Zero-carbon park emission reduction path planning method based on coupling modeling of technology penetration rate

Abstract

The application relates to the technical field of power system planning, comprehensive energy system optimization and low-carbon park construction, in particular to a zero-carbon park emission reduction path planning method based on coupling modeling of technical penetration rates. The method comprises the following steps: S1, obtaining target park basic data; S2, constructing a low-carbon technology candidate library; S3, constructing a unified variable system; S4, establishing a penetration rate-physical scale coupling conversion model; S5, constructing a double-target emission reduction path optimization model; S6, constructing a constraint system; and S7, designing an improved multi-objective discrete particle swarm algorithm. By constructing a unified variable system and establishing a coupling mapping relationship between the technical penetration rate and the installed capacity or the physical implementation scale, the application effectively solves the problem that the variable system of the capacity technology and the non-capacity technology is split and difficult to be cooperatively modeled in the traditional planning method.

Description

Technical Field

[0001] This invention relates to the fields of power system planning, integrated energy system optimization, and low-carbon industrial park construction, specifically a zero-carbon industrial park emission reduction path planning method based on technology penetration rate coupling modeling. Background Technology

[0002] Zero-carbon industrial park emission reduction pathway planning needs to simultaneously consider low-carbon technology selection, technology implementation scale, energy supply and demand balance, carbon emission constraints, investment and operating costs, and the interaction model of power generation, grid, load, and storage. Existing park-level optimization methods typically use installed capacity, investment scale, or single technology penetration rate as modeling variables, making it difficult to uniformly represent different types of low-carbon technologies in the same model. In particular, it is difficult to be compatible with capacity-dependent technologies such as photovoltaics, wind power, energy storage, and CCUS, as well as capacity-independent technologies such as electric vehicle replacement, building energy-saving retrofits, ecological carbon sinks, and demand response.

[0003] Existing methods suffer from the following main problems: First, the variable systems for technologies with and without capacity are fragmented. Technologies with capacity are often represented by installed capacity, while those without capacity are often represented by proportions or implementation rates. It is difficult to solve these two types of technologies collaboratively within the same optimization framework, affecting the overall optimality of the planning results. Second, there is insufficient coupling between technology selection variables and implementation scale variables. Even when binary variables indicating whether a technology is selected are set, they often fail to effectively link with penetration rate, installed capacity, or physical implementation scale, potentially resulting in situations where a technology is not selected but a non-zero scale still exists. Third, existing optimization algorithms are insufficiently adapted to discrete decision-making. Park planning includes not only continuous variables such as penetration rate, energy flow, and energy storage charging and discharging power, but also discrete variables such as technology selection and the choice of source-grid-load-storage interaction modes, which traditional continuous optimization algorithms struggle to handle effectively. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling.

[0005] A zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling includes the following steps: S1. Obtain basic data of the target park; S2. Construct a candidate library of low-carbon technologies: Based on whether the technology has a clearly measurable installed capacity attribute, candidate low-carbon technologies are divided into technologies with capacity and technologies without capacity. S3. Construct a unified variable system: Define the following variables for the k-th low-carbon technology in the i-th planning year: technology selection variable, technology penetration rate variable, installed capacity variable for technologies with capacity, and physical implementation scale variable for technologies without capacity. S4. Establish a penetration rate-physical scale coupling transformation model: For technologies with capacity, establish a mapping relationship between technology penetration rate and installed capacity; for technologies without capacity, establish a mapping relationship between technology penetration rate and physical implementation scale; and set linkage constraints between technology selection variables and penetration rate variables, installed capacity variables, or physical implementation scale variables to force the penetration rate and implementation scale of unselected technologies to zero. S5. Construct a dual-objective emission reduction path optimization model: With the optimization objectives of maximizing total carbon emission reduction and minimizing comprehensive cost within the planning period, construct a dual-objective mixed variable optimization model that is compatible with discrete and continuous variables; S6. Construct a constraint system; S7. Design an improved multi-objective discrete particle swarm optimization algorithm: construct a particle encoding structure adapted to mixed variables; establish an adaptive velocity update mechanism; adopt a binary discrete update method for technology selection variables, a multi-class probability mapping update method for source-grid-load-storage interaction mode variables, and a continuous update and boundary correction method for technology penetration rate variables; establish an external Pareto archive and output the optimization results.

[0006] By classifying low-carbon technologies into those with and without capacity and establishing a unified variable system and coupled transformation model, the technical problem of the separation of variables between the two types of technologies and the inability to optimize them collaboratively in traditional methods is solved. By constructing a dual-objective mixed variable optimization model and designing an improved multi-objective discrete particle swarm optimization algorithm adapted to discrete decision-making, the joint optimization of technology selection, penetration rate, implementation scale and source-grid-load-storage interaction mode is achieved, thereby improving the systematicness and feasibility of the planning results.

[0007] Preferably, in step S3, The variables selected for the technology are represented as follows:

[0008] in, This indicates that the k-th technology is selected in year i. This indicates that the technology will not be used in year i. The technology penetration rate variable is represented as follows: For technologies with capacity, the penetration rate represents the proportion of implementation relative to the maximum adaptable capacity; for technologies without capacity, the penetration rate represents the proportion of promotion relative to the benchmark size of the park. in, This represents the application rate of the k-th technology in year i. For technologies with capacity, the penetration rate represents the implementation rate relative to the maximum adaptable capacity; for technologies without capacity, the penetration rate represents the promotion rate relative to the benchmark size of the park. The installed capacity variable of the capacity technology mentioned:

[0009] in, This represents the actual installed capacity or processing power of the k-th capacity technology in year i. The physical implementation scale variables of the capacity-free technology:

[0010] in, This represents the actual implementation scale of the k-th capacity-less technology in year i.

[0011] By defining the hierarchical variables of technology selection, penetration rate, installed capacity, and implementation scale, a variable basis is provided for the unified expression of technologies with and without capacity in the same optimization model, making the decision variables of different types of low-carbon technologies comparable and synergistic.

[0012] Preferably, the penetration rate-installed capacity mapping relationship of the capacity technology described in step S4 is as follows:

[0013] in: Let k be the actual installed capacity of the k-th capacity-enabled technology in year i. Let be the penetration rate of the k-th technology in year i; This represents the maximum adaptability capacity of the k-th technology within the target park. Let k be the selection variable for the k-th technology in year i; The linkage constraint is:

[0014] when hour, and further enable This forces the penetration rate and installed capacity of unselected technologies to zero.

[0015] By expressing installed capacity as the product of technology selection variables, penetration rate, and maximum adaptable capacity, the organic linkage between technology selection decisions and implementation scale decisions is achieved. This eliminates the logical contradiction in traditional methods where a non-zero installed capacity still exists even if the technology is not selected, ensuring the feasibility of the planning results in engineering.

[0016] Preferably, the penetration rate-physical implementation scale mapping relationship of the capacity-free technology in step S4 is as follows:

[0017] in: Let k be the actual implementation scale of the capacity-free technology in year i. Let k be the benchmark size of the park corresponding to the k-th technology; Let be the penetration rate of the k-th technology in year i; The capacity-free technologies include electric vehicle replacement technology, building energy-saving retrofit technology, ecological carbon sink technology, and demand response technology. For electric vehicle replacement technology, the benchmark scale of the park is the total number of vehicles in the park, and the actual implementation scale is the actual number of vehicles replaced. For building energy-saving renovation technology, the benchmark scale of the park is the total building area of ​​the park, and the actual implementation scale is the actual renovation area. For ecological carbon sequestration technology, the baseline scale of the park is the greenable area of ​​the park, and the actual implementation scale is the actual green area; For demand response technology, the baseline scale of the park is the total interruptible load of the park, and the actual implementation scale is the actual load scale participating in demand response.

[0018] By unifying the physical implementation scale of different dimensions such as the number of electric vehicles replaced, the area of ​​building energy-saving renovation, the area of ​​ecological carbon sink, and the demand response load into the product of penetration rate and benchmark scale, the capacity-less technology, which cannot be modeled with installed capacity in traditional methods, can be solved collaboratively with the capacity-based technology within the same optimization framework, thus expanding the applicability of this invention to different types of parks.

[0019] Preferably, the objective function of the dual-objective emission reduction path optimization model in step S5 is: Objective 1: Maximize total carbon emission reductions within the planning period

[0020] Where Y is the planning period in years, and K is the number of candidate technologies. Let be the carbon emission reduction generated by the k-th technology in year i; For technologies with capacity:

[0021] in, This refers to the annual equivalent output or emission reduction coefficient per unit capacity. For corresponding energy or emission factors; For capacity-less technologies:

[0022] in, Emission reduction coefficients for implementation scale; Objective 2: Minimize the overall cost within the planning period

[0023] in: For investment costs; Operating costs; To maintain costs; Costs related to carbon trading, carbon offsetting, or carbon assets.

[0024] With the dual optimization objectives of maximizing carbon emission reduction and minimizing overall cost, it avoids planning biases that may be caused by single-objective optimization (such as over-pursuing emission reduction while ignoring economic efficiency or over-pursuing low cost while weakening emission reduction effect). This enables the output emission reduction path to achieve Pareto optimality between emission reduction effect and economic efficiency, meeting the multi-objective trade-off needs in the actual decision-making of the park.

[0025] Preferably, the constraint system in step S6 includes: The carbon balance constraint is:

[0026] in, This represents the baseline carbon emissions for the park in year i. To reduce emissions through technology, Carbon sequestration amount External carbon offset amount The target carbon emissions for year i; For zero-carbon industrial parks, the following can be further set up: .

[0027] Energy supply and demand balance constraints: For each time period t, establish the balance relationship of multiple energy flows of electricity, heat, cold and hydrogen; Technology Scale Boundary Constraints: ; ; ; Source-grid-load-storage interaction ratio constraint: set according to the source-grid-load-storage structure type of the target park, which includes source-dominant, load-dominant, storage-dominant and grid-dominant types.

[0028] Carbon balance constraints ensure the achievability of emission reduction targets, energy supply and demand balance constraints ensure the reliability of energy supply, technology scale boundary constraints ensure the feasibility of planning schemes under park resource conditions, energy storage operation constraints ensure the physical operation rationality of energy storage systems, and source-grid-load-storage interaction ratio constraints achieve adaptability to different types of park scenarios. The combined effect of multiple constraints makes the final emission reduction path technically feasible and physically operable.

[0029] Preferably, the particle encoding structure described in step S7 is represented as follows:

[0030] in, This is the technology selection coding segment, used to indicate whether each low-carbon technology is selected; This is the encoding segment for the source-grid-load-storage interaction mode, used to represent interaction modes such as source-led, load-led, storage-led, or grid-led. This is the technology penetration rate coding segment, used to represent the implementation ratio of each low-carbon technology within the planning cycle.

[0031] By encoding discrete variables of technology selection, discrete variables of source-grid-load-storage interaction mode, and continuous variables of technology penetration rate into the same particle, the algorithm can simultaneously handle discrete and continuous decisions in one optimization process. This avoids the suboptimal solution problem caused by optimizing discrete and continuous variables separately in traditional methods and improves the global optimization capability.

[0032] Preferably, the adaptive speed update mechanism in step S7 is as follows:

[0033] in, Let be the velocity of the nth particle in the d-th dimension at the t-th iteration. For the corresponding position, For the optimal position of an individual, The global boot location selected from the external Pareto archive. , It is a random number; The adaptive inertia weights and dynamic learning factors are dynamically adjusted with the number of iterations, so that the algorithm focuses on global search in the early stage and local convergence in the later stage.

[0034] By adaptively changing the inertia weight and learning factor with the number of iterations, a large inertia weight and individual learning factor are maintained in the early stage of the algorithm to enhance the global search capability and avoid getting trapped in local optima. In the later stage of the algorithm, the inertia weight is reduced and the social learning factor is increased to accelerate local convergence. This effectively balances the algorithm's exploration and development capabilities, and improves the optimization efficiency and the quality of the Pareto solution set.

[0035] Preferably, in step S7: A binary discrete update method is used for the technology selection variable, mapping particle velocity to the probability of technology being selected:

[0036] This technique should be used when the probability of generating a random number is less than this value; otherwise, it should not be used. The variables of the source-grid-load-storage interaction mode are updated using a multi-class probability mapping method. The particle velocity is mapped to the selection probability of each interaction mode through the softmax function, and the interaction mode is determined according to the maximum probability or the roulette wheel method. The technology penetration rate variable is continuously updated and its boundary is corrected, and the update results are corrected for the boundary:

[0037] Simultaneously, combined with the technology selection variables, a linkage correction is performed to force the penetration rate, installed capacity, and implementation scale of unselected technologies to zero.

[0038] By employing sigmoid probability mapping for binary variables, softmax probability mapping for multi-class discrete variables, and boundary correction for continuous variables, the improved multi-objective discrete particle swarm optimization algorithm can effectively adapt to the characteristics of the mixed discrete and continuous variables in the model of this invention. At the same time, the linkage correction mechanism ensures that the penetration rate and implementation scale of the unselected technology are forced to zero, ensuring that the physical meaning of all particles remains consistent throughout the algorithm iteration process.

[0039] Preferably, the external Pareto archive mentioned in step S7 is used to store non-dominated solutions; after each iteration, the new particle is compared with the existing particles in the external Pareto archive for non-domination, the dominated solutions are deleted and the non-dominated solutions are retained; when the archive size exceeds the preset upper limit, the solution set is filtered according to the crowding distance, and the set of Pareto optimal paths with uniform distribution is retained; finally, the candidate emission reduction path that takes into account both maximizing carbon emission reduction and minimizing comprehensive cost is output according to the external Pareto archive, and the optimal or near-optimal path scheme under different source-grid-load-storage interaction scenarios is formed.

[0040] By preserving the optimal solution set under multi-objective significance through non-dominated sorting and ensuring the uniformity of Pareto front distribution through crowding distance screening, this invention provides decision-makers with a diverse selection space of candidate solutions. At the same time, by outputting four differentiated emission reduction paths—source-dominated, load-dominated, storage-dominated, and grid-dominated—based on different scenarios, this invention can be adapted to the source-grid-load-storage structure characteristics of different types of parks, enhancing the universality and engineering practical value of the method.

[0041] Compared with existing technologies, this invention has the following advantages: Firstly, by constructing a unified variable system and establishing a coupled mapping relationship between technology penetration rate and installed capacity or physical implementation scale, this invention effectively solves the problem of fragmented variable systems and difficulty in collaborative modeling between technologies with and without capacity in traditional planning methods. Secondly, by linking the technology selection variable with penetration rate and implementation scale, it eliminates the logical contradiction of non-selected technologies still taking non-zero values, enabling the planning results to directly correspond to specific implementation content such as photovoltaic installed capacity, energy storage capacity, vehicle replacement quantity, building renovation area, and carbon sink scale, significantly improving the feasibility of the plan. Thirdly, the dual-objective hybrid variable optimization model constructed in this invention incorporates discrete variables such as technology selection and interaction mode selection with continuous variables such as technology penetration rate and energy output into a unified framework, avoiding the limitations of a single objective. The invention addresses planning biases caused by target optimization. Furthermore, it proposes an improved multi-objective discrete particle swarm optimization algorithm designed for discrete decision-making scenarios. This algorithm effectively improves the solution efficiency and Pareto solution quality of mixed-variable optimization problems by adapting to a particle encoding structure for mixed variables, using adaptive inertia weights and dynamic learning factors, discrete variable update rules, and a Pareto profile screening mechanism based on congestion distance. Simultaneously, the invention systematically verifies multiple constraints during the optimization process, including carbon balance, energy supply and demand balance, technology scale boundaries, energy storage operation, and the ratio of source-grid-load-storage interaction. It can also output differentiated optimal or near-optimal emission reduction paths for four types of park scenarios: source-led, load-led, storage-led, and grid-led. This provides universally applicable and practically useful technical support for low-carbon technology selection, source-grid-load-storage collaborative configuration, and emission reduction investment decisions for different types of zero-carbon parks. Attached Figure Description

[0042] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0043] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0044] Reference Figure 1 A zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling includes the following steps: S1. Obtain basic data for the target park: First, determine the planning period Y, typically set to 5-15 years, with one year as one planning phase, for a total of Y planning phases. The specific basic data obtained includes: (1) Park type data: including information such as park functional positioning, land use composition, industry type and output value scale; (2) Energy system data: including information such as the current energy structure of the park, total consumption of various types of energy and consumption by type, and major energy supply facilities; (3) Source-side data: including the existing distributed power types and installed capacity in the park, the available new energy resources in the surrounding area and the maximum access capacity, etc. (4) Network side data: including information such as the structure of the industrial park's distribution network, voltage level, line scale, and power supply capacity of the upstream power grid; (5) Load-side data: including typical daily load curves of the park, seasonal load characteristics, composition and proportion of various loads, type and scale of adjustable loads, etc. (6) Storage-side data: including information such as the types and capacities of existing energy storage facilities in the park, and the maximum planned energy storage capacity; (7) Carbon emission data: including total carbon emissions in the park in the baseline year and the composition of emissions by product, carbon emission distribution in various industries or links, etc. (8) Cost data: including unit investment cost, operation and maintenance cost, carbon trading price or carbon tax rate of various low-carbon technologies; (9) Time series data: including hourly load data, hourly renewable energy output data and hourly meteorological data. The time granularity is usually 1 hour, i.e. 8760 hours / year.

[0045] The aforementioned basic data can be obtained through channels such as park energy audit reports, statistical yearbooks, on-site surveys, and professional databases.

[0046] S2. Constructing a Low-Carbon Technology Candidate Library: The construction of the candidate low-carbon technology library follows these principles: covering the main carbon emission links in the park, meeting the implementation requirements within the planning period in terms of technology maturity, and being compatible with the park's resource endowment and infrastructure conditions. Candidate technologies should include at least the following categories: (1) Renewable energy and energy system technologies: including photovoltaic power generation, wind power generation, biomass energy utilization, ground source heat pumps, gas turbine cogeneration, etc.; (2) Energy-saving technologies for buildings and infrastructure: including building exterior wall insulation, replacement of energy-saving doors and windows, high-efficiency air conditioning systems, green lighting renovation, etc.; (3) Digital and intelligent management technologies: including intelligent energy management platforms, source-grid-load-storage coordinated dispatching systems, and intelligent microgrid control systems; (4) Circular economy and resource utilization technologies: including waste heat recovery and utilization, waste-to-energy conversion, water resource recycling, etc.; (5) Carbon sinks and carbon management technologies: including ecological carbon sinks (greening, forest construction), CCUS (carbon capture, utilization and storage), etc.; (6) Transportation and logistics technologies: including electric vehicle replacement, shore power systems, intelligent traffic scheduling, etc. After the candidate low-carbon technology library is completed, the candidate low-carbon technologies are divided into technologies with capacity and technologies without capacity based on whether the technologies have a clearly measurable installed capacity attribute.

[0047] Capacity-based technologies refer to technologies with a clearly defined installed capacity (usually measured in kW, MW, or kWh), including but not limited to: photovoltaic power generation, wind power generation, electrochemical energy storage, pumped storage, CCUS, gas turbine cogeneration, and shore power systems.

[0048] Capacity-free technologies refer to technologies that do not have a clear installed capacity attribute and whose implementation scale is characterized by physical quantities (such as quantity, area, volume, etc.), including but not limited to: electric vehicle replacement, building energy-saving renovation, ecological carbon sink, demand response, smart energy management platform, green lighting renovation, etc.

[0049] S3. Construct a unified variable system: For the k-th low-carbon technology in the i-th planning year, define the following variables: (1) Technology selection variables:

[0050] in, This indicates that the k-th technology is selected in year i. This indicates that the technology will not be used in year i.

[0051] (2) Technology penetration rate variable:

[0052] in, This represents the application rate of the k-th technology in year i. For technologies with capacity, the penetration rate represents the implementation rate relative to the maximum adaptable capacity; for technologies without capacity, the penetration rate represents the promotion rate relative to the benchmark size of the park.

[0053] (3) Variables in installed capacity of capacity technologies:

[0054] in, This represents the actual installed capacity or processing power of the k-th capacity technology in year i.

[0055] (4) Variables related to the physical implementation scale of capacity-free technologies:

[0056] in, This represents the actual implementation scale of the k-th capacity-less technology in year i.

[0057] The above variables were established for all candidate low-carbon technologies and for each planning year, forming a complete variable system.

[0058] S4. Establish a penetration rate-physical scale coupling transformation model: (1) The conversion relationship of capacity technology For technologies with capacity, the following conversion formula is established: Formula 1 in: Let k be the actual installed capacity of the k-th capacity-enabled technology in year i. Let be the penetration rate of the k-th technology in year i; This represents the maximum adaptability capacity of the k-th technology within the target park. Let be the selection variable for the k-th technology in year i.

[0059] Simultaneously set linkage constraints: Formula 2 when hour, and further enable This forces the penetration rate and installed capacity of unselected technologies to zero.

[0060] (2) Transformation relationship of capacity-free technology For capacity-less technologies, the following conversion formula is established: Formula 3 in: Let k be the actual implementation scale of the capacity-free technology in year i. Let k be the benchmark size of the park corresponding to the k-th technology; Let be the penetration rate of the k-th technology in year i.

[0061] For example: Among electric vehicle replacement technologies, This represents the total number of vehicles in the park. This refers to the actual number of vehicles replaced. In building energy-saving renovation technologies The total building area of ​​the park This refers to the actual area to be renovated. In ecological carbon sequestration technologies, This refers to the area of ​​the park that can be greened. This refers to the actual green area. In demand response technology, S_{k,i} represents the total interruptible load of the park, and S_{k,i} represents the actual load scale participating in demand response.

[0062] For other types of capacity-less technologies, the corresponding baseline scale parameters are determined based on their technical characteristics.

[0063] S5. Construct a dual-objective emission reduction path optimization model This invention constructs a dual-objective optimization model, the objectives of which include: Objective 1: Maximize total carbon emission reductions within the planning period Formula 4 Where Y is the planning period in years, and K is the number of candidate technologies. Let be the carbon emission reduction generated by the k-th technology in year i. For technologies with capacity: Formula 5 in, This refers to the annual equivalent output or emission reduction coefficient per unit capacity. This corresponds to the energy or emission factor.

[0064] For capacity-less technologies: Formula 6 in, Emission reduction coefficients are applied on a per-unit scale.

[0065] Objective 2: Minimize the overall cost within the planning period Formula 7 in: For investment costs; Operating costs; To maintain costs; Costs related to carbon trading, carbon offsetting, or carbon assets.

[0066] For technologies with capacity: Formula 8 For capacity-less technologies: Formula 9 S6. Constructing a constraint system This invention establishes the following constraints: (1) Carbon balance constraints Formula 10 in, This represents the baseline carbon emissions for the park in year i. To reduce emissions through technology, Carbon sequestration amount External carbon offset amount Let be the target carbon emissions for year i.

[0067] For zero-carbon industrial parks, the following can be further set up: Formula 11 (2) Constraints on energy supply and demand balance For each time period t, establish the balance relationship of multiple energy flows such as electricity, heat, cold, and hydrogen: Formula 12 in, Contribute to the source side of the park For purchased electricity or external energy input, For energy storage discharge capacity, To meet load demand, For energy storage charging capacity, This refers to system losses.

[0068] (3) Technology scale boundary constraints

[0069]

[0070]

[0071] (4) Constraints on Energy Storage Operation

[0072]

[0073] It also sets mutual exclusion constraints for charging and discharging to prevent simultaneous charging and discharging at the same time.

[0074] (5) Constraints on the ratio of source-grid-load-storage interaction Based on the source-grid-load-storage structure type of the target industrial park, set interaction ratio constraints. For example: In energy-dominated industrial parks, constraints on the renewable energy consumption rate can be set:

[0075] In load-dominated industrial parks, constraints on the proportion of flexible load participation can be set:

[0076] In energy storage-dominated industrial parks, constraints on the proportion of energy storage capacity can be set:

[0077] In grid-dominated industrial parks, a coverage constraint for source-grid-load-storage coordinated scheduling can be set:

[0078] S7. Design an improved multi-objective discrete particle swarm optimization algorithm To adapt to discrete decision-making processes such as technology selection and source-grid-load-storage interaction models, this invention designs an improved multi-objective discrete particle swarm optimization algorithm. This algorithm is used to solve the aforementioned dual-objective mixed-variable optimization model.

[0079] (1) Construct a particle encoding structure that adapts to mixed variables.

[0080] Given that the model of this invention simultaneously includes discrete variables such as technology selection and source-grid-load-storage interaction patterns, as well as continuous variables such as technology penetration rate, each particle is represented as a complete emission reduction path scheme for the industrial park, and its particle position is represented as follows:

[0081] in, This is the technology selection coding segment, used to indicate whether each low-carbon technology is selected; This is the encoding segment for the source-grid-load-storage interaction mode, used to represent interaction modes such as source-led, load-led, storage-led, or grid-led. This is the technology penetration rate coding segment, used to represent the implementation ratio of each low-carbon technology within the planning cycle.

[0082] (2) Establish an adaptive speed update mechanism.

[0083] Particle velocity is updated according to the following formula: Formula 13 in, Let be the velocity of the nth particle in the d-th dimension at the t-th iteration. For the corresponding position, For the optimal position of an individual, The global boot location selected from the external Pareto archive. , It is a random number.

[0084] To improve the algorithm's search and convergence capabilities, adaptive inertia weights and dynamic learning factors are set:

[0085]

[0086]

[0087] Where T is the maximum number of iterations. This setting allows the algorithm to focus on global search in the early stages and local convergence in the later stages.

[0088] (3) Discretely update the technology selection variables.

[0089] For technology selection variables A binary discrete update method is used to map particle velocities to the probability of a technology being selected.

[0090] When random number season: Otherwise, let:

[0091] This step is used to adapt to the discrete decision of "selecting / not selecting" low-carbon technologies.

[0092] (4) Perform multi-category discrete updates on the variables of the source-grid-load-storage interaction model.

[0093] For the variables of the source-grid-load-storage interaction mode A multi-class probability mapping approach is adopted. Suppose there are L interaction modes in the i-th planning stage, then the probability of selecting the l-th interaction mode is:

[0094] The interaction mode for the i-th planning stage is determined based on the maximum probability or roulette wheel method, enabling discrete selection of modes such as source-led, load-led, storage-led, and grid-led.

[0095] (5) Continuously update and correct the boundary of the technology penetration rate variable.

[0096] For the technology penetration rate variable Pk,i, a continuous update method is adopted, and boundary correction is performed on the update results:

[0097] Simultaneously, adjustments are made in conjunction with technology selection variables:

[0098]

[0099] in, For the installed capacity of capacity-enabled technologies, This refers to the physical implementation scale of technologies without capacity. This revised rule ensures that the unselected technologies will not result in non-zero penetration, installed capacity, or implementation scale.

[0100] (6) Calculate the fitness of the two objectives and the constraint violation.

[0101] After each iteration, calculate the bi-target fitness of the particle:

[0102] in, () represents the total carbon emission reduction during the planning period. This represents the comprehensive cost over the planning period. (Using...) This indicates the goal of maximizing carbon emission reductions.

[0103] Simultaneously calculate the constraint violation degree of the particles:

[0104] in, , , , , These represent the degree of violation of constraints related to carbon balance, energy supply and demand balance, technology scale boundaries, energy storage operation, and the ratio of source-grid-load-storage interaction. The particles are subjected to constraint repair or penalty processing.

[0105] (7) Create an external Pareto archive and output the optimization results.

[0106] An external Pareto archive is created to store non-dominated solutions. After each iteration, new particles are compared with existing particles in the archive for non-domination, dominated solutions are deleted and non-dominated solutions are retained. When the archive size exceeds a preset limit, the solution set is filtered according to the crowding distance, and the set of uniformly distributed Pareto optimal paths is retained.

[0107] Ultimately, based on the output of external Pareto archives, candidate emission reduction paths that balance maximizing carbon emission reductions and minimizing overall costs are identified, and optimal or near-optimal path schemes are formed for different source-grid-load-storage interaction scenarios.

[0108] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling, characterized in that, Includes the following steps: S1. Obtain basic data of the target park; S2. Construct a candidate library of low-carbon technologies: Based on whether the technology has a clearly measurable installed capacity attribute, candidate low-carbon technologies are divided into technologies with capacity and technologies without capacity. S3. Construct a unified variable system: Define the following variables for the k-th low-carbon technology in the i-th planning year: technology selection variable, technology penetration rate variable, installed capacity variable for technologies with capacity, and physical implementation scale variable for technologies without capacity. S4. Establish a penetration rate-physical scale coupling transformation model: For technologies with capacity, establish a mapping relationship between technology penetration rate and installed capacity; for technologies without capacity, establish a mapping relationship between technology penetration rate and physical implementation scale; and set linkage constraints between technology selection variables and penetration rate variables, installed capacity variables, or physical implementation scale variables to force the penetration rate and implementation scale of unselected technologies to zero. S5. Construct a dual-objective emission reduction path optimization model: With the optimization objectives of maximizing total carbon emission reduction and minimizing comprehensive cost within the planning period, construct a dual-objective mixed variable optimization model that is compatible with discrete and continuous variables; S6. Construct a constraint system; S7. Design an improved multi-objective discrete particle swarm optimization algorithm: construct a particle encoding structure adapted to mixed variables; establish an adaptive velocity update mechanism; adopt a binary discrete update method for technology selection variables, a multi-class probability mapping update method for source-grid-load-storage interaction mode variables, and a continuous update and boundary correction method for technology penetration rate variables; establish an external Pareto archive and output the optimization results.

2. The zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling according to claim 1, characterized in that, In step S3, The variables selected for the technology are represented as follows: ; in, This indicates that the k-th technology is selected in year i. This indicates that the technology will not be used in year i. The technology penetration rate variable is represented as follows: For technologies with capacity, the penetration rate represents the proportion of implementation relative to the maximum adaptable capacity; for technologies without capacity, the penetration rate represents the proportion of promotion relative to the benchmark size of the park. in, This represents the application rate of the k-th technology in year i. For technologies with capacity, the penetration rate represents the implementation rate relative to the maximum adaptable capacity; for technologies without capacity, the penetration rate represents the promotion rate relative to the benchmark size of the park. The installed capacity variable of the capacity technology mentioned: ; in, This represents the actual installed capacity or processing power of the k-th capacity technology in year i. The physical implementation scale variables of the capacity-free technology: ; in, This represents the actual implementation scale of the k-th capacity-less technology in year i.

3. The zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling according to claim 1, characterized in that, The penetration rate-installed capacity mapping relationship of the capacity technology mentioned in step S4 is as follows: ; in: Let k be the actual installed capacity of the k-th capacity-enabled technology in year i. Let k be the penetration rate of the k-th technology in year i. This represents the maximum adaptability capacity of the k-th technology within the target park. Let k be the selection variable for the k-th technology in year i; The linkage constraint is: ; when hour, and further enable This forces the penetration rate and installed capacity of unselected technologies to zero.

4. The zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling according to claim 1, characterized in that, The penetration rate-physical implementation scale mapping relationship of the capacity-free technology described in step S4 is as follows: ; in: Let k be the actual implementation scale of the capacity-free technology in year i. Let k be the benchmark size of the park corresponding to the k-th technology; Let be the penetration rate of the k-th technology in year i; The capacity-free technologies include electric vehicle replacement technology, building energy-saving retrofit technology, ecological carbon sink technology, and demand response technology. For electric vehicle replacement technology, the benchmark scale of the park is the total number of vehicles in the park, and the actual implementation scale is the actual number of vehicles replaced. For building energy-saving renovation technology, the benchmark scale of the park is the total building area of ​​the park, and the actual implementation scale is the actual renovation area. For ecological carbon sequestration technology, the baseline scale of the park is the greenable area of ​​the park, and the actual implementation scale is the actual green area; For demand response technology, the baseline scale of the park is the total interruptible load of the park, and the actual implementation scale is the actual load scale participating in demand response.

5. The zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling according to claim 1, characterized in that, The objective function of the dual-objective emission reduction path optimization model described in step S5 is: Objective 1: Maximize total carbon emission reductions within the planning period ； Where Y is the planning period in years, and K is the number of candidate technologies. Let be the carbon emission reduction generated by the k-th technology in year i; For technologies with capacity: ； in, This refers to the annual equivalent output or emission reduction coefficient per unit capacity. For corresponding energy or emission factors; For capacity-less technologies: ; in, Emission reduction coefficients for implementation scale; Objective 2: Minimize the overall cost within the planning period ； in: For investment costs; Operating costs; To maintain costs; Costs related to carbon trading, carbon offsetting, or carbon assets.

6. The zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling according to claim 1, characterized in that, The constraint system mentioned in step S6 includes: The carbon balance constraint is: ; in, This represents the baseline carbon emissions for the park in year i. To reduce emissions through technology, Carbon sequestration amount External carbon offset amount The target carbon emissions for year i; For zero-carbon industrial parks, the following can be further set up: ; Energy supply and demand balance constraints: For each time period t, establish the balance relationship of multiple energy flows of electricity, heat, cold and hydrogen; Technology Scale Boundary Constraints: ; ; ; Source-grid-load-storage interaction ratio constraint: set according to the source-grid-load-storage structure type of the target park, which includes source-dominant, load-dominant, storage-dominant and grid-dominant types.

7. The zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling according to claim 1, characterized in that, The particle encoding structure described in step S7 is represented as follows: ; in, This is the technology selection coding segment, used to indicate whether each low-carbon technology is selected; This is the encoding segment for the source-grid-load-storage interaction mode, used to represent interaction modes such as source-led, load-led, storage-led, or grid-led. This is the technology penetration rate coding segment, used to represent the implementation ratio of each low-carbon technology within the planning cycle.

8. The zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling according to claim 1, characterized in that, The adaptive speed update mechanism described in step S7 is as follows: ; in, Let be the velocity of the nth particle in the d-th dimension at the t-th iteration. For the corresponding position, For the optimal position of an individual, The global boot location selected from the external Pareto archive. , It is a random number; The adaptive inertia weights and dynamic learning factors are dynamically adjusted with the number of iterations, so that the algorithm focuses on global search in the early stage and local convergence in the later stage.

9. The zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling according to claim 1, characterized in that, In step S7: A binary discrete update method is used for the technology selection variable, mapping particle velocity to the probability of technology being selected: ； This technique should be used when the probability of generating a random number is less than this value; otherwise, it should not be used. The variables of the source-grid-load-storage interaction mode are updated using a multi-class probability mapping method. The particle velocity is mapped to the selection probability of each interaction mode through the softmax function, and the interaction mode is determined according to the maximum probability or the roulette wheel method. The technology penetration rate variable is continuously updated and its boundary is corrected, and the update results are corrected for the boundary: ； Simultaneously, combined with the technology selection variables, a linkage correction is performed to force the penetration rate, installed capacity, and implementation scale of unselected technologies to zero.

10. The zero-carbon industrial park emission reduction pathway planning method based on technology penetration rate coupling modeling according to claim 1, characterized in that, The external Pareto archive mentioned in step S7 is used to store non-dominated solutions. After each iteration, the new particles are compared with the existing particles in the external Pareto archive for non-domination, and dominated solutions are deleted while non-dominated solutions are retained. When the archive size exceeds the preset upper limit, the solution set is filtered according to the crowding distance, and the set of Pareto optimal paths with uniform distribution is retained. Finally, the candidate emission reduction paths that take into account both maximizing carbon emission reduction and minimizing comprehensive cost are output according to the external Pareto archive, and the optimal or near-optimal path schemes under different source-grid-load-storage interaction scenarios are formed.

Images