Automobile seal strip production parameter collaborative control method based on environmental load constraint

Abstract

The application discloses a car sealing strip production parameter collaborative control method based on environmental load constraints, and particularly relates to the field of production parameter collaborative control, which comprises the following steps: collecting and preprocessing five-dimensional data such as environmental load and equipment operation, building a prediction model, and constructing a dynamic threshold system in combination with multi-dimensional requirements; then carrying out hierarchical coupling analysis to obtain a full-system coupling characteristic set, constructing three types of coupling functions and outputting quantitative values; subsequently comprehensively judging the production state, positioning the root cause of the optimized parameters and outputting a standardized optimization instruction set; finally, completing the initialization of the optimized parameters, executing collaborative optimization by using a particle swarm optimization algorithm with constraints, verifying the effect and closing the loop for iteration. The application realizes multi-dimensional coupling analysis based on the physical logic of the production process, discards subjective weights, realizes dynamic self-adaptive control of parameters, takes into account environmental protection compliance, production efficiency and product consistency, and is suitable for the large-scale and multi-specification production scene of sealing strips.

Description

Technical Field

[0001] The present invention relates to the technical field of coordinated control of production parameters, and more specifically, to a method for coordinated control of production parameters of automotive sealing strips based on environmental load constraints. Background Art

[0002] As a core component for automotive sealing, sound insulation, and shock absorption, the production process of automotive sealing strips covers multiple processes such as raw material mixing, extrusion, vulcanization, cooling, and inspection. The control accuracy of production parameters directly affects product consistency and production efficiency. Related parameter control technologies have become the focus of research and application in the field of automotive parts manufacturing. Currently, the automotive manufacturing industry is developing towards green, intelligent, and large-scale directions, and environmental protection constraint indicators have been gradually incorporated into the whole-process control system of production. The monitoring and consideration of environmental load-related data have become an important part of sealing strip production. In the production data collection link, various sensors, online detection equipment combined with offline supplementary recording methods have been widely used in the industry to carry out time-series data collection in dimensions such as production, equipment, and quality. The digitization and standardization of data collection are continuously improving. At the same time, intelligent algorithms and data analysis models are gradually applied to the optimization of production parameters. Various coupling analysis methods, threshold setting strategies, and intelligent optimization algorithms are gradually introduced into the production parameter control process. The industry as a whole is developing towards the direction of multi-dimensional data fusion and multi-process linkage control.

[0003] However, when it is actually used, there are still some disadvantages, such as:

[0004] 1. The control of production parameters and the monitoring of environmental load are in a separated state. Only process parameters are controlled separately or environmental protection indicators are monitored, and no associated control system between the two is established. It is impossible to achieve dynamic adjustment of parameters under environmental protection constraints, and it is difficult to balance environmental protection compliance and production efficiency.

[0005] 2. The dimension of production data analysis is single, mostly focusing on a single level of production parameters or product quality, without integrating key dimensions such as equipment operation and raw material characteristics. The analysis results lack comprehensiveness and cannot accurately locate the core root cause of production anomalies.

[0006] 3. The determination of parameter optimization depends on fixed thresholds or manual experience, the determination logic is rigid, there is no quantitative coupling index as a basis, it is easy to have misjudgment or deviation in the optimization direction, and the optimization amplitude has no scientific calculation, which is easy to cause production system oscillation.

[0007] 4. There is a lack of a full-process closed-loop iterative control mechanism. After parameter optimization, the production status is not verified in real time, and there is no dynamic iterative strategy based on the verification results. The optimization effect cannot be effectively guaranteed, and it is difficult to achieve dynamic adaptive control of production parameters. Summary of the Invention

[0008] To overcome the aforementioned deficiencies of the prior art, this invention provides a collaborative control method for automotive sealing strip production parameters based on environmental load constraints, which addresses the problems mentioned in the background art through the following solutions.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a collaborative control method for automotive sealing strip production parameters based on environmental load constraints, comprising:

[0010] Multi-dimensional data acquisition and processing module: Collects data from five dimensions: environmental load, equipment operation, core production parameters, raw material characteristics, and product quality prediction; performs preprocessing operations to build a preprocessed dataset; builds a prediction model to output predicted values; and constructs a dynamic threshold system by combining historical data, environmental standards, and production conditions to output the constraint threshold range for each dimension.

[0011] Multi-dimensional hierarchical coupling analysis module: Based on the preprocessed dataset, it performs single-dimensional time series feature analysis and outputs feature vectors; it constructs a coupling adjacency matrix, obtains the coupling strength matrix through mutual information entropy quantization and normalization, and identifies strongly coupled subspaces; it expands the coupling matrix, calculates the global coupling transfer coefficient, and outputs the coupling feature set of the entire system;

[0012] Strongly Coupled Multi-Objective Function Construction Module: Based on the coupling feature set of the whole system, threshold deviation and dynamic threshold, the environmental load coupling compliance function and the production parameter process coordination function are constructed in sequence. Then, the coupling strength between quality and load and parameters is integrated to construct the global coupling function and output the three types of function quantification values.

[0013] The comprehensive judgment module: Based on the coupling feature set of the whole system, the quantization values ​​of three types of functions and the constraint threshold range of each dimension, it initially screens strong anomaly dimensions and locates their respective strong coupling subspaces, calculates the anomaly propagation contribution, and locates the root cause optimization parameters; based on the value of the global coupling function, it determines the global production status and optimization constraint direction, and integrates the results to output a standardized judgment instruction set;

[0014] Collaborative tuning and closed-loop iteration module: Initializes the tuning parameter step size and search interval based on a standardized judgment instruction set, with the optimization objective of minimizing the global coupling function value, and uses a constrained particle swarm optimization algorithm to perform collaborative tuning of multi-process parameters and output the optimal parameter combination; verifies the tuning effect, and performs closed-loop iterative control based on the verification results.

[0015] The technical effects and advantages of this invention are as follows:

[0016] 1. Based on the physical logic of the production process, a multi-dimensional hierarchical coupling analysis is carried out, embedding auxiliary dimensions such as equipment and raw materials. There is no human subjective weighting involved, and the coupling strength and transmission coefficient are both quantified by data. The analysis results are consistent with the actual production conditions, which greatly improves the scientificity and objectivity of parameter control.

[0017] 2. Construct a multi-objective coupling function system to integrate the full-dimensional correlation of environmental load, production parameters, and product quality. Through function quantification, the coupling state of each system and the whole system is characterized, providing accurate and quantifiable basis for production judgment and avoiding the one-sidedness of single indicator judgment.

[0018] 3. Achieve closed-loop management of the entire process from data collection to optimization iteration, verify the effect in real time after optimization and dynamically adjust the strategy, iterate and optimize if the target is not met, and link manual intervention when anomalies occur to ensure the continuity and adaptability of parameter management.

[0019] 4. Integrate environmental load constraints into the entire process of parameter collaborative management, taking into account environmental compliance, production efficiency and product consistency, adapting to large-scale and multi-specification sealing strip production scenarios, and conforming to the green and intelligent development trend of automobile manufacturing. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall structure of the present invention.

[0021] Figure 2 This is a schematic diagram of the data acquisition and preprocessing module of the present invention.

[0022] Figure 3 This is a schematic diagram of the multi-dimensional hierarchical coupling analysis module of the present invention.

[0023] Figure 4 This is a schematic diagram of the comprehensive determination module of the present invention.

[0024] Figure 5 This is a schematic diagram of the collaborative optimization and closed-loop iteration module of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] refer to Figure 1 - Figure 5 The illustrated method for coordinated control of automotive sealing strip production parameters based on environmental load constraints includes:

[0027] Multidimensional data acquisition and processing module: It uses a combination of methods to complete five-dimensional data acquisition and define standardized parameters; it constructs three types of standardized time-series datasets based on the acquired data; it performs four preprocessing steps on the original dataset: noise reduction, deduplication, completion, and standardization, and outputs a preprocessed dataset; it builds a load-parameter-quality linkage prediction model based on the preprocessed dataset and outputs the predicted values ​​of load and quality; it combines historical data, environmental standards, and production conditions to construct a dynamic threshold system and output the constraint threshold ranges for each dimension.

[0028] It should be further explained that the data collection for the five dimensions adopts a combination of real-time sensor acquisition, online detection, and offline supplementary recording. Data collection is completed in five dimensions, and the acquisition device and acquisition object are clearly defined for each type of data, with standardized parameter symbols defined as follows:

[0029] Environmental load data: Collected through energy consumption sensors, VOCs online monitors, dust sensors, and equipment load detectors, covering environmental constraints related to the entire production process, and recorded as follows: ,in:

[0030] Energy consumption per unit of process (kWh / h): Real-time power consumption of extrusion and vulcanization processes is collected by energy consumption sensors and converted to hourly average.

[0031] VOCs emission concentration (mg / m³) is the real-time concentration of exhaust gas from the sulfurization process, collected by an online VOCs monitoring instrument.

[0032] Dust emission concentration (mg / m³) is the real-time dust emission concentration collected by dust sensors during the raw material mixing and cutting processes.

[0033] The overall operating load of the equipment (%) is calculated by the equipment load detector, which collects the real-time load of each production equipment (combined with speed and pressure).

[0034] Equipment operating data: collected through vibration sensors, temperature sensors, speed sensors, and pressure sensors, focusing on the impact of equipment operating status on parameters and load, and recorded as follows. ,in:

[0035] The equipment vibration value (mm / s) is collected by vibration sensors from the real-time vibration data of the extruder and vulcanizing machine body.

[0036] Equipment temperature rise (°C): The temperature difference between the equipment body temperature and the ambient temperature during operation is collected by a temperature sensor.

[0037] The actual rotational speed (r / min) of the equipment is collected by a speed sensor, which measures the real-time rotational speed of the extruder screw and the raw material mixer.

[0038] The equipment operating pressure (MPa) is the real-time pressure of the extruder die and vulcanizing cylinder, which is collected by pressure sensors.

[0039] Production core parameters: These are collected via a PLC controller and an online parameter acquisition unit, focusing on key process parameters in the production of sealing strips, and are denoted as follows: ,in:

[0040] Extrusion speed (m / min): The real-time extrusion speed of the extruder is collected by the PLC controller and precisely controlled to 0.01m / min;

[0041] Vulcanization temperature (°C): The real-time temperature inside the vulcanizing machine cavity is collected by a temperature sensor and PLC linkage, with a control accuracy of ±1°C.

[0042] Cooling water temperature (°C): The real-time water temperature of the cooling water tank is collected by a temperature sensor to meet the cooling requirements of the sealing strip after vulcanization.

[0043] Raw material characteristic data: Data is collected offline using hardness testers and viscometers, and supplemented online, focusing on the implicit impact of raw material characteristics on production. This data is recorded as follows: ,in:

[0044] The Shore hardness (HA) of rubber is measured offline by a Shore hardness tester for each batch of raw materials and then recorded online into the system.

[0045] : Raw material mixing viscosity (Pa·s) is measured offline by a rotational viscometer and synchronously recorded to the system.

[0046] Product quality prediction data: Data is collected in real time using online dimensional measuring instruments and tensile sensors to predict product quality status, and recorded as follows. ,in:

[0047] The real-time dimensional accuracy (mm) of the sealing strip is obtained by collecting the cross-sectional dimensional deviation of the newly formed sealing strip using an online dimensional measuring instrument.

[0048] Real-time tensile strength (MPa): The real-time tensile strength of the sealing strip after cooling is collected by a tensile sensor.

[0049] Based on the five types of collected data mentioned above, three standardized datasets were constructed to meet the usage requirements of different subsequent modules. All datasets adopt a time-series format (timestamp + parameter value), as follows:

[0050] Basic Data Collection , where t is the collection timestamp, which contains all the original collected data and is used for data preprocessing and tracing;

[0051] It should be further explained that the preprocessing performs four standardization steps on the basic collected dataset to ensure that the data is consistent in scale, free of anomalies, and free of redundancy. The specific processing steps are as follows:

[0052] Noise reduction: Wavelet threshold noise reduction algorithm is used to remove instantaneous interference data collected by the sensor (such as extreme values ​​caused by voltage fluctuations).

[0053] Deduplication: Remove data with duplicate timestamps and retain the latest collected values ​​to avoid data redundancy;

[0054] Data completion processing: Linear interpolation is used to complete missing data caused by sensor failures and offline acquisition, ensuring temporal continuity;

[0055] Standardization: Z-score standardization is used to normalize data of different magnitudes to the [0, 1] interval. The standardization formula is: ;in For this dimension of data The historical average, where x is the original collected data. The standard deviation of the data in this dimension. This is the standardized data.

[0056] Based on the standardized data output from the data preprocessing unit, a preprocessed dataset is constructed to meet the needs of subsequent coupled analysis and function computation. The dataset adopts a time-series format (timestamp + standardized parameter values), specifically: Preprocessed dataset ;

[0057] It should be further explained that constructing the load-parameter-quality linkage prediction model includes using preprocessed datasets. An improved LSTM model that integrates CNN and attention mechanisms takes preprocessed multi-dimensional time-series data as input and outputs predicted environmental load values. Product quality prediction value T represents the predicted duration (5-10 minutes).

[0058] It should be further noted that the dynamic threshold system is based on the preprocessed dataset. Based on historical statistical results and combined with regional environmental protection standards and production conditions, a dynamic threshold system driven by "environmental protection standards + production conditions + historical data" is constructed. This system outputs the constraint threshold ranges for each dimension of data and constructs a threshold dataset. The details are as follows:

[0059] Dynamic threshold for environmental load:

[0060] The values ​​are dynamically adjusted based on environmental standards and historical load data; they are, in order, the minimum and maximum energy consumption per unit of process, the minimum and maximum VOCs emission concentration, the minimum and maximum dust emission concentration, and the minimum and maximum comprehensive operating load of the equipment.

[0061] Production parameter standard thresholds:

[0062] The parameters are dynamically adjusted based on production process requirements and historical best parameters; these are, in order, minimum and maximum extrusion speed, minimum and maximum vulcanization temperature, and minimum and maximum cooling water temperature.

[0063] Product quality pass threshold: The values ​​are dynamically adjusted based on product quality standards and historical testing data; they are, in order, the minimum and maximum real-time dimensional accuracy of the sealing strip, and the minimum and maximum real-time tensile strength.

[0064] By integrating the preprocessed dataset with load and quality predictions, a coupled modeling dataset is constructed to accommodate subsequent linked modeling and threshold update requirements. The dataset adopts a time-series format, as detailed below:

[0065] Coupled modeling dataset ,in , These are the predicted values ​​for load and quality.

[0066] Multi-dimensional hierarchical coupling analysis module: preprocessing datasets With threshold dataset First, perform single-dimensional time series feature analysis on standardized time series data of each independent dimension, calculate the sliding window standard deviation and threshold deviation, and output a single-dimensional state feature vector; then focus on the core coupling chain of environmental load-production parameters-product quality, construct a coupling adjacency matrix, quantify the coupling strength through mutual information entropy, cluster to identify strongly coupled subspaces, and output a coupling strength matrix; finally, embed auxiliary dimensions of equipment and raw materials to expand the coupling matrix, calculate the global coupling transmission coefficient, and output the coupling feature set of the entire system.

[0067] It should be further explained that the single-dimensional time series feature analysis is carried out as follows:

[0068] Base layer: Single-dimensional temporal feature analysis, analyzing preprocessed datasets. Standardized time series data for each independent dimension Specifically, it is as follows:

[0069] Perform sliding window standard deviation calculation on single-dimensional time series data to extract fluctuation characteristics: ,in:

[0070] : The standard deviation of the i-th data dimension within the sliding window. Its physical meaning is to characterize the real-time fluctuation of the data in that dimension. The larger the value, the worse the stability of the production data.

[0071] The sliding window length is set to 30 seconds. Physically, it matches the data acquisition frequency (1 time / second). It selects 30 consecutive sampling points to form an analysis window, reflecting short-term data fluctuations.

[0072] t: Data sampling timestamp, k is the current sampling time, t = k - +1 to k represent the sampling time range within the window;

[0073] : The standardized value of the i-th data dimension at time t, which physically means the real-time data collected in a single dimension after normalization;

[0074] : The standardized mean of the i-th data dimension within the current sliding window, physically representing the central tendency value of the data within the window.

[0075] Calculate the single-dimensional threshold deviation, which represents the degree of deviation from the compliance range:

[0076] ;

[0077] in, : Threshold deviation of the i-th data dimension, which physically means the degree of deviation of the real-time data of this dimension from the compliance threshold range. The value is [0, 1], and the larger the value, the more serious the deviation.

[0078] , : The lower and upper limits of the dynamic threshold for the i-th data dimension, which are physically the compliance boundary values ​​of the data in that dimension;

[0079] ϵ: Minimal constant for preventing zero denominator, with a value of 10. -6 The physical meaning is to avoid calculation errors when the upper and lower limits of the threshold interval are equal, but it has no practical physical meaning in production.

[0080] Output a single-dimensional state feature vector: The physical meaning is to integrate the fluctuation characteristics of single-dimensional data with the threshold deviation characteristics, and use them as the basic input unit for upper-level coupling analysis.

[0081] Core layer: cross-dimensional coupled correlation analysis; base layer outputs a single-dimensional state vector. It focuses on the core coupling chain of "environmental load X - production parameters Y - product quality Z"; specifically as follows:

[0082] Based on the physical logic of the sealing strip production process, a coupling adjacency matrix A is constructed. For matrix elements, the physical meaning is to characterize whether there is a direct production process coupling relationship between the i-th data dimension and the j-th data dimension. =1 indicates that there is direct coupling. =0 indicates no direct coupling.

[0083] Mutual information entropy is used to quantify the coupling strength, and the elements of the coupling strength matrix are obtained by normalization. Step 1: Calculate the original mutual information entropy between dimension i and dimension j:

[0084] ;

[0085] Step 2: Normalize the original mutual information entropy to obtain the elements of the coupling strength matrix. : ;

[0086] Where m and n are the total number of dimensions involved in the coupling analysis, after normalization. ∈[0,1], ensuring that the coupling strength between different dimensions is comparable;

[0087] Dimension With dimension The original mutual information entropy value quantifies the degree of nonlinear coupling between two data dimensions. The original value has no fixed range and only reflects the relative strength of coupling.

[0088] : The normalized coupling strength value between the i-th dimension and the j-th dimension, representing the nonlinear coupling strength between dimensions i and j, with a value of [0, 1]. The larger the value, the stronger the production correlation between the two.

[0089] Dimension , The joint probability density, the probability distribution of the simultaneous occurrence of data from two dimensions;

[0090] , Dimension , Marginal probability density, independent probability distribution of data in a single dimension;

[0091] The maximum value of the original mutual information entropy for all dimension pairs is used as the normalization benchmark, making... It is limited to the interval [0, 1].

[0092] Final output coupling strength matrix The matrix dimension is the same as the coupling adjacency matrix A, only for =1 (direct coupling exists) dimension pair computation , =0 =0.

[0093] Perform connectivity clustering (K-means clustering algorithm) on the non-zero elements in the coupling strength matrix C, with a clustering threshold set to 0.7 (i.e., (≥0.7 is considered strong coupling), identify the strongly coupled subspace, that is, group the related dimensions with high coupling strength into the same subspace, and locate the core coupled parameter group that has the most significant impact on production load and quality;

[0094] Association Layer: The core layer's strongly coupled subspace is connected to the auxiliary dimensions of equipment (S) and raw materials (M); its details are as follows:

[0095] Embedding the equipment and raw material auxiliary dimensions into the coupling adjacency matrix A expands it into a system-wide coupling matrix. By incorporating auxiliary dimensions such as equipment operation and raw material characteristics into the coupling system, a dimensional correlation matrix for the entire production process is constructed.

[0096] Calculate the global coupling transfer coefficient to characterize the transmission effect of parameters / loads along the process chain:

[0097] ;

[0098] in: The global coupling transmission coefficient of the i-th dimension to the j-th dimension quantifies the indirect coupling contribution on a single straight chain, reflecting the degree of implicit influence of a certain parameter / load on downstream dimensions after being transmitted along the production process.

[0099] n: Total number of data dimensions in the entire system, the number of all data dimensions involved in the coupling analysis;

[0100] , : Elements of the coupling strength matrix, representing the direct coupling strength between adjacent dimensions, and the product represents the superposition effect of indirect coupling transmission.

[0101] Output the entire system coupling feature set It integrates the coupling strength, transmission relationship and core coupling group of the entire system to provide the process coupling logic basis for subsequent function construction modules.

[0102] Strongly Coupled Multi-Objective Function Construction Module: Receives the entire system's coupling feature set Single-dimensional threshold deviation and dynamic threshold dataset First, an environmental load coupling compliance function is constructed based on the threshold deviation and coupling strength matrix C of the environmental load dimension. Furthermore, combining the coupling transmission coefficients of production parameters... Constructing production parameter process coordination functions with dynamic thresholds Finally, by integrating the coupling strength between the quality dimension and the load and parameters, a global coupling function for quality, load, and parameters is constructed. Output the quantized values ​​of three types of functions.

[0103] It should be further explained that the environmental load coupling compliance function is constructed. This includes: integrating the threshold deviation and inter-dimensional coupling strength of various dimensions of the environmental load to quantify the overall coupling compliance status of the environmental load system. The higher the deviation and coupling strength, the larger the function value, and the worse the compliance. The specific mathematical function is as follows:

[0104]

[0105] in, : Environmental load coupling compliance function value, a non-negative real number, whose physical meaning is to quantify the overall coupling compliance degree of the environmental load system. The smaller the value, the better the compliance of the load dimension.

[0106] N: Total number of environmental load dimensions (N=4, corresponding to energy consumption, VOCs, dust, and equipment load), physically representing the number of core environmental load dimensions involved in the calculation;

[0107] : Threshold deviation of the i-th environmental load dimension (from the single-dimensional time series feature analysis of the base layer), which physically means the degree of deviation of the real-time data of this load dimension from the compliance threshold;

[0108] : Normalized coupling strength between the i-th and j-th environmental load dimensions, physically representing the nonlinear correlation strength between the two load dimensions;

[0109] : The sum of the coupling strengths of the i-th load dimension with all other load dimensions, physically representing the total coupling weight of that dimension.

[0110] It should be further explained that constructing the production parameter process coordination function FY includes: based on the coupling transmission coefficient of the production parameter dimension, quantifying the coordination deviation of the production parameters along the process chain. The higher the deviation of the transmission coefficient from the threshold, the larger the function value, and the worse the process coordination. Its specific mathematical function is as follows:

[0111]

[0112] in, : Production parameter process coordination function value, a non-negative real number. Its physical meaning is to quantify the degree of overall coordination deviation of production parameters along the process chain. The smaller the value, the better the coordination of process parameters.

[0113] M: Total number of production parameter dimensions (M=3, corresponding to extrusion speed, vulcanization temperature, and cooling water temperature), physically representing the number of core dimensions of production parameters involved in the calculation;

[0114] : The standardized value of the j-th production parameter dimension at time t (from the preprocessed dataset);

[0115] : The standard threshold mean of the j-th production parameter The physical meaning is the optimal target value of this production parameter;

[0116] , : The lower and upper limits of the dynamic threshold for the j-th production parameter (from the dynamic threshold dataset), which physically represent the compliance boundary values ​​of the parameter;

[0117] : The global coupling transmission coefficient of the i-th production parameter to the j-th production parameter (from the coupling analysis of the entire system in the correlation layer), which physically represents the degree of indirect transmission influence of the upstream parameter on the downstream parameter;

[0118] The sum of the transmission coefficients of all upstream production parameters to the j-th parameter, which physically represents the total process transmission weight of that parameter.

[0119] It should be further explained that the global coupling function for quality, load, and parameters is constructed. This includes: integrating the coupling strength between product quality dimensions and environmental load and production parameters; quantifying the global coupling state of the entire system; the worse the load / parameter coupling compliance and the higher the quality deviation, the larger the function value, indicating a worse overall system production state. The specific mathematical function is as follows:

[0120]

[0121] in, : Global coupling function value, a non-negative real number, whose physical meaning is to quantify the overall coupling state of the entire system of quality, load and parameters. The smaller the value, the better the production state of the entire system.

[0122] I(Z; X): Normalized mutual information entropy (i.e., in the coupling strength matrix) between the product quality dimension and the environmental load dimension. The physical meaning of is the nonlinear coupling strength between mass and load dimensions;

[0123] I(Z;Y): Normalized mutual information entropy between the product quality dimension and the production parameter dimension (i.e., in the coupling strength matrix) The physical meaning of is the nonlinear coupling strength between mass and parameter dimensions;

[0124] Overall threshold deviation in product quality dimensions. ( For dimensional accuracy deviation, (Tensile strength deviation), physically meaning the overall degree of deviation in the mass dimension.

[0125] The comprehensive judgment module receives the system's coupling feature set, strongly coupled subspace set, strongly coupled multi-objective function values, single-dimensional state feature vectors, and dynamic threshold datasets. It first screens strong anomaly dimensions based on single-dimensional state feature vectors and locates the corresponding strongly coupled subspaces. Then, it calculates the anomaly propagation contribution by combining the coupling strength matrix and global coupling propagation coefficients, identifying root cause optimization parameters. Subsequently, using the global coupling function value as the core, it combines load compliance and parameter coordination function values ​​to output global state judgment conclusions and optimization constraint directions. Finally, it integrates all judgment results to output a standardized judgment instruction set, which is then transmitted to the collaborative optimization and closed-loop iteration module.

[0126] It should be further noted that this module takes a single-dimensional state feature vector as input. Coupling strength matrix C, strongly coupled subspace set, global coupling transfer coefficient Global coupling function Load compliance function , parameter cooperative function Dynamic threshold dataset .

[0127] It should be further explained that the progressive integrated judgment process is as follows:

[0128] Step 1: Initial screening of single-dimensional anomalies: The judgment logic is based on the fluctuation features and threshold deviation in the single-dimensional state feature vector to screen out strong anomaly dimensions. These dimensions are only used as the basic feature input for comprehensive judgment and do not output conclusions independently.

[0129] Judgment formula: ;in, The fluctuation anomaly threshold for the i-th dimension is obtained based on historical statistics of the preprocessed dataset, and the maximum fluctuation threshold for stable operation of the data in this dimension is also given. is the deviation anomaly threshold for the i-th dimension, and is the maximum acceptable threshold deviation for data in this dimension; ∩ is a logical AND operation, indicating that a dimension is considered strongly anomalous only when both fluctuation and deviation exceed the limit, avoiding misjudgment based on a single indicator. Action: Map the selected strongly anomalous dimensions to the coupling strength matrix C, locating their respective strongly coupled subspaces.

[0130] Step 2: Coupling Anomaly Source Tracing: The judgment logic is based on the strong anomaly dimension and its corresponding strongly coupled subspace, calculating the anomaly propagation contribution, quantifying the influence of each parameter on the anomaly, and locating the root cause to optimize parameters. The specific mathematical calculation formula is as follows: ;in: Ω represents the contribution of the j-th parameter to the anomaly propagation, which is the total contribution of this parameter to the system anomaly. The larger the value, the more likely it is to be a core root cause parameter. Ω represents the set of strong anomaly dimensions, which is all the anomaly dimensions selected in the first step.

[0131] Perform action: Press Sort the parameters in descending order and select the top 3-5 parameters as the root cause tuning parameter set to distinguish between core root causes and secondary root causes.

[0132] Step 3: Global State Comprehensive Determination: The determination logic is based on a globally coupled function. With as the core, combined , The system integrates the root cause parameter set and outputs a global production status judgment conclusion. The judgment threshold is optimized based on historical production data.

[0133] Judgment rules:

[0134]

[0135] in: This is the global coupling compliance threshold, which is the maximum global coupling function value for stable operation of the entire system; The global coupling mismatch threshold is the critical global coupling function value that requires urgent optimization of the entire system. This means that production parameter optimization must be within the dynamic threshold range, and optimization must not exceed the compliance boundary of production parameters; , This indicates that optimization must simultaneously meet environmental load compliance and product quality standards, prioritizing environmental and quality constraints. Additional determination of optimization constraint direction: If... If the corresponding limit is exceeded, prioritize constraints on the environmental load dimension and optimize to reduce it. As the core; if If the corresponding limit is exceeded, prioritize the coordination of repair processes and optimize to reduce the impact. With the core as the core.

[0136] Step 4: Closed-loop output of the judgment result:

[0137] The output is a standardized judgment instruction set: D = {Global State Level. Root Cause Tuning Parameter Set. Tuning Magnitude. Tuning Priority. Tuning Constraint Boundaries}. The global state level is divided into three levels: compliance, local deviation, and global mismatch. The root cause tuning parameter set consists of the core and secondary root cause parameters located in the second step. The tuning magnitude is determined based on the contribution of anomaly propagation. The tuning priority is based on... Sort in descending order, prioritizing optimization of core root causes; optimization constraints and boundaries are determined in step three. , , Threshold range.

[0138] Execution action: The judgment instruction set D is transmitted in real time to the collaborative optimization and closed-loop iteration module as the core basis for optimization execution.

[0139] Collaborative optimization and closed-loop iteration module: Receives the standardized judgment instruction set output by the comprehensive judgment module, and combines it with the entire system's coupling feature set and dynamic threshold dataset. First, it initializes the optimization parameter step size based on optimization priority and coupling transfer coefficient. Then, it uses a constrained particle swarm optimization (PSO) algorithm to perform collaborative optimization of multi-process parameters, ensuring that the optimization process meets environmental, quality, and process threshold constraints. Subsequently, it verifies whether the optimized global coupling function value meets the standards. Finally, based on the verification results, it performs closed-loop iterative control until the global production state returns to the compliant range, achieving dynamic adaptive optimization of production parameters.

[0140] It should be further explained that the initial tuning parameters are initialized based on the priority of the root cause tuning parameter set and the contribution of anomaly propagation, combined with the tuning magnitude, to initialize the tuning step size and search interval of each parameter, ensuring that the initial step size matches the degree of anomaly, and the search interval is strictly limited within the dynamic threshold boundary; the specific mathematical function is as follows: ,in, : The initial tuning step size of the j-th root cause tuning parameter, which physically represents the maximum amplitude of a single tuning of the parameter. The step size is positively correlated with the parameter's contribution to the anomaly.

[0141] α: Step size coefficient, with a value of 0.02~0.05 (0.02 for "local deviation" and 0.05 for "global mismatch"). Its physical meaning is the step size scaling factor, which avoids excessive single optimization amplitude that could cause production system oscillation.

[0142] Initialization constraints: The optimization search interval is strictly limited to [ , Simultaneously, the environmental load dimension X∈[ , ]、Quality dimension Z∈[ , To ensure that the initialization phase does not exceed the boundaries of environmental protection, quality, and process compliance. Actions to be performed: Prioritize optimization based on core root cause parameters (…). ≥0.5) Assign a larger initial step size, secondary root cause parameter (0.2≤ <0.5) Allocate a small step size to complete the initialization of all tuning parameters.

[0143] It should be further explained that constrained particle swarm optimization (PSO) is performed to "minimize the global coupling function". With "as the core optimization objective, a constrained particle swarm optimization algorithm is used to perform multi-process parameter collaborative tuning. During the optimization process, the parameters are checked in real time to ensure they meet the threshold constraints, thus avoiding unauthorized tuning. The specific mathematical function of the optimization objective is as follows:

[0144]

[0145] Constraints:

[0146] Production parameter constraints: Root cause optimization parameter set, which physically means that the optimization parameters cannot exceed the process compliance boundary;

[0147] Environmental load constraints: < ( =0.6), which means that the environmental load coupling compliance after optimization must meet environmental protection requirements;

[0148] Quality constraints: < ( =0.3 (Comprehensive Deviation Threshold for Quality Dimensions), which means that the product quality must be maintained within the qualified range after optimization.

[0149] PSO Algorithm Coupling Adaptation Adjustment: To ensure the particle search direction matches the process coupling transfer logic, a coupling transfer coefficient is introduced into the velocity update formula. :

[0150]

[0151] : The velocity of particle i in the (k+1)th generation, which physically represents the search direction and magnitude for parameter tuning;

[0152] ω: Inertial weight, with a value of 0.7-0.9, physically meaning the weight that preserves the search direction of the particle's history, balancing the global search and the local search;

[0153] , : Learning factor, all of which are 2, physically representing the weights by which particles learn from their individual optimal and global optimal solutions;

[0154] , A random number between 0 and 1, which physically introduces randomness into the search and prevents the algorithm from getting trapped in local optima;

[0155] The optimal position of particle i is, in physical terms, the optimal combination of parameters found by the particle.

[0156] The global optimal position, in physical terms, is the optimal combination of parameters found by all particles.

[0157] β: Coupling adaptation coefficient, with a value of 0.1. Its physical meaning is the correction weight of the coupling transfer coefficient on the search direction, so that the search fits the production process logic.

[0158] Actions: Set the number of iterations to 50-100 (50 for local deviations, 100 for global mismatches). After each iteration, check the constraints, discard violating particles, and output the optimal parameter combination that satisfies all constraints. .

[0159] It should be further explained that the optimization effect was verified as follows:

[0160] Decision logic: Combine the optimal parameters Substitute the data into the production system, collect new operational data, and recalculate the global coupling function value. Verify whether the compliance threshold requirements have been met.

[0161] Verification rules:

[0162]

[0163] Physical meaning of the parameters:

[0164] Global coupling compliance threshold (value 0.8). The global coupling mismatch threshold (value 1.5) is consistent with the definition of the comprehensive judgment module.

[0165] Actions to be performed: If the optimization meets the target, record the optimal parameters and synchronize them to the production control system; if it does not meet the target, execute closed-loop iteration; if ineffective, reinitialize and reset the step size coefficient. Increase (maximum not exceeding 0.08).

[0166] It should be further explained that the closed-loop iterative control is as follows:

[0167] Judgment logic: Based on the optimization effect verification results, execute a closed-loop iteration of "optimization-verification-feedback-re-optimization" until the global state returns to the compliance range or reaches the maximum number of iterations (set to 10 times) to avoid infinite loop;

[0168] Iteration rules:

[0169] First iteration failed to meet the target: adjust the step size coefficient. Reduce the search range by 50%, keep the search range unchanged, and re-execute PSO tuning;

[0170] If the target is not met after three consecutive iterations: the coupling strength of environmental load and quality dimensions is adjusted to optimize the objective function, the weight of load / quality constraints is increased, and compliance is prioritized.

[0171] If the maximum number of iterations is reached but the target is still not met: trigger a manual intervention command, output an abnormal alarm, pause automatic optimization, and restart after manual investigation.

[0172] Action: After each iteration, recalculate the entire system coupling feature set U (coupling strength, transfer coefficient) to ensure that the iteration process closely matches the real-time production coupling state until the termination condition is met.

[0173] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0174] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for coordinated control of automotive sealing strip production parameters based on environmental load constraints, characterized in that, include: Multi-dimensional data acquisition and processing module: Collects data from five dimensions: environmental load, equipment operation, core production parameters, raw material characteristics, and product quality prediction; performs preprocessing operations to build a preprocessed dataset; builds a prediction model to output predicted values; and constructs a dynamic threshold system by combining historical data, environmental standards, and production conditions to output the constraint threshold range for each dimension. Multi-dimensional hierarchical coupling analysis module: Based on the preprocessed dataset, it performs single-dimensional time series feature analysis and outputs feature vectors; Construct a coupling adjacency matrix, obtain the coupling strength matrix by mutual information entropy quantization and normalization, and identify strongly coupled subspaces; Extend the coupling matrix, calculate the global coupling transfer coefficients, and output the coupling feature set of the entire system. Strongly Coupled Multi-Objective Function Construction Module: Based on the coupling feature set of the whole system, threshold deviation and dynamic threshold, the environmental load coupling compliance function and the production parameter process coordination function are constructed in sequence. Then, the coupling strength between quality and load and parameters is integrated to construct the global coupling function and output the three types of function quantification values. The comprehensive judgment module: Based on the coupling feature set of the whole system, the quantization values ​​of three types of functions and the constraint threshold range of each dimension, it initially screens strong anomaly dimensions and locates their respective strong coupling subspaces, calculates the anomaly propagation contribution, and locates the root cause optimization parameters; based on the value of the global coupling function, it determines the global production status and optimization constraint direction, and integrates the results to output a standardized judgment instruction set; Collaborative tuning and closed-loop iteration module: Initializes the tuning parameter step size and search interval based on a standardized judgment instruction set, with the optimization objective of minimizing the global coupling function value, and uses a constrained particle swarm optimization algorithm to perform collaborative tuning of multi-process parameters and output the optimal parameter combination; verifies the tuning effect, and performs closed-loop iterative control based on the verification results.

2. The method for coordinated control of automotive sealing strip production parameters based on environmental load constraints according to claim 1, characterized in that: The single-dimensional time-series feature analysis includes: The sliding window standard deviation is calculated for each independent dimension of the standardized time series data in the preprocessed dataset to extract the real-time fluctuation features of the data; then, the deviation of each dimension of the data relative to the corresponding constraint threshold interval is calculated; finally, the data fluctuation features and threshold deviation features are integrated to output the single-dimensional state feature vector corresponding to each dimension.

3. The method for coordinated control of automotive sealing strip production parameters based on environmental load constraints according to claim 1, characterized in that: The coupling strength matrix includes: Based on the physical transfer logic of the production process, a coupling adjacency matrix is ​​constructed to represent the direct process coupling relationship between dimensions; the original mutual information entropy between each dimension is calculated to quantify the degree of nonlinear coupling between dimensions; the original mutual information entropy is normalized to obtain the normalized coupling strength value between each dimension; only the dimension pairs with direct process coupling relationship in the coupling adjacency matrix are assigned the corresponding normalized coupling strength value, and the coupling strength value of the dimension pairs without direct process coupling relationship is set to 0, thereby forming the coupling strength matrix.

4. The method for coordinated control of automotive sealing strip production parameters based on environmental load constraints according to claim 1, characterized in that: The calculation of the global coupling transfer coefficient includes: The auxiliary dimensions of equipment and raw materials are embedded into the coupling adjacency matrix, which is then expanded into a system-wide coupling matrix. Combined with the corresponding coupling strength matrix, for any two dimensions, all data dimensions of the entire system are traversed and the product of the pairwise direct coupling strengths of the two dimensions transmitted through the intermediate dimension is accumulated to obtain the global coupling transmission coefficient between the two dimensions. In this way, the global coupling transmission coefficients of all dimension pairs in the entire system are calculated to form the global coupling transmission coefficient set of the entire system.

5. The method for coordinated control of automotive sealing strip production parameters based on environmental load constraints according to claim 1, characterized in that: The environmental load coupling compliance function includes: Based on the coupling strength between various dimensions of environmental load in the system coupling feature set, and the degree of deviation of each environmental load dimension relative to the corresponding constraint threshold range, the deviation of a single environmental load dimension is multiplied by the sum of the coupling strength of that dimension and all other environmental load dimensions. The above calculation results for all environmental load dimensions are then averaged to obtain the environmental load coupling compliance function.

6. The method for coordinated control of automotive sealing strip production parameters based on environmental load constraints according to claim 1, characterized in that: The production parameter process coordination function includes: Based on the global coupling transfer coefficients between production parameter dimensions in the system coupling feature set, and the degree of deviation of each production parameter relative to its constraint threshold range, the deviation of a single production parameter from the mean of its corresponding threshold is calculated and standardized. The standardized deviation value is multiplied by the sum of the global coupling transfer coefficients of all upstream production parameters received by the production parameter. The above calculation results of all production parameters are summed to obtain the production parameter process coordination function.

7. The method for coordinated control of automotive sealing strip production parameters based on environmental load constraints according to claim 1, characterized in that: The construction of the global coupling function includes: Based on the coupling strength between the product quality dimension and the environmental load and production parameter dimensions in the system-wide coupling feature set, combined with the values ​​of the constructed environmental load coupling compliance function and the production parameter process coordination function, as well as the overall deviation of the product quality dimension from the corresponding constraint threshold range, the above elements are integrated and calculated to construct a global coupling function.

8. The method for coordinated control of automotive sealing strip production parameters based on environmental load constraints according to claim 1, characterized in that: The parameters for calculating the contribution of anomaly propagation to locate root causes and optimize performance include: First, determine the set of anomalous dimensions composed of the strong anomalous dimensions obtained from the initial screening. Then, combine the global coupling transmission coefficient and coupling strength matrix in the coupling feature set of the whole system, calculate the sum of the products of the global coupling transmission coefficient and the corresponding coupling strength of all dimensions in the anomalous dimension set to the parameter for each production parameter, and obtain the anomalous transmission contribution of each production parameter. Sort the production parameters from largest to smallest according to the anomalous transmission contribution, and select a preset number of parameters at the top of the sort as root cause tuning parameters.

9. The method for coordinated control of automotive sealing strip production parameters based on environmental load constraints according to claim 1, characterized in that: The initialization tuning parameter step size and search range include: Based on the priority, anomaly propagation contribution, and tuning magnitude of root cause tuning parameters in the standardized judgment instruction set, and combined with the width of the constraint threshold interval corresponding to each root cause tuning parameter, the initial tuning step size of each root cause tuning parameter is calculated, and the step size coefficient is adjusted according to the global production status. The tuning search interval of each root cause tuning parameter is limited to its corresponding constraint threshold interval, while simultaneously constraining the threshold intervals of environmental load and product quality dimensions. Differentiated initial step sizes are assigned to root cause tuning parameters according to the magnitude of anomaly propagation contribution, thus completing the initialization of the step size and search interval of all tuning parameters.

10. The method for coordinated control of automotive sealing strip production parameters based on environmental load constraints according to claim 1, characterized in that: The process of performing multi-process parameter collaborative optimization and outputting the optimal parameter combination includes: With the goal of minimizing the global coupling function value, a constrained particle swarm optimization algorithm is used to perform multi-process parameter collaborative tuning. A global coupling transmission coefficient is introduced into the velocity update formula of the particle swarm algorithm to adjust the coupling adaptation, so that the particle search direction matches the coupling transmission logic of the production process. The number of algorithm iterations is set according to the global production state adaptation. After each iteration, it is verified in real time whether the production parameters, environmental load, and product quality dimensions meet the corresponding constraint threshold ranges, and particles that do not meet the constraints are discarded. After the iteration is completed, the particle with the smallest global coupling function value is selected from the particles that meet all constraints. The production parameter combination corresponding to this particle is the optimal parameter combination, which is output as the tuning result.

Images