Scholarlink

<p>In recent years, precision agriculture integrated the portfolio of technologies for continuous monitoring of the soil parameters and recommend crops based on soil fertility to enhance the crop yield. Specifically, the crop recommendation system monitors the significant parameters including the soil moisture, pH, nitrogen, phosphorous, potassium, and temperature, for decision making based on the soil conditions and recommends the suitable crops. However, the existing soil moisture prediction techniques suffered from time complexity and yielded inaccurate detection, due to the dynamic nature of soil moisture controlled by the spatial and temporal variability associated with the soil parameters. Hence, in this research, Intelligent Hunting Optimized Adaptive Light Gradient Boosting Ensemble Deep Neural Network (InHtO-AE-LNN) model is proposed for achieving the accurate soil moisture prediction. Additionally, optimized SMOTE is proposed for managing the data imbalance issues associated with the aggregated data. The proposed Intelligent Hunting Optimizer (InHtO) not only fine-tunes the model parameters but also optimizes the SMOTE parameters in balancing the aggregated data. The proposed InHtO-AE-LNN model acquires 98.88%, 98.94%, and 97.64%, 1.74, 3.43, and 1.85 for accuracy, sensitivity, specificity, MAE, MSE, and RMSE with 90% of training for the crop recommendation dataset and attained 98.23% accuracy, 98.00% sensitivity, 98.00%, 1.88 MAE, 4.00MSE, and 2.00 RMSE specificity for Harvard crop dataset.</p>

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

Soil moisture prediction and crop recommendation in IoT-based smart agricultural monitoring using intelligent hunting based adaptive light gradient boosting ensemble deep neural network

  • 期刊论文

  • 发表日期：

    2025年7月26日

查看全文
  • Uttam R. Patole,
  • Manish Shrivastava

摘要

In recent years, precision agriculture integrated the portfolio of technologies for continuous monitoring of the soil parameters and recommend crops based on soil fertility to enhance the crop yield. Specifically, the crop recommendation system monitors the significant parameters including the soil moisture, pH, nitrogen, phosphorous, potassium, and temperature, for decision making based on the soil conditions and recommends the suitable crops. However, the existing soil moisture prediction techniques suffered from time complexity and yielded inaccurate detection, due to the dynamic nature of soil moisture controlled by the spatial and temporal variability associated with the soil parameters. Hence, in this research, Intelligent Hunting Optimized Adaptive Light Gradient Boosting Ensemble Deep Neural Network (InHtO-AE-LNN) model is proposed for achieving the accurate soil moisture prediction. Additionally, optimized SMOTE is proposed for managing the data imbalance issues associated with the aggregated data. The proposed Intelligent Hunting Optimizer (InHtO) not only fine-tunes the model parameters but also optimizes the SMOTE parameters in balancing the aggregated data. The proposed InHtO-AE-LNN model acquires 98.88%, 98.94%, and 97.64%, 1.74, 3.43, and 1.85 for accuracy, sensitivity, specificity, MAE, MSE, and RMSE with 90% of training for the crop recommendation dataset and attained 98.23% accuracy, 98.00% sensitivity, 98.00%, 1.88 MAE, 4.00MSE, and 2.00 RMSE specificity for Harvard crop dataset.