Rotating Machinery Fault Diagnosis Based On Acoustic Emission Technology

With the rapid development of modern industry,all kinds of large-scale rotating machinery in the industrial production have a pivotal position.Gear can support,guide and reduce the friction between fixed machinery and sports machinery,which is an important part of rotating machinery.Gear transmission is of the following characteristics,large carrying capacity,high precision of transmission and constant power transmission etc.,so it has been widely used in the modern large-scale machinery and equipment.But because of the complicated structure of gear transmission,and poor working conditions,and other factors,makes it become a easy to the failure of the parts.Once the gear failure will affect the entire operation of equipment,and even lead to the major security incidents.So the development of the gear fault diagnosis technology has a very important significance.Acoustic emission is an effective monitoring and diagnosis technique in mechanical fault diagnosis.It is used to study the state of the object by detecting the stress wave released by the object itself.Acoustic emission technology does not need to be close to the measured object,the deformation of the measured object will not easily cause changes in acoustic emissions.It involves large areas,can detect both macro and the internal changes in the organization of various states.Therefore,acoustic emission technology for mechanical early failure of real-time monitoring and diagnosis is essential.This paper introduces the significance of fault diagnosis of rotary machinery,the principle of acoustic emission technology and the development and application of acoustic emission technology,and summarizes the methods of fault diagnosis of rotating machinery.According to the actual situation of rotating machinery failure,the application of support vector machine on the acoustic emission signal recognition effect of the study.Support vector machine is a machine learning algorithm,which is based on the principle of structural risk minimization,and this principle proves in mathematical theory that the generalization ability of this model can be maximized.In the aspect of experimental data acquisition,the gear data acquisition system is composed of PCI-2 acoustic emission system and rotating machine fault test bed.The gear for a long time wear and groove experiments,through the experimental equipment to observe and collect different fault sound emission signal.The classification model of support vector machine established by experimental data wear data,trench data and normal data is used to verify the classification effect of support vector machine.The RBF function is used as the kernel function of the support vector machine.Three parameters optimization methods are proposed for the RBF kernel function: grid search method,genetic algorithm optimization and particle swarm optimization(PSO).The key parameters of RBF kernel function are analyzed by three parameter optimization methods.The accuracy of the application support vector machine fault classification is about 10% higher than that before optimization.The results show that this method can be used for reference to the acoustic emission signal of rotating machinery.