policy optimization integrated with Bayesian optimization for stand structure optimization in Pinus yunnanensis secondary forests

Jianming Wang, Chenyang Lv, Jiting Yin, Yuling Chen, Shuangqing Lv, Baoguo Wu

Frontiers in Plant Science · 26.06.2026 · 17

DOI: 10.3389/fpls.2026.1840683 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Plant Science, online erschienen am 26.06.2026. Verfasst von Jianming Wang und 5 weiteren Autor:innen. Beteiligte Einrichtung: School of Mathematics and Computer Science, Dali University, Dali, Yunnan, China. Die Arbeit nennt 54 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

Secondary forests are often characterized by irrational stand structures and spatial imbalances, which constrain the full realization of their ecological functions. Existing deep learning approaches for stand structure optimization still face several limitations, including low computational efficiency, Q-value overestimation, and insufficient consideration of site-specific factors in felling constraints.

Methods

To address these issues, this study focused on secondary Pinus yunnanensis forests on Cangshan Mountain in Dali. Based on field survey data from typical plots, a stand structure optimization model was developed with individual trees as decision-making units. At the algorithmic level, the proximal policy optimization (PPO) algorithm was introduced to replace the Deep Q-Network (DQN), employing proximal clipping to reduce Q-value overestimation and stabilize high variance. This approach was further integrated with Bayesian optimization (BO) for automatic hyperparameter tuning, thereby establishing a PPO-BO framework. In terms of regulatory constraints, slope was incorporated as a key site factor, with trees located on slopes steeper than 45° excluded from felling to ensure ecological suitability. Additionally, six optimization schemes were designed, comprising three tree selection strategies combined with a multi-objective optimization framework, and their performance was systematically evaluated.

Results

The results demonstrated that the PPO-BO algorithm consistently achieved 19 higher maximum objective function values across five circular plots (0.4475, 0.5075, 0.5146, 0.5133, and 0.5255) compared with the DQN algorithm (0.4421, 0.4973, 0.5051, 0.4947, and 0.5057), while also reducing overall training costs. Following selective felling, the spatial structure indices of each plot improved markedly, with the overall stand structure gradually approaching that of natural forests.

Discussion

Overall, this study provides new insights into the optimization of stand structure in secondary forests and enriches the theoretical and methodological framework for sustainable forest management.

Schlagwörter

Stichwörter der Autor:innen

Bayesian OptimizationProximal Policy Optimizationdeep reinforcement learningsite factorsstand structure optimization

Autor:innen und Institutionen

  • School of Mathematics and Computer Science, Dali University, Dali, Yunnan, China.
  • Dali Forestry and Grassland Science Research Institute, Dali, Yunnan, China.
  • Institute of Remote Sensing and Geographic Information System, School of Earth and Space Sciences, Peking University, Beijing, China.
  • Faculty of Surveying and Information Engineering, West Yunnan University of Applied Sciences, Dali, Yunnan, China.
  • School of Information Science and Technology, Beijing Forestry University, Beijing, China.

Zitieren

APA

Wang, J., Lv, C., Yin, J., Chen, Y., Lv, S., & Wu, B. (2026). Proximal policy optimization integrated with Bayesian optimization for stand structure optimization in Pinus yunnanensis secondary forests. Frontiers in Plant Science, 17. https://doi.org/10.3389/fpls.2026.1840683

Vancouver

Wang J, Lv C, Yin J, Chen Y, Lv S, Wu B. Proximal policy optimization integrated with Bayesian optimization for stand structure optimization in Pinus yunnanensis secondary forests. Frontiers in Plant Science. 2026;17. doi:10.3389/fpls.2026.1840683

BibTeX
@article{Wang2026,
  author = {Wang, Jianming and Lv, Chenyang and Yin, Jiting and Chen, Yuling and Lv, Shuangqing and Wu, Baoguo},
  title = {Proximal policy optimization integrated with Bayesian optimization for stand structure optimization in Pinus yunnanensis secondary forests},
  journal = {Frontiers in Plant Science},
  year = {2026},
  volume = {17},
  doi = {10.3389/fpls.2026.1840683}
}
Kostenlos starten →