Carbon footprint modeling method for green electrical and green certificate equipment based on full life cycle evaluation

Abstract

This invention discloses a carbon footprint modeling method for electrical equipment based on green electricity and green certificates, belonging to the field of power equipment carbon tracking technology. The method includes: Step 1: Building a four-scale model based on a pre-constructed digital twin coupled model of the electrical equipment, and bidirectionally coupling the interaction and nesting of the four-scale model through a bidirectional parameter transfer interface; wherein the four-scale model includes a process-level energy consumption time-varying model, a workshop-level scheduling constraint model, an enterprise-level green electricity and green certificate balance model, and a grid-level dynamic carbon intensity model; Step 2: Based on the bidirectionally coupled four-scale model, acquiring multi-source dynamic data through the Internet of Things, and after preprocessing, using a consortium blockchain to timestamp and store the preprocessed multi-source dynamic data, generating a standardized trusted dataset; wherein the consortium blockchain consists of multiple trusted nodes; Step 3: Based on the standardized trusted dataset, combining pre-acquired green electricity output characteristics, green certificate verification rules, and grid purchase carbon intensity.

Description

Technical Field

[0001] This invention relates to the field of carbon tracking technology for power equipment, specifically a method for modeling the carbon footprint of electrical equipment based on green electricity and green certificates in the context of full life cycle assessment. Background Technology

[0002] With the acceleration and deepening of global climate governance, the green and low-carbon transformation of electrical equipment has become a requirement for sustainable industrial development, as the power system is a key hub for energy transition. As a fundamental component of the power grid, the electrical equipment manufacturing industry not only bears the important function of energy transmission and distribution, but also faces the urgent pressure to transform from the traditional high-energy-consuming and high-emission production model to a green and low-carbon manufacturing system. In the process of transformation, effectively reducing the carbon footprint of products in the production process and improving environmental benefits is one of the core tasks. Promoting the clean energy structure of the production process is an important path to achieve carbon reduction at this stage.

[0003] In existing technologies, carbon footprint assessment and emission reduction control for electrical equipment suffer from severe lag and static defects in carbon footprint accounting. Current mainstream carbon footprint assessment methods typically rely on historical statistical data and industry average emission factors for ex-post calculations. This means that companies often only obtain a carbon footprint report for a product several months after production is completed or even delivered. This ex-post evaluation mechanism struggles to guide real-time decision-making during the production process, preventing companies from promptly identifying and correcting high-energy-consuming and high-emission production behaviors. Furthermore, there is a lack of accurate capture of the dynamic characteristics of green electricity and green certificates. Existing carbon footprint models typically treat electricity emissions as a static constant, but in reality, with the large-scale grid connection of renewable energy, the output of self-built distributed photovoltaic systems exhibits significant fluctuations, and the price of green certificates in the electricity market also changes in real time. Existing technologies cannot correlate the dynamic information of using a specific batch of green electricity on a specific day and time with the carbon footprint of a particular batch of products, resulting in the environmental rights of green electricity being averaged in the calculation, failing to reflect the true emission reduction effect of companies using green electricity. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] Step 1: Based on the pre-built digital twin coupled model of electrical equipment, a four-scale model is built, and the interaction and nesting of the four-scale model are bidirectionally coupled through the bidirectional parameter transfer interface. The four-scale model includes a process-level energy consumption time-varying model, a workshop-level scheduling constraint model, an enterprise-level green electricity and green certificate balance model, and a grid-level dynamic carbon intensity model.

[0006] Step 2: Based on the four-scale model after bidirectional coupling, multi-source dynamic data is acquired through the Internet of Things. After preprocessing, the preprocessed multi-source dynamic data is timestamped and stored using a consortium blockchain to generate a standardized trusted dataset. The consortium blockchain consists of multiple trusted nodes.

[0007] Step 3: Based on a standardized and reliable dataset, and combined with pre-acquired green electricity output characteristics, green certificate verification rules, and grid purchase carbon intensity, construct a real-time carbon intensity quantification model for the manufacturing period, and output time-varying carbon intensity data that is strongly correlated with the production period.

[0008] Step 4: Using time-varying carbon intensity data as constraints, construct a multi-objective optimization model with the objectives of minimizing the carbon footprint of the manufacturing stage, minimizing total production costs, and minimizing delivery delay defaults. Solve the model using an improved dual-population genetic algorithm and output the Pareto optimal production scheduling scheme.

[0009] Furthermore, the process of building a four-scale model is as follows:

[0010] The process-level energy consumption time-varying model distinguishes between the ground-state energy consumption when there is no workpiece processing and the dynamic energy consumption during workpiece processing. It introduces a weather sensitivity coefficient and an energy efficiency feedback coefficient, which directly links the green electricity usage strategy with the energy efficiency of the process.

[0011] The workshop-level scheduling constraint model defines process sequence, equipment capacity, delivery time, and energy consumption by constructing a correlation matrix of processes, equipment, time, and carbon footprint.

[0012] The enterprise-level green electricity and green certificate balance model establishes a dynamic balance logic for green electricity production, consumption, procurement, and verification.

[0013] The grid-level dynamic carbon intensity model outputs time-of-use grid carbon intensity, green electricity zero carbon, and green certificate deduction equivalent carbon intensity, which is obtained by multiplying the carbon intensity by the deduction coefficient.

[0014] Furthermore, the bidirectional parameter transfer interface enables bidirectional coupling of the interaction and nesting of the four-scale model as follows:

[0015] The bidirectional parameter transmission coupling interface enables dynamic nesting and data interaction of four-scale models—process-level energy consumption time-varying model, workshop-level scheduling constraint model, enterprise-level green electricity and green certificate balance model, and grid-level dynamic carbon intensity model—through a downward transmission and upward feedback mechanism.

[0016] Furthermore, the process of acquiring multi-source dynamic data through the Internet of Things is as follows:

[0017] Based on the self-operation logic of the four-scale model after bidirectional coupling, data acquisition tasks are generated, and the acquisition method is matched according to the type of data acquisition task to perform hierarchical and classified data acquisition.

[0018] Furthermore, the process of generating a standardized, trustworthy dataset is as follows:

[0019] The preprocessed multi-source dynamic data is structured and encapsulated to generate a unified format evidence storage data package. Before the data package is submitted, the enterprise node completes the access verification through the identity authentication mechanism of the consortium blockchain. The enterprise node submits the encapsulated evidence storage data package to the consortium blockchain network in the form of an on-chain transaction.

[0020] The consortium blockchain adopts a practical Byzantine fault-tolerant consensus mechanism. After an enterprise node submits a transaction, the verification nodes in the consortium blockchain network verify the transaction. When more than 2 / 3 of the verification nodes pass the verification, the transaction reaches a consensus and is packaged into a candidate block. Otherwise, the consensus of the notarized transaction is not established, and it is directly judged as an invalid transaction and will not be packaged into a candidate block.

[0021] Candidate blocks that have reached a consensus on transactions are written into the distributed ledger of the consortium blockchain. After the notarization is completed, the consortium blockchain network automatically generates a notarization index table, and a standardized trusted dataset is obtained based on the notarization index table.

[0022] Furthermore, the process of outputting time-varying carbon intensity data that is strongly correlated with the production period is as follows:

[0023] Based on the standardized and reliable dataset, the total energy consumption of the workshop, the direct consumption of green electricity in each process, the sales volume of green certificates, and the electricity purchased from the grid are extracted. According to the characteristics of green electricity output and the rules for green certificate verification, the total energy consumption of the workshop in each time period is broken down. Based on the total energy consumption after the breakdown, combined with the carbon intensity of electricity purchased from the grid, the real-time carbon intensity of the manufacturing period in each time period is calculated and dynamically updated every 15 minutes. The real-time carbon intensity of each time period is matched one-to-one with the production period to generate a time-varying carbon intensity dataset.

[0024] Furthermore, the construction process of the multi-objective optimization model is as follows:

[0025] Using time-varying carbon intensity data as constraints, including basic scheduling constraints at the workshop level, time-varying carbon intensity constraints, supplementary green electricity and green certificate constraints, and energy consumption balance constraints, the objectives of carbon footprint, total cost, and default duration with different dimensions are normalized and assigned corresponding weights. The three independent objectives are integrated into a comprehensive optimization objective. The integrated comprehensive optimization objective and all constraints are embedded into the workshop-level scheduling constraint model, linking the process-level energy consumption model, the enterprise-level green electricity and green certificate balance model, and the grid-level dynamic carbon intensity model to form a closed-loop optimization model.

[0026] Furthermore, the process of outputting the Pareto optimal production scheduling scheme is as follows:

[0027] Based on the allocation of processing equipment, selection of processing time periods, green electricity allocation ratio, and number of green certificate cancellations for each process, a segmented coding rule is designed to map each decision variable to a coding segment. The initial population is split into a low-carbon oriented population and a cost-oriented population, and a heuristic method is used for differentiated initialization. For each individual in the two populations, the target value is normalized by the total carbon footprint of the manufacturing stage, the total production cost, and the total duration of delivery default. The comprehensive fitness is calculated by combining the target weights preset by the enterprise. At the same time, all constraints are verified. Individuals that violate the constraints are directly set to extremely low fitness. Dedicated crossover and mutation operators are designed for the characteristics of the two populations.

[0028] All individuals in the low-carbon oriented population and the cost oriented population are merged. Non-dominated solutions with no objective inferior to other solutions are selected through non-dominated sorting. The high-quality non-dominated solutions are then injected back into the two populations to replace the individuals with the lowest fitness in each population. At the same time, the orientation characteristics of each population are preserved. The process is terminated when the number of iterations reaches a preset value.

[0029] The carbon footprint modeling method for green electricity and green certificates electrical equipment based on full life cycle assessment provided by this invention has the following beneficial effects:

[0030] (1) This invention proposes an innovative real-time carbon intensity quantification system for manufacturing periods by constructing a four-scale digital twin coupling model of power grid, enterprise, workshop and process. It deeply couples the dynamic output of green electricity, the rules for green certificate verification with the time-of-use power grid carbon intensity, and completely abandons the traditional calculation mode of static carbon intensity factor for full life cycle evaluation. At the same time, it relies on the consortium blockchain to complete the timestamp storage of multi-source data, ensuring the credibility and immutability of carbon accounting data from the source, greatly improving the accuracy of carbon footprint accounting, and effectively solving the industry pain points of green electricity, green certificates and carbon accounting being separated, difficult to dynamically match and high accounting error in traditional technologies, so that carbon footprint accounting is fully in line with the actual production energy consumption and green electricity consumption scenarios.

[0031] (2) By introducing the manufacturing energy efficiency transfer coefficient, this invention dynamically links the green electricity use strategy and process optimization effect in the manufacturing stage with the carbon footprint of equipment operation loss in the operation and maintenance stage, thereby realizing the coupling superposition and overall optimization of the carbon footprint of the entire life cycle of manufacturing, operation and maintenance and scrapping. This avoids the problem of one-sided emission reduction optimization caused by independent accounting of stages in traditional methods, and can formulate the optimal low-carbon strategy from the perspective of the entire life cycle.

[0032] (3) This invention incorporates carbon footprint management, green certificate cost optimization and production scheduling into a unified multi-objective optimization model. With time-varying carbon intensity as the core constraint, it uses an improved dual-population genetic algorithm to output the Pareto optimal scheduling scheme, thereby achieving a synergistic balance of the three objectives of low carbon, cost reduction and delivery assurance. This solves the pain point of traditional production decisions that only focus on efficiency and cost and ignore carbon management. It maximizes the benefits of green electricity consumption and green certificate deduction without affecting the delivery cycle, and improves the integration of carbon management and production operation of electrical equipment manufacturing enterprises. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the method of the present invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] Please see Figure 1 This application provides a method for modeling the carbon footprint of green electricity and green certificates electrical equipment based on full life cycle assessment. The method includes:

[0036] Step 1: Based on the pre-built digital twin coupled model of electrical equipment, a four-scale model is built, and the interaction and nesting of the four-scale model are bidirectionally coupled through the bidirectional parameter transfer interface. The four-scale model includes a process-level energy consumption time-varying model, a workshop-level scheduling constraint model, an enterprise-level green electricity and green certificate balance model, and a grid-level dynamic carbon intensity model.

[0037] The process of building a four-scale model is as follows:

[0038] Construct a process-level time-varying sub-model of energy consumption:

[0039] For core processes in electrical equipment manufacturing, such as transformer core lamination, winding, insulation casting, assembly, and testing, the basic energy consumption of each process when there is no workpiece processing is first determined through process calibration experiments, such as the basic energy consumption of equipment standby and workshop constant temperature maintenance. Then, the dynamic power and actual processing time during the workpiece processing are collected in real time through the equipment's Internet of Things sensors, forming dual-dimensional data of basic and dynamic energy consumption.

[0040] Furthermore, a weather sensitivity coefficient is introduced. By statistically analyzing process energy consumption data under different ambient temperatures, the influence of temperature on energy consumption is fitted to determine the pattern. For example, in winter, the pouring process requires additional heating and insulation, resulting in a significant increase in energy consumption. This allows process energy consumption to dynamically change with ambient temperature. At the same time, an energy efficiency feedback coefficient is introduced: process energy efficiency is adjusted according to the proportion of green electricity used. The energy efficiency of variable frequency heating and processing equipment driven by green electricity is higher than that of traditional mains-powered equipment. Therefore, the higher the proportion of green electricity used, the lower the overall energy consumption of the process, thus realizing a direct correlation between green electricity use strategy and process energy efficiency.

[0041] Construct a workshop-level scheduling constraint sub-model:

[0042] Based on the process-level energy consumption model, the logical relationship between processes, equipment, time, and carbon footprint is established. The range of processes that each piece of equipment can process, the standard processing time of each process, and the sequence of processes for the same workpiece are clearly defined, such as stacking the iron core before winding the winding, which cannot be reversed. At the same time, equipment capacity constraints are set so that a single piece of equipment can only process one process at a time. Delivery period constraints are set so that the completion time of all processes does not exceed the delivery node agreed by the customer. Energy consumption constraints are set so that the total energy consumption of a single piece of equipment during a time period does not exceed the rated power.

[0043] This sub-model also links the processing time of each step to carbon intensity, allowing scheduling decisions to naturally take into account the differences in carbon emissions from electricity consumption at different times, providing a constraint basis for subsequent carbon footprint optimization.

[0044] Constructing an enterprise-level green electricity and green certificate balancing sub-model:

[0045] First, the source of green electricity for enterprises is determined, including real-time output of their own photovoltaic or wind power. This data is collected through photovoltaic or wind power monitoring systems and dynamically calculated using weather data such as irradiance and wind speed. The amount of green electricity purchased is also obtained from the power trading platform to form the total amount of green electricity that the enterprise can control. Next, a binding rule is established between green electricity consumption and green certificate verification: green electricity actually used in the workshop is prioritized to offset both owned and purchased green electricity. If green electricity is insufficient, carbon emissions must be offset by verifying green certificates. The number of green certificates verified is strictly correlated with the amount of non-green electricity consumed, with 1 green certificate corresponding to 1 MWh of non-green electricity consumption. At the same time, the purchase and holding of green certificates are tracked to ensure that the number of green certificates verified does not exceed the current holding amount, thus avoiding unauthorized verification.

[0046] Constructing a grid-level dynamic carbon intensity sub-model:

[0047] Connecting with the power trading center and the Ministry of Ecology and Environment's time-sharing and regional power carbon intensity data platform, the system acquires dynamic carbon intensity data on grid-purchased electricity every 15 minutes. The carbon intensity fluctuates in real time due to variations in the proportions of thermal power, hydropower, wind power, and solar power generation within the grid. It clarifies that green electricity itself emits no carbon, and the equivalent carbon intensity for green certificate deductions is adjusted based on market acceptance. For example, green certificates recognized by international authoritative institutions can fully deduct carbon emissions; when some regions do not recognize them, the deduction ratio is reduced accordingly, ensuring that carbon intensity data aligns with actual policies and market rules.

[0048] The bidirectional coupling process of interaction and nesting in the four-scale model is as follows:

[0049] Based on the standardized data interaction protocol, real-time response mechanism and hierarchical linkage rules preset by the interface, a closed-loop coupling system of top-down constraint transmission and bottom-up feedback correction is constructed to carry out the interaction and dynamic nesting process of the four-scale model.

[0050] Top-down constraint propagation process:

[0051] The bidirectional parameter transfer interface pushes the core constraint parameters of the upper-level model to the lower-level model in real time according to the hierarchical order of power grid level, enterprise level, workshop level and process level, and completes the standardized parsing of the parameters, so that the lower-level model can use them as the core basis for operation, thus realizing the downward nesting of models.

[0052] During the constraint transfer process from the grid level to the enterprise level, the interface extracts parameters such as time-sharing and zone-based dynamic carbon intensity and green certificate deduction coefficient from the grid-level dynamic carbon intensity model. After format verification, these parameters are pushed to the enterprise-level green electricity and green certificate balance model. The enterprise-level model incorporates these parameters into the decision-making logic of green electricity procurement and green certificate verification, forming a green electricity and green certificate balance rule under the grid carbon intensity constraint.

[0053] During the constraint transfer process from the enterprise level to the workshop level, the interface extracts constraint parameters such as real-time green electricity availability, green certificate holdings, and minimum green electricity consumption ratio from the enterprise-level green electricity and green certificate balance model. After parsing, these parameters are synchronized to the workshop-level scheduling constraint model. The workshop-level model uses these parameters as pre-constraints for scheduling, and nests them to form production scheduling rules under green electricity and green certificate constraints. This ensures that the scheduling scheme adapts to the enterprise's green electricity supply capacity and carbon emission reduction requirements in terms of process arrangement and time period selection.

[0054] During the constraint transfer process from the workshop level to the process level, the interface extracts scheduling constraint parameters such as process processing time, equipment allocation scheme and energy consumption limit from the workshop-level scheduling constraint model and pushes them to the process-level energy consumption time-varying model. The process-level model combines these parameters with its own energy consumption calculation logic to form process energy consumption time-varying rules under scheduling constraints, so that the process energy consumption calculation matches the time period and equipment requirements of the workshop scheduling.

[0055] Bottom-up feedback and correction process:

[0056] The bidirectional parameter transmission interface collects actual operating data and deviation information of the next-level model in the hierarchical order of process level, workshop level, enterprise level and power grid level. After authenticity verification and data aggregation, the data is fed back to the next-level model. The next-level model corrects its own parameters and operating results based on the feedback data, and performs upward interaction between models.

[0057] During the feedback and correction process from the process level to the workshop level, the interface collects data such as actual process energy consumption, actual green electricity consumption, and energy efficiency feedback coefficient from the process-level energy consumption time-varying model in real time. After removing outliers, the data is aggregated into workshop-level energy consumption data and fed back to the workshop-level scheduling constraint model. The workshop-level model compares the deviation between planned energy consumption and actual energy consumption and corrects the scheduling energy consumption constraint parameters. If the actual energy consumption of a certain process exceeds expectations, the processing time or equipment allocation of subsequent processes is immediately adjusted.

[0058] During the feedback and correction process from the workshop level to the enterprise level, the interface aggregates data such as the actual green electricity consumption, the actual green certificate issuance, and the energy consumption demand after scheduling adjustments from the workshop-level model into enterprise-level energy consumption data, which is then fed back to the enterprise-level green electricity and green certificate balance model. Based on the actual consumption data, the enterprise-level model adjusts the green electricity procurement plan and the green certificate procurement quantity. If the actual green electricity consumption exceeds the available quantity, it initiates a decision adjustment to purchase external green electricity or supplement the procurement of green certificates to maintain the dynamic balance of green electricity and green certificates.

[0059] During the feedback correction process from the enterprise level to the grid level, the interface feeds back data such as the actual demand for electricity purchases by the grid, the capacity for green energy absorption, and the carbon intensity adaptation deviation from the enterprise-level model to the grid-level dynamic carbon intensity model. The grid-level model combines this actual demand data from the enterprise side to optimize the prediction accuracy of time-of-use and zone-of-use carbon intensity. If the carbon intensity adaptation deviation is large in a certain period of time on the enterprise side, the power trading platform will be linked to update the carbon intensity data for that period.

[0060] Step 2: Based on the four-scale model after bidirectional coupling, multi-source dynamic data is acquired through the Internet of Things. After preprocessing, the preprocessed multi-source dynamic data is timestamped and stored using a consortium blockchain to generate a standardized trusted dataset. The consortium blockchain consists of multiple trusted nodes.

[0061] The process of acquiring multi-source dynamic data is as follows:

[0062] The four-scale model after multi-scale bidirectional coupling generates a data acquisition task list according to its own operating logic, and sends it to the Internet of Things acquisition terminal in a unified manner through the bidirectional parameter transmission interface, including the physical layer, energy layer, market layer and environment layer.

[0063] The physical layer data acquisition is adapted to process-level and workshop-level models. For process processing and equipment operation data, it collects core indicators through IoT sensors deployed on the production site, such as current transformers, processing time counters, equipment status acquisition modules, and production execution systems. The data is collected based on the processing time of each process, real-time power of the equipment, number of workpieces processed, and equipment operating status. The data collection frequency is once every 15 minutes, which is consistent with the minimum time granularity of the model, so that the data can accurately support the time-varying calculation of process energy consumption and the verification of workshop scheduling constraints.

[0064] The energy layer data acquisition is adapted to enterprise-level and workshop-level models. It collects data on green electricity and grid purchases through smart meters, photovoltaic or wind power monitoring systems, and enterprise energy management systems. It calculates the real-time output of self-owned photovoltaic or wind power by combining inverter data, irradiance or wind speed sensor data, and synchronizes the purchased green electricity volume from the power trading platform interface. The grid purchase volume is statistically analyzed in real time through the main meter, and the actual green electricity consumption is statistically analyzed by workshop and time period. The data collection frequency is once every 15 minutes, which enables the enterprise-level green electricity and green certificate balance model to obtain the dynamics of energy supply and consumption in real time.

[0065] Market-level data collection is adapted to enterprise-level and grid-level models. For green certificate and electricity market data, it obtains the following through the green certificate trading platform API and the electricity trading center interface: time-of-use green certificate trading price, time-of-use grid electricity price, real-time holding and write-off records of green certificates, and electricity market transaction settlement data. The collection frequency is once every 15 minutes to ensure that the grid-level carbon intensity model calculates the equivalent carbon intensity, and the enterprise-level model dynamically adjusts the green certificate procurement and write-off strategies.

[0066] The environmental layer data acquisition is adapted to process-level and power grid-level models. For weather and environmental data, it acquires workshop ambient temperature, photovoltaic irradiance, wind speed and air humidity through weather stations deployed in the workshop, satellite remote sensing data interfaces and third-party weather forecasting platforms. The acquisition frequency is once every 15 minutes, providing real-time input for the weather sensitivity coefficient of the process-level energy consumption time-varying model, while supporting the green power output prediction and correction of the power grid-level carbon intensity model.

[0067] The preprocessing process is as follows:

[0068] Data from different acquisition terminals and sources, such as millisecond-level data from sensors and hourly-level data from trading platforms, are uniformly mapped onto a 15-minute granularity time axis. Data for each time period is aggregated according to the start time of the time period, so that the time dimension of all data is completely consistent, avoiding model calculation deviations caused by time misalignment. For example, the process energy consumption data collected by workshop sensors from 10:02 to 10:17 is aggregated into the time period of 10:00 to 10:15, keeping the time granularity synchronized with the green electricity output and carbon intensity data.

[0069] Missing value imputation employs differentiated imputation strategies for different types of data missing values:

[0070] If environmental data, such as temperature, irradiance, and wind speed, are temporarily missing, such as due to sensor failure causing data loss for 1-2 time periods, a time series prediction algorithm is used to fit and complete the missing data using historical data from adjacent time periods, thus ensuring the continuity of weather sensitivity coefficient calculation.

[0071] If production data, such as process duration and equipment power, is missing, it is filled in using process rules, such as filling in the missing data according to the standard process duration and rated equipment power. At the same time, the confidence level of the filled data is marked for subsequent model verification.

[0072] If energy or market data, green electricity output, and green certificate prices are missing, the previous period's data will be used for a smooth transition, while triggering a data re-collection mechanism. If three consecutive periods are missing, it will be marked as abnormal to avoid affecting the green electricity and green certificate balance calculation.

[0073] Outlier removal employs a dual verification mechanism of statistical rules and process thresholds to eliminate invalid data.

[0074] The statistical rules use the 3σ criterion to remove extreme values ​​that exceed the mean ± 3 times the standard deviation from the continuously collected time-series data, such as equipment power and green electricity output.

[0075] The process threshold is determined by combining the process specifications for electrical equipment manufacturing to eliminate abnormal data that exceeds the rated power of the equipment and the standard energy consumption range of the process. For example, if the processing power of a certain process far exceeds the rated value of the equipment, it is determined to be invalid data.

[0076] For the excluded abnormal data, record the reasons for the abnormality, such as sensor failure or data acquisition terminal disconnection, and generate an abnormality report for maintenance personnel to investigate.

[0077] Dimensional normalization unifies data of different dimensions and magnitudes, such as equipment power (kW), green electricity output (kWh), and green certificate price (yuan / certificate), into the [0,1] interval, eliminating the impact of dimensional differences on subsequent modeling. During normalization, the historical maximum and minimum values ​​of each data type are used as benchmarks to ensure that the normalized data accurately reflects the relative differences of the original data, while preserving the trend characteristics of the data and avoiding model weight imbalances due to dimensional issues.

[0078] Consistency checks perform cross-dimensional consistency checks on the preprocessed data to ensure that the data conforms to the constraint logic of the multi-scale coupled model.

[0079] Energy balance is verified by checking whether the total energy consumption of the workshop equals the sum of the energy consumption of each process, and whether the actual consumption of green electricity does not exceed the company's available green electricity. Process logic is verified by checking whether the processing sequence of the process conforms to the process specifications, such as whether the winding process is after the core stacking process, and whether the delivery period is within a reasonable range. Market rules are verified by checking whether the sales volume of green certificates does not exceed the amount of green certificates held, and whether the electricity purchased from the grid matches the carbon intensity data. If the verification fails, a data re-sampling or correction process is triggered until the data meets the consistency requirements.

[0080] The core data that passes consistency verification—green electricity consumption, green certificate sales, process energy consumption, and carbon intensity data—will be structured and encapsulated to generate standardized data packages containing data content, collection time, preprocessing identifiers, and verification results, ensuring the immutability and traceability of the data.

[0081] The process of timestamping and storing multi-source dynamic data is as follows:

[0082] First, the core data that has passed consistency verification after preprocessing, including actual green electricity consumption, green certificate sales volume, process energy consumption, grid purchase electricity, and real-time carbon intensity during the manufacturing period, is structured and encapsulated to generate a unified format certificate data package. The data package includes: data content, such as the green electricity consumption of 500kWh during the period of 10:00-10:15; data type identifier, such as energy, production, and market; data collection source, such as sensors, trading platforms, and EMS systems; preprocessing verification results, including whether missing data was filled or anomaly was removed; and data hash value. The data content is encrypted using the SHA-256 algorithm to ensure data integrity. Each data package is bound to a unique data ID, which corresponds one-to-one with the data traceability identifier in the multi-scale coupling model.

[0083] The consortium blockchain network consists of multiple trusted nodes, including: enterprise data submitters, third-party carbon certification verification agencies, energy data providers from power trading centers, green certificate data providers from green certificate trading platforms, and compliance supervisors from industry regulatory agencies. Before submitting data, enterprise nodes must complete access verification through the consortium blockchain's identity authentication mechanism, such as digital certificates and public-private key pairs, to ensure that only authorized nodes can submit evidence-based data.

[0084] The permissions of each participating node are strictly defined: enterprises can submit data, third-party institutions can verify data, and regulatory agencies can audit data, to prevent data tampering or unauthorized operations.

[0085] Enterprise nodes submit the encapsulated evidence storage data package to the consortium blockchain network in the form of on-chain transactions. Each evidence storage transaction includes: data hash value, data ID, submitter identity identifier, transaction timestamp (generated uniformly by the consortium blockchain network with millisecond accuracy), and data type label. After the transaction is submitted, the consortium blockchain network automatically assigns a unique transaction hash to the transaction, which serves as the core identifier for subsequent evidence storage queries and verifications.

[0086] The consortium blockchain employs a practical Byzantine fault-tolerant consensus mechanism suitable for industrial scenarios, ensuring the consistency and immutability of the stored data. After an enterprise node submits a transaction, verification nodes in the consortium blockchain network, such as third-party certification bodies and power trading center nodes, verify the transaction. This verification includes checking whether the data hash value matches the pre-processed original data to ensure the data has not been tampered with; verifying the legitimacy of the submitter and whether the transaction complies with the consortium blockchain's permission rules; and verifying whether the data content conforms to carbon footprint accounting logic, such as ensuring that the number of green certificates issued does not exceed the holding amount. Once more than two-thirds of the verification nodes pass the verification, the transaction reaches consensus and is packaged into a candidate block.

[0087] The approved candidate blocks are written into the distributed ledger of the consortium blockchain. Each block contains the hash value of the previous block in the block header, the timestamp of this block, the block height, and the hash values ​​of multiple evidence-based transactions contained in the block body. After the block is written, it is linked to the previous block through a chain structure to form an immutable link of blocks, transactions, and data. Any modification to the evidence-based data will cause the block hash to break and be recognized by all nodes in the network.

[0088] Once the notarization is completed, the consortium blockchain network automatically generates a notarization index table, associating data IDs, transaction hashes, timestamps, and data types. Enterprises can query the notarization time, submitter, and verification results of the corresponding data using the data ID or transaction hash, enabling full-process traceability of carbon footprint data. Third-party certification bodies or regulatory agencies can verify whether the data was notarized at the specified time and whether it has been tampered with through the consortium blockchain interface, ensuring the compliance and credibility of carbon footprint accounting results. The notarized data can be directly exported as a standard-compliant carbon footprint report attachment, resulting in a standardized and trustworthy dataset.

[0089] Step 3: Based on a standardized and reliable dataset, and combined with pre-acquired green electricity output characteristics, green certificate verification rules, and grid purchase carbon intensity, construct a real-time carbon intensity quantification model for the manufacturing period, and output time-varying carbon intensity data that is strongly correlated with the production period.

[0090] The process of constructing a real-time carbon intensity quantification model for the manufacturing period is as follows:

[0091] Extract core data for each time period from a standardized and trusted dataset: total energy consumption in the workshop, direct green electricity consumption of each process, green certificate issuance and sales, and electricity purchased from the grid; simultaneously acquire three types of rules and parameters.

[0092] The characteristics of green electricity output include the time-of-day output patterns of the enterprise's own photovoltaic or wind power, such as photovoltaic output accounting for more than 80% during the daytime (9:00-17:00), and the available time periods for purchased green electricity, such as the green electricity supply time periods agreed upon by the power trading platform, which are output by the enterprise-level green electricity and green certificate balance model. The green certificate redemption rules clearly stipulate that one green certificate corresponds to 1 MWh of green electricity, which can only offset non-green electricity consumption, and the redemption volume cannot exceed the enterprise's current green certificate holdings, which is defined by the green electricity and green certificate balance logic of the enterprise-level model. The grid-purchased electricity carbon intensity adopts the grid-level dynamic carbon intensity model, and the output time-division and zone values ​​are updated every 15 minutes to reflect the carbon emission levels of grid-purchased electricity at different times.

[0093] Based on the characteristics of green electricity output and the rules for green certificate verification, the total energy consumption of the workshop in each time period is precisely broken down. Priority is given to allocating direct green electricity consumption. For example, during peak photovoltaic output periods in the daytime, priority is given to using both self-owned and purchased green electricity. The direct green electricity consumption for this period is calculated and does not exceed the company's available green electricity. Regarding green certificate deductions, if direct green electricity consumption is insufficient to cover total energy consumption, the remaining non-green electricity portion is deducted according to the green certificate verification rules. The amount of green electricity deducted is calculated to ensure that the deducted amount does not exceed the company's green certificate holdings and only deducts the non-green electricity consumption portion. Finally, the amount of electricity purchased from the grid is determined by deducting the direct green electricity consumption and green certificate deduction from the total energy consumption, ensuring that the sum of all energy consumption components equals the total energy consumption of the workshop.

[0094] Based on the three parts of energy consumption after decomposition, and combining the carbon intensity of electricity purchased from the grid with the green certificate deduction coefficient, the real-time carbon intensity of the manufacturing period is calculated for each time period. In the direct consumption of green electricity, green electricity itself has no carbon emissions, so the carbon emissions for this part are 0. For the green certificate deduction part, the carbon emissions are obtained by multiplying the carbon intensity of electricity purchased from the grid by the green certificate deduction coefficient, which represents the market acceptance of green certificates. For example, green certificates recognized by international authoritative institutions can be fully deducted, while in some regions they are deducted proportionally. The carbon emissions of the electricity purchased from the grid are obtained by multiplying the carbon intensity of electricity purchased from the grid by the amount of electricity purchased from the grid. The real-time carbon intensity is calculated by adding the carbon emissions of the green certificate deduction part and the carbon emissions of the electricity purchased from the grid to obtain the total carbon emissions for that period, and then dividing by the total energy consumption of the workshop for that period to obtain the real-time carbon intensity of the manufacturing period for that period.

[0095] Carbon intensity is strongly correlated with production time periods. A dynamic update mechanism with a 15-minute granularity is constructed, synchronously updating the available green electricity, green certificate holdings, and grid-purchased electricity carbon intensity every 15 minutes. The direct consumption of green electricity, green certificate deductions, and grid-purchased electricity are readjusted for each time period. If the production schedule is adjusted, such as changing the processing time from daytime to nighttime, the energy consumption breakdown and carbon intensity for the corresponding time period are recalculated to ensure that carbon intensity changes dynamically with the production time period. If the green certificate deduction coefficient is adjusted, such as due to increased international market acceptance, the carbon emission calculation for the green certificate deduction portion is updated synchronously to ensure that carbon intensity conforms to the latest market rules. The real-time carbon intensity of each time period is mapped one-to-one with the production time period to generate a time-varying carbon intensity dataset. For example, the carbon intensity for the time period from 10:00 to 10:15 is 0.2 kg CO2e / kWh, with a high proportion of green electricity, while the carbon intensity for the time period from 18:00 to 18:15 is 0.5 kg CO2e / kWh, with a low proportion of green electricity. This dataset is output to the workshop-level scheduling constraint model as the core constraint for subsequent multi-objective optimization.

[0096] Step 4: Using time-varying carbon intensity data as constraints, construct a multi-objective optimization model with the objectives of minimizing the carbon footprint of the manufacturing stage, minimizing total production costs, and minimizing delivery delays. Solve the model using an improved dual-population genetic algorithm and output the Pareto optimal production scheduling scheme.

[0097] The specific process of constructing a multi-objective optimization model is as follows:

[0098] To minimize the carbon footprint during the manufacturing phase, the total energy consumption of the workshop in each production period is multiplied by the time-varying carbon intensity corresponding to that period, and the carbon emissions of the entire manufacturing process are accumulated over time. The optimization objective is to reduce the total carbon emissions as much as possible, and to automatically reduce the production load of high-energy-consuming processes during periods of high carbon intensity.

[0099] To minimize total production costs, the total cost is broken down into four items: electricity procurement cost, green certificate procurement cost, equipment operation and maintenance cost, and delivery default penalty cost. Electricity procurement cost includes the real-time costs of purchased green electricity and electricity purchased from the grid. Green certificate procurement cost is calculated based on real-time transaction prices and procurement volume. Equipment operation and maintenance cost is positively correlated with total production energy consumption. Default cost is calculated based on the actual overdue period. The goal is to minimize the sum of these four costs.

[0100] To minimize delivery delays, the actual total time from the start of the first process to the completion of the last process for each piece of electrical equipment is calculated and compared with the agreed delivery date. Only the total time exceeding the delivery date is calculated, with the goal of minimizing the overdue time and prioritizing on-time delivery.

[0101] Constructing multi-layered constraints with time-varying carbon intensity as the core:

[0102] During the process of incorporating basic workshop scheduling constraints, the same workpiece must be processed in the order of process sequence, a single machine can only process one process at a time, and the total energy consumption of a single machine during a time period cannot exceed the rated power; a time-varying carbon intensity core constraint is added, the real-time carbon intensity of each production period must not exceed the upper limit of the enterprise's preset target carbon intensity, and the arrangement of high-energy-consuming processes is strictly restricted during high-carbon emission periods; at the same time, special constraints for green electricity and green certificates are added, the direct consumption of green electricity cannot exceed the real-time available amount during that period, the sales volume of green certificates cannot exceed the current holding amount of the enterprise, and the total energy consumption ratio of green electricity consumption and green certificate deduction is not less than the minimum ratio stipulated by the enterprise; when setting energy consumption balance constraints, the total production energy consumption of each period is equal to the sum of the direct consumption of green electricity, the amount of electricity deducted by green certificates, and the electricity purchased from the grid, ensuring that the energy consumption data is closed-loop and unbiased;

[0103] The three objectives—carbon footprint, total cost, and duration of default—with different dimensions are unified and normalized to eliminate optimization bias caused by differences in numerical magnitude. Then, based on the company's low-carbon strategy, cost control efforts, and delivery priorities, corresponding weights are assigned to integrate the three independent objectives into a directly solvable comprehensive optimization objective. This achieves optimal balance among the three objectives while ensuring carbon intensity constraints. The integrated comprehensive optimization objective and all constraints are embedded into the workshop-level scheduling constraint model, synchronously linking it with the process-level energy consumption model, the company-level green electricity and green certificate balance model, and the grid-level dynamic carbon intensity model to form a closed-loop optimization model. The model's decision variables include process equipment, processing time, green electricity allocation ratio, and green certificate write-off quantity, ensuring that each output scheduling scheme meets time-varying carbon intensity constraints while simultaneously considering the three core objectives of low carbon, low cost, and guaranteed delivery.

[0104] The process of outputting the Pareto optimal production scheduling scheme is as follows:

[0105] The decision variables of the algorithm are obtained, which are the core adjustment dimensions of the scheduling scheme. The decision variables include the allocation of processing equipment for each process, the selection of processing time, the green electricity allocation ratio, and the number of green certificates to be revoked. At the same time, a segmented coding rule is designed to map each decision variable to a coding segment, such as workpiece 1, iron core stacking process, equipment 3 time period from 10:00 to 10:15, green electricity allocation of 80%, and 2 green certificates to be revoked. This ensures that the coding corresponds one-to-one with the scheduling scheme and is compatible with all constraints such as carbon intensity, green electricity and green certificates, and process sequence, providing a basic carrier for population evolution.

[0106] The initial population is split into a low-carbon oriented population and a cost-oriented population, with both populations of equal size. A heuristic method is used for differentiated initialization to avoid the local optima of a single population.

[0107] The low-carbon-oriented population initialization prioritizes the goal of minimizing the carbon footprint. High-energy-consuming processes such as transformer core lamination and insulation casting are scheduled during peak green electricity output periods, such as photovoltaic power generation from 9:00 to 17:00 and periods when the grid carbon intensity is low. Green certificates are prioritized for the cancellation of non-green electricity consumption during high-carbon periods to ensure that the initial solution has low-carbon characteristics.

[0108] Cost-oriented population initialization prioritizes minimizing total cost, selecting equipment and processing procedures with low maintenance costs, and matching the purchase volume of green electricity or green certificates to periods of low market prices. At the same time, it prioritizes processes with near-delivery dates to ensure that the initial solution balances cost and delivery requirements.

[0109] For each individual in both populations, i.e. a scheduling scheme, three major target values ​​are first calculated: total carbon footprint of the manufacturing stage, total production cost, and total duration of delivery default. Then, the target values ​​are normalized to eliminate dimensional differences, and the overall fitness is calculated in combination with the target weights preset by the enterprise. At the same time, all constraints are verified, such as the upper limit of carbon intensity, the amount of green electricity available, and the process sequence. Individuals that violate the constraints are directly set to extremely low fitness to ensure that only feasible solutions participate in subsequent evolution.

[0110] Dedicated crossover and mutation operators are designed for the characteristics of the two population types to preserve the core advantages of the populations while avoiding evolutionary stagnation. For low-carbon oriented populations: during crossover, priority is given to retaining the coding segments of low-carbon time period process arrangements and high green electricity allocation ratios; during mutation, only the processing time of high-energy-consuming processes is randomly adjusted, and only time periods with lower carbon intensity are replaced to ensure that the mutation still aligns with low-carbon goals. For cost-oriented populations: during crossover, priority is given to retaining the coding segments of low-cost equipment and low green certificate purchase prices; during mutation, the green electricity purchase volume is adjusted, only matching processing equipment for lower-priced time periods or processes, and only replacing equipment with lower operation and maintenance costs to ensure that the mutation still aligns with cost goals.

[0111] After each generation of evolution is completed, a population collaboration mechanism is activated to merge all individuals in the low-carbon-oriented population and the cost-oriented population. Non-dominated solutions, i.e. Pareto front solutions, are selected through non-dominated sorting without any objective inferior to other solutions. These high-quality non-dominated solutions are then injected back into the two populations to replace the individuals with the lowest fitness in each population. At the same time, the core orientation characteristics of each population are preserved. For example, the low-carbon population still focuses on low carbon. This allows the two populations to share the optimal solution while avoiding a single population from getting trapped in local optima, thus improving the diversity and optimality of the solution set.

[0112] Set iteration termination conditions: if the number of iterations reaches a preset value or there is no significant optimization in the solution set for 5 consecutive generations, the process terminates. Then, all non-dominated solutions from both populations are merged, duplicate solutions are removed, and the final Pareto optimal solution set is formed. For each scheduling scheme in the solution set, the corresponding manufacturing stage carbon footprint, total production cost, and delivery delay default duration are labeled and output in a structured form. For example, Scheme 1: carbon footprint reduced by 58%, cost increased by 8%, no default; Scheme 2: carbon footprint reduced by 35%, cost reduced by 2%, no default. Based on actual needs, such as choosing a low-carbon scheme to cope with carbon tariffs or a cost-saving scheme to increase efficiency, the final production scheduling scheme to be executed is selected from the solution set and synchronized to the workshop-level scheduling constraint model to guide production.

[0113] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0114] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0115] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for modeling the carbon footprint of green electricity and green certificates electrical equipment based on life cycle assessment, characterized in that, The method includes: Step 1: Based on the pre-built digital twin coupled model of electrical equipment, a four-scale model is built, and the interaction and nesting of the four-scale model are bidirectionally coupled through the bidirectional parameter transfer interface. The four-scale model includes a process-level energy consumption time-varying model, a workshop-level scheduling constraint model, an enterprise-level green electricity and green certificate balance model, and a grid-level dynamic carbon intensity model. Step 2: Based on the four-scale model after bidirectional coupling, multi-source dynamic data is acquired through the Internet of Things. After preprocessing, the preprocessed multi-source dynamic data is timestamped and stored using a consortium blockchain to generate a standardized trusted dataset. The consortium blockchain consists of multiple trusted nodes. Step 3: Based on a standardized and reliable dataset, and combined with pre-acquired green electricity output characteristics, green certificate verification rules, and grid purchase carbon intensity, construct a real-time carbon intensity quantification model for the manufacturing period, and output time-varying carbon intensity data that is strongly correlated with the production period. Step 4: Using time-varying carbon intensity data as constraints, construct a multi-objective optimization model with the objectives of minimizing the carbon footprint of the manufacturing stage, minimizing total production costs, and minimizing delivery delay defaults. Solve the model using an improved dual-population genetic algorithm and output the Pareto optimal production scheduling scheme.

2. The carbon footprint modeling method for green electricity and green certificates electrical equipment based on life cycle assessment as described in claim 1, characterized in that, The process of building a four-scale model is as follows: The process-level energy consumption time-varying model distinguishes between the ground-state energy consumption when there is no workpiece processing and the dynamic energy consumption during workpiece processing. It introduces a weather sensitivity coefficient and an energy efficiency feedback coefficient, which directly links the green electricity usage strategy with the energy efficiency of the process. The workshop-level scheduling constraint model defines process sequence, equipment capacity, delivery time, and energy consumption by constructing a correlation matrix of processes, equipment, time, and carbon footprint. The enterprise-level green electricity and green certificate balance model establishes a dynamic balance logic for green electricity production, consumption, procurement, and verification. The grid-level dynamic carbon intensity model outputs time-of-use grid carbon intensity, green electricity zero carbon, and green certificate deduction equivalent carbon intensity, which is obtained by multiplying the carbon intensity by the deduction coefficient.

3. The carbon footprint modeling method for green electricity and green certificates electrical equipment based on life cycle assessment as described in claim 2, characterized in that, The bidirectional parameter transfer interface enables bidirectional coupling of the interaction and nesting of the four-scale model as follows: The bidirectional parameter transmission coupling interface enables dynamic nesting and data interaction of four-scale models—process-level energy consumption time-varying model, workshop-level scheduling constraint model, enterprise-level green electricity and green certificate balance model, and grid-level dynamic carbon intensity model—through a downward transmission and upward feedback mechanism.

4. The carbon footprint modeling method for green electricity and green certificates electrical equipment based on life cycle assessment as described in claim 1, characterized in that, The process of acquiring multi-source dynamic data through the Internet of Things is as follows: Based on the self-operation logic of the four-scale model after bidirectional coupling, data acquisition tasks are generated, and the acquisition method is matched according to the type of data acquisition task to perform hierarchical and classified data acquisition.

5. The carbon footprint modeling method for green electricity and green certificates electrical equipment based on life cycle assessment as described in claim 4, characterized in that, The process of generating a standardized, trustworthy dataset is as follows: The preprocessed multi-source dynamic data is structured and encapsulated to generate a unified format evidence storage data package. Before the data package is submitted, the enterprise node completes the access verification through the identity authentication mechanism of the consortium blockchain. The enterprise node submits the encapsulated evidence storage data package to the consortium blockchain network in the form of an on-chain transaction. The consortium blockchain adopts a practical Byzantine fault-tolerant consensus mechanism. After an enterprise node submits a transaction, the verification nodes in the consortium blockchain network verify the transaction. When more than 2 / 3 of the verification nodes pass the verification, the transaction reaches a consensus and is packaged into a candidate block. Otherwise, the consensus of the notarized transaction is not established, and it is directly judged as an invalid transaction and will not be packaged into a candidate block. Candidate blocks that have reached a consensus on transactions are written into the distributed ledger of the consortium blockchain. After the notarization is completed, the consortium blockchain network automatically generates a notarization index table, and a standardized trusted dataset is obtained based on the notarization index table.

6. The carbon footprint modeling method for green electricity and green certificates electrical equipment based on life cycle assessment as described in claim 5, characterized in that, The process of outputting time-varying carbon intensity data that is strongly correlated with the production period is as follows: Based on the standardized and reliable dataset, the total energy consumption of the workshop, the direct consumption of green electricity in each process, the sales volume of green certificates, and the electricity purchased from the grid are extracted. According to the characteristics of green electricity output and the rules for green certificate verification, the total energy consumption of the workshop in each time period is broken down. Based on the total energy consumption after the breakdown, combined with the carbon intensity of electricity purchased from the grid, the real-time carbon intensity of the manufacturing period in each time period is calculated and dynamically updated every 15 minutes. The real-time carbon intensity of each time period is matched one-to-one with the production period to generate a time-varying carbon intensity dataset.

7. The carbon footprint modeling method for green electricity and green certificates electrical equipment based on life cycle assessment as described in claim 1, characterized in that, The process of constructing a multi-objective optimization model is as follows: Using time-varying carbon intensity data as constraints, including basic scheduling constraints at the workshop level, time-varying carbon intensity constraints, supplementary green electricity and green certificate constraints, and energy consumption balance constraints, the objectives of carbon footprint, total cost, and default duration with different dimensions are normalized and assigned corresponding weights. The three independent objectives are integrated into a comprehensive optimization objective. The integrated comprehensive optimization objective and all constraints are embedded into the workshop-level scheduling constraint model, linking the process-level energy consumption model, the enterprise-level green electricity and green certificate balance model, and the grid-level dynamic carbon intensity model to form a closed-loop optimization model.

8. The carbon footprint modeling method for green electricity and green certificates electrical equipment based on life cycle assessment as described in claim 6, characterized in that, The process of outputting the Pareto optimal production scheduling scheme is as follows: Based on the allocation of processing equipment, selection of processing time periods, green electricity allocation ratio, and number of green certificate cancellations for each process, a segmented coding rule is designed to map each decision variable to a coding segment. The initial population is split into a low-carbon oriented population and a cost-oriented population, and a heuristic method is used for differentiated initialization. For each individual in the two populations, the target value is normalized by the total carbon footprint of the manufacturing stage, the total production cost, and the total duration of delivery default. The comprehensive fitness is calculated by combining the target weights preset by the enterprise. At the same time, all constraints are verified. Individuals that violate the constraints are directly set to extremely low fitness. Dedicated crossover and mutation operators are designed for the characteristics of the two populations. All individuals in the low-carbon oriented population and the cost oriented population are merged. Non-dominated solutions with no objective inferior to other solutions are selected through non-dominated sorting. The high-quality non-dominated solutions are then injected back into the two populations to replace the individuals with the lowest fitness in each population. At the same time, the orientation characteristics of each population are preserved. The process is terminated when the number of iterations reaches a preset value.

Images