Intelligent Fault Diagnosis of Hydraulic Multi-Way Valve Using the Improved SECNN-GRU Method with mRMR Feature...

Area of Science:

• Mechanical Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Hydraulic multi-way valves are critical in heavy industries but prone to hidden faults due to harsh environments.
• Acquiring true fault data is challenging due to valve cost and experimental replication difficulties.
• Existing fault diagnosis methods struggle with the complexity and hidden nature of these faults.

Purpose of the Study:

• To develop an effective intelligent fault diagnosis method for hydraulic multi-way valves.
• To address the challenges of hidden faults and limited true fault data.
• To improve the accuracy and reliability of fault diagnosis in critical industrial components.

Main Methods:

• Feature extraction using shallow statistical methods and Maximum Relevance Minimum Redundancy (mRMR).
• Spatial feature extraction via Convolutional Neural Network (CNN) with Squeeze-Excitation (SE) blocks for weighted feature enhancement.
• Temporal feature extraction and fusion using Gated Recurrent Unit (GRU) for classification.

Main Results:

• Simulation data achieved an average diagnostic accuracy of 98.94%.
• Experimental data from a directional valve reached an average accuracy of 92.10% (A1 sensor).
• The proposed SECNN-GRU method demonstrated superior stationarity and diagnostic accuracy compared to other algorithms.

Conclusions:

• The SECNN-GRU method provides a feasible and accurate solution for hydraulic multi-way valve fault diagnosis.
• The approach effectively handles complex fault data and improves diagnostic reliability.
• This intelligent method offers significant potential for predictive maintenance in industrial machinery.