Data analysis-based pull-together production energy efficiency optimization method

Abstract

This invention discloses a data analysis-based method for optimizing energy efficiency in bonded lamination production, belonging to the field of production optimization technology. The specific steps are as follows: S1, real-time acquisition of equipment operation data, energy consumption data, process parameters, and quality data from the bonded lamination production line; after data preprocessing, a standardized dataset is constructed; S2, an energy efficiency KPI system is established based on the standardized dataset. A grey relational analysis and fuzzy clustering coupled algorithm is used. Grey relational analysis is used to screen out core parameters strongly correlated with energy efficiency, and fuzzy clustering is used to refine the classification of inefficient operating conditions. Based on the process energy efficiency entropy value evaluation method, inefficient operating conditions, key influencing parameters, and energy efficiency bottleneck processes are identified. This invention utilizes a grey relational analysis and fuzzy clustering coupled algorithm, combined with the process energy efficiency entropy value evaluation method, to accurately identify core influencing parameters, inefficient operating conditions, and energy efficiency bottleneck processes; it aligns with the trend of green transformation in the manufacturing industry and can reduce enterprise energy cost expenditures.

Description

Technical Field

[0001] This invention relates to the field of production optimization technology, specifically to a data analysis-based method for optimizing the energy efficiency of lamination production. Background Technology

[0002] Wound closure tape, also known as wound closure tape or closure adhesive tape, is a medical device used for wound care. Its core function is to close and fix the wound, reduce skin tension on both sides of the wound, aid wound healing, and prevent scar hyperplasia and widening. It can replace or supplement traditional suture methods. Current technology for wound closure tape production relies on the coordinated operation of multiple specialized production equipment, and the production process requires strict control of process parameters such as temperature, pressure, rotation speed, and application time to ensure core quality indicators such as adhesion strength and flatness. With the transformation of manufacturing towards green and intelligent manufacturing, and the continuous fluctuation of energy prices, wound closure tape manufacturers face the dual pressures of improving energy efficiency, reducing energy consumption, and maintaining quality stability. Maximizing production energy efficiency and reducing unit product energy consumption while ensuring product quality and production capacity has become crucial for companies to enhance their core competitiveness.

[0003] Currently, energy efficiency management in the laminate manufacturing industry remains largely rudimentary. Existing energy efficiency control methods rely heavily on operator experience for manual adjustments, lacking scientific and systematic data analysis and intelligent optimization techniques. Specifically, this results in the following technical deficiencies: Laminate manufacturing energy efficiency is influenced by a combination of factors, including equipment operating status, process parameters, and material characteristics. The relationships between these factors and energy efficiency are complex and exhibit non-linear characteristics. Current technologies lack effective analytical methods to screen core parameters strongly correlated with energy efficiency, making it difficult to categorize inefficient operating conditions in the production process and accurately identify energy efficiency bottlenecks. This leads to a lack of targeted energy efficiency optimization, often resulting in only general, extensive adjustments with limited effectiveness, failing to achieve precise optimization at the process or parameter levels. To address this, we propose a data analysis-based energy efficiency optimization method for laminate manufacturing. Summary of the Invention

[0004] To address the aforementioned technical issues, this technical solution provides a data-driven method for optimizing the energy efficiency of lamination and bonding production. This solution resolves the problem of the inability to achieve precise optimization at the process and parameter levels.

[0005] To achieve the above objectives, the technical solution adopted in this invention is: a data analysis-based method for optimizing the energy efficiency of lamination and bonding production, the specific steps of which are as follows: S1. Real-time acquisition of equipment operation data, energy consumption data, process parameters and quality data of the bonding production line, and construction of standardized datasets after data preprocessing; S2. Establish an energy efficiency KPI system based on a standardized dataset. Use a grey relational analysis and fuzzy clustering coupled algorithm to screen out core parameters that are strongly related to energy efficiency through grey relational analysis. Use fuzzy clustering to refine the classification of inefficient working conditions. Based on the process energy efficiency entropy value evaluation method, identify inefficient working conditions, key influencing parameters and energy efficiency bottleneck processes. S3. Construct a correlation model of parameters, energy consumption, capacity and quality based on bidirectional LSTM and attention mechanism, and output the optimal combination of process parameters that meets the constraints of quality and capacity. S4. Send the optimal parameters to the production line control system for automatic execution and collect the execution results in real time for verification. Based on the energy efficiency deviation traceability model, when there is a deviation between the actual energy efficiency and the optimal energy efficiency, quickly locate the source of the deviation, dynamically fine-tune the parameters and perform secondary verification, and finally determine the optimal energy efficiency solution for the lamination production, thereby optimizing the energy efficiency of lamination production.

[0006] Preferably, in step S1, the acquisition of equipment operation data, energy consumption data, process parameters, and quality data is achieved by deploying corresponding data acquisition devices at key nodes of the production line. These devices include sensors, smart meters, and machine vision inspection equipment. The acquisition frequency is set. The key nodes include the feeding end, bonding section, drying section, inspection section, and discharge end. The data acquisition devices are activated to capture various types of data in real time and transmit the raw data via industrial Ethernet and wireless transmission. Preprocessing includes data cleaning, data standardization, and data fusion.

[0007] Preferably, the energy efficiency KPI system in step S2 is obtained through industry standards, equipment characteristics and production processes; the KPI system includes unit product energy consumption, comprehensive energy efficiency ratio, process energy efficiency coefficient and equipment load rate indicators. The core parameters strongly correlated with energy efficiency were identified through grey relational analysis as follows: The energy efficiency KPI system is used as a reference sequence, and the operating data of various equipment are used as a comparison sequence. Calculate the grey relational coefficient and grey relational degree between the operating data of various equipment and the energy efficiency KPI system; Based on the degree of correlation, a threshold is set to filter out the parameters that are strongly related to energy efficiency and remove the parameters that are weakly related or irrelevant; thus obtaining a subset of core parameters.

[0008] Preferably, the reference sequence for the energy efficiency KPI system is set as follows: , No. The operating data of each device is ;in The number of samples; The number of equipment operating parameters; The formula for calculating the grey relational coefficient is: ; in For the first The parameter in the first... Grey relational coefficient of each sample point ; For the first Absolute difference; It is the minimum difference between two levels; The maximum difference between the two levels; The resolution coefficient; The grey relational degree is obtained by taking the arithmetic mean of the correlation coefficients. The calculation formula is: ; in The larger the value, the better the parameter. The stronger the correlation with energy efficiency KPIs; according to Sort by size from largest to smallest, and set a threshold. ; like If , then it is a strongly correlated parameter; if If the parameter is weakly correlated or irrelevant, it should be removed.

[0009] Preferably, the detailed classification of inefficient operating conditions using fuzzy clustering is as follows: A feature matrix is ​​constructed from a subset of core parameters for historical operating condition samples; Establish a fuzzy similarity matrix and perform fuzzy clustering based on the fuzzy equivalence matrix; Combined threshold The operating conditions are divided into high-efficiency, medium-efficiency and low-efficiency, and the low-efficiency operating condition cluster is further subdivided into subcategories. Multiple typical inefficient operating conditions were obtained, and the parameter characteristics and operating status of each type of operating condition were clarified. Based on the process energy efficiency entropy value evaluation method, the inefficient operating conditions, key influencing parameters, and energy efficiency bottleneck processes are identified as follows: Calculate the energy efficiency entropy value for each production process. The smaller the entropy value, the more concentrated the energy efficiency fluctuation and the higher the energy efficiency level; the larger the entropy value, the more unstable the energy efficiency and the higher the risk of inefficiency. By combining core parameters with operating condition clustering results, the distribution characteristics of key influencing parameters in various inefficient operating conditions are located. By combining the magnitude of entropy, the proportion of energy consumption in each process, and the clustering categories of operating conditions, the energy efficiency bottleneck processes and typical inefficient operating modes are identified; the inefficient operating condition types, key influencing parameters, and energy efficiency bottleneck processes are output.

[0010] Preferably, the operating condition samples are selected from historical operating condition samples that have been running continuously and stably, and samples from start-up and shutdown phases, fault shutdowns, and abnormal fluctuations and unsteady states are removed, with the number of samples n≥30; the parameters in the feature matrix are normalized.

[0011] Preferably, step S3 is as follows: A bidirectional LSTM coupled model with an attention mechanism is constructed, and the optimal combination of process parameters that satisfies quality and capacity constraints is solved based on the coupled model. Core process parameters and energy consumption, capacity, and quality data are screened, preprocessed, divided into datasets, and reshaped into time-series format. A bidirectional LSTM is built to capture temporal features, key features are enhanced based on the attention mechanism, and a correlation model is constructed by combining a fully connected layer. The prediction accuracy is ensured through training, validation and optimization. Quantifiable constraints on quality and production capacity are set, with the goal of minimizing energy consumption. The optimal combination of process parameters is solved and verified using the particle swarm optimization algorithm. Regularly supplement the model with new data to ensure that the parameter set is appropriate for changes in operating conditions.

[0012] Preferably, the constraints include quality constraints and production capacity constraints. The quality constraints are based on product quality standards, setting the constraint range of core quality indicators and using inequality constraints. When there are multiple quality indicators, all quality constraints are satisfied simultaneously. Capacity constraints are set based on the company's production plan and the rated capacity of the equipment, defining the range of capacity constraints. To find the optimal combination of process parameters, the correlation model is used as the fitness function and input into the particle swarm optimization algorithm. The particles represent different combinations of process parameters. By iteratively updating the particle positions, the particle that satisfies all constraints and minimizes energy consumption is selected as the optimal combination of process parameters.

[0013] Preferably, in step S4, the optimal parameters are sent in batches to the control systems of each station on the bonding production line via an industrial communication protocol. The completeness and rationality of the parameters are automatically verified, and once verified, the instructions are automatically executed, and the production line starts running according to the optimal parameters. Real-time data collection and verification of execution results: By collecting data in real time during the production of the adhesive bonding process, the collected data is compared with the predicted data corresponding to the optimal parameters to verify the execution effect. If the deviation between the collected data and the predicted data is within the allowable range and all constraints are met, the execution is deemed qualified and the production line continues to operate with optimal parameters. If there is a deviation or the constraints are not met, a deviation warning is triggered and the source of the deviation is traced.

[0014] Preferably, in step S4, the source of the positioning deviation is combined with the characteristics of the bonding production process, and an allowable threshold for energy efficiency deviation is set. When the deviation between the actual energy efficiency and the optimal energy efficiency exceeds the threshold, the energy efficiency deviation tracing model is activated. Based on the energy efficiency deviation tracing model and combined with real-time collected data, the source of the deviation is located. Identify the core sources and extent of impact of deviations, generate a deviation tracing report, and customize fine-tuning plans for the core sources of deviations; The fine-tuning plan is sent to the production line control system to execute the fine-tuning instructions; If the actual energy efficiency, production capacity and quality meet the requirements after parameter fine-tuning, and there are no abnormal deviations after continuous operation, then the parameter combination after the current parameter fine-tuning is judged to be the optimal energy efficiency solution for pudding production.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention collects and preprocesses equipment, energy consumption, process, and quality data from the production line in real time, constructing a standardized dataset to provide a precise foundation for analysis. It employs a grey relational analysis and fuzzy clustering coupling algorithm, combined with a process energy efficiency entropy evaluation method, to accurately identify core influencing parameters, inefficient operating conditions, and energy efficiency bottleneck processes. By refining the classification of inefficient operating conditions and locating energy efficiency bottleneck processes, and through the construction of a correlation model and closed-loop control, it achieves precise output and dynamic fine-tuning of optimal process parameters. This significantly improves the overall energy efficiency of the bonding production line, reduces energy consumption per unit product, minimizes energy waste, helps enterprises achieve green production goals, aligns with the trend of green transformation in the manufacturing industry, and simultaneously reduces enterprise energy costs. Attached Figure Description

[0016] Figure 1 This is a flowchart of the steps of the present invention. Detailed Implementation

[0017] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0018] Reference Figure 1 As shown, the specific steps of the energy efficiency optimization method for laminate production based on data analysis are as follows: S1. Real-time acquisition of equipment operation data, energy consumption data, process parameters and quality data of the bonding production line, and construction of standardized datasets after data preprocessing; S2. Establish an energy efficiency KPI system based on a standardized dataset. Use a grey relational analysis and fuzzy clustering coupled algorithm to screen out core parameters that are strongly related to energy efficiency through grey relational analysis. Use fuzzy clustering to refine the classification of inefficient working conditions. Based on the process energy efficiency entropy value evaluation method, identify inefficient working conditions, key influencing parameters and energy efficiency bottleneck processes. S3. Construct a correlation model of parameters, energy consumption, capacity and quality based on bidirectional LSTM and attention mechanism, and output the optimal combination of process parameters that meets the constraints of quality and capacity. S4. Send the optimal parameters to the production line control system for automatic execution and collect the execution results in real time for verification. Based on the energy efficiency deviation traceability model, when there is a deviation between the actual energy efficiency and the optimal energy efficiency, quickly locate the source of the deviation, dynamically fine-tune the parameters and perform secondary verification, and finally determine the optimal energy efficiency solution for the lamination production, thereby optimizing the energy efficiency of lamination production.

[0019] This application provides multi-dimensional value to the production of laminated adhesives through standardized data collection and advanced algorithm application throughout the entire process. First, it breaks down traditional data barriers through comprehensive data collection and standardized processing. Combined with a coupled algorithm, it accurately identifies energy efficiency bottlenecks. Then, through the construction of a correlation model and closed-loop control, it optimizes process parameters, effectively improving production line energy efficiency, reducing energy consumption, and supporting green production for enterprises. Quality indicators are incorporated into the entire process analysis. Intelligent models capture the correlation between parameters and quality, achieving synergistic improvement in energy efficiency and quality, and reducing waste. In terms of cost control, this method can reduce energy, equipment maintenance, raw material, and labor costs, significantly improving the economic benefits for enterprises. Intelligent design throughout the entire production process promotes the transformation of production from experience-driven to data-driven, improving the level of production automation and management refinement, and assisting enterprises in intelligent transformation. The dynamic fine-tuning mechanism can quickly respond to production fluctuations, and the correlation model can flexibly adapt to changes in market demand, enhancing production stability and market responsiveness.

[0020] This embodiment discloses a data analysis-based method for optimizing energy efficiency in the production of laminated adhesive tape. The overall process includes: real-time acquisition and standardized processing of multi-source data, construction of energy efficiency KPIs and identification of key parameters and bottleneck processes, multi-objective correlation modeling based on bidirectional LSTM-attention mechanism, optimization and execution of optimal process parameters, real-time tracing and dynamic fine-tuning of energy efficiency deviations, and finally achieving optimal energy efficiency control throughout the entire production process of laminated adhesive tape.

[0021] Multi-source data acquisition and standardized dataset construction: At key process nodes of the lamination production line, such as the feeding end, lamination section, drying section, inspection section, and discharge end, corresponding data acquisition equipment is deployed: The equipment's operating status is collected using current, voltage, speed, temperature, and pressure sensors; Energy consumption data is collected using smart meters, smart gas meters, and smart water meters. Process parameters such as bonding pressure, drying temperature, linear speed, and tension are collected via PLC. Product quality data is collected using machine vision inspection equipment, which collects indicators such as appearance defects, dimensional accuracy, and bonding strength.

[0022] Each data acquisition device starts real-time data acquisition according to the preset acquisition frequency, and the raw data is uploaded to the production line data platform in a unified manner through a combination of industrial Ethernet and wireless transmission.

[0023] The collected raw data undergoes data cleaning, data standardization, and data fusion processing sequentially, specifically as follows: Data cleaning removes missing, outlier, and duplicate values, and eliminates noisy data that is clearly outside the equipment's operating range; Data standardization normalizes or standardizes parameters of different dimensions and orders of magnitude, unifying data ranges; Data fusion aligns multi-source heterogeneous data by timestamp to form a standardized dataset for pull-and-stick production with a unified time-series format.

[0024] Select historical operating condition samples that are continuously and stably in operation, and remove non-steady-state data such as start-up and shutdown phases, fault shutdowns, and abnormal fluctuations to ensure that the sample size n≥30, so as to provide an effective data foundation for subsequent association analysis and clustering.

[0025] Energy efficiency KPI system construction and inefficient operating conditions and bottleneck identification: Based on the energy efficiency standards of the bonding industry, equipment operating characteristics and production processes, an energy efficiency KPI system is constructed, including: unit product energy consumption, comprehensive energy efficiency ratio, process energy efficiency coefficient and key equipment load rate; The above indicators serve as the benchmark for energy efficiency evaluation and are used for subsequent correlation analysis and entropy evaluation. Grey relational analysis was used to screen core influencing parameters, and the energy efficiency KPI system was set as a reference sequence. ; No. The operating / process parameters of each device are a comparison sequence: ; in For the sample size, The number of parameters to be filtered; Calculate the grey relational coefficient: ; In the formula: For the first The parameter in the first... The correlation coefficient of each sample point; For the resolution coefficient, this embodiment takes... ; Calculate the grey relational degree: ; according to Sort by size from largest to smallest, and set a correlation threshold. ,like It was determined to be a strongly correlated core parameter; like Parameters that are determined to be weakly correlated or irrelevant are discarded; finally, a subset of core parameters that are strongly correlated with energy efficiency is obtained. Fuzzy clustering is used to refine the classification of inefficient operating conditions. A normalized operating condition feature matrix is ​​constructed using a subset of core parameters as features. The similarity between features is calculated to establish a fuzzy similarity matrix and construct a fuzzy equivalence matrix. A classification threshold is set to divide the operating conditions into three categories: high efficiency, medium efficiency, and low efficiency. The low efficiency operating condition cluster is further subdivided into subclasses to obtain multiple typical low efficiency operating condition patterns. The parameter range and operating state characteristics corresponding to each pattern are clarified.

[0026] Calculate the process energy efficiency entropy value for each process: feeding, bonding, drying, testing, and discharging. The smaller the entropy value, the more concentrated the energy efficiency fluctuations, the more stable the operation, and the higher the energy efficiency level of the process. The larger the entropy value, the more drastic the energy efficiency fluctuations and the higher the risk of inefficiency. By combining the distribution of core parameters with the clustering results of operating conditions, the contribution characteristics of key parameters in inefficient operating conditions are analyzed; by combining the magnitude of entropy, the proportion of energy consumption of processes, and the clustering categories, the bottleneck processes and typical inefficient operating modes are located, and the types of inefficient operating conditions, key influencing parameters, and a list of bottleneck processes are output.

[0027] Solving for optimal process parameters based on bidirectional LSTM-attention mechanism: Using the selected core process parameters, energy consumption, capacity, and quality time series data as input, the standardized dataset is divided into training set, validation set, and test set according to the proportion, and then reshaped into a time series input format that the model can accept. Build a bidirectional LSTM network to capture the sequential dependencies of production data in both forward and reverse directions simultaneously. An attention mechanism is introduced to assign adaptive weights to features at different times and in different dimensions, strengthening key features that are strongly correlated with energy efficiency and quality. The prediction results of energy consumption, quality, and production capacity are output through a fully connected layer to complete the multi-dimensional correlation modeling of parameters, energy consumption, production capacity, and quality. Iterative training on the training set, optimization on the validation set, and validation on the test set are adopted to ensure that the model prediction accuracy meets the requirements of production control.

[0028] With minimizing production energy consumption as the optimization objective, the following constraints are set: The quality constraints are based on the quality standards of the adhesive tape product, and inequality constraint ranges are set for core indicators such as appearance, size and adhesion. Multiple quality indicators must be met simultaneously. Capacity constraints combine enterprise production plans with equipment rated capacity to set lower and upper limits for capacity. The trained correlation model is used as the fitness function, and a particle swarm optimization algorithm is introduced: a set of process parameter combinations is represented by the particle position; the particle velocity and position are iteratively updated, and the particle that meets the quality and capacity constraints and has the lowest energy consumption is selected; the process parameter combination corresponding to the particle is output, which is the optimal process parameter combination.

[0029] Regularly import new production data to incrementally train and iteratively update the bidirectional LSTM-attention mechanism model, ensuring that the model and the optimal parameter combination can adapt to changes in working conditions.

[0030] Parameter delivery and execution, energy efficiency deviation tracing and dynamic optimization: The optimal combination of process parameters is sent in batches to the control systems of each station on the bonding production line through a standard industrial communication protocol. The system automatically verifies the integrity, rationality and safety of the parameters. After the verification is passed, the parameter configuration is automatically executed and the production line operates stably with the optimal parameters.

[0031] Real-time feedback and verification of execution results: By collecting actual operating data from the production line in real time, and comparing it with the optimal energy efficiency, quality, and capacity indicators predicted by the model: If the deviation between the actual data and the predicted data is within the allowable range, and all quality and capacity constraints are met, the execution is deemed qualified, and the current optimal parameters are maintained; if the deviation exceeds the limit or the constraints are not met, an energy efficiency deviation warning is automatically triggered and deviation tracing is initiated. A preset energy efficiency deviation threshold is set. When the deviation between the actual energy efficiency and the optimal energy efficiency exceeds the threshold, the energy efficiency deviation tracing model is activated. Combining real-time operating data, core parameter status, and equipment operating status, the source of the deviation is located, the degree of influence of each factor is quantified, and a tracing report is generated. Targeted fine-tuning plans are formulated for the core deviation sources, and the fine-tuned process parameter combination is generated and issued for execution. After fine-tuning the parameters, the production line operation status is continuously monitored: if the actual energy efficiency, capacity and quality indicators meet the constraints and the continuous and stable operation is free from abnormal deviations, the fine-tuned parameter combination is determined as the optimal energy efficiency solution for the bonding production, thereby achieving continuous optimization of production energy efficiency.

[0032] This embodiment, through data-driven modeling, intelligent optimization, and closed-loop feedback control, significantly reduces unit product energy consumption and improves the overall energy efficiency of the production line while ensuring the quality and production capacity of the bonding products. It can also adapt to different working conditions and has strong engineering applicability.

[0033] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the present invention is defined by the appended technical solutions and their equivalents.

Claims

1. A method for optimizing energy efficiency in a draw-together production based on data analysis, characterized in that, The specific steps are as follows: S1. Real-time acquisition of equipment operation data, energy consumption data, process parameters and quality data of the bonding production line, and construction of standardized datasets after data preprocessing; S2. Establish an energy efficiency KPI system based on a standardized dataset. Use a grey relational analysis and fuzzy clustering coupled algorithm to screen out core parameters that are strongly related to energy efficiency through grey relational analysis. Use fuzzy clustering to refine the classification of inefficient working conditions. Based on the process energy efficiency entropy value evaluation method, identify inefficient working conditions, key influencing parameters and energy efficiency bottleneck processes. S3. Construct a correlation model of parameters, energy consumption, capacity and quality based on bidirectional LSTM and attention mechanism, and output the optimal combination of process parameters that meets the constraints of quality and capacity. S4. Send the optimal parameters to the production line control system for automatic execution and collect the execution results in real time for verification. Based on the energy efficiency deviation traceability model, when there is a deviation between the actual energy efficiency and the optimal energy efficiency, quickly locate the source of the deviation, dynamically fine-tune the parameters and perform secondary verification, and finally determine the optimal energy efficiency solution for the lamination production, thereby optimizing the energy efficiency of lamination production.

2. The data analysis-based energy efficiency optimization method for bonding production according to claim 1, characterized in that: In step S1, data acquisition of equipment operation, energy consumption, process parameters, and quality is achieved by deploying corresponding data acquisition devices at key nodes of the production line. These devices include sensors, smart meters, and machine vision inspection equipment. The acquisition frequency is also set. Key nodes include the feeding end, bonding section, drying section, inspection section, and discharge end. The data acquisition devices are then activated to capture various types of data in real time, and the raw data is transmitted via industrial Ethernet and wireless transmission. Preprocessing includes data cleaning, data standardization, and data fusion.

3. The data analysis-based energy efficiency optimization method for bonding production according to claim 1, characterized in that: The energy efficiency KPI system in step S2 is derived from industry standards, equipment characteristics, and production processes. The KPI system includes unit product energy consumption, comprehensive energy efficiency ratio, process energy efficiency coefficient, and equipment load rate indicators. The core parameters strongly correlated with energy efficiency were identified through grey relational analysis as follows: The energy efficiency KPI system is used as a reference sequence, and the operating data of various equipment are used as a comparison sequence. Calculate the grey relational coefficient and grey relational degree between the operating data of various equipment and the energy efficiency KPI system; Based on the degree of correlation, a threshold is set to filter out the parameters that are strongly related to energy efficiency and remove the parameters that are weakly related or irrelevant; thus obtaining a subset of core parameters.

4. The data analysis-based energy efficiency optimization method for bonding production according to claim 3, characterized in that: The reference sequence for setting the energy efficiency KPI system is as follows: , No. The operating data of each device is ;in The number of samples; The number of equipment operating parameters; The formula for calculating the grey relational coefficient is: ； in For the first The parameter in the first... Grey relational coefficient of each sample point ; For the first Absolute difference; It is the minimum difference between two levels; The maximum difference between the two levels; The resolution coefficient; The grey relational degree is obtained by taking the arithmetic mean of the correlation coefficients. The calculation formula is: ； in The larger the value, the more likely it is to be a parameter. The stronger the correlation with energy efficiency KPIs; according to Sort by size from largest to smallest, and set a threshold. ; like If , then it is a strongly correlated parameter; if If the parameter is weakly correlated or irrelevant, it should be removed.

5. The data analysis-based energy efficiency optimization method for bonding production according to claim 1, characterized in that, The specific steps for refining the classification of inefficient operating conditions using fuzzy clustering are as follows: A feature matrix is ​​constructed from a subset of core parameters for historical operating condition samples; Establish a fuzzy similarity matrix and perform fuzzy clustering based on the fuzzy equivalence matrix; Combined threshold The operating conditions are divided into high-efficiency, medium-efficiency and low-efficiency, and the low-efficiency operating condition cluster is further subdivided into subcategories. Multiple typical inefficient operating conditions were obtained, and the parameter characteristics and operating status of each type of operating condition were clarified. Based on the process energy efficiency entropy value evaluation method, the inefficient operating conditions, key influencing parameters, and energy efficiency bottleneck processes are identified as follows: Calculate the energy efficiency entropy value for each production process. The smaller the entropy value, the more concentrated the energy efficiency fluctuation and the higher the energy efficiency level; the larger the entropy value, the more unstable the energy efficiency and the higher the risk of inefficiency. By combining core parameters with operating condition clustering results, the distribution characteristics of key influencing parameters in various inefficient operating conditions are located. By combining the magnitude of entropy, the proportion of energy consumption in each process, and the clustering categories of operating conditions, the energy efficiency bottleneck processes and typical inefficient operating modes are identified; the inefficient operating condition types, key influencing parameters, and energy efficiency bottleneck processes are output.

6. The data analysis-based energy efficiency optimization method for bonding production according to claim 5, characterized in that: The operating condition samples were selected from historical operating condition samples that were continuously and stably in operation. Samples from start-up and shutdown phases, fault shutdowns, and abnormal fluctuations were removed. The number of samples n≥30. The parameters in the feature matrix were normalized.

7. The data analysis-based energy efficiency optimization method for bonding production according to claim 1, characterized in that, The S3 steps are as follows: A bidirectional LSTM coupled model with an attention mechanism is constructed, and the optimal combination of process parameters that satisfies quality and capacity constraints is solved based on the coupled model. Core process parameters and energy consumption, capacity, and quality data are screened, preprocessed, divided into datasets, and reshaped into time-series format. A bidirectional LSTM is built to capture temporal features, key features are enhanced based on the attention mechanism, and a correlation model is constructed by combining a fully connected layer. The prediction accuracy is ensured through training, validation and optimization. Quantifiable constraints on quality and production capacity are set, with the goal of minimizing energy consumption. The optimal combination of process parameters is solved and verified using the particle swarm optimization algorithm. Regularly supplement the model with new data to ensure that the parameter set is appropriate for changes in operating conditions.

8. The energy efficiency optimization method for bonding production based on data analysis according to claim 7, characterized in that: The constraints include quality constraints and production capacity constraints. The quality constraints are based on product quality standards, setting the constraint range of core quality indicators and using inequality constraints. When there are multiple quality indicators, all quality constraints are satisfied simultaneously. Capacity constraints are set based on the company's production plan and the rated capacity of the equipment, defining the range of capacity constraints. To find the optimal combination of process parameters, the correlation model is used as the fitness function and input into the particle swarm optimization algorithm. The particles represent different combinations of process parameters. By iteratively updating the particle positions, the particle that satisfies all constraints and minimizes energy consumption is selected as the optimal combination of process parameters.

9. The data analysis-based energy efficiency optimization method for bonding production according to claim 1, characterized in that: In step S4, the optimal parameters are sent out in batches to the control systems of each station on the bonding production line via an industrial communication protocol. The completeness and rationality of the parameters are automatically verified. Once verified, the instructions are automatically executed and the production line starts running according to the optimal parameters. Real-time data collection and verification of execution results: By collecting data in real time during the production of the adhesive bonding process, the collected data is compared with the predicted data corresponding to the optimal parameters to verify the execution effect. If the deviation between the collected data and the predicted data is within the allowable range and all constraints are met, the execution is deemed qualified and the production line continues to operate with optimal parameters. If there is a deviation or the constraints are not met, a deviation warning is triggered and the source of the deviation is traced.

10. The data analysis-based energy efficiency optimization method for bonding production according to claim 1, characterized in that: In step S4, the source of the positioning deviation is combined with the characteristics of the bonding production process, and an allowable threshold for energy efficiency deviation is set. When the deviation between the actual energy efficiency and the optimal energy efficiency exceeds the threshold, the energy efficiency deviation tracing model is activated. Based on the energy efficiency deviation tracing model and combined with real-time collected data, the source of the deviation is located. Identify the core sources and extent of impact of deviations, generate a deviation tracing report, and customize fine-tuning plans for the core sources of deviations; The fine-tuning plan is sent to the production line control system to execute the fine-tuning instructions; If the actual energy efficiency, production capacity and quality meet the requirements after parameter fine-tuning, and there are no abnormal deviations after continuous operation, then the parameter combination after the current parameter fine-tuning is judged to be the optimal energy efficiency solution for pudding production.

Images