Abstract
Firefighting pumps provide vital roles in maintaining the pressure of fluid power in the firefighting system and are broadly utilized in the majority of residential, commercial, or industrial buildings. However, the reliance on periodic inspections to ensure their availability during emergency scenarios poses challenges such as high maintenance costs and the need for expert knowledge. This study addresses these challenges by proposing a cloud-based AIoT intelligent infrastructure for diagnosing faults and identifying potential hidden failure conditions in firefighting pumps. The framework integrates IoT devices with different sensors installed on pumps to accumulate real-time data under normal and failure conditions. A hybrid convolutional neural network-gate recurrent unit (CNN-GRU) deep learning model is constructed to analyze this data, leveraging hyperparameter optimization to enhance performance. Through the hyperparameter process, a hybrid CNN-GRU algorithm is constructed for analyzing and validating with other traditional approaches, including the recurrent unit (RNN), long short-term memory (LSTM), GRU, CNN, and CNN-RNN methods. The experiments demonstrate that the AIoT framework-based CNN-GRU algorithm could accurately provide intelligent fault diagnosis of firefighting pumps and effectively decrease the operation and maintenance cost of firefighting pump manufacturing companies in Taiwan. This innovative approach offers a practical solution for firefighting pump manufacturers and has broader implications for smart maintenance systems in critical infrastructure.
Thông tin:
| Thuộc danh mục | SCIE |
| Tạp chí | The Journal of Supercomputing |
| Nhà xuất bản | Springer |
| Mã định danh bài | 462 |
| Tập | 81 |
| Số | |
| Ngày xuất bản | 04/02/2025 |
| DOI | https://doi.org/10.1007/s11227-025-06965-w |