Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM63389
Volume:
10 |
5 |
May 2026
Pages:
1-10
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Computer Science and Engineering
Government college of Technology, Coimbatore
Computer Science and Engineering
Government college of Technology, Coimbatore
Pigeon pea (Cajanus cajan) is an important pulse crop grown in the tropical and subtropical regions of the world, especially in India. The productivity of pigeon pea is severely impacted by insect pests such as pod borers, aphids, whiteflies, and leaf webbers, which are highly dependent upon climatic conditions. Traditional pest monitoring and forecasting methods rely on manual observation and simple statistical models which often fail to produce accurate predictions about pest outbreaks in pigeon pea crops because they cannot handle nonlinear relationships. In the present study, we introduce PigeonXNet which is a deep learning system based on climate that predicts outbreak of pests in pigeon pea plants with the use of historical climate and pest data. In our proposed system, we have used various statistical models as well as deep learning models like Auto-Regressive Integrated Moving Average (ARIMA), CNN-LSTM, and Bi-LSTM which can handle both linear and non-linear relationships. While the ARIMA algorithm works as a base model for time series prediction, CNN-LSTM and Bi-LSTM model capture short-term as well as long-term dependencies and enhances the prediction accuracy of a Bi-LSTM model through the process of learning temporal information in bi-directional form. To make the prediction model transparent and interpretable, we utilize Explainable Artificial Intelligence approaches like SHapley Additive exPlanations (SHAP) in our approach, through which we can understand the contribution of various climatic parameters like temperature, humidity, and precipitation in pest outbreaks. As a result, experimental outcomes indicate the effectiveness of our proposed approach in forecasting accuracy. Our proposed system helps farmers make decisions timely regarding pest management, prevent unnecessary use of pesticides, and improve production of crops.
KEYWORDS
In this study, an intelligent approach based on deep learning was designed for the purpose of pigeon pea pest prediction through time-series modeling of environmental variables along with historical pest data. It is important to note that this study efficiently leveraged and compared ARIMA, CNN-LSTM, and Bi-LSTM models to incorporate both linear and nonlinear temporal effects on the pest infestation pattern. ARIMA model offered robust benchmark results for the trend detection of pests while CNN-LSTM model accounted for feature interaction and temporal dependency of data. In comparison to ARIMA and CNN-LSTM, the performance of Bi-LSTM was further enhanced by considering both the past and future temporal information leading to superior accuracy. The empirical results indicated that the performance of Bi-LSTM was higher than CNN-LSTM with MAE: 10.244 and RMSE: 10.460 while being much better than ARIMA model having MAE: 17.687 and RMSE: 21.524. Besides, the employment of Explainable AI techniques like SHAP helped identify the most influencing environmental factors contributing to pests infestations.
[1] FAOSTAT, “Pigeonpea production statistics and cultivation trends,” Food and Agriculture Organization of the United Nations (FAO), Rome, Italy, 2024. [2] R. Sharma, P. Verma, and S. Kulkarni, “Climate effects on pigeonpea pests and seasonal outbreak analysis,” Journal of Agricultural Science, vol. 15, no. 3, pp. 112–120, 2023. [3] Agricultural Research Council, Pulse Crop Productivity Report 2022, New Delhi, India: ARC Publications, 2022. [4] K. Patel, M. Reddy, and A. Singh, “Insect pests in pigeonpea cultivation and their management practices,” International Journal of Pest Management, vol. 67, no. 4, pp. 255–263, 2021. [5] P. Singh and R. Kumar, “Yield loss assessment due to pod borers in pigeonpea crops,” Crop Protection, vol. 135, pp. 104–110, 2020. [6] V. Rao, S. Mishra, and N. Gupta, “Weather-based pest forecasting systems for sustainable agriculture,” Agricultural Informatics Journal, vol. 10, no. 2, pp. 78–89, 2022. [7] Indian Council of Agricultural Research (ICAR), Annual Report on Climate and Agriculture 2023, New Delhi, India: ICAR, 2023. [8] A. Kumar, D. Prasad, and S. Roy, “Traditional pest monitoring methods in crop protection: A review,” Journal of Agricultural Extension, vol. 29, no. 1, pp. 45–53, 2021. [9[ Food and Agriculture Organization (FAO), Pest Surveillance Handbook, Rome, Italy: FAO Publications, 2020. [10] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436–444, May 2015.
Publication Date
5/20/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
0
Total views since publication
PDF Downloads
2
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...