Electromagnetic commutating valve fault diagnosis method based on physical information neural network

By using a physical information neural network-based method, combined with the physical structure and electromechanical-hydraulic coupling dynamic model of the electromagnetic directional valve, the mechanistic characteristic parameters of the electromagnetic directional valve are inverted and classified. This solves the problems of lack of mechanistic depth and insensitivity to early faults in the existing technology, and achieves high-precision and low-cost fault diagnosis.

CN122432783APending Publication Date: 2026-07-21YANSHAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
YANSHAN UNIV
Filing Date
2026-04-30
Publication Date
2026-07-21

AI Technical Summary

Technical Problem

Existing fault diagnosis methods for electromagnetic directional valves lack mechanistic depth, failing to delve into mechanistic characteristic parameters that cannot be directly measured, such as the viscous friction coefficient of the valve core and the stiffness of the reset spring. Furthermore, they are insensitive to early, minor faults, have limited classification accuracy, rely on a large amount of fault data to train models, exhibit poor generalization ability under small sample conditions, and result in diagnostic results lacking physical meaning and high implementation costs.

Method used

A physical information neural network-based approach is adopted. By constructing a multi-source dataset and combining the physical structure of the electromagnetic directional valve with the electromechanical-hydraulic coupling dynamic model, simulation data is generated. An inversion model is constructed and the physical information neural network is trained to realize the inversion of 5-dimensional mechanism features and fault classification. Wavelet operator algorithm and backpropagation algorithm are used to optimize the model and combine existing sensors for signal acquisition.

Benefits of technology

It enables in-depth diagnosis of the internal physical state of electromagnetic directional valves, accurately locates the root cause of faults, precisely captures early and subtle faults, reduces reliance on fault data, adapts to complex operating conditions, improves diagnostic accuracy and coverage, and reduces implementation costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122432783A_ABST
    Figure CN122432783A_ABST
Patent Text Reader

Abstract

The present application relates to the field of electromagnetic reversing valve fault diagnosis, and particularly relates to a kind of electromagnetic reversing valve fault diagnosis method based on physical information neural network.It includes: S1, constructing multi-source data set: S2, using the physical information neural network model based on wavelet operator algorithm, constructs electromagnetic reversing valve inversion model;S3, using classification algorithm constructs and trains fault classification model;S4, the dynamic response signal of electromagnetic reversing valve to be diagnosed is collected, in turn input to electromagnetic reversing valve inversion model, fault classification model, obtains electromagnetic reversing valve fault diagnosis result.The present application adopts the two-stage diagnostic framework of response data inversion mechanism characteristic parameter+mechanism characteristic parameter fault identification, effectively solves the problems that existing electromagnetic reversing valve diagnosis method lacks mechanism depth, is not sensitive to early weak fault and small sample generalization ability is poor.
Need to check novelty before this filing date? Find Prior Art