Scholarlink

<p>In emergency response systems, firefighting pumps are essential, and their unplanned failures can have disastrous results. Current monitoring techniques frequently do not adequately capture complex temporal and geographical failure patterns under operational unpredictability in the actual world. This study develops a sophisticated Artificial Intelligence of Things (AIoT) framework for monitoring and troubleshooting firefighting pumps, an essential component of emergency response systems. The suggested framework makes use of IPSO-BiGRU-CNN, a hybrid deep learning model that combines a Bidirectional Gated Recurrent Unit-Convolutional Neural Network (BiGRU-CNN) with Improved Particle Swarm Optimization (IPSO). By adjusting the BiGRU-CNN model’s parameters, the IPSO method improves diagnostic precision and resilience in challenging real-world situations. To evaluate the performance of the IPSO-BiGRU-CNN, extensive experiments were conducted on failure datasets, comparing the proposed model with several traditional techniques, including Recurrent Neural Network, GRU, Long Short-Term Memory, BiGRU, CNN, and CNN-BiGRU. Results demonstrate that the IPSO-BiGRU-CNN model outperforms these traditional models with supreme improvements of 22.96% loss, 65.45% ValLoss, 16.89% CP, 93.76% ValCP, 17.39% MAE, 46.22% ValMAE, 23.68% MSE, 63.12% ValMSE, 4.46% PRE, 4.72% ValPRE, 4.73% REC, and 4.83% ValREC. This model’s adaptability to different data sampling rates further increases its usefulness for real-time AIoT applications. The suggested framework provides a dependable and effective method for firefighting pump preventative maintenance, allowing for prompt failure diagnosis and minimizing downtime. This study underscores the potential of combining metaheuristic optimization with deep learning for intelligent fault diagnosis in critical infrastructure.</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 End-to-End AIoT Maintenance Framework for Fighting Pumps Failure Monitoring Based on Metaheuristic Improved Particle Swarm Algorithm and Combining BiGRU-CNN Models

  • 期刊论文

  • 发表日期：

    2025年11月6日

查看全文
  • Thanh-Phuong Nguyen

摘要

In emergency response systems, firefighting pumps are essential, and their unplanned failures can have disastrous results. Current monitoring techniques frequently do not adequately capture complex temporal and geographical failure patterns under operational unpredictability in the actual world. This study develops a sophisticated Artificial Intelligence of Things (AIoT) framework for monitoring and troubleshooting firefighting pumps, an essential component of emergency response systems. The suggested framework makes use of IPSO-BiGRU-CNN, a hybrid deep learning model that combines a Bidirectional Gated Recurrent Unit-Convolutional Neural Network (BiGRU-CNN) with Improved Particle Swarm Optimization (IPSO). By adjusting the BiGRU-CNN model’s parameters, the IPSO method improves diagnostic precision and resilience in challenging real-world situations. To evaluate the performance of the IPSO-BiGRU-CNN, extensive experiments were conducted on failure datasets, comparing the proposed model with several traditional techniques, including Recurrent Neural Network, GRU, Long Short-Term Memory, BiGRU, CNN, and CNN-BiGRU. Results demonstrate that the IPSO-BiGRU-CNN model outperforms these traditional models with supreme improvements of 22.96% loss, 65.45% ValLoss, 16.89% CP, 93.76% ValCP, 17.39% MAE, 46.22% ValMAE, 23.68% MSE, 63.12% ValMSE, 4.46% PRE, 4.72% ValPRE, 4.73% REC, and 4.83% ValREC. This model’s adaptability to different data sampling rates further increases its usefulness for real-time AIoT applications. The suggested framework provides a dependable and effective method for firefighting pump preventative maintenance, allowing for prompt failure diagnosis and minimizing downtime. This study underscores the potential of combining metaheuristic optimization with deep learning for intelligent fault diagnosis in critical infrastructure.