Scholarlink

<p>Soil fertility plays a crucial role in sustainable agriculture, influencing nutrient availability, water retention, and overall crop productivity. Accurate assessment of soil fertility allows farmers to make informed decisions on irrigation, fertilization, and crop selection. Traditional crop prediction methods often rely on historical data and manual soil testing, which are time-consuming, labor-intensive, and prone to errors. Additionally, conventional approaches may fail to capture real-time variations in soil and environmental conditions, limiting prediction accuracy. To address these limitations, the research presents an IoT-enabled real-time crop prediction framework utilizing a Depth-wise Separable Convolutional Neural Network with Nizar Optimization Algorithm (DepSCNN-NOA). The methodology begins with IoT sensors deployed in agricultural fields, continuously collecting real-time data on soil moisture, temperature, pH, humidity, rainfall, light intensity, and nutrient levels (N, P, K). Collected data are transmitted to the cloud for secure storage and centralized access. The raw data undergoes pre-processing using Missing Value Imputation based on Denoising Autoencoders (MVI-DA) to handle noise and missing values. Relevant soil and environmental features are then selected using the Sculptor Optimization Algorithm (SOA) to reduce dimensionality and enhance predictive performance. The selected features are processed by the Depth-wise Separable Convolutional Neural Network (DepSCNN), which efficiently extracts spatial and channel-wise patterns, while the Nizar Optimization Algorithm (NOA) optimizes model’s weight to minimize prediction errors. Experimental evaluations demonstrate that the proposed DSCNN-NOA framework achieves superior accuracy (99.89%), precision (99.71%), recall (99.66%), and F1-score (99.78%) compared to traditional methods. This integrated IoT-deep learning system enables accurate, real-time crop prediction and soil fertility assessment, supporting precision agriculture and sustainable farming practices.</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

An IoT-Enabled Real-Time Crop Prediction Framework Utilizing a Depth-Wise Separable Convolutional Neural Network with Nizar Optimization Algorithm

  • 期刊论文

  • 发表日期：

    2026年4月30日

查看全文
  • Harsharani Kote,
  • S. P. Siddique Ibrahim

摘要

Soil fertility plays a crucial role in sustainable agriculture, influencing nutrient availability, water retention, and overall crop productivity. Accurate assessment of soil fertility allows farmers to make informed decisions on irrigation, fertilization, and crop selection. Traditional crop prediction methods often rely on historical data and manual soil testing, which are time-consuming, labor-intensive, and prone to errors. Additionally, conventional approaches may fail to capture real-time variations in soil and environmental conditions, limiting prediction accuracy. To address these limitations, the research presents an IoT-enabled real-time crop prediction framework utilizing a Depth-wise Separable Convolutional Neural Network with Nizar Optimization Algorithm (DepSCNN-NOA). The methodology begins with IoT sensors deployed in agricultural fields, continuously collecting real-time data on soil moisture, temperature, pH, humidity, rainfall, light intensity, and nutrient levels (N, P, K). Collected data are transmitted to the cloud for secure storage and centralized access. The raw data undergoes pre-processing using Missing Value Imputation based on Denoising Autoencoders (MVI-DA) to handle noise and missing values. Relevant soil and environmental features are then selected using the Sculptor Optimization Algorithm (SOA) to reduce dimensionality and enhance predictive performance. The selected features are processed by the Depth-wise Separable Convolutional Neural Network (DepSCNN), which efficiently extracts spatial and channel-wise patterns, while the Nizar Optimization Algorithm (NOA) optimizes model’s weight to minimize prediction errors. Experimental evaluations demonstrate that the proposed DSCNN-NOA framework achieves superior accuracy (99.89%), precision (99.71%), recall (99.66%), and F1-score (99.78%) compared to traditional methods. This integrated IoT-deep learning system enables accurate, real-time crop prediction and soil fertility assessment, supporting precision agriculture and sustainable farming practices.