Reserch On Measurement Method Of Copper Layer Thickness Of Metal Laminates Based On Deep Learning

With the development of science and technology,the requirements for material properties in various industries are gradually increasing.Metal laminates have been widely used in many fields due to their excellent performance.The thickness and uniformity of the coating material greatly affect the performance of the laminated board,therefore,non-destructive testing of the coating thickness of metal laminated boards is of great significance.At the same time,with the continuous improvement of deep learning models,the use of deep learning technology can solve problems such as identifying and classifying defect types and locating defect positions in non-destructive testing.However,there are still technical issues such as "difficulty in thickness measurement","difficulty in thickness measurement",and "difficulty in prediction" in detecting the thickness of metal laminates.In response to the issue of detecting the thickness of metal laminates,this paper first uses simulation methods to study the effects of the distance between the laser emission center and the receiving point,as well as changes in copper layer thickness,on the dispersion characteristics of R-L type guided waves.A copper aluminum laminates copper layer thickness detection experimental platform is built to verify and analyze the simulation results,and a recurrent neural network model is built based on simulation and experimental data.The R-L type guided wave dispersion characteristics were extracted and the detection of copper layer thickness was achieved.The specific research is as follows:In response to the problem of "thickness measurement difficulty" in multi-layer composite material coatings,and "thickness measurement difficulty" when the coating material thickness is relatively thin.A finite element simulation combined with experimental verification was used to propose an inversion method for copper layer thickness in copper aluminum laminates based on R-L type guided wave dispersion characteristics.A finite element model of laser ultrasonic propagation in copper aluminum laminates was established to study the propagation law of R-L type guided waves in copper aluminum laminates.Research has found that the distance between the laser emitting point and the receiving point affects the dispersion mode of R-L type guided waves.As the distance increases,the distance between the laser emitting center and the receiving point,as well as the thickness of the copper layer,changes and the R-L type guided waves transition from normal dispersion to anomalous dispersion.The optimal distance between the receiving point is 8mm;The thickness variation of the coating material will affect the amplitude and frequency of the R-L type guided wave.When the thickness exceeds 0.2mm,the R-L type guided wave will gradually transform into a surface wave.We have built an experimental platform for thickness detection of copper aluminum laminated plates,and conducted a comparative study on the dispersion mode,amplitude,and frequency changes of R-L guided waves when the distance between the laser emission center and the receiving point,as well as the thickness of the copper layer,were simulated.The results show that the simulation results are consistent with the fitting relationship between the receiving point distance,copper layer thickness,and R-L guided wave amplitude,as well as the dispersion mode transformation process in the experimental results,R-L guided waves can be used to characterize the thickness of thin copper coatings on copper aluminum laminates.In response to the problem of "difficult to predict" the thickness of metal laminates.By pre-processing the simulation and experimental data to reduce the interference of other waveforms on model training,the method of data splitting is used to expand the small data set and expand the sample size to avoid overfitting of the model.A recurrent neural network model based on Bi LSTM-LSTM was constructed,and model parameters were adjusted to improve model performance.The model was trained and predicted using simulation and experimental datasets,and different evaluation criteria were used to evaluate the test results.The results showed that the average accuracy of the model exceeded 97%,achieving accurate prediction of the thickness of thin copper cladding on copper aluminum laminated plates.This article uses a recurrent neural network model to predict the thickness of thin copper coatings on copper aluminum laminates,assisting in the development of thin copper coating thickness detection technology for metal laminates,and providing strong support for the performance reliability of metal laminates in industrial manufacturing and application.