A life cycle-based polylactic acid green evaluation method and system

Abstract

The application discloses a kind of based on life cycle's polylactic acid green evaluation method and system, including the monitoring data and working data of the preset polylactic acid in life cycle are collected, carbon emission data is obtained by carbon emission deduction to the monitoring data based on emission factor, dynamic genetic self-correction is introduced to the carbon emission data is dynamically corrected, according to the working data, life cycle is divided into multiple decision units, the carbon emission data is as the label of the decision unit;According to the decision unit and the label, input-output correlation analysis is carried out to obtain cross-period dependent data, the production process of the decision unit is decomposed into multiple sub-stages, based on dynamic network, according to the cross-period dependent data and the sub-stage, construct carbon emission efficiency green evaluation function;According to the carbon emission efficiency green evaluation function, construct life cycle polylactic acid green evaluation model, input the data to be evaluated into the life cycle polylactic acid green evaluation model, and output evaluation result.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of green evaluation, and in particular to a polylactic acid green evaluation method and system based on a life cycle. BACKGROUND

[0002] With the increasing global attention to sustainable development, polylactic acid (PLA) as a biobased degradable material has shown great potential in replacing traditional petrochemical plastics and reducing environmental pollution. However, the "green attribute" of PLA is not absolute. Its life cycle covers raw material planting, production and processing, use, and waste disposal, and there are significant differences in resource consumption, pollutant emissions, and energy utilization efficiency at each link. Optimization of a single link cannot fully reflect the overall green performance. Traditional evaluation methods focus on a certain stage or a single indicator of PLA, making it difficult to systematically consider the dynamic changes and cross-stage correlations in the whole life cycle, resulting in incomplete and inaccurate evaluation results.

[0003] In recent years, although the life cycle assessment (LCA) theory has been widely applied to material sustainability evaluation, it still faces many challenges in practical application. On the one hand, the data of each stage of the PLA life cycle has significant time series and complexity, and the traditional LCA model is difficult to capture the cross-period dependence between data. On the other hand, carbon emissions as a core indicator to measure green degree are affected by fluctuations in raw materials, process improvements, environmental factors, etc. The traditional static evaluation method cannot adapt to its dynamic change characteristics. In addition, existing evaluation models rely on fixed parameters and preset weights, lack of adaptive correction ability for data uncertainty and environmental dynamics, resulting in evaluation results lagging behind the actual situation, making it difficult to meet the needs of rapid iteration and upgrading of the industry.

[0004] Therefore, there is an urgent need for a method that can integrate time series data analysis, dynamic parameter correction, and cross-stage correlation evaluation to scientifically and comprehensively quantify the green performance of the PLA life cycle, and provide reliable basis for industry optimization and policy making. SUMMARY

[0005] The purpose of the present application is to provide a polylactic acid green evaluation method based on a life cycle.

[0006] To achieve the above-mentioned purpose, the present application is implemented according to the following technical solutions:

[0007] The present application comprises the following steps:

[0008] Collect monitoring data and working data of pre-set polylactic acid in a life cycle, and pre-process the monitoring data and the working data; the life cycle comprises raw material planting, lactic acid production, PLA polymerization, product manufacturing, circulation use, waste gas treatment and recycling stages; the monitoring data comprises nitrogen fertilizer consumption, pesticide consumption, agricultural machinery diesel consumption, irrigation consumption, crop yield, fermenter power consumption, steam consumption, wastewater COD concentration, polymerization reactor heat energy consumption, solvent recovery rate, catalyst consumption, injection molding machine power, running time, waste material generation amount, transportation distance, load, packaging material weight, compost plant PLA degradation rate, chemical depolymerization energy consumption and recycled material performance retention rate;

[0009] Deduce carbon emission data from the monitoring data based on emission factors, dynamically correct the carbon emission data by introducing dynamic genetic self-correction, divide the life cycle into multiple decision units according to the working data, and take the carbon emission data as labels of the decision units;

[0010] Perform input-output correlation analysis according to the decision units and the labels to obtain cross-period dependence data, decompose the production process of the decision units into multiple sub-stages, and construct a carbon emission efficiency green evaluation function based on a dynamic network according to the cross-period dependence data and the sub-stages;

[0011] Construct a life cycle polylactic acid green evaluation model according to the carbon emission efficiency green evaluation function, input to-be-evaluated data into the life cycle polylactic acid green evaluation model, and output evaluation results.

[0012] Further, the method for deducing carbon emission data from the monitoring data based on emission factors comprises:

[0013] Obtain emission factors of each life cycle, and calculate carbon emission of the raw material planting stage:

[0014] ,

[0015] Wherein, the nitrous oxide emission factor is , the conversion coefficient is , the global warming potential of nitrous oxide is converted into carbon dioxide equivalent, the nitrogen fertilizer consumption is , the irrigation diesel consumption volume is , the carbon dioxide emission amount generated by burning one liter of diesel is , the crop yield is , the average carbon content ratio in crops is , the carbon conversion coefficient into carbon dioxide is , and the carbon emission of the raw material planting stage is ;

[0016] Carbon emissions of the lactic acid production stage are calculated as:

[0017] ,

[0018] where the power consumption of the fermenter is , the grid emission factor is , the amount of steam used is , the amount of carbon dioxide emitted per kilogram of steam produced is , the amount of chemical oxygen demand removed in wastewater treatment is , the amount of methane emitted per kilogram of chemical oxygen demand is , the global warming potential of methane is , and the carbon emissions of the lactic acid production stage are ;

[0019] Carbon emissions of the PLA polymerization stage are calculated as:

[0020] ,

[0021] where the carbon emissions of the PLA polymerization stage are , the heat energy consumption of the polymerization reactor is , the emission factor of the fuel is , the amount of unrecovered solvent is , and the amount of carbon dioxide emitted per kilogram of solvent lost is ;

[0022] Carbon emissions of the product manufacturing stage are calculated as:

[0023] ,

[0024] where the carbon emissions of the product manufacturing stage are , the power of the injection molding machine is P, the operating time is s, the grid emission factor is , the amount of waste produced is , and the amount of carbon dioxide emitted per kilogram of waste incinerated is ;

[0025] Carbon emissions of the circulation use stage are calculated as:

[0026] ,

[0027] where the carbon emissions of the circulation use stage are , the transportation distance is d, the transportation emission factor is , the load is , the weight of the packaging material is , and the amount of carbon dioxide emitted per kilogram of packaging material is ;

[0028] Carbon emissions of waste gas treatment stage:

[0029] ,

[0030] wherein the degradation amount is , the methane emissions per kilogram of PLA degradation is , the amount of PLA incinerated is , the carbon dioxide emissions per kilogram of PLA incinerated is , the energy recovery deduction is , the energy recovered during the incineration process is converted into carbon dioxide emissions deduction;

[0031] Carbon emissions of the regeneration cycle stage:

[0032] ,

[0033] wherein the regeneration power consumption is , the regenerated material yield is , the virgin PLA carbon emissions are , and the carbon deduction coefficient is .

[0034] Further, a dynamic genetic self-correction method is introduced to dynamically correct the carbon emission data, comprising:

[0035] Extract the input parameter pool by monitoring data; input parameters include heat transfer coefficient, catalyst activity coefficient, solvent recovery rate, and polymerization time;

[0036] Using a global sensitivity screening method, calculate the sensitivity index of each parameter on carbon emissions:

[0037] ,

[0038] ,

[0039] wherein the i-th sensitivity index is , the parameter perturbation step is , the i-th input parameter is , the carbon emissions caused by parameter changes are , the number of randomly sampled paths is , the i-th basic effect value of the z-th sampling path is , and the standard value of the i-th basic effect is ;

[0040] Input parameters with a sensitivity index greater than 0.472 are considered sensitive parameters, and sensitive parameters enter the genetic algorithm optimization. Initialize the population according to the sensitive parameters, and the individual represents the potential solution corresponding to the adjusted sensitive parameters;

[0041] Given fitness, the expression is:

[0042] ,

[0043] Where the fitness function is , the vth predicted carbon emission is , the vth measured carbon emission is ;

[0044] Iterate until the fitness function value reaches the maximum, output the adjusted sensitivity coefficient, and dynamically adjust the self-corrected adjusted sensitivity coefficient, the expression is:

[0045] ,

[0046] Where the genetic optimized weight coefficient is , the original carbon emission is , the adjusted ith sensitivity coefficient is , and the corrected carbon emission is .

[0047] Further, the method of dividing the life cycle into multiple decision units according to the working data, comprising:

[0048] The initial unit is obtained by multi-dimensional initial division of the life cycle, including process dimension, time dimension and carbon source dimension; the process dimension is divided according to the physical equipment boundary; the time dimension is segmented according to the operation stage for intermittent production; the carbon source dimension distinguishes direct emission and indirect emission;

[0049] The related parameter sensitivity variance of the aggregation unit is obtained from the monitoring data, including catalyst concentration and stirring speed, and the difference degree of the related parameters is calculated:

[0050] ,

[0051] Where the sensitivity variance of the ath related parameter is , the difference degree of the ath related parameter is , the maximum value of the sensitivity variance is , and the minimum value of the sensitivity variance is ;

[0052] When the difference degree is greater than the determination threshold, the related parameter contribution unit carbon emission is calculated according to the working data, and the sensitivity variance proportion of the related parameter is calculated according to the carbon emission;

[0053] The string function is used to split the elements in the unit list according to the catalyst concentration in the aggregation unit, to obtain a first new unit and a second new unit; the first new unit is the catalyst part in the polymerization reaction, and the second new unit is the mixed part in the polymerization reaction;

[0054] If the catalyst concentration of the split sub-unit exceeds the concentration threshold, the stirring speed is multiplied by 1.02, otherwise the split sub-unit is combined into the adjacent unit; the final result is taken as the decision unit.

[0055] Further, a method for input-output correlation analysis according to the decision unit and the label to obtain cross-period dependence data, comprising:

[0056] Based on the dynamic Bayesian network, a dynamic causal relationship model is established to describe the dependence relationship between decision units at different time points, and the same specification time series is used to capture the influence of upstream units on the current unit;

[0057] The carbon emissions of the upstream unit are used as input by using the recursive method, which is gradually transmitted to the downstream unit to simulate the actual flow process, and is converted into a dynamic recursive relationship, and the expression is:

[0058] ,

[0059] Wherein the carbon emission recursive relationship function is The carbon emission of the i-th decision unit at the t-th time is The carbon emission of the j-th decision unit at the t-1 time is The degree dependence coefficient is The dependence degree of the r-th decision unit on the j-th decision unit, the constant term is The basic carbon emission of the i-th decision unit without the influence of upstream units;

[0060] The time series network analysis technology is used to model the dependence relationship at different time points in different life stages, the dependence strength change is analyzed by time slicing, and the Kalman filter and long-short term neural network model are used to dynamically predict the carbon emission change of each stage to capture the cross-period dependence and trend;

[0061] The time is added to expand the input-output matrix into a 3D matrix The dependence degree of the i-th decision unit on the j-th decision unit at the t time; the dynamic adjustment weight is used to calculate the cross-period cumulative dependence matrix:

[0062] ,

[0063] Wherein the discount factor is The total number of time periods is The cross-period cumulative dependence matrix of the r-th decision unit is ;

[0064] According to the cross-period cumulative dependency matrix, a dependency value is generated, and the dependency value is output as cross-period dependency data.

[0065] Further, a method for constructing a carbon emission efficiency green evaluation function based on a dynamic network according to the cross-period dependency data and the sub-phases, comprises:

[0066] Taking the sub-phases as nodes and the cross-period dependency data as edges, the attribute of the nodes is carbon emission, and the cross-period cumulative carbon emission is calculated:

[0067] ,

[0068] Wherein the cross-period cumulative carbon emission of the bth sub-phase is , the carbon emission of the kth sub-phase is , and the cross-period cumulative dependency data of the bth sub-phase to the kth sub-phase is ;

[0069] The objective weighted unit output carbon emission efficiency is calculated:

[0070] ,

[0071] Wherein the unit output carbon emission efficiency is , the performance index of the bth sub-phase is , the performance index includes efficiency, total energy consumption and water footprint, the importance weight of the bth sub-phase is ;

[0072] The carbon emission efficiency green evaluation function is constructed, and the expression is:

[0073] ;

[0074] Wherein the efficiency index is , the total energy consumption is , the water footprint index is , the total consumption of water resources is , the energy consumption index is , the efficiency weight is , the energy consumption weight is , the water footprint weight is , and the carbon emission efficiency green evaluation function is .

[0075] Further, a method for constructing a life cycle polylactic acid green evaluation model according to the carbon emission efficiency green evaluation function, comprises:

[0076] The objective function is an objective weighting sum of the carbon emission efficiency green evaluation function and the loss function, the life cycle polylactic acid green evaluation model comprises a self-encoder, a time attention mechanism and a recurrent neural network algorithm;

[0077] The self-encoder compresses input data into a low-dimensional hidden layer feature vector through an encoder, and reconstructs the input through a decoder, so that the model automatically extracts green features of the data core in the compression process, with the objective of minimizing reconstruction error;

[0078] The time attention mechanism focuses on the environmental impact features of key time periods, analyzes the time sequence correlation of green performance, and obtains time sequence green feature characteristics by giving dynamic weights of green features of the life cycle polylactic acid at different time sequences;

[0079] The recurrent neural network algorithm captures the dependency relationship of the time sequence green feature characteristics in the life cycle through the recurrent connection of the hidden layer neurons, optimizes the network parameters based on the minimization of the evaluation error of the objective function, and dynamically models and comprehensively evaluates the green performance of each stage.

[0080] In a second aspect, a life cycle-based polylactic acid green evaluation system comprises:

[0081] A data acquisition module is configured to acquire monitoring data and working data of a predetermined polylactic acid in a life cycle, and to preprocess the monitoring data and the working data; the life cycle comprises raw material planting, lactic acid production, PLA polymerization, product manufacturing, circulation use, waste gas treatment and regeneration cycle stages;

[0082] A deduction correction module is configured to deduce carbon emissions based on emission factors to obtain carbon emission data, introduce dynamic genetic self-correction to dynamically correct the carbon emission data, and divide the life cycle into multiple decision units according to the working data, and use the carbon emission data as labels of the decision units;

[0083] An input-output correlation efficiency module is configured to perform input-output correlation analysis according to the decision units and the labels to obtain cross-period dependency data, decompose the production process of the decision units into multiple sub-stages, and construct a carbon emission efficiency green evaluation function based on dynamic networks according to the cross-period dependency data and the sub-stages;

[0084] A modeling output module is configured to construct a life cycle polylactic acid green evaluation model according to the carbon emission efficiency green evaluation function, input to-be-evaluated data into the life cycle polylactic acid green evaluation model, and output an evaluation result.

[0085] The present application has the following advantages:

[0086] The present application is a kind of based on life cycle's polylactic acid green evaluation method and system, compared with prior art, the present application has the following technical effects:

[0087] The present application can improve the accuracy of life cycle's polylactic acid green evaluation, thereby improving the precision of life cycle's polylactic acid green evaluation, optimize life cycle's polylactic acid green evaluation, greatly save resources and improve work efficiency, realize scientific evaluation of life cycle's polylactic acid green, real-time life cycle's polylactic acid green evaluation dynamic correction and input-output correlation analysis, which is of great significance to life cycle's polylactic acid green evaluation, can adapt to different standards of life cycle's polylactic acid green evaluation, different life cycle's polylactic acid green evaluation needs, and has certain universality. BRIEF DESCRIPTION OF DRAWINGS

[0088] Figure 1 The present application is a kind of based on life cycle's polylactic acid green evaluation method and system, compared with prior art, the present application has the following technical effects: DETAILED DESCRIPTION

[0089] The present application will be further described below through specific examples, the illustrative examples of the present application and the description are used to explain the present application, but not as a limitation of the present application.

[0090] The present application is a kind of based on life cycle's polylactic acid green evaluation method and system, compared with prior art, the present application has the following technical effects:

[0091] As shown in Figure 1 In the present embodiment, the following steps are included:

[0092] Collect monitoring data and working data of polylactic acid in life cycle, and pretreat the monitoring data and the working data;The life cycle includes raw material planting, lactic acid production, PLA polymerization, product manufacturing, circulation use, waste gas treatment and regeneration cycle stage;The monitoring data includes nitrogen fertilizer consumption, pesticide consumption, farm machinery diesel consumption, irrigation volume, crop yield, fermenter power consumption, steam consumption, wastewater COD concentration, polymerization reactor heat consumption, solvent recovery rate, catalyst consumption, injection molding machine power, running time, waste material production, transportation distance, load, packaging material weight, compost plant PLA degradation rate, chemical depolymerization energy consumption and regenerated material performance retention rate;

[0093] In the actual evaluation, the green evaluation is performed on process route A and process route B of a PLA enterprise, process route A adopts traditional solvent method, and process route B adopts solvent-free method; the solvent recovery rate of process route A is 85%, and the energy consumption of chemical depolymerization is 500 kWh / t; the catalyst consumption of process route B is reduced by 30%, and the performance retention rate of regenerated material is 95%;

[0094] The monitoring data includes nitrogen fertilizer consumption, agricultural machinery diesel consumption, crop yield, fermentation tank power consumption 5000 kWh, steam consumption, wastewater COD removal, polymerization reactor heat consumption, solvent recovery rate, solvent unrecovered amount, injection molding machine power, operation time, and waste material generation amount;

[0095] The carbon emission data is obtained by deducing the carbon emission based on the emission factor on the monitoring data, the carbon emission data is dynamically corrected by introducing dynamic genetic self-correction, the life cycle is divided into multiple decision units according to the working data, and the carbon emission data is taken as the label of the decision unit;

[0096] In the actual evaluation, the carbon emission data of process route A in the raw material planting, lactic acid production, PLA polymerization and product manufacturing stages are 1798.5 kgCO2e, 12001.8 kgCO2e, 1493 kgCO2e and 812 kgCO2e respectively; the carbon emission data of process route B in the raw material planting, lactic acid production, PLA polymerization and product manufacturing stages are 1791.6 kgCO2e, 11987.8 kgCO2e, 1031.3 kgCO2e and 604.1 kgCO2e respectively

[0097] The solvent recovery rate is taken as the sensitivity index, the carbon emission of the PLA polymerization stage of process route A after dynamic correction is 1795.8 kgCO2e; the carbon emission of the PLA polymerization stage of process route B after dynamic correction is 1206.4 kgCO2e;

[0098] The inter-period dependence data is obtained by input-output correlation analysis according to the decision unit and the label, the production process of the decision unit is decomposed into multiple sub-stages, and the carbon emission efficiency green evaluation function is constructed based on the dynamic network according to the inter-period dependence data and the sub-stages;

[0099] In the actual evaluation, the dependence coefficient of PLA polymerization on the upstream unit of process route B is reduced by 30% due to the solvent-free process, the output efficiency of process route A is 0.0585, the output efficiency of process route B is 0.0778, and the carbon emission efficiency green evaluation function values of process route A and process route B are 0.2606 and 0.2473 respectively;

[0100] The carbon emission efficiency green evaluation function is used to construct a life cycle polylactic acid green evaluation model, and the evaluation data is input into the life cycle polylactic acid green evaluation model, and an evaluation result is output.

[0101] In the embodiment, the method for deducing carbon emission data from the monitoring data based on the emission factor comprises:

[0102] Obtain the emission factor of each life cycle, and calculate the carbon emission of the raw material planting stage:

[0103] ,

[0104] The nitrous oxide emission factor is , the conversion coefficient is , the global warming potential of nitrous oxide is converted into carbon dioxide equivalent, the nitrogen fertilizer consumption is , the irrigation diesel oil consumption volume is , the carbon dioxide emission amount generated by burning one liter of diesel oil is , the crop yield is , the average carbon content ratio in the crop is , the carbon conversion coefficient into carbon dioxide is , and the carbon emission of the raw material planting stage is ;

[0105] Calculate the carbon emission of the lactic acid production stage:

[0106] ,

[0107] The fermentation tank power consumption is , the power grid emission factor is , the steam consumption is , the carbon dioxide emission amount generated by one kilogram of steam is , the chemical oxygen demand removal amount in the wastewater treatment is , the methane emission amount generated by one kilogram of chemical oxygen demand is , the global warming potential of methane is , and the carbon emission of the lactic acid production stage is ;

[0108] Calculate the carbon emission of the PLA polymerization stage:

[0109] ,

[0110] The carbon emission of the PLA polymerization stage is , the polymerization reaction kettle heat energy consumption is , the emission factor of the fuel is , and the unrecovered solvent amount is CO2 emissions per kg of solvent loss ;

[0111] Calculate the carbon emissions of the product manufacturing stage:

[0112] ,

[0113] Where the carbon emissions of the product manufacturing stage are , the power of the injection molding machine is P, the running time is s, the grid emission factor is , the waste production is , the CO2 emissions per kg of waste incineration are ;

[0114] Calculate the carbon emissions of the circulation use stage:

[0115] ,

[0116] Where the carbon emissions of the circulation use stage are , the transportation distance is d, the transportation emission factor is , the load is , the packaging material weight is , the CO2 emissions per kg of packaging material are ;

[0117] Calculate the carbon emissions of the waste gas treatment stage:

[0118] ,

[0119] Where the degradation amount is , the methane emissions per kg of PLA degradation are , the amount of PLA incinerated is , the CO2 emissions per kg of PLA incineration are , the energy recovery deduction is , the energy recovered in the incineration process is converted into CO2 emissions deduction;

[0120] Calculate the carbon emissions of the recycling stage:

[0121] ,

[0122] Where the regeneration power consumption is , the regenerated material production is , the virgin PLA carbon emissions are , and the carbon deduction coefficient is .

[0123] In this embodiment, a dynamic genetic self-correction method is introduced to dynamically correct the carbon emission data, which includes:

[0124] The input parameter pool is extracted by monitoring data; the input parameters include heat transfer coefficient, catalyst activity coefficient, solvent recovery rate, and polymerization time;

[0125] The global sensitivity screening method is used to calculate the sensitivity index of each parameter on carbon emission:

[0126] ,

[0127] ,

[0128] The i-th sensitivity index is , the parameter perturbation step is , the i-th input parameter is , the carbon emission caused by parameter variation is , the number of randomly sampled paths is , the i-th basic effect value of the z-th sampling path is , and the standard value of the i-th basic effect is ;

[0129] The input parameters with a sensitivity index greater than 0.472 are regarded as sensitive parameters, and the sensitive parameters enter the genetic algorithm optimization; the population is initialized according to the sensitive parameters, and the individual represents the potential solution corresponding to the adjusted sensitive parameters;

[0130] Given the fitness, the expression is:

[0131] ,

[0132] The fitness function is , the v-th predicted carbon emission is , and the v-th measured carbon emission is ;

[0133] Iterate until the fitness function value reaches the maximum, output the adjusted sensitive coefficient, and perform self-correction dynamic adjustment on the adjusted sensitive coefficient, the expression is:

[0134] ,

[0135] The weight coefficient of genetic optimization is , the original carbon emission is , the adjusted i-th sensitive coefficient is , and the corrected carbon emission is .

[0136] In this embodiment, the method of dividing the life cycle into multiple decision units according to the working data comprises:

[0137] The initial unit is obtained by multi-dimensionally initially dividing the life cycle, and the multi-dimensions include a process dimension, a time dimension, and a carbon source dimension; the process dimension is divided according to the physical equipment boundary; the time dimension is segmented according to the operation stage for batch production; and the carbon source dimension distinguishes direct emission and indirect emission;

[0138] The related parameter sensitivity variance of the aggregation unit is obtained from the monitoring data, the related parameters including catalyst concentration and stirring speed, and the difference degree of the related parameters is calculated:

[0139] ,

[0140] The sensitivity variance of the a-th related parameter is The difference degree of the a-th related parameter is The maximum value of the sensitivity variance is The minimum value of the sensitivity variance is ;

[0141] When the difference degree is greater than the determination threshold, the carbon emission amount of the related parameter contribution unit is calculated according to the working data, and the sensitivity variance proportion of the related parameter is calculated according to the carbon emission amount;

[0142] The string function is used to split the elements in the unit list according to the catalyst concentration in the aggregation unit, to obtain a first new unit and a second new unit; the first new unit is the catalyst part in the polymerization reaction, and the second new unit is the mixed part in the polymerization reaction;

[0143] If the catalyst concentration of the split sub-unit exceeds the concentration threshold, the stirring speed is multiplied by 1.02, otherwise the split sub-unit is merged into the adjacent unit; and the final result is taken as the decision unit.

[0144] In the embodiment, a method for obtaining cross-period dependence data according to the decision unit and the label includes:

[0145] A dynamic causal relationship model is established based on a dynamic Bayesian network, to describe the dependence relationship between decision units at different time points, and a same-specification time sequence is used to capture the influence of an upstream unit on a current unit;

[0146] A recursive method is used to gradually transfer the carbon emission amount of the upstream unit to the downstream unit as input, to simulate the actual flow process, convert it into a dynamic recursive relationship, and the expression is:

[0147] ,

[0148] The carbon emission recursive relationship function is The carbon emission amount of the i-th decision unit at the t-th time point is The carbon emission amount of the j-th decision unit at the t-1-th time point is , the degree-dependent coefficient is , the degree of dependence of the rth decision unit on the jth decision unit, the constant term is , the basic carbon emission of the ith decision unit without the influence of upstream units;

[0149] The time series network analysis technology is used to model the dependence relationship at different time points in different life stages, the dependence strength change is analyzed through time slicing, Kalman filtering, long short-term neural network model is used to dynamically predict the carbon emission change of each stage to capture the inter-period dependence and trend;

[0150] The time input is used to expand the input-output matrix into a 3D matrix , the degree of dependence of the ith decision unit on the jth decision unit at time t; the dynamic adjustment weight is used to calculate the inter-period cumulative dependence matrix:

[0151] ,

[0152] , wherein the discount factor is , the total number of time periods is , the inter-period cumulative dependence matrix of the rth decision unit is ;

[0153] The dependence relationship value is generated according to the inter-period cumulative dependence matrix, and the dependence relationship value is output as inter-period dependence data.

[0154] In this embodiment, the method for constructing a carbon emission efficiency green evaluation function based on a dynamic network according to the inter-period dependence data and the sub-stages, comprising:

[0155] The sub-stages are taken as nodes, the inter-period dependence data are taken as edges, the attributes of the nodes are carbon emissions, and the inter-period cumulative carbon emissions are calculated:

[0156] ,

[0157] , wherein the inter-period cumulative carbon emissions of the bth sub-stage are , the carbon emissions of the kth sub-stage are , and the inter-period cumulative dependence data of the bth sub-stage on the kth sub-stage are ;

[0158] The objective weighted unit output carbon emission efficiency is calculated:

[0159] ,

[0160] , wherein the unit output carbon emission efficiency is , and the performance index of the bth sub-stage is , performance indicators include efficiency, total energy consumption and water footprint, the importance weight of the bth sub-stage ;

[0161] The carbon emission efficiency green evaluation function is constructed, and the expression is:

[0162] ;

[0163] The efficiency indicator is , the total energy consumption is , the water footprint indicator is , the total consumption of water resources is , the energy consumption indicator is , the efficiency weight is , the energy consumption weight is , the water footprint weight is , and the carbon emission efficiency green evaluation function is .

[0164] In this embodiment, the method for constructing the life cycle polylactic acid green evaluation model according to the carbon emission efficiency green evaluation function comprises:

[0165] The objective weighted sum of the carbon emission efficiency green evaluation function and the loss function is taken as the target function, and the life cycle polylactic acid green evaluation model comprises an autoencoder, a time attention mechanism and a recurrent neural network algorithm;

[0166] The autoencoder compresses the input data into a low-dimensional hidden layer feature vector through an encoder, and reconstructs the input through a decoder, so as to minimize the reconstruction error as the target, so that the model automatically extracts the green features of the data core in the compression process;

[0167] The time attention mechanism focuses on the environmental impact features of the key period by giving the life cycle polylactic acid a dynamic weight of green features at different time sequences, analyzes the time sequence correlation of the green performance, and obtains the time sequence green feature characteristics;

[0168] The recurrent neural network algorithm captures the dependency relationship of the time sequence green feature characteristics in the life cycle through the recurrent connection of the hidden layer neurons, optimizes the network parameters based on the minimum evaluation error of the target function, and dynamically models and comprehensively evaluates the green performance of each stage.

[0169] In a second aspect, a polylactic acid green evaluation system based on a life cycle comprises:

[0170] A data acquisition module is configured to acquire monitoring data and working data of a predetermined polylactic acid in a life cycle, and to preprocess the monitoring data and the working data; the life cycle includes raw material planting, lactic acid production, PLA polymerization, product manufacturing, circulation use, waste gas treatment and regeneration cycle stages;

[0171] Inference correction module: used for carbon emission data obtained by inferring carbon emission based on emission factor on the monitoring data, introducing dynamic genetic self-correction to dynamically correct the carbon emission data, and dividing the life cycle into multiple decision units according to the working data, and taking the carbon emission data as the label of the decision unit;

[0172] Input-output correlation efficiency module: used for input-output correlation analysis according to the decision unit and the label to obtain cross-period dependence data, decomposing the production process of the decision unit into multiple sub-stages, and constructing a carbon emission efficiency green evaluation function based on a dynamic network according to the cross-period dependence data and the sub-stages;

[0173] Modeling output module: used for constructing a life cycle polylactic acid green evaluation model according to the carbon emission efficiency green evaluation function, inputting to-be-evaluated data into the life cycle polylactic acid green evaluation model, and outputting an evaluation result.

[0174] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A life cycle-based polylactic acid green evaluation method characterized by comprising: The method comprises the following steps: Collecting monitoring data and working data of preset polylactic acid in a life cycle, and preprocessing the monitoring data and the working data; the life cycle comprises raw material planting, lactic acid production, PLA polymerization, product manufacturing, circulation use, waste gas treatment and recycling stages; the monitoring data comprises nitrogen fertilizer consumption, pesticide consumption, agricultural machinery diesel oil consumption, irrigation consumption, crop yield, fermenter power consumption, steam consumption, wastewater COD concentration, polymerization reactor heat energy consumption, solvent recovery rate, catalyst consumption, injection molding machine power, running time, waste material production, transportation distance, load, packaging material weight, compost plant PLA degradation rate, chemical depolymerization energy consumption and recycled material performance retention rate; Based on the emission factor, carbon emission data is obtained by deducing the monitoring data, dynamic genetic self-correction is introduced to dynamically correct the carbon emission data, and the life cycle is divided into multiple decision units according to the working data, and the carbon emission data is taken as the label of the decision unit; According to the decision unit and the label, input-output correlation analysis is performed to obtain cross-period dependence data, the production process of the decision unit is divided into multiple sub-stages, and based on the dynamic network, a carbon emission efficiency green evaluation function is constructed according to the cross-period dependence data and the sub-stages; According to the carbon emission efficiency green evaluation function, a life cycle polylactic acid green evaluation model is constructed, and evaluation results are output by inputting to-be-evaluated data into the life cycle polylactic acid green evaluation model; The method for introducing dynamic genetic self-correction to dynamically correct the carbon emission data comprises: Extracting an input parameter pool through monitoring data; the input parameters include heat transfer coefficient, catalyst activity coefficient, solvent recovery rate and polymerization time; Using a global sensitivity screening method to calculate the sensitivity index of each parameter to carbon emission: ； ； wherein the i-th sensitivity index is , the parameter perturbation step is , the i-th input parameter is , the carbon emission caused by the parameter change is , the number of randomly sampled paths is , the i-th basic effect value of the z-th sampled path is , and the standard value of the i-th basic effect is ; Taking the input parameters with a sensitivity index greater than 0.472 as sensitive parameters, and inputting the sensitive parameters into a genetic algorithm optimization; initializing a population according to the sensitive parameters, and taking an individual to represent a potential solution corresponding to the adjusted sensitive parameters; Given a fitness, the expression is: ； where the fitness function is , the vth predicted carbon emission is , the vth measured carbon emission is ; Iterate until the fitness function value reaches the maximum, output the adjusted sensitive coefficient, and perform self-correction dynamic adjustment on the adjusted sensitive coefficient, the expression is: ； wherein the genetically optimized weight coefficient is , the original carbon emission is , the adjusted i-th sensitivity coefficient is , and the corrected carbon emission is .

2. The life cycle-based polylactic acid green evaluation method according to claim 1, characterized by, The method for obtaining carbon emission data based on the emission factor and deducing the monitoring data comprises: Obtaining the emission factor of each life cycle, and calculating the carbon emission of the raw material planting stage: ； wherein the nitrous oxide emission factor is , the conversion factor is , the global warming potential of nitrous oxide is converted to carbon dioxide equivalent, the nitrogen fertilizer use is , the irrigated diesel fuel consumption volume is , the carbon dioxide emission per liter of diesel fuel burned is , the crop yield is , the average proportion of carbon content in the crop is , the carbon conversion to carbon dioxide factor is , and the carbon emission during the raw material planting stage is ; Calculating the carbon emission of the lactic acid production stage: ； wherein the power consumption of the fermenter is , the grid emission factor is , the steam usage is , the carbon dioxide emission per kilogram of steam produced is , the removal of chemical oxygen demand in wastewater treatment is , the methane emission per kilogram of chemical oxygen demand produced is , the global warming potential of methane is , and the carbon footprint of the lactic acid production stage is ; Calculating the carbon emission of the PLA polymerization stage: ； wherein the carbon footprint of the PLA polymerization stage is , the heat energy consumption of the polymerization reactor is , the emission factor of the fuel is , the amount of unrecycled solvent is , and the carbon dioxide emissions per kilogram of solvent lost is ; Calculating the carbon emission of the product manufacturing stage: ； where the carbon footprint of the product manufacturing stage is , the power of the injection molding machine is P, the operating time is s, the grid emission factor is , the amount of waste produced is , and the amount of carbon dioxide emitted per kilogram of waste incinerated is ; Calculating the carbon emission of the circulation use stage: ； wherein the carbon emission of the flow-through use stage is , the transportation distance is d, the transportation emission factor is , the load is , the packaging material weight is , and the carbon dioxide emission per kilogram of packaging material is ; Calculating the carbon emission of the waste gas treatment stage: ； wherein the degradation amount is the methane emission amount per kilogram of PLA produced by degradation is the amount of PLA incinerated is the carbon dioxide emission amount per kilogram of PLA produced by incineration is the energy recovery deduction is the energy recovered during incineration is converted into a carbon dioxide emission deduction; Calculating the carbon emission of the recycling stage: ； wherein the regeneration power consumption is , the regenerated material yield is , the virgin PLA carbon emission is , and the carbon offset coefficient is .

3. The life cycle-based polylactic acid green evaluation method according to claim 1, characterized by, The method for dividing the life cycle into multiple decision units according to the working data comprises: Carrying out multi-dimensional initial division on the life cycle to obtain an initial unit; the multi-dimensions include process dimension, time dimension and carbon source dimension; the process dimension is divided according to the physical equipment boundary; the time dimension is segmented according to the operation stage for intermittent production; and the carbon source dimension distinguishes direct emission and indirect emission; Obtain the variance of the sensitivity of the relevant parameters of the aggregation unit from the monitoring data, the relevant parameters including the catalyst concentration and the stirring speed, and calculate the difference degree of the relevant parameters: ； wherein the sensitivity variance of the a-th correlation parameter is , the difference degree of the a-th correlation parameter is , the maximum value of the sensitivity variance is , and the minimum value of the sensitivity variance is ; When the difference degree is greater than the determination threshold, calculate the carbon emission of the relevant parameter contribution unit according to the working data, and calculate the variance proportion of the sensitivity of the relevant parameter according to the carbon emission; Split the elements in the unit list according to the catalyst concentration in the aggregation unit by using the string function to obtain a first new unit and a second new unit; the first new unit is the catalyst part in the polymerization reaction, and the second new unit is the mixed part in the polymerization reaction; If the catalyst concentration of the split sub-unit exceeds the concentration threshold, the stirring speed is multiplied by 1.02, otherwise the split sub-unit is merged into the adjacent unit; the final result is taken as the decision unit.

4. The life cycle-based polylactic acid green evaluation method according to claim 1, characterized by, A method for input-output correlation analysis according to the decision unit and the label to obtain cross-period dependence data, comprising: Based on the dynamic Bayesian network, a dynamic causal relationship model is established to describe the dependence relationship between decision units at different time points, and the same specification time series is used to capture the influence of upstream units on the current unit; Using a recursive method, the carbon emissions of upstream units are taken as inputs and gradually transmitted to downstream units to simulate the actual flow process, and are converted into a dynamic recursive relationship, which is expressed as: ； wherein the carbon emission recursive relationship function is , the carbon emission of the ith decision unit at the tth time is , the carbon emission of the jth decision unit at the (t-1)th time is , the degree-dependent coefficient is , the degree of dependence of the rth decision unit on the jth decision unit, the constant term is , the basic carbon emission of the ith decision unit without the influence of upstream units; Using time series network analysis technology to model the dependence relationship at different time points in different life stages, analyzing the dependence intensity change through time slicing, and using Kalman filtering and long-short term neural network model to dynamically predict the carbon emission change of each stage to capture the cross-period dependence and trend; The time of joining extends the input-output matrix into a 3D matrix The degree of dependence of the ith decision unit on the jth decision unit at time t; the cross-period cumulative dependence matrix is calculated using dynamic adjustment weights: ； wherein the discount factor is , the total number of time periods is , the inter-period cumulative dependence matrix of the rth decision unit is ; According to the cross-period cumulative dependence matrix, a dependence relationship value is generated, and the dependence relationship value is output as cross-period dependence data.

5. The life cycle-based polylactic acid green evaluation method according to claim 1, characterized by, A method for constructing a carbon emission efficiency green evaluation function based on a dynamic network according to the cross-period dependence data and the sub-stage, comprising: Taking the sub-stage as a node and the cross-period dependence data as an edge, the attribute of the node is the carbon emission, and the cross-period cumulative carbon emission is calculated: ； The cross-period cumulative carbon emission amount of the bth sub-stage is The carbon emission amount of the kth sub-stage is The cross-period cumulative dependency data of the bth sub-stage on the kth sub-stage is ; Calculating the objectively weighted unit output carbon emission efficiency: ； Wherein the unit output carbon emission efficiency is , the performance index of the bth sub-stage is , the performance index includes efficiency, total energy consumption and water footprint, the importance weight of the bth sub-stage is ; Constructing a carbon emission efficiency green evaluation function, which is expressed as: ； The efficiency index is , the total energy consumption is , the water footprint index is , the total consumption of water resources is , the energy consumption index is , the efficiency weight is , the energy consumption weight is , the water footprint weight is , and the carbon emission efficiency green evaluation function is .

6. The life cycle-based polylactic acid green evaluation method according to claim 1, characterized by, A method for constructing a life cycle polylactic acid green evaluation model according to the carbon emission efficiency green evaluation function, comprising: Taking the objective weighted sum of the carbon emission efficiency green evaluation function and the loss function as the objective function, the life cycle polylactic acid green evaluation model includes an autoencoder, a time attention mechanism and a recurrent neural network algorithm; The autoencoder compresses the input data into a low-dimensional hidden feature vector through an encoder, and reconstructs the input through a decoder, with the objective of minimizing the reconstruction error, so that the model automatically extracts the green features of the data core in the compression process; The time attention mechanism focuses on the environmental impact features of the key period by giving the life cycle polylactic acid different time sequence stages dynamic weights, and analyzes the time sequence correlation of the green performance to obtain the time sequence green feature characteristics; The recurrent neural network algorithm captures the dependence relationship of the time sequence green feature characteristics in the life cycle through the recurrent connection of the hidden layer neurons, optimizes the network parameters based on the minimization of the evaluation error of the objective function, and dynamically models and evaluates the green performance of each stage.

7. A life cycle based polylactic acid green evaluation system for performing the method of any one of claims 1 to 6, characterized by, ​ The data acquisition module is used for collecting monitoring data and working data of preset polylactic acid in a life cycle, and pre-processing the monitoring data and the working data; the life cycle includes raw material planting, lactic acid production, PLA polymerization, product manufacturing, circulation use, waste gas treatment and recycling stages. The deduction correction module is used for deducing carbon emission data based on the monitoring data, introducing dynamic genetic self-correction to dynamically correct the carbon emission data, and dividing the life cycle into multiple decision units according to the working data, and taking the carbon emission data as a label of the decision unit. The production correlation efficiency module is used for input-output correlation analysis according to the decision unit and the label to obtain cross-period dependence data, decomposing a production process of the decision unit into multiple sub-stages, and constructing a carbon emission efficiency green evaluation function based on a dynamic network according to the cross-period dependence data and the sub-stages. The modeling output module is used for constructing a life cycle polylactic acid green evaluation model according to the carbon emission efficiency green evaluation function, inputting to-be-evaluated data into the life cycle polylactic acid green evaluation model, and outputting an evaluation result.

Images