A multi-objective optimization configuration method for the entire green hydrogen production-transportation-storage-use chain in a province

Abstract

This invention proposes a multi-objective optimization configuration method for the entire green hydrogen production-transmission-storage-use chain at the provincial level, relating to the field of new energy configuration technology. The method includes: determining the number of provinces and the planning period for optimization configuration, and collecting data for each province; constructing a decision variable system and establishing a multi-objective optimization model; generating an initial solution population that meets physical constraints using an initialization method based on the time-series matching of renewable energy output and hydrogen demand; iteratively optimizing the initial solution population using an improved NSGA-III algorithm to obtain a Pareto front solution set; and determining the hydrogen production capacity, hydrogen storage capacity, hydrogen transmission channel configuration, and corresponding operation strategies for each province based on the Pareto front solution set, outputting a multi-objective optimization configuration scheme for the entire green hydrogen production-transmission-storage-use chain at the provincial level. This invention can provide a high-quality multi-objective optimization solution set for the planning and decision-making of the entire green hydrogen chain system at the provincial level.

Description

Technical Field

[0001] This invention relates to the field of new energy configuration technology, and in particular to a multi-objective optimization configuration method for the entire chain of green hydrogen production, transportation, storage and utilization in a province. Background Technology

[0002] Green hydrogen is hydrogen produced through water electrolysis using renewable energy electricity. Compared to hydrogen production from fossil fuels, it has advantages such as lower carbon emissions throughout its entire life cycle and the ability to be deeply coupled with fluctuating power sources like wind and solar. It is widely considered a key energy carrier for deep decarbonization. The production, transportation, storage, and end-use of green hydrogen at the provincial scale exhibit significant spatial and temporal coupling characteristics: on the one hand, wind and solar resources vary significantly across different provinces, and the layout of wind and solar power installations, along with inter-provincial power and hydrogen transmission networks, jointly determine the spatial flow of green hydrogen; on the other hand, hydrogen production, storage, and use change over time, and hydrogen storage equipment plays a "peak-shaving and valley-filling" role in time series, resulting in strong coupling of decision variables at different times through inventory status. Therefore, the optimized configuration of the entire green hydrogen production-transmission-storage-use chain at the provincial level needs to consider the coupling relationship between the power system and the hydrogen energy system simultaneously in space and time. Under the premise of satisfying hydrogen energy system constraints, electricity-hydrogen synergy constraints, power system constraints, and resource constraints, it is necessary to comprehensively optimize the entire life cycle cost and the level of renewable energy consumption.

[0003] Existing research often employs mixed-integer programming or evolutionary algorithms to optimize the hydrogen energy supply chain in a single province or region, focusing on localized issues such as hydrogen production station site selection and transportation mode choice, lacking a unified model for the entire supply chain across multiple provinces and time periods. In terms of multi-objective optimization solutions, most works directly apply general optimization algorithms without designing dedicated initialization strategies, genetic operators, and constraint handling mechanisms tailored to the spatiotemporal coupling and dual-balance characteristics of the green hydrogen system. This results in slow convergence speed, low feasible solution ratios, and uneven Pareto front distribution in high-dimensional, strongly constrained scenarios.

[0004] Chinese patent application CN115759360A discloses a two-stage optimization planning method for a wind-solar-hydrogen-storage coupled system. This method optimizes the electrolysis hydrogen production equipment and hydrogen storage tanks in two stages, establishing an operational state model for the electrolysis hydrogen production equipment and a capacity constraint model for the hydrogen storage tanks. With the goal of maximizing overall system benefits, it uses time-series simulation to simulate the power system's operation periodically, calculating the optimal ratio of hydrogen production capacity to hydrogen storage tank capacity. This patent reduces the scale of variables in a single optimization by using two-stage decomposition, thus improving calculation speed. However, the optimization object of this patent is limited to a single-region wind-solar-hydrogen-storage coupled system, and it does not address the problem of hydrogen energy transmission and coordinated configuration across multiple provinces. Its optimization method focuses on time-series simulation and staged solutions, making it difficult to directly apply to the multi-objective optimization configuration of the entire green hydrogen production-transmission-storage-use chain across provinces. Summary of the Invention

[0005] In view of this, the present invention provides a multi-objective optimization configuration method for the entire chain of green hydrogen production-transmission-storage-use in a province. By designing an initialization method based on the time-series matching of renewable energy output and hydrogen demand, a hybrid crossover operator that maintains the hydrogen storage state chain, an adaptive constraint penalty function mechanism based on a comprehensive imbalance index, and a reference point migration and update strategy, the method improves the convergence speed and feasible solution ratio of the algorithm in high-dimensional and strongly constrained scenarios, improves the coverage uniformity of the Pareto front, and provides a high-quality multi-objective optimization solution set for the planning and decision-making of the entire green hydrogen chain system in a province.

[0006] The technical solution of this invention is implemented as follows:

[0007] This invention provides a multi-objective optimization configuration method for the entire chain of green hydrogen production, transportation, storage, and utilization in a province, including:

[0008] S1. During the planning period for determining the number of provinces and optimizing their allocation, collect time-series data on renewable energy output, hydrogen demand, inter-provincial adjacency relationships, and technical parameters for hydrogen production, storage, and transportation for each province.

[0009] S2. Construct a decision variable system covering all aspects of hydrogen production, storage, transportation and use, and establish a multi-objective optimization model with total life cycle cost and renewable energy hydrogen production and consumption rate as multiple objectives, while satisfying the balance of electricity and hydrogen supply and demand and capacity constraints.

[0010] S3. An initialization method based on the time-series matching of renewable energy output and hydrogen demand is adopted to perform time-series allocation and perturbation of decision variables, thereby generating an initial solution population that meets the physical constraints.

[0011] S4. The improved NSGA-III algorithm is used to iteratively optimize the initial solution population. The hydrogen production, storage and transportation configurations of each province are coordinated and optimized through the time consistency maintenance strategy and constraint processing mechanism to obtain the Pareto front solution set that meets the preset termination conditions.

[0012] S5. Based on the Pareto frontier solution set, determine the hydrogen production capacity, hydrogen storage capacity, hydrogen transmission channel configuration and corresponding operation strategy for each province, and output a multi-objective optimization configuration scheme for the entire green hydrogen production-transmission-storage-use chain in the province.

[0013] Preferably, in step S2, the decision variable system includes spatial decision variables and temporal decision variables. Spatial decision variables include the installed capacity of each hydrogen production technology in each province, the upper limit of hydrogen storage installed capacity, the installed capacity marker of hydrogen production technology, and the allocation ratio of hydrogen storage technology. Temporal decision variables include the total hydrogen production in each province and time period, hydrogen storage inventory, hydrogen storage charging and discharging volume, technology allocation ratio, inter-provincial hydrogen transmission flow, hydrogen production electricity consumption, and conventional generator output. The objective function of the multi-objective optimization model includes minimizing the total cost over the entire life cycle and maximizing the renewable energy consumption rate. The constraints include hydrogen energy supply and demand balance constraints, hydrogen storage state equation, electricity supply and demand balance constraints, installed capacity constraints, hydrogen storage inventory boundary constraints, and hydrogen transmission flow capacity constraints.

[0014] Preferably, in step S3, the initialization method based on the time-series matching of renewable energy output and hydrogen demand uses a hydrogen production time-series allocation kernel function to allocate the decision variables in time series. The hydrogen production time-series allocation kernel function is defined as follows:

[0015] ;

[0016] In the formula, i is the provincial index; t and This is a time period index for the planning period; To contribute to renewable energy in province i during time period t; To meet the demand for hydrogen; For the weighting parameters of renewable energy; For hydrogen demand-side weighting parameters; This refers to the set of time periods within the planning period.

[0017] Preferably, step S4 includes:

[0018] S41. Perform a selection operation on the initial solution population or the previous generation population to generate a mating pool;

[0019] S42. The hybrid crossover operator is executed on individuals in the mating pool to maintain the hydrogen storage state chain and generate offspring. The hybrid crossover operator uses simulated binary crossover for spatial layer decision variables and inherits the time-series variables of the parent individuals alternately according to continuous time segments for temporal layer decision variables.

[0020] S43. Perform polynomial mutation on the offspring and evaluate the objective function and the overall imbalance index;

[0021] S44. After merging the parent and offspring generations, perform non-dominated sorting and environment selection. Adaptively adjust the strength of the constraint penalty function according to the average population imbalance. When the average population imbalance is higher than the expected value, increase the penalty coefficient to enhance the constraint penalty. When the average population imbalance is lower than the expected value, decrease the penalty coefficient to relax the constraint pressure.

[0022] S45. Perform migration update on the reference points every preset number of algebras, identify dense reference points with too many associated solutions and sparse reference points with too few associated solutions, and migrate dense reference points to sparse reference points to improve the uniformity of Pareto front coverage.

[0023] S46. Determine if the termination condition is met. If it is, output the first layer non-dominated solution set as the Pareto front solution set. Otherwise, return to S41 to continue iterating.

[0024] Preferably, in step S42, the cross-operation of the hybrid cross-operator on the time-level decision variables to maintain the hydrogen storage state chain includes:

[0025] The planning period is divided into several consecutive segments in chronological order, and the offspring inherit time-series variables from two parents alternately according to the parity of the segment numbers.

[0026] After crossover, the hydrogen production of the offspring is scaled proportionally to make the sum consistent with the new installed capacity. Starting from the initial period of the planning period, the inventory is calculated hourly based on the hydrogen storage state equation. If the inventory is negative at a certain period, the hydrogen release is reduced and the inventory is set to zero. If the inventory exceeds the upper limit of the installed capacity, the hydrogen charging is reduced and the inventory is set to the upper limit value.

[0027] Check whether each pair of adjacent provinces is simultaneously transmitting hydrogen in both directions during the same time period. If so, retain the net flow direction and set the reverse flow to zero.

[0028] Preferably, in step S43, the comprehensive imbalance index is used to quantify the degree of violation of the electro-hydrogen dual balance constraint by the candidate solution. The comprehensive imbalance index is defined as follows:

[0029] ;

[0030] In the formula, i is the provincial index; t is the time period index during the planning period; x represents the set of decision variables for candidate solutions; and N is the total number of provinces. For hydrogen energy balance residuals; For power balance residuals; and These are the scale normalization factors for hydrogen energy and electricity, respectively; and These are the weighting coefficients for hydrogen energy and electricity balance residuals, respectively; and The exponential parameter is not less than 1.

[0031] Preferably, in step S44, the method for adaptively adjusting the strength of the constraint penalty function includes: calculating the average imbalance of the g-th generation population, introducing the expected imbalance level as the convergence target, and updating the penalty coefficient by multiplying the current penalty coefficient by an exponential function, wherein the exponent term of the exponential function is the product of the adjustment coefficient and the proportion of the deviation of the average imbalance, increasing the penalty coefficient when the average imbalance is higher than the expected value, decreasing the penalty coefficient when the average imbalance is lower than the expected value, and applying upper and lower limit constraints to the updated penalty coefficient to ensure stability.

[0032] Preferably, in step S45, the reference point migration and update step includes:

[0033] The initial reference point is generated in the normalized target space using the Das-Dennis method. Every preset algebra, the normalized target vector of the first layer of non-dominated solution set is analyzed, and the number of solutions associated with each reference point is counted.

[0034] Calculate the population average number of correlated solutions and its standard deviation, and identify sparse reference points with a number of correlated solutions lower than the average minus one standard deviation and dense reference points with a number of correlated solutions higher than the average plus one standard deviation;

[0035] For each dense reference point, find the nearest sparse reference point, calculate the direction vector from the dense reference point to the sparse reference point, and set the migration step size according to the deviation ratio between the number of associated solutions of the dense reference point and the average number of associated solutions.

[0036] The updated reference point position is the original position plus the product of the step size and the direction vector, and boundary constraints are applied to the coordinates of the migrated reference point.

[0037] Preferably, the specific steps of the initialization method include:

[0038] For each province, Latin hypercube sampling is used to generate spatial layer decision variables, including the hydrogen production capacity, hydrogen storage capacity limit, and hydrogen storage technology ratio for each technology;

[0039] The hydrogen production volume of each province is allocated in the time dimension using the hydrogen production time-series allocation kernel function. The initial hydrogen production volume is set as the product of the upper bound of the total hydrogen production volume, the allocation kernel function, and the random disturbance term. The random disturbance term follows a normal distribution with a mean of zero.

[0040] The initial hydrogen storage charging and discharging volume is heuristically generated based on the deviation between hydrogen demand in each period and the average value of the planning period. Hydrogen charging is arranged when hydrogen demand is lower than the average value, and hydrogen discharging is arranged when hydrogen demand is higher than the average value.

[0041] The initial value of hydrogen transport flow is allocated according to the hydrogen supply and demand gap of each province in each time period. For provinces with oversupply, the excess hydrogen will be transported to adjacent provinces with shortages according to the inverse weight of distance.

[0042] The overall imbalance index is calculated for the generated initial individuals. If it exceeds the preset threshold, the hydrogen production, hydrogen storage and release, and hydrogen transportation flow rate are iteratively adjusted according to the sign and magnitude of the residuals until the threshold requirement is met or the maximum number of repair iterations is reached.

[0043] Preferably, the termination condition in step S46 includes any of the following conditions:

[0044] (1) Reach the maximum number of iterations;

[0045] (2) The hypervolume index of the first-layer non-dominated solution set changes less than the preset threshold for several consecutive generations, indicating that the Pareto front distribution has become stable;

[0046] (3) If the average imbalance of the population is lower than the expected imbalance level and the number of individuals in the first layer of non-dominated solution set does not exceed the preset percentage within several consecutive generations, it indicates that both the constraint satisfaction and frontier stability criteria have been met.

[0047] The present invention has the following advantages over the prior art:

[0048] This invention constructs a multi-objective optimization model for the entire green hydrogen supply chain across a province. It designs a hydrogen production time-series allocation kernel function, a hybrid crossover operator to maintain the hydrogen storage state chain, a comprehensive imbalance index, an adaptive constraint penalty function mechanism, and a reference point migration strategy, all tailored to the spatiotemporal coupling characteristics of the green hydrogen system. The hydrogen production time-series allocation kernel function adopts the product of renewable energy output and hydrogen demand, and achieves nonlinear enhancement through power parameters. This guides the initialization phase to prioritize the allocation of hydrogen production to periods of high supply and demand, reducing the generation of physically unreasonable solutions and increasing the proportion of feasible solutions in the initial population. The hybrid crossover operator to maintain the hydrogen storage state chain alternates between continuous time segments. The inheritance mechanism and the time-progressive repair mechanism for hydrogen storage avoid the disruption of temporal coupling by the standard crossover operator, reducing the proportion of constraint-violation offspring generated by genetic operations. The comprehensive imbalance index aggregates the electro-hydrogen dual-balance residuals and amplifies the degree of constraint violation through normalization and exponential forms, providing a quantitative basis for the adaptive penalty function. This enables the algorithm to dynamically adjust the constraint penalty intensity based on the population average imbalance, balancing constraint satisfaction and objective optimization. The reference point migration strategy identifies sparse and dense reference points based on the number of associated solutions and performs directional migration, improving the coverage uniformity of the Pareto front under complex shapes and increasing the diversity of the solution set. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart of the method of the present invention;

[0051] Figure 2 This is a hierarchical structure diagram of the decision variable system of the present invention;

[0052] Figure 3 This is a schematic diagram of the optimization algorithm flow of the present invention;

[0053] Figure 4 This is a schematic diagram of the time layer operation of the hybrid crossover operator of the present invention;

[0054] Figure 5 This is a schematic diagram of the adaptive penalty coefficient adjustment mechanism of the present invention;

[0055] Figure 6 This is a schematic diagram of the reference point migration strategy of the present invention. Detailed Implementation

[0056] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0057] like Figure 1 As shown, this invention provides a multi-objective optimization configuration method for the entire chain of green hydrogen production, transportation, storage, and utilization in a province, including:

[0058] S1. During the planning period for determining the number of provinces and optimizing their allocation, collect time-series data on renewable energy output, hydrogen demand, inter-provincial adjacency relationships, and technical parameters for hydrogen production, storage, and transportation for each province.

[0059] S2. Construct a decision variable system covering all aspects of hydrogen production, storage, transportation and use, and establish a multi-objective optimization model with total life cycle cost and renewable energy hydrogen production and consumption rate as multiple objectives, while satisfying the balance of electricity and hydrogen supply and demand and capacity constraints.

[0060] S3. An initialization method based on the time-series matching of renewable energy output and hydrogen demand is adopted to perform time-series allocation and perturbation of decision variables, thereby generating an initial solution population that meets the physical constraints.

[0061] S4. The improved NSGA-III algorithm is used to iteratively optimize the initial solution population. The hydrogen production, storage and transportation configurations of each province are coordinated and optimized through the time consistency maintenance strategy and constraint processing mechanism to obtain the Pareto front solution set that meets the preset termination conditions.

[0062] S5. Based on the Pareto frontier solution set, determine the hydrogen production capacity, hydrogen storage capacity, hydrogen transmission channel configuration and corresponding operation strategy for each province, and output a multi-objective optimization configuration scheme for the entire green hydrogen production-transmission-storage-use chain in the province.

[0063] In one embodiment of the present invention, step S1 includes:

[0064] The total number of provinces N involved in the optimization problem is determined, and the planning period is discretized into a set of time periods. Where T is the total number of time periods;

[0065] Collect wind power output of each province i during time period t. With photovoltaic power The sum of the two is recorded as renewable energy contribution. ,in Collect power load data for each province. Total hydrogen demand ;

[0066] Obtain the upper and lower limits of the installed capacity and output of conventional generators in each province, denoted as set. And its related parameters, defining renewable energy installed capacity and total installed capacity ;

[0067] Hydrogen production technologies include alkaline electrolyzers, proton exchange membrane electrolyzers, and solid oxide electrolyzers, denoted as a collection. Each technology The unit installed capacity investment cost is The unit hydrogen production operation and maintenance cost is Electrolysis efficiency is Technological and economic life is ;

[0068] Hydrogen storage technologies include gaseous hydrogen storage and liquid hydrogen storage, denoted as a set. Each hydrogen storage technology The unit capacity investment cost is The annual maintenance cost per unit capacity is Technological and economic life is The lower heating value of hydrogen is... The unit is kWh / kg. The water consumption per unit of hydrogen production by water electrolysis is... The unit is kg-H2O / kg-H2, and the total available water resources in each province during the planning period are... The discount rate is ;

[0069] Constructing a provincial spatial adjacency matrix , of which elements This indicates that province i and province j are adjacent. This indicates that they are not adjacent. The inter-provincial hydrogen transport distance matrix is ​​as follows: Let the potential upper limit of hydrogen transport capacity between adjacent provinces be... The cost of transporting hydrogen per unit distance and per unit mass of pipeline is The unit hydrogen transport loss rate is The equipment investment and operation and maintenance cost per unit of hydrogen consumption for end-use hydrogen equipment is... The lower and upper limits for the wind-solar-hydrogen storage ratio are respectively... , The target for the proportion of renewable energy installed capacity is ;

[0070] In addition, time-series electricity price data were collected from various provinces and time periods. The total number of years in the plan is For conventional generator sets, record the data for each unit. Lower limit of output With upper limit of output .

[0071] like Figure 2 As shown, in one embodiment of the present invention, step S2 includes:

[0072] The spatial layer decision variables include: the installed capacity of hydrogen production technology k in province i. The hydrogen production capacity is measured in kg / h; the upper limit of hydrogen storage capacity in province i. In kg; technical installation binary labeled variable When a province invests in the construction of hydrogen production equipment using technology k Otherwise, it is 0; hydrogen storage technology ratio This indicates the proportion of hydrogen storage capacity in province i using hydrogen storage technology h to the total hydrogen storage capacity of the province, satisfying the requirement... ;

[0073] The time-level decision variables are defined as follows: the total hydrogen production in province i during time period t. In kg; the proportion of hydrogen production using technology k in the total hydrogen production during the i-th time period of a province. ,satisfy and Hydrogen storage capacity in the province during time period t Hydrogen storage charge and release amount All figures are in kg; inter-provincial hydrogen transport flow rate This represents the mass of hydrogen transported from province i to province j during time period t, in kg; the electricity consumption for hydrogen production. In kWh; output of conventional generator sets Power input across provinces, measured in kW. With output power , in kW;

[0074] Technology selection and timing allocation satisfy logical constraints: if Then there must be The installed capacity of each technology is linked to its maximum output during the planning period, thereby constraining... To ensure that the installed capacity is sufficient to support the actual hydrogen production demand, among which This refers to the number of hours in a single time period. The total hydrogen production capacity of each province during the planning period is... .

[0075] The objective functions of the multi-objective optimization model include minimizing the total life-cycle cost and maximizing the renewable energy integration rate.

[0076] Total lifecycle cost objective function This includes investment costs for hydrogen production equipment, operation and maintenance costs for hydrogen production, electricity costs for hydrogen production, investment and operation and maintenance costs for hydrogen storage equipment, hydrogen transportation costs, and end-use costs for hydrogen.

[0077] The investment cost of hydrogen production equipment, after being amortized over its economic life over all provinces and technologies, is defined as:

[0078] ;

[0079] Hydrogen production operation and maintenance costs are calculated as a weighted average of actual hydrogen production output and the operation and maintenance costs of each technical unit, and are defined as follows:

[0080] ;

[0081] The electricity cost for hydrogen production is determined by the amount of electricity consumed and the electricity price at different times, where the electricity price for time period t in province i is... The electricity consumption for hydrogen production is calculated based on hydrogen production, lower heating value, and technical efficiency.

[0082] ;

[0083] Electricity costs can be expressed as:

[0084] ;

[0085] The investment and operation and maintenance costs of hydrogen storage equipment are amortized annually based on the installed capacity and are defined as follows:

[0086] ;

[0087] Hydrogen transportation cost, taking into account the transportation mileage and flow rate of pipeline hydrogen transport between adjacent provinces, is defined as:

[0088] ;

[0089] The cost of hydrogen end-use, calculated based on the total hydrogen consumption, can be expressed as:

[0090] ;

[0091] The objective function for total life-cycle cost at the discount rate The present value of each year in the planning period is then summed using discounted cash flow. The total number of planning years is: Defined as:

[0092] ;

[0093] The subscript y indicates that the costs for year y are accumulated over time periods.

[0094] Renewable energy consumption rate objective function Defined as the ratio of electricity consumed in hydrogen production to total renewable energy electricity, expressed as:

[0095] .

[0096] In this embodiment, the constraint system includes four main categories.

[0097] The first type is the hydrogen energy system constraint. For each province i and time period t, the hydrogen energy supply and demand balance can be expressed as:

[0098] ;

[0099] The hydrogen storage equation of state is used to characterize the time-coupled relationship of the storage, and is defined as:

[0100] ;

[0101] Initial inventory It can be set to zero or given according to the actual situation, and should satisfy the following: and , ;

[0102] Water resource constraints are defined as:

[0103] ;

[0104] Hydrogen transport flow direction constraints require that each pair of adjacent provinces cannot simultaneously transport hydrogen in both directions at the same time period, through logical constraints. The expression is implemented through genetic operations followed by repair during algorithm execution. The hydrogen transport capacity constraint is... .

[0105] The second category is the constraint of synergistic interaction between electricity and hydrogen. The constraint on the efficiency of the electricity-to-hydrogen conversion is given by the aforementioned formula for hydrogen production and electricity consumption. The constraint on green hydrogen consumption and green electricity limits the electricity consumption for hydrogen production to not exceed the output of renewable energy in the province, that is... .

[0106] The third category is power system constraints. The power supply and demand balance of province i in time period t can be expressed as:

[0107] ;

[0108] in Energy is converted into power. The output of conventional units should meet upper and lower limit constraints. .

[0109] The fourth category is resource constraints. The wind-solar-storage-hydrogen ratio constraint for each province i is defined as:

[0110] ;

[0111] The constraint on the proportion of renewable energy installed capacity is: .

[0112] In one embodiment of the present invention, step S3 employs an initialization method based on the time-series matching of renewable energy output and hydrogen demand. This method utilizes a hydrogen production time-series allocation kernel function to allocate decision variables according to time series. The hydrogen production time-series allocation kernel function is defined as follows:

[0113] ;

[0114] In the formula, i is the provincial index, t and For the time period index during the planning period, To contribute to renewable energy in province i during time period t; To meet the demand for hydrogen; For the weighting parameters of renewable energy; For hydrogen demand-side weighting parameters; This refers to the set of time periods during the planning period;

[0115] The kernel function satisfies And when the renewable energy output in a certain period t With hydrogen demand At the same time, when the weight is high, the weight of that period is... Larger, and vice versa;

[0116] The kernel function uses a product form instead of a weighted sum. When either renewable energy output or hydrogen demand is zero in a given period, the weight for that period approaches zero, avoiding the allocation of hydrogen production tasks during meaningless periods and ensuring the physical rationality of the timing scheme. This is achieved through power-law parameters. and To achieve nonlinear enhancement, when the power is greater than 1, the weighting is concentrated in the period of high supply and high demand, so that hydrogen production is preferentially allocated to the period with the highest matching degree between resources and demand, which is in line with the actual operation law.

[0117] Weight parameters and The choice reflects decision-making preferences: if the policy places more emphasis on the local consumption of renewable energy, then [the following option] can be chosen. If a greater emphasis is placed on ensuring a stable supply of hydrogen for the load, then... If both are considered in balance, then we can choose... ;

[0118] In the initialization phase, Latin hypercube sampling is first used for each province i to generate spatial layer decision variables, including the hydrogen production capacity of each technology. Upper limit of hydrogen storage capacity and hydrogen storage technology ratio The sampling range was uniformly covered based on the upper and lower limits determined by the preliminary techno-economic analysis. The upper limit of the total hydrogen production for each province was calculated based on the installed capacity, denoted as... ;

[0119] Then, the hydrogen production timing assignment kernel function is used. The hydrogen production capacity of province i is allocated along the time dimension, with the initial hydrogen production capacity set as follows:

[0120] ;

[0121] in To conform to a mean of 0 and a standard deviation of The random perturbation term of the normally distributed kernel is used to introduce moderate diversity while maintaining the backbone structure of the allocation kernel function;

[0122] The initial hydrogen storage charging and discharging volume is heuristically generated based on the deviation between hydrogen demand and average demand in each period: hydrogen charging is prioritized during periods when hydrogen demand is lower than the daily average of the planning period, and hydrogen discharging is arranged during periods when hydrogen demand is higher than the daily average. The initial value of the charging and discharging volume can be set according to the proportion of demand deviation, and the initial hydrogen storage inventory sequence is updated recursively starting from t=1, with boundary correction ensuring... ;

[0123] The initial hydrogen transport flow rate is allocated based on the hydrogen supply and demand gap of each province in each time period. For provinces with oversupply, the excess hydrogen is transported to adjacent provinces with shortages according to an inverse distance weighting. For provinces that still cannot achieve balance, adjustments are made by regulating the hydrogen storage and release capacity or scaling up hydrogen production. (Technology allocation ratio) The initial value can be set according to the proportion of installed capacity of each technology. If multiple technologies are built in province i, then... If a certain technology is not developed, then the corresponding ;

[0124] Finally, the overall imbalance index is calculated for the generated initial individuals. ,like Exceeding the preset threshold The hydrogen production rate, hydrogen storage and discharge rate, and hydrogen transport flow rate are then adjusted iteratively based on the sign and magnitude of the residuals, and a maximum number of repair iterations is set. ,For example In each round of repair: for provinces and time periods with positive hydrogen balance residuals, priority is given to increasing hydrogen storage input, decreasing hydrogen production, or increasing output flow; for provinces and time periods with negative residuals, priority is given to increasing hydrogen storage release, increasing hydrogen production, or increasing input flow. The adjustment magnitude decreases with the number of iterations to enhance convergence stability. Recalculation is performed after each round of adjustment. ,like The repair process will terminate if the maximum number of iterations is reached. If the problem persists after repair... Significantly exceeding the threshold, for example If the individual is not found, it is discarded and a new one is generated to prevent severely infeasible solutions from entering the initial population.

[0125] The initial population generated by the above-mentioned initialization method based on the time-series allocation kernel function reflects the simultaneity of wind and solar power output and hydrogen demand in the time dimension. It is physically closer to the feasible region, reduces the generation of unreasonable solutions, increases the proportion of feasible solutions in the initial population, and is conducive to the rapid convergence of subsequent genetic operations.

[0126] like Figure 3 As shown, in one embodiment of the present invention, step S4 uses an improved NSGA-III algorithm to iteratively optimize the initial solution population. This algorithm includes operations such as selection, crossover, mutation, non-dominated sorting, environment selection, adaptive constraint processing, and reference point migration.

[0127] Step S41 performs a selection operation on the initial solution population or the previous generation population to generate a mating pool. A tournament selection is performed based on the non-dominated sorting hierarchy and crowding, starting from the current population. Select Each parent individual forms a mating pool ;in, Indicates population size.

[0128] Step S42 executes a hybrid crossover operator on the individuals in the mating pool to generate offspring while maintaining the hydrogen storage state chain. In this invention, the decision variables are divided into two levels based on spatiotemporal relationships during the crossover stage. The spatial layer variables include the hydrogen production capacity of various technologies in different provinces. Upper limit of hydrogen storage capacity and technical installation markings Matching with hydrogen storage technology Time-level variables include hydrogen production per time period. Hydrogen storage , charge / discharge , Technology allocation ratio and hydrogen transport flow rate ;

[0129] At the spatial level, the hybrid crossover operator employs simulated binary crossover for continuous installed capacity variables, with the distribution index taking typical values ​​such as... For binary labeled variables Using random inheritance or a logical OR operation, offspring can inherit one or more combinations of techniques from their parents. Specifically, for each (i,k) combination, with probability... Inherited randomly from parent 1 or parent 2, with probability. Execute the logical OR operation, i.e. ,in 0.5 is acceptable;

[0130] like Figure 4 As shown, at the time level, time-level decision variables inherit the time-series variables of parent individuals alternately in continuous time segments. To avoid disrupting the hydrogen storage state chain and seasonal structure, the entire planning period (containing T fine-grained time periods) is divided into several continuous segments in chronological order, with the number of segments... It can be set according to the length of the planning period, for example, divided by seasons. Divided by month Each sub-segment contains a series of consecutive time periods such that the sum of the time periods in all sub-segments equals T. For example, when T = 8760 (annual planning period discrete in hours) and divided into 4 sub-segments by season, the first sub-segment contains time periods 1 to 2190, the second sub-segment contains time periods 2191 to 4380, and so on. For child generation 1, it inherits all time-series variables from all time periods in odd-numbered sub-segments from parent generation 1, and inherits all time-series variables from all time periods in even-numbered sub-segments from parent generation 2; for child generation 2, the reverse inheritance method is used.

[0131] After crossover, the hydrogen production of the offspring generation is proportionally scaled to match the total hydrogen production of the new installed capacity. First, the total hydrogen production of each province's offspring generation is corrected to match the total hydrogen production supported by the new installed capacity. This is achieved through proportional scaling:

[0132] ;

[0133] in This is the target value for total hydrogen production calculated based on the new installed capacity;

[0134] Starting from the initial planning period t=1, the inventory is calculated hourly based on the hydrogen storage state equation:

[0135] ;

[0136] Initial inventory Set to zero. Once it is detected that inventory exceeds the limit for a certain period... If the inventory is negative for a certain period, then the following repair will be performed: This indicates that the hydrogen release is too large, so reduce the hydrogen release and reduce the inventory to zero. , If the inventory exceeds the installed capacity limit, that is... This indicates that the hydrogen charging amount is too high. Therefore, reduce the hydrogen charging amount and set the inventory to the upper limit. , Continue recursively until t=T to complete the full-time repair;

[0137] The hydrogen transport flow rate is checked to see if each pair of adjacent provinces is simultaneously transporting hydrogen bidirectionally within the same time period. Specifically, it is checked whether each pair of adjacent provinces (i,j) exists simultaneously within the same time period t. and If it exists, retain the direction of net flow and set the reverse flow to zero, then calculate the net flow. ,like Then retain Direction, , ;like Then retain Direction, , ;like Then both are set to zero;

[0138] By combining the aforementioned spatial-temporal separation crossover with state chain and directional constraint repair, the hybrid crossover operator not only preserves the temporal structure characteristics of the parent individuals in different seasonal periods, but also ensures that the offspring meet the physical feasibility in hydrogen storage and transportation directions. This avoids the standard crossover operator from destroying the temporal coupling relationship and reduces the proportion of offspring that violate constraints generated by genetic operations.

[0139] Step S43 performs polynomial mutation on the offspring and evaluates the objective function and the overall imbalance index. For each offspring individual, a probability... Perform polynomial mutation, with the mutation distribution index taking typical values ​​such as... After mutation, boundary checks are performed on the decision variables, and those exceeding the upper and lower limits are resampled or pruned to the boundary.

[0140] The comprehensive imbalance index is used to quantify the degree to which candidate solutions violate the electro-hydrogen dual balance constraint, and is defined as follows:

[0141] ;

[0142] In the formula, i is the provincial index; t is the time period index during the planning period; x represents the set of decision variables for candidate solutions; and N is the total number of provinces. For hydrogen energy balance residuals; For power balance residuals; and These are the scale normalization factors for hydrogen energy and electricity, respectively; and These are the weighting coefficients for hydrogen energy and electricity balance residuals, respectively; and An exponential parameter of not less than 1;

[0143] For each province i and time period t, the hydrogen energy balance residual is defined as:

[0144] ;

[0145] The power balance residual is defined as:

[0146] ;

[0147] The scale normalization factor is defined as and ,in and To prevent small positive numbers from being divided by zero, for example The weighting coefficients can be set according to the importance of the two types of constraints in the planning process, for example, taking... The exponential parameter is used to amplify the effect of larger residuals, and can be taken as... ;

[0148] This indicator aggregates the hydrogen-electric balance deviation across all provinces and time periods, possessing uniform dimensions and clear physical meaning. It is used in subsequent algorithms to adaptively adjust the penalty function strength and the genetic operator's repair intensity. This applies to all offspring individuals. Evaluation objective function and imbalance index.

[0149] Step S44 involves merging the parent and offspring generations and then performing non-dominated sorting and environmental selection. The parent population... With offspring population Merge into candidate set ,right Perform a non-dominated sort. From Individuals are added sequentially from the lowest to the highest non-dominated hierarchy, until a certain level is reached. The addition of [something] will cause the total to exceed [a certain number]. At this point, the distance from each individual in the normalized target space to each reference point is calculated. Based on the number of reference points associated with each reference point and the crowding degree, a niche retention selection is performed, prioritizing the retention of individuals associated with reference points with fewer associated reference points, thus forming a new generation of population.

[0150] The strength of the constraint penalty function is adaptively adjusted based on the average imbalance of the population. This invention employs a hybrid strategy of hard repair and soft penalty function. Constraints such as the hydrogen storage state chain and hydrogen transport direction, which can be directly repaired through local adjustments, are forcibly satisfied using the repair operator in step S42. Residuals and resource constraints in the hydrogen-electricity balance equation are incorporated into the overall fitness through the penalty function. The penalty value is defined as... ,in The penalty coefficient;

[0151] like Figure 5 As shown, this invention adaptively adjusts the penalty coefficient based on the average imbalance of the population over the most recent generations. Let the average imbalance of the population in generation g be... ,in For the g-th generation population, introduce the expected imbalance level. As a convergence target;

[0152] When the average population imbalance is higher than the expected value, the penalty coefficient is increased to strengthen the constraint punishment; when the average population imbalance is lower than the expected value, the penalty coefficient is decreased to relax the constraint pressure. The penalty coefficient update rule is to multiply the current penalty coefficient by an exponential function, defined as:

[0153] ;

[0154] The exponential term of the exponential function is the product of the adjustment coefficient and the proportion of the average imbalance deviation, where As an adjustment coefficient, a smaller value can be taken, such as... , To prevent division by zero for small positive numbers, this update rule implements an adaptive constraint handling mechanism: when the average population imbalance... Higher than the target value When the exponent term is positive, the penalty coefficient is... Increasing the size of the population strengthens the punishment for individuals who violate the constraints, prompting the population to move towards the feasible region; when When the value falls below the target value, the exponential term becomes negative, the penalty coefficient decreases, the constraint pressure is relaxed, and the population is encouraged to explore a better target value within the feasible region. The exponential form achieves a smooth adjustment of the penalty coefficient, avoiding oscillations caused by step adjustments.

[0155] To ensure stability, upper and lower bound constraints are applied to the updated penalty coefficients, expressed as follows:

[0156] ;

[0157] in and It can be set according to the size of the problem, for example ;

[0158] When ranking individuals, the degree of constraint satisfaction, i.e., the degree of imbalance, is compared first. Smaller individuals are prioritized over those with greater imbalance. For feasible solutions with similar imbalance, the objective function values ​​are then compared. Through the adaptive penalty coefficient adjustment based on the bi-balance imbalance index, the algorithm can dynamically adjust the constraint satisfaction pressure according to the population evolution state, balancing constraint satisfaction and objective optimization.

[0159] Step S45 performs a migration update on the reference point every preset number of generations. For example... Figure 6 As shown, this invention introduces a reference point migration strategy based on congestion and target spatial distribution, on the basis of standard NSGA-III uniform reference point generation.

[0160] The initial reference points are generated in the normalized target space using the Das-Dennis method. Let the total number of reference points be . Every preset algebra, the normalized objective vector of the first-level non-dominated solution set is analyzed, and the number of solutions associated with each reference point is counted. It is then determined whether the current algebra g+1 is an integer multiple of the reference point update interval. If so, then for the first-level non-dominated solution set Analyze the normalized target vector and statistically analyze each reference point. Number of associated solutions The solution x is related to the reference point. If and only if It is the distance normalization target The nearest reference point;

[0161] Calculate the average number of correlated solutions in the population and its standard deviation. and its standard deviation ;

[0162] Identify sparse reference points with too few associated solutions and dense reference points with too many associated solutions. Identify sparse reference point sets with fewer than one standard deviation below the average number of associated solutions. and a dense set of reference points with a number of correlated solutions exceeding the mean plus one standard deviation. ;

[0163] Migrate dense reference points toward sparse reference points to improve Pareto front cover uniformity. For each dense reference point, find the nearest sparse reference point. Find the nearest sparse reference point. ;

[0164] Calculate the direction vector from the dense reference point to the sparse reference point. ;

[0165] The migration step size is set according to the deviation ratio between the number of associated solutions at dense reference points and the average number of associated solutions; the migration step size is set to... ,in As the reference step size parameter, it can be taken as follows: , To prevent small positive numbers from being divided by zero;

[0166] The updated reference point position is the original position plus the product of the step size and the direction vector, which can be written as: Furthermore, boundary constraints are applied to the coordinates of the migrated reference points. To ensure that the reference points remain within a reasonable target space region, boundary constraints are applied to the coordinates of the migrated reference points to keep each component within the interval [0, 1.5].

[0167] By employing the aforementioned reference point migration strategy, the algorithm identifies sparse and dense reference points based on the number of associated solutions and performs directional migration, thereby improving the coverage uniformity of the Pareto front under complex shapes and enhancing the diversity of the solution set.

[0168] Step S46 determines whether the termination condition is met. If it is, the first-level non-dominated solution set is output as the Pareto front solution set; otherwise, it returns to S41 to continue iteration. The algorithm terminates iteration when either of the following conditions is met: the maximum number of iterations is reached, i.e. The change in the hypervolume index of the first-level non-dominated solution set is less than a preset threshold for several consecutive generations, indicating that the Pareto front distribution has stabilized. The average population imbalance is lower than the expected imbalance level, and the change in the number of individuals in the first-level non-dominated solution set does not exceed a preset percentage over several consecutive generations, indicating that both the constraint satisfaction and front stability criteria have been met, such as the average imbalance. Furthermore, the number of individuals in the first-level non-dominated solution set does not vary by more than 5% over several consecutive generations. If no termination condition is met, then let... Continue the loop;

[0169] The algorithm outputs the final generation population after it terminates. The first level of non-dominated solution set As the Pareto optimal solution set;

[0170] Through the improved NSGA-III algorithm described above, this invention makes three core improvements to address the spatiotemporal coupling characteristics of green hydrogen systems: First, it constructs a hydrogen production time-series allocation kernel function to guide initialization and repair, improving the quality of the initial population and evolutionary efficiency; second, it designs a hybrid crossover operator to maintain the hydrogen storage state chain, preserving temporal coupling and physical constraints during genetic operations; and third, it establishes an adaptive constraint handling and reference point adjustment mechanism based on the electro-hydrogen dual-balance imbalance index, achieving dynamic balance constraint satisfaction and multi-objective convergence. These improvements enable the algorithm to exhibit better convergence performance and constraint handling capabilities in high-dimensional, strongly constrained scenarios.

[0171] In one embodiment of the present invention, step S5 outputs a multi-objective optimized configuration scheme for the entire chain of green hydrogen production, transportation, storage and utilization in the province based on the Pareto front solution set;

[0172] Each individual solution in this solution set corresponds to a provincial-level green hydrogen full-chain configuration plan, including the province's hydrogen production capacity, hydrogen storage capacity, technology selection combinations, time-based hydrogen production volume, hydrogen storage and release plans, inter-provincial hydrogen transportation flow arrangements, and the corresponding life-cycle costs and renewable energy integration rates. Decision-makers can select a satisfactory solution from the Pareto frontier based on cost budgets, integration rate targets, policy preferences, etc., or further screen compromise solutions through multi-criteria decision-making methods such as the entropy weight method and the approximation of ideal solution ranking method.

[0173] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A provincial green hydrogen production, transportation, storage and use full-chain multi-objective optimization configuration method, characterized in that, The method comprises the following steps: S1, determining the number of provinces and the planning period of optimization configuration, collecting renewable energy output time series data, hydrogen demand time series data, inter-provincial adjacency relationship data and technical parameters of hydrogen production, hydrogen storage and hydrogen transportation of each province; S2, constructing a decision variable system covering hydrogen production, hydrogen storage, hydrogen transportation and hydrogen use, and establishing a multi-objective optimization model with total life cycle cost and renewable energy hydrogen production consumption rate as multiple objectives, and meeting the balance of electricity and hydrogen supply and demand and capacity constraints; In step S2, the decision variable system includes spatial layer decision variables and time layer decision variables, the spatial layer decision variables include the installed capacity of each hydrogen production technology in each province, the upper limit of the installed capacity of hydrogen storage, the installed capacity mark of hydrogen production technology and the proportion of hydrogen storage technology, the time layer decision variables include the total amount of hydrogen production, hydrogen storage inventory, hydrogen storage charging and discharging amount, technology allocation proportion, cross-provincial hydrogen transportation flow, hydrogen production electricity consumption and conventional generator set output of each province in each period, the objective function of the multi-objective optimization model includes minimization of total life cycle cost and maximization of renewable energy consumption rate, and the constraint conditions include hydrogen energy supply and demand balance constraint, hydrogen storage state equation, power supply and demand balance constraint, installed capacity constraint, hydrogen storage inventory boundary constraint and hydrogen transportation flow capacity constraint; S3, using an initialization method based on the matching of renewable energy output and hydrogen demand time series to perform time series allocation and disturbance on the decision variables to generate an initial solution population meeting the physical constraint requirements; S4, using an improved NSGA-III algorithm to iteratively optimize the initial solution population, and using a time series consistency maintenance strategy and a constraint processing mechanism to cooperatively optimize the hydrogen production, hydrogen storage and hydrogen transportation configuration of each province to obtain a Pareto frontier solution set meeting the preset termination condition; Step S4 comprises: S41, performing a selection operation on the initial solution population or the last generation population to generate a mating pool; S42, performing a mixed crossover operator that maintains the hydrogen storage state chain on the individuals in the mating pool to generate offspring, the mixed crossover operator uses a simulated binary crossover on the spatial layer decision variables and alternately inherits the time series variables of the parent individual according to the continuous time sub-section on the time layer decision variables; S43, performing a polynomial mutation on the offspring and evaluating the objective function and the comprehensive imbalance index; S44, combining the parent and offspring and performing non-dominated sorting and environmental selection, adaptively adjusting the constraint penalty function strength according to the average imbalance degree of the population, increasing the penalty coefficient to enhance the constraint punishment when the average imbalance degree of the population is higher than the expected value, and reducing the penalty coefficient to relax the constraint pressure when the average imbalance degree of the population is lower than the expected value; S45, migrating and updating the reference points every preset number of generations, identifying dense reference points with too many associated solutions and sparse reference points with too few associated solutions, and migrating the dense reference points to the sparse reference points to improve the uniformity of the Pareto frontier coverage; S46, determining whether the termination condition is met, if yes, outputting the first layer non-dominated solution set as the Pareto frontier solution set, otherwise returning to S41 for iteration; In step S43, the comprehensive imbalance index is used to quantify the violation degree of the candidate solution in the balance constraint of electricity and hydrogen, and the comprehensive imbalance index is defined as: ； where i is the province index; t is the time period index of the planning horizon; x represents the set of decision variables of the candidate solution; N is the total number of provinces; is the hydrogen energy balance residual; is the power balance residual; and are the scale normalization factors for hydrogen energy and power, respectively; and are the weight coefficients for the hydrogen energy and power balance residuals, respectively; and are the exponential parameters not less than 1. S5, according to the Pareto frontier solution set, determine the hydrogen production capacity, hydrogen storage capacity, hydrogen transmission channel configuration and corresponding operation strategy of each province, and output the multi-objective optimization configuration scheme of the whole chain of green hydrogen production-transmission-storage-use in the province.

2. The regional green hydrogen production, transportation, storage and use full-chain multi-objective optimization configuration method according to claim 1, characterized in that, In step S3, the initialization method based on the time sequence matching of renewable energy output and hydrogen demand adopts a hydrogen production time sequence allocation kernel function to allocate the decision variables in time sequence, and the hydrogen production time sequence allocation kernel function is defined as: ； where i is a regional index; t and is a period index for the planning period; is the renewable energy output of region i at period t; is the hydrogen demand; is the renewable energy side weight parameter; is the hydrogen demand side weight parameter; is a set of periods for the planning period.

3. The regional green hydrogen production, transportation, storage and use full-chain multi-objective optimization configuration method according to claim 1, characterized in that, In step S42, the mixed crossover operator of the hydrogen storage state chain performs crossover operation on the time layer decision variables, including: Divide the planning period into several continuous sub-sections in time sequence, and the offspring alternately inherit the time sequence variables of different sub-sections from the two parents according to the parity of the sub-section number; After crossover, the hydrogen production of the offspring is proportionally scaled to make the sum consistent with the new installed capacity, and the inventory is calculated recursively according to the hydrogen storage state equation from the initial period of the planning period, if the inventory is negative, the hydrogen discharge amount is reduced and the inventory is set to zero, if the inventory exceeds the upper limit of the installed capacity, the hydrogen charging amount is reduced and the inventory is set to the upper limit value; The hydrogen transmission flow checks whether each pair of adjacent provinces simultaneously transmits hydrogen in the same period, if there is, the net flow direction is retained and the reverse flow is set to zero.

4. The regional green hydrogen production, transportation, storage and use full-chain multi-objective optimization configuration method according to claim 1, characterized in that, In step S44, the method of self-adaptive adjustment of constraint penalty function strength includes: calculating the average imbalance degree of the gth generation population, introducing the expected imbalance degree level as the convergence target, the penalty coefficient updating rule is the current penalty coefficient multiplied by the exponential function, the exponential term of the exponential function is the product of the adjustment coefficient and the imbalance degree deviation ratio, when the average imbalance degree is higher than the expected value, the penalty coefficient increases, when the average imbalance degree is lower than the expected value, the penalty coefficient decreases, the updated penalty coefficient is subjected to upper and lower limit constraints to ensure stability.

5. The regional green hydrogen production, transportation, storage and use full-chain multi-objective optimization configuration method according to claim 1, characterized in that, In step S45, the reference point migration update step includes: The initial reference point is generated in the normalized target space by the Das-Dennis method, the normalized target vectors of the first layer non-dominated solution set are analyzed every preset number of generations, and the number of solutions associated with each reference point is counted; Calculate the average number of associated solutions and its standard deviation, identify the sparse reference points with less than one standard deviation below the average number of associated solutions and the dense reference points with more than one standard deviation above the average number of associated solutions; For each dense reference point, find the nearest sparse reference point, calculate the direction vector from the dense reference point to the sparse reference point, and set the migration step according to the deviation ratio of the number of associated solutions of the dense reference point to the average number of associated solutions; The position of the updated reference point is the original position plus the product of the step and the direction vector, and the coordinates of the migrated reference point are subjected to boundary constraints.

6. The regional green hydrogen production, transportation, storage and use full-chain multi-objective optimization configuration method according to claim 2, characterized in that, The specific steps of the initialization method include: For each province, use Latin hypercube sampling to generate spatial layer decision variables, including hydrogen production capacity, hydrogen storage capacity upper limit and hydrogen storage technology ratio of each technology; Distribute the hydrogen production of each province in the time dimension by using the hydrogen production time sequence allocation kernel function, set the initial period hydrogen production to the upper limit of the total hydrogen production multiplied by the allocation kernel function and a random disturbance term, and the random disturbance term follows a normal distribution with a mean of zero; The initial hydrogen storage charging and discharging amount is generated heuristically according to the deviation of hydrogen demand in each period from the average value in the planning period, and hydrogen charging is arranged in the period when the hydrogen demand is lower than the average value, and hydrogen discharging is arranged in the period when the hydrogen demand is higher than the average value; The initial hydrogen delivery flow value is allocated according to the hydrogen supply and demand gap of each province in each period, and the excess hydrogen in the province with excess supply is delivered to the adjacent short supply province according to the inverse distance weight; The comprehensive imbalance index of the generated initial individual is calculated, and if the index exceeds the preset threshold, the hydrogen production amount, the hydrogen storage charging and discharging amount, and the hydrogen delivery flow are iteratively adjusted according to the residual sign and size until the threshold requirement is met or the maximum repair iteration number is reached.

7. The regional green hydrogen production, transportation, storage and use full-chain multi-objective optimization configuration method according to claim 1, characterized in that, The termination conditions in step S46 include any of the following conditions: (1) the maximum iteration number is reached; (2) the change of the hypervolume index of the first layer non-dominated solution set in consecutive generations is less than a preset threshold, indicating that the Pareto front distribution has tended to be stable; (3) the average imbalance degree of the population is lower than the expected imbalance degree level, and the number of individuals in the first layer non-dominated solution set changes by no more than a preset percentage in consecutive generations, indicating that both the constraint satisfaction and the front stability criteria have met the standards.

Images