A laser micropore processing parameter optimization method and system for double-sided film ice bag production

Abstract

This invention relates to the field of micropore processing technology, and discloses a method and system for optimizing laser micropore processing parameters for the production of double-sided ice packs. The method involves denoising, imputing missing values, and standardizing the original data to obtain preprocessed data. This preprocessed data is then input into a BKA-LSTM-SVM model, which outputs prediction results. A multi-objective optimization model is established with the optimization objectives of maximizing micropore permeation rate, maximizing pore size uniformity, minimizing membrane breakage rate, and maximizing processing efficiency. Using this multi-objective optimization model, the prediction results are used as fitness values ​​to iteratively update particle velocity and position, introducing a nonlinear convergence factor and adaptive inertia weights to balance global exploration and local development. When the maximum number of iterations is reached, the optimal combination of laser micropore processing parameters is output and sent to the laser processing system. This invention significantly reduces manual debugging costs and improves yield and production efficiency.

Description

Technical Field

[0001] This invention relates to the field of micropore processing technology, specifically to a method and system for optimizing laser micropore processing parameters for the production of double-sided ice packs. Background Technology

[0002] Double-sided ice packs, as commonly used self-absorbing preservation and cold compress consumables, are made of PE composite smooth film. They rely on the micropores on the film surface to achieve self-absorption function and are widely used in scenarios such as express cold chain and post-operative facial cold compresses. Currently, the laser micropore processing parameters for double-sided ice packs rely on manual experience to set, lacking a systematic optimization method. This easily leads to problems such as uneven pore size, film layer damage, and low permeation rate, directly resulting in slow water absorption, easy leakage after freezing, and reduced skin adhesion and reusability. At the same time, it is difficult to balance processing efficiency and micropore quality. A single optimization goal cannot meet the multiple requirements of water absorption speed, uniformity, integrity rate, and production efficiency, which restricts the improvement of double-sided ice pack quality and cost control, and makes it difficult to adapt to the high-precision and high-efficiency processing requirements of laser micropores for double-sided ice packs. Summary of the Invention

[0003] The purpose of this invention is to solve the above-mentioned problems by designing a method and system for optimizing laser micropore processing parameters for the production of double-sided ice packs.

[0004] The first aspect of this invention provides a method for optimizing laser micropore processing parameters for the production of double-sided ice packs, the method comprising the following steps:

[0005] The process parameters and micropore quality indicators of laser micropore processing for double-sided ice packs were collected to obtain raw data. The raw data were then denoised, missing values ​​were filled in, and the data was standardized to obtain preprocessed data.

[0006] The preprocessed data is input into the BKA-LSTM-SVM model. First, the process parameter features and micropore quality features are extracted through the dual-channel LSTM layer. Then, the nonlinear regression prediction of the processing parameters and micropore quality is performed through the SVM layer, and the prediction results are output. The hyperparameters of the LSTM layer and SVM layer are optimized through the Black Kite optimization algorithm.

[0007] A multi-objective optimization model is established with the optimization objectives of maximizing micropore permeation rate, maximizing pore size uniformity, minimizing membrane breakage rate, and maximizing processing efficiency.

[0008] A multi-objective optimization model is adopted, using the prediction results as the fitness value. The particle velocity and position are iteratively updated, and a nonlinear convergence factor and adaptive inertia weight are introduced to balance global exploration and local development. When the maximum number of iterations is reached, the optimal combination of laser micro-hole processing parameters is output and sent to the laser processing system.

[0009] Optionally, in a first implementation of the first aspect of the present invention, the process parameters and micropore quality indicators of the laser micropore processing of the double-sided ice pack are collected to obtain raw data. The raw data is then subjected to noise reduction, missing value completion, and standardization to obtain preprocessed data, including:

[0010] The original data is subjected to multi-level wavelet decomposition, the high-frequency coefficients after decomposition are subjected to threshold quantization, and the data is reconstructed by inverse wavelet transform to obtain the denoised data.

[0011] The missing positions in the denoised data are located, and the continuous valid data before and after the missing position are selected. The value of the missing position is calculated using Lagrange interpolation, and the calculation result is filled into the corresponding position.

[0012] The completed data is then subjected to Z-score standardization to obtain preprocessed data.

[0013] Optionally, in a second implementation of the first aspect of the present invention, the step of inputting the preprocessed data into the BKA-LSTM-SVM model, first extracting process parameter features and micropore quality features through the LSTM layer, and then performing nonlinear regression prediction of processing parameters and micropore quality through the SVM layer, and outputting the prediction results, includes:

[0014] The preprocessed data is input into the BKA-LSTM-SVM model. The LSTM layer sequentially performs forget gate filtering, input gate updating, memory cell state adjustment and output gate calculation to extract process parameter features and micropore quality features.

[0015] The extracted process parameter features and micropore quality features are fused to obtain a comprehensive feature vector, which is then input into the SVM layer.

[0016] By using kernel functions to map the comprehensive feature vectors to a high-dimensional feature space, an optimal hyperplane is constructed and relaxation variables are introduced to handle noise interference.

[0017] Within the SVM layer, a dual problem is solved to perform nonlinear regression fitting calculations between processing parameters and micropore quality. The regression calculation results output by the SVM layer are then processed to obtain the prediction results.

[0018] Optionally, in a third implementation of the first aspect of the present invention, the step of mapping the comprehensive feature vector to a high-dimensional feature space using a kernel function, constructing an optimal hyperplane, and introducing relaxation variables to handle noise interference includes:

[0019] The radial basis function kernel is used to perform a dimension-by-dimensional transformation on the comprehensive feature vector, mapping it to a high-dimensional feature space;

[0020] In a high-dimensional feature space, with the goal of maximizing the margin, we traverse and search for hyperplanes that can distinguish the data distribution, and initially determine the set of candidate hyperplanes.

[0021] Calculate the interval distance for candidate hyperplanes, select the hyperplane that maximizes the interval, and determine it as the current optimal hyperplane;

[0022] Traverse all feature data points in the high-dimensional space, identify noisy data points that deviate from the normal distribution and are disturbed, and mark the location of the noisy data.

[0023] In the process of constructing the optimal hyperplane, slack variables are introduced to compensate for the bias of the marked noisy data points. Based on the compensation results of the slack variables, the position and orientation of the optimal hyperplane are readjusted to construct the optimal hyperplane.

[0024] Optionally, in a fourth implementation of the first aspect of the present invention, the optimization of the hyperparameters of the LSTM and SVM layers using the Black-winged Kite optimization algorithm includes:

[0025] An initial population of black-winged kites is randomly generated, such that each individual in the population corresponds to a set of hyperparameter combinations of LSTM and SVM layers;

[0026] The fitness of all individuals in the initial population is calculated, the individual with the best fitness is selected from the initial population and set as the initial leader individual, and the optimal hyperparameter combination in the current iteration cycle is determined.

[0027] Based on the globally optimal individual position and the random individual position, update the positions of all non-leader individuals in the population and perform a global search with hyperparameter combinations;

[0028] Local adjustments are made to the positions of individuals in the current population, and local optimization operations are performed on the updated hyperparameter combinations.

[0029] By comparing the fitness values ​​of the current individual with those of random individuals, the leader individual is dynamically adjusted, and the globally optimal hyperparameter combination is updated.

[0030] Determine if the current iteration count has reached the maximum iteration count. If not, return to fitness calculation and continue iterating.

[0031] If the optimal combination of hyperparameters for the globally optimal individual is achieved, it will be used as the final optimization hyperparameters for the LSTM and SVM layers.

[0032] Optionally, in the fifth implementation of the first aspect of the present invention, the multi-objective optimization model is adopted, using the prediction result as the fitness value, iteratively updating the particle velocity and position, introducing a nonlinear convergence factor and an adaptive inertia weight to balance global exploration and local development, and when the maximum number of iterations is reached, outputting the optimal combination of laser micro-hole processing parameters, and sending the optimal combination of laser micro-hole processing parameters to the laser processing system, including:

[0033] Initialize the particle swarm algorithm parameters, set the population size, maximum number of iterations and search interval boundaries, and map the laser micro-hole processing parameters into particle position vectors;

[0034] The prediction results output by the BKA-LSTM-SVM model are input into the multi-objective optimization model to calculate the fitness value of the corresponding particle.

[0035] In the iterative optimization process, a nonlinear convergence factor is introduced to dynamically adjust the search range of the particles, and an adaptive inertial weight is introduced to dynamically adjust the particle movement amplitude according to the iteration process.

[0036] Update the particle's flight velocity based on its current fitness value, individual best position, and global best position;

[0037] Based on the updated particle velocity, the position of the particle in the solution space is updated synchronously to complete the iterative calculation of the position of the next generation of particles;

[0038] Determine if the current iteration count has reached the preset maximum iteration count. If not, return to the fitness calculation step to continue the loop iteration.

[0039] If the optimal position information of the global best particle is obtained, it is converted into the optimal combination of laser micro-hole processing parameters and then sent to the laser processing system.

[0040] Optionally, in a sixth implementation of the first aspect of the present invention, the step of introducing a nonlinear convergence factor to dynamically adjust the search range of particles during the iterative optimization process, and simultaneously introducing an adaptive inertia weight to dynamically adjust the particle movement amplitude according to the iteration process, includes:

[0041] Set the initial value, termination value and maximum number of iterations of the nonlinear convergence factor. In each iteration, calculate and update the convergence factor value of the current iteration in real time according to the ratio of the current number of iterations to the maximum number of iterations and the nonlinear change. Use the updated nonlinear convergence factor to dynamically adjust the global search range of the particles.

[0042] Set the initial and final weights for the adaptive inertia weights. In each iteration, calculate and update the adaptive inertia weight value in real time according to the current iteration progress. Substitute the adaptive inertia weights into the particle velocity update calculation process and dynamically adjust the particle's movement amplitude according to the weight size to control the speed of the particle's movement.

[0043] A second aspect of the present invention provides a laser micropore processing parameter optimization system for the production of double-sided ice packs, the system comprising:

[0044] The data processing module is used to collect the process parameters and micropore quality indicators of the laser micropore processing of the double-sided ice pack, obtain the raw data, and perform noise reduction, missing value completion and standardization on the raw data to obtain the preprocessed data.

[0045] The model prediction module is used to input the preprocessed data into the BKA-LSTM-SVM model. First, the process parameter features and micropore quality features are extracted through the dual-channel LSTM layer. Then, the nonlinear regression prediction of the processing parameters and micropore quality is performed through the SVM layer, and the prediction results are output. The hyperparameters of the LSTM layer and SVM layer are optimized through the Black Kite Optimization Algorithm.

[0046] The model building module is used to establish a multi-objective optimization model with the optimization objectives of maximizing micropore permeation rate, maximizing pore size uniformity, minimizing membrane damage rate, and maximizing processing efficiency.

[0047] The parameter optimization module is used to employ a multi-objective optimization model, using the prediction results as fitness values ​​to iteratively update particle velocity and position. It introduces a nonlinear convergence factor and adaptive inertial weight to balance global exploration and local development. When the maximum number of iterations is reached, it outputs the optimal combination of laser micro-hole processing parameters and sends the optimal combination of laser micro-hole processing parameters to the laser processing system.

[0048] A third aspect of the present invention provides a laser micro-perforation processing parameter optimization device for the production of double-sided ice packs. The laser micro-perforation processing parameter optimization device for the production of double-sided ice packs includes a memory and at least one processor. The memory stores instructions. The at least one processor calls the instructions in the memory to cause the laser micro-perforation processing parameter optimization device for the production of double-sided ice packs to perform the various steps of the laser micro-perforation processing parameter optimization method for the production of double-sided ice packs as described in any of the preceding claims.

[0049] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the steps of the laser micropore processing parameter optimization method for producing double-sided ice packs as described in any of the preceding claims.

[0050] In the technical solution provided by this invention, process parameters and micropore quality indicators for laser micropore processing of double-sided ice packs are collected to obtain raw data. The raw data is then denoised, missing values ​​are filled in, and standardized to obtain preprocessed data. The preprocessed data is input into a BKA-LSTM-SVM model. First, process parameter features and micropore quality features are extracted using a dual-channel LSTM layer. Then, a nonlinear regression prediction of processing parameters and micropore quality is performed using an SVM layer, outputting the prediction results. The hyperparameters of the LSTM and SVM layers are optimized using the Black Kite optimization algorithm, with the optimization goals being to maximize micropore permeation rate, pore size uniformity, membrane breakage rate, and processing efficiency. The objective is to establish a multi-objective optimization model. Using this model, prediction results are used as fitness values ​​to iteratively update particle velocity and position. A nonlinear convergence factor and adaptive inertia weight are introduced to balance global exploration and local development. When the maximum number of iterations is reached, the optimal combination of laser micropore processing parameters is output and sent to the laser processing system. This invention maximizes pore size uniformity, minimizes membrane breakage rate, reduces the risk of freezing leakage, improves the stability and skin comfort of reusable ice packs, balances processing efficiency and micropore quality, significantly reduces manual adjustment costs, improves yield and production efficiency, and comprehensively improves the self-absorption performance, durability, and user experience of double-layer ice packs. Attached Figure Description

[0051] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0052] Figure 1 A flowchart of a method for optimizing laser micropore processing parameters for producing double-sided ice packs provided in an embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of the laser micropore processing parameter optimization system for producing double-sided ice packs provided in an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram of the laser micropore processing parameter optimization equipment for producing double-sided ice packs provided in an embodiment of the present invention. Detailed Implementation

[0055] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0056] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The flowchart of the laser micropore processing parameter optimization method for producing double-sided ice packs provided in this embodiment of the invention includes the following steps:

[0057] Step 101: Collect the process parameters and micropore quality indicators of laser micropore processing for double-sided ice packs to obtain raw data. Perform noise reduction, missing value completion and standardization on the raw data to obtain preprocessed data.

[0058] In this embodiment, key process parameters such as laser power, scanning speed, pulse frequency, defocusing amount, and number of processing steps are collected in real time through the sensing and acquisition unit of the on-site processing equipment, the industrial control system, and dedicated testing instruments. At the same time, micropore quality indicators such as average pore size, pore size uniformity, pore depth, porosity, micropore permeation rate, and membrane damage rate are also collected.

[0059] In this embodiment, the original data undergoes multi-level wavelet decomposition. Based on the characteristics of the double-film ice pack laser micro-hole processing data, a suitable wavelet basis function is selected and the number of decomposition levels is determined. The original data is decomposed layer by layer into low-frequency approximation coefficients and high-frequency detail coefficients at different scales. Threshold quantization is then performed on the decomposed high-frequency coefficients, and noise-corresponding coefficient components are removed or compressed according to a set threshold. Finally, the processed high- and low-frequency coefficients are inversely synthesized using wavelet inverse transform to complete data reconstruction and obtain denoised data. The denoised data is then traversed point by point and its integrity is checked to accurately locate and mark data locations with missing values ​​or information gaps. Valid, continuous, and anomaly-free data points are selected as the basis for interpolation, centered on the missing locations. The Lagrange interpolation algorithm is used to fit the values ​​at the missing locations, and the calculated fitted values ​​are accurately filled into the corresponding missing locations to complete the missing data. After the missing values ​​are filled, Z-score standardization is performed on the data. The mean and standard deviation of each data sequence are calculated to convert all data into a standard normal distribution, eliminating the dimensional differences and numerical magnitude gaps between different process parameters and quality indicators, and standardizing the data to a uniform interval. Finally, preprocessed data that can be directly input into the model is obtained.

[0060] Step 102: Input the preprocessed data into the BKA-LSTM-SVM model. First, extract the process parameter features and micropore quality features through the dual-channel LSTM layer, and then perform nonlinear regression prediction of the processing parameters and micropore quality through the SVM layer. Output the prediction results. The hyperparameters of the LSTM layer and SVM layer are optimized through the Black Kite optimization algorithm.

[0061] In this embodiment, the preprocessed data is input into the BKA-LSTM-SVM hybrid model according to the sample order and dimensional specification. Inside the LSTM layer, forget gate filtering, input gate updating, memory cell state adjustment and output gate calculation operations are executed sequentially according to the time-series processing logic. The forget gate filters out historical information with low correlation to the current processing state. The input gate determines the effective process and quality information that needs to be stored in the memory cell. The memory cell completes the preservation and updating of long-term effective information. The output gate determines the feature information that needs to be output at the current moment. Thus, process parameter features and micropore quality features with strong characterization ability are deeply mined and extracted from the process parameter sequence and micropore quality sequence, respectively.

[0062] The process parameter features and micropore quality features extracted from the LSTM layer are subjected to dimension alignment, weight allocation and information fusion processing. Redundant information and repeated expressions between features are removed, and the core information that can fully reflect the laser processing process, micropore structure state and quality performance is retained to form a comprehensive feature vector with unified dimensions, complete information and full expression. The comprehensive feature vector is then normalized according to the input format of the SVM layer and sent to the SVM layer for subsequent calculation.

[0063] A radial basis function kernel is used to perform a dimensional transformation on the comprehensive feature vector. According to the preset kernel function parameters, nonlinear mapping calculations are performed on each dimension of the feature vector sequentially, completely mapping the comprehensive feature vector originally in the low-dimensional space to the high-dimensional feature space, thus improving the data's distinguishability and separability in the high-dimensional space. Within the high-dimensional feature space, maximizing the classification margin is the core optimization objective. All feasible hyperplane combinations are traversed according to the set search step size and search range, and hyperplanes that can effectively distinguish different data distributions and meet basic fitting requirements are selected, forming a preliminary candidate hyperplane set. The corresponding data margin distance is calculated for each hyperplane in the candidate hyperplane set, and the margins of each hyperplane are compared one by one. The system first selects the hyperplane that maximizes the margin and identifies it as the optimal hyperplane under the current iteration conditions. Then, it iterates through all feature data points in the high-dimensional feature space, comparing the relative position and distance deviations between each data point and the current optimal hyperplane. Noisy data points that significantly deviate from the normal data distribution, exhibit random disturbances, or abnormal fluctuations are identified and their positions are accurately marked. During the construction and optimization of the optimal hyperplane, relaxation variables are introduced to compensate for deviations and provide fault tolerance for the marked noisy data points. The position, angle, and orientation of the optimal hyperplane are recalculated and adjusted based on the compensation results provided by the relaxation variables, ultimately completing the construction of a stable and robust optimal hyperplane.

[0064] The computational complexity of high-dimensional calculations is reduced by transforming the problem into a dual problem within the SVM layer, thereby improving computational efficiency and numerical stability. Based on the optimal hyperplane in the high-dimensional space, regression fitting calculations are performed to complete the complex nonlinear relationship between processing parameters and micropore quality, enabling the prediction of key indicators such as micropore permeation rate, pore size uniformity, and membrane damage rate. The original regression results output by the SVM layer are then numerically calibrated, format standardized, and validity screened to finally obtain stable, accurate, and reliable micropore quality prediction results.

[0065] In this embodiment, based on the preset population size, hyperparameter search range, and dimensional information, an initial population for the Black-winged Kite algorithm is randomly generated to ensure that each individual in the population uniquely corresponds to a complete hyperparameter combination containing LSTM layer network structure parameters and SVM layer kernel function parameters, thus constructing an initial solution space. Fitness is calculated for each individual in the initial population, using model prediction accuracy as the fitness evaluation criterion to objectively assess the quality of each hyperparameter combination. The individual with the best fitness value is then selected from all individuals and designated as the initial leader, determining the optimal hyperparameter combination for the current iteration cycle. Based on the currently determined global optimal individual position and the randomly selected individual positions, the positions of all non-leader individuals in the population are uniformly updated according to the Black-winged Kite algorithm's flight search rules. A large-scale global search is conducted throughout the entire hyperparameter search space, gradually expanding the distribution area of ​​high-quality hyperparameter combinations. After completing the global position update... After the update, the positions of all individuals in the current population are finely adjusted locally. Local optimization operations are performed on various updated hyperparameter combinations to further explore better hyperparameter configurations in the current region and improve the local search accuracy of the algorithm. The fitness values ​​of the current individual and random individuals are compared one by one. The leader individual is dynamically judged and adjusted based on the comparison results. The globally optimal hyperparameter combination is updated in real time during the iteration process to ensure that the leader individual always represents the optimal level of the current population. After each iteration, it is determined in real time whether the current iteration number has reached the preset maximum iteration number. If the maximum iteration number has not been reached, the fitness calculation step is returned, and fitness evaluation, position update and optimization operations are carried out again to continue the iteration loop. If the current iteration number has reached the preset maximum iteration number, the entire iteration process of the Black-winged Kite algorithm is terminated immediately, the hyperparameter combination corresponding to the globally optimal individual is extracted, and this combination is used as the final optimization hyperparameter of the LSTM layer and SVM layer.

[0066] Step 103: Establish a multi-objective optimization model with the optimization objectives of maximizing micropore permeation rate, maximizing pore size uniformity, minimizing membrane damage rate, and maximizing processing efficiency.

[0067] In this embodiment, the core optimization objectives are to maximize the micropore permeation rate, pore size uniformity, membrane damage rate, and processing efficiency during the laser micropore processing of double-sided ice packs. Combining actual processing constraints, equipment operation limitations, and quality control requirements, each objective is uniformly quantified and standardized, the constraint relationships and weight allocation rules between each objective are clarified, and a multi-objective optimization model that can simultaneously take into account multiple indicators and meet actual production needs is constructed.

[0068] Step 104: Using a multi-objective optimization model, the predicted results are used as fitness values ​​to iteratively update particle velocity and position. Nonlinear convergence factors and adaptive inertial weights are introduced to balance global exploration and local development. When the maximum number of iterations is reached, the optimal combination of laser micro-hole processing parameters is output and sent to the laser processing system.

[0069] In this embodiment, the relevant operating parameters of the particle swarm optimization algorithm are initialized. Based on the optimization requirements of the laser micro-hole processing parameters, the population size, maximum number of iterations, and search interval boundaries of each processing parameter are reasonably set. The laser micro-hole processing parameters such as laser power, scanning speed, pulse frequency, and defocusing amount are uniformly mapped to the position vectors of particles in the solution space, thus completing the initial solution space construction of the particle swarm optimization algorithm. The relevant prediction results output by the BKA-LSTM-SVM model are input into the pre-constructed multi-objective optimization model. The fitness value of each particle position vector is calculated according to the multi-objective comprehensive evaluation rules, providing a unified evaluation standard for particle iterative optimization.

[0070] The algorithm sets initial, final, and maximum iteration values ​​for the nonlinear convergence factor. In each iteration, the convergence factor value is calculated and updated in real time according to the ratio of the current iteration number to the maximum iteration number, based on the nonlinear changes. The updated nonlinear convergence factor is used to dynamically adjust the global search range of the particles. The algorithm also sets initial and final weights for the adaptive inertia weight. In each iteration, the adaptive inertia weight value is calculated and updated in real time according to the current iteration progress. The adaptive inertia weight is substituted into the particle velocity update calculation process, and the particle's movement amplitude is dynamically adjusted according to the weight value to control the speed of particle movement and balance the algorithm's global exploration capability in the early stage with its local exploration capability in the later stage.

[0071] Based on the particle's current fitness value, its own best position obtained from historical searches, and the global best position obtained from the entire population's historical searches, the particle's flight speed is calculated and updated according to the particle swarm optimization algorithm's speed update rules, guiding the particle to move towards a better region. Based on the calculated and updated particle flight speed, combined with the particle's current position, the particle's new position in the solution space is updated synchronously, completing the iterative calculation of the next generation of particle positions, enabling the particle to gradually approach the optimal parameter region within the search space.

[0072] After each update of particle position and velocity, it is determined in real time whether the current iteration count has reached the preset maximum iteration count. If the set value has not been reached, the fitness calculation step is returned to recalculate the particle fitness and the loop iteration continues. If the current iteration count has reached the preset maximum iteration count, the entire particle swarm iterative optimization process is terminated, the position vector information corresponding to the globally optimal particle in the swarm is extracted, it is converted into the optimal laser micro-hole processing parameter combination that can be directly applied, and the optimal laser micro-hole processing parameter combination is sent to the laser processing system according to the device communication protocol to complete parameter optimization and command execution.

[0073] Please see Figure 2 A schematic diagram of the laser micropore processing parameter optimization system for the production of double-sided ice packs provided in this embodiment of the invention. The system includes:

[0074] The data processing module is used to collect the process parameters and micropore quality indicators of the laser micropore processing of the double-sided ice pack, obtain the raw data, and perform noise reduction, missing value completion and standardization on the raw data to obtain the preprocessed data.

[0075] The model prediction module is used to input the preprocessed data into the BKA-LSTM-SVM model. First, the process parameter features and micropore quality features are extracted through the dual-channel LSTM layer. Then, the nonlinear regression prediction of the processing parameters and micropore quality is performed through the SVM layer, and the prediction results are output. The hyperparameters of the LSTM layer and SVM layer are optimized through the Black Kite Optimization Algorithm.

[0076] The model building module is used to establish a multi-objective optimization model with the optimization objectives of maximizing micropore permeation rate, maximizing pore size uniformity, minimizing membrane damage rate, and maximizing processing efficiency.

[0077] The parameter optimization module is used to employ a multi-objective optimization model, using the prediction results as fitness values ​​to iteratively update particle velocity and position. It introduces a nonlinear convergence factor and adaptive inertial weight to balance global exploration and local development. When the maximum number of iterations is reached, it outputs the optimal combination of laser micro-hole processing parameters and sends the optimal combination of laser micro-hole processing parameters to the laser processing system.

[0078] Figure 3This is a schematic diagram of a laser micro-perforation parameter optimization device for producing double-sided ice packs, provided in an embodiment of the present invention. This laser micro-perforation parameter optimization device 300 for producing double-sided ice packs can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) storing application programs 333 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the laser micro-perforation parameter optimization device 300 for producing double-sided ice packs. Furthermore, the processor 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the laser micro-perforation parameter optimization device 300 for producing double-sided ice packs to implement the method provided in the above embodiment.

[0079] The laser micropore processing parameter optimization equipment 300 for producing double-sided ice packs may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The structure of the laser micropore processing parameter optimization equipment for producing double-sided ice packs shown does not constitute a limitation on the computer equipment provided by this invention. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0080] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform each step of the laser micropore processing parameter optimization method for producing double-sided ice packs provided in the above embodiments.

[0081] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described equipment or apparatus / unit can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0082] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0083] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing laser micropore processing parameters for the production of double-sided ice packs, characterized in that, The method includes the following steps: The process parameters and micropore quality indicators of laser micropore processing for double-sided ice packs were collected to obtain raw data. The raw data were then denoised, missing values ​​were filled in, and the data was standardized to obtain preprocessed data. The preprocessed data is input into the BKA-LSTM-SVM model. First, the process parameter features and micropore quality features are extracted through the dual-channel LSTM layer. Then, the nonlinear regression prediction of the processing parameters and micropore quality is performed through the SVM layer, and the prediction results are output. The hyperparameters of the LSTM layer and SVM layer are optimized through the Black Kite optimization algorithm. A multi-objective optimization model is established with the optimization objectives of maximizing micropore permeation rate, maximizing pore size uniformity, minimizing membrane breakage rate, and maximizing processing efficiency. A multi-objective optimization model is adopted, using the prediction results as the fitness value. The particle velocity and position are iteratively updated, and a nonlinear convergence factor and adaptive inertia weight are introduced to balance global exploration and local development. When the maximum number of iterations is reached, the optimal combination of laser micro-hole processing parameters is output and sent to the laser processing system.

2. The method for optimizing laser micro-hole processing parameters for producing double-sided ice packs as described in claim 1, characterized in that, The process parameters and micropore quality indicators of the laser micropore processing of the double-sided ice pack were collected to obtain raw data. The raw data underwent noise reduction, missing value completion, and standardization to obtain preprocessed data, including: The original data is subjected to multi-level wavelet decomposition, the high-frequency coefficients after decomposition are subjected to threshold quantization, and the data is reconstructed by inverse wavelet transform to obtain the denoised data. The missing positions in the denoised data are located, and the continuous valid data before and after the missing position are selected. The value of the missing position is calculated using Lagrange interpolation, and the calculation result is filled into the corresponding position. The completed data is then subjected to Z-score standardization to obtain preprocessed data.

3. The method for optimizing laser micro-hole processing parameters for producing double-sided ice packs as described in claim 1, characterized in that, The preprocessed data is input into the BKA-LSTM-SVM model. First, process parameter features and micropore quality features are extracted through the LSTM layer. Then, nonlinear regression prediction of processing parameters and micropore quality is performed through the SVM layer, outputting the prediction results, including: The preprocessed data is input into the BKA-LSTM-SVM model. The LSTM layer sequentially performs forget gate filtering, input gate updating, memory cell state adjustment and output gate calculation to extract process parameter features and micropore quality features. The extracted process parameter features and micropore quality features are fused to obtain a comprehensive feature vector, which is then input into the SVM layer. By using kernel functions to map the comprehensive feature vectors to a high-dimensional feature space, an optimal hyperplane is constructed and relaxation variables are introduced to handle noise interference. Within the SVM layer, a dual problem is solved to perform nonlinear regression fitting calculations between processing parameters and micropore quality. The regression calculation results output by the SVM layer are then processed to obtain the prediction results.

4. The method for optimizing laser micro-hole processing parameters for producing double-sided ice packs as described in claim 3, characterized in that, The process of mapping the comprehensive feature vector to a high-dimensional feature space using a kernel function, constructing an optimal hyperplane, and introducing relaxation variables to handle noise interference includes: The radial basis function kernel is used to perform a dimension-by-dimensional transformation on the comprehensive feature vector, mapping it to a high-dimensional feature space; In a high-dimensional feature space, with the goal of maximizing the margin, we traverse and search for hyperplanes that can distinguish the data distribution, and initially determine the set of candidate hyperplanes. Calculate the interval distance for candidate hyperplanes, select the hyperplane that maximizes the interval, and determine it as the current optimal hyperplane; Traverse all feature data points in the high-dimensional space, identify noisy data points that deviate from the normal distribution and are disturbed, and mark the location of the noisy data. In the process of constructing the optimal hyperplane, slack variables are introduced to compensate for the bias of the marked noisy data points. Based on the compensation results of the slack variables, the position and orientation of the optimal hyperplane are readjusted to construct the optimal hyperplane.

5. The method for optimizing laser micro-hole processing parameters for producing double-sided ice packs as described in claim 1, characterized in that, The optimization of hyperparameters of LSTM and SVM layers using the Black-winged Kite optimization algorithm includes: An initial population of black-winged kites is randomly generated, such that each individual in the population corresponds to a set of hyperparameter combinations of LSTM and SVM layers; The fitness of all individuals in the initial population is calculated, the individual with the best fitness is selected from the initial population and set as the initial leader individual, and the optimal hyperparameter combination in the current iteration cycle is determined. Based on the globally optimal individual position and the random individual position, update the positions of all non-leader individuals in the population and perform a global search with hyperparameter combinations; Local adjustments are made to the positions of individuals in the current population, and local optimization operations are performed on the updated hyperparameter combinations. By comparing the fitness values ​​of the current individual with those of random individuals, the leader individual is dynamically adjusted, and the globally optimal hyperparameter combination is updated. Determine if the current iteration count has reached the maximum iteration count. If not, return to fitness calculation and continue iterating. If the optimal combination of hyperparameters for the globally optimal individual is achieved, it will be used as the final optimization hyperparameters for the LSTM and SVM layers.

6. The method for optimizing laser micro-hole processing parameters for producing double-sided ice packs as described in claim 1, characterized in that, The process employs a multi-objective optimization model, using the prediction results as fitness values. It iteratively updates particle velocity and position, introducing a nonlinear convergence factor and adaptive inertia weights to balance global exploration and local development. When the maximum number of iterations is reached, the optimal combination of laser micro-hole processing parameters is output and sent to the laser processing system. This includes: Initialize the particle swarm algorithm parameters, set the population size, maximum number of iterations and search interval boundaries, and map the laser micro-hole processing parameters into particle position vectors; The prediction results output by the BKA-LSTM-SVM model are input into the multi-objective optimization model to calculate the fitness value of the corresponding particle. In the iterative optimization process, a nonlinear convergence factor is introduced to dynamically adjust the search range of the particles, and an adaptive inertial weight is introduced to dynamically adjust the particle movement amplitude according to the iteration process. Update the particle's flight velocity based on its current fitness value, individual best position, and global best position; Based on the updated particle velocity, the position of the particle in the solution space is updated synchronously to complete the iterative calculation of the position of the next generation of particles; Determine if the current iteration count has reached the preset maximum iteration count. If not, return to the fitness calculation step to continue the loop iteration. If the optimal position information of the global best particle is obtained, it is converted into the optimal combination of laser micro-hole processing parameters and then sent to the laser processing system.

7. The method for optimizing laser micro-hole processing parameters for producing double-sided ice packs as described in claim 6, characterized in that, The process of introducing a nonlinear convergence factor to dynamically adjust the search range of particles during iterative optimization, and simultaneously introducing an adaptive inertia weight to dynamically adjust the particle movement amplitude according to the iteration progress, includes: Set the initial value, termination value and maximum number of iterations of the nonlinear convergence factor. In each iteration, calculate and update the convergence factor value of the current iteration in real time according to the ratio of the current number of iterations to the maximum number of iterations and the nonlinear change. Use the updated nonlinear convergence factor to dynamically adjust the global search range of the particles. Set the initial and final weights for the adaptive inertia weights. In each iteration, calculate and update the adaptive inertia weight value in real time according to the current iteration progress. Substitute the adaptive inertia weights into the particle velocity update calculation process and dynamically adjust the particle's movement amplitude according to the weight size to control the speed of the particle's movement.

8. A laser micro-hole processing parameter optimization system for the production of double-sided ice packs, characterized in that, The system includes: The data processing module is used to collect the process parameters and micropore quality indicators of the laser micropore processing of the double-sided ice pack, obtain the raw data, and perform noise reduction, missing value completion and standardization on the raw data to obtain the preprocessed data. The model prediction module is used to input the preprocessed data into the BKA-LSTM-SVM model. First, the process parameter features and micropore quality features are extracted through the dual-channel LSTM layer. Then, the nonlinear regression prediction of the processing parameters and micropore quality is performed through the SVM layer, and the prediction results are output. The hyperparameters of the LSTM layer and SVM layer are optimized through the Black Kite Optimization Algorithm. The model building module is used to establish a multi-objective optimization model with the optimization objectives of maximizing micropore permeation rate, maximizing pore size uniformity, minimizing membrane damage rate, and maximizing processing efficiency. The parameter optimization module is used to employ a multi-objective optimization model, using the prediction results as fitness values ​​to iteratively update particle velocity and position. It introduces a nonlinear convergence factor and adaptive inertial weight to balance global exploration and local development. When the maximum number of iterations is reached, it outputs the optimal combination of laser micro-hole processing parameters and sends the optimal combination of laser micro-hole processing parameters to the laser processing system.

9. A laser micropore processing parameter optimization device for the production of double-sided ice packs, characterized in that, The laser micro-perforation processing parameter optimization device for producing double-sided ice packs includes a memory and at least one processor. The memory stores instructions. The at least one processor calls the instructions in the memory to cause the laser micro-perforation processing parameter optimization device for producing double-sided ice packs to perform each step of the laser micro-perforation processing parameter optimization method for producing double-sided ice packs as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement each step of the laser micropore processing parameter optimization method for producing double-sided ice packs as described in any one of claims 1-7.

Images