Research On Crack Classification And Recognition Method Based On Eddy Current Signal

With the continuous progress and development of science and technology,country is also stepping forward to become a manufacturing power.In this process,the normal operation of mechanical parts is a solid foundation for its stable forward.However,some parts may suffer from wear and tear under complex working conditions,leading to cracking and even failure,which seriously threatens the safety of people’s lives and property.Therefore,it is urgent to adopt appropriate nondestructive testing methods to effectively inspect and evaluate equipment parts.compared with conventional testing methods,eddy current testing method can not only carry out efficient testing for equipment parts,but also carry out online testing,which can effectively deal with the above situation.In application,it is found that the eddy current testing method has a good quantitative identification effect for shallow cracks within 5mm,but it is difficult to quantitatively identify deep cracks between 5mm and 15 mm.The reason is that the internal branching structure has a great influence on the testing,which makes the quantitative testing effect is not good enough.In order to solve this problem,a quantitative method of multi-fork crack classification is proposed in this paper,and the feasibility and accuracy of this method are verified by experiments.The specific research process is as follows:Firstly,in order to reduce the influence of coil parameters(coil inner diameter,coil outer diameter,number of turns,height,excitation current,excitation frequency,phase distance between coils,lifting height,etc.)on eddy current detection sensitivity.the relationship between eddy current coil parameters and detection sensitivity is studied by ANSYS simulation calculation,and the optimal value and range value of each parameter are obtained based on the principle of high sensitivity in this paper,which laid a foundation for subsequent research.Secondly,in order to find the optimal parameter combination from the excitation coil parameter range under the same sensitivity condition,the response surface method combined with ANSYS simulation calculation to optimize the excitation coil parameters is used in this paper,so as to determine the various parameters of the coil,and applies them to the following paper.Thirdly,in order to quantitatively identify the simplified rectangular crack,a quantitative crack identification method is proposed in this paper.The specific process is as follows: firstly,the optimal parameter combination mentioned above is used to simulate the rectangular crack of different sizes;Secondly,the simulation results are processed by wavelet transform.Then BP neural network is used to quantitatively identify the crack size of processed data.Finally,the feasibility and accuracy of this method are verified by experimental data.Finally,in order to classify and identify the multi-fork crack with complex shape,a classification and identification method is proposed.The specific process is as follows:firstly,the optimal parameter combination above is used to simulate and calculate the multi-fork crack with complex shape;Secondly,the crack classification method constructed in this paper is used to classify the simulation results.Then,the surface width and depth of the classified multi-branch crack are quantitatively identified by using the crack quantitative identification network described above.Finally,the feasibility and effectiveness of the proposed multi-fork crack classification and identification method are verified by the experimental data.