Aero-engine Fault Diagnosis Research Articles

Research on a Fault Diagnosis Method for Aero-Engine Based on Improved SVM and Information Fusion

This paper describes a support vector machine(SVM) approach to improve the test validity and accuracy for Aero-engine fault diagnosis. A new concept called classification rate has been introduced. The paper presents a new information fusion fault diagnosis method based on SVM. The diagnostic decision rules have been improved and applied to aero-engine gas path fault diagnosis. The test result manifests SVM has distinguish limitation to a strong linear correlation of two types of fault samples, it may also have diagnostic difficulties caused by Maximum number of classifications. Therefore, this paper has then proposed a corresponding solution. The simulation results verify that the method is feasible and it can reduce the defect caused by small sample data. It has high capacity of resisting disturbance and high accuracy.

Gas path fault diagnosis system of aero-engine based on Grey Relationship Degree

In consideration of the problem in traditional aero-engine fault diagnosis, a new algorithm was proposed, which does not rely on engine math model nor numerous data samples. This method based on grey relationship degree theory, engine gas path fault symptoms shifts were needed first, and then the relationship degree was calculated between ambiguous faults and specific faults to achieve the fault diagnosis. Finally, the CFM56-3 fault example validates the effectiveness and feasibility of this method.

Research on Identification Method of Multi-Mixed Vibration Signal of Aeroengine

The effectiveness of separation and identification of mechanical signals vibrations is crucial to successful fault diagnosis in the condition monitoring and diagnosis of complex machines.Aeroengine vibration signals always include many complicated components, blind source separation (BSS) provides a efficient way to separate the independent component.In order to get the most effective algorithm of vibration signal separation,experiment has been done to acquire plenty of multi-mixed rotor vibration signals,three sets of vibration data generated from aeroengine rotating shafts were separated from the synthetic vibration signal. The results prove that blind source separation is effective and can be applied for vibration signal processing and fault diagnosis of aeroengine.

Constructing a Web-Based Fuzzy Expert System for Aeroengine Fault Diagnosis

Based on the analysis of current state of aeroengine fault diagnosis, this paper proposes the design and implementation of a web-based fuzzy expert system called WFESAFD, which is focused on the monitoring and fuzzy fault diagnosis of aeroengine. The system architecture and its components, such as web interface layer, monitor, knowledge acquisition and management, integrated data base, reasoning mechanism and knowledge base are illustrated. The detail process of system implementation and practical applications of WFESAFD are presented to illustrate and prove the proposed system’s applicability. Running results show the feasibility and reliability of the framework, which will be helpful to integrate the aeroengine diagnosis knowledge, improve the diagnosis efficiency and decrease the aeroengine diagnosis cost remarkably.

Comparison of RBF Neural Network and Support Vector Machine on Aero-Engine Vibration Fault Diagnosis

RBF neural network and support vector machine (SVM), two Artificial Intelligent (AI) methods, have been extensively applied on machinery fault diagnosis. Aero-engine, as one kind of rotating machine with complex structure and high rotating speed, has complicated vibration faults. As one kind of AI methods, RBF neural network has the advantages of fast learning, high accuracy and strong self-adapting ability. Support vector machine, another AI method, only needs a small quantity of fault data samples to train the classifier and does not need to extract signal features. In this paper, the applications of two AI methods on aero-engine vibration fault diagnosis are introduced. Firstly, the principles and algorithm of both two methods are presented. Secondly the fundamentals of two-shaft aero-engine vibration fault diagnosis are described and gotten the standard fault samples (training samples) and simulation samples (testing samples). Third, two AI methods are applied to the vibration fault diagnosis and obtained the diagnostic results. Finally, the advantages and disadvantages of the two methods are compared such as the computing speed, accuracy of diagnosis and complexity of algorithm, and given a suggestion of selecting the diagnostic methods.