Screen plate soaking pool cleaning control parameter adaptation method and system

Abstract

The invention relates to the technical field of artificial intelligence, computer vision and automation, in particular to a screen soaking pool cleaning control parameter adaptation method and system.The method comprises the steps that current structural strength data and deformation data of a to-be-cleaned screen are obtained to determine the health state of the to-be-cleaned screen, a collaborative strategy knowledge base is inquired based on the health state, and the to-be-cleaned screen is obtained; obtaining a corresponding optimal cleaning control parameter combination; wherein the collaborative strategy knowledge base performs Pareto frontier analysis by taking maximization of a cleaning effect quantized value and minimization of a health loss quantized value as optimization targets, and identifies a non-dominated solution set; deleting the solutions of which the health loss values are greater than a first threshold value in the non-dominated solutions, and deleting the non-dominated solutions of which the cleaning effect quantized values are smaller than a preset second threshold value; and for different health states, based on the residual non-dominated solution set, on the Pareto frontier, different strategies are adopted to determine an optimal cleaning control parameter combination, and the optimal cleaning control parameter combination is constructed. According to the invention, collaborative optimization of the screen cleaning effect and the safety effect is realized.

Description

Technical Field

[0001] This application relates to the fields of artificial intelligence, computer vision and automation technology, and in particular to a method and system for adapting cleaning control parameters for a screen soaking tank. Background Technology

[0002] In modern livestock and poultry farming, wire mesh screens are exposed to high humidity and high organic load environments for extended periods, making them highly susceptible to the accumulation of uneaten feed, feces, and other contaminants. To ensure the cleanliness and reusability of these screens, a soaking tank is crucial equipment in the cleaning process. Cleaning involves immersing used wire mesh screens in a tank containing a cleaning agent, utilizing both chemical dissolution and physical cleaning. However, existing methods rely solely on the difference in contaminant removal between adjacent cleaning cycles to determine whether the current cleaning intensity and soaking time are effectively washing away the contaminants, and then adjust these settings accordingly. However, this method ignores the health condition of different wire mesh screens, easily damaging them and shortening their lifespan, especially for aged screens that have become brittle, cracked, or deformed. Therefore, achieving a synergistic optimization of cleaning effectiveness and safety for wire mesh screens is a pressing technical challenge that needs to be addressed. Summary of the Invention

[0003] To address the aforementioned technical problems, the purpose of this application is to provide a method and system for adapting cleaning control parameters in a screen immersion tank, aiming to achieve synergistic optimization of screen cleaning effect and safety effect.

[0004] In a first aspect, embodiments of this application provide a method for adapting cleaning control parameters for a screen soaking tank, the method comprising: Obtain the current structural strength and deformation data of the mesh panel to be cleaned; The health status of the mesh panel to be cleaned is determined based on the current structural strength data and deformation data. The corresponding optimal combination of cleaning control parameters is obtained by querying the collaborative strategy knowledge base based on the health status. The collaborative strategy knowledge base is constructed in the following way: Obtain multiple sets of sample meshes with different health states; For each healthy sample mesh, cleaning tests were conducted using multiple combinations of different cleaning control parameters. For each combination of cleaning control parameters, collect and record the corresponding quantitative values ​​of cleaning effect and quantified values ​​of screen health loss; With the optimization objective of maximizing the quantitative value of cleaning effect and minimizing the quantitative value of health loss, Pareto front analysis is performed in a two-dimensional decision space composed of the quantitative values ​​of cleaning effect and health loss to identify the non-dominated solution set. Delete the non-dominated solutions whose health loss value is greater than a preset first threshold and the non-dominated solutions whose cleaning effect quantification value is less than a preset second threshold, and obtain the remaining non-dominated solution set; For different health states, based on the residual nondominated solution set, different strategies are used on the Pareto front to determine the optimal combination of cleaning control parameters; By associating and mapping each health status with its corresponding optimal cleaning control parameter combination, a collaborative strategy knowledge base is constructed.

[0005] In one embodiment, the immersion tank for the wire mesh includes a wire mesh health status detection device, which includes an array-type ultrasonic flaw detector. The step of acquiring the current structural strength data of the wire mesh to be cleaned includes: The low-frequency array ultrasonic array probe in the array ultrasonic flaw detection device is used to perform non-contact scanning of the mesh plate to be cleaned, and the ultrasonic echo signal timing data of the mesh plate to be cleaned is obtained. Based on the ultrasonic echo signal time series data and the preset signal interval division rules, the mid-section echo signal in the signal is identified; Within the mid-section echo signal, time-domain and frequency-domain features are extracted and input into a trained classification model to determine whether there is an abnormal reflection pattern caused by surface contaminants. The time-domain features include root mean square amplitude, zero-crossing rate, and signal entropy, while the frequency-domain features include dominant frequency, spectral entropy, and high-low frequency energy ratio. If an abnormal reflection pattern is detected, the mid-section echo signal is weighted and attenuated or interpolated to generate a denoised echo signal. Based on the denoised echo signal, the material attenuation coefficient, sound velocity change rate, and defect density index are extracted as structural strength data.

[0006] In one embodiment, the health status detection device further includes a binocular vision camera, and the four corner points of the screen to be cleaned are pre-set with corrosion-resistant reflective marking points. The step of obtaining the current deformation data of the screen to be cleaned includes: The binocular vision camera is used to acquire images of the mesh plate to be cleaned, thus obtaining binocular images; Identify reflective markers in each eye of the binocular image; Based on the reflective markers in each image, the spatial three-dimensional coordinates of each reflective marker in the mesh to be cleaned are calculated using the principle of binocular stereo vision triangulation. The diagonal error is calculated based on the spatial three-dimensional coordinates of each reflective marker point in the screen to be cleaned and the diagonal length of the standard screen, and the diagonal error is used as deformation data.

[0007] In one embodiment, prior to the step of identifying reflective markers in each eye of the binocular images, the method further includes: The four corners of the screen to be cleaned are cleaned using a water spray gun until each camera in the binocular vision camera detects all reflective markers; wherein the reflective markers are determined based on brightness comparison and the theoretical coordinates of the reflective markers.

[0008] In one embodiment, the step of determining the optimal combination of cleaning control parameters on the Pareto front using different strategies based on the residual nondominated solution set for different health states includes: The strategy includes a first strategy and a second strategy. When the health status is lower than a preset health status level, the first strategy is used; otherwise, the second strategy is used. The first strategy is: Based on the remaining nondominated solutions, on the Pareto front, we start with the solution with the lowest health loss and move sequentially towards the solutions with higher cleaning effects. When the improvement in cleaning effect is lower than the preset third threshold, stop moving and select the current solution as the optimal combination of cleaning control parameters; The second strategy is: Based on the residual nondominated solutions, the substitution slope between adjacent solutions is calculated on the Pareto front; where the formula for calculating the substitution slope is: the increase in cleanliness / the increase in health loss; When the replacement slope is lower than the preset fourth threshold, the solution before the slope change is used as the optimal cleaning control parameter combination.

[0009] In one embodiment, the step of determining the health status of the mesh panel to be cleaned based on current structural strength data and deformation data includes: The current structural strength data and deformation data are input into the mesh panel health status prediction model to obtain the health status of the mesh panel to be cleaned; wherein, the mesh panel health status prediction model is trained according to the following method: Obtain a training dataset consisting of multiple sample mesh panels; each training data point includes structural strength data, deformation data, and a health status true value label determined by experts based on the comprehensive performance of the sample mesh panel for each sample mesh panel. The machine learning regression model is trained using the training dataset to obtain a trained health status assessment model.

[0010] Secondly, embodiments of this application provide a system for adapting cleaning control parameters to a screen soaking tank, the system comprising: The acquisition module is used to acquire the current structural strength data and deformation data of the mesh panel to be cleaned; The determination module is used to determine the health status of the mesh panel to be cleaned based on the current structural strength data and deformation data, and to query the collaborative strategy knowledge base based on the health status to obtain the corresponding optimal combination of cleaning control parameters. The collaborative strategy knowledge base is constructed in the following way: Obtain multiple sets of sample meshes with different health states; For each healthy sample mesh, cleaning tests were conducted using multiple combinations of different cleaning control parameters. For each combination of cleaning control parameters, collect and record the corresponding quantitative values ​​of cleaning effect and quantified values ​​of screen health loss; With the optimization objective of maximizing the quantitative value of cleaning effect and minimizing the quantitative value of health loss, Pareto front analysis is performed in a two-dimensional decision space composed of the quantitative values ​​of cleaning effect and health loss to identify the non-dominated solution set. Delete the non-dominated solutions whose health loss value is greater than a preset first threshold and the non-dominated solutions whose cleaning effect quantification value is less than a preset second threshold, and obtain the remaining non-dominated solution set; For different health states, based on the residual nondominated solution set, different strategies are used on the Pareto front to determine the optimal combination of cleaning control parameters; By associating and mapping each health status with its corresponding optimal cleaning control parameter combination, a collaborative strategy knowledge base is constructed.

[0011] This application embodiment considers two types of data reflecting the health status of the stencil: current structural strength data and deformation data. This lays the foundation for subsequent synergistic optimization of the stencil's cleaning and safety effects. By optimizing to maximize the quantified value of cleaning effect and minimize the quantified value of health loss, Pareto front analysis is performed within a two-dimensional decision space composed of these two values. This identifies a set of non-dominated solutions, where each non-dominated solution represents a combination of cleaning control parameters that achieves the optimal balance between cleaning and safety effects under a given health condition. These cleaning control parameter combinations include cleaning intensity and soaking time. Cleaning intensity includes jet water pressure, nozzle scanning speed, and pulse frequency. To avoid selecting solutions with poor cleaning effect or excessive health loss from the non-dominated solutions, a first threshold and a second threshold are predefined. Non-dominated solutions with health loss values ​​greater than the first threshold and non-dominated solutions with cleaning effect values ​​less than the second threshold are deleted, resulting in the remaining set of non-dominated solutions. To determine the optimal combination of cleaning control parameters from the remaining non-dominated solutions, and to better achieve synergistic optimization of screen cleaning and safety effects, different strategies are formulated for different health states. Based on the remaining non-dominated solution set, the optimal combination of cleaning control parameters for each health state is determined using the corresponding strategy on the Pareto front. This constructs a synergistic strategy knowledge base, which can determine the optimal combination of control parameters based on the screen health state. In summary, the embodiments of this application achieve synergistic optimization of screen cleaning and safety effects, achieving excellent cleaning results while effectively protecting the screen. Attached Figure Description

[0012] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of a method for adapting cleaning control parameters to a mesh screen soaking tank, provided in an embodiment of this application. Figure 2 This is a schematic diagram of a cleaning control parameter adaptation system for a mesh plate soaking tank provided in an embodiment of this application. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0015] Those skilled in the art will understand that, unless explicitly stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in the specification of this application means the presence of features, integers, steps, operations, elements, modules, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, modules, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any modules and all combinations of one or more associated listed items.

[0016] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0017] Please see Figure 1 This application provides a method for adapting cleaning control parameters for a screen soaking tank, the method comprising: S1. Obtain the current structural strength data and deformation data of the mesh plate to be cleaned; S2. Determine the health status of the mesh panel to be cleaned based on the current structural strength data and deformation data, and query the collaborative strategy knowledge base based on the health status to obtain the corresponding optimal combination of cleaning control parameters; The collaborative strategy knowledge base is constructed in the following way: Obtain multiple sets of sample meshes with different health states; For each healthy sample mesh, cleaning tests were conducted using multiple combinations of different cleaning control parameters. For each combination of cleaning control parameters, collect and record the corresponding quantitative values ​​of cleaning effect and quantified values ​​of screen health loss; With the optimization objective of maximizing the quantitative value of cleaning effect and minimizing the quantitative value of health loss, Pareto front analysis is performed in a two-dimensional decision space composed of the quantitative values ​​of cleaning effect and health loss to identify the non-dominated solution set. Delete the non-dominated solutions whose health loss value is greater than a preset first threshold and the non-dominated solutions whose cleaning effect quantification value is less than a preset second threshold, and obtain the remaining non-dominated solution set; For different health states, based on the residual nondominated solution set, different strategies are used on the Pareto front to determine the optimal combination of cleaning control parameters; By associating and mapping each health status with its corresponding optimal cleaning control parameter combination, a collaborative strategy knowledge base is constructed.

[0018] This application embodiment considers two types of data reflecting the health status of the stencil: current structural strength data and deformation data. This lays the foundation for subsequent synergistic optimization of the stencil's cleaning and safety effects. By optimizing to maximize the quantified value of cleaning effect and minimize the quantified value of health loss, Pareto front analysis is performed within a two-dimensional decision space composed of these two values. This identifies a set of non-dominated solutions, where each non-dominated solution represents a combination of cleaning control parameters that achieves the optimal balance between cleaning and safety effects under a given health condition. These cleaning control parameter combinations include cleaning intensity and soaking time. Cleaning intensity includes jet water pressure, nozzle scanning speed, and pulse frequency. To avoid selecting solutions with poor cleaning effect or excessive health loss from the non-dominated solutions, a first threshold and a second threshold are predefined. Non-dominated solutions with health loss values ​​greater than the first threshold and non-dominated solutions with cleaning effect values ​​less than the second threshold are deleted, resulting in the remaining set of non-dominated solutions. To determine the optimal combination of cleaning control parameters from the remaining non-dominated solutions, and to better achieve synergistic optimization of screen cleaning and safety effects, different strategies are formulated for different health states. Based on the remaining non-dominated solution set, the optimal combination of cleaning control parameters for each health state is determined using the corresponding strategy on the Pareto front. This constructs a synergistic strategy knowledge base, which can determine the optimal combination of control parameters based on the screen health state. In summary, the embodiments of this application achieve synergistic optimization of screen cleaning and safety effects, achieving excellent cleaning results while effectively protecting the screen.

[0019] In one embodiment, the immersion tank for the wire mesh includes a wire mesh health status detection device, which includes an array-type ultrasonic flaw detector. The step of acquiring the current structural strength data of the wire mesh to be cleaned includes: The low-frequency array ultrasonic array probe in the array ultrasonic flaw detection device is used to perform non-contact scanning of the mesh plate to be cleaned, and the ultrasonic echo signal timing data of the mesh plate to be cleaned is obtained. Based on the ultrasonic echo signal time series data and the preset signal interval division rules, the mid-section echo signal in the signal is identified; Within the mid-section echo signal, time-domain and frequency-domain features are extracted and input into a trained classification model to determine whether there is an abnormal reflection pattern caused by surface contaminants. The time-domain features include root mean square amplitude, zero-crossing rate, and signal entropy, while the frequency-domain features include dominant frequency, spectral entropy, and high-low frequency energy ratio. If an abnormal reflection pattern is detected, the mid-section echo signal is weighted and attenuated or interpolated to generate a denoised echo signal. Based on the denoised echo signal, the material attenuation coefficient, sound velocity change rate, and defect density index are extracted as structural strength data.

[0020] In this embodiment, low-frequency ultrasound (20-100kHz) has strong penetrating power, capable of penetrating a certain thickness of adhered dirt layer to detect the structural strength information inside the metal or plastic mesh. Based on the ultrasonic echo signal timing data and preset signal interval division rules, the mid-range echo signal is identified. Specifically, based on the ultrasonic echo signal timing data and preset signal interval division rules, the surface echo signal, mid-range echo signal, and bottom echo signal are identified. The surface echo refers to the reflected signal that occurs when the ultrasonic wave first encounters the surface of the mesh to be cleaned (including surface coverings) after being triggered by the probe; it is the first echo to appear in the entire echo sequence. The bottom echo signal refers to the signal that the ultrasonic wave is reflected at the bottom surface of the mesh after passing through the entire thickness of the mesh to be cleaned and returns to the probe; it is the last strong reflection peak in the echo sequence. The mid-range echo signal refers to the segment of ultrasonic wave signal located after the surface echo signal and before the bottom echo signal. It is mainly composed of the superposition of scattered signals generated by the sound wave encountering microscopic defects during propagation within the material. The preset signal interval division rule is based on the energy integral sliding window detection method. The principle is that the surface echo signal and the bottom echo signal have concentrated energy, while the middle echo signal has dispersed but continuous energy. Specifically, a short sliding window is used to calculate the local energy. The first detected high-energy pulse is taken as the surface echo signal, the last detected high-energy pulse is taken as the bottom echo signal, and the continuous energy region in between is taken as the middle echo signal.

[0021] Furthermore, in this embodiment, although the mid-range echo signal mainly reflects the internal structural state of the material, the presence of surface contaminants, such as feed or residue, can distort the ultrasonic incident field, thereby altering the statistical characteristics of the mid-range echo signal. Therefore, time-domain and frequency-domain features are extracted from the mid-range echo signal to form a feature vector. This feature vector is then input into a pre-trained classification model to identify atypical scattering modes caused by surface contaminants, thus indirectly determining the state of the contaminants. The mid-range echo signal reflects the statistical response of the interaction between ultrasonic waves and the microstructure during propagation within the material. Its time-domain and frequency-domain features are sensitive to incident field distortions caused by surface contaminants but exhibit strong robustness to environmental disturbances. Determining the presence of abnormal reflection modes caused by surface contaminants based on the mid-range echo signal is stable and reliable. Time-domain features include root-mean-square amplitude, zero-crossing rate, and signal entropy; frequency-domain features include dominant frequency, spectral entropy, and high-low frequency energy ratio. Based on these time-domain and frequency-domain features, atypical scattering modes caused by surface contaminants can be effectively identified. It should be understood that the root mean square amplitude (RMS) is used to characterize the overall energy intensity of the signal, and its calculation formula is as follows:

[0022] in, Let be the signal amplitude at the i-th sampling point, and N be the total number of signal sampling points. When feces or biofilm adhere to the surface of the mesh, the ultrasound undergoes multiple scattering within the dirt layer, resulting in enhanced energy of the mid-range echo signal and a significant increase in the RMS value.

[0023] Zero-crossing rate (ZCR): Defined as the number of times a signal crosses a zero point per unit time, calculated as follows:

[0024] in, ( () is the indicator function. Clutter caused by dirt will make the signal waveform more oscillating, leading to an increase in ZCR.

[0025] Signal entropy H: Used to measure the randomness and complexity of a signal waveform, calculated using Shannon entropy.

[0026] in, This represents the probability that the signal amplitude falls within the k-th quantization interval. Noise signals introduced by contaminants are more dispersed and have higher entropy values.

[0027] Dominant frequency: defined as the energy-weighted average frequency of the spectrum, calculated using the following formula:

[0028] in, This is the Fourier transform result of the signal. Sticky pollutants such as biofilms absorb high-frequency components, causing the spectral energy to shift to lower frequencies, resulting in a significant decrease in the dominant frequency.

[0029] High-low frequency energy ratio: Calculate the energy ratio of the high-frequency band (e.g., 400-800kHz) to the low-frequency band (100-300kHz): Where f is the frequency, Indicates high frequency band, Corresponding to the low frequency band.

[0030] Spectral entropy : Reflects the uniformity of frequency component distribution, and is calculated as follows: , ; in, This represents the normalized energy percentage at frequency f.

[0031] Multiple scattering disperses the energy distribution across a wider frequency band, flattens the spectrum, and increases the frequency domain entropy.

[0032] In this embodiment, if an abnormal reflection pattern is detected, the mid-range echo signal is subjected to weighted attenuation or interpolation compensation. Based on the classification label, the above classification model can also classify the type of contaminants. Through extensive experimental calibration, a contaminant-frequency band influence mapping table is established, thereby enabling weighted attenuation or interpolation compensation of the mid-range echo signal based on the contaminant-frequency band influence mapping table.

[0033] In this embodiment, the degree to which ultrasonic energy attenuates with distance when propagating in a material is significantly increased by aging, microcracks, and porosity. The method for calculating the material attenuation coefficient is as follows:

[0034] in, Indicates the amplitude of the transmitted pulse. d represents the amplitude of the first bottom surface echo, and d represents the thickness of the screen.

[0035] It should be understood that a decrease in the elastic modulus of a material (such as in plastic catalysis or metal fatigue) will lead to a decrease in the speed of sound. The formula for calculating the rate of change of the speed of sound is:

[0036] in, This represents the round-trip time of the ultrasound. The reference sound velocity for the standard mesh plate. The rate of change of the speed of sound This represents the currently measured speed of sound.

[0037] Defect density refers to the number of scattering particles such as cracks and pores per unit volume, reflecting the degree of internal damage. The formula for calculating defect density is:

[0038] in, It is the time period between the surface echo and the bottom echo. This is the mid-range echo signal.

[0039] Through the above embodiments, the present invention can extract reliable structural strength data such as material attenuation coefficient, sound velocity change rate, and defect density index.

[0040] In one embodiment, the health status detection device further includes a binocular vision camera, and the four corner points of the screen to be cleaned are pre-set with corrosion-resistant reflective marking points. The step of obtaining the current deformation data of the screen to be cleaned includes: The binocular vision camera is used to acquire images of the mesh plate to be cleaned, thus obtaining binocular images; Identify reflective markers in each eye of the binocular image; Based on the reflective markers in each image, the spatial three-dimensional coordinates of each reflective marker in the mesh to be cleaned are calculated using the principle of binocular stereo vision triangulation. The diagonal error is calculated based on the spatial three-dimensional coordinates of each reflective marker point in the screen to be cleaned and the diagonal length of the standard screen, and the diagonal error is used as deformation data.

[0041] In this embodiment, reflective markers are installed at the four corners of the mesh panel. These markers can be reflective nails, which are corrosion-resistant to withstand farm manure, cleaning agents, and humid environments. The reflective properties make them highly visible under camera flash or active lighting, and the method of using reflective nails is low-cost and yields more accurate deformation data. Diagonal error: reflects whether the mesh panel has undergone tensile, compressive, or shear deformation. Let the four corners of the mesh panel to be cleaned be the upper left corner. Top right corner B bottom right corner C D in the bottom left corner Calculate the lengths of the two diagonals: ;

[0042] Calculate the diagonal error: ; in, The diagonal length of the standard mesh plate. This is the diagonal error. Let AC be the length of the diagonal. Let be the length of the diagonal of point BD.

[0043] In one embodiment, prior to the step of identifying reflective markers in each eye of the binocular images, the method further includes: The four corners of the screen to be cleaned are cleaned using a water spray gun until each camera in the binocular vision camera detects all reflective markers; wherein the reflective markers are determined based on brightness comparison and the theoretical coordinates of the reflective markers.

[0044] In this embodiment, feces and dirt may obscure the reflective markers. To ensure smooth deformation detection, after fixing the mesh to be cleaned at a designated position in the soaking tank, four miniature water spray guns positioned at the fixed position are used to clean the four corners. A binocular vision camera captures images of the mesh to be cleaned. When each camera in the binocular vision camera detects all reflective markers, the water spray guns are stopped. The reflective markers are determined based on brightness comparison and the theoretical coordinates of the reflective markers. Specifically, for each monocular image, the reflective markers of each monocular image are determined as follows: the monocular image is converted to grayscale to obtain a grayscale image; the N bright spots with the highest average brightness in the grayscale image are selected as candidate targets; where N is greater than 4; the candidate targets are spatially matched with the theoretical coordinates of the reflective markers, and the four candidate targets with the highest matching degree are determined as the four reflective markers of the mesh to be cleaned.

[0045] In one embodiment, the step of determining the optimal combination of cleaning control parameters on the Pareto front using different strategies based on the residual nondominated solution set for different health states includes: The strategy includes a first strategy and a second strategy. When the health status is lower than a preset health status level, the first strategy is used; otherwise, the second strategy is used. The first strategy is: based on the remaining nondominated solutions, on the Pareto front, start from the solution with the lowest health loss and move sequentially towards the solutions with higher cleaning effects; When the improvement in cleaning effect is lower than the preset third threshold, stop moving and select the current solution as the optimal combination of cleaning control parameters; The second strategy is: Based on the residual nondominated solutions, the substitution slope between adjacent solutions is calculated on the Pareto front; where the formula for calculating the substitution slope is: the increase in cleanliness / the increase in health loss; When the replacement slope is lower than the preset fourth threshold, the solution before the slope change is used as the optimal cleaning control parameter combination.

[0046] In this embodiment, after deleting non-dominated solutions with health loss values ​​greater than a preset first threshold and non-dominated solutions with cleaning effect quantification values ​​less than a preset second threshold, for screens with poor health, the process starts with the solution with the lowest health loss and moves sequentially towards solutions with higher cleaning effects. When the improvement in cleaning effect falls below a preset third threshold, the movement stops, and the current solution is selected as the optimal cleaning control parameter combination. This allows for better protection of screens with poor health during cleaning while achieving a better cleaning effect. For screens with good health, the replacement slope between adjacent solutions is calculated, and the formula for the replacement slope is: improvement in cleaning effect / increase in health loss. When the replacement slope falls below a preset fourth threshold, the solution before the slope change is selected as the optimal cleaning control parameter combination. This allows for better cleaning effect while protecting screens with good health. This invention employs different strategies to determine the optimal cleaning control parameter combination for different health states, enabling the acquisition of the optimal cleaning control parameter combination for each health state.

[0047] In one embodiment, the step of determining the health status of the mesh panel to be cleaned based on current structural strength data and deformation data includes: The current structural strength data and deformation data are input into the mesh panel health status prediction model to obtain a health score for the mesh panel to be cleaned, and the health status is determined based on the health score; wherein, the mesh panel health status prediction model is trained according to the following method: Obtain a training dataset consisting of multiple sample mesh panels; each training dataset includes structural strength data, deformation data, and a health rating label determined by experts based on the overall performance of the sample mesh panel for each sample mesh panel. The machine learning regression model is trained using the training dataset to obtain a trained health status assessment model.

[0048] This application's embodiments utilize a supervised machine learning regression-based method for quantifying stencil health. Its continuous numerical values ​​accurately characterize subtle declines in stencil health, laying the foundation for calculating the optimal combination of cleaning control parameters. Its powerful nonlinear fitting capability automatically learns the intrinsic relationship between various health indicators and the overall health status from massive amounts of data, eliminating reliance on difficult-to-determine empirical formulas. Its data-driven nature makes the evaluation structure more objective, and the model can be continuously optimized and become more accurate with data accumulation.

[0049] In one embodiment, the quantification value of the screen health loss is calculated based on the health score before and after cleaning. The quantification value of the screen cleaning effect is calculated based on the cleaning effect score before and after cleaning. The cleaning effect score is calculated based on the dirt coverage area, which is determined by computer vision. The formula for calculating the effect score is as follows: ; ; in, For the effect score, This represents the percentage of the area covered by dirt.

[0050] In one embodiment, the cleaning and soaking tank includes multiple chambers, each chamber being independent of the others, water flow is not allowed between the chambers, and each chamber is used to place a mesh plate.

[0051] like Figure 2 As shown in the embodiment of this application, a cleaning control parameter adaptation system for a screen soaking tank is also provided. The system includes: Module 1 is used to acquire the current structural strength data and deformation data of the mesh panel to be cleaned; Module 2 is used to determine the health status of the mesh panel to be cleaned based on the current structural strength data and deformation data, and to query the collaborative strategy knowledge base based on the health status to obtain the corresponding optimal combination of cleaning control parameters. The collaborative strategy knowledge base is constructed in the following way: Obtain multiple sets of sample meshes with different health states; For each healthy sample mesh, cleaning tests were conducted using multiple combinations of different cleaning control parameters. For each combination of cleaning control parameters, collect and record the corresponding quantitative values ​​of cleaning effect and quantified values ​​of screen health loss; With the optimization objective of maximizing the quantitative value of cleaning effect and minimizing the quantitative value of health loss, Pareto front analysis is performed in a two-dimensional decision space composed of the quantitative values ​​of cleaning effect and health loss to identify the non-dominated solution set. Delete the non-dominated solutions whose health loss value is greater than a preset first threshold and the non-dominated solutions whose cleaning effect quantification value is less than a preset second threshold, and obtain the remaining non-dominated solution set; For different health states, based on the residual nondominated solution set, different strategies are used on the Pareto front to determine the optimal combination of cleaning control parameters; By associating and mapping each health status with its corresponding optimal cleaning control parameter combination, a collaborative strategy knowledge base is constructed.

[0052] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0053] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0054] The above description is only a preferred embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural changes made based on the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for adapting cleaning control parameters in a screen immersion tank, characterized in that, The method includes: Obtain the current structural strength and deformation data of the mesh panel to be cleaned; The health status of the mesh panel to be cleaned is determined based on the current structural strength data and deformation data. The corresponding optimal combination of cleaning control parameters is obtained by querying the collaborative strategy knowledge base based on the health status. The collaborative strategy knowledge base is constructed in the following way: Obtain multiple sets of sample meshes with different health states; For each healthy sample mesh, cleaning tests were conducted using multiple combinations of different cleaning control parameters. For each combination of cleaning control parameters, collect and record the corresponding quantitative values ​​of cleaning effect and quantified values ​​of screen health loss; With the optimization objective of maximizing the quantitative value of cleaning effect and minimizing the quantitative value of health loss, Pareto front analysis is performed in a two-dimensional decision space composed of the quantitative values ​​of cleaning effect and health loss to identify the non-dominated solution set. Delete the non-dominated solutions whose health loss value is greater than a preset first threshold and the non-dominated solutions whose cleaning effect quantification value is less than a preset second threshold, and obtain the remaining non-dominated solution set; For different health states, based on the residual nondominated solution set, different strategies are used on the Pareto front to determine the optimal combination of cleaning control parameters; By associating and mapping each health status with its corresponding optimal cleaning control parameter combination, a collaborative strategy knowledge base is constructed.

2. The method for adapting cleaning control parameters of a screen soaking tank according to claim 1, characterized in that, The immersion tank for the wire mesh includes a wire mesh health status detection device, which includes an array-type ultrasonic flaw detector. The step of acquiring the current structural strength data of the wire mesh to be cleaned includes: The low-frequency array ultrasonic array probe in the array ultrasonic flaw detection device is used to perform non-contact scanning of the mesh plate to be cleaned, and the ultrasonic echo signal timing data of the mesh plate to be cleaned is obtained. Based on the ultrasonic echo signal time series data and the preset signal interval division rules, the mid-section echo signal in the signal is identified; Within the mid-section echo signal, time-domain and frequency-domain features are extracted and input into a trained classification model to determine whether there is an abnormal reflection pattern caused by surface contaminants. The time-domain features include root mean square amplitude, zero-crossing rate, and signal entropy, while the frequency-domain features include dominant frequency, spectral entropy, and high-low frequency energy ratio. If an abnormal reflection pattern is detected, the mid-section echo signal is weighted and attenuated or interpolated to generate a denoised echo signal. Based on the denoised echo signal, the material attenuation coefficient, sound velocity change rate, and defect density index are extracted as structural strength data.

3. The method for adapting cleaning control parameters of a screen soaking tank according to claim 2, characterized in that, The health status detection device also includes a binocular vision camera. Corrosion-resistant reflective markers are pre-set at the four corners of the screen to be cleaned. The steps for obtaining the current deformation data of the screen to be cleaned include: The binocular vision camera is used to acquire images of the mesh plate to be cleaned, thus obtaining binocular images; Identify reflective markers in each eye of the binocular image; Based on the reflective markers in each image, the spatial three-dimensional coordinates of each reflective marker in the mesh to be cleaned are calculated using the principle of binocular stereo vision triangulation. The diagonal error is calculated based on the spatial three-dimensional coordinates of each reflective marker point in the screen to be cleaned and the diagonal length of the standard screen, and the diagonal error is used as deformation data.

4. The method for adapting cleaning control parameters of the screen soaking tank according to claim 3, characterized in that, Before the step of identifying reflective markers in each eye of the binocular images, the method further includes: The four corners of the screen to be cleaned are cleaned using a water spray gun until each camera in the binocular vision camera detects all reflective markers; wherein the reflective markers are determined based on brightness comparison and the theoretical coordinates of the reflective markers.

5. The method for adapting cleaning control parameters of a screen soaking tank according to claim 1, characterized in that, The steps for determining the optimal combination of cleaning control parameters on the Pareto front using different strategies based on the residual nondominated solution set for different health states include: The strategy includes a first strategy and a second strategy. When the health status is lower than a preset health status level, the first strategy is used; otherwise, the second strategy is used. The first strategy is: Based on the remaining nondominated solutions, on the Pareto front, we start with the solution with the lowest health loss and move sequentially towards the solutions with higher cleaning effects. When the improvement in cleaning effect is lower than the preset third threshold, stop moving and select the current solution as the optimal combination of cleaning control parameters; The second strategy is: Based on the residual nondominated solutions, the substitution slope between adjacent solutions is calculated on the Pareto front; where the formula for calculating the substitution slope is: the increase in cleanliness / the increase in health loss; When the replacement slope is lower than the preset fourth threshold, the solution before the slope change is used as the optimal cleaning control parameter combination.

6. The method for adapting cleaning control parameters of a screen soaking tank according to claim 1, characterized in that, The step of determining the health status of the mesh panel to be cleaned based on the current structural strength data and deformation data includes: The current structural strength data and deformation data are input into the mesh panel health status prediction model to obtain the health status of the mesh panel to be cleaned; wherein, the mesh panel health status prediction model is trained according to the following method: Obtain a training dataset consisting of multiple sample mesh panels; each training data point includes structural strength data, deformation data, and a health status true value label determined by experts based on the comprehensive performance of the sample mesh panel for each sample mesh panel. The machine learning regression model is trained using the training dataset to obtain a trained health status assessment model.

7. A system for adapting cleaning control parameters to a screen immersion tank, characterized in that, The system includes: The acquisition module is used to acquire the current structural strength data and deformation data of the mesh panel to be cleaned; The determination module is used to determine the health status of the mesh panel to be cleaned based on the current structural strength data and deformation data, and to query the collaborative strategy knowledge base based on the health status to obtain the corresponding optimal combination of cleaning control parameters. The collaborative strategy knowledge base is constructed in the following way: Obtain multiple sets of sample meshes with different health states; For each healthy sample mesh, cleaning tests were conducted using multiple combinations of different cleaning control parameters. For each combination of cleaning control parameters, collect and record the corresponding quantitative values ​​of cleaning effect and quantified values ​​of screen health loss; With the optimization objective of maximizing the quantitative value of cleaning effect and minimizing the quantitative value of health loss, Pareto front analysis is performed in a two-dimensional decision space composed of the quantitative values ​​of cleaning effect and health loss to identify the non-dominated solution set. Delete the non-dominated solutions whose health loss value is greater than a preset first threshold and the non-dominated solutions whose cleaning effect quantification value is less than a preset second threshold, and obtain the remaining non-dominated solution set; For different health states, based on the residual nondominated solution set, different strategies are used on the Pareto front to determine the optimal combination of cleaning control parameters; By associating and mapping each health status with its corresponding optimal cleaning control parameter combination, a collaborative strategy knowledge base is constructed.

Images