Optimization control method and system for plywood production

Abstract

The present invention discloses an optimization control method and system for plywood production, which relates to the technical field of plywood production, including: obtaining the gluing evaluation parameters of plywood, obtaining quality influencing factors; using an association rule mining algorithm to find the correlation of quality influencing factors, determining the influence correlation and the influence relationship; constructing a stamping cycle chain; configuring the optimization target of each stamping cycle node, fitting the objective function relationship according to the influence correlation and the influence relationship; constructing an optimization model according to the objective function, fitting the optimization model to the stamping cycle chain, optimizing the quality influencing factors of each stamping cycle node, determining the optimization control information of the cycle node, performing full-cycle fitting evaluation on the optimization control information of each cycle node, and obtaining cycle optimization control information. The present invention solves the technical problems of the limitations of the parameter fixed control strategy and the optimization control limitations in the prior art, and achieves the technical effect of full-cycle optimization of the production process.

Description

Technical Field

[0001] The invention relates to the technical field of plywood production, and in particular to an optimization control method and system for plywood production. Background Art

[0002] At present, a certain degree of automation has been achieved in the production of plywood, through the use of sensors, PLC and other automation equipment to monitor and control each link of the production process. These systems can collect parameters in the production process in real time and automatically adjust them according to the preset control logic to maintain the stability and efficiency of production. However, these automation systems are usually based on pre-set fixed parameters and control strategies, lacking the ability to deeply analyze and dynamically optimize the parameter changes and complex correlations in the production process. Summary of the invention

[0003] The present application provides an optimization control method and system for plywood production, which are used to solve the technical problems of the limitations of the parameter fixed control strategy and the optimization control limitations in the prior art.

[0004] In view of the above problems, the present application provides an optimization control method and system for plywood production.

[0005] In a first aspect of the present application, an optimization control method for plywood production is provided, the method comprising:

[0006] The gluing evaluation parameters of plywood are obtained, and based on historical data, the gluing evaluation parameters are used as the target for analysis to obtain quality influencing factors, wherein the quality influencing factors are production control information that has an impact on the gluing evaluation parameters; the correlation of the quality influencing factors is found by using an association rule mining algorithm, and the impact correlation and the impact relationship are determined, wherein the impact relationship includes a constraint parameter relationship and a promotion parameter relationship; the stamping state and stamping value of the plywood are analyzed, and a stamping cycle chain is constructed, wherein the stamping cycle chain includes a stamping cycle node, a node stamping state, and a stamping impact incentive relationship; the optimization target of each stamping cycle node is configured with the gluing evaluation parameters, and the objective function relationship is fitted according to the impact correlation and the impact relationship; an optimization model is constructed according to the objective function, and the optimization model is fitted to the stamping cycle chain, and the quality influencing factors of each stamping cycle node are optimized, and the optimization control information of the cycle node is determined; the optimization control information of each cycle node is evaluated for full-cycle fitting based on the stamping cycle chain, and the cycle optimization control information is obtained.

[0007] In a second aspect of the present application, an optimization control system for plywood production is provided, the system comprising:

[0008] A gluing evaluation parameter acquisition module, wherein the gluing evaluation parameter acquisition module acquires the gluing evaluation parameters of plywood, and based on historical data, takes the gluing evaluation parameters as the target to analyze and obtain quality influencing factors, wherein the quality influencing factors are production control information that has an impact on the gluing evaluation parameters; a correlation determination module, wherein the correlation determination module uses an association rule mining algorithm to find the correlation of the quality influencing factors, and determines the impact correlation degree and the impact relationship, wherein the impact relationship includes a constraint parameter relationship and a promotion parameter relationship; a stamping cycle chain construction module, wherein the stamping cycle chain construction module analyzes the stamping state and stamping value of the plywood, and constructs a stamping cycle chain, wherein It includes stamping cycle nodes, node stamping states, and stamping influence incentive relationships; a relationship fitting module, which configures the optimization target of each stamping cycle node with the bonding evaluation parameters, and fits the objective function relationship according to the influence correlation and the influence relationship; a cycle optimization control information acquisition module, which constructs an optimization model according to the objective function, fits the optimization model to the stamping cycle chain, optimizes the quality influencing factors of each stamping cycle node, determines the optimization control information of the cycle node, and performs full-cycle fitting evaluation on the optimization control information of each cycle node based on the stamping cycle chain to obtain the cycle optimization control information.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] The present application obtains the gluing evaluation parameters of plywood, and based on historical data, obtains the quality influencing factors by taking the gluing evaluation parameters as the target for analysis, wherein the quality influencing factors are the production control information that has an impact on the gluing evaluation parameters; uses the association rule mining algorithm to find the correlation of the quality influencing factors, determines the impact correlation and the impact relationship, wherein the impact relationship includes the constraint parameter relationship and the promotion parameter relationship; analyzes the stamping state and stamping value of the plywood, and constructs a stamping cycle chain, wherein the stamping cycle chain includes stamping cycle nodes, node stamping states, and stamping impact incentive relationships; configures the optimization target of each stamping cycle node by the gluing evaluation parameters, and performs objective function relationship fitting according to the impact correlation and the impact relationship; constructs an optimization model according to the objective function, fits the optimization model to the stamping cycle chain, optimizes the quality influencing factors of each stamping cycle node, determines the optimization control information of the cycle node, performs full-cycle fitting evaluation on the optimization control information of each cycle node based on the stamping cycle chain, and obtains the cycle optimization control information. The present invention solves the technical problems of the limitations of the prior art in the parameter fixed control strategy and the optimization control limitations. It collects production data in real time and conducts correlation analysis of quality influencing factors, dynamically analyzes the stamping status and constructs the stamping cycle chain, generates optimization targets and constructs an optimization model, and finally optimizes and controls each stamping cycle node to achieve the technical effect of full-cycle optimization of the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A schematic flow chart of an optimization control method for plywood production provided in an embodiment of the present application.

[0013] Figure 2 This is a schematic diagram of the structure of an optimized control system for plywood production provided in an embodiment of the present application.

[0014] Explanation of the reference numerals: bonding evaluation parameter acquisition module 11 , correlation determination module 12 , stamping cycle chain construction module 13 , relationship fitting module 14 , cycle optimization control information acquisition module 15 . DETAILED DESCRIPTION

[0015] The present application provides an optimization control method and system for plywood production, which aims to solve the technical problems of the limitations of the parameter fixed control strategy and the optimization control limitations in the prior art. By collecting production data in real time and performing correlation analysis on quality influencing factors, dynamically analyzing the stamping status and building a stamping cycle chain, generating optimization targets and constructing an optimization model, the application finally optimizes and controls each stamping cycle node, thereby achieving the technical effect of full-cycle optimization of the production process.

[0016] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0017] It should be noted that any variations of the terms "include" and "have" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0018] Embodiment 1

[0019] like Figure 1 As shown, the present application provides an optimization control method for plywood production, the method comprising:

[0020] Step S100: Obtaining gluing evaluation parameters of plywood, and obtaining quality influencing factors by analyzing the gluing evaluation parameters as a target based on historical data, wherein the quality influencing factors are production control information that has an impact on the gluing evaluation parameters.

[0021] In the embodiment of the present application, the key parameters in the plywood production process, including temperature, humidity, pressure, time, glue coating uniformity and thickness, are monitored and collected in real time by sensors and data acquisition equipment. At the same time, the plywood is quality tested to obtain gluing evaluation parameters, such as gluing strength, gluing uniformity, glue layer thickness, etc. Parameter identification is generated to record the gluing evaluation results of each batch of plywood. The collected production process parameters and gluing evaluation parameters are recorded to build a historical database.

[0022] Using historical data, the collected parameter labels are clustered for quality. According to the differences in bonding evaluation parameters, a multi-level quality case set is constructed to classify and organize production data of different quality levels. The data in the multi-level quality case set is analyzed to fit the data quantity change relationship of temperature, humidity, pressure, time, glue coating uniformity and thickness characteristics. Through fitting, the influence relationship of parameter quantity change on bonding evaluation parameters, that is, the quality-parameter quantity relationship, is obtained. Quantization coefficients are configured for quality cases of different levels, and the multi-dimensional parameter quantity relationship is reduced in dimension. The quality-parameter quantity relationship is reduced in dimension and fused to simplify the data model and determine the quality influence of each factor. The mapping relationship between bonding evaluation parameters and each factor is established. According to the quality influence of each factor, the production control information that has a significant impact on the bonding evaluation parameters, that is, the quality influencing factors, is determined.

[0023] Furthermore, step S100 in the method provided in the application embodiment also includes:

[0024] Sensors and data acquisition equipment are deployed to collect temperature, humidity, pressure, time, glue uniformity and thickness characteristics in plywood production, and the gluing evaluation parameters of plywood are detected and evaluated to generate parameter identifications and construct the historical data; quality clustering is performed based on the historical data using the parameter labels to construct a multi-level quality case set; the relationship between the change in the temperature, humidity, pressure, time, glue uniformity and thickness characteristic data quantities is fitted based on the multi-level quality case set to obtain the quality-parameter quantity relationship; multi-level quality quantization coefficients are configured, and the quality-parameter quantity relationship is dimensionality reduced and fused based on the quantization coefficients to determine the quality influence of each factor; a mapping relationship between the gluing evaluation parameter and each factor is established, and the quality influence factor of the gluing evaluation parameter is obtained based on the quality influence of each factor.

[0025] In the embodiment of the present application, in order to optimize the production process of plywood, sensors and data acquisition equipment are first deployed. High-precision sensors are installed at key locations of the production line, such as thermocouple sensors to measure temperature, humidity sensors to measure air humidity, pressure sensors to measure pressing pressure, and laser thickness gauges to measure board thickness. Through the Internet of Things technology, these sensors and data acquisition equipment achieve real-time data transmission and monitoring.

[0026] Data from the production process, including temperature, humidity, pressure, time, glue coating uniformity and thickness characteristics, are continuously collected through data acquisition equipment such as PLC and SCADA systems and transmitted to the central database in real time for storage and processing. At the same time, quality inspection is carried out on each batch of plywood to obtain evaluation parameters such as bonding strength, bonding uniformity, and glue layer thickness. Automated quality inspection equipment, such as universal material testing machines, are used to test the mechanical properties of plywood, and parameter identifiers are generated based on the test results. The quality evaluation data of each batch of plywood is associated with the production process data and stored to form a complete historical database, completing the construction of historical data.

[0027] Next, the historical data was clustered and analyzed. The quality data of plywood was divided into three levels: high quality, medium quality and low quality by K-means clustering algorithm. A multi-level quality case set was constructed, and each case set contained all production data of the corresponding quality level. In order to determine the influence of each production parameter on the gluing quality, regression analysis was performed on the data in the multi-level quality case set. A multivariate linear regression model was used to fit the data volume change relationship of temperature, humidity, pressure, time, gluing uniformity and thickness characteristics, and the regression coefficient of each parameter on gluing quality was obtained.

[0028] In order to simplify the analysis process, dimensionality reduction techniques such as principal component analysis are used to reduce the dimensionality of multi-dimensional parameter quantity relationships. Through PCA, multiple variables such as temperature, humidity, and pressure are converted into several main components, thereby simplifying the data model and determining the most important quality influencing factors. After dimensionality reduction, a mapping relationship between bonding evaluation parameters and various production parameters is established. Using techniques such as path analysis, the weight of the influence of each parameter on the bonding quality is calculated, and the key quality influencing factors are determined. For example, through path analysis, the direct and indirect effects of temperature and humidity on bonding strength are determined, and key quality influencing factors such as temperature control accuracy and humidity stability are obtained.

[0029] Step S200: using an association rule mining algorithm to find the correlation of the quality influencing factors, and determine the influence correlation and influence relationship, wherein the influence relationship includes a constraint parameter relationship and a promotion parameter relationship.

[0030] Furthermore, in the method provided in the embodiment of the application, the method of finding the correlation of the quality influencing factors using the association rule mining algorithm, determining the influence correlation and the influence relationship, further includes:

[0031] Based on the historical data, the continuous data in the quality influencing factors are discretized, including discretization processing according to the multi-level configuration of the influencing factors; the discretized data is converted into a transaction data set, each transaction represents a parameter combination of a production cycle; association mining is performed based on the transaction data set to determine frequent item sets and association rules; for the association rules, the support and confidence of the frequent item sets are calculated; key rules are screened according to the support and confidence, and the key rules are association rules whose support and confidence are higher than a preset threshold; the lift is calculated according to the support and confidence, and the association metric is quantitatively calculated according to the support, confidence and lift to obtain the impact correlation of each quality influencing factor, and at the same time, the impact relationship is divided according to the association rules.

[0032] In the embodiment of the present application, in order to find the correlation of quality influencing factors and determine the influence correlation and influence relationship, the continuous data in the quality influencing factors are first discretized based on historical data using equal-width discretization, equal-frequency discretization and other methods, and multiple levels are configured for discretization according to different influencing factors. For example, the temperature data is divided into low temperature (0-20°C), medium temperature (21-30°C) and high temperature (31-40°C); the humidity data is divided into low humidity (0-40%), medium humidity (41-60%) and high humidity (61-100%).

[0033] The discretized data is converted into a transaction data set, and each transaction represents a parameter combination of a production cycle. This transaction representation makes the data of each production cycle an independent transaction record, which is convenient for the application of association rule mining algorithms. Using data preprocessing technology, all discretized data are integrated into a transaction data set, and each transaction contains a combination of all parameters, such as temperature is medium temperature; humidity is high humidity, etc.

[0034] Based on the transaction data set, the association rule mining algorithm is applied for analysis. Common association rule mining algorithms include the Apriori algorithm and the FP-Growth algorithm. Taking the Apriori algorithm as an example, the algorithm generates association rules by iteratively searching for frequent item sets. First, the minimum support threshold is determined, such as 0.1, and the algorithm is applied to mine frequent item sets that meet the support requirements, and association rules are generated from the frequent item sets. For the generated association rules, the support and confidence of the frequent item sets are calculated. The support represents the frequency of the rule appearing in all transactions, and the confidence represents the frequency of the consequent, such as high humidity, appearing when the antecedent, such as low temperature, appears. The calculation method of support and confidence is based on statistical principles and is calculated by traversing the transaction data set.

[0035] Filter key rules based on support and confidence. Key rules refer to association rules whose support and confidence are higher than the preset thresholds. Technical experts set appropriate support and confidence thresholds, such as support greater than 0.05 and confidence greater than 0.7, and only retain association rules that meet these conditions.

[0036] Then, lift is calculated based on support and confidence. Lift reflects the ratio of the frequency of the antecedent and the consequent appearing together to the frequency of their separate occurrences. Lift is obtained by dividing the support of the consequent by the confidence. A rule with lift greater than 1 indicates a positive correlation between the antecedent and the consequent.

[0037] According to the support, confidence and lift, the association rules are quantitatively calculated to determine the influence correlation of each quality influencing factor. The influence correlation indicates the degree of influence of a certain parameter on the gluing quality. By comprehensively considering the support, confidence and lift, the correlation of each quality influencing factor is quantified.

[0038] According to the association rules and the influence correlation, the influence relations are divided into the constraint parameter relations and the promotion parameter relations. The constraint parameter relations are that some parameters have a negative impact on the gluing quality. The promotion parameter relations are that some parameters have a positive impact on the gluing quality. For example, through the association rule mining, it is found that the combination of low temperature and high humidity frequently appears in the plywood with poor quality, which indicates that there is a constraint parameter relation between low temperature and high humidity. On the other hand, the combination of high temperature and moderate humidity frequently appears in the plywood with good quality, which indicates that there is a promotion parameter relation between high temperature and moderate humidity.

[0039] Furthermore, in the method provided in the embodiment of the application, the correlation quantification calculation is performed according to the support, confidence and lift to obtain the influence correlation of each quality influencing factor, and further includes:

[0040] ,in, is the support, which represents the frequency of the rule appearing in item sets A and B. is the confidence level, which represents the frequency with which the rule appears in A and also in B. is the weight coefficient, To improve the degree, , is the frequency of occurrence of phase set B.

[0041] In the embodiment of the present application, the above formula is used to perform quantitative calculation of association according to support, confidence, and lift, where: is the weight coefficient, which is preset.

[0042] By substituting the support, confidence and lift into the above formula, the quantitative calculation of the association is completed, and the association of each association rule is obtained. According to the association, it is determined which parameter combinations have a significant impact on the quality of plywood.

[0043] Step S300: analyzing the stamping state and stamping value of the plywood, and constructing a stamping cycle chain, wherein the stamping cycle chain includes stamping cycle nodes, node stamping states, and stamping influence incentive relationships.

[0044] In the embodiment of the present application, the key parameters of the stamping process are monitored and collected in real time by sensors. These sensors include pressure sensors, displacement sensors, acceleration sensors, temperature sensors, etc. These data are transmitted in real time through data acquisition equipment such as PLC or SCADA system and stored in a central database. The data includes detailed records of each stamping cycle, such as stamping start time, end time, maximum pressure value, stamping depth, etc.

[0045] Next, the collected data is parsed to determine the stamping state and stamping value in each stamping cycle. Through time series analysis and signal processing technology, the specific state stages in each stamping cycle are identified and parsed. For example, each stamping cycle is divided into the initial stage, pressurization stage, maintenance stage and pressure relief stage, and the parameter values ​​corresponding to each stage are recorded in the database respectively. On the basis of parsing the stamping state and stamping value, a stamping cycle chain is constructed. The stamping cycle chain is a representation of the time and state serialization of the stamping process, specifically including stamping cycle nodes, node stamping states, and stamping influence incentive relationships. Each stamping cycle node contains information such as start time, end time, maximum pressure, minimum pressure, and stamping depth. The stamping state in the node includes the initial stage, pressurization stage, maintenance stage, and pressure relief stage, and each stage records the corresponding time period and pressure changes. For example, the pressurization stage of node 1 is from 10:00:01 to 10:00:05, and the pressure changes from 1000N to 5000N.

[0046] Through further analysis of these data, the stamping influence excitation relationship is identified and described, that is, the influence of the state and parameters of one node on the subsequent nodes. For example, the maximum pressure of node 1 has a direct influence on the initial pressure of node 2, and the temperature of node 1 has an indirect influence on the material properties of node 2. This analysis is achieved through data analysis techniques such as path analysis and regression analysis, so as to accurately describe the relationship between nodes.

[0047] Step S400: configuring the optimization target of each stamping cycle node with the bonding evaluation parameter, and fitting the objective function relationship according to the influence correlation and influence relationship.

[0048] Furthermore, step S400 in the method provided in the application embodiment also includes:

[0049] According to the bonding evaluation parameters, the influence relationship of each stamping cycle node is analyzed, and the optimization target of each stamping cycle node is configured. The optimization target includes one or more. When there are multiple optimization targets, the weight of each target corresponds to the influence relationship; according to the influence relationship, the negative correlation influence parameters are screened, and the penalty items are added to fit the negative correlation influence degree; according to the influence relationship, the positive correlation influence parameters are screened, and the reward items are added to fit the positive correlation influence degree; the positive correlation objective function and the negative correlation objective function are respectively fitted, and the positive correlation objective function and the negative correlation objective function are merged to obtain the objective function.

[0050] In the embodiment of the present application, the influence relationship of each stamping cycle node is first analyzed based on the existing bonding evaluation parameters. This step uses data analysis methods such as association rule mining and regression analysis to identify which production control parameters, such as temperature, humidity, pressure and time, have a significant impact on bonding evaluation parameters, such as bonding strength, bonding uniformity and glue layer thickness.

[0051] Next, configure the optimization objectives for each stamping cycle node. These optimization objectives can be single objectives, such as maximizing the bonding strength, or a combination of multiple objectives, such as maximizing the bonding strength and bonding uniformity at the same time, and minimizing the fluctuation of the glue layer thickness. When configuring multiple optimization objectives, use multi-objective optimization methods, such as the linear weighted sum method, to assign weights to each objective, ensuring that the weight of each objective corresponds to its importance in the influence relationship.

[0052] According to the results of the influence relationship analysis, the parameters that have a negative impact on the bonding evaluation parameters are screened out. These parameters are called negative correlation influence parameters. These negative correlation influences are fitted by introducing penalty terms to reduce their negative impact. The penalty term is fitted using a regularization method, such as L2 regularization. Similarly, the parameters that have a positive impact on the bonding evaluation parameters are screened out, namely positive correlation influence parameters. These positive correlation influences are fitted by introducing reward terms to enhance their positive impact. The reward term is fitted using the lift analysis method.

[0053] After fitting the positive correlation objective function and the negative correlation objective function respectively, they are fused to construct a comprehensive objective function.

[0054] Furthermore, in the method provided in the embodiment of the application, respectively fitting a positive correlation objective function and a negative correlation objective function, fusing the positive correlation objective function and the negative correlation objective function to obtain the objective function, further includes:

[0055] The expression of the positive correlation objective function is: ;

[0056] The expression of the negative correlation objective function is: ;

[0057] The positive correlation objective function and the negative correlation objective function are integrated to obtain the expression of the objective function: , where x is the vector of quality influencing factors, R is the set of parameter pairs with mutually restrictive relationships, C is the set of parameter pairs with mutually promoting relationships, is any parameter pair in the parameter pair set, is the mutual influence of the parameter pair, is the penalty coefficient, is the reward coefficient.

[0058] In the embodiment of the present application, association rule mining technology is used to identify the mutual constraints and mutual promotion relationships between quality influencing factors, and the mined association rules are analyzed to determine which quality influencing factors have mutual constraints and which have mutual promotion relationships. Mutually restrictive relationships mean that when one factor changes, the quality evaluation of another factor may decrease; mutually promoting relationships mean that when one factor changes, the quality evaluation of another factor may increase.

[0059] For mutually restrictive relationships, the penalty coefficient is determined based on the strength of the association rules, such as the confidence level. The higher the confidence level, the stronger the restrictive relationship, and the larger the penalty coefficient. For mutually promoting relationships, the reward coefficient is also determined based on the strength of the association rules. The higher the confidence level, the stronger the promoting relationship, and the larger the reward coefficient. For example, when the confidence level is 0.8, the baseline penalty coefficient is multiplied by 0.8 to determine the final penalty coefficient. When the confidence level is 0.7, the baseline reward coefficient is multiplied by 0.7 to determine the final reward coefficient. The baseline reward coefficient and the baseline penalty coefficient are pre-set.

[0060] The positive correlation objective function and the negative correlation objective function are combined to form a comprehensive objective function. .

[0061] Step S500: construct an optimization model according to the objective function, fit the optimization model to the stamping cycle chain, optimize the quality influencing factors of each stamping cycle node, determine the optimization control information of the cycle node, perform full-cycle fitting evaluation on the optimization control information of each cycle node based on the stamping cycle chain, and obtain the cycle optimization control information.

[0062] In an embodiment of the present application, an optimization model is constructed according to the expression of the comprehensive objective function. The constructed optimization model is fitted into the stamping cycle chain. The stamping cycle chain consists of a series of cycle nodes, and each node contains multiple influencing factors, such as temperature, pressure, humidity, time, etc. By applying the optimization model to each node, the optimization control information of each cycle node is determined. For each cycle node, an optimization algorithm, such as gradient descent method, genetic algorithm, particle swarm optimization algorithm, is used to solve the optimization model and determine the optimal control parameters. These parameters are adjusted in real time to optimize the production process.

[0063] Based on the stamping cycle chain, the optimization control information of each cycle node is evaluated through full-cycle fitting. Time series analysis and prediction models, such as ARIMA model and LSTM neural network, are used to evaluate the entire production cycle and obtain cycle optimization control information to ensure that the parameter adjustment of each node is optimal throughout the production process, thereby improving the overall production quality.

[0064] Furthermore, step S500 in the method provided in the application embodiment also includes:

[0065] According to historical data, periodic nodes are classified and labeled, node features in different states are identified, and a time series training data set is constructed; the time series training data set is used to train a time series model, and the control information of each periodic node is evaluated for full-cycle fitting to obtain a periodic prediction model, wherein the time series model uses a hidden Markov model as a model architecture; the trained periodic prediction model outputs a periodic prediction result, and the optimized control information of each node is evaluated according to the periodic prediction result, and the node is corrected according to the evaluation result to meet the periodic control requirements.

[0066] In the embodiment of the present application, a data analysis tool is used to extract features from the data of each periodic node to identify node features under different states. These features include maximum pressure, minimum pressure, average pressure, temperature range, etc. Then, a clustering algorithm, such as a K-means clustering algorithm, is used to classify the extracted features, classify nodes with similar features into one category, and label them. For example, the nodes are classified as "high pressure-high temperature", "low pressure-low temperature", etc. According to the classification and labeling results, a time series training data set is constructed. Each data set contains feature data of multiple continuous periodic nodes to form a time series. These time series data sets provide a basis for training hidden Markov models. Hidden Markov models are selected as time series model architectures. HMM is suitable for processing time series data, and can model the state transitions between nodes and the dependencies between observations. HMM is trained using a time series training data set, and the training process includes estimating the initial state probability, state transition probability, and observation probability of HMM. The commonly used training algorithm is the Baum-Welch algorithm, which maximizes the likelihood function of the model through iterative calculation. The trained HMM is used to perform full-cycle fitting evaluation on the control information of each periodic node. HMM predicts the most likely state sequence and observation value sequence of each node, thereby evaluating the performance of control information throughout the production cycle. The trained HMM outputs the cycle prediction results, including the most likely state and control parameters of each cycle node.

[0067] Based on the cycle prediction results of HMM, the optimization control information of each node is evaluated. The control parameters of each node are evaluated to see whether they meet the production requirements, and potential optimization space is identified. Based on the evaluation results, node corrections are made. Feedback control methods are used to adjust the parameters of nodes that do not meet the cycle control requirements to achieve overall production optimization.

[0068] In an actual plywood production process, it is assumed that the main gluing evaluation parameters are gluing strength and gluing uniformity. Through feature extraction and classification annotation of historical data, two main node types, high pressure-high temperature and low pressure-low temperature, are identified. After constructing the time series training data set, HMM is selected as the time series model architecture, and the Baum-Welch algorithm is used for model training. The trained HMM predicts the most likely state sequence and observation value sequence of each node, thereby performing a full-cycle fitting evaluation on the control information of each node. For example, when the HMM predicts that the pressure and temperature changes of a certain node do not meet expectations, corrections are made by adjusting the corresponding control parameters, such as lowering the temperature or increasing the pressure, to meet the cycle control requirements.

[0069] In the embodiments of the present application, in summary, the embodiments of the present application have at least the following technical effects:

[0070] The present application obtains the gluing evaluation parameters of plywood, and based on historical data, obtains the quality influencing factors by taking the gluing evaluation parameters as the target for analysis, wherein the quality influencing factors are the production control information that has an impact on the gluing evaluation parameters; uses the association rule mining algorithm to find the correlation of the quality influencing factors, determines the impact correlation and the impact relationship, wherein the impact relationship includes the constraint parameter relationship and the promotion parameter relationship; analyzes the stamping state and stamping value of the plywood, and constructs a stamping cycle chain, wherein the stamping cycle chain includes stamping cycle nodes, node stamping states, and stamping impact incentive relationships; configures the optimization target of each stamping cycle node by the gluing evaluation parameters, and performs objective function relationship fitting according to the impact correlation and the impact relationship; constructs an optimization model according to the objective function, fits the optimization model to the stamping cycle chain, optimizes the quality influencing factors of each stamping cycle node, determines the optimization control information of the cycle node, performs full-cycle fitting evaluation on the optimization control information of each cycle node based on the stamping cycle chain, and obtains the cycle optimization control information. The present invention solves the technical problems of the limitations of the prior art in the parameter fixed control strategy and the optimization control limitations. It collects production data in real time and conducts correlation analysis of quality influencing factors, dynamically analyzes the stamping status and constructs the stamping cycle chain, generates optimization targets and constructs an optimization model, and finally optimizes and controls each stamping cycle node to achieve the technical effect of full-cycle optimization of the production process.

[0071] Embodiment 2

[0072] Based on the same inventive concept as the optimization control method for plywood production in the above-mentioned embodiment, Figure 2 As shown, the present application provides an optimization control system for plywood production, and the system and method embodiments in the present application are based on the same inventive concept. The system includes:

[0073] A gluing evaluation parameter acquisition module 11, wherein the gluing evaluation parameter acquisition module 11 acquires the gluing evaluation parameters of the plywood, and obtains quality influencing factors by analyzing the gluing evaluation parameters as a target based on historical data, wherein the quality influencing factors are production control information that has an impact on the gluing evaluation parameters;

[0074] A correlation determination module 12, wherein the correlation determination module 12 uses an association rule mining algorithm to find the correlation of the quality influencing factors, and determines the influence correlation and the influence relationship, wherein the influence relationship includes a constraint parameter relationship and a promotion parameter relationship;

[0075] A stamping cycle chain building module 13, wherein the stamping cycle chain building module 13 analyzes the stamping state and stamping value of the plywood and builds a stamping cycle chain, wherein the stamping cycle chain includes stamping cycle nodes, node stamping states, and stamping influence incentive relationships;

[0076] A relationship fitting module 14, wherein the relationship fitting module 14 configures the optimization target of each stamping cycle node with the bonding evaluation parameter, and performs objective function relationship fitting according to the influence correlation and the influence relationship;

[0077] The cycle optimization control information acquisition module 15 constructs an optimization model according to the objective function, fits the optimization model to the stamping cycle chain, optimizes the quality influencing factors of each stamping cycle node, determines the optimization control information of the cycle node, performs full-cycle fitting evaluation on the optimization control information of each cycle node based on the stamping cycle chain, and obtains the cycle optimization control information.

[0078] Furthermore, the system is also used to implement the following functions:

[0079] Deploy sensors and data acquisition equipment to collect temperature, humidity, pressure, time, gluing uniformity and thickness characteristics in plywood production, and detect and evaluate the gluing evaluation parameters of plywood, generate parameter identifiers, and construct the historical data;

[0080] According to the historical data, quality clustering is performed with the parameter labels to construct a multi-level quality case set;

[0081] Fitting the relationship between the temperature, humidity, pressure, time, glue coating uniformity and thickness characteristic data quantity changes according to the multi-level quality case set to obtain the quality-parameter quantity relationship;

[0082] Configuring quantization coefficients of multiple levels of quality, performing dimensionality reduction fusion on the quality-parameter quantity relationship according to the quantization coefficients, and determining the quality influence of each factor;

[0083] A mapping relationship between the bonding evaluation parameter and each factor is established, and the quality impact factor of the bonding evaluation parameter is obtained according to the quality impact of each factor.

[0084] Furthermore, the system is also used to implement the following functions:

[0085] Discretizing the continuous data in the quality impact factor based on the historical data, including configuring multiple levels for discretization according to the impact factor;

[0086] Convert the discretized data into a transaction data set, where each transaction represents a parameter combination of a production cycle;

[0087] Perform association mining based on the transaction data set to determine frequent item sets and association rules;

[0088] For the association rule, calculating the support and confidence of the frequent item set;

[0089] Filtering key rules according to the support and confidence, wherein the key rules are association rules whose support and confidence are higher than a preset threshold;

[0090] The lifting degree is calculated according to the support and confidence, and the correlation degree is quantitatively calculated according to the support, confidence and lifting degree to obtain the influence correlation degree of each quality influencing factor, and the influence relationship is divided according to the association rule.

[0091] Furthermore, the system is also used to implement the following functions:

[0092] ,in, is the support, which represents the frequency of the rule appearing in item sets A and B. is the confidence level, which represents the frequency with which the rule appears in A and also in B. is the weight coefficient, To improve the degree, , is the frequency of occurrence of phase set B.

[0093] Furthermore, the system is also used to implement the following functions:

[0094] Performing an influence relationship analysis of each stamping cycle node according to the bonding evaluation parameters, and configuring an optimization target for each stamping cycle node, wherein the optimization target includes one or more optimization targets, and when there are multiple optimization targets, the weight of each target corresponds to the influence relationship;

[0095] According to the influence relationship, negative correlation influence parameters are screened, and a penalty item is added to fit the negative correlation influence degree;

[0096] According to the influence relationship, positive correlation influence parameters are screened, and reward items are added to fit the positive correlation influence degree;

[0097] The positive correlation objective function and the negative correlation objective function are fitted respectively, and the positive correlation objective function and the negative correlation objective function are fused to obtain the objective function.

[0098] Furthermore, the system is also used to implement the following functions:

[0099] The expression of the positive correlation objective function is: ;

[0100] The expression of the negative correlation objective function is: ;

[0101] The positive correlation objective function and the negative correlation objective function are integrated to obtain the expression of the objective function: , where x is the vector of quality influencing factors, R is the set of parameter pairs with mutually restrictive relationships, C is the set of parameter pairs with mutually promoting relationships, is any parameter pair in the parameter pair set, is the mutual influence of the parameter pair, is the penalty coefficient, is the reward coefficient.

[0102] Furthermore, the system is also used to implement the following functions:

[0103] Classify and label periodic nodes based on historical data, identify node features in different states, and build a time series training data set;

[0104] The time series training data set is used to train the time series model, and the control information of each periodic node is subjected to full-cycle fitting evaluation to obtain a periodic prediction model, wherein the time series model uses a hidden Markov model as a model architecture;

[0105] The trained cycle prediction model outputs the cycle prediction result, the optimization control information of each node is evaluated according to the cycle prediction result, and the node is corrected according to the evaluation result to meet the cycle control requirement.

[0106] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0107] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0108] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. An optimization control method for plywood production, characterized in that: The optimization control method for plywood production comprises: Obtaining a gluing evaluation parameter of the plywood, and obtaining a quality influencing factor by analyzing the gluing evaluation parameter as a target based on historical data, wherein the quality influencing factor is production control information that has an impact on the gluing evaluation parameter; Using an association rule mining algorithm to find the correlation of the quality influencing factors, determine the influence correlation and the influence relationship, wherein the influence relationship includes a constraint parameter relationship and a promotion parameter relationship; Analyze the stamping state and stamping value of plywood, and construct a stamping cycle chain, which includes stamping cycle nodes, node stamping states, and stamping influence incentive relationships. Specifically, on the basis of analyzing the stamping state and stamping value, construct a stamping cycle chain, which is a representation of the stamping process in terms of time and state serialization, including stamping cycle nodes, node stamping states, and stamping influence incentive relationships. The incentive relationship refers to the influence of the state and parameters of a node on subsequent nodes. The optimization target of each stamping cycle node is configured with the bonding evaluation parameter, and the objective function relationship is fitted according to the influence correlation and influence relationship; Constructing an optimization model according to the objective function, fitting the optimization model to the stamping cycle chain, optimizing the quality influencing factors of each stamping cycle node, determining the optimization control information of the cycle node, performing full-cycle fitting evaluation on the optimization control information of each cycle node based on the stamping cycle chain, and obtaining cycle optimization control information; Wherein, the quality influencing factors are obtained by analyzing the bonding evaluation parameters based on historical data, including: Deploy sensors and data acquisition equipment to collect temperature, humidity, pressure, time, gluing uniformity and thickness characteristics in plywood production, and detect and evaluate the gluing evaluation parameters of plywood, generate parameter identifiers, and construct the historical data; According to the historical data, quality clustering is performed with the parameter labels to construct a multi-level quality case set; Fitting the relationship between the temperature, humidity, pressure, time, glue coating uniformity and thickness characteristic data quantity changes according to the multi-level quality case set to obtain the quality-parameter quantity relationship; Configuring quantization coefficients of multiple levels of quality, performing dimensionality reduction fusion on the quality-parameter quantity relationship according to the quantization coefficients, and determining the quality influence of each factor; Establishing a mapping relationship between the bonding evaluation parameter and each factor, and obtaining a quality impact factor of the bonding evaluation parameter according to the quality impact of each factor; Wherein, based on the stamping cycle chain, the optimization control information of each cycle node is subjected to full cycle fitting evaluation to obtain cycle optimization control information, including: Classify and label periodic nodes based on historical data, identify node features in different states, and build a time series training data set; The time series training data set is used to train the time series model, and the control information of each periodic node is subjected to full-cycle fitting evaluation to obtain a periodic prediction model, wherein the time series model uses a hidden Markov model as a model architecture; The trained cycle prediction model outputs the cycle prediction result, the optimization control information of each node is evaluated according to the cycle prediction result, and the node is corrected according to the evaluation result to meet the cycle control requirement.

2. The optimization control method for plywood production according to claim 1, characterized in that: The method of finding the correlation of the quality influencing factors by using an association rule mining algorithm and determining the influence correlation and influence relationship includes: Discretizing the continuous data in the quality impact factor based on the historical data, including configuring multiple levels for discretization according to the impact factor; Convert the discretized data into a transaction data set, where each transaction represents a parameter combination of a production cycle; Perform association mining based on the transaction data set to determine frequent item sets and association rules; For the association rule, calculating the support and confidence of the frequent item set; Filtering key rules according to the support and confidence, wherein the key rules are association rules whose support and confidence are higher than a preset threshold; The lifting degree is calculated according to the support and confidence, and the correlation degree is quantitatively calculated according to the support, confidence and lifting degree to obtain the influence correlation degree of each quality influencing factor, and the influence relationship is divided according to the association rule.

3. The optimization control method for plywood production according to claim 2, characterized in that: According to the support, confidence, and lift, the correlation is quantitatively calculated to obtain the impact correlation of each quality impact factor, including: ,in, is the support, which represents the frequency of the rule appearing in item sets A and B. is the confidence level, which represents the frequency with which the rule appears in A and also in B. is the weight coefficient, To improve the degree, , is the frequency of occurrence of phase set B.

4. The optimization control method for plywood production according to claim 1, characterized in that: The optimization target of each stamping cycle node is configured with the bonding evaluation parameter, and the objective function relationship is fitted according to the influence correlation and influence relationship, including: Performing an influence relationship analysis of each stamping cycle node according to the bonding evaluation parameters, and configuring an optimization target for each stamping cycle node, wherein the optimization target includes one or more optimization targets, and when there are multiple optimization targets, the weight of each target corresponds to the influence relationship; According to the influence relationship, negative correlation influence parameters are screened, and a penalty item is added to fit the negative correlation influence degree; According to the influence relationship, positive correlation influence parameters are screened, and reward items are added to fit the positive correlation influence degree; The positive correlation objective function and the negative correlation objective function are fitted respectively, and the positive correlation objective function and the negative correlation objective function are fused to obtain the objective function.

5. The optimization control method for plywood production according to claim 4, characterized in that: The respectively fitting of the positive correlation objective function and the negative correlation objective function, and fusing the positive correlation objective function and the negative correlation objective function to obtain the objective function, comprises: The expression of the positive correlation objective function is: ; The expression of the negative correlation objective function is: ; The positive correlation objective function and the negative correlation objective function are integrated to obtain the expression of the objective function: , where x is the vector of quality influencing factors, R is the set of parameter pairs with mutually restrictive relationships, C is the set of parameter pairs with mutually promoting relationships, is any parameter pair in the parameter pair set, is the mutual influence of the parameter pair, is the penalty coefficient, is the reward coefficient.

6. An optimized control system for plywood production, characterized in that: For executing the method according to any one of claims 1 to 5, the system comprises: A gluing evaluation parameter acquisition module, wherein the gluing evaluation parameter acquisition module acquires the gluing evaluation parameters of the plywood, and obtains quality influencing factors by analyzing the gluing evaluation parameters as a target based on historical data, wherein the quality influencing factors are production control information that has an impact on the gluing evaluation parameters; A correlation determination module, wherein the correlation determination module uses an association rule mining algorithm to find the correlation of the quality influencing factors, and determines the influence correlation degree and the influence relationship, wherein the influence relationship includes a constraint parameter relationship and a promotion parameter relationship; A stamping cycle chain construction module, wherein the stamping cycle chain construction module analyzes the stamping state and stamping value of the plywood and constructs a stamping cycle chain, wherein the stamping cycle chain includes stamping cycle nodes, node stamping states, and stamping influence incentive relationships. Specifically, on the basis of analyzing the stamping state and stamping value, a stamping cycle chain is constructed, wherein the stamping cycle chain is a representation of the stamping process in terms of time and state serialization, and includes stamping cycle nodes, node stamping states, and stamping influence incentive relationships, wherein the incentive relationship refers to the influence of the state and parameters of a node on subsequent nodes; A relationship fitting module, wherein the relationship fitting module configures the optimization target of each stamping cycle node with the bonding evaluation parameter, and performs objective function relationship fitting according to the influence correlation and the influence relationship; A cycle optimization control information acquisition module, wherein the cycle optimization control information acquisition module constructs an optimization model according to the objective function, fits the optimization model to the stamping cycle chain, optimizes the quality influencing factors of each stamping cycle node, determines the optimization control information of the cycle node, performs full-cycle fitting evaluation on the optimization control information of each cycle node based on the stamping cycle chain, and obtains the cycle optimization control information.

Images