A method and system for accurately preparing micro-textured slope surface with laser beam

Abstract

The present invention belongs to the field of laser beam processing technology, and discloses a method and system for accurately preparing micro-textured slopes by laser beam. The method establishes a BP neural network prediction model for the morphology of laser-etched materials based on the morphology of micro-pits at different laser powers; inputs the geometric model parameters of the micro-texture to be processed, extracts the characteristics of the micro-texture, and obtains the correction coefficients of the geometric model and the theoretical micro-texture; arranges the input overlap rate, outputs the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate; calculates the total diameter length of the focus coordinates, and determines the method of using method 1 polarizer or method 2 to control the stepper motor for processing; inputs the laser beam processing focus coordinate DXF image in the laser or controls the stepper motor and the laser beam power corresponding to each laser beam processing focus for processing. The present invention automatically controls and adjusts the laser parameters, realizes more accurate focus control, and improves the efficiency and accuracy of laser beam processing micro-textures.

Description

Technical Field

[0001] The present invention belongs to the technical field of laser beam processing, and in particular relates to a method and system for accurately preparing a micro-textured slope surface with a laser beam. Background Art

[0002] Laser beam processing technology has become one of the important means of preparing micro-textures. The laser beam has the advantages of not needing to contact the target object during the processing process, having high processing speed, precision and flexibility, and being able to generate high-quality micro-texture structures on the surfaces of different materials. Therefore, laser beam processing has become the mainstream technology in the field of micro-texture processing. The traditional processing method adjusts the laser beam of the laser through artificial experience, which has poor adaptability to the environment, seriously affecting the precision and processing efficiency of the micro-texture slope, and may affect the wear-resistant performance of the micro-texture itself, and reduce the life of the micro-texture. At the same time, the traditional processing method also has problems such as low processing efficiency, low precision, and poor surface quality. In addition, in the current processing process, due to the problem of micro-texture scale, it is easy to cause inaccurate processing and shape errors, affecting the performance of the micro-texture itself.

[0003] Through the above analysis, the problems and defects of the existing technology are: the micro-texture slope surface accuracy obtained by the traditional laser beam processing method is low, the processing efficiency is low, the wear reduction and wear resistance performance is poor, and the service life is short. Summary of the invention

[0004] In order to overcome the problems existing in the related art, the disclosed embodiment of the present invention provides a method and system for accurately preparing a micro-textured slope surface by a laser beam, and the technical solution is as follows:

[0005] The present invention is achieved by a method for accurately preparing a micro-textured slope surface using a laser beam, comprising:

[0006] S1, based on the micro-pit morphology formed by lasers with different powers, a BP neural network prediction model for the laser ablation material morphology is established;

[0007] S2, input the geometric model parameters of the micro-texture to be processed, extract the characteristics of the micro-texture, and obtain the correction coefficient between the geometric model and the theoretical micro-texture;

[0008] S3, arranging the input overlap ratios, outputting the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate;

[0009] S4, calculating the total diameter length of the focal coordinates, and selecting a polarizer or a method of controlling a stepper motor for processing;

[0010] S5, inputting the laser beam processing focus coordinate DXF image into the laser or controlling the stepper motor to process the laser beam power corresponding to each laser beam processing focus.

[0011] In step S1, a BP neural network prediction model for laser ablation material morphology is established, including:

[0012] Obtain the micro-pit depth and diameter corresponding to the laser beam parameters, and use Python language to input the laser beam parameters, micro-pit depth and diameter into an array; import the preprocessing module in the sklearn library, use sklearn's MinMaxScaler to normalize the data, and save the normalized data;

[0013] The normalized data is divided into a training set and a test set, and the features and labels of the training set and the test set are converted into numpy arrays; the neural network model MTSSULB is used to train the neural network model, and the trained neural network model is used for prediction, and the prediction results are denormalized;

[0014] The training of the neural network model includes establishing an initialization function _init_, feed_forward forward transmission to calculate the input neurons, MSGD function to perform a small batch stochastic gradient descent method, update_WB_by_mini_batch function to use a small sample training set to update the w weight vector and b bias vector, back_propagation function to use the error back propagation algorithm to solve the update amount of the w weight vector and b bias vector for each sample, evaluate function to evaluate the training data, predict function to predict the given test input data, cost_function function to calculate the difference between the model output value and the true value; copy the inverse normalization result y through the inverse_transform_col function, subtract the minimum value used in normalization or standardization from the inverse normalization result y to cancel the translation effect in the normalization process; divide the inverse normalization result y by the scaling factor used in normalization or standardization to cancel the scaling effect in the normalization process; return the inverse normalized y value; establish a BP neural network prediction model and determine the relationship between the laser beam parameters and the depth and diameter of the micro-pits.

[0015] Furthermore, the input layer of the BP neural network prediction model is laser beam power, pit depth and pit diameter, the hidden layer is the extracted correlation features between laser power, diameter and depth, and the output layer is the prediction results of laser beam power, pit depth and pit diameter.

[0016] In step S2, the geometric model parameters of the micro-texture to be processed are input, the characteristics of the micro-texture are extracted, and the correction coefficients of the geometric model and the theoretical micro-texture are obtained, including:

[0017] The NURBS curve is used for fitting to obtain the geometric model, which is expressed as follows:

[0018] ;

[0019] In the formula, is a point on the curve, is the index of the control point, starting from 0 and gradually increasing until , is the number of control points, is the B-Spline basis function, is the weight of the control point, As parameters, is the order of the B-Spline basis function, is the control point; is the weight function;

[0020] Use Python language to establish a function curve_fitting, call the curve_fitting function and input the characteristic points of the micro-texture to be processed, wherein the characteristic points of the micro-texture to be processed include the beginning, the end and the deepest point;

[0021] The feature points are divided into curve fitting and straight line fitting. The curve fitting part uses the np.polyfit function to fit a given polynomial. The straight line fitting part uses plt.scatter and plt.plot to draw the original coordinate points and the fitted straight line respectively, thereby generating an image of the micro-textured slope surface.

[0022] Adjust the order in the polynomial, perform image processing through third-party libraries such as torchvision, PIL, and collections in Pytorch, convert the image path into image data, and convert the image into RGB format; convert the image into a tensor, and scale the pixel value from the integer range [0,255] to the floating point range [0.0,1.0], and determine the correction coefficient through the micro-texture geometry model combined with the theoretical micro-texture , the expression is:

[0023] ;

[0024] In the formula, is the correction factor, is the length of the upper surface of the micro texture in the geometric model, is the upper surface in the theoretical microtexture;

[0025] The real value of the theoretical microtexture is obtained by multiplying the microtexture focus coordinates, microtexture diameter and depth in the microtexture geometric model with the correction coefficient.

[0026] Furthermore, the functions of the function curve_fitting include: generating parameter values ​​of points, x and y values ​​of the fitting curve, generating new parameter values ​​for evaluating the curve, and obtaining new x and y coordinates of the evaluation curve.

[0027] In step S3, the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate are output, including:

[0028] Retrieve the depth from all laser focuses of each upper surface to be processed to the bottom of the slope, predict the spot diameter corresponding to the depth of the upper surface to be processed through the BP neural network, establish the laser beam model corresponding to each focus, and store it in the array S[coordinates, power, diameter, depth] in the form of [focus coordinates, laser beam power, laser beam spot diameter, laser beam depth]; determine the basic position obtained by the first laser beam processing, and traverse the stored laser focus coordinates and diameter arrays; obtain the laser prediction model that best meets the overlap rate through the calculation of the deviation Q, and then obtain the corresponding laser power and laser focus coordinates;

[0029] Each Q value is stored in the newly created error array through the append() function, and the error array is traversed through the built-in function abs in the python language to convert each number in the error array into a positive number; the error array is processed using the argmin function in the python Numpy library to obtain the number with the smallest difference from the overlap rate I; the index i in the array error is obtained through the _index_ function, and the stored laser focus, power, depth and diameter array S[i] corresponding to the index i is retrieved, and the focus coordinates S[i][0], laser beam power S[i][1] and laser diameter S[i][3] stored in S[i] are added to the array to be processed.

[0030] Furthermore, the stored laser focus coordinates and diameter arrays are traversed, and the expression is:

[0031] ;

[0032] In the formula, The deviation caused by arranging the input values ​​according to the overlap ratio is is the diameter corresponding to the previous laser focus, is the diameter of the laser focus, is the input overlap ratio, is the actual coordinate value corresponding to the laser focus, is the radius corresponding to the laser focus.

[0033] In step S4, the total diameter length of the focal coordinate is calculated, and a polarizer or a stepper motor is selected for processing, including:

[0034] Calculate the total diameter length of the focal coordinate D1=S1[0][0]+S1[0][-1], where the maximum value of the polarizer adjustment is D2. The returned maximum value is named Maximum=max(D1,D2). The max function is used to return the maximum value. The control flow statement if is used for judgment. The returned result is D2. The polarizer is used for processing.

[0035] The diameter of the micro-textured slope to be processed is determined by D1, and D1 is compared with the adjustment range of the polarizer. If D1 is smaller than the adjustment range of the polarizer, the ezdxf third-party library in Python is used to create a DXF document object doc, and the model space of the DXF file is obtained by calling the doc.modelspace() function; the array S1[i1][0] is used in a for loop to take all values ​​of i1 from 0 to the total amount of the S1 array minus 1, and the laser beam processing focus is added to the DXF file in combination with the add_point(x,y) function, and the saveas function is used. function to save the DXF file; if D1 is greater than the adjustment range of the polarizer, the stage controls the processing position, a double-displacement XY micro-motion platform controlled by a stepper motor is set at the bottom of the stage, the stepper motor is connected to the controller, and the code for controlling the stepper motor is written in Python; by setting the step angle of the stepper motor, it is set according to the unit of the coordinate axis and the specifications of the stepper motor; according to the proportional relationship between the required moving distance and the focal coordinate system, the number of steps of the stepper motor is calculated; the movement of the stepper motor is controlled by sending instructions, so that the stepper motor moves according to the calculated number of steps.

[0036] In step S5, inputting a DXF image of the laser beam processing focus coordinates into the laser or controlling a stepper motor and the laser beam power corresponding to each laser beam processing focus for processing includes:

[0037] If a polarizing filter is used for processing, the DXF drawing of the laser beam focus to be processed is input into the laser, and the initial direction and position of the laser beam are determined by the information of the DXF drawing; the angle and position of the polarizing filter are adjusted so that the laser beam is properly reflected, refracted and polarized, and finally focused to the position of the laser focus to be processed; the i1 in the power array S1[i1][1] is set from 0 to the total value of the S1 array minus 1, and the obtained laser power is input into the laser for processing;

[0038] If a stepper motor is used for processing, the coordinates in S1[0] are displaced by the stepper motor, and the position and power of the laser beam are adjusted by changing the value of i1 in the array S1[1][i1] from 0 to the total value of the S1 array minus 1, so as to realize the processing operation of the micro-texture geometric morphology.

[0039] Another object of the present invention is to provide a system for accurately preparing micro-textured slopes with a laser beam, the system being used to control the method for accurately preparing micro-textured slopes with a laser beam, the system comprising:

[0040] A prediction model building module is used to establish a BP neural network prediction model for the laser ablation material morphology based on the micro-pit morphology formed by lasers with different powers;

[0041] A correction coefficient calculation module is used to input the geometric model parameters of the micro-texture to be processed, extract the characteristics of the micro-texture, and obtain the correction coefficient of the geometric model and the theoretical micro-texture;

[0042] An overlap ratio arrangement module is used to arrange the input overlap ratios and output the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate;

[0043] A polarizer or stepper motor selection module is used to calculate the total diameter length of the focal coordinates and select a polarizer or a method of controlling a stepper motor for processing;

[0044] The laser processing module is used to input the laser beam processing focus coordinate DXF image into the laser or control the stepper motor to process the laser beam power corresponding to each laser beam processing focus.

[0045] In combination with all the above technical solutions, the beneficial effects of the present invention are as follows:

[0046] The method for accurately preparing a micro-textured slope surface with a laser beam of the present invention establishes a prediction model of a laser-etched material morphology BP neural network model based on the micro-pit morphology at different laser powers; inputs the geometric model parameters of the micro-texture to be processed, extracts the characteristics of the micro-texture, and obtains the correction coefficient of the geometric model and the theoretical micro-texture; arranges them according to the input overlap rate, outputs the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate; determines the method of using a polarizer or controlling a stepper motor by calculating the total diameter length of the focus coordinates; inputs a laser beam processing focus coordinate DXF image into the laser or controls the stepper motor and the laser beam power corresponding to each laser beam processing focus for processing.

[0047] The present invention can obtain the characteristics of micro-textures by inputting the geometric model parameters of the micro-texture to be processed and analyzing the geometric model of the micro-textures. By establishing a neural network algorithm prediction model, the input overlap rate is combined with the micro-texture processing requirements to determine the combination of laser focus coordinates and power values, and the laser parameters can be automatically controlled and adjusted during the processing process to achieve more accurate focus control, eliminate the errors caused by manual parameter adjustment, and improve the efficiency of laser beam processing micro-textures and the accuracy of the surface quality of processed micro-textures. Compared with the traditional laser beam processing micro-texture method, the present invention can improve the processing accuracy, greatly improve its own processing efficiency, and realize cross-platform application by inputting the parameters of different lasers.

[0048] The present invention can adjust the laser parameters according to the needs, reduce the influence of human factors and subjective judgment, and thus effectively improve the accuracy and consistency of the shape of the micro-texture. Through the precise parameter adjustment method of the present invention, common errors in the manual adjustment process can be avoided, ensuring that the shape of the micro-texture meets the requirements. The present invention improves the adjustment efficiency and working stability. The present invention can adjust the laser focus coordinates and power required by different lasers by processing different laser beam parameters, saving the time of manual attempts and improving work efficiency. The method can quickly and accurately adjust the parameters of the laser according to environmental changes and target requirements, eliminating the cumbersome steps and time consumption of manual adjustment. In this way, the adjustment efficiency can be improved, the preparation cycle can be shortened, and the good working stability of the micro-texture can be always maintained under different environmental conditions. Improve the performance and life of the micro-texture and ensure that the shape of the micro-texture is more accurate. Through the precise preparation effect of the present invention, the wear-reducing and wear-resistant properties of the micro-texture can be optimized and the service life of the micro-texture can be extended. The present invention has a good effect in preparing micro-textures with a large aspect ratio slope, is good at processing array patterns with high density and high precision, and has significant advantages in cross-scale preparation of micro-textures. After the technology transformation of the present invention, the precision of micro-texture processing can be improved, and in production and processing, the product quality and processing precision can be improved, the production efficiency can be improved, and the energy consumption and production cost can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The accompanying drawings herein are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description, serve to explain the principles of the present disclosure;

[0050] Figure 1 is a flow chart of a method for accurately preparing a micro-textured slope surface using a laser beam provided in an embodiment of the present invention;

[0051] Figure 2 This is a schematic diagram of a method for accurately preparing a micro-textured slope surface using a laser beam provided in an embodiment of the present invention;

[0052] Figure 3is a schematic diagram of a micro-texture geometric model provided by an embodiment of the present invention;

[0053] Figure 4 is a schematic diagram of the fitting effect provided by an embodiment of the present invention;

[0054] Figure 5 This is a comparison chart between the actual numerical value and the prediction of the BP neural network experiment provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific implementation disclosed below.

[0056] The innovation of the present invention is as follows: based on the different laser powers for the micro-pit morphology, the present invention establishes a prediction model of the laser ablation material morphology BP neural network model; inputs the geometric model parameters of the micro-texture to be processed, extracts the characteristics of the micro-texture, and obtains the correction coefficient of the geometric model and the theoretical micro-texture; arranges by the input overlap rate, outputs the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate; by calculating the total diameter length of the focus coordinates, determines the method of using method 1 polarizer or method 2 to control the stepper motor; inputs the laser beam processing focus coordinate DXF image in the laser or controls the stepper motor and the laser beam power corresponding to each laser beam processing focus for processing. The present invention can realize automatic control and adjustment of laser parameters during processing, can realize more accurate focus control, eliminate the error caused by manual parameter adjustment, and improve the efficiency of laser beam processing micro-texture and the accuracy of processing micro-texture surface quality.

[0057] Embodiment 1, as Figure 1 and Figure 2 As shown, the method for accurately preparing a micro-textured slope surface with a laser beam provided in an embodiment of the present invention specifically includes the following steps:

[0058] S1, based on the micro-pit morphology formed by lasers with different powers, a BP neural network prediction model for the laser ablation material morphology is established;

[0059] Through experiments, the micro-pit depth and diameter corresponding to 1000 sets of laser beam parameters were obtained. The laser beam parameters, micro-pit depth and diameter were input into the array Data using Python. The preprocessing module in the sklearn library was imported, and the data was normalized using sklearn's MinMaxScaler, and the normalized data was saved in the array Data1. The normalized data set was divided into a training set and a test set. Finally, number=600 groups were used as the test set, the first number data samples of Data1 were used as the test set, and the remaining data samples were used as the training set. The features and labels of the training set and the test set were saved in the variables x_train, y_train and x_test, y_test, respectively. The features and labels of the training set and the test set were converted into numpy arrays to facilitate the training and prediction of the neural network model.

[0060] Use the custom neural network model MTSSULB for training. The training data is train_data, where each sample includes input features and output labels. Use the trained neural network model for prediction. The prediction data is test_data, where each sample only includes input features. Convert the prediction result y_predict to an array and perform denormalization to obtain the denormalized prediction result y_pre. Figure 5 As shown in the figure, the prediction accuracy reaches more than 95%, and the BP neural network prediction model is established to establish the relationship between laser beam parameters, depth and diameter. Among them, the input layer of the BP neural network prediction model is the laser beam power, pit depth and pit diameter, the hidden layer is the extracted related features between laser power, diameter and depth, and the output layer is the prediction results of laser beam power, pit depth and pit diameter.

[0061] S2, input the geometric model parameters of the micro-texture to be processed, extract the characteristics of the micro-texture, and obtain the correction coefficient between the geometric model and the theoretical micro-texture;

[0062] like Figure 3 As shown, the NURBS curve is used for fitting to obtain the geometric model, which is expressed as:

[0063] ;

[0064] In the formula, is a point on the curve, is the index of the control point, starting from 0 and gradually increasing until , is the number of control points, is the B-Spline basis function, is the weight of the control point, As parameters, is the order of the B-Spline basis function, is the control point; is the weight function;

[0065] Use Python language to establish a function curve_fitting, call the curve_fitting function and input the characteristic points of the micro-texture to be processed, wherein the characteristic points of the micro-texture to be processed include the beginning, the end and the deepest point;

[0066] The feature points are divided into curve fitting and straight line fitting. The curve fitting part uses the np.polyfit function to fit a given polynomial. The straight line fitting part uses plt.scatter and plt.plot to draw the original coordinate points and the fitted straight line respectively, thereby generating an image of the micro-textured slope surface.

[0067] Adjust the order in the polynomial, perform image processing through third-party libraries such as torchvision, PIL, and collections in Pytorch, convert the image path into image data, and convert the image into RGB format; convert the image into a tensor, and scale the pixel value from the integer range [0,255] to the floating point range [0.0,1.0], and determine the correction coefficient through the micro-texture geometry model combined with the theoretical micro-texture , the expression is:

[0068] ;

[0069] In the formula, is the correction factor, is the length of the upper surface of the micro texture in the geometric model, is the upper surface in the theoretical microtexture;

[0070] The real value of the theoretical microtexture is obtained by multiplying the microtexture focus coordinates, microtexture diameter and depth in the microtexture geometric model with the correction coefficient.

[0071] S3, arranging the input overlap ratios, outputting the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate;

[0072] The depth from all laser focuses of each upper surface to be processed to the bottom of the slope is retrieved, and the spot diameter corresponding to the depth of the upper surface to be processed is predicted through the BP neural network. The laser beam model corresponding to each focus is established and stored in the array S[coordinates,power,diameter,depth] in the form of [focus coordinates, laser beam power, laser beam spot diameter, laser beam depth].

[0073] By determining the basic position obtained by the first laser beam processing, the previously stored laser focus coordinates and diameter arrays are traversed;

[0074] ;

[0075] In the formula, The deviation caused by arranging the input values ​​according to the overlap ratio is is the diameter corresponding to the previous laser focus, is the diameter of the laser focus, is the input overlap ratio, is the actual coordinate value corresponding to the laser focus, is the radius corresponding to the laser focus.

[0076] By calculating the deviation Q, the laser prediction model that best matches the overlap rate can be obtained, and then the corresponding laser power and laser focus coordinates can be obtained.

[0077] The values ​​calculated by the formula are stored in the newly created error array through the append() function for each Q value. The error array is traversed through the built-in function abs in the Python language, and each number in the error array is converted to a positive number. The error array is processed by the argmin function in the Numpy library in Python to obtain the number with the smallest difference from the overlap rate I. The index i in the error array is obtained through the _index_ function, and the laser focus, power, depth, and diameter array S[i] corresponding to the index i is retrieved. The focus coordinates S[i][0], laser beam power S[i][1], and laser diameter S[i][3] stored in S[i] are added to the array to be processed.

[0078] S4, calculating the total diameter length of the focal coordinates, and selecting a polarizer or a method of controlling a stepper motor for processing;

[0079] Calculate the total diameter length of the focal coordinate D1=S1[0][0]+S1[0][-1], where the maximum value of the polarizer adjustment is D2. The returned maximum value is named Maximum=max(D1,D2). The max function is used to return the maximum value. The control flow statement if is used for judgment. The returned result is D2. Therefore, method 1 is used for polarizer processing.

[0080] By determining D1, the diameter of the micro-textured slope to be processed can be determined. By comparing D1 with the adjustment range of the polarizer, if D1 is less than the adjustment range of the polarizer, the ezdxf third-party library in Python is used to create a DXF document object doc, and the model space of the DXF file is obtained by calling the doc.modelspace() function. The array S1[i1][0] is used in a for loop to take all the values ​​of i1 from 0 to the total amount of the S1 array minus 1, and the laser beam processing focus is added to the DXF file in combination with the add_point(x,y) function. The DXF file is saved using the saveas function. If D1 is greater than the adjustment range of the polarizer, the stage controls the processing position, and a dual-displacement XY micro-motion platform controlled by a stepper motor is set at the bottom of the stage. The stepper motor is connected to the controller, and the code for controlling the stepper motor is written in Python. By setting the step angle of the stepper motor, it is set according to the unit of the coordinate axis and the specifications of the stepper motor. According to the proportional relationship between the required moving distance and the focus coordinate system, the number of steps that the stepper motor needs to rotate is calculated. The movement of the stepper motor is controlled by sending instructions so that it moves according to the calculated number of steps.

[0081] S5, inputting the laser beam processing focus coordinate DXF image into the laser or controlling the stepper motor to process the laser beam power corresponding to each laser beam processing focus;

[0082] If the polarizer is used in method 1, the laser beam focus DXF drawing to be processed is input into the laser, the initial direction and position of the laser beam are determined by the information of the DXF drawing, and the angle and position of the polarizer are adjusted so that the laser beam can be properly reflected, refracted and polarized, and finally focused to the position of the laser focus to be processed. The laser power obtained is input into the laser for processing by taking all the values ​​of i1 in the power array S1[i1][1] from 0 to the total amount of the S1 array minus 1.

[0083] If method 2 of stepper motor processing is adopted, the coordinates in S1[0] are displaced by the stepper motor and the position and power of the laser beam are adjusted by adjusting the values ​​of i1 in the array S1[1][i1] from 0 to the total value of the S1 array minus 1, so as to achieve precise processing operations.

[0084] By combining the two methods, high-quality, high-efficiency and high-precision micro-texture geometric morphology can be achieved.

[0085] Embodiment 2, the system for accurately preparing micro-textured slope surface with laser beam provided in the embodiment of the present invention comprises:

[0086] A prediction model building module is used to establish a BP neural network prediction model for the laser ablation material morphology based on the micro-pit morphology formed by lasers with different powers;

[0087] A correction coefficient calculation module is used to input the geometric model parameters of the micro-texture to be processed, extract the characteristics of the micro-texture, and obtain the correction coefficient of the geometric model and the theoretical micro-texture;

[0088] An overlap ratio arrangement module is used to arrange the input overlap ratios and output the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate;

[0089] A polarizer or stepper motor selection module is used to calculate the total diameter length of the focal coordinates and select a polarizer or a method of controlling a stepper motor for processing;

[0090] The laser processing module is used to input the laser beam processing focus coordinate DXF image into the laser or control the stepper motor to process the laser beam power corresponding to each laser beam processing focus.

[0091] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principles of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for accurately preparing micro-textured slope surface by laser beam, characterized in that: The method includes: S1, based on the micro-pit morphology formed by lasers with different powers, a BP neural network prediction model for the laser ablation material morphology is established; S2, input the geometric model parameters of the micro-texture to be processed, extract the characteristics of the micro-texture, and obtain the correction coefficient between the geometric model and the theoretical micro-texture; S3, arranging the input overlap ratios, outputting the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate; S4, calculating the total diameter length of the focal coordinates, and selecting a polarizer or a method of controlling a stepper motor for processing; S5, inputting the laser beam processing focus coordinate DXF image into the laser or controlling the stepper motor to process the laser beam power corresponding to each laser beam processing focus; In step S3, the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate are output, including: Retrieve the depth from all laser focuses of each upper surface to be processed to the bottom of the slope, predict the spot diameter corresponding to the depth of the upper surface to be processed through the BP neural network, establish the laser beam model corresponding to each focus, and store it in the array S[coordinates, power, diameter, depth] in the form of [focus coordinates, laser beam power, laser beam spot diameter, laser beam depth]; determine the basic position obtained by the first laser beam processing, and traverse the stored laser focus coordinates and diameter arrays; obtain the laser prediction model that best meets the overlap rate through the calculation of the deviation Q, and then obtain the corresponding laser power and laser focus coordinates; Each Q value is stored in the newly created error array through the append() function, and the error array is traversed through the built-in function abs in the Python language to convert each number in the error array into a positive number; the error array is processed using the argmin function in the Python Numpy library to obtain the number with the smallest difference from the overlap rate I; the index i in the array error is obtained through the _index_ function, and the stored laser focus, power, depth and diameter array S[i] corresponding to the index i is retrieved, and the focus coordinate S[i][0], laser beam power S[i][1] and laser diameter S[i][3] stored in S[i] are added to the array to be processed; Traverse the stored laser focus coordinates and diameter arrays, the expression is: ； In the formula, The deviation caused by arranging the input values ​​according to the overlap ratio is is the diameter corresponding to the previous laser focus, is the diameter of the laser focus, is the input overlap ratio, is the actual coordinate value corresponding to the laser focus, is the radius corresponding to the laser focus; In step S4, the total diameter length of the focal coordinate is calculated, and a polarizer or a stepper motor is selected for processing, including: Calculate the total diameter length of the focal coordinate D1=S1[0][0]+S1[0][-1], where the maximum value of the polarizer adjustment is D2. The returned maximum value is named Maximum=max(D1,D2). The max function is used to return the maximum value. The control flow statement if is used for judgment. The returned result is D2. The polarizer is used for processing. The diameter of the micro-textured slope to be processed is determined by D1, and D1 is compared with the adjustment range of the polarizer. If D1 is smaller than the adjustment range of the polarizer, the ezdxf third-party library in Python is used to create a DXF document object doc, and the model space of the DXF file is obtained by calling the doc.modelspace() function; the array S1[i1][0] is used in a for loop to take all values ​​of i1 from 0 to the total amount of the S1 array minus 1, and the laser beam processing focus is added to the DXF file in combination with the add_point(x,y) function, and the saveas function is used. function to save the DXF file; if D1 is greater than the adjustment range of the polarizer, the stage controls the processing position, a double-displacement XY micro-motion platform controlled by a stepper motor is set at the bottom of the stage, the stepper motor is connected to the controller, and the code for controlling the stepper motor is written in Python; the step angle of the stepper motor is set according to the unit of the coordinate axis and the specifications of the stepper motor; the number of steps of the stepper motor is calculated according to the proportional relationship between the required moving distance and the focal coordinate system; the movement of the stepper motor is controlled by sending instructions, so that the stepper motor moves according to the calculated number of steps; In step S5, a DXF image of the laser beam processing focus coordinates is input into the laser or a stepper motor is controlled to process the laser beam power corresponding to each laser beam processing focus, including: If a polarizing filter is used for processing, the DXF drawing of the laser beam focus to be processed is input into the laser, and the initial direction and position of the laser beam are determined by the information of the DXF drawing; the angle and position of the polarizing filter are adjusted so that the laser beam is properly reflected, refracted and polarized, and finally focused to the position of the laser focus to be processed; the i1 in the power array S1[i1][1] is set from 0 to the total value of the S1 array minus 1, and the obtained laser power is input into the laser for processing; If a stepper motor is used for processing, the coordinates in S1[0] are displaced by the stepper motor, and the position and power of the laser beam are adjusted by changing the values ​​of i1 in the array S1[1][i1] from 0 to the total value of the S1 array minus 1, so as to realize the processing operation of the micro-texture geometric morphology.

2. The method for accurately preparing micro-textured slope surface by laser beam according to claim 1, characterized in that: In step S1, a BP neural network prediction model for laser ablation material morphology is established, including: Obtain the micro-pit depth and diameter corresponding to the laser beam parameters, and use Python language to input the laser beam parameters, micro-pit depth and diameter into an array; import the preprocessing module in the sklearn library, use sklearn's MinMaxScaler to normalize the data, and save the normalized data; The normalized data is divided into a training set and a test set, and the features and labels of the training set and the test set are converted into numpy arrays; the neural network model MTSSULB is used to train the neural network model, and the trained neural network model is used for prediction, and the prediction results are denormalized; The training of the neural network model includes establishing an initialization function _init_, feed_forward forward transmission to calculate the input neurons, MSGD function to perform a small batch stochastic gradient descent method, update_WB_by_mini_batch function to update the w weight vector and b bias vector using a small sample training set, back_propagation function to solve the update amount of the w weight vector and b bias vector for each sample using an error back propagation algorithm, evaluate function to evaluate the training data, predict function to predict the given test input data, cost_function function to calculate the difference between the model output value and the true value; copy the inverse normalization result y through the inverse_transform_col function, subtract the minimum value used in normalization or standardization from the inverse normalization result y to cancel the translation effect in the normalization process; divide the inverse normalization result y by the scaling factor used in normalization or standardization to cancel the scaling effect in the normalization process; return the inverse normalized y value; establish a BP neural network prediction model and determine the relationship between the laser beam parameters and the depth and diameter of the micro-pits.

3. The method for accurately preparing micro-textured slope surface by laser beam according to claim 2, characterized in that: The input layer of the BP neural network prediction model is the laser beam power, pit depth and pit diameter, the hidden layer is the extracted correlation features between laser power, diameter and depth, and the output layer is the prediction results of laser beam power, pit depth and pit diameter.

4. The method for accurately preparing micro-textured slope surface by laser beam according to claim 1, characterized in that: In step S2, the geometric model parameters of the micro-texture to be processed are input, the characteristics of the micro-texture are extracted, and the correction coefficients of the geometric model and the theoretical micro-texture are obtained, including: The NURBS curve is used for fitting to obtain the geometric model, which is expressed as follows: ； In the formula, is a point on the curve, is the index of the control point, starting from 0 and gradually increasing until , is the number of control points, is the B-Spline basis function, is the weight of the control point, As parameters, is the order of the B-Spline basis function, is the control point; is the weight function; Use Python language to establish a function curve_fitting, call the curve_fitting function and input the characteristic points of the micro-texture to be processed, wherein the characteristic points of the micro-texture to be processed include the beginning, the end and the deepest point; The feature points are divided into curve fitting and straight line fitting. The curve fitting part uses the np.polyfit function to fit a given polynomial. The straight line fitting part uses plt.scatter and plt.plot to draw the original coordinate points and the fitted straight line respectively, thereby generating an image of the micro-textured slope surface. Adjust the order in the polynomial, perform image processing through third-party libraries such as torchvision, PIL, and collections in Pytorch, convert the image path into image data, and convert the image into RGB format; convert the image into a tensor, and scale the pixel value from the integer range [0,255] to the floating point range [0.0,1.0], and determine the correction coefficient through the micro-texture geometry model combined with the theoretical micro-texture , the expression is: ； In the formula, Correction factor, is the length of the upper surface of the micro texture in the geometric model, is the upper surface in the theoretical microtexture; The real value of the theoretical microtexture is obtained by multiplying the microtexture focus coordinates, microtexture diameter and depth in the microtexture geometric model with the correction coefficient.

5. The method for accurately preparing micro-textured slope surface by laser beam according to claim 4, characterized in that: The functions of the function curve_fitting include: generating parameter values ​​of points, fitting the x and y values ​​of the curve, generating new parameter values ​​for evaluating the curve, and obtaining new x and y coordinates of the evaluated curve.

6. A system for accurately preparing micro-textured slope surfaces using a laser beam, characterized in that: The system is used to control the method for accurately preparing a micro-textured slope surface with a laser beam as described in any one of claims 1 to 5, and the system comprises: A prediction model building module is used to establish a BP neural network prediction model for the laser ablation material morphology based on the micro-pit morphology formed by lasers with different powers; A correction coefficient calculation module is used to input the geometric model parameters of the micro-texture to be processed, extract the characteristics of the micro-texture, and obtain the correction coefficient of the geometric model and the theoretical micro-texture; An overlap ratio arrangement module is used to arrange the input overlap ratios and output the laser beam processing focus coordinates of the micro-texture to be processed and the power corresponding to each focus coordinate; A polarizer or stepper motor selection module is used to calculate the total diameter length of the focal coordinates and select a polarizer or a method of controlling a stepper motor for processing; The laser processing module is used to input the laser beam processing focus coordinate DXF image into the laser or control the stepper motor to process the laser beam power corresponding to each laser beam processing focus.

Images