Threat Detection for IoT-Based Smart Farming Systems Based on Bidirectional Gated Recurrent Unit with Long Short-Term Memory

Now a days, Agriculture and farming explore the significance of driving with in the economic growth and ensuring food security, and promoting worlds sustainable development. By examining the role of agriculture in various sectors. However, in most of the cases the agriculture sector leads to the uncontrolled utilizing of herbicides which gets harmful to the human words. To overcome these issues this research proposes an ensemble method merging Bidirectional Gated Recurrent Unit - Long short-term memory (BiGRU-LSTM) for threat detection in Internet of Things (IoT) based on smart farming. Using the Smart Agriculture Dataset (SAD), the approach preprocesses data with One-Hot Encoding and Minimum-Maximum scaling models analyze sequential data by identifying potential threats. Their outputs are combined and classified by SoftMax function. Leveraging the strengths of both models, the ensemble method improves threat detection accuracy. The proposed method demonstrates a robust approach to threat detection, combining BiGRU-LSTM for effective threat identification in IoT-based smart farming. The proposed BiGRU-LSTM model obtained high results in the context of precision 91.67%, F1 score 95.89% and accuracy 92.67%, when compared to existing methods like Statistical Process Control (SPC).