Research On Laser Ultrasonic Nondestructive Testing Method For Grain Size Of Metal Strips

As an important parameter of the internal microstructure of the metal,the grain size of polycrystalline material has an important influence on the mechanical and electromagnetic properties of the material.On-line rapid non-destructive testing of the grain size of metal sheets and strips is a necessary means to improve material properties and realize intelligent production of metal sheets and strips.It is of great significance to achieve high stability and the accurate hit of product mechanical properties through online detection and real-time control of microscopic grain structure.However,for a long time,due to the lack of online direct detection technology of metal internal structure performance parameters,only offline and destructive methods can be used for sampling detection(such as the metallographic method,electron backscatter diffraction method,etc.).Although accurate microstructure performance parameters can be obtained,the efficiency is low and the time period is long.It is difficult to identify and intervene and adjust the sudden product quality fluctuations in a timely manner,.which becomes a bottleneck for the improvement of internal quality and stability of the metal sheet and strip in the production process.Based on laser ultrasonic nondestructive testing technology,taking pure aluminum plate as an example,this paper studies the detection method of microscopic grain size and distribution of metal sheet and strip products,builds a set of laser ultrasonic nondestructive testing platform and optimizes the nondestructive excitation detection method,attenuation characteristic value extraction,and characterization model.Non-destructive testing of grain size and its distribution was achieved.The main contents and research results of this paper are as follows:(1)A set of laser ultrasonic testing platform for the grain size of metal materials was built.The ultrasonic wave was excited inside the metal aluminum plate by a pulsed laser,and the ultrasonic signal carrying the grain size information was received on the opposite side of the metal plate with a two-wave mixing interferometer.The long-distance excitation and reception of ultrasonic waves were realized.When the power of the continuous laser is set to 800mw,the interferometer can achieve better interference effect and avoid damage to the interferometer components.The high-speed signal acquisition system based on the PXIe bus was adopted,and the sampling rate can reach 2.5GS/s.The data was accessed by the method of the queue.It can match the high-speed acquisition and storage of ultrasonic signals excited by 10Hz.The MATLAB program was embedded to realize the real-time high-speed continuous acquisition,storage,and analysis of ultrasonic signals.A non-destructive testing platform based on a ring pulse laser was built,and the non-destructive testing optimization of the laser ultrasonic testing system was realized.(2)For the problem of surface damage caused by the ablation mechanism in the traditional laser ultrasonic detection of grain size,the excitation of the thermoelastic nondestructive testing mechanism of ring pulsed laser ultrasound was realized by using the optical component axicon.And the nondestructive signal was detected by the two-wave mixing interferometer.Using the established finite element model of the ring laser excitation and the ring pulse laser ultrasound experiment platform,the directionality of ultrasonic energy was analyzed,and the angle of 35.5 degrees was the best angle.For a 3 mm sample,4(33.69 degrees)to 5(39.81 degrees)mm diameter of the laser ring were used as suitable detection parameters for the experiments.The width of the ring was chosen to be 0.2mm.The ultrasonic focusing effect of the ring excitation and the influence of the size parameters of the ring pulsed laser on the ultrasonic signal were analyzed.According to the simulation and experimental results,the optimal ring pulsed laser parameters were obtained.Through theoretical calculation,a grain size detection model under the thermoelastic mechanism was established,which simultaneously solved the problems of low energy and the directionality of ultrasonic intensity in the thermoelastic nondestructive mechanism.(3)The ultrasonic signal obtained by the laser ultrasonic system was decomposed and reconstructed through the ensemble empirical mode decomposition algorithm.Through the frequency domain analysis of Intrinsic Mode Functions of different orders,an ensemble empirical mode decomposition algorithm was used to decompose the ultrasonic signal to eliminate mode mixing.By analyzing the correlation between the ultrasonic signal and the Intrinsic Mode Functions of different orders,the high-frequency noise and low-frequency linear trend were discarded,and the key signal components were extracted and the signal was recombined,which realized the high-frequency noise reduction and the removal of the low-frequency linear trend of the ultrasonic signal.Using the ultrasonic attenuation information obtained from the laser ultrasonic experiment and the grain size information obtained by EBSD,a prediction model of the average grain size was established.Through experimental analysis,the average error of the established model on the grain size was 2.96%.The effectiveness of the ensemble empirical mode decomposition algorithm for the prediction model of laser ultrasonic signal and grain size was verified.(4)A detection model of microscopic grain size distribution based on an artificial neural network was established.Through the Fourier transform frequency domain calculation method,the acquired laser ultrasonic signal was preprocessed to obtain the spectral attenuation information of the ultrasonic signal,and the ultrasonic attenuation information was used as an input parameter.The grain size distribution information of different 1060 aluminum samples obtained by the EBSD experiment was extracted,and the mean and variance information of grain size distribution were used as the output parameters of the prediction system.The artificial neural network system was established by using the input and output information,and the initial parameters of the artificial neural network system were optimized by particle swarm optimization.The artificial neural network was trained by using the input and output information,and the representation model of the grain size distribution was established.The accuracy of the established grain size distribution model was verified by using the laser ultrasonic experimental data.The mean absolute values of the expected value μ and standard deviation σ were 3.15%and 2.10%,respectively.