발간년도 : [2024]
| 논문정보 |
|
| 논문명(한글) |
[Vol.19, No.4] Comparison of Smart Farm Complex Environment Sensor Failure Prediction Algorithms and Development of Control System |
|
| 논문투고자 |
Jeong Hun Seo, Meong Hun Lee, Hyeon O Choe |
|
| 논문내용 |
This study focuses on developing a failure prediction and diagnosis system for smart farm complex environmental sensors. Smart farm is a technology that improves productivity and resource management efficiency by incorporating the latest technologies such as the Internet of Things, big data, and artificial intelligence into agriculture. Existing sensor failure detection methods have mainly relied on the diagnosis function of the sensor itself or the method of manually checking obvious abnormalities, but these methods are inefficient in large-scale systems and have limitations in early detection. The study goes through the stages of data collection and preprocessing, failure condition definition, model training and evaluation, and system implementation and verification. A model that predicts failures is developed using various machine learning algorithms, and performance indicators such as precision, recall, F1 score, and accuracy are compared and evaluated. The best model is implemented in a real-time monitoring system to visualize the results and build a warning system. Missing value processing, outlier removal, and data normalization are performed using past data from temperature and humidity sensors. Each model is trained and evaluated to derive the optimal failure prediction algorithm. The results of the study show that the CatBoost model showed the best performance in failure prediction and provides high accuracy and reliability. This system can optimize crop growth conditions in smart farms, reduce unnecessary crop loss, and reduce sensor maintenance costs. |
|
| 첨부논문 |
|
|
|
|
|
자료실 > 논문지 
