Automatic correction and compensation method for material cutting and weighing

Abstract

The invention relates to an automatic correction and compensation method for cut material weighing, and belongs to the technical field of cut material processing. The method comprises the following steps: obtaining various types of cutting related parameters, and generating a standardized data set through noise filtering and abnormal value elimination processing; calculating a weight deviation based on the standardized data, judging a deviation type through a deviation characteristic adaptive identification and classification mechanism, and generating related characteristic data; corresponding compensation models are matched for different deviation types, and fusion compensation quantity parameters are generated by means of a multi-modal compensation model dynamic fusion mechanism; and material cutting parameters are corrected based on the compensation quantity parameters, control instructions are issued, and meanwhile historical data are obtained periodically to iteratively update compensation model regulation and control parameters. The control precision of the material cutting weight is improved, the method adapts to the dynamic change of material characteristics and equipment states, the manual intervention cost is reduced, the stability and reliability of material cutting machining are enhanced, and the method can be widely applied to various industrial machining scenes needing precise material cutting and weighing.

Description

Technical Field

[0001] This invention belongs to the field of material cutting technology, specifically relating to an automatic correction and compensation method for material cutting and weighing. Background Technology

[0002] In the field of material cutting and processing, precise control of material weight directly affects product quality, material utilization, and production efficiency. This is especially true in industries with high requirements for material weight accuracy, such as food processing, plastic molding, and metal processing. Existing material cutting and weighing compensation technologies mostly employ fixed parameter compensation or simple deviation feedback adjustment methods, which have many shortcomings.

[0003] Existing technologies handle the collected cutting parameters rather crudely, often failing to adequately consider high-frequency noise from equipment vibration and outlier interference during data acquisition, resulting in poor input data quality and consequently affecting the accuracy of deviation calculations. Current deviation determination methods mostly rely on single threshold judgments, failing to differentiate between different types of deviations such as inherent system deviations, material fluctuation deviations, and equipment wear deviations. Using a uniform compensation model for adjustment makes it difficult to adapt to the personalized compensation needs of different deviation causes, resulting in limited compensation accuracy.

[0004] Existing compensation fusion mechanisms mostly use fixed weight superposition, which cannot dynamically adjust the weight of each deviation compensation amount according to the fluctuation of material characteristics and equipment operating status, easily leading to over-compensation or under-compensation. In addition, existing technologies lack an effective closed-loop iterative optimization mechanism, and the control parameters of the compensation model cannot be continuously updated based on historical processing data. When material characteristics change or equipment wears out, the compensation accuracy will gradually decrease, requiring frequent manual intervention to adjust parameters, increasing production and maintenance costs, and making it difficult to guarantee the stability of the production process.

[0005] Therefore, in order to address the problems of low data processing accuracy, inaccurate identification of deviation types, poor adaptability of compensation fusion, and lack of closed-loop optimization in existing material cutting and weighing compensation technologies, an automatic correction and compensation method for material cutting and weighing that can realize data preprocessing, adaptive deviation classification, dynamic compensation fusion, and continuous iterative optimization is needed. Summary of the Invention

[0006] To address the aforementioned problems in the existing technology, this invention provides an automatic correction and compensation method for material cutting and weighing. The objective of this invention can be achieved through the following technical solutions: An automatic correction and compensation method for material cutting and weighing includes: S1: Obtain preset cutting parameters, actual cutting and weighing feedback data, material characteristic parameters and equipment operating status parameters, perform noise filtering and outlier removal, and generate a standardized data set; S2: Based on the target cutting weight and actual cutting weighing feedback data in the standardized data set, calculate the weight deviation; introduce an adaptive identification and classification mechanism for deviation characteristics, and combine the preset deviation judgment benchmark library, the fluctuation of material characteristic parameters in the standardized data set, and the changing trend of equipment operating status parameters to determine the type of the weight deviation, and generate deviation type judgment results and deviation feature data. S3: Based on the deviation type determination result and deviation feature data, match the corresponding weight deviation compensation model for the deviation type; the weight deviation includes system inherent deviation, material fluctuation deviation and equipment wear deviation, and generate an adaptive compensation amount based on the dynamic fusion mechanism of the multi-modal compensation model; perform fusion optimization on the compensation amount corresponding to the weight deviation, set the upper limit constraint of the compensation amount, and generate fusion compensation amount parameters; S4: Based on the fusion compensation parameters, the preset cutting parameters are corrected, and the corrected cutting control command is generated and sent to the cutting execution mechanism; historical data is periodically acquired, and the adjustment parameters of the weight deviation compensation model are iteratively updated in combination with the standardized data set, deviation type determination results, and fusion compensation parameters to generate a compensation model parameter set.

[0007] Specifically, the noise filtering process includes: extracting the material viscosity data from the material characteristic parameters and the cutter rotation speed data from the equipment operating status parameters; calculating the linear correlation coefficient between the two through correlation analysis; dynamically adjusting the time span of the filtering window based on the magnitude of the linear correlation coefficient; and performing a moving average smoothing process on the actual material cutting and weighing feedback data based on the adjusted filtering window to filter high-frequency noise signals.

[0008] Specifically, the outlier removal process includes: statistically analyzing historical data related to cutting of similar materials, calculating the mean and standard deviation of each parameter, and constructing normal data distribution intervals for each parameter; substituting the acquired parameters into the corresponding distribution intervals for comparison, and further verifying the initially identified outliers by combining the parameter change trends at adjacent times; for data finally identified as outliers, using the valid data at two adjacent times before and after the outlier as interpolation nodes, and calculating the supplementary data corresponding to the outlier time based on the time interval and numerical difference between the two nodes.

[0009] Specifically, the process of setting the deviation judgment benchmark includes: classifying and sorting historical cutting data, dividing the data into subsets according to material type, and further dividing each data subset into different time period groups according to equipment runtime; performing deviation statistics on qualified cutting data in each time period group, and outputting the qualified cutting deviation range for each group in combination with the allowable cutting accuracy requirements of the project; and organizing and archiving the material type, equipment runtime and corresponding qualified cutting deviation range of all groups to construct the deviation judgment benchmark library.

[0010] Specifically, when determining the type of weight deviation, the determination process for material fluctuation deviation includes: based on the standardized data set, extracting material characteristic parameters and corresponding weight deviation data to form a material characteristic-deviation data sequence; calculating the change difference between two adjacent sets of material characteristic parameters in the material characteristic-deviation data sequence, respectively statistically analyzing the continuous change difference of material density, material moisture content, and material viscosity and summing them, comparing the summed result with a preset material fluctuation benchmark, and recording the type and magnitude of the material characteristic parameter that triggered the determination.

[0011] Specifically, when determining the type of weight deviation, the process for determining equipment wear deviation includes: extracting the cumulative runtime data and corresponding weight deviation data from the equipment operating status parameters in the standardized data set, organizing and constructing the equipment runtime-weight deviation dataset in chronological order; analyzing the changing trend of the weight deviation, comparing the cumulative runtime of the equipment with a preset runtime benchmark, and recording the current cumulative runtime of the equipment and the rate of change of deviation.

[0012] Specifically, the implementation process of the weight deviation compensation model for matching the inherent deviation of the system includes: based on the deviation characteristic data, extracting the real-time value of the weight deviation, the cumulative effect of the deviation, and the rate of change of the deviation, and substituting them into the compensation model constructed based on proportional-integral-derivative control logic; calculating the proportional adjustment component based on the real-time value, calculating the integral adjustment component based on the cumulative effect of the deviation, and calculating the derivative adjustment component based on the rate of change of the deviation; and superimposing the three adjustment components according to preset weights to obtain the compensation amount corresponding to the inherent deviation of the system.

[0013] Specifically, the implementation process of the weight deviation compensation model for matching material fluctuation deviations includes: acquiring data on cutting weight deviations and compensation amounts corresponding to different material characteristic parameters under different degrees of difference; constructing a correlation mapping table between each material characteristic parameter and the compensation amount, including parameter difference level, corresponding basic compensation amount, and influence weight; during actual compensation calculation, extracting the actual material characteristic parameters from the standardized dataset, comparing them with preset standard material characteristic parameters, determining the degree of difference, and classifying the difference level; retrieving the corresponding basic compensation amount from the correlation mapping table according to the difference level, and simultaneously extracting the influence weight on the cutting weight from the mapping table to calculate the preliminary correction compensation amount; summing the preliminary correction compensation amounts corresponding to all material characteristic parameters to obtain the compensation amount for the material fluctuation deviation.

[0014] Specifically, the implementation process of the multimodal compensation model dynamic fusion mechanism includes: determining the compensation amount corresponding to the system inherent deviation, material fluctuation deviation, and equipment wear deviation based on the deviation type determination result; determining the weight coefficient corresponding to each deviation according to the fluctuation of material characteristic parameters and equipment operating status parameters in the standardized data set; multiplying the compensation amount corresponding to each deviation by its own weight coefficient to obtain the weighted compensation amount of each deviation; summing all weighted compensation amounts to obtain the initial fusion compensation amount; and smoothing the initial fusion compensation amount to generate the final adaptive compensation amount.

[0015] Specifically, the process of setting the upper limit constraint of the compensation amount includes: pre-setting the compensation amount constraint range to form the upper limit constraint standard of the compensation amount; comparing the adaptive compensation amount with the upper limit constraint of the compensation amount, and synchronously recording the comparison result of the compensation amount and the final determined compensation amount value.

[0016] Specifically, the implementation process of periodically acquiring historical data includes: triggering a historical data acquisition command according to a preset cycle, extracting the standardized data set, deviation type judgment results, fusion compensation parameters, historical cutting data, deviation data, and compensation effect data within the cycle; classifying and organizing the extracted data, adding time, material, and equipment identification information to the organized data, and storing it in the historical database according to a preset format.

[0017] Specifically, the implementation process of iteratively updating the control parameters of the weight deviation compensation model includes: constructing a parameter update dataset based on the historical cutting data, deviation data, and compensation effect data, combined with the standardized data sets generated at each stage, deviation type determination results, and fusion compensation amount parameters; adjusting the core control parameters of the compensation model corresponding to the system's inherent deviation, material fluctuation deviation, and equipment wear deviation in a data-driven manner with the goal of achieving the best compensation effect; and summarizing all updated control parameters to generate a compensation parameter set.

[0018] The beneficial effects of this invention are as follows: Noise filtering and outlier removal steps are added during the data acquisition stage. By dynamically adjusting the filtering window based on the correlation between material viscosity and cutter rotation speed, the actual material cutting and weighing feedback data is smoothed, effectively filtering high-frequency noise. At the same time, a normal data distribution range is constructed based on historical data, and outliers are verified a second time by combining the changing trends of adjacent parameters, ensuring the accuracy and reliability of the standardized data set.

[0019] By introducing an adaptive identification and classification mechanism for deviation characteristics, and combining a preset deviation judgment benchmark library, material characteristic parameter fluctuations, and equipment operating status parameter change trends, this invention can accurately distinguish different types of deviations, such as inherent system deviations, material fluctuation deviations, and equipment wear deviations, and generate corresponding deviation feature data. Compared to existing single threshold judgment methods, this invention can match dedicated compensation models to different deviation causes, improving the targeting and accuracy of compensation.

[0020] A dynamic fusion mechanism based on a multimodal compensation model is adopted. According to the deviation type determination results and material and equipment status information in standardized data, the weight coefficients of each deviation compensation amount are dynamically adjusted. An adaptive compensation amount is generated through weighted superposition and smoothing. Simultaneously, an upper limit constraint on the compensation amount is set to ensure that the compensation amount is within a reasonable range. This mechanism solves the problem of poor adaptability in existing fixed-weight fusion methods, can adapt to fluctuations in material characteristics and changes in equipment status, and ensures the stability of the compensation effect.

[0021] By periodically acquiring historical processing data, combining standardized data sets generated at each stage, deviation type determination results, and fusion compensation parameters, the adjustment parameters of the compensation model are iteratively updated using a data-driven approach, generating an optimized compensation model parameter set. This closed-loop mechanism enables the compensation model to continuously adapt to long-term trends such as changes in material characteristics and equipment wear and aging, reducing the need for manual intervention, lowering operation and maintenance costs, and ensuring long-term stability of cutting and weighing accuracy. Attached Figure Description

[0022] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0023] Figure 1 This is a flowchart illustrating an automatic correction and compensation method for material cutting and weighing according to the present invention. Figure 2 This is a schematic diagram of the dynamic fusion of multimodal compensation logic in this invention. Detailed Implementation

[0024] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0025] Please see Figures 1-2 An automatic correction and compensation method for material cutting and weighing, comprising: S1: Obtain preset cutting parameters, actual cutting and weighing feedback data, material characteristic parameters and equipment operating status parameters, perform noise filtering and outlier removal, and generate a standardized data set; S2: Based on the target cutting weight and actual cutting weighing feedback data in the standardized data set, calculate the weight deviation; introduce an adaptive identification and classification mechanism for deviation characteristics, and combine the preset deviation judgment benchmark library, the fluctuation of material characteristic parameters in the standardized data set, and the changing trend of equipment operating status parameters to determine the type of the weight deviation, and generate deviation type judgment results and deviation feature data. S3: Based on the deviation type determination result and deviation feature data, match the corresponding weight deviation compensation model for the deviation type; the weight deviation includes system inherent deviation, material fluctuation deviation and equipment wear deviation, and generate an adaptive compensation amount based on the dynamic fusion mechanism of the multi-modal compensation model; perform fusion optimization on the compensation amount corresponding to the weight deviation, set the upper limit constraint of the compensation amount, and generate fusion compensation amount parameters; S4: Based on the fusion compensation parameters, the preset cutting parameters are corrected, and the corrected cutting control command is generated and sent to the cutting execution mechanism; historical data is periodically acquired, and the adjustment parameters of the weight deviation compensation model are iteratively updated in combination with the standardized data set, deviation type determination results, and fusion compensation parameters to generate a compensation model parameter set.

[0026] Specifically, the noise filtering process includes: using an adaptive filtering method, first extracting material viscosity data from the material characteristic parameters and cutter rotation speed data from the equipment operating status parameters, calculating the linear correlation coefficient between the two using a correlation analysis algorithm, and dynamically adjusting the time span of the filtering window based on the magnitude of the correlation coefficient—when the absolute value of the correlation coefficient is greater than a preset correlation threshold, it indicates a strong correlation between material viscosity and cutter rotation speed, and the equipment vibration noise is significantly affected by the coupling between the two, so the filtering window is increased to enhance the noise suppression effect; when the absolute value of the correlation coefficient is less than or equal to the preset correlation threshold, it indicates a weak correlation between the two, and the noise source is relatively singular, so the filtering window is decreased to ensure data real-time performance; based on the adjusted filtering window, performing moving average smoothing processing on the collected actual material cutting and weighing feedback data, traversing the data sequence point by point, taking the mean of all data points within the window as the center of the current data point as the smoothed output data, thereby filtering out high-frequency noise signals generated by equipment vibration, while retaining the true trend of the weighing data.

[0027] Specifically, the outlier removal process includes: first, collecting historical data related to cutting of similar materials within the past three months; statistically analyzing the historical data to calculate the mean and standard deviation of each parameter; using the mean plus or minus three times the standard deviation as boundaries to construct normal data distribution intervals for each parameter; substituting each parameter collected in this instance into the corresponding distribution interval for comparison; if a parameter value exceeds the upper or lower limit threshold of its distribution interval, it is initially determined to be an outlier; for the initially determined outliers, further verification is performed by combining the parameter change trend at adjacent times—if the parameters at adjacent times are all within the normal range, and the difference between the outlier and the adjacent normal parameters exceeds a preset mutation threshold, it is finally determined to be an outlier; for the data finally determined to be outliers, linear interpolation is used to supplement them, using the effective data at two adjacent times before and after the outlier as interpolation nodes, and calculating the supplementary data corresponding to the outlier time based on the time interval and numerical difference between the two nodes; after substituting the supplementary data into the original data sequence, the correlation between the data point and the surrounding data is verified again to ensure the continuity and reliability of the data.

[0028] Specifically, the process of setting the deviation judgment benchmark includes: first, classifying and sorting historical cutting data, dividing it into several data subsets according to material type, and further dividing each data subset into different time period groups according to equipment runtime; performing deviation statistics on qualified cutting data in each time period group, calculating the maximum, minimum, and average values ​​of cutting weight deviation for each group, and determining the qualified cutting deviation range for each group based on the average value and the allowable cutting accuracy requirements of the project; organizing and archiving the material type, equipment runtime, and corresponding qualified cutting deviation range for all groups to construct a deviation judgment benchmark library, which supports the addition of benchmark data for new material types and equipment operating conditions; before determining the deviation type, extracting the material characteristic parameters of the current cutting to determine the material type, and extracting the cumulative runtime from the equipment operating status parameters to determine the time period group; accurately matching the corresponding qualified cutting deviation range from the benchmark library and using it as the benchmark for this deviation judgment to achieve adaptive matching of the benchmark.

[0029] Specifically, when determining the type of weight deviation, the process for determining material fluctuation deviation includes: setting a continuous acquisition period, synchronously acquiring a set of material characteristic parameters and corresponding weight deviation data in each acquisition period, and continuously acquiring multiple sets to form a material characteristic-deviation data sequence; calculating the change difference between two adjacent sets of material characteristic parameters in the data sequence, and statistically analyzing the continuous change difference of material density, material moisture content, and material viscosity respectively; summing the continuous change differences of each characteristic parameter, and if the summation result of any material characteristic parameter exceeds the preset material fluctuation benchmark, then extracting the weight deviation change trend of the corresponding time period; using a trend fitting algorithm to linearly fit the weight deviation data of the time period to obtain the deviation change slope; if the slope is positive and the absolute value is greater than the preset slope threshold, it indicates that the change trend of the material characteristic parameter and the change trend of the weight deviation are positively correlated; when both of the above conditions are met, it is determined to be a material fluctuation deviation, and the type and magnitude of the material characteristic parameter that triggered the determination are recorded.

[0030] Specifically, when determining the type of weight deviation, the process for determining equipment wear deviation includes: extracting cumulative runtime data from equipment operating status parameters, recording the cumulative runtime and weight deviation data corresponding to each cutting operation in chronological order, and constructing an equipment runtime-weight deviation dataset; plotting a weight deviation change curve based on this dataset, and preprocessing the curve using a curve smoothing algorithm to eliminate the influence of random noise on trend judgment; analyzing the smoothed curve using a trend recognition algorithm, if the overall trend of the curve shows a continuous monotonically increasing or decreasing trend, and the duration corresponding to this trend covers a preset proportion of the equipment runtime; then the control system pauses the current cutting process and performs three repeated cutting tests under the same material characteristic parameters and the same preset cutting parameters, collecting the weight deviation data corresponding to the tests; if the test data still maintains the same continuous monotonically changing trend, it indicates that the influence of fluctuations in material characteristic parameters has been eliminated; after considering the above conditions, it is determined to be equipment wear deviation, and the current cumulative runtime and deviation change rate of the equipment are recorded.

[0031] Specifically, the implementation process of the weight deviation compensation model for matching the inherent deviation of the system includes: continuously acquiring real-time values ​​of the weight deviation and constructing a deviation time series; analyzing the deviation time series using a trend analysis algorithm to identify the trend of deviation change (increasing, decreasing, or stable); dynamically adjusting the control parameters in the proportional-integral-derivative control logic according to the identified trend—when the deviation shows an increasing trend, increasing the coefficient of the proportional control component to accelerate the response speed, and simultaneously increasing the coefficient of the integral control component to eliminate accumulated deviation; when the deviation shows a decreasing trend, appropriately decreasing the coefficient of the proportional control component to avoid overshoot and keeping the coefficient of the integral control component stable; calculating the proportional control component (the product of the real-time deviation value and the proportional coefficient), the integral control component (the product of the integral value of the deviation time series and the integral coefficient), and the derivative control component (the product of the deviation change rate and the derivative coefficient) based on the adjusted coefficients; superimposing the three components according to preset weights to obtain the compensation amount corresponding to the inherent deviation of the system; after superposition, performing range verification on the compensation amount; if it exceeds the preset single-type deviation compensation range, then adjusting the weights of the three components a second time until the compensation amount meets the requirements.

[0032] Specifically, the implementation process of the weight deviation compensation model for matching material fluctuation deviations includes: obtaining data on cutting weight deviations and compensation amounts corresponding to different material characteristic parameters (density, moisture content, viscosity) under varying degrees of difference through numerous orthogonal experiments; constructing a correlation mapping table between each material characteristic parameter and the compensation amount based on the experimental data, the mapping table containing parameter difference levels, corresponding basic compensation amounts, and influence weights; during actual compensation calculations, extracting actual material characteristic parameters from the standardized dataset and comparing them with preset standard material characteristic parameters to determine the degree of difference and classify the difference levels; retrieving the corresponding basic compensation amount from the correlation mapping table according to the difference level; simultaneously extracting the influence weight of the characteristic parameter on the cutting weight from the mapping table, multiplying the basic compensation amount by the influence weight to obtain the preliminary correction compensation amount; summing the preliminary correction compensation amounts corresponding to all material characteristic parameters to obtain the final compensation amount corresponding to the material fluctuation deviation; verifying the matching between the compensation amount and the variation range of the material characteristic parameters after summation; if they do not match, retrieving the weights from the mapping table for adjustment.

[0033] Specifically, the implementation process of the multimodal compensation model dynamic fusion mechanism includes: first, establishing a deviation type weight allocation model, in which initial weight coefficients for different deviation types under different cutting conditions are preset in the model; substituting the deviation type determination results into the model, and combining the material characteristic parameter types and equipment operating status parameters in the standardized data set, dynamically correcting the initial weight coefficients—when the material characteristic parameters fluctuate significantly, increasing the weight coefficient of material fluctuation deviation; when the cumulative operating time of the equipment approaches the wear threshold, increasing the weight coefficient of equipment wear deviation; after correction, ensuring that the sum of the weight coefficients of the system inherent deviation, material fluctuation deviation, and equipment wear deviation is one; multiplying the compensation amount corresponding to each deviation by its own corrected weight coefficient to obtain the weighted compensation amount of each deviation; summing all weighted compensation amounts to obtain the initial fusion compensation amount; using a moving average algorithm to smooth the initial fusion compensation amount to eliminate the fluctuations caused by the superposition of different deviation compensation amounts; after smoothing, obtaining the final adaptive compensation amount, which is output to the subsequent fusion optimization stage.

[0034] Specifically, the process of setting the upper limit constraint of the compensation amount includes: firstly, calculating the mechanical load-bearing limit of the cutting system through mechanical analysis, and determining the processable range of the material by combining the physical properties of the material (such as brittleness and toughness); using the maximum compensation capacity corresponding to the mechanical load-bearing limit as the upper limit basis, and the minimum compensation requirement corresponding to the processable range of the material as the lower limit basis, and taking into account the engineering safety margin, determining the upper and lower limits of the compensation amount constraint range; comparing the fused and optimized compensation amount with the upper and lower limits of the constraint range one by one; if the compensation amount exceeds the upper limit, the upper limit value is taken as the final compensation amount; if the compensation amount is lower than the lower limit, the lower limit value is taken as the final compensation amount; if the compensation amount is within the range, the fused and optimized compensation amount is directly taken as the final compensation amount; synchronously recording the over-limit information, including the over-limit time, over-limit type (upper limit / lower limit), original compensation amount value, final determined compensation amount value and corresponding cutting condition parameters, and storing the over-limit information in the abnormal data log to provide a basis for subsequent equipment maintenance.

[0035] Specifically, the implementation process of periodically acquiring historical data includes: setting a fixed historical data collection cycle based on the batch characteristics of cutting production; the collection cycle can be adaptively adjusted according to the production batch size (shorter cycle for larger batches, longer cycle for smaller batches); after each collection cycle, the system automatically triggers a data aggregation command to extract all standardized data sets, deviation type judgment results, fusion compensation parameters, and cutting execution result data (including corrected cutting parameters, actual cutting weight, cutting pass rate, etc.) within that cycle from the data storage module; classifying and archiving the aggregated data according to data type, adding timestamps, material numbers, equipment numbers, and other identification information to each type of data for easy subsequent querying and retrieval; storing the classified and archived data in a preset storage format to the historical database; simultaneously starting a data cleaning program to query data in the historical database that exceeds the preset storage time and delete it in order of data generation time from earliest to latest; backing up and verifying the data before deletion to ensure that important historical data has been backed up to avoid data loss; and recording a detailed log of this data aggregation, archiving, and cleaning process after data cleaning is completed.

[0036] Specifically, the implementation process of iteratively updating the control parameters of the weight deviation compensation model includes: first, collecting historical cutting data, deviation data, and compensation effect data within a preset time period; cleaning the data to remove invalid and abnormal data; dividing the cleaned dataset into a training dataset and a validation dataset according to a preset ratio, whereby the training dataset is used for parameter adjustment and the validation dataset is used for effect verification; based on the training dataset, using the gradient descent algorithm as the core data-driven algorithm, and aiming to minimize the compensation effect error, iteratively adjusting the control parameters of the weight deviation compensation model; after each adjustment, substituting the adjusted parameters into the validation dataset, calculating the error value of the compensation effect (the difference between the actual compensation effect and the ideal compensation effect); setting an error threshold, if the calculated error value is less than or equal to the error threshold, then the compensation effect is determined to be satisfactory, and the currently adjusted parameters are determined to be the updated control parameters; if the error value is greater than the error threshold, returning to the training stage, adjusting the step size of the gradient descent algorithm, and readjusting the parameters; repeating the above training-validation process until the error value is satisfactory; summarizing all satisfactory control parameters to generate a compensation model parameter set, and simultaneously recording the parameter update process data (number of adjustments, error value each time, and final error value).

[0037] This embodiment takes the plastic granule cutting processing scenario as an example, where: the preset cutting parameters are denoted as set P={P1,P2,P3} (P1 is the target cutting weight, P2 is the preset cutting speed, P3 is the preset cutting length); the actual cutting weighing feedback data is denoted as W; the material characteristic parameters are denoted as M={M1,M2,M3} (M1 is the material density, M2 is the material moisture content, M3 is the material viscosity); and the equipment operating status parameters are denoted as E={E1,E2,E3} (E1 is the cutter rotation speed, E2 is the feeding mechanism propulsion speed, and E3 is the cumulative equipment running time). The specific implementation steps are as follows: S1 data acquisition and preprocessing to generate standardized datasets S1.1 Data Acquisition: The parameter acquisition module of the cutting system acquires the preset cutting parameter set P, the actual cutting weighing feedback data W, the material characteristic parameter set M, and the equipment operating status parameter set E in real time, forming the original data matrix D=[P,W,M,E].

[0038] S1.2 Noise Filtering: (1) Extract the material viscosity M3 in the material characteristic parameters and the cutter rotation speed E1 in the equipment operating status parameters. Use the Pearson correlation analysis algorithm to calculate the linear correlation coefficient r between the two. The calculation process is: r=Cov(M3,E1) / (σ(M3)×σ(E1)), where Cov(M3,E1) is the covariance of M3 and E1, σ(M3) is the standard deviation of M3, and σ(E1) is the standard deviation of E1. (2) Dynamically adjust the time span of the filtering window according to the magnitude of the linear correlation coefficient r: Set the correlation coefficient threshold to r0. If |r|>r0, adjust the time span of the filtering window to T1; if |r|≤r0, adjust the time span of the filtering window to T2 (T1>T2). (3) Based on the adjusted filter window, the actual material cutting and weighing feedback data W is smoothed by moving average to filter high-frequency noise signals. The calculation process is as follows: Assume that the adjusted filter window contains n continuous data points, and the smoothed weighing data W'_k=∑(i=k-n+1 to k)W_i / n, where k is the current data point number and W_i is the i-th original weighing data point. Finally, the denoised weighing data sequence W' is obtained.

[0039] S1.3 Outlier Removal: (1) Collect historical data on the cutting and processing of similar plastic granules, and perform statistical analysis on the preset cutting parameters P, material characteristic parameters M, and equipment operating status parameters E in the historical data. Calculate the mean μ and standard deviation σ of each parameter, and construct the normal data distribution interval [μ-3σ, μ+3σ] for each parameter. (2) Substitute the various parameters in the original data matrix D obtained in S1.1 into the corresponding normal data distribution interval for comparison. If a parameter value x∉[μ-3σ,μ+3σ], then x is initially determined to be an outlier. (3) Perform secondary verification on the initially identified outliers: extract the parameter data x₋1 (the previous time) and x₊1 (the next time) of the time adjacent to the outlier, calculate the change between the outlier and the adjacent data Δx₋1=|xx₋1| and Δx₊1=|xx₊1|, set the mutation threshold Δ0, if Δx₋1>Δ0 and Δx₊1>Δ0, then x is finally determined to be an outlier; (4) Supplement the data that is finally determined to be outliers using linear interpolation: Let the time corresponding to the outlier be t, the previous valid time be t₋1 and the data be x₋1, the next valid time be t₊1 and the data be x₊1, and the supplemented data be x_t=x₋1+(x₊1-x₋1)×(tt₋1) / (t₊1-t₋1); S1.4 Generate a standardized dataset: Normalize all parameters after noise filtering and outlier removal. The normalization formula is x''=(x'-x_min) / (x_max-x_min) (x' is the processed data, x_min is the historical minimum value of the parameter, and x_max is the historical maximum value of the parameter). Finally, a standardized dataset S=[P'',W'',M'',E'' is generated.

[0040] S2: Deviation calculation and type determination, generating deviation type determination results and deviation characteristic data. S2.1 Weight Deviation Calculation: Based on the target cutting weight P1'' in the standardized data set S and the noise-reduced actual cutting weight data W'', the weight deviation e is calculated. The calculation process is e=W''-P1''.

[0041] S2.2 Construction of the Deviation Judgment Benchmark Library: (1) The historical cutting data is divided into subsets according to the material type. In this embodiment, it is the plastic granule subset. Within this subset, it is divided into multiple time periods according to the cumulative running time of the equipment E3, denoted as G1 (E3∈[0,T). a ]), G2 (E3∈(T a 2T a ]), ..., G n (E3∈((n-1)T) a ,nT a ]); (2) Perform deviation statistics on the qualified cutting data in each time period group, calculate the maximum deviation e_max and minimum deviation e_min for each group, set the correction coefficient k in combination with the allowable cutting accuracy requirements of the project, and output the qualified cutting deviation range of each group as [e_min×k, e_max×k]; (3) Group the plastic granules by type and time period G1~G n The corresponding acceptable cutting deviation ranges are compiled and archived to construct a deviation judgment benchmark library B.

[0042] S2.3 Deviation Type Determination: An adaptive identification and classification mechanism for deviation characteristics is introduced, combining the fluctuation of material characteristic parameters in the deviation determination benchmark library B and the standardized data set S, as well as the changing trends of equipment operating status parameters, as detailed below: (1) Determination of material fluctuation deviation: ① Based on the standardized dataset S, extract m consecutive sets of material characteristic parameters M'' and corresponding weight deviations e to form a material characteristic-deviation data sequence S_ME=[(M''1,e1),(M''2,e2),...,(M''_m,e_m)]; ② Calculate the variation difference ΔM''_i=M''_i-M''_{i-1} (i=2 to m) between two adjacent sets of material characteristic parameters. Calculate the continuous variation differences of material density M1'', material moisture content M2'', and material viscosity M3'' respectively, and sum them up to obtain the cumulative variation ΣΔM1''=∑(i=2 to m)|ΔM1''_i|, ΣΔM2''=∑(i=2 to m)|ΔM2''_i|, ΣΔM3''=∑(i=2 to m)|ΔM3''_i|; ③ Set a preset material fluctuation benchmark ΔM0, and compare ΣΔM1'', ΣΔM2'', and ΣΔM3'' with ΔM0 respectively. If any cumulative summation result ΣΔM''_j>ΔM0 (j=1,2,3), it is determined to be a material fluctuation deviation. Record the material characteristic parameter type (M1'' / M2'' / M3'') and the cumulative change ΣΔM''_j that triggered the judgment.

[0043] (2) Determination of equipment wear deviation: ① Extract the cumulative running time E3'' and the corresponding weight deviation e from the standardized dataset S, and organize them in chronological order to construct the equipment running time-weight deviation dataset S_EE=[(E3''1,e1),(E3''2,e2),...,(E3''_m,e_m)]; ② The linear fitting algorithm is used to analyze the trend of weight deviation. The fitting function is e=a×E3''+b (a is the trend slope and b is the intercept). If a>0, the deviation shows an increasing trend, and if a<0, it shows a decreasing trend, which means it is determined to be a continuous unidirectional trend. ③ Set the preset duration baseline E 30 The cumulative running time of the device, E3'', is compared with E 30 Comparison, if E3''>E 30 If the deviation shows a continuous unidirectional trend, it is determined to be a wear deviation of the equipment. Record the current cumulative running time of the equipment E3'' and the deviation change rate v=a (the fitting slope is the change rate).

[0044] (3) System inherent deviation judgment: If neither of the above two deviation judgment conditions is met, and the weight deviation e exceeds the range of qualified cutting deviation matched in the deviation judgment benchmark library B, then it is judged as a system inherent deviation.

[0045] S2.4 Generation Results: Output deviation type determination result R (R=1 is the system inherent deviation, R=2 is the material fluctuation deviation, R=3 is the equipment wear deviation) and deviation characteristic data F=[e,ΣΔM''_j (if R=2),E3'',v (if R=3)].

[0046] S3: Compensation calculation and fusion, generating fusion compensation parameters. S3.1 Matching weight deviation compensation model: Based on the deviation type determination result R and deviation characteristic data F, the corresponding weight deviation compensation model is matched: R=1 matches the proportional-integral-derivative control logic compensation model, R=2 matches the material characteristic parameter correction compensation model, R=3 matches the equipment operating state attenuation law compensation model (this embodiment takes the simultaneous existence of three types of deviation as an example).

[0047] S3.2 Calculation of various deviation compensation amounts: (1) Calculation of inherent deviation compensation for the system: ① Based on the deviation characteristic data F, extract the real-time value of weight deviation e_t, the cumulative effect of deviation Σe (Σe=∑(τ=0 to t)e_τ, where τ is a time variable) and the deviation change rate v_e=de / dt; ② Substitute the above parameters into the compensation model constructed by the proportional-integral-derivative control model, and calculate each control component: proportional control component u_p=K_p×e_t (K_p is the proportional coefficient), integral control component u_i=K_i×Σe (K_i is the integral coefficient), and derivative control component u_d=K_d×v_e (K_d is the derivative coefficient). ③ Set the weights of each component ω_p, ω_i, and ω_d (ω_p+ω_i+ω_d=1), and superimpose the three adjustment components to obtain the system's inherent deviation compensation amount ΔW1=ω_p×u_p+ω_i×u_i+ω_d×u_d.

[0048] (2) Calculation of material fluctuation deviation compensation: ① Obtain cutting weight deviation and compensation data corresponding to different material characteristic parameters through orthogonal experiments in advance, and construct an association mapping table T, which includes parameter difference level L (L=1 to 5) and corresponding basic compensation amount ΔW. 20 Influence weight ω_M; ② Extract the actual material characteristic parameters M'' from the standardized data set S, compare them with the preset standard material characteristic parameters M0'', calculate the degree of difference d_j=|M''_j-M0''_j| (j=1,2,3), and classify the difference level L_j according to d_j; ③ Retrieve the basic compensation amount ΔW for each parameter at level L_j from the mapping table T. 20 Given the influence weights _j and ω_Mj, calculate the initial correction compensation amount ΔW2_j = ΔW 20 _j×ω_Mj; ④ Sum the preliminary correction compensation amounts for all material characteristic parameters to obtain the material fluctuation deviation compensation amount ΔW2=∑(j=1 to 3)ΔW2_j.

[0049] (3) Calculation of equipment wear deviation compensation: The compensation model based on the attenuation law of equipment operation status is adopted. The calculation process is ΔW3=K_w×E3''+b_w (K_w is the wear compensation coefficient, and b_w is the initial wear compensation benchmark value).

[0050] S3.3 Dynamic Fusion of Multimodal Compensation Model: (1) Based on the deviation type determination result R, determine the compensation amounts ΔW1, ΔW2, and ΔW3 corresponding to the three types of deviations; (2) Based on the fluctuation of material characteristic parameters ΣΔM''_j and the cumulative running time of equipment E3'' in the standardized data set S, dynamically determine the weight coefficients ω1, ω2, and ω3 of each deviation (ω2 increases when the fluctuation is large, ω3 increases when E3'' is large, and ω1+ω2+ω3=1). (3) Calculate the weighted compensation for each deviation: ω1×ΔW1, ω2×ΔW2, ω3×ΔW3; (4) Summing up the initial fusion compensation amount ΔW_initial = ω1×ΔW1 + ω2×ΔW2 + ω3×ΔW3; (5) The initial fusion compensation amount is smoothed by the moving average algorithm. The smoothing formula is ΔW_flat = (ΔW_initial + ΔW_initial-1 + ΔW_initial-2) / 3 (ΔW_initial-1 and ΔW_initial-2 are the initial fusion compensation amounts in the first two moments), and the final adaptive compensation amount ΔW_self = ΔW_flat is generated.

[0051] S3.4 Upper Limit Constraint on Compensation Amount: (1) Pre-set the compensation amount constraint range as [ΔW_min, ΔW_max] to form the upper limit constraint standard for compensation amount; (2) Compare the adaptive compensation amount ΔW_self with the constraint range: if ΔW_self > ΔW_max, then the final compensation amount ΔW = ΔW_max; if ΔW_self < ΔW_min, then ΔW = ΔW_min; otherwise ΔW = ΔW_self; (3) Synchronously record the comparison result C (C=1 is the upper limit of the limit, C=2 is the lower limit of the limit, and C=3 is normal) and the final compensation amount ΔW, and generate the fusion compensation amount parameter Q=[ΔW,C].

[0052] S4: Parameter correction and iterative update to generate the compensation model parameter set. S4.1 Cutting Parameter Correction and Command Issuance: Based on the final compensation amount ΔW in the fusion compensation parameter Q, the target cutting weight P1 in the preset cutting parameter set P is corrected, and the correction formula is P1'=P1+ΔW; the corrected preset cutting parameter set P'=[P1',P2,P3] is used to generate cutting control commands, which are then sent to the cutting actuators (cutter drive module, feeding drive module).

[0053] S4.2 periodically retrieves historical data: (1) Set a preset period T, and trigger the historical data collection command every T interval; (2) Extract the standardized data set S, deviation type judgment result R, fusion compensation parameter Q, historical cutting data D_history, deviation data e_history, and compensation effect data Y_history within the period T (Y_history is the degree of fit between the corrected actual cutting weight and the target weight). (3)Classify and organize the extracted data, add time identifier t, material identifier M_id, and equipment identifier E_id, and store it in the historical database in the format F_st = [t, M_id, E_id, S, R, Q, D_his, e_his, Y_his].

[0054] S4.3 Iteratively update the compensation model control parameters: (1)Data collection and cleaning: Collect historical cutting data D_his, deviation data e_his, and compensation effect data Y_his within the preset time period T_total, clean the data, and use S1.2 - S1.3 to eliminate invalid data and abnormal data to obtain the cleaned dataset D_cln; (2)Dataset division: Divide the cleaned dataset D_cln into a training dataset D_trn and a validation dataset D_val according to the preset ratio k_div (e.g., k_div = 0.8), that is, D_trn contains 80% of the data in D_cln, and D_val contains the remaining 20% of the data. Among them, D_trn is used for parameter adjustment, and D_val is used for effect verification; (3)Parameter training: Based on the training dataset D_trn, use the gradient descent algorithm as the core algorithm driven by data, and aim to minimize the compensation effect error to iteratively adjust the core control parameters (K_p, K_i, K_d; K_w, b_w; ω_Mj, etc.) of the compensation model corresponding to the system inherent deviation, material fluctuation deviation, and equipment wear deviation; ① Set the initial parameter value θ0 = [K_p0, K_i0, K_d0, K_w0, b_w0, ω_Mj0,...], the learning step size is η, and the initial value of the iteration number n = 0; ② Calculate the compensation effect error value Loss = ∑(x∈D_trn)|Y_act(x) - Y_th(x)| / N_trn at the current parameter, where Y_act(x) is the actual compensation effect corresponding to the x data, Y_th(x) is the ideal compensation effect, and N_trn is the data volume of D_trn; ③ Calculate the gradient ∇Loss(θ n ) of the error Loss with respect to each parameter, and update the parameter according to the formula θ n+1 =θ n -η×∇Loss(θ n ), complete one iteration, and n = n + 1; (4)Effect verification: Substitute the updated parameter θ n+1 into the validation dataset D_val, and calculate the validation error Loss_val = ∑(x∈D_val)|Y_act(x) - Y_th(x)| / N_val (N_val is the data volume of D_val); (5)Iteration judgment: Set the error threshold Loss_0. If Loss_val ≤ Loss_0, it is determined that the compensation effect meets the standard, and the current parameter θ n+1is the updated regulation parameter; if Loss_verification > Loss_0, then adjust the step size η of the gradient descent algorithm (e.g., η = η × 0.9), and return to step (3) to re-adjust the parameters; (6) Repeat the above training-verification process until Loss_verification ≤ Loss_0; (7) Parameter summary: Organize and summarize all the qualified regulation parameters to generate the compensation model parameter set θ_update = [K_p_update, K_i_update, K_d_update, K_w_update, b_w_update, ω_Mj_update,...], and at the same time record the process data of parameter update: the number of iterations n, the error value Loss_n of each iteration, and the final verification error value Loss_verification.

[0055] S4.4 Parameter application: Store the generated compensation model parameter set θ_update in the parameter cache module for parameter calls of various compensation models in the subsequent S3 stage, to achieve continuous optimization of the compensation model.

[0056] The above are only the preferred embodiments of the present invention, and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments by using the disclosed technical content within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. An automatic correction and compensation method for material cutting and weighing, characterized in that, include: S1: Obtain preset cutting parameters, actual cutting and weighing feedback data, material characteristic parameters and equipment operating status parameters, perform noise filtering and outlier removal, and generate a standardized data set; S2: Based on the target cutting weight and actual cutting weighing feedback data in the standardized data set, calculate the weight deviation; introduce an adaptive identification and classification mechanism for deviation characteristics, and combine the preset deviation judgment benchmark library, the fluctuation of material characteristic parameters in the standardized data set, and the changing trend of equipment operating status parameters to determine the type of the weight deviation, and generate deviation type judgment results and deviation feature data. S3: Based on the deviation type determination results and deviation feature data, match the corresponding weight deviation compensation model for the deviation type; the weight deviation includes system inherent deviation, material fluctuation deviation and equipment wear deviation, and generate adaptive compensation amount based on the dynamic fusion mechanism of multimodal compensation model; The compensation amount corresponding to the weight deviation is fused and optimized, an upper limit constraint on the compensation amount is set, and fused compensation amount parameters are generated. S4: Based on the fusion compensation parameters, the preset cutting parameters are corrected, and the corrected cutting control command is generated and sent to the cutting execution mechanism; Historical data is periodically acquired, and the adjustment parameters of the weight deviation compensation model are iteratively updated by combining the standardized data set, deviation type determination results, and fusion compensation parameters to generate a compensation model parameter set.

2. The method according to claim 1, characterized in that, In S1, the noise filtering process specifically includes: extracting the material viscosity data from the material characteristic parameters and the cutter rotation speed data from the equipment operating status parameters; calculating the linear correlation coefficient between the two through correlation analysis; dynamically adjusting the time span of the filtering window according to the magnitude of the linear correlation coefficient; and performing moving average smoothing processing on the actual cutting and weighing feedback data based on the adjusted filtering window to filter high-frequency noise signals.

3. The method according to claim 1, characterized in that, In S1, the specific implementation process of outlier removal includes: statistical analysis of historical data related to cutting of similar materials, calculation of the mean and standard deviation of each parameter, and construction of normal data distribution intervals for each parameter; comparison of the obtained parameters into the corresponding distribution intervals, and further verification of the initially determined outliers by combining the parameter change trends at adjacent times; for the data finally determined to be outliers, using the valid data at two adjacent times before and after the outlier as interpolation nodes, and calculating the supplementary data corresponding to the outlier time based on the time interval and numerical difference between the two nodes.

4. The method according to claim 1, characterized in that, In S2, the specific process of setting the deviation judgment benchmark includes: classifying and sorting historical cutting data, dividing the data into subsets according to material type, and further dividing each data subset into different time period groups according to equipment running time; performing deviation statistics on qualified cutting data in each time period group, and outputting the qualified cutting deviation range for each group in combination with the allowable cutting accuracy requirements of the project; and organizing and archiving the material type, equipment running time and corresponding qualified cutting deviation range of all groups to construct the deviation judgment benchmark library.

5. The method according to claim 1, characterized in that, In S2, when determining the type of weight deviation, the determination process for material fluctuation deviation includes: based on the standardized data set, extracting material characteristic parameters and corresponding weight deviation data to form a material characteristic-deviation data sequence; calculating the change difference between two adjacent sets of material characteristic parameters in the material characteristic-deviation data sequence, respectively statistically analyzing the continuous change difference of material density, material moisture content, and material viscosity and summing them, comparing the summed result with a preset material fluctuation benchmark, and recording the type and magnitude of the material characteristic parameter that triggered the determination.

6. The method according to claim 1, characterized in that, In S2, when determining the type of weight deviation, the process for determining equipment wear deviation includes: extracting the cumulative running time data and corresponding weight deviation data from the equipment operating status parameters in the standardized data set, organizing and constructing the equipment running time-weight deviation dataset in chronological order; analyzing the changing trend of the weight deviation, comparing the cumulative running time of the equipment with the preset time benchmark, and recording the current cumulative running time of the equipment and the rate of change of deviation.

7. The method according to claim 1, characterized in that, In S3, the implementation process of the weight deviation compensation model for matching the inherent deviation of the system includes: based on the deviation characteristic data, extracting the real-time value of the weight deviation, the cumulative effect of the deviation, and the rate of change of the deviation, and substituting them into the compensation model constructed based on proportional-integral-derivative control logic; calculating the proportional adjustment component based on the real-time value, calculating the integral adjustment component based on the cumulative effect of the deviation, and calculating the derivative adjustment component based on the rate of change of the deviation; and superimposing the three adjustment components according to preset weights to obtain the compensation amount corresponding to the inherent deviation of the system.

8. The method according to claim 1, characterized in that, In S3, the implementation process of the weight deviation compensation model for matching material fluctuation deviations includes: acquiring data on cutting weight deviations and compensation amounts corresponding to different material characteristic parameters under different degrees of difference; constructing a correlation mapping table between each material characteristic parameter and the compensation amount, including parameter difference level, corresponding basic compensation amount, and influence weight; during actual compensation calculation, extracting the actual material characteristic parameters from the standardized data set, comparing them with preset standard material characteristic parameters, determining the degree of difference, and classifying the difference level; retrieving the corresponding basic compensation amount from the correlation mapping table according to the difference level, and simultaneously extracting the influence weight on the cutting weight from the mapping table to calculate the preliminary correction compensation amount; summing the preliminary correction compensation amounts corresponding to all material characteristic parameters to obtain the compensation amount for the material fluctuation deviation.

9. The method according to claim 1, characterized in that, In S3, the implementation process of the dynamic fusion mechanism of the multimodal compensation model includes: determining the compensation amount corresponding to the system inherent deviation, material fluctuation deviation, and equipment wear deviation based on the deviation type determination result; determining the weight coefficient corresponding to each deviation according to the fluctuation of material characteristic parameters and equipment operating status parameters in the standardized data set; multiplying the compensation amount corresponding to each deviation by its own weight coefficient to obtain the weighted compensation amount of each deviation; summing all weighted compensation amounts to obtain the initial fusion compensation amount; and smoothing the initial fusion compensation amount to generate the final adaptive compensation amount.

10. The method according to claim 1, characterized in that, In S3, the process of setting the upper limit constraint of the compensation amount includes: pre-setting the compensation amount constraint range to form the upper limit constraint standard of the compensation amount; comparing the adaptive compensation amount with the upper limit constraint of the compensation amount, and synchronously recording the comparison result of the compensation amount and the final determined compensation amount value.

11. The method according to claim 1, characterized in that, In S4, the implementation process of periodically acquiring historical data includes: triggering a historical data acquisition command according to a preset cycle, extracting the standardized data set, deviation type judgment result, fusion compensation parameter, historical cutting data, deviation data and compensation effect data within the cycle; classifying and organizing the extracted data, adding time, material and equipment identification information to the organized data, and storing it in the historical database according to a preset format.

12. The method according to claim 1, characterized in that, In S4, the implementation process of iteratively updating the control parameters of the weight deviation compensation model includes: constructing a parameter update dataset based on the historical cutting data, deviation data, and compensation effect data, combined with the standardized data sets generated at each stage, deviation type determination results, and fusion compensation amount parameters; adjusting the control parameters of the compensation model corresponding to the system's inherent deviation, material fluctuation deviation, and equipment wear deviation using a data-driven approach with the goal of achieving the best compensation effect; and summarizing all updated control parameters to generate a compensation parameter set.

Images