Scholarlink

<p>The cycloidal hydraulic motor is a critical actuating component in hydraulic systems. However, due to harsh operating conditions, it is prone to malfunctions, which directly compromise the safety and reliability of the hydraulic system. Therefore, achieving accurate fault diagnosis for cycloidal motors is of paramount importance. To address the issue of fault diagnosis and identification in cycloidal motors, a fault diagnosis method combining fusion feature vector and particle swarm optimization-backpropagation neural network(PSO-BPNN) is proposed. First, the variational mode decomposition (VMD) method is utilized to decompose and reconstruct vibration signals for noise reduction, leveraging its superior capability in processing non-stationary signals. Subsequently, sensitivity analysis (SA) is applied to construct a vibration feature vector <i>V</i>, which serves as the input for the model. Next, This study proposes a fault diagnosis model combining PSO with a BP neural network. By leveraging the global optimization capability of PSO, the initial weights and thresholds of the BP neural network are optimized. This approach effectively mitigates the issue of the BP network converging to local optima while significantly improving diagnostic accuracy. This paper designs a fused feature vector <i>E</i> by concatenating the vibration feature vector <i>V</i> and the pressure feature vector <i>P</i> in a vector stacking manner. Finally, <i>V</i>, <i>P</i>, and <i>E</i> are used as inputs to the PSO-BP neural network model, and 10 trials are conducted for each. The results demonstrate that, compared to the vibration feature vector and pressure feature vector, the fused feature vector <i>E</i> better reflects the fault characteristics of the cycloidal motor.</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

A fault diagnosis method for cycloidal hydraulic motors based on improved BP neural network and fusion feature vectors

  • 期刊论文

  • 发表日期：

    2025年9月2日

查看全文
  • Bing Zhang,
  • Pengzhi Han,
  • Hongyu Liu,
  • Lijie Zhang,
  • Yonggan Wang,
  • Xinrui Tang,
  • Jie Kong

摘要

The cycloidal hydraulic motor is a critical actuating component in hydraulic systems. However, due to harsh operating conditions, it is prone to malfunctions, which directly compromise the safety and reliability of the hydraulic system. Therefore, achieving accurate fault diagnosis for cycloidal motors is of paramount importance. To address the issue of fault diagnosis and identification in cycloidal motors, a fault diagnosis method combining fusion feature vector and particle swarm optimization-backpropagation neural network(PSO-BPNN) is proposed. First, the variational mode decomposition (VMD) method is utilized to decompose and reconstruct vibration signals for noise reduction, leveraging its superior capability in processing non-stationary signals. Subsequently, sensitivity analysis (SA) is applied to construct a vibration feature vector V, which serves as the input for the model. Next, This study proposes a fault diagnosis model combining PSO with a BP neural network. By leveraging the global optimization capability of PSO, the initial weights and thresholds of the BP neural network are optimized. This approach effectively mitigates the issue of the BP network converging to local optima while significantly improving diagnostic accuracy. This paper designs a fused feature vector E by concatenating the vibration feature vector V and the pressure feature vector P in a vector stacking manner. Finally, V, P, and E are used as inputs to the PSO-BP neural network model, and 10 trials are conducted for each. The results demonstrate that, compared to the vibration feature vector and pressure feature vector, the fused feature vector E better reflects the fault characteristics of the cycloidal motor.