Ocean island low-carbon scheduling method based on improved grey wolf algorithm

Abstract

The invention innovatively provides an ocean island low-carbon scheduling method based on an improved grey wolf algorithm. The method focuses on improving the self-sufficiency capability of the island, integrates and shares hydrogen storage, and plans a hydrogen fuel cell ship transportation network and island new energy system scheme by constructing a bidirectional coupling framework. In the aspect of energy utilization optimization, an electricity-hydrogen-traffic ocean island low-carbon scheduling model and a double-layer optimization scheduling model are established in combination with constraints of island new energy resources, ocean traffic, hydrogen energy storage and short-distance transportation. And solving the model by using an improved grey wolf optimization algorithm to realize an optimal scheduling scheme. The method is characterized in that new energy and hydrogen energy storage systems in the marine transportation network are optimized, use of diesel generators is reduced, carbon emission is reduced, and transformation of low-carbon green energy is assisted. According to the method, the improved grey wolf algorithm is verified by adopting an improved IEEE 6-node grid structure, and the method is proved to be effective in the aspects of improving island energy potential and transportation network efficiency, reducing fossil energy consumption and realizing green and low-carbon targets.

Description

Technical field:

[0001] The present invention belongs to the field of comprehensive energy optimization dispatching of offshore islands Background technology:

[0002] The islands are located in areas with abundant renewable energy resources. Existing research mainly focuses on the inter-island discrete energy transmission mode based on battery swapping ships, that is, by transporting the clean electricity produced by the enriched islands in the form of mobile battery packs to solve the problem of energy complementarity between islands. The advantage of this mode is that it can flexibly dispatch electricity and ensure the optimal economic operation of electricity when the load fluctuates. It is especially suitable for islands with strong renewable energy resources (such as wind energy and solar energy). However, this model still faces certain bottlenecks, including battery energy density, battery swapping efficiency, and energy loss during transportation. Especially in ocean transportation, due to the limitations of battery technology, its endurance, charging time and energy density are all affected to a certain extent.

[0003] Compared with traditional battery technology, the application of hydrogen fuel cell ships in the energy supply system of offshore islands provides a new solution. Compared with traditional battery packs, hydrogen fuel cells can effectively reduce energy loss during transportation, and because hydrogen can be stored through liquefaction or high pressure, it has higher flexibility in energy storage and transportation. Although hydrogen fuel cell technology has been applied to other fields, there are few studies on the use of hydrogen energy as a power source for new energy storage and transportation equipment, especially in the context of offshore island energy supply. Therefore, exploring the application of hydrogen fuel cell ships as energy transportation carriers can further improve the efficiency of energy storage and transportation, and provide a greener and more sustainable solution for the energy supply of offshore islands. Summary of the invention:

[0004] In order to solve the problem of natural energy utilization on offshore islands, the present invention proposes a low-carbon scheduling method for offshore islands based on an improved grey wolf optimization algorithm.

[0005] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions:

[0006] Step 1: Construct a low-carbon dispatch model for electricity, hydrogen and transportation in remote ocean islands

[0007] The active-grid-load integrated power supply system structure is configured by the load center island, and the constraints under various actual conditions are considered to construct the load center island model and the marine transportation network model respectively.

[0008] The constraints for constructing the load center island model are as follows:

[0009] 1) Power flow constraints:

[0010]

[0011] Where: j∈i means that the source, storage, load, etc. are directly connected to node i; U i is the voltage amplitude of node i; θ ij is the power angle difference between nodes ij; G ij +B ij is the corresponding admittance parameter; P DG , Q DG , They are the active power and reactive power of distributed renewable energy, diesel generators, energy storage and loads respectively.

[0012] 2) Line power constraints:

[0013] -P l,max ≤U i U j (G ij cosθ ij +B ij sinθ ij )≤P l,max (2)

[0014] Where: -P l,max is the maximum transmission power of line l (nodes i and j at both ends).

[0015] 3) Voltage constraint:

[0016] U min ≤U i ≤U max (3)

[0017] Here: U min , U max They are the upper and lower limits of the node voltage amplitude, respectively, and are taken as 1.1 and 0.9 here.

[0018] 4) Diesel generator output constraints:

[0019]

[0020] Where: The maximum output of the diesel generator, generally the rated power is the minimum output of the diesel generator.

[0021] 5) Wind and solar distributed energy output constraints

[0022]

[0023] Where: are the minimum output of wind and solar distributed energy, They are the maximum outputs of wind and solar distributed energy respectively.

[0024] 6) Hydrogen production model using surplus electricity from wind and solar distributed energy:

[0025]

[0026] Where: P ex (t) is the power of wind and solar distributed energy surplus power in period t, P eP2G (t) is the power consumption of the hydrogen electrolysis module during period t, P pv (t), P wind (t) is the output of wind and solar distributed energy in period t, P e-load (t) is the electric load in period t, η p2G The energy conversion efficiency of hydrogen production from electricity is is the hydrogen storage capacity during period t, It is the upper limit of the capacity of hydrogen energy storage equipment.

[0027] 7) Construction of hydrogen energy storage model:

[0028] Hydrogen energy storage consists of hydrogen storage tanks, electrolytic hydrogen production devices, and hydrogen fuel cells. Hydrogen storage tanks store hydrogen, electrolytic hydrogen production devices consume electricity to produce hydrogen, and fuel cells provide electricity by burning hydrogen. The calculation process of the hydrogen energy stored in the hydrogen storage tank is shown below:

[0029]

[0030] In the formula, t represents the sequence number of the scheduling period; represents the amount of hydrogen energy stored in the scheduling period t, η p2G , η FC They represent the energy conversion efficiency of the hydrogen electrolysis device and the hydrogen fuel cell respectively; Δt represents the time span of the scheduling period.

[0031] Lifetime attenuation cost corresponding to hydrogen energy storage power fluctuation:

[0032]

[0033] In the formula, κ λ_FL P represents the life attenuation cost corresponding to the unit fluctuating power of the electrolytic hydrogen production device and the hydrogen fuel cell; λ_t Represents the electrical power of the hydrogen electrolysis device and hydrogen fuel cell; r λ_t Indicates the start and stop status of the electrolytic hydrogen production device and the hydrogen fuel cell. The start-up and shutdown costs of the electrolytic hydrogen production device and hydrogen fuel cell are shown in the formula:

[0034]

[0035]

[0036] In the formula, κ λ_SU and κ λ_SD They are respectively represented by the unit start-up and shutdown costs of the electrolytic hydrogen production device and the hydrogen fuel cell; r λ_SU_t and r λ_SD_t They respectively indicate the start and stop status of the hydrogen electrolysis device and the hydrogen fuel cell.

[0037] Maintenance costs of hydrogen electrolysis equipment and hydrogen fuel cells

[0038]

[0039] In the formula, κ λ_OM Represents the unit time maintenance cost of the hydrogen electrolysis device and hydrogen fuel cell.

[0040] The life attenuation cost corresponding to the historical operating time of the electrolytic hydrogen production device and the hydrogen fuel cell is shown in the formula:

[0041]

[0042] In the formula, κ λ_AGE T represents the life attenuation cost of the hydrogen electrolysis device / hydrogen fuel cell per unit operating time; λ_LT Indicates the expected life of the hydrogen electrolysis device / hydrogen fuel cell.

[0043] The total operating cost of hydrogen energy storage is:

[0044]

[0045] The marine transportation network model is constructed as follows:

[0046] The size of HFCS hydrogen load is mainly affected by factors such as driving mileage and return time as well as the supply and demand of hydrogen refueling stations. The transportation network model is established as follows.

[0047] The probability density function of the last return time of a user every day is:

[0048]

[0049] Where: x t is the user's last return time, ψ t is the expected value, γ t is the standard deviation.

[0050] The probability density function of HFCS user mileage is defined as:

[0051]

[0052] Where: xs is the user's last return time, ψ s is the expected value, γ t is the standard deviation.

[0053] HFCS hydrogenation time:

[0054]

[0055] Where: T s is the hydrogen refueling time, S is the daily mileage, P is s is the amount of hydrogenation, σ s is the hydrogen refueling efficiency, and E is the hydrogen consumption per unit kilometer.

[0056] Single station hydrogen refueling demand

[0057] N m,HFCS ≤N m,H (17)

[0058] Where: N m,HFCS is the number of HFCS currently available at the mth hydrogen refueling station, N m,H is the amount of HFCS that the mth hydrogen refueling station can accept.

[0059] Total hydrogen refueling demand in the transportation network

[0060]

[0061] Where: M is the total number of hydrogen refueling stations in the transportation network, M m,cqz is the capacity of the hydrogen storage station, M e,HFCS is the e-th HFCS capacity, N HFCS is the amount of HFCS.

[0062] Short-term hydrogen storage model

[0063]

[0064] Where: and They are the hydrogen storage per unit time, hydrogen release per unit time and the total amount of remaining hydrogen in short-term hydrogen energy storage. and are the hydrogen storage efficiency, hydrogen release efficiency and self-release rate of STHS, C STHS is the configuration capacity of STHS, and Δt is the unit time step.

[0065] Shared Hydrogen Storage Model

[0066]

[0067] Where: and They are the amount of hydrogen stored per unit time, the amount of hydrogen released per unit time and the total amount of hydrogen remaining in hydrogen energy storage. and are the hydrogen storage efficiency, hydrogen release efficiency and self-release rate of SHS, respectively, C SHS The configuration capacity of SHS. SHS is used when STHS cannot meet the hydrogen load, and the order of use is: real-time hydrogen replenishment > STHS > SHS. HFCS hydrogenation must meet certain conditions. The actual amount of hydrogen cannot exceed the total capacity of the battery, but at the same time it cannot be less than the amount of hydrogen expected by the user. Assuming that when the ship owner connects the HFCS to the hydrogenation pile, the user will provide his or her expected information, such as the desired hydrogen state when leaving the hydrogenation pile is S. h , which means that when the user decides to leave, the actual amount of hydrogen in HFCS should be greater than or equal to the desired value, which can be expressed as

[0068]

[0069] Where: S st,i is the hydrogen charge state of the i-th HFCS vehicle when it starts to be refueled, S en,i It is the hydrogen charge state when the i-th amount of HFCS leaves.

[0070] Step 2: Construct a low-carbon dual-layer optimization dispatch model for electricity, hydrogen and transportation in ocean islands

[0071] A) Upper-level transportation network - offshore island operation model

[0072] Objective Function

[0073]

[0074] Where: f1 represents the operating cost of the offshore island, C HS , C WT , C PV They represent the operating cost of hydrogen energy storage, wind power generation device, and photovoltaic power generation device, respectively. f2 represents the hydrogenation cost; it is the hydrogenation status of the i-th ship in the j-th period, and has only two values: 0 or 1, 0 means no hydrogenation, 1 means hydrogenation; p i is the power of the i-th ship; p j represents the hydrogen price in period j; N represents the total number of HFCS ships.

[0075] Constraints

[0076] Hydrogen energy storage constraints, power balance constraints, HFCS hydrogenation constraints.

[0077] B) Lower-level transportation network-distant ocean island self-consistent model

[0078] Based on the bidirectional coupling architecture of ocean island new energy and transportation network, a low-carbon economic optimization scheduling objective function is established, which includes two parts: one is the highest self-consistency rate of system electricity and hydrogen energy; the other is the lowest carbon emission cost of the system.

[0079]

[0080]

[0081]

[0082]

[0083] Where: η is the daily self-consistency rate of electricity-hydrogen-transportation coupling on the ocean island, F E is the electrical load on the island, F DG is the output power of distributed energy, F H is the hydrogen load of the transportation network, F q is the energy output of the hydrogen refueling station, F is the system carbon emission cost, α is the carbon emission tax amount, is the carbon emission cost. c is the carbon trading price, c is the basic unit carbon price, β is the step price increase, v is the length of the carbon emission interval. γ is the incentive coefficient in the carbon trading price; E g is the actual carbon emission generated by starting the diesel generator, and a1, b1, and c1 are the carbon emission calculation coefficients of the diesel generator. c is the carbon emission quota for electricity generated by diesel generators, χ is the carbon emission quota per unit of electricity, P t E It is the amount of electricity generated by the diesel generator per unit time.

[0084] Constraints

[0085] Distribution network operation constraints, electrolyzer equipment operation constraints, hydrogen fuel cell equipment operation constraints, HFCS and hydrogen refueling station constraints, and hydrogen energy storage equipment operation constraints.

[0086] Step 3: Solve the model based on the improved gray wolf optimization algorithm

[0087] The Gray Wolf Algorithm is a heuristic algorithm based on the analysis of the hierarchy and hunting behavior of wolves. It can be used to solve the two-layer model constructed in this paper. In the algorithm, the α wolf is the leader, and the β, γ, and ω wolves are assistants. The entire group gradually finds the optimal solution under the leadership of the α wolf. The traditional Gray Wolf Algorithm has the disadvantage of slow convergence speed. This paper introduces chaotic adversarial learning, differential evolution, and individual perturbation mechanism to improve the algorithm's optimization ability and convergence speed.

[0088] 1) Tent adversarial learning optimizes population initialization

[0089] In the search space, the position of the gray wolf is initialized using chaotic tent mapping to form the initial population X. Then, the opposing population X′ is generated through adversarial learning. The fitness of these two populations is sorted, and the individuals with the highest ranking will be determined as the final initial population.

[0090] The initial population created using the Tent map is:

[0091]

[0092] Where: i represents the population size; i = 1, 2, 3, ..., N; j represents the index of the chaotic sequence, j = 1, 2, 3, ..., d; μ represents the chaotic parameter, and its value is in the range of [0, 2].

[0093] The adversarial solution generated by the adversarial learning mechanism is:

[0094] x i,j = l i,j +y i,j ×(u i,j -l i,j ),x i ∈X (28)

[0095] Where: l ij 、u ij Represents element x i ∈X. And compare the fitness after it is generated, and retain the optimal solution.

[0096] 2) Position update based on differential evolution

[0097] The traditional gray wolf algorithm first generates the initial gray wolf population position with a random position, and then the α, β, and γ wolves hunt for prey and update their own positions through equations (29) to (31).

[0098]

[0099]

[0100]

[0101] Where: D is the distance between the wolf and its prey; C and A are coefficient vectors. The wolf gradually approaches its prey by adjusting A and D. When the number of iterations reaches the maximum value T max , the algorithm ends running; t-current number of iterations; represents the location of the prey that α, β, and γ wolves want to catch; X t Represents the position of the gray wolf itself; X1, X2, and X3 represent the updated positions of α, β, and γ wolves.

[0102] Based on the differential mutation method, this paper introduces mutation operators, crossover operators and selection operators to increase the population size and category richness and optimize the position update process.

[0103] ① Mutation operator

[0104] The mutation solution is expressed by the following formula, where the difference changes between different individual solutions are determined by the mutation operator.

[0105]

[0106] Where: v represents the variant solution; t represents the number of iterations; x represents the original population solution; N represents the population size; r1, r2, r3 represent random integers in [1, N]; ω represents the mutation operator that affects the differential changes between individual solutions.

[0107] ② Crossover operator

[0108] The crossover operator performs crossover processing on the individual solutions that have completed the mutation to increase the richness of the population, as shown in the following formula.

[0109]

[0110] Where: rand represents a random value in [0,1]; C R represents the individual cross-coefficient.

[0111] ③Select operator

[0112] A greedy strategy is used to compare parent and offspring individuals, and new individuals are selected by the following formula:

[0113]

[0114] Where: f(z) represents the fitness function.

[0115]

[0116] Where: P i Represents the probability that the gray wolf chooses the differential evolution strategy to update the position. If P≥rand, the differential evolution method is not used to update the position, but the traditional gray wolf algorithm is used to update the position; if P<rand, the differential evolution method is used to update the position.

[0117] 3) Individual disturbance

[0118] Individual perturbations enhance the global search capability. When two gray wolves gradually approach each other, individual perturbations will make the gray wolves converge to the initial point, thereby maintaining the diversity of the population, as shown in the following formula.

[0119]

[0120] Where: D op represents the perturbation operator; represents the Euclidean distance between the locations of gray wolf i and gray wolf j; Represents a random quantity uniformly distributed in [a,b]. Description of the drawings:

[0121] Figure 1 Improved Grey Wolf Algorithm Algorithm Flowchart

[0122] Figure 2 Improved IEEE 6-bus system diagram

[0123] Figure 3 Basic data algorithm iteration

[0124] Figure 4 Operation of the terminal in Scheme 4

[0125] Figure 5 Operation status of substation 1 in scheme 4

[0126] Figure 6 Scheme 4 Substation 2 operation Specific implementation method:

[0127] Step 1: Construct a low-carbon dispatch model for electricity, hydrogen and transportation in remote ocean islands

[0128] The active-grid-load integrated power supply system structure is configured by the load center island, and the constraints under various actual conditions are considered to construct the load center island model and the marine transportation network model respectively.

[0129] The constraints for constructing the load center island model are as follows:

[0130] 1) Power flow constraints:

[0131]

[0132] Where: j∈i means that the source, storage, load, etc. are directly connected to node i; U i is the voltage amplitude of node i; θ ij is the power angle difference between nodes ij; G ij +B ij is the corresponding admittance parameter; P DG , Q DG , They are the active power and reactive power of distributed renewable energy, diesel generators, energy storage and loads respectively.

[0133] 2) Line power constraints:

[0134] -P l,max ≤U i Uj (G ij cosθ ij +B ij sinθ ij )≤P l,max (2)

[0135] Where: -P l,max is the maximum transmission power of line l (nodes i and j at both ends).

[0136] 3) Voltage constraint:

[0137] U min ≤U i ≤U max (3)

[0138] Here: U min , U max They are the upper and lower limits of the node voltage amplitude, respectively, and are taken as 1.1 and 0.9 here.

[0139] 4) Diesel generator output constraints:

[0140]

[0141] Where: The maximum output of the diesel generator, generally the rated power is the minimum output of the diesel generator.

[0142] 5) Wind and solar distributed energy output constraints

[0143]

[0144] Where: are the minimum output of wind and solar distributed energy, They are the maximum outputs of wind and solar distributed energy respectively.

[0145] 6) Hydrogen production model using surplus electricity from wind and solar distributed energy:

[0146]

[0147] Where: P ex (t) is the power of wind and solar distributed energy surplus power in period t, P eP2G (t) is the power consumption of the hydrogen electrolysis module during period t, P pv (t), P wind (t) is the output of wind and solar distributed energy in period t, P e-load (t) is the electric load in period t, η p2G The energy conversion efficiency of hydrogen production from electricity is is the hydrogen storage capacity during period t, It is the upper limit of the capacity of hydrogen energy storage equipment.

[0148] 7) Construction of hydrogen energy storage model:

[0149] Hydrogen energy storage consists of hydrogen storage tanks, electrolytic hydrogen production devices, and hydrogen fuel cells. Hydrogen storage tanks store hydrogen, electrolytic hydrogen production devices consume electricity to produce hydrogen, and fuel cells provide electricity by burning hydrogen. The calculation process of the hydrogen energy stored in the hydrogen storage tank is shown below:

[0150]

[0151] In the formula, t represents the sequence number of the scheduling period; represents the amount of hydrogen energy stored in the scheduling period t, η p2G , η FC They represent the energy conversion efficiency of the hydrogen electrolysis device and the hydrogen fuel cell respectively; Δt represents the time span of the scheduling period.

[0152] Lifetime attenuation cost corresponding to hydrogen energy storage power fluctuation:

[0153]

[0154] In the formula, κ λ_FL P represents the life attenuation cost corresponding to the unit fluctuating power of the electrolytic hydrogen production device and the hydrogen fuel cell; λ_t Represents the electrical power of the hydrogen electrolysis device and hydrogen fuel cell; r λ_t Indicates the start and stop status of the electrolytic hydrogen production device and the hydrogen fuel cell. The start-up and shutdown costs of the electrolytic hydrogen production device and hydrogen fuel cell are shown in the formula:

[0155]

[0156]

[0157] In the formula, κ λ_SU and κ λ_SD They are respectively represented by the unit start-up and shutdown costs of the electrolytic hydrogen production device and the hydrogen fuel cell; r λ_SU_t and r λ_SD_t They respectively indicate the start and stop status of the hydrogen electrolysis device and the hydrogen fuel cell.

[0158] Maintenance costs of hydrogen electrolysis equipment and hydrogen fuel cells

[0159]

[0160] In the formula, κ λ_OM Represents the unit time maintenance cost of the hydrogen electrolysis device and hydrogen fuel cell.

[0161] The life attenuation cost corresponding to the historical operating time of the electrolytic hydrogen production device and the hydrogen fuel cell is shown in the formula:

[0162]

[0163] In the formula, κ λ_AGE T represents the life attenuation cost of the hydrogen electrolysis device / hydrogen fuel cell per unit operating time; λ_LT Indicates the expected life of the hydrogen electrolysis device / hydrogen fuel cell.

[0164] The total operating cost of hydrogen energy storage is:

[0165]

[0166] The marine transportation network model is constructed as follows:

[0167] The size of HFCS hydrogen load is mainly affected by factors such as driving mileage and return time as well as the supply and demand of hydrogen refueling stations. The transportation network model is established as follows.

[0168] The probability density function of the last return time of a user every day is:

[0169]

[0170] Where: x t is the user's last return time, ψ t is the expected value, γ t is the standard deviation.

[0171] The probability density function of HFCS user mileage is defined as:

[0172]

[0173] Where: x s is the user's last return time, ψ s is the expected value, γ t is the standard deviation.

[0174] HFCS hydrogenation time:

[0175]

[0176] Where: T s is the hydrogen refueling time, S is the daily mileage, P is s is the amount of hydrogenation, σ s is the hydrogen refueling efficiency, and E is the hydrogen consumption per unit kilometer.

[0177] Single station hydrogen refueling demand

[0178] N m,HFCS ≤N m,H(17)

[0179] Where: N m,HFCS is the number of HFCS currently available at the mth hydrogen refueling station, N m,H is the amount of HFCS that the mth hydrogen refueling station can accept.

[0180] Total hydrogen refueling demand in the transportation network

[0181]

[0182] Where: M is the total number of hydrogen refueling stations in the transportation network, M m,cqz is the capacity of the hydrogen storage station, M e,HFCS is the e-th HFCS capacity, N HFCS is the amount of HFCS.

[0183] Short-term hydrogen storage model

[0184]

[0185] Where: and They are the hydrogen storage per unit time, hydrogen release per unit time and the total amount of remaining hydrogen in short-term hydrogen energy storage. and are the hydrogen storage efficiency, hydrogen release efficiency and self-release rate of STHS, C STHS is the configuration capacity of STHS, and Δt is the unit time step.

[0186] Shared Hydrogen Storage Model

[0187]

[0188] Where: and They are the amount of hydrogen stored per unit time, the amount of hydrogen released per unit time and the total amount of hydrogen remaining in hydrogen energy storage. and are the hydrogen storage efficiency, hydrogen release efficiency and self-release rate of SHS, respectively, C SHS The configuration capacity of SHS. SHS is used when STHS cannot meet the hydrogen load, and the order of use is: real-time hydrogen replenishment > STHS > SHS. HFCS hydrogenation must meet certain conditions. The actual amount of hydrogen cannot exceed the total capacity of the battery, but at the same time it cannot be less than the amount of hydrogen expected by the user. Assuming that when the ship owner connects the HFCS to the hydrogenation pile, the user will provide his or her expected information, such as the desired hydrogen state when leaving the hydrogenation pile is S. h , which means that when the user decides to leave, the actual amount of hydrogen in HFCS should be greater than or equal to the desired value, which can be expressed as

[0189]

[0190] Where: S st,i is the hydrogen charge state of the i-th HFCS vehicle when it starts to be refueled, S en,i It is the hydrogen charge state when the i-th amount of HFCS leaves.

[0191] Step 2: Construct a low-carbon dual-layer optimization dispatch model for electricity, hydrogen and transportation in ocean islands

[0192] A) Upper-level transportation network - offshore island operation model

[0193] Objective Function

[0194]

[0195] Where: f1 represents the operating cost of the offshore island, C HS , C WT , C PV They represent the operating cost of hydrogen energy storage, wind power generation device, and photovoltaic power generation device, respectively. f2 represents the hydrogenation cost; it is the hydrogenation status of the i-th ship in the j-th period, and has only two values: 0 or 1, 0 means no hydrogenation, 1 means hydrogenation; p i is the power of the i-th ship; p j represents the hydrogen price in period j; N represents the total number of HFCS ships.

[0196] Constraints

[0197] Hydrogen energy storage constraints, power balance constraints, HFCS hydrogenation constraints.

[0198] B) Lower-level transportation network-distant ocean island self-consistent model

[0199] Based on the bidirectional coupling architecture of ocean island new energy and transportation network, a low-carbon economic optimization scheduling objective function is established, which includes two parts: one is the highest self-consistency rate of system electricity and hydrogen energy; the other is the lowest carbon emission cost of the system.

[0200]

[0201]

[0202]

[0203]

[0204] Where: η is the daily self-consistency rate of electricity-hydrogen-transportation coupling on the ocean island, F E is the electrical load on the island, F DG is the output power of distributed energy, F H is the hydrogen load of the transportation network, Fq is the energy output of the hydrogen refueling station, F is the system carbon emission cost, α is the carbon emission tax amount, is the carbon emission cost. c is the carbon trading price, c is the basic unit carbon price, β is the step price increase, v is the length of the carbon emission interval. γ is the incentive coefficient in the carbon trading price; E g is the actual carbon emission generated by starting the diesel generator, and a1, b1, and c1 are the carbon emission calculation coefficients of the diesel generator. c is the carbon emission quota for electricity generated by diesel generators, χ is the carbon emission quota per unit of electricity, P t E It is the amount of electricity generated by the diesel generator per unit time.

[0205] Constraints

[0206] Distribution network operation constraints, electrolyzer equipment operation constraints, hydrogen fuel cell equipment operation constraints, HFCS and hydrogen refueling station constraints, and hydrogen energy storage equipment operation constraints.

[0207] Step 3: Solve the model based on the improved gray wolf optimization algorithm

[0208] The Gray Wolf Algorithm is a heuristic algorithm based on the analysis of the hierarchy and hunting behavior of wolves. It can be used to solve the two-layer model constructed in this paper. In the algorithm, the α wolf is the leader, and the β, γ, and ω wolves are assistants. The entire group gradually finds the optimal solution under the leadership of the α wolf. The traditional Gray Wolf Algorithm has the disadvantage of slow convergence speed. This paper introduces chaotic adversarial learning, differential evolution, and individual perturbation mechanism to improve the algorithm's optimization ability and convergence speed.

[0209] 1) Tent adversarial learning optimizes population initialization

[0210] In the search space, the position of the gray wolf is initialized using chaotic tent mapping to form the initial population X. Then, the opposing population X′ is generated through adversarial learning. The fitness of these two populations is sorted, and the individuals with the highest ranking will be determined as the final initial population.

[0211] The initial population created using the Tent map is:

[0212]

[0213] Where: i represents the population size; i = 1, 2, 3, ..., N; j represents the index of the chaotic sequence, j = 1, 2, 3, ..., d; μ represents the chaotic parameter, and its value is in the range of [0, 2].

[0214] The adversarial solution generated by the adversarial learning mechanism is:

[0215] x i,j = l i,j +y i,j ×(u i,j -l i,j ),x i ∈X (28)

[0216] Where: l ij 、u ij Represents element x i ∈X. And compare the fitness after it is generated, and retain the optimal solution.

[0217] 2) Position update based on differential evolution

[0218] The traditional gray wolf algorithm first generates the initial gray wolf population position with a random position, and then the α, β, and γ wolves hunt for prey and update their own positions through equations (29) to (31).

[0219]

[0220]

[0221]

[0222] Where: D is the distance between the wolf and its prey; C and A are coefficient vectors. The wolf gradually approaches its prey by adjusting A and D. When the number of iterations reaches the maximum value T max , the algorithm ends running; t-current number of iterations; represents the location of the prey that α, β, and γ wolves want to catch; X t Represents the position of the gray wolf itself; X1, X2, and X3 represent the updated positions of α, β, and γ wolves.

[0223] Based on the differential mutation method, this paper introduces mutation operators, crossover operators and selection operators to increase the population size and category richness and optimize the position update process.

[0224] ① Mutation operator

[0225] The mutation solution is expressed by the following formula, where the difference changes between different individual solutions are determined by the mutation operator.

[0226]

[0227] Where: v represents the variant solution; t represents the number of iterations; x represents the original population solution; N represents the population size; r1, r2, r3 represent random integers in [1, N]; ω represents the mutation operator that affects the differential changes between individual solutions.

[0228] ② Crossover operator

[0229] The crossover operator performs crossover processing on the individual solutions that have completed the mutation to increase the richness of the population, as shown in the following formula.

[0230]

[0231] Where: rand represents a random value in [0,1]; C R represents the individual cross-coefficient.

[0232] ③Select operator

[0233] A greedy strategy is used to compare parent and offspring individuals, and new individuals are selected by the following formula:

[0234]

[0235] Where: f(z) represents the fitness function.

[0236]

[0237] Where: P i Represents the probability that the gray wolf chooses the differential evolution strategy to update the position. If P≥rand, the differential evolution method is not used to update the position, but the traditional gray wolf algorithm is used to update the position; if P<rand, the differential evolution method is used to update the position.

[0238] 3) Individual disturbance

[0239] Individual perturbations enhance the global search capability. When two gray wolves gradually approach each other, individual perturbations will make the gray wolves converge to the initial point, thereby maintaining the diversity of the population, as shown in the following formula.

[0240]

[0241] Where: D op represents the perturbation operator; represents the Euclidean distance between the locations of gray wolf i and gray wolf j; Represents a random quantity uniformly distributed in [a,b].

[0242] The algorithm flow is as follows Figure 1 Shown

[0243] Example:

[0244] This paper takes an island as the research object and analyzes its power grid and transportation network. The power grid adopts an improved IEEE 6-node grid structure (such as Figure 2), rated voltage 10kV, load based on typical island conventional load and proportional scaling. The island is equipped with a 400kW diesel generator, wind power and photovoltaic power are 180kW and 200kW respectively, representing the multi-form renewable energy of offshore islands. The transportation network is designed with 1 main station and 2 substations, and energy coupling is carried out through hydrogen refueling stations. The system aims to study the coordinated application of renewable energy and traditional energy, and improve the reliability and sustainability of island energy supply.

[0245] In the electricity-hydrogen-transportation coupling calculation example, the hydrogenation station is equipped with STHS and SHS. The hydrogenation substation only has STHS, that is, the hydrogenation substation cannot transport hydrogen energy to the hydrogenation station, and the hydrogenation substations cannot provide hydrogen energy to each other. If the hydrogen production from the surplus electricity of wind and solar distributed energy cannot meet the hydrogen load demand, the STHS of the hydrogenation station will first make up for it. If there is still a shortage of hydrogen energy, the SHS of the hydrogenation station will be called by the transport ship.

[0246] First, the improved grey wolf algorithm is compared with the traditional grey wolf algorithm. Figure 3 As shown. Figure 3 It can be seen that the improved Grey Wolf Algorithm converges faster and has higher accuracy than the traditional Grey Wolf Algorithm. Therefore, the improved Grey Wolf Algorithm is used to optimize the original hydrogenation data, and the comparison before and after optimization is shown in Table 1.

[0247] Table 1 Optimization comparison results

[0248]

[0249] By comparison, the total station hydrogenation cost was reduced by 21.5%, and the peak-to-valley difference was reduced by 15.9%; the substation 1 hydrogenation cost was reduced by 38.1%, and the peak-to-valley difference was reduced by 38.5%; the substation 2 hydrogenation cost was reduced by 37.5%, and the peak-to-valley difference was reduced by 28.9%. It can be seen that the transportation network hydrogenation model can effectively reduce users' hydrogenation costs and peak-to-valley differences.

[0250] Comparative analysis of solutions

[0251] Four schemes are set up in the example, and the hydrogen energy storage operation modes of each scheme are shown in Table 2.

[0252] Table 2 Hydrogen energy storage solutions

[0253]

[0254] Note: Indicates that this situation is not taken into consideration. Indicates that this situation is taken into consideration.

[0255] When the distributed energy output of the energy self-consistent island meets the electricity load demand, the electricity self-consistency rate is 100%. The self-consistency rate of hydrogen energy under the hydrogen load energy demand is analyzed, the optimized hydrogenation data and the hydrogen production curve of the hydrogenation station are selected, and the energy self-consistency rate and carbon emission cost calculation results of each scheme are shown in Table 3.

[0256] Table 3 Calculation results of hydrogen energy storage scheme

[0257]

[0258] Note: Indicates that this situation is not considered.

[0259] The example analysis results in Table 3 show that:

[0260] 1) The energy self-consistency rate of the hydrogen refueling station in Scheme 2 is increased by 20.56% compared with Scheme 1, and the carbon emission cost is reduced by 91.55%. The calculation results show that STHS can effectively improve the energy self-consistency rate and reduce carbon emissions.

[0261] 2) In Scheme 3, the energy self-consistency rate of hydrogenation substation 1 is increased by 19.51% compared with Scheme 1, and the carbon emission cost is reduced by 42.33%. The energy self-consistency rate of substation 2 is increased by 15.74% compared with Scheme 1, and the carbon emission cost is reduced by 37.56%. The calculation results show that SHS can improve the energy self-consistency rate and reduce carbon emissions.

[0262] 3) Comparative analysis of Scheme 2 and Scheme 3 shows that, in the case of a single hydrogen energy storage configuration, STHS is better than SHS because STHS has the characteristics of local energy utilization and cross-time balance, which meets the needs of local energy consumption and high system self-consistency.

[0263] 4) Comparative analysis of the four schemes shows that Scheme 4 is the best calculation result. It can be seen that the combination of STHS and SHS is the best choice to improve energy self-consistency and reduce carbon emissions. The order of hydrogen energy supply is: hydrogen production from wind and solar distributed energy surplus power, STHS, SHS, and hydrogen production through diesel generators.

[0264] In Scheme 4, the hydrogen refueling station has STHS and SHS. If the remaining hydrogen energy is less than 0.8kg, all of it is stored in STHS; if the remaining hydrogen energy is greater than 0.8kg, 0.8kg is STHS, and the rest of the hydrogen energy is shared energy storage, such as Figure 4 As shown, the operation of the main station and substations 1 and 2 is as follows Figure 5 , Figure 6 shown.

[0265] The above optimization results show that:

[0266] 1) Figure 4The energy self-consistency rate of Scheme 4 and Scheme 2 is 95.15% both because in both schemes, the hydrogen refueling station only has STHS and needs to rely on diesel generators to generate electricity and produce hydrogen.

[0267] 2) Figure 5 The STHS and SHS jointly support the operation of hydrogen refueling substation 1, with an energy self-consistency rate of 97.07%. 14.28 kg of hydrogen is produced by purchasing electricity from the upper-level power grid, resulting in a carbon emission cost of 3.13 yuan.

[0268] 3) Figure 6 The STHS and SHS jointly support the operation of the hydrogen refueling substation 2, and the energy self-consistency rate reaches 100%. No diesel generator is needed for power generation, and no carbon emission costs will be generated. It can be seen that improving the energy self-consistency rate can effectively reduce carbon emission costs.

[0269] 4) During the early morning hours, the output of wind and solar distributed energy is small and hydrogen reserves are insufficient, so the island needs to rely on diesel generators to generate electricity.

Claims

1. Research on the optimization scheduling of electric-hydrogen-transportation coupled ocean islands based on the improved Gray Wolf optimization algorithm, which features the following steps: Step 1: Construct a low-carbon dispatch model for electricity, hydrogen and transportation in oceanic islands Step 2: Construct a low-carbon dual-layer optimization dispatch model for electricity, hydrogen and transportation in ocean islands Step 3: Use the improved Grey Wolf Algorithm to solve the electricity-hydrogen-transportation ocean island low-carbon scheduling model and the island two-layer optimization scheduling model constructed in Step 2, and output the optimal solution.

2. The island electricity-hydrogen transmission coupling optimization scheduling method based on the improved grey wolf algorithm according to claim 1 is characterized by: It also includes step 3: This step proposes an improved IEEE standard 6-node grid structure to verify the feasibility of the method of improving energy self-consistency and reducing the carbon emission cost of the island, and evaluates the effectiveness of the improved grey wolf algorithm.

3. The island electricity-hydrogen transmission coupling optimization scheduling method based on the improved Grey Wolf optimization algorithm according to claim 1 is characterized by: The electric-hydrogen-transportation coupling model in step 1 includes a load center island model, as follows: The constraints for constructing the load center island model are as follows: 1) Power flow constraints: Where: j∈i means that the source, storage, load, etc. are directly connected to node i; U i is the voltage amplitude of node i; θ ij is the power angle difference between nodes ij; G ij +B ij is the corresponding admittance parameter; P DG , Q DG , They are the active power and reactive power of distributed renewable energy, diesel generators, energy storage and loads respectively. 2) Line power constraints: -P l,max ≤U i U j (G ij cosθ ij +B ij sinθ ij )≤P l,max (2) Where: -P l,max is the maximum transmission power of line l (nodes i and j at both ends). 3) Voltage constraint: IN min ≤U i ≤U max (3) Here: U min , U max They are the upper and lower limits of the node voltage amplitude, respectively, and are taken as 1.1 and 0.9 here. 4) Diesel generator output constraints: Where: The maximum output of the diesel generator, generally the rated power is the minimum output of the diesel generator, take 0.3 5) Output constraints of wind and solar distributed energy Where: are the minimum output of wind and solar distributed energy, They are the maximum outputs of wind and solar distributed energy respectively.

4. The island electricity-hydrogen transmission coupling optimization scheduling method based on the improved grey wolf algorithm according to claim 5 is characterized by: The electric-hydrogen transportation coupling model in step 1 includes a hydrogen energy storage model and an operation and maintenance cost model, as follows: 1) Hydrogen production model using surplus electricity from wind and solar distributed energy: Where: P ex (t) is the power of the wind and solar energy surplus power during period t, P eP2G (t) is the power consumption of the hydrogen electrolysis module during period t, P pv (t), P wind (t) is the output of wind and solar distributed energy in period t, P e-load (t) is the electric load in period t, η p2G The energy conversion efficiency of hydrogen production from electricity is is the hydrogen storage capacity during period t, It is the upper limit of the capacity of hydrogen energy storage equipment. 2) Construction of hydrogen energy storage model: Hydrogen energy storage consists of hydrogen storage tanks, electrolytic hydrogen production devices, and hydrogen fuel cells. Hydrogen storage tanks store hydrogen, electrolytic hydrogen production devices consume electricity to produce hydrogen, and fuel cells provide electricity by burning hydrogen. The calculation process of the hydrogen energy stored in the hydrogen storage tank is shown below: In the formula, t represents the sequence number of the scheduling period; represents the amount of hydrogen energy stored in the scheduling period t, η p2G , η FC They represent the energy conversion efficiency of the hydrogen electrolysis device and the hydrogen fuel cell respectively; Δt represents the time span of the scheduling period. 3) Hydrogen energy storage cost model In the formula, κ λ_FL P represents the life attenuation cost corresponding to the unit fluctuating power of the electrolytic hydrogen production device and the hydrogen fuel cell; λ_t Represents the electrical power of the hydrogen electrolysis device and hydrogen fuel cell; r λ_t Indicates the start and stop status of the electrolytic hydrogen production device and the hydrogen fuel cell. The start-up and shutdown costs of the hydrogen electrolysis device and hydrogen fuel cell are as follows: In the formula, κ λ_SU and κ λ_SD They are respectively represented by the unit start-up and shutdown costs of the electrolytic hydrogen production device and the hydrogen fuel cell; r λ_SU_t and r λ_SD_t They respectively indicate the start and stop status of the hydrogen electrolysis device and the hydrogen fuel cell. Maintenance costs of hydrogen electrolysis equipment and hydrogen fuel cells In the formula, κ λ_OM Represents the unit time maintenance cost of the hydrogen electrolysis device and hydrogen fuel cell. The life attenuation cost corresponding to the historical operating time of the electrolytic hydrogen production device and the hydrogen fuel cell is shown in the formula: In the formula, κ λ_AGE T represents the life attenuation cost of the hydrogen electrolysis device / hydrogen fuel cell per unit operating time; λ_LT Indicates the expected life of the hydrogen electrolysis device / hydrogen fuel cell. The total operating cost of hydrogen energy storage is:

5. According to claim 6, the island electricity-hydrogen transmission coupling optimization scheduling method based on the improved grey wolf algorithm is characterized in that the electricity-hydrogen-transportation ocean island low-carbon scheduling model includes a marine transportation network model, which is as follows: The size of HFCS hydrogen load is mainly affected by factors such as driving mileage and return time as well as the supply and demand of hydrogen refueling stations. The transportation network model is established as follows. The probability density function of the last return time of a user every day is: Where: x t is the user's last return time, ψ t is the expected value, γ t is the standard deviation. The probability density function of HFCS user mileage is defined as: Where: x s is the user's last return time, ψ s is the expected value, γ t is the standard deviation. HFCS hydrogenation time: Where: T s is the hydrogen refueling time, S is the daily mileage, P is s is the amount of hydrogenation, σ s is the hydrogen refueling efficiency, and E is the hydrogen consumption per unit kilometer. Single station hydrogen refueling demand N m,HFCS ≤N m,H (17) Where: N m,HFCS is the number of HFCS currently available at the mth hydrogen refueling station, N m,H is the amount of HFCS that the mth hydrogen refueling station can accept. Total hydrogen demand in the transportation network Where: M is the total number of hydrogen refueling stations in the transportation network, M m,cqz is the capacity of the hydrogen storage station, M e,HFCS is the e-th HFCS capacity, N HFCS is the amount of HFCS. Short-term hydrogen energy storage model Where: and They are the hydrogen storage per unit time, hydrogen release per unit time and the total amount of remaining hydrogen in short-term hydrogen energy storage. and are the hydrogen storage efficiency, hydrogen release efficiency and self-release rate of STHS, C STHS is the configuration capacity of STHS, and Δt is the unit time step. Shared Hydrogen Storage Model Where: and They are the amount of hydrogen stored per unit time, the amount of hydrogen released per unit time and the total amount of hydrogen remaining in hydrogen energy storage. and are the hydrogen storage efficiency, hydrogen release efficiency and self-release rate of SHS, respectively, C SHS The configuration capacity of SHS. SHS is used when STHS cannot meet the hydrogen load, and the order of use is: real-time hydrogen replenishment > STHS > SHS. HFCS hydrogenation must meet certain conditions. The actual amount of hydrogen cannot exceed the total capacity of the battery, but at the same time it cannot be less than the amount of hydrogen expected by the user. Assuming that when the ship owner connects the HFCS to the hydrogenation pile, the user will provide his or her expected information, such as the desired hydrogen state when leaving the hydrogenation pile is S. h , which means that when the user decides to leave, the actual amount of hydrogen in HFCS should be greater than or equal to the desired value, which can be expressed as Where: S st,i is the hydrogen charge state of the i-th HFCS vehicle when it starts to be refueled, S en,i It is the hydrogen charge state when the i-th amount of HFCS leaves.

6. The island electricity-hydrogen transmission coupling optimization scheduling method based on the improved grey wolf algorithm according to claim 6 is characterized by: The island two-layer optimization scheduling model includes a self-consistent model of the upper ocean island operation model and the lower ocean island self-consistent model; including: 1) Upper-level transportation network-ocean island operation model Objective Function Where: f1 represents the operating cost of the offshore island, C HS , C WT , C PV They represent the operating cost of hydrogen energy storage, wind power generation device, and photovoltaic power generation device, respectively. f2 represents the hydrogenation cost; it is the hydrogenation status of the i-th ship in the j-th period, and has only two values: 0 or 1, 0 means no hydrogenation, 1 means hydrogenation; p i is the power of the i-th ship; p j represents the hydrogen price in period j; N represents the total number of HFCS ships. Constraints Hydrogen energy storage constraints, power balance constraints, HFCS hydrogenation constraints. 2) Lower-level transportation network-self-consistent model of ocean islands Based on the bidirectional coupling architecture of ocean island new energy and transportation network, a low-carbon economic optimization scheduling objective function is established, which includes two parts: one is the highest self-consistency rate of system electricity and hydrogen energy; the other is the lowest system carbon emission cost. The objective function is as follows: Where: η is the daily self-consistency rate of electricity-hydrogen-transportation coupling on the ocean island, F E is the electrical load on the island, F DG is the output power of distributed energy, F H is the hydrogen load of the transportation network, F q is the energy output of the hydrogen refueling station, F is the system carbon emission cost, α is the carbon emission tax amount, is the carbon emission cost. c is the carbon trading price, c is the basic unit carbon price, β is the step price increase, and v is the length of the carbon emission interval. γ is the reward coefficient in the carbon trading price; E g is the actual carbon emission generated by starting the diesel generator, and a1, b1, and c1 are the carbon emission calculation coefficients of the diesel generator. c is the carbon emission quota for electricity generated by diesel generators, χ is the carbon emission quota per unit of electricity, P t E It is the amount of electricity generated by the diesel generator per unit time. Constraints Distribution network operation constraints, electrolyzer equipment operation constraints, hydrogen fuel cell equipment operation constraints, HFCS and hydrogen refueling station constraints, and hydrogen energy storage equipment operation constraints.

7. The island electricity-hydrogen transmission coupling optimization scheduling method based on the improved grey wolf algorithm according to claim 1 is characterized by: In step 3, the grey wolf algorithm is improved by introducing chaotic adversarial learning, differential evolution and individual perturbation mechanism to improve the optimization ability and convergence speed of the algorithm. 1) Tent adversarial learning optimizes population initialization In the search space, the position of the gray wolf is initialized using chaotic tent mapping to form the initial population X. Then, the opposing population X′ is generated through adversarial learning. The fitness of these two populations is sorted, and the individuals with the highest ranking will be determined as the final initial population. The initial population created using the Tent map is: Where: i represents the population size; i = 1, 2, 3, ..., N; j represents the index of the chaotic sequence, j = 1, 2, 3, ..., d; μ represents the chaotic parameter, and its value is in the range of [0, 2]. The adversarial solution generated by the adversarial learning mechanism is: x i,j =l i,j +y i,j ×(u i,j -l i,j ),x i ∈X (28) Where: l ij 、u ij Represents element x i ∈X. And compare the fitness after it is generated, and retain the optimal solution. 2) Position update based on differential evolution The traditional gray wolf algorithm first generates the initial gray wolf population position with a random position, and then the α, β, and γ wolves hunt for prey and update their own positions through equations (29) to (31). Where: D is the distance between the wolf and its prey; C and A are coefficient vectors. The wolf gradually approaches its prey by adjusting A and D. When the number of iterations reaches the maximum value T max , the algorithm ends running; t-current number of iterations; represents the location of the prey that α, β, and γ wolves want to catch; X t Represents the position of the gray wolf itself; X1, X2, and X3 represent the updated positions of α, β, and γ wolves. Based on the differential mutation method, this paper introduces mutation operators, crossover operators and selection operators to increase the population size and category richness and optimize the position update process. ① Mutation operator The mutation solution is expressed by the following formula, where the difference changes between different individual solutions are determined by the mutation operator. Where: v represents the variant solution; t represents the number of iterations; x represents the original population solution; N represents the population size; r1, r2, r3 represent random integers in [1, N]; ω represents the mutation operator that affects the differential changes between individual solutions. ② Crossover operator The crossover operator performs crossover processing on the individual solutions that have completed the mutation to increase the richness of the population, as shown in the following formula. Where: rand represents a random value in [0,1]; C R represents the individual cross-coefficient. ③Select operator A greedy strategy is used to compare parent and offspring individuals, and new individuals are selected by the following formula: Where: f(z) represents the fitness function. Where: P i Represents the probability that the gray wolf chooses the differential evolution strategy to update the position. If P≥rand, the differential evolution method is not used to update the position, but the traditional gray wolf algorithm is used to update the position; if P<rand, the differential evolution method is used to update the position. 3) Individual disturbance Individual perturbations enhance the global search capability. When two gray wolves gradually approach each other, individual perturbations will make the gray wolves converge to the initial point, thereby maintaining the diversity of the population, as shown in the following formula. Where: D op represents the perturbation operator; represents the Euclidean distance between the locations of gray wolf i and gray wolf j; Represents a random quantity uniformly distributed in [a,b].

Images