Scholarlink

This chapter presents a long-short-term memory (LSTM) model based on Harris Hawks optimization (HHO) for guava yield prediction under variable climate conditions. A framework of four phases is introduced. The first phase includes data description and preprocessing. Data was collected over two years from 2019 to 2021, applying pruning techniques at different depths (0, 15, 30, 45) cm during the spring, monsoon, and autumn seasons. Data included fruit number and yield per plant, as well as chemical properties of the fruits such as soluble solids, acidity, sugars, vitamin C, and fruit density. Data analysis results regarding the periods of June–August and September–November were identified as the most productive, while March–May had the highest fruit quality. Pruning to a depth of 30 cm was found to be most productive during June–August, whereas pruning to 45 cm was optimal for fruit quality in March–May. The second phase includes parameter optimization using HHO with LSTM for time series analysis to determine the best timing for pruning branches to improve guava production. HHO used to optimize feature selection to identify the best features for model training. The third phase regarding stacking ensemble learning is developed to improve the performance of the LSTM model for time series prediction, combines gradient boosting for better prediction. In the last phase, reinforcement learning (RL) is used to optimize the described pruning techniques by formulating the pruning process as a decision-making problem. The RL agent is designed to dynamically adjust pruning strategies based on environmental conditions and historical performance data to maximize guava production and quality across varying seasons. In addition, the carbon footprint of the proposed model was calculated to validate its performance. Overall, this chapter provides practical recommendations for guava farmers, providing a scientific foundation for enhancing crop yields and maintaining fruit quality through strategic pruning, as well as achieving food security and sustainability in guava production.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Long Short Term Memory Model Based on Harris Hawks Optimization for Guava Yield Prediction

  • 章节

  • 发表日期：

    2026年

查看全文
  • Ahmed Khedr,
  • Heba Askr,
  • Aboul Ella Hassanien

摘要

This chapter presents a long-short-term memory (LSTM) model based on Harris Hawks optimization (HHO) for guava yield prediction under variable climate conditions. A framework of four phases is introduced. The first phase includes data description and preprocessing. Data was collected over two years from 2019 to 2021, applying pruning techniques at different depths (0, 15, 30, 45) cm during the spring, monsoon, and autumn seasons. Data included fruit number and yield per plant, as well as chemical properties of the fruits such as soluble solids, acidity, sugars, vitamin C, and fruit density. Data analysis results regarding the periods of June–August and September–November were identified as the most productive, while March–May had the highest fruit quality. Pruning to a depth of 30 cm was found to be most productive during June–August, whereas pruning to 45 cm was optimal for fruit quality in March–May. The second phase includes parameter optimization using HHO with LSTM for time series analysis to determine the best timing for pruning branches to improve guava production. HHO used to optimize feature selection to identify the best features for model training. The third phase regarding stacking ensemble learning is developed to improve the performance of the LSTM model for time series prediction, combines gradient boosting for better prediction. In the last phase, reinforcement learning (RL) is used to optimize the described pruning techniques by formulating the pruning process as a decision-making problem. The RL agent is designed to dynamically adjust pruning strategies based on environmental conditions and historical performance data to maximize guava production and quality across varying seasons. In addition, the carbon footprint of the proposed model was calculated to validate its performance. Overall, this chapter provides practical recommendations for guava farmers, providing a scientific foundation for enhancing crop yields and maintaining fruit quality through strategic pruning, as well as achieving food security and sustainability in guava production.