Adaptive parameter adjustment method, system and laser marking equipment

Abstract

This invention discloses an adaptive parameter tuning method, system, and laser marking equipment. The method includes the following steps: constructing a neural network model; training the neural network model using marking parameter samples of laser marking materials to obtain a target network model; inputting material parameters and marking target values ​​as input parameters into the target network model, performing iterative calculations to obtain the optimal process parameters of the laser marking equipment; adaptively adjusting the laser marking equipment using the obtained optimal process parameters; deriving and calculating the marking effect value using the target network model, and calculating the color difference value using a color space color difference function, thereby obtaining the optimal solution for the process parameters and adaptively adjusting the parameters of the laser marking equipment. This solves the problems of requiring employees to spend a lot of time and energy debugging laser marking equipment parameters, the inability to standardize related skills, and the difficulty in technology transfer.

Description

Technical Field

[0001] This invention relates to the field of laser processing technology, and in particular to an adaptive parameter tuning method, system, and laser marking equipment. Background Technology

[0002] Laser processing is the largest application area in the laser industry. With rapid economic and technological development, high-end manufacturing demands increasingly higher precision and quality from processed products. Laser processing, with its high efficiency and precision, has replaced traditional processing methods in many fields and has gained widespread application. Laser processing technology is widely used in consumer electronics, new energy, and semiconductor industries. As a major manufacturing country in consumer electronics and new energy, China is also seeing increasingly widespread application of laser processing across various sectors domestically.

[0003] Laser manufacturing equipment is an advanced manufacturing tool integrating optics, mechanics, electronics, software, and processes. Controlling the production process and application technology requires extensive experience and skill accumulation. Currently, the challenges and difficulties in the field of industrial laser intelligent manufacturing include:

[0004] In their daily work, laser application engineers often rely on experience to determine laser parameters and adjust the effect when testing products from customers in different industries. Since different industries involve a wide variety of products and evaluation standards, which is very complicated, this process consumes a lot of time and energy, and there are many human factors that affect the judgment.

[0005] The accumulation and transmission of these debugging skills suffer from severe path dependence. The experience of senior employees cannot be effectively passed on, and they all need to be trained by apprentices, which is time-consuming and laborious.

[0006] The production process and application technology of laser manufacturing equipment require a certain amount of experience and skill accumulation. Employees cannot form an objective and unique standard, and experience and skills are difficult to standardize, resulting in problems with technology transfer. Summary of the Invention

[0007] In existing technologies, the parameters of laser marking equipment are adjusted based on the experience of employees. The relevant skills cannot be standardized, which consumes a lot of time and energy and also poses a problem of technology transfer.

[0008] To address the aforementioned issues, an adaptive parameter tuning method, system, and laser marking equipment are proposed. This method constructs neural network models for various marking materials, trains these models using collected marking parameter samples, derives and calculates the marking effect value from the target network model, and calculates the color difference using a color space color difference function. This yields the optimal solution for the process parameters, enabling adaptive parameter adjustment of the laser marking equipment. This solves the problems of requiring employees to spend significant time and effort debugging laser marking equipment parameters, the lack of standardization of related skills, and difficulties in technology transfer.

[0009] Firstly, an adaptive parameter tuning method includes the following steps:

[0010] A neural network model is constructed, and the neural network model is trained using marking parameter samples of laser marking material to obtain the target network model;

[0011] The material parameters and the target value for marking are input into the target network model as input parameters, and iterative calculations are performed to obtain the optimal process parameters for the laser marking equipment.

[0012] The laser marking equipment is adaptively adjusted using the obtained optimal process parameters;

[0013] The marking parameters include material parameters, marking effect values, and corresponding process parameters.

[0014] In conjunction with the adaptive parameter tuning method described in the first aspect of the present invention, in a first possible implementation, the step of: inputting material parameters and the target marking value as input parameters into the target network model, performing iterative calculations, and obtaining the optimal process parameters of the laser marking equipment includes:

[0015] The color difference value in the color space is calculated using the fitness function;

[0016] Obtain the color difference between the color space color difference value and the target value;

[0017] If the color difference is less than the specified threshold, then the process parameters corresponding to the marking effect value are the optimal solution, and the iteration stops.

[0018] If the color difference value is greater than a specified threshold, the marking parameters of this iteration are mutated to obtain a new parameter population to participate in the next iteration.

[0019] In conjunction with the first possible embodiment of the first aspect of the present invention, in the second possible embodiment, the step of calculating the color space color difference value using a fitness function includes:

[0020] The labeling effect value is calculated using the target network model.

[0021] Calculate the marking difference between the marking effect value and the target marking value;

[0022] Using the color difference function formula in color space:

[0023] = (1)

[0024] Calculate the color difference value of the color space;

[0025] in, Indicates psychological brightness. =L (Effect Value) - L (Target Value) Positive values ​​indicate a brighter result, and negative values ​​indicate a darker result;

[0026] For psychological chromaticity, = a (Effect Value) - a (Target Value) Positive values ​​indicate a reddish tint, and negative values ​​indicate a greenish tint;

[0027] For psychological chromaticity, = b (effect value) - b (target value). A positive value indicates a yellowish tint, and a negative value indicates a bluish tint.

[0028] In conjunction with the first possible implementation of the first aspect of the present invention, in the third possible implementation, the step of: if the color difference value is greater than a predetermined threshold, then performing mutation processing on the marking parameters of this iteration to obtain a new parameter population to participate in the next iteration includes:

[0029] Using formula (2)

[0030] (2)

[0031] Calculate the selection probability of an individual among the labeling parameters, where, Let represent the fitness of the i-th individual. Let be the probability that the i-th individual is selected;

[0032] The selected parameters are randomly cross-referenced and mutated at random locations to form a new parameter population.

[0033] In conjunction with the third possible implementation of the first aspect of the present invention, in the fourth possible implementation, before the step of calculating the selection probability of an individual in the marking parameters, the step of: if the color difference value is greater than a predetermined threshold, then performing mutation processing on the marking parameters of this iteration to obtain a new parameter population to participate in the next iteration, further includes:

[0034] The optimal parameter population for this iteration is retained.

[0035] In conjunction with the fourth possible implementation of the first aspect of the present invention, in the fifth possible implementation, before the step of calculating the color space color difference value using a fitness function, the step of inputting material parameters and the marking target value as input parameters into the target network model, performing iterative calculations, and obtaining the optimal process parameters of the laser marking equipment, further includes:

[0036] Classify the process parameters of laser marking equipment and obtain empirical parameters;

[0037] The empirical parameters of the marking material are obtained using the target network model.

[0038] Among them, the empirical parameters are those that have a relatively large impact on the marking effect and require multiple adjustments by process engineers to determine.

[0039] In conjunction with the fourth possible implementation of the first aspect of the present invention, in the sixth possible implementation, before the step of classifying the process parameters of the laser marking equipment and obtaining empirical parameters, the step of inputting material parameters and marking target values ​​as input parameters into the target network model, performing iterative calculations, and obtaining the optimal process parameters of the laser marking equipment, further includes:

[0040] Set the maximum number of iterations for the target network model.

[0041] Secondly, an adaptive parameter tuning system, employing the method described in the first aspect, includes:

[0042] Training and learning modules;

[0043] Adaptive adjustment module;

[0044] The training and learning module is used to train the constructed neural network model using the marking parameter samples of the laser marking material to obtain the target network model;

[0045] The adaptive adjustment module is used to input material parameters and marking target values ​​as input parameters into the target network model, perform iterative calculations, obtain the optimal process parameters of the laser marking equipment, and adaptively adjust the laser marking equipment.

[0046] The marking parameters include material parameters, marking effect values, and corresponding process parameters.

[0047] In conjunction with the adaptive parameter tuning system described in the second aspect of the present invention, in a first possible embodiment, the adaptive adjustment module is used for:

[0048] The color difference value in the color space is calculated using the fitness function;

[0049] Obtain the color difference between the color space color difference value and the target value;

[0050] The color difference value is compared with a specified threshold:

[0051] If the color difference is less than a specified threshold, the process parameters corresponding to the marking effect value are the optimal solution, and the iteration stops. If the color difference is greater than the specified threshold, the marking parameters of this iteration are mutated to obtain a new parameter population to participate in the next iteration.

[0052] Thirdly, a laser marking device includes the adaptive parameter adjustment system described in the second aspect, used for marking different materials.

[0053] The adaptive parameter tuning method, system, and laser marking equipment described in this invention construct neural network models for various marking materials, train these models using collected marking parameter samples, derive and calculate marking effect values ​​from the target network model, and calculate color difference values ​​using color space color difference functions. This allows for the acquisition of optimal solution for process parameters and adaptive adjustment of laser marking equipment parameters. This solves the problems of requiring employees to spend a lot of time and energy debugging laser marking equipment parameters, the lack of standardization of related skills, and the difficulty in technology transfer. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a schematic diagram of the first process of the adaptive parameter tuning method in this invention;

[0056] Figure 2 This is a schematic diagram of the second process of the adaptive parameter tuning method in this invention;

[0057] Figure 3 This is a schematic diagram of the third process of the adaptive parameter tuning method in this invention;

[0058] Figure 4 This is a schematic diagram of the fourth process of the adaptive parameter tuning method in this invention;

[0059] Figure 5 This is a schematic diagram of parameter classification in the adaptive parameter tuning method of this invention; Detailed Implementation

[0060] The technical solutions of this invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are all within the scope of protection of this invention.

[0061] In existing technologies, the parameters of laser marking equipment are adjusted based on the experience of employees. The relevant skills cannot be standardized, which consumes a lot of time and energy and also poses a problem of technology transfer.

[0062] To address the above problems, an adaptive parameter tuning method, system, and laser marking equipment are proposed.

[0063] Example 1

[0064] Firstly, an adaptive parameter tuning method, such as Figure 1 , Figure 1 This is a schematic diagram of the first process of the adaptive parameter tuning method in this invention; preferably, it includes the following steps:

[0065] A neural network model is constructed, and the neural network model is trained using the marking parameter samples of the laser marking material to obtain the target network model;

[0066] The material parameters and the target value for marking are input into the target network model as input parameters, and iterative calculations are performed to obtain the optimal process parameters for the laser marking equipment.

[0067] The laser marking equipment is adaptively adjusted using the obtained optimal process parameters;

[0068] The marking parameters include material parameters, marking effect values, and corresponding process parameters.

[0069] Commonly used marking materials in industry, such as PMMA, ABS, and PS, are used to construct corresponding neural network models.

[0070] The neural network model is trained in advance using corresponding marking parameter samples. Machine learning capabilities are then used to obtain the required target network model and predict process parameters.

[0071] The process of predicting process parameters using a target network model is also an iterative process of obtaining the optimal solution for process parameters.

[0072] Before the iteration begins, the preferred implementation steps are:

[0073] Classify the process parameters of laser marking equipment and obtain empirical parameters;

[0074] Empirical parameters of the marking material are obtained using the target network model;

[0075] Among them, the empirical parameters are those that have a relatively large impact on the marking effect and require multiple adjustments by process engineers to determine.

[0076] Before the iteration begins, the process parameters are classified. Only those parameters that have a greater impact on the marking effect, and even if the material type is known, require multiple adjustments by process engineers to determine the parameters, can be predicted using neural networks, which can reduce the computational load of the target network model.

[0077] The process parameters can be classified using a quadrant method, such as... Figure 5 Figure 5 is a schematic diagram of parameter classification in the adaptive parameter tuning method of this invention. Specifically, it can be implemented as follows:

[0078] Once the material is determined, parameters with relatively fixed values ​​can be determined by an expert system designed by experienced process engineers and used as parameters for the first and fourth quadrants, respectively.

[0079] For parameters that have a relatively small impact on the marking effect and are less certain, their impact on marking efficiency is more direct. Therefore, they need to be determined according to the specific marking requirements and are thus considered as parameters in the third quadrant.

[0080] The impact on the marking effect is greater, and even if the material type is known, the process engineer needs to make multiple adjustments to determine the parameters. Therefore, these parameters are classified as parameters in the second quadrant. These process parameters are predicted using neural networks.

[0081] Before the iteration begins, in case the algorithm fails to converge, to avoid invalid loops, preferably, the following steps are also required:

[0082] Set the maximum number of iterations for the target network model.

[0083] After completing the maximum number of iterations, stop iterating regardless of whether the optimal solution has been found.

[0084] In this embodiment, a fitness function is used to calculate the color difference value corresponding to the process parameters, such as... Figure 2 , Figure 2 This is a schematic diagram of the second process of the adaptive parameter tuning method in this invention. Preferably, the specific steps include:

[0085] The color difference value in the color space is calculated using the fitness function;

[0086] like Figure 3 , Figure 3 This is a schematic diagram of the third process of the adaptive parameter tuning method in this invention. In this embodiment, the preferred step for calculating the color difference value is as follows:

[0087] The marking effect value is calculated using the target network model;

[0088] Calculate the marking difference between the marking effect value and the target marking value;

[0089] Using the color difference function formula in color space:

[0090] = (1)

[0091] Calculate the color difference value in the color space;

[0092] in, Indicates psychological brightness. =L (Effect Value) - L (Target Value) Positive values ​​indicate a brighter result, and negative values ​​indicate a darker result;

[0093] For psychological chromaticity, = a (Effect Value) - a (Target Value) Positive values ​​indicate a reddish tint, and negative values ​​indicate a greenish tint;

[0094] For psychological chromaticity, = b (effect value) - b (target value). A positive value indicates a yellowish tint, and a negative value indicates a bluish tint.

[0095] After obtaining the color difference value, it is necessary to solve for the optimal solution of the process parameters:

[0096] Obtain the color difference between the color space color difference value and the target value;

[0097] If the color difference is less than the specified threshold, the process parameters corresponding to the marking effect value are the optimal solution, and the iteration stops.

[0098] If the color difference exceeds a specified threshold, the marking parameters for this iteration are mutated to obtain a new parameter population for the next iteration. By constructing neural network models for various marking materials and training the models with collected marking parameter samples, the marking effect value is calculated using the target network model. The color difference is then calculated using a color space color difference function, thereby obtaining the optimal solution for the process parameters. This allows for adaptive adjustment of the laser marking equipment parameters, solving the problems of requiring employees to spend significant time and effort debugging laser marking equipment parameters, the lack of standardization of related skills, and the difficulty in technology transfer.

[0099] If the optimal solution cannot be obtained in this iteration, the data from this iteration needs to be selected, crossovered, and mutated to obtain a new data population to participate in the next iteration, such as... Figure 4 , Figure 4 This is a schematic diagram of the fourth process of the adaptive parameter tuning method in this invention. The preferred steps are as follows:

[0100] Using formula (2)

[0101] (2)

[0102] Calculate the selection probability of an individual in the labeling parameters, where, Let represent the fitness of the i-th individual. Let be the probability that the i-th individual is selected;

[0103] The selected parameters are randomly cross-referenced and mutated at random locations to form a new parameter population.

[0104] Each parameter individual is selected, cross-referenced, and mutated at random locations to form a new parameter population, which participates in the next iteration.

[0105] The optimal parameter population for this iteration is retained.

[0106] In a preferred embodiment, in order to allow the optimal parameter combination to vary, it is necessary to retain the parameter combination that performed relatively well in the current iteration.

[0107] Example 2

[0108] An adaptive parameter tuning system, employing the method of the first aspect, includes:

[0109] Training and learning modules;

[0110] Adaptive adjustment module;

[0111] The training and learning module is used to train the constructed neural network model using the marking parameter samples of the laser marking material to obtain the target network model;

[0112] The adaptive adjustment module is used to input material parameters and marking target values ​​as input parameters into the target network model, perform iterative calculations, obtain the optimal process parameters of the laser marking equipment, and adaptively adjust the laser marking equipment.

[0113] The marking parameters include material parameters, marking effect values, and corresponding process parameters.

[0114] In conjunction with the adaptive parameter tuning system of the second aspect of the present invention, in a first possible embodiment, the adaptive adjustment module is used for:

[0115] The color difference value in the color space is calculated using the fitness function;

[0116] Obtain the color difference between the color space color difference value and the target value;

[0117] Compare the color difference value with a specified threshold:

[0118] If the color difference is less than the specified threshold, the process parameters corresponding to the marking effect value are the optimal solution, and the iteration stops. If the color difference is greater than the specified threshold, the marking parameters of this iteration are mutated to obtain a new parameter population to participate in the next iteration.

[0119] Example 3

[0120] A laser marking device includes a second aspect of an adaptive parameter adjustment system, and marks different materials.

[0121] The adaptive parameter tuning method, system, and laser marking equipment described in this invention construct neural network models for various marking materials, train these models using collected marking parameter samples, derive and calculate marking effect values ​​from the target network model, and calculate color difference values ​​using color space color difference functions. This allows for the acquisition of optimal solution for process parameters and adaptive adjustment of laser marking equipment parameters. This solves the problems of requiring employees to spend a lot of time and energy debugging laser marking equipment parameters, the lack of standardization of related skills, and the difficulty in technology transfer.

[0122] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An adaptive parameter tuning method, characterized in that, Including the following steps: A neural network model is constructed, and the neural network model is trained using marking parameter samples of laser marking material to obtain the target network model; The material parameters and the target value for marking are input into the target network model as input parameters, and iterative calculations are performed to obtain the optimal process parameters for the laser marking equipment. The laser marking equipment is adaptively adjusted using the obtained optimal process parameters; The marking parameters include material parameters, marking effect values, and corresponding process parameters; Before the iteration begins, the process parameters are classified using a quadrant classification method: laser type and focal position are used as parameters in the first quadrant; the number of markings, marking speed, Q frequency, Q release time, current, and fill spacing are used as parameters in the second quadrant; the air jump speed, laser switch delay, jump delay, and turn delay are used as parameters in the third quadrant; the field lens focal length and whether air blowing is required are used as parameters in the fourth quadrant; and the parameters in the second quadrant are predicted using a neural network.

2. The adaptive parameter tuning method according to claim 1, characterized in that, The steps include: inputting material parameters and the target marking value as input parameters into the target network model, performing iterative calculations, and obtaining the optimal process parameters for the laser marking equipment, including: The color difference value in the color space is calculated using the fitness function; Obtain the color difference between the color space color difference value and the target value; If the color difference is less than the specified threshold, then the process parameters corresponding to the marking effect value are the optimal solution, and the iteration stops. If the color difference value is greater than a specified threshold, the marking parameters of this iteration are mutated to obtain a new parameter population to participate in the next iteration.

3. The adaptive parameter tuning method according to claim 2, characterized in that, The steps include: calculating the color space color difference value using a fitness function, including: The labeling effect value is calculated using the target network model. Calculate the marking difference between the marking effect value and the target marking value; Using the color difference function formula in color space: = （1） Calculate the color difference value of the color space; in, Indicates psychological brightness. =L (Effect Value) - L (Target Value) Positive values ​​indicate a brighter result, and negative values ​​indicate a darker result; For psychological chromaticity, = a (Effect Value) - a (Target Value) Positive values ​​indicate a reddish tint, and negative values ​​indicate a greenish tint; For psychological chromaticity, = b (effect value) - b (target value). A positive value indicates a yellowish tint, and a negative value indicates a bluish tint.

4. The adaptive parameter tuning method according to claim 2, characterized in that, The steps are as follows: If the color difference value is greater than a specified threshold, the marking parameters for this iteration are mutated to obtain a new parameter population to participate in the next iteration, including: Using formula (2) （2） Calculate the selection probability of an individual among the labeling parameters, where, Let represent the fitness of the i-th individual. Let be the probability that the i-th individual is selected; The selected parameters are randomly cross-referenced and mutated at random locations to form a new parameter population.

5. The adaptive parameter tuning method according to claim 4, characterized in that, Before the step of calculating the selection probability of an individual in the labeling parameters, the step of performing mutation processing on the labeling parameters of this iteration to obtain a new parameter population to participate in the next iteration if the color difference value is greater than a specified threshold also includes: The optimal parameter population for this iteration is retained.

6. The adaptive parameter tuning method according to claim 5, characterized in that, Before the step of calculating the color space difference value using the fitness function, the step of inputting material parameters and the marking target value as input parameters into the target network model and performing iterative calculations to obtain the optimal process parameters for the laser marking equipment also includes: Classify the process parameters of laser marking equipment and obtain empirical parameters; The empirical parameters of the marking material are obtained using the target network model. Among them, the empirical parameters are those that have a relatively large impact on the marking effect and require multiple adjustments by process engineers to determine.

7. The adaptive parameter tuning method according to claim 6, characterized in that, In the step of classifying the process parameters of the laser marking equipment and obtaining empirical parameters, the previous step of inputting material parameters and marking target values ​​as input parameters into the target network model for iterative calculation to obtain the optimal process parameters of the laser marking equipment also includes: Set the maximum number of iterations for the target network model.

8. An adaptive parameter tuning system, employing the method described in any one of claims 1-7, characterized in that, include: Training and learning modules; Adaptive adjustment module; The training and learning module is used to train the constructed neural network model using the marking parameter samples of the laser marking material to obtain the target network model; The adaptive adjustment module is used to input material parameters and marking target values ​​as input parameters into the target network model, perform iterative calculations, obtain the optimal process parameters of the laser marking equipment, and adaptively adjust the laser marking equipment. The marking parameters include material parameters, marking effect values, and corresponding process parameters.

9. The adaptive parameter tuning system according to claim 8, characterized in that, The adaptive adjustment module is used for: The color difference value in the color space is calculated using the fitness function; Obtain the color difference between the color space color difference value and the target value; The color difference value is compared with a specified threshold: If the color difference is less than a specified threshold, the process parameters corresponding to the marking effect value are the optimal solution, and the iteration stops. If the color difference is greater than the specified threshold, the marking parameters of this iteration are mutated to obtain a new parameter population to participate in the next iteration.

10. A laser marking device, characterized in that, The system includes the adaptive parameter tuning system as described in claim 9, used for marking different materials.

Images