Scholarlink

<p>Heat pump air conditioning systems are widely used in the thermal management of new energy vehicles. Faults in the heat pump air conditioning system can lead to degraded thermal management performance, increased energy consumption, and even safety hazards. To address these issues, a fault diagnosis strategy based on modulated signals and Conv-Bayes (MCB) is proposed. Fault-modulated components in multi-directional vibration signals are extracted and analyzed using demodulation principal component analysis. A data fusion framework based on convolutional neural networks is employed to fusion modulation features from horizontal, vertical, and axial directions, enabling multi-dimensional fault characterization. Furthermore, based on Bayesian theory, a multi-dimensional Gaussian distribution matrix is constructed using the fused data as a bridge, thereby completing the fault diagnosis model. Experimental validation was conducted on four types of faults: condenser fan failure, refrigerant undercharge, refrigerant overcharge, and compressor spindle wear. The MCB model achieves nearly 100% accuracy and significantly outperforms comparative models in terms of precision, recall, and F1-score. Additionally, the proposed method reduces computation time by 75–92%, improving diagnostic efficiency. This approach provides an efficient and reliable solution for real-time fault diagnosis in heat pump air conditioning systems.</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

Research on fault diagnosis strategy of heat pump air conditioning system based on modulation signal and Conv-Bayes

  • 期刊论文

  • 发表日期：

    2025年8月2日

查看全文
  • Yongxing Song,
  • Yanjie Zhao,
  • Qiang Liu,
  • Zhichen Song,
  • Tonghe Zhang,
  • Linhua Zhang

摘要

Heat pump air conditioning systems are widely used in the thermal management of new energy vehicles. Faults in the heat pump air conditioning system can lead to degraded thermal management performance, increased energy consumption, and even safety hazards. To address these issues, a fault diagnosis strategy based on modulated signals and Conv-Bayes (MCB) is proposed. Fault-modulated components in multi-directional vibration signals are extracted and analyzed using demodulation principal component analysis. A data fusion framework based on convolutional neural networks is employed to fusion modulation features from horizontal, vertical, and axial directions, enabling multi-dimensional fault characterization. Furthermore, based on Bayesian theory, a multi-dimensional Gaussian distribution matrix is constructed using the fused data as a bridge, thereby completing the fault diagnosis model. Experimental validation was conducted on four types of faults: condenser fan failure, refrigerant undercharge, refrigerant overcharge, and compressor spindle wear. The MCB model achieves nearly 100% accuracy and significantly outperforms comparative models in terms of precision, recall, and F1-score. Additionally, the proposed method reduces computation time by 75–92%, improving diagnostic efficiency. This approach provides an efficient and reliable solution for real-time fault diagnosis in heat pump air conditioning systems.