A plastic woven bag full-process production carbon efficiency collaborative optimization method

Abstract

The present application relates to a kind of plastic woven bag whole-process production carbon efficiency collaborative optimization method applied to carbon emission optimization technical field, the method first constructs the whole caliber carbon emission accounting model containing equipment load rate nonlinear dynamic emission factor and process coupling carbon emission, quantifies process inter-parameter coupling degree by correlation analysis, identifies including the first strong coupling process group such as melt extrusion wire drawing to round weaving cloth as core carbon emission reduction node;On this basis, with the minimum carbon intensity of unit product as core, coupling qualified rate, capacity, cost constructs multi-objective optimization model, uses improved NSGA-Ⅱ algorithm to carry out collaborative solution, exports Pareto optimal process parameter set.Last, through the trigger rule such as pre-set raw material, order, working condition change, drives model update and closed-loop iterative optimization.The present application significantly improves the carbon emission accounting precision of plastic woven bag production, realizes whole-process carbon efficiency global optimization, and can be widely applied to the low-carbonization management and control of plastic woven bag production.

Description

Technical Field

[0001] This invention relates to the field of carbon emission optimization technology, and more specifically, to a method for synergistic optimization of carbon efficiency in the entire process of plastic woven bag production. Background Technology

[0002] Plastic woven bags (referred to as woven bags) are a major industrial packaging material widely used in chemical, building materials, agriculture, and food industries. The production of woven bags is a typical process-oriented manufacturing process, encompassing multiple continuous steps such as raw material mixing, melt extrusion and drawing, cooling and shaping, circular weaving, coating / lamination, printing, cutting, and sewing. This production process is characterized by a close inter-process quality transfer relationship: fluctuations in the process parameters of the preceding steps (such as extrusion and drawing) directly affect the defect rate and waste volume of the subsequent steps (such as circular weaving), thereby causing additional energy consumption and carbon emissions.

[0003] In carbon emission accounting, existing technologies generally employ the static emission factor method, which calculates the carbon emissions per unit of product for each process using a fixed coefficient. This method has two main drawbacks:

[0004] First, it fails to consider the non-linear variation of carbon emissions per unit product under different load rates. In actual production, when equipment operates below full load, the proportion of energy consumption of the equipment itself (such as motor no-load loss and heat loss from the heating system) to total energy consumption increases, resulting in carbon emissions per unit product that are often significantly higher than those at full load. Furthermore, this increase exhibits a non-linear accelerating trend. The static emission factor method uses fixed coefficients and cannot capture this dynamic change, leading to a significant discrepancy between the calculated carbon emissions and the actual emissions under low load conditions.

[0005] Second, the coupling relationship of carbon emissions between processes is not considered. Existing accounting methods usually simply add up the carbon emissions of each process, ignoring the additional carbon emissions caused by the increase in defect rate and waste in subsequent processes due to fluctuations in process parameters of the preceding process. For continuous process production such as woven plastic bags, where the quality transfer between processes is closely related, this "secondary carbon emissions between processes" accounts for a significant proportion of the total carbon emissions, yet it is not measured by the existing accounting system.

[0006] Chinese patent CN118521100B discloses a method for joint control of material flow, energy flow, and carbon emission flow in long-process steel enterprises. This method couples the material-energy consumption characteristic models of each production process in the steel enterprise with a carbon emission model to construct a coupled material flow-energy flow-carbon emission flow model for joint control. However, this method only considers the carbon emissions of each process as being related to the output of that process and a fixed carbon emission coefficient. It does not consider the nonlinear impact of equipment load rate changes on carbon emissions per unit product, nor does it address the coupling and quantification of the additional carbon emissions in subsequent processes caused by fluctuations in process parameters of upstream processes through the mass transfer chain.

[0007] The aforementioned deficiencies result in insufficient accuracy of existing carbon emission accounting methods in the entire production process of woven plastic bags, thus affecting the effectiveness of subsequent carbon efficiency optimization decisions. Therefore, there is an urgent need for a method that can accurately and dynamically calculate carbon emissions throughout the entire woven plastic bag production process, tailored to the specific characteristics of woven plastic bag production, and then perform multi-process collaborative optimization based on this calculation. Summary of the Invention

[0008] This invention aims to solve the aforementioned technical problems in the existing technology and provides a method for synergistic optimization of carbon efficiency throughout the entire production process of woven plastic bags. This method achieves accurate dynamic accounting of carbon emissions throughout the entire process by establishing a process-level dynamic carbon emission factor model and an inter-process coupled carbon emission quantification mechanism; and based on this, it achieves dynamic optimization of carbon efficiency throughout the entire process through multi-objective synergistic optimization and closed-loop iterative control.

[0009] To achieve the above objectives, this invention provides a method for synergistic optimization of carbon efficiency in the entire production process of woven plastic bags, comprising the following steps:

[0010] S1. Identification of dynamic carbon emission accounting and process coupling relationship throughout the entire process:

[0011] Define the accounting boundaries for the entire plastic woven bag production process and construct a comprehensive carbon emission accounting model covering all processes; the formula for calculating the total carbon emissions of the entire process using the comprehensive carbon emission accounting model is as follows:

[0012] ;

[0013] In the formula, This represents the total carbon emissions throughout the entire process. The total number of processes. For the first Each process is under equipment load rate Below, the dynamic carbon emission factor of qualified products from production units. For the first The yield of qualified products in each process. The carbon emissions resulting from inter-process coupling caused by fluctuations in process parameters of the preceding process, leading to an increase in the defect rate and waste amount in the subsequent process.

[0014] Among them, dynamic emission factors Calculate using the following formula:

[0015] ;

[0016] In the formula, For the first The baseline emission factor for each process when operating at full load. For the first Equipment load rate of each process For the first The no-load emission coefficient of each process, For the first Load rate of each process - emission sensitivity index, and ;

[0017] Inter-process carbon emissions The general calculation formula is: In the formula, This represents the total number of processes that generate secondary waste. For the first The amount of additional waste generated in each process due to fluctuations in the process parameters of the preceding process. For the first Carbon emission factors corresponding to waste recycling and treatment in each process;

[0018] The above dynamic emission factor formula introduces The nonlinear term describes the physical law of accelerated increase in carbon emissions per unit product when the equipment is running under partial load, and solves the problem of large calculation deviation in the static emission factor method under low load conditions. The proposed method explicitly incorporates the additional carbon emissions generated by the increase in the defect rate and waste amount in the subsequent process caused by the fluctuation of process parameters in the preceding process through the quality transmission chain into the accounting model, thus making up for the shortcomings of the existing method that simply adds up the carbon emissions of each process.

[0019] Based on historical big data of production, correlation analysis is used to quantify the correlation coefficients between key process parameters of each process and carbon emissions and product quality indicators of upstream and downstream processes. The absolute value of the correlation coefficient is the parameter coupling degree. The process coupling level is divided according to the magnitude of the parameter coupling degree. Based on the process coupling level, the core carbon emission reduction nodes of the whole process are identified, and the core carbon emission reduction nodes are used as the core objects of subsequent collaborative optimization.

[0020] S2, Full-process, multi-dimensional, multi-objective coupled modeling:

[0021] Using the core controllable process parameters of each process as the input variable set, and taking the minimum carbon emission intensity per unit of qualified product as the core optimization objective, the product qualification rate, production line capacity, and unit product production cost are simultaneously coupled as auxiliary optimization objectives to construct an optimization objective set. A constraint condition set is constructed by combining production process constraints, product quality constraints, safety production constraints, and environmental emission constraints. A nonlinear coupling model between the input variables and each optimization objective is constructed based on production test data and historical production big data.

[0022] The formula for calculating the carbon emission intensity per unit of qualified product is: ;in, Carbon emission intensity per unit of qualified product This refers to the output of qualified products.

[0023] S3. Multi-objective collaborative optimization solution:

[0024] A multi-objective intelligent optimization algorithm is used to solve the nonlinear coupled model, and the Pareto optimal solution set that satisfies all constraints is output. Based on the enterprise's production needs, the Pareto optimal solution set is comprehensively sorted, the optimal compromise solution is selected, and the optimal combination of process parameters for each process is output.

[0025] S4. Closed-loop management and dynamic iterative optimization throughout the entire process:

[0026] The optimal combination of process parameters is distributed to each process equipment for execution. Real-time collection of production operation data, carbon emission data, quality data, capacity data and cost data throughout the entire process. When the production conditions change in accordance with preset rules, the nonlinear coupling model is automatically updated and steps S2 to S4 are re-executed to achieve closed-loop control and dynamic iterative optimization of carbon efficiency.

[0027] Preferably, in step S1, the correlation analysis method is the Pearson correlation coefficient method; the process coupling level includes first-level strong coupling, where the parameter coupling degree of first-level strong coupling is ≥0.75, and the process group corresponding to first-level strong coupling includes at least the melt extrusion drawing process → cooling and setting process → circular weaving process. The selection of this process group is based on the actual process characteristics of woven plastic bags: in the woven plastic bag production process, the extrusion temperature and draw ratio of the melt extrusion drawing process and the cooling water temperature of the cooling and setting process directly affect the physical properties and dimensional uniformity of the flat yarn; while the quality uniformity of the flat yarn directly affects the breakage rate and defect rate of the subsequent circular weaving process. Therefore, these three processes constitute the most closely coupled process group in the woven plastic bag production process.

[0028] Furthermore, for the first-level strongly coupled process group, a separate parameter linkage sub-model can be constructed. The parameter linkage sub-model quantifies the coupling relationship between extrusion temperature, stretch ratio, cooling water temperature and circular weaving breakage rate and defect rate, so as to realize the parameter linkage carbon efficiency optimization of the core process.

[0029] Preferably, in step S1, the dynamic emission factor In the middle, the no-load emission coefficient The value ranges from 0.2 to 0.5, and the load factor-emissions sensitivity index is... The value range is from 1.0 to 1.5. This parameter range is set based on the measured energy consumption data of key equipment such as drawing machines, circular looms, coating machines, printing machines, and bag making machines on the plastic woven bag production line. It is obtained through nonlinear regression fitting and reflects the typical operating characteristics of plastic woven bag production equipment.

[0030] Preferably, in step S2, the nonlinear coupling model is constructed using Box-Behnken response surface methodology combined with a backpropagation (BP) neural network. The construction steps include: designing multi-factor, multi-level orthogonal experiments for the core coupled process group to obtain effective experimental data, and combining this with historical production big data to form a model training dataset; constructing a BP neural network prediction model containing three hidden layers, using the input variable set as the input layer and the optimization target set as the output layer, and employing the ReLU activation function and Adam optimization algorithm to complete model training; and determining the model's prediction determination coefficient. The model is deemed compliant when the carbon emission intensity prediction error is ≤3% and the product qualification rate prediction error is ≤1%. This construction method organically combines the uniform sampling capability of response surface methodology with the nonlinear fitting capability of BP neural network, enabling the acquisition of a high-precision process parameter-carbon efficiency-mass mapping model with a limited number of experiments. This solves the problem of insufficient accuracy of traditional regression models caused by the high dimensionality and complex interaction relationships of process parameters in woven plastic bag production.

[0031] Preferably, in step S3, the multi-objective intelligent optimization algorithm is an improved NSGA-II algorithm. The improvements include: introducing adaptive crossover and mutation operators to dynamically adjust the crossover and mutation probabilities based on the population's generational evolution and individual fitness values; introducing a crowding-adaptive adjustment strategy to optimize the uniformity of the Pareto optimal solution set; and adding a constraint gradient penalty function to ensure that all output optimal solutions meet the constraint requirements. Specifically, the adaptive crossover and mutation operators adapt to the high-dimensional characteristics of the plastic weaving process input variables, dynamically balancing global search and local optimization capabilities during the evolutionary process; the crowding-adaptive adjustment strategy optimizes the uniformity of the Pareto optimal solution set under multiple optimization objectives; and the constraint gradient penalty function ensures that all output solutions comply with the mandatory constraints of safe operation of the plastic weaving equipment and product quality.

[0032] Preferably, in step S3, the Pareto optimal solution set is comprehensively sorted based on the enterprise's production needs, and the entropy weight method-TOPSIS comprehensive evaluation method is adopted to complete the screening of the optimal compromise solution according to the enterprise's preset priority weights for carbon reduction, production capacity maintenance, cost control and quality improvement.

[0033] Preferably, in step S4, the preset rules for triggering dynamic iterative optimization include satisfying any of the following conditions: changes in raw material batches or raw material performance; changes in order specifications or product quality requirements; changes in equipment operating status; deviations between actual carbon emissions and model predictions exceeding a preset threshold; and when it is detected that the actual output of a downstream process is consistently lower than that of a upstream process due to non-fault reasons, and the duration exceeds a preset duration. The preset threshold is set based on the industry's allowable error range for carbon emission calculation in woven plastic bag production; the preset duration is set based on the maximum buffer duration of the semi-finished product buffer rack in the woven plastic production line to avoid equipment downtime due to full buffer load.

[0034] Furthermore, in step S2, the carbon cost factor of the carbon market is also incorporated into the optimization target set, and the carbon trading price is included in the calculation of carbon emission intensity, so as to achieve synergy between carbon efficiency optimization and carbon cost control.

[0035] Furthermore, for different product types such as general woven bags, food-grade woven bags, container bags, and flood control bags, a differentiated constraint and weight system is constructed to output the optimal combination of process parameters suitable for the corresponding product type.

[0036] The spatial boundary of the full-process accounting boundary covers the entire production chain from raw material warehousing to finished product qualification warehousing, including main processes and supporting auxiliary processes. The main processes include raw material proportioning and mixing, melt extrusion and drawing, cooling and shaping, circular weaving into fabric, coating / composite, printing, cutting and sewing, online waste recycling and granulation. Supporting auxiliary processes include cooling water circulation system, waste heat recovery system, and compressed air power system. The full-caliber carbon emission accounting model covers greenhouse gas accounting scope 1 (direct emissions from fuel combustion in the production process and direct emissions from the process), scope 2 (indirect emissions from purchased electricity, heat, and steam), and scope 3 (carbon emissions implied in raw material production, carbon emissions from waste recycling and treatment, and carbon emissions from in-plant transportation).

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. This invention achieves a significant improvement in the accuracy of carbon emission accounting through a dual mechanism: First, it introduces a dynamic emission factor. ,Establish The nonlinear model captures the physical law of accelerated increase in carbon emissions per unit product when the equipment is running under partial load, overcoming the defect of large deviation in carbon emission accounting under low load conditions of the static emission factor method; secondly, it introduces inter-process coupled carbon emissions. This study, for the first time, quantifies in a mathematical model the additional carbon emissions generated by fluctuations in process parameters of upstream processes leading to increased defect rates and waste in downstream processes through a quality transmission chain. This fills a theoretical gap in existing accounting methods that simply add up carbon emissions from each process while neglecting the carbon transfer effect between processes caused by defects. Both mechanisms are integrated into a unified accounting formula. The organic integration of these technologies upgrades the entire process of carbon emission accounting from the traditional static model of "summing up fixed coefficients of each process" to a dynamic system model of "capturing dynamic load rates + coupling effects between processes".

[0039] 2. This invention identifies the coupling relationship of the process flow (such as identifying melt extrusion drawing → cooling and shaping → circular weaving as a first-level strongly coupled process group), and uses the identification results as input to a multi-objective optimization model, so that the optimization solution is no longer a mechanical combination of independent optimization of each process, but a collaborative decision oriented towards the optimal carbon efficiency of the entire process. The introduction of this item enables the optimization model to predict the chain reaction of carbon emissions from downstream processes caused by adjustments to process parameters, thereby avoiding the risk of "local optimization leading to global deterioration" at the source.

[0040] 3. This invention utilizes a model reconstruction-based closed-loop iterative mechanism triggered by preset rules. This mechanism automatically updates the nonlinear coupled model and re-solves for the optimal combination of process parameters when production conditions change, such as variations in raw material batches, product specification switching, or equipment status changes. Unlike existing static optimization methods that rely solely on periodic parameter adjustments based on a fixed model, this invention's "model reconstruction-based" closed loop ensures that the optimization model remains synchronized with the actual production state, enabling real-time adaptive adjustment of the carbon efficiency optimization strategy to dynamic production changes.

[0041] 4. This invention fully considers the industry characteristics of woven plastic bag production, such as discrete but sequential processes, variable material forms, dispersed energy consumption points, and close quality transmission relationships between processes. By using the Pearson correlation coefficient method, the most tightly coupled process group in the woven plastic bag production process is identified, and a parameter linkage sub-model is constructed for this process group. This overcomes the technical obstacle of general carbon emission optimization methods failing to effectively handle industry-specific coupling relationships when applied to the woven plastic bag industry. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating the overall process of carbon efficiency synergistic optimization for the entire production process of woven plastic bags in this embodiment of the invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the scope of protection of the invention.

[0044] Example 1

[0045] Figure 1 This paper presents a method for synergistic optimization of carbon efficiency in the entire production process of woven plastic bags. This embodiment uses a standardized general-purpose woven bag production line of a woven plastic bag manufacturer as an example to describe the complete implementation process of the method. The production line is designed to produce 30 million standard general-purpose woven bags per year. The main production processes include: raw material mixing, melt extrusion drawing, cooling and shaping, circular weaving, coating / lamination, printing, and cutting and sewing. Supporting auxiliary processes include a cooling water circulation system, a waste heat recovery system, and a compressed air power system. The key equipment for each process and its designed capacity and rated power parameters are shown in Table 1.

[0046] Table 1: Design and Rated Parameters of Key Equipment for Each Process

[0047] Melt extrusion drawing Single screw extruder + stretching unit 150kg / h 55 Cooling and shaping Cooling water tank + shaping roller assembly 150kg / h 15 Circular woven fabric 10 four-shuttle circular looms 15kg / h / unit 7.5 / unit Coating / Lamination Extrusion coating laminating machine 120kg / h 45 print Flexographic printing press 3000 messages / hour 12 Cutting and sewing Automatic cutting and sewing machine 2000 items / hour 8

[0048] Note: In the supporting auxiliary processes, the total rated power of the cooling water circulation system is 20 kW, and the total rated power of the air compressor power system is 30 kW.

[0049] Step S1: Dynamic accounting of carbon emissions throughout the entire process and identification of the coupling relationship between processes

[0050] The first step is to define the spatial boundaries of the full-process accounting boundary, covering the entire production chain from raw material entry into the factory and warehousing to finished product inspection and warehousing. This includes all main processes listed in Table 1, as well as supporting auxiliary processes such as cooling water circulation systems, waste heat recovery systems, and compressed air power systems. The accounting scope covers Scope 1 (direct emissions from fuel combustion in the production process and direct emissions from the process), Scope 2 (indirect emissions from purchased electricity and heat), and Scope 3 (carbon emissions implicit in raw material production, carbon emissions from waste recycling and treatment, and carbon emissions from in-plant transportation).

[0051] The second step is to construct dynamic carbon emission factor models for each process. Taking the melt extrusion fiber drawing process as an example, based on the production history data of this process over the past year (no less than 300 effective working days), the load rates of different equipment are directly obtained. Total energy consumption (converted to) ) and the output of qualified products Calculate the actual carbon emissions per unit of product at each load rate. .by As the independent variable, with As the dependent variable (i.e., the incremental ratio of carbon emissions per unit of product relative to the full-load baseline), the nonlinear least squares method is used to modify the formula. Perform fitting to obtain parameters and .

[0052] Taking the melt extrusion fiber drawing process as an example, the following results were obtained through fitting: The corresponding dynamic emission factor formula is: Statistical tests showed that the coefficient of determination of the fitted model was... Root mean square error This indicates that the model fits well.

[0053] Similarly, parameter fitting was performed on each process, including cooling and setting, circular weaving, coating / composite, printing, cutting and sewing. The results of the dynamic emission factor model parameter fitting for each process are summarized in Table 2.

[0054] Table 2: Fitting Results of Dynamic Emission Factor Model Parameters for Each Process

[0055] Process i <![CDATA[ (kgCO2e / t)]]> Melt extrusion drawing (1) 285.6 0.32 1.28 0.94 Cooling and shaping (2) 78.3 0.25 1.15 0.91 Circular knitted fabric (3) 312.5 0.41 1.42 0.96 Coating / Lamination (4) 198.7 0.28 1.18 0.93 Print(5) 65.2 0.22 1.08 0.89 Cutting and sewing (6) 52.8 0.20 1.05 0.88

[0056] As can be seen from Table 2, each process... The values ​​are all greater than 1, verifying that the carbon emissions per unit product show a non-linear accelerating growth trend when the equipment is operating under partial load. Among them, the circular weaving process... The value is the highest (1.42), reflecting that the carbon emissions of the circular loom are most sensitive to changes in the load rate. This is mainly because the circular loom is a multi-station collaborative operation equipment, and the cumulative effect of the idling energy consumption of the remaining stations when some stations are stopped is significant.

[0057] The third step involves using historical big data on production to quantify the correlation coefficients between key process parameters of each process and carbon emissions and product quality indicators of upstream and downstream processes, and to calculate the parameter coupling degree.

[0058] Taking the melt extrusion drawing → cooling and setting → circular weaving process as an example, the key process parameters selected include: extrusion temperature. , stretch ratio Cooling water temperature Product quality indicators include: tensile strength of flat wire Elongation at break of flat wire Circular weaving breakage rate Circular knitting defect rate Valid batch data (no less than 500 batches) from the past two years were extracted from the production history database. The Pearson correlation coefficient method was used to calculate the correlation coefficient between each parameter, and the absolute value of the correlation coefficient was used as the parameter coupling degree.

[0059] The calculation results show that:

[0060] Extrusion temperature With circular weaving breakage rate The Pearson correlation coefficient between them was -0.83, and the parameter coupling degree was 0.83 (≥0.75).

[0061] stretch ratio With circular weaving breakage rate The Pearson correlation coefficient between them was 0.79, and the parameter coupling degree was 0.79 (≥0.75).

[0062] Cooling water temperature With circular weaving breakage rate The Pearson correlation coefficient between them is -0.76, and the parameter coupling degree is 0.76 (≥0.75).

[0063] Extrusion temperature Tensile strength of flat wire The Pearson correlation coefficient between them was 0.81, and the parameter coupling degree was 0.81 (≥0.75).

[0064] stretch ratio Tensile strength of flat wire The Pearson correlation coefficient between them is -0.77, and the parameter coupling degree is 0.77 (≥0.75).

[0065] Based on the above calculation results, the parameter coupling degree of the melt extrusion drawing → cooling and shaping → circular weaving process group is ≥0.75, which meets the judgment criteria of first-level strong coupling. Therefore, it is identified as a first-level strong coupling process group, serving as the core carbon emission reduction node and the core object of subsequent collaborative optimization in this embodiment.

[0066] Coupling degree calculations were also performed for other adjacent process groups. The parameter coupling degree of the coating / lamination → printing → cutting and sewing process group was between 0.4 and 0.7, which belongs to the second-order coupling; the parameter coupling degree between the raw material proportioning and mixing, online waste recycling and granulation processes was 0.55.

[0067] Fourthly, based on the above process coupling level identification results, the core carbon emission reduction node of the entire process is the first-level strongly coupled process group of "melt extrusion drawing → cooling and shaping → circular weaving". Subsequent multi-objective collaborative optimization will be carried out with the process parameters of this process group as the main decision variables.

[0068] Step S2: Full-process multi-dimensional and multi-objective coupled modeling

[0069] The first step is to determine the set of input variables. Based on the core carbon emission reduction nodes and process coupling levels identified in step S1, this embodiment selects the core controllable process parameters of each process as input variables, including but not limited to: the recycled material ratio in the raw material mixing process, the temperature and stretch ratio of each section of the extruder in the melt extrusion and drawing process, the circular loom speed and shed angle in the circular weaving process, the coating thickness and laminating temperature in the coating / composite process, and the printing speed and drying temperature in the printing process. A total of 11 input variables are included.

[0070] The second step is to determine the optimization objective set. The optimization objective set consists of four objectives:

[0071] Core optimization objective: Carbon emission intensity per unit of qualified product The minimum value is calculated using the following formula: ,in , The final qualified product output throughout the entire process;

[0072] Optimization Objective 1: Maximize product qualification rate;

[0073] Secondary optimization objective 2: Maximize production line capacity (output of qualified products per unit time);

[0074] Auxiliary optimization objective 3: Minimize the unit product production cost.

[0075] The third step is to determine the set of constraints. Combining production process constraints, product quality constraints, safe production constraints, and environmental emission constraints, the main constraints set in this embodiment include:

[0076] (1) Product quality constraints: tensile strength of flat wire (In this embodiment) ), elongation at break of flat wire (In this embodiment) ), circular knitting breakage rate (In this embodiment) Coating adhesion (In this embodiment) Sewing strength (In this embodiment) ).

[0077] (2) Equipment operation constraints: Equipment load rate of each process (Where, the lower limit of 0.3 represents the minimum economic operating load rate of the equipment), extruder temperature stretch ratio Circular loom speed .

[0078] (3) Environmental emission constraints: total carbon emissions throughout the entire process The carbon emission quota of the enterprise shall not be exceeded. (In this embodiment) Based on the annual quota issued by the local ecological and environmental protection department, the daily quota is 2850 kg CO2e.

[0079] The fourth step is to construct a nonlinear coupling model. A Box-Behnken response surface methodology combined with a backpropagation (BP) neural network is used to construct a nonlinear coupling model between the input variables and the optimization objectives. The specific steps are as follows:

[0080] (1) Design a multi-factor, multi-level orthogonal experimental scheme. Using the five key process parameters (extrusion temperature, stretch ratio, cooling water temperature, circular loom speed, and shed angle) of the core coupled process group (melt extrusion drawing → cooling and setting → circular weaving) identified in step S1 as experimental factors, each factor was assigned three levels (low, medium, and high). Box-Behnken design was used to generate 46 experimental schemes (including 6 sets of center point repeated experiments). The experiments were conducted on the production line according to the experimental schemes. After each set of experiments stabilized, production data for no less than 2 hours was collected, and the values ​​of each optimization objective were recorded.

[0081] (2) Expand the training dataset. The above 46 sets of valid experimental data were merged with 1,000 sets of historical production data selected from the production history database to form a model training dataset of 1,046 samples. The dataset was randomly divided into training set (732 sets), validation set (209 sets) and test set (105 sets) in a ratio of 7:2:1.

[0082] (3) Constructing a BP neural network model. The model adopts a 4-layer structure: the input layer contains 11 neurons (corresponding to 11 input variables), the 3 hidden layers contain 64, 128, and 64 neurons respectively, and the output layer contains 4 neurons (corresponding to 4 optimization objectives: carbon emission intensity, product qualification rate, production line capacity, and unit product production cost). The activation function is the ReLU function, the optimization algorithm is Adam, the initial learning rate is set to 0.001, the batch size is 32, and the maximum number of training rounds is set to 500 rounds.

[0083] (4) Model Training and Validation. The BP neural network is trained using the training set, and its performance is evaluated on the validation set after each training round. The model is considered to have met the standards and training is stopped when the prediction determination coefficient $R^2$ on the validation set reaches 0.95 or higher, the carbon emission intensity prediction error is within 3%, and the product qualification rate prediction error is within 1%. In this embodiment, the model reaches the above accuracy indicators after the 387th training round, and training is terminated. The final evaluation result on the test set is: carbon emission intensity prediction... The mean prediction error is 2.3%; product qualification rate prediction. The average forecast error is 0.7%; capacity forecast. Production cost forecast All indicators meet the preset model compliance standards.

[0084] Step S3: Solving multi-objective collaborative optimization

[0085] The first step involves using the nonlinear coupled model constructed in step S2 as the fitness function and employing the improved NSGA-II algorithm for multi-objective optimization. The algorithm parameters are set as follows: population size... Maximum number of iterations .

[0086] Improvement measures include:

[0087] (1) Introduce adaptive crossover and mutation operators. In the early stages of evolution (the first 30% of iterations), the crossover probability... It remained at a high level (0.85-0.90), with a high probability of mutation. The crossover probability is maintained at a low level (0.02-0.03) to enhance global search capabilities. During the mid-evolutionary phase (30%-70% of iterations), the crossover probability gradually decreases to 0.70-0.80, while the mutation probability gradually increases to 0.04-0.06. In the late-evolutionary phase (the last 30% of iterations), the crossover probability further decreases to 0.60-0.70, while the mutation probability increases to 0.07-0.10, to enhance local search capabilities and maintain population diversity. This adaptive mechanism adapts to the high-dimensional characteristics of the plastic weaving process input variables (11 dimensions in this embodiment), dynamically balancing global search and local optimization capabilities.

[0088] (2) Introduce a crowding degree adaptive adjustment strategy. After each non-dominated sorting, the sorting weights of individuals within the same non-dominated level are dynamically adjusted according to their crowding degree, so that individuals with lower crowding degree will have higher priority in the next round of selection, thereby optimizing the uniformity of the distribution of the Pareto optimal solution set in the target space composed of the four optimization objectives of carbon emission intensity, product qualification rate, production capacity and cost.

[0089] (3) Add a constraint gradient penalty function. For solutions that violate the constraints, the fitness value is corrected towards the feasible region through the gradient penalty function to ensure that all solutions in the final Pareto optimal solution set meet the mandatory constraints of safe operation of plastic weaving equipment and product quality.

[0090] The algorithm converges after 300 iterations, outputting a Pareto optimal solution set containing 45 feasible solutions.

[0091] The second step involves using the entropy weight method-TOPSIS comprehensive evaluation method to comprehensively rank the Pareto optimal solution set based on the company's current production needs. The company sets the priority weights of the evaluation indicators according to its current business strategy as follows: carbon reduction 40%, maintaining production capacity 25%, cost control 20%, and quality improvement 15%. After objectively determining the information weights of each indicator using the entropy weight method, the TOPSIS method is used to calculate the comprehensive closeness of each solution to the ideal solution, combined with the aforementioned subjective priority weights. After ranking, the solution with the highest comprehensive closeness (numbered P28) is determined as the optimal compromise solution. The optimal process parameter combinations for each process corresponding to this solution are shown in Table 3.

[0092] Table 3: Optimal process parameter combinations for each process corresponding to the optimal compromise solution

[0093] Melt extrusion drawing Extrusion temperature (°C) 215 208 -3.3% Melt extrusion drawing stretch ratio 5.5 5.2 -5.5% Cooling and shaping Cooling water temperature (°C) 25 22 -12.0% Circular woven fabric Circular loom speed (rpm) 150 158 +5.3% Circular woven fabric Roofing angle (°) 32 30 -6.3% Coating / Lamination Coating thickness (mm) 0.025 0.023 -8.0% Coating / Lamination Coating temperature (°C) 185 178 -3.8% print Printing speed (strips / hour) 3000 3200 +6.7% print Drying temperature (°C) 65 62 -4.6% Cutting and sewing Heat sealing temperature (°C) 160 155 -3.1% Cutting and sewing Bag making speed (bags / hour) 2000 2100 +5.0%

[0094] As shown in Table 3, under the premise of meeting quality constraints, the extrusion temperature of the drawing-cooling-circular weaving process group was reduced by 3.3%, the draw ratio by 5.5%, the cooling water temperature by 12%, and the circular loom speed by 5.3%, effectively reducing the energy consumption and carbon emissions of this process group. Simultaneously, optimizing the cooling water temperature reduced the circular weaving breakage rate, thus reducing waste generated from breakage and carbon emissions from waste reprocessing.

[0095] In this embodiment, the optimization objective set in step S2 also incorporates the carbon cost factor of the carbon market. Taking the closing price of 82.5 yuan / ton CO2 in the national carbon emission trading market on that day as an example, if this carbon trading price is included in the calculation of carbon emission intensity, the carbon cost corresponding to the optimal compromise solution (P28) is: 342.8 kgCO2e / ton of product × 82.5 yuan / ton CO2 ÷ 1000 = 28.28 yuan / ton of product. In contrast, the carbon emission intensity of the unoptimized scheme (original operating parameters) is 368.5 kgCO2e / ton of product, and the carbon cost is 30.41 yuan / ton of product. The optimized carbon cost is reduced by about 7.0%, which translates to an annual carbon cost saving of about 16,000 yuan based on an annual production of 30 million general-purpose woven bags (about 7,500 tons of product). When the carbon trading price rises to 150 yuan / ton CO2 (for example, when the carbon market compliance period is approaching and quotas are tightening), the optimized annual carbon cost saving will expand to about 29,000 yuan.

[0096] Step S4: Closed-loop management and dynamic iterative optimization of the entire process

[0097] The optimal process parameter combination shown in Table 3 is distributed to each process equipment through the production execution system. The system collects real-time production operation data, carbon emission data, quality data, capacity data, and cost data throughout the entire process at a 5-minute interval, and performs online preprocessing and storage of the collected data.

[0098] During continuous production, the system automatically monitors the following preset rules for triggering conditions:

[0099] Has the batch or properties of the raw materials changed?

[0100] Have the order specifications or product quality requirements changed?

[0101] Has the equipment's operating status changed?

[0102] Whether the deviation between the actual carbon emission value and the model prediction value exceeds the preset threshold (set to 5% in this embodiment, which is based on the industry's allowable error range for carbon emission accounting of woven plastic bags).

[0103] If the actual output of the subsequent process is consistently lower than that of the previous process due to non-fault reasons, and the duration exceeds the preset time (set to 30 minutes in this embodiment, which is based on the maximum buffer time of the semi-finished buffer rack of the plastic weaving production line to avoid equipment shutdown caused by full load of the buffer).

[0104] To illustrate the closed-loop process, take a raw material batch switchover as an example: When the raw material supplier is changed or different batches of polypropylene raw materials from the same supplier are put into use, the online near-infrared spectrometer detects that the melt index of the new batch of raw materials has changed by 12% compared to the original batch (exceeding the company's internal control standard of 10% batch difference threshold), and it is determined that the raw material batch has changed, thus the triggering condition is met.

[0105] The system automatically executes the following closed-loop iterative operations: First, a full data update is performed on the nonlinear coupled model constructed in step S2, merging the production data accumulated since the new batch of raw materials was put into production (no less than 100 sets) with the original training dataset, and the BP neural network model is retrained according to the method described in step S2. Then, the updated model is re-solved using the improved NSGA-II algorithm described in step S3 to find the Pareto optimal solution set, and a new optimal compromise solution is selected according to the current enterprise priority weights (carbon reduction 40%, maintaining capacity 25%, cost control 20%, quality improvement 15%). Finally, the new optimal process parameter combination is distributed to each process equipment for execution, completing one round of closed-loop iterative optimization.

[0106] In this embodiment, the new optimal process parameter combination obtained after the raw material batch switch, compared with the optimal process parameters before the switch, further adjusts the extrusion temperature to 205℃ (a decrease of 1.4%) and the draw ratio to 5.1 (a decrease of 1.9%) to adapt to the changes in the flowability of the new batch of raw materials. The carbon emission intensity of the entire process is reduced from 368.5 kgCO2e / ton of product before optimization to 342.8 kgCO2e / ton of product, a decrease of 7.0%.

[0107] During the above optimization process, The accounting of this item played a crucial role. The following detailed explanation uses the single-stage strongly coupled process group of drawing → cooling → circular weaving as an example. The calculation process.

[0108] The calculation model is as follows: ,in, The fluctuation of parameters in the preceding process leads to the subsequent process The amount of additional waste generated. For processing procedures Carbon emission factor per unit of waste.

[0109] Specifically, based on extrusion temperature Taking the impact of fluctuations on the circular knitting process as an example:

[0110] The first step is to establish the extrusion temperature. With circular weaving breakage rate A regression prediction model was developed based on historical production data (no fewer than 500 batches). A univariate nonlinear regression model was used to obtain the following results: (times / thousand shuttles), where 210℃ is the long-term operating process reference temperature of this production line, and the deviation from the optimized 208℃ is within the allowable range of the process and does not affect the prediction accuracy of the regression model. (Fitness determination coefficient) .

[0111] The second step is to calculate the deviation due to extrusion temperature. The resulting additional waste. Predicted filament breakage rate when the actual extrusion temperature is 205℃ (3℃ lower than the optimal value of 208℃). Count / thousand shuttles; when the extrusion temperature fluctuates to 213℃ (5℃ higher than the optimal value of 208℃), predict the filament breakage rate. Based on this model, the change in filament breakage rate corresponding to any extrusion temperature deviation can be quantified.

[0112] The third step is to convert the change in breakage rate into additional waste. The waste generated per thousand shuttles of broken yarn is approximately 0.12 kg (estimated based on the weight of a single flat yarn and the average length of broken yarn), therefore: Under the original operating conditions, fluctuations in extrusion temperature caused... The rate is 1.8 times per thousand shuttles (compared to the optimal value of 1.2 times per thousand shuttles). .

[0113] Step 4, Calculation $. Carbon emission factors of waste recycling and granulation processes. Then the extrusion temperature fluctuations will cause .

[0114] Similarly, the effects of factors such as stretch ratio fluctuation and cooling water temperature fluctuation can be calculated separately. and corresponding The components are summarized to obtain the total for this process group. Value. In this embodiment, the total value of the drawing → cooling → circular weaving process group under the original operating state. The calculated CO2e is 0.39 kgCO2e / kiloshuttle, accounting for approximately 3.2% of the total carbon emissions of this process group. The optimized extrusion temperature control accuracy has been improved from ±8℃ to ±5℃. It decreased to 0.24 kgCO2e / kiloshunt, a reduction of 38.5%.

[0115] Example 2

[0116] This embodiment uses a food-grade woven bag production line from the same plastic woven bag manufacturer as an example to demonstrate the implementation process and effects of the carbon efficiency synergistic optimization method of the present invention in the scenario of adjusting the proportion of recycled materials (PCR, Post-Consumer Recycled). Compared with general woven bags, food-grade woven bags have stricter requirements for flat yarn strength, hygiene indicators, and appearance quality, while allowing the use of a certain proportion of recycled polypropylene (PP) raw materials to achieve carbon reduction goals, provided that performance is guaranteed.

[0117] The basic configuration of the production line is the same as in Example 1, but the product quality constraints are more stringent: flat wire tensile strength (0.22 N / tex in Example 1), elongation at break of flat filament (12% in Example 1), circular knitting breakage rate (In Example 1, the number of applications was 2 per thousand shuttles), coating adhesion (3 N / 15 mm in Example 1), sewing strength (In Example 1, it is 200 N / 50 mm).

[0118] Step S1: Dynamic accounting of carbon emissions throughout the entire process and identification of the coupling relationship between processes

[0119] Following the method described in Example 1, the boundary delineation of the entire process accounting and the construction of dynamic emission factor models for each process were completed. Unlike general woven bag production, this example adds a recycled material ratio variable to the raw material mixing process. Its value ranges from 0% to 30% (the upper limit of recycled material content in food-grade packaging bags is set according to relevant food safety standards).

[0120] Because recycled materials differ from virgin polypropylene in melt flow index, impurity content, and thermal stability, changes in the recycled material ratio directly affect the process parameter window and carbon emission characteristics of the melt extrusion and fiber drawing process. When conducting Pearson correlation analysis on the key process parameters of each step and their relationship with carbon emissions and product quality indicators of upstream and downstream processes, the recycled material ratio was added in addition to the parameters analyzed in Example 1. Calculation of correlation coefficients with various downstream indicators.

[0121] The calculation results show that the proportion of recycled materials is With circular weaving breakage rate The Pearson correlation coefficient between them was 0.78 (parameter coupling degree 0.78, ≥0.75), and the tensile strength of the flat wire was... The Pearson correlation coefficient between the parameters was -0.74 (parameter coupling degree 0.74, close to the first-order strong coupling threshold), while the Pearson correlation coefficient between the parameters and the carbon emissions from the melt extrusion drawing process was 0.71. These results indicate that the recycled material ratio is a key upstream variable affecting the carbon efficiency of the first-order strong coupling process group. Its fluctuations significantly impact the breakage rate and defect rate of the circular weaving process through the mass transfer chain, thereby generating additional coupled carbon emissions. .

[0122] Therefore, in this embodiment, the identification results of the core carbon emission reduction nodes are expanded based on those of Embodiment 1: in addition to identifying melt extrusion drawing → cooling and shaping → circular weaving as a first-level strongly coupled process group, the recycled material ratio of the raw material mixing process is also included. As a key upstream control variable for this core carbon emission reduction node, it is included in the set of input variables for subsequent collaborative optimization.

[0123] Step S2: Full-process multi-dimensional and multi-objective coupled modeling

[0124] This embodiment adds a 12th input variable to the 11 variables in Embodiment 1: the proportion of recycled material in the raw material mixing process. .

[0125] The optimization objective set is consistent with that of Example 1, and includes four objectives: carbon emission intensity per unit of qualified product. Minimum (core objective), maximum product qualification rate, maximum production line capacity, and minimum unit product production cost.

[0126] The constraint set is adjusted based on Example 1 as follows to meet the quality requirements of food-grade woven bags:

[0127] flat wire tensile strength ;

[0128] Elongation at break of flat wire ;

[0129] Circular weaving breakage rate ;

[0130] Coating adhesion ;

[0131] Sewing strength .

[0132] The nonlinear coupling model was constructed using the same Box-Behnken response surface methodology as in Example 1, combined with a backpropagation (BP) neural network. The number of neurons in the input layer was adjusted to 12 (due to the addition of a recycled material ratio variable), and the number of neurons in the three hidden layers were adjusted to 72, 144, and 72 respectively (to appropriately increase network capacity based on the increased input dimension). The output layer remained at 4 neurons. After training and validation, the model demonstrated its ability to predict carbon emission intensity on the test set. The average prediction error is 2.6%; product qualification rate prediction. The mean prediction error is 0.8%, which meets the model qualification standard. Carbon emission intensity prediction error Product qualification rate prediction error ).

[0133] Step S3: Solving multi-objective collaborative optimization

[0134] The improved NSGA-II algorithm was used for the solution, with the same algorithm parameters as in Example 1. The priority weights for enterprises were adjusted as follows: carbon reduction 45%, quality assurance 25% (the original "ensuring production capacity" and "improving quality" were merged and the quality weight was increased to adapt to the higher quality requirements of food-grade woven bags), cost control 20%, and production capacity assurance 10%.

[0135] After 300 iterations, the algorithm converged, outputting the Pareto optimal solution set. The optimal compromise solution was selected using the entropy weight method-TOPSIS comprehensive evaluation method. The optimization results of the main process parameters of this optimal solution under different recycled material ratios are shown in Table 4.

[0136] Table 4: Optimal combination of process parameters for food-grade woven bags with different recycled material ratios

[0137] Recycled material ratio 0% (raw material) 15% 25% 30% Extrusion temperature (°C) 212 218 226 233 stretch ratio 5.3 5.1 4.8 4.5 Cooling water temperature (°C) 22 20 18 16 Circular loom speed (rpm) 160 156 150 145 Roofing angle (°) 30 31 32 33 <![CDATA[Full-process carbon emission intensity (kgCO2e / t)]]> 356.2 368.5 385.7 402.3

[0138] As can be observed from Table 4, as the proportion of recycled material gradually increases from 0% to 30%, the optimal process parameters show a clear trend: the extrusion temperature gradually increases (from 212℃ to 233℃, an increase of 21℃), the draw ratio gradually decreases (from 5.3 to 4.5), and the cooling water temperature gradually decreases (from 22℃ to 16℃).

[0139] The changes in the aforementioned process parameters have a clear technical mechanism: due to molecular chain degradation, recycled materials typically have a higher melt index than virgin materials, resulting in better flowability, but decreased melt strength and tensile properties. Increasing the extrusion temperature can improve the melt uniformity of recycled materials, reducing the draw ratio can reduce the risk of filament breakage during the stretching process, and lowering the cooling water temperature can accelerate melt cooling and shaping, compensating for the decreased dimensional stability caused by insufficient melt strength in recycled materials. Synergistic adjustment of these three factors can achieve optimal carbon emission intensity while ensuring that the flat filament tensile strength (≥0.25 N / tex) and elongation at break (≥15%) meet food-grade standards.

[0140] From a carbon efficiency perspective, although the overall carbon emission intensity of the process using 30% recycled materials (402.3 kgCO2e / t) is slightly higher than that of the virgin material process (356.2 kgCO2e / t), if we consider the implicit carbon emissions from raw material production in range 3 from a life cycle perspective, the overall carbon reduction benefit of 30% recycled materials is significant—the implicit carbon emissions from raw material production of recycled materials are only about 25% of those of virgin materials (about 450 kgCO2e / t vs 1800 kgCO2e / t), and the overall life cycle carbon emissions can be reduced by about 18%.

[0141] Step S4: Closed-loop management and dynamic iterative optimization of the entire process

[0142] The closed-loop mechanism is the same as in Example 1. This example focuses on demonstrating the dynamic iterative optimization effect in the scenario of batch switching of recycled materials.

[0143] During a production run, the company adjusted the proportion of recycled materials from 15% to 25% (due to ample supply and a significant price advantage). The system detected a change in the raw material batch (the melt flow index of the new batch of recycled materials changed by approximately 15% compared to the original batch, exceeding the preset 10% batch-to-batch difference threshold), triggering closed-loop iterative optimization.

[0144] After the system updates the model and resolves, the process parameters are automatically adjusted to those in Table 4. The corresponding optimal values ​​are: extrusion temperature increased from 218℃ to 226℃, draw ratio decreased from 5.1 to 4.8, cooling water temperature decreased from 20℃ to 18℃, and circular loom speed decreased from 156 rpm to 150 rpm. After the adjustments, the product qualification rate remained at 98.2% (meeting the enterprise's internal control standard of ≥98%), and the overall carbon emission intensity was 385.7 kgCO2e / t, compared to the unoptimized state (maintaining...). The carbon emission intensity (when operating under the specified process parameters) was 414.6 kgCO2e / t, a reduction of 7.0%.

[0145] During the above optimization process, the changes in the proportion of recycled materials resulted in... The changes are significant. As described in Example 1... The calculation model shows that when the recycled material ratio is increased from 15% to 25%, if the original process parameters remain unchanged (extrusion temperature 218℃, draw ratio 5.1), due to changes in the fluidity and melt strength of the recycled material, the circular weaving breakage rate increases from 1.3 times / thousand shuttles to 2.4 times / thousand shuttles. The CO2e content increased from 0.27 kgCO2e / thousand shuttles to 0.54 kgCO2e / thousand shuttles, more than doubling. After dynamically optimizing the process parameters using this method (extrusion temperature increased to 226℃, draw ratio decreased to 4.8, and cooling water temperature decreased to 18℃), the circular weaving breakage rate decreased to 1.4 times / thousand shuttles. The CO2e level was reduced to 0.31 kgCO2e / kiloshut, an increase of only 15%, which effectively suppressed the increase in carbon emissions from inter-process coupling caused by the increase in the proportion of recycled materials.

[0146] This embodiment illustrates the adaptive optimization capability of the present invention for different recycled material ratio scenarios. By incorporating the recycled material ratio into the input variable set and combining it with quality constraints for multi-objective optimization, the invention can provide enterprises with a quantitative decision-making basis for adjusting process parameters under the carbon reduction goals of the circular economy, achieving optimal carbon efficiency while ensuring product quality.

[0147] Example 3

[0148] This embodiment demonstrates the closed-loop dynamic iterative optimization capability of the present invention under operating condition switching scenarios. Plastic woven bag manufacturers frequently face changes in operating conditions such as product specification switching and equipment restarts after malfunctions in actual operation. These changes can cause the original optimization model to become mismatched with the current production state, requiring automatic updates and re-optimization of the model.

[0149] The production line configuration is the same as in Example 1, taking the production of general woven bags as an example.

[0150] Scenario 1: Closed-loop iteration triggered by product specification switching

[0151] One day, the company received a new order that required the company to switch the general woven bag specifications from type A (width 550 mm, weft density 40 threads / 100 mm) to type B (width 600 mm, weft density 48 threads / 100 mm), and at the same time switch the product printing colors from single color to two-color printing.

[0152] The system detects changes in order specifications and product quality requirements and automatically triggers closed-loop iterative optimization:

[0153] First, the system updates the product parameters (width 600 mm, weft density 48 threads / 100 mm, two-color printing) of the new specification type B to the constraint set. Specifically, this includes updating the width constraint to 600 mm ± 3 mm (originally 550 mm ± 3 mm), the weft density constraint to 48 ± 1 threads / 100 mm (originally 40 ± 1 threads / 100 mm), and adding a registration accuracy requirement (registration deviation ≤ 0.3 mm) to the printing quality constraint.

[0154] Secondly, the nonlinear coupling model is incrementally updated. Since product type B is a regular product for the company, the production history database already contains approximately 280 sets of valid historical production data for product type B. The system merges this historical data with recent data for product type A (approximately 150 sets, reflecting the current state of the equipment) as a training dataset. This dataset is then used to rapidly incrementally train the BP neural network model (fine-tuning the model weights using transfer learning based on the previous model weights; approximately 80 training epochs are sufficient to reach the target), forming a nonlinear coupling model adapted to product type B.

[0155] After the model meets the requirements, the multi-objective optimization solution in step S3 is executed again. The enterprise's priority weights are adjusted according to the delivery date and profit margin of the order as follows: maintain production capacity 35%, reduce carbon emissions 30%, control costs 25%, and improve quality 10%. After solving using the NSGA-II algorithm and sorting using the entropy weight method-TOPSIS, the optimal combination of process parameters adapted to type B products is output and compared with the optimal parameters of type A products, as shown in Table 5.

[0156] Table 5: Comparison of Optimal Process Parameters Before and After Product Specification Switching

[0157] Extrusion temperature (°C) 208 215 +3.4% stretch ratio 5.2 5.6 +7.7% Cooling water temperature (°C) 22 24 +9.1% Circular loom speed (rpm) 158 148 -6.3% Weft density (threads / 100mm) 40 48 +20.0% Roofing angle (°) 30 28 -6.7% Printing speed (strips / hour) 3200 2600 -18.8% Drying temperature (°C) 62 68 +9.7% <![CDATA[Carbon emission intensity (kgCO2e / t)]]> 342.8 378.5 +10.4%

[0158] As shown in Table 5, after the specification switch, the overall carbon emission intensity increased from 342.8 kgCO2e / t to 378.5 kgCO2e / t, while the carbon efficiency decreased slightly year-on-year. This is mainly due to the increased weft density of the Type B product (from 40 threads / 100 mm to 48 threads / 100 mm), requiring a corresponding reduction in the circular loom speed, extending the weaving time per unit product, and consequently increasing the carbon emissions per unit product. Through the optimization solution of this method, the combination of process parameters has reached the optimal carbon efficiency result under the constraints of the new specification, ensuring that the product switching losses (carbon efficiency losses caused by the production change process and readjustment) after the specification switch are minimized.

[0159] Scenario 2: Closed-loop iteration triggered by output imbalance between upstream and downstream processes

[0160] During the same production batch, the circular weaving process experienced a change in operating status after equipment maintenance (the loom's operating status improved after maintenance, with the weft density qualification rate increasing from 98.5% to 99.8%, but the corresponding loom speed adjustment range changed), resulting in a continuous decrease in actual output from 150 kg / h to 142 kg / h. Meanwhile, the drawing process continued to operate normally at a rate of 150 kg / h, causing the output of the preceding process (drawing) to consistently exceed that of the following process (circular weaving).

[0161] During continuous monitoring, the system detected that the actual output of the downstream process (circular weaving) (142 kg / h) was consistently lower than that of the upstream process (filament drawing) (150 kg / h) for a non-faulty reason, and this state persisted for more than 30 minutes (a preset duration threshold). The system determined that the closed-loop trigger condition was met and automatically performed the following operations:

[0162] First, an analysis of the semi-finished product buffer inventory was conducted. The flat yarn semi-finished products produced in the drawing process are stored in an intermediate buffer rack before the circular weaving process. Currently, the buffer rack occupancy rate has increased from the normal operating rate of 55% to 82%, exceeding the safe occupancy threshold of 80%. If this situation continues, the buffer rack is expected to reach full capacity in 45 minutes, forcing the drawing process to shut down, resulting in greater energy waste and carbon emissions.

[0163] Secondly, the system triggered a model update and re-optimization. The current actual output rate of the circular loom (142 kg / h) and equipment status parameters were incorporated into the constraint set, and the nonlinear coupled model was updated and optimized. The optimization results showed that, under this operating condition, the extrusion speed of the drawing process should be reduced from 150 kg / h to 144 kg / h to make the capacity matching between drawing and circular weaving more reasonable. Simultaneously, the circular loom speed was reduced from 158 rpm to 150 rpm (to adapt to the actual improvement in shed running accuracy after maintenance), and the draw ratio and cooling water temperature were fine-tuned to adapt to the new drawing speed. After the adjustment, the overall carbon emission intensity was 347.3 kgCO2e / t, a 1.2% reduction compared to the previous 351.6 kgCO2e / t. Although the adjustment was limited, from the perspective of continuous production line operation, it avoided the large amount of additional carbon emissions caused by the drawing machine stopping and restarting due to full load of the buffer rack (a single stop and restart is expected to generate approximately 85 kgCO2e of additional carbon emissions).

[0164] This embodiment illustrates the significant advantages of the "model reconstruction type" closed-loop mechanism of the present invention compared with the prior art which only performs periodic parameter fine-tuning based on a fixed model: when the production conditions change substantially, it does not only perform parameter fine-tuning, but automatically updates the entire optimization model and solves it again, ensuring that the carbon efficiency optimization strategy always maintains the best match with the current actual production state.

[0165] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production, characterized in that, Includes the following steps: S1. Identification of dynamic carbon emission accounting and process coupling relationship throughout the entire process: Define the accounting boundaries for the entire process of woven plastic bag production, and construct a comprehensive carbon emission accounting model covering all processes; the formula for calculating the total carbon emissions of the entire process using the comprehensive carbon emission accounting model is as follows: ； In the formula, This represents the total carbon emissions throughout the entire process. The total number of processes. For the first Each process is under equipment load rate Below, the dynamic carbon emission factor of qualified products from production units. For the first The yield of qualified products in each process. The carbon emissions resulting from inter-process coupling caused by fluctuations in process parameters of the preceding process, leading to an increase in the defect rate and waste amount in the subsequent process. Wherein, the dynamic emission factor Calculate using the following formula: ； In the formula, For the first The baseline emission factor for each process when operating at full load. For the first Equipment load rate of each process For the first The no-load emission coefficient of each process, For the first Load rate of each process - emission sensitivity index, and ; Based on historical big data of production, correlation analysis is used to quantify the correlation coefficients between key process parameters of each process and carbon emissions and product quality indicators of upstream and downstream processes. The absolute value of the correlation coefficient is the parameter coupling degree. The process coupling level is divided according to the magnitude of the parameter coupling degree. Based on the process coupling level, the core carbon emission reduction nodes of the whole process are identified. The core carbon emission reduction nodes are used as the core objects of subsequent collaborative optimization. S2, Full-process, multi-dimensional, multi-objective coupled modeling: Using the core controllable process parameters of each process as the input variable set, and taking the minimum carbon emission intensity per unit of qualified product as the core optimization objective, the product qualification rate, production line capacity, and unit product production cost are simultaneously coupled as auxiliary optimization objectives to construct an optimization objective set. A constraint condition set is constructed by combining production process constraints, product quality constraints, safety production constraints, and environmental emission constraints. A nonlinear coupling model between the input variables and each optimization objective is constructed based on production test data and historical production big data. The formula for calculating the carbon emission intensity per unit of qualified product is as follows: ,in, Carbon emission intensity per unit of qualified product The output of qualified products; S3. Multi-objective collaborative optimization solution: A multi-objective intelligent optimization algorithm is used to solve the nonlinear coupled model, and the Pareto optimal solution set that satisfies all constraints is output. Based on the enterprise's production needs, the Pareto optimal solution set is comprehensively sorted, the optimal compromise solution is selected, and the optimal combination of process parameters for each process is output. S4. Closed-loop management and dynamic iterative optimization throughout the entire process: The optimal combination of process parameters is distributed to each process equipment for execution. Real-time collection of full-process production operation data, carbon emission data, quality data, capacity data, and cost data is conducted. When the production conditions change in accordance with preset rules, the nonlinear coupling model is automatically updated and steps S2 to S4 are re-executed to achieve closed-loop control and dynamic iterative optimization of carbon efficiency.

2. The method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production according to claim 1, characterized in that, In step S1, the correlation analysis method is the Pearson correlation coefficient method; the process coupling level includes first-level strong coupling, the parameter coupling degree of the first-level strong coupling is ≥0.75, and the process group corresponding to the first-level strong coupling includes at least melt extrusion drawing process → cooling and shaping process → circular weaving process.

3. The method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production according to claim 2, characterized in that, The method further includes: constructing a separate parameter linkage sub-model for the first-level strongly coupled process group, wherein the parameter linkage sub-model quantifies the coupling relationship between extrusion temperature, draw ratio, cooling water temperature and circular weaving breakage rate and defect rate, thereby realizing parameter linkage carbon efficiency optimization of the core process.

4. The method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production according to claim 1, characterized in that, In step S1, the dynamic emission factor In the middle, the no-load emission coefficient The value ranges from 0.2 to 0.5, and the load factor-emissions sensitivity index is... The value range is from 1.0 to 1.

5.

5. The method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production according to claim 1, characterized in that, In step S2, the nonlinear coupling model is constructed using Box-Behnken response surface methodology combined with a backpropagation (BP) neural network. The construction steps include: Multi-factor, multi-level orthogonal experiments were designed for the core coupled process group to obtain effective experimental data, which were then combined with historical production big data to form a model training dataset. A BP neural network prediction model with three hidden layers is constructed, with the input variable set as the input layer and the optimization target set as the output layer. The ReLU activation function and Adam optimization algorithm are used to complete the model training. When the model predicts the coefficient of determination The model is deemed to meet the standards when the carbon emission intensity prediction error is ≤3% and the product qualification rate prediction error is ≤1%.

6. The method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production according to claim 1, characterized in that, In step S3, the multi-objective intelligent optimization algorithm is an improved NSGA-II algorithm. The improvements include: introducing adaptive crossover and mutation operators to dynamically adjust the crossover and mutation probabilities based on the population evolution generation and individual fitness values; introducing a crowding degree adaptive adjustment strategy; and adding a constraint condition gradient penalty function.

7. The method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production according to claim 1, characterized in that, In step S3, the Pareto optimal solution set is comprehensively sorted based on the enterprise's production needs. The entropy weight method-TOPSIS comprehensive evaluation method is used to screen the optimal compromise solution according to the enterprise's preset priority weights for carbon reduction, production capacity maintenance, cost control, and quality improvement.

8. The method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production according to claim 1, characterized in that, In step S4, the preset rules for triggering dynamic iterative optimization include satisfying any of the following conditions: changes in raw material batch or raw material performance; Changes in order specifications or product quality requirements; changes in equipment operating status; deviations between actual carbon emissions and model predictions exceeding a preset threshold; when it is detected that the actual output of a downstream process is consistently lower than that of a upstream process due to non-fault reasons, and the duration exceeds a preset time.

9. The method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production according to claim 1, characterized in that, In step S2, the carbon cost factor of the carbon market is also included in the optimization target set, and the carbon trading price is also included in the calculation of carbon emission intensity. Furthermore, for different product types such as general woven bags, food-grade woven bags, container bags, and flood control bags, a differentiated constraint and weight system is constructed to output the optimal combination of process parameters suitable for the corresponding product types.

10. The method for synergistic optimization of carbon efficiency in the entire process of woven plastic bag production according to claim 1, characterized in that, The spatial boundary of the full-process accounting boundary covers the entire production chain from raw material warehousing to finished product qualification warehousing, including main processes and supporting auxiliary processes; the main processes include raw material proportioning and mixing, melt extrusion and drawing, cooling and shaping, circular weaving into fabric, coating / lamination, printing, cutting and sewing, online waste recycling and granulation; the supporting auxiliary processes include cooling water circulation system, waste heat recovery system, and compressed air power system; the full-caliber carbon emission accounting model covers scope 1, scope 2, and scope 3 of greenhouse gas accounting.

Images