MPPT and capacity matching design optimization method for wind power hydrogen production system

Abstract

This invention discloses a method for coordinated optimization of MPPT (Multi-Level Testing) and capacity in a wind power-to-hydrogen system. Meteorological data and system operating parameters are collected, a component mathematical model is established, and the optimization objective and operating boundary are determined. A two-layer collaborative algorithm (upper quantum layer and lower myxomycete layer) is employed to optimize MPPT control parameters and system capacity configuration respectively, and a two-layer collaborative feedback update is performed. The parent generation's non-dominated optimal solution, quantum layer offspring, and myxomycete layer offspring are merged to form a joint population. The Pareto front solution set is screened through non-dominated sorting and crowding distance calculation. After iterative convergence, the Pareto optimal solution set is output, generating a matching MPPT control parameter and capacity configuration scheme. This invention can solve the problems of response lag, susceptibility to local optima, and disconnect between control and capacity matching under wind speed fluctuations, thereby improving wind energy utilization.

Description

Technical Field

[0001] This invention relates to the field of wind power hydrogen production technology, and in particular to a method for optimizing MPPT and capacity matching design of a wind power hydrogen production system. Background Technology

[0002] As the energy structure shifts towards low-carbon and distributed energy, the coupled application of wind power generation and water electrolysis for hydrogen production has become an important way to solve the problem of renewable energy consumption and large-scale green hydrogen production. However, wind energy has extremely strong randomness and volatility, which brings dual challenges to the stable and reliable operation and system planning of off-grid wind power hydrogen production systems.

[0003] In terms of energy control, wind speed fluctuations cause rapid fluctuations in wind turbine output power. Therefore, wind turbines need to use Maximum Power Point Tracking (MPPT) technology to cope with rapid wind speed changes and reduce issues such as DC bus voltage deviation, increased converter losses, and electrolytic cell polarization losses and aging. Traditional hill-climbing search (P&O) or power curve (PC) methods often suffer from response lag, oscillations around the optimal operating point, and a tendency to get trapped in local optima when facing turbulent winds or gusts, resulting in reduced wind energy utilization efficiency (C). p The low capacity makes it difficult to provide maximum energy input for hydrogen production. At the system planning level, existing research on capacity matching of wind power hydrogen production systems (including electrolyzers, hydrogen storage tanks, energy storage systems, and compressors) mostly adopts independent optimization or segmented optimization methods, which separates maximum power point tracking control from capacity matching, ignoring the dynamic coupling relationship and synergistic effects between the two. This results in "optimal design but poor operation," leading to high wind curtailment rates or hydrogen load shortages in the system. Summary of the Invention

[0004] The purpose of this invention is to provide a method for optimizing MPPT and capacity matching design in a wind power hydrogen production system, which solves the problems of existing technologies such as slow response to wind speed fluctuations, getting trapped in local optima leading to low wind energy utilization, and the disconnect between capacity matching and MPPT control.

[0005] To achieve the above objectives, this invention provides a method for optimizing the MPPT and capacity matching design of a wind power hydrogen production system, comprising the following steps: S1. Collect meteorological data and operating parameters of each component of the wind power hydrogen production system in the target area to establish mathematical models of each component of the wind power hydrogen production system, and establish the optimization target system and operating boundary conditions of the wind power hydrogen production system. S2. Based on the mathematical models of each component of the wind power hydrogen production system, the optimization target system and the operating boundary conditions, a two-layer collaborative algorithm consisting of an upper quantum layer and a lower slime mold layer is adopted to optimize the MPPT control parameters and system capacity configuration parameters respectively, and to perform two-layer collaborative feedback updates. S3. Merge the parent generation's non-dominated optimal solution, the quantum layer's offspring MPPT control parameters, and the slime mold layer's offspring capacity configuration parameters to form a joint population. Filter the Pareto front solution set through non-dominated sorting and crowding distance calculation, and return to S2 for iteration. After the iteration is complete, output the current Pareto front solution set as the Pareto optimal solution set. S4. Based on the Pareto optimal solution set, generate matching MPPT control parameters and system capacity configuration scheme.

[0006] In one possible implementation, in S2, the upper quantum layer optimizes the MPPT control parameters, including: U1. Constructing a quantum population using qubits; the quantum population includes multiple quantum individuals, each corresponding to a set of MPPT control parameters; Among them, the The encoding matrix of each quantum individual for The complex matrix is ​​expressed as: ; ; In the formula, Indicates the first Among the individuals, the first The probability magnitude of a quantum bit collapsing to a state of 0; Indicates the first Among the individuals, the first The probability magnitude of a qubit collapsing to a state of 1. , The length of the quantum bit encoding is equal to the total number of MPPT control parameters; where the collapse states of 0 and 1 correspond to the lower and upper limits of the MPPT control parameters, respectively. U2. Based on the current quantum probability amplitude, a set of determined MPPT control parameters are generated through observation. The specific observation mechanism is as follows: for the first... The first individual One qubit, producing one Uniformly distributed random numbers within the interval ,like If the state of the qubit is 1, then the qubit collapses to 1; otherwise, the qubit collapses to 0, resulting in a binary sequence. The binary sequence is then decoded into MPPT control parameters using a linear mapping to obtain the corresponding MPPT control parameters. U3. Dynamically update the probability amplitude of the qubit through the quantum rotation gate to optimize the generation of MPPT control parameters; U4. The optimized MPPT control parameters are converted into a fan power curve that takes into account the actual operating characteristics of the fan, and then output to the lower slime mold layer.

[0007] In one possible implementation, the expression for the probability magnitude of updating the qubit is: ; ; In the formula, and These represent the values ​​after the quantum rotation gate update. Among the individuals, the first The new probability magnitudes of each qubit collapsing into 0 and 1 states; Indicates the first Among the individuals, the first The rotation angle of each qubit; Indicates the step size of the rotation angle; This represents the direction determination function, ensuring that the probability amplitude converges towards the optimal solution region.

[0008] In one possible implementation, in S2, the lower slime mold layer optimizes the system capacity configuration parameters, including: D1. Obtain the wind turbine power curve output from the upper quantum layer, perform time-series matching with the system load demand, and calculate the energy flow and balance state of the wind power hydrogen production system; construct the objective function and constraint function for optimizing the capacity configuration of the wind power hydrogen production system. D2. Initialize the slime mold population location vectors, with each location vector corresponding to a set of capacity configuration parameters; D3. Based on the energy flow and balance state of the wind power hydrogen production system, and based on the objective function and constraint function, calculate the fitness value of each individual in the slime mold population, and sort the fitness values ​​of all individuals in ascending order to obtain the fitness value sequence. D4. Based on the step fitness ranking results, calculate the adaptive weight of each slime mold individual; D5. Simulate the foraging behavior of slime molds. Based on adaptive weights and combined with the slime mold position update formula, update the slime mold population position within the capacity configuration target space, optimize the capacity configuration parameters round by round, and feed them back to the upper quantum layer.

[0009] In one possible implementation, the adaptive weight of the slime mold individual is calculated using the following formula: ; In the formula, Let r represent the adaptive weight of the i-th slime mold individual after sorting. Random numbers within the interval; This represents the fitness value of the i-th individual in the slime mold population; Indicates the first The fitness value of the best solution in the next iteration of the population; Indicates the first The fitness value of the worst solution in the next iteration of the population.

[0010] In one possible implementation, the slime mold location update formula is: ; In the formula, This represents the new position vector of the generated slime mold individual after the update; and These represent the upper and lower boundaries of the search range, respectively. This indicates that the slime mold generated in the lower layer is... A random number that is uniformly distributed within an interval; The probability threshold for initializing a slime mold individual at a random location; This represents the dynamic probability threshold for controlling the selection of different search mechanisms by slime mold individuals. , This represents the fitness value of the i-th slime mold individual. Indicates the population size of the slime mold algorithm; Indicates until the The position vector of the slime mold individual with the best fitness in the next iteration; The biological oscillation weights of slime molds; Indicates the first A slime mold individual randomly selected in the next iteration; Indicates the first Another slime mold individual randomly selected in the next iteration; express A vector of oscillation control parameters randomly generated within the interval. The boundary control parameter represents a non-linearly decreasing parameter that varies with the number of iterations. ; Indicates in A vector of shrinkage control parameters that decreases within the interval; Indicates the first The location of the slime mold individual in the next iteration; This represents the optimal fitness value obtained in all iterations so far. This represents the maximum number of iterations.

[0011] In one possible implementation, the two-layer collaborative feedback update is as follows: the upper quantum layer transmits the MPPT control parameters to the lower layer to correct the optimization direction of the slime mold population capacity; the lower slime mold layer feeds back the capacity configuration parameters to the upper layer to correct the update direction of the quantum population rotation angle, thereby realizing bidirectional collaborative iteration of control and capacity.

[0012] In one possible implementation, the two-layer collaborative feedback update further includes: if the MPPT control parameters output by the upper quantum layer make the capacity matching cost of the lower slime mold layer lower than the cost threshold and the efficiency higher than the efficiency threshold, then the positive rotation angle is increased to enhance the probability amplitude of the MPPT control parameters in the upper quantum layer population; otherwise, it indicates capacity configuration redundancy and suppresses the probability amplitude of the MPPT control parameters in the upper quantum layer population.

[0013] In one possible implementation, the dynamic correction of rotation angle compensation is expressed as: ; In the formula, Indicates the corrected rotation angle; Indicates the rotation angle before correction; Indicates the cooperative gain coefficient; This indicates the hypervolume increment of the Pareto front in the lower slime mold layer.

[0014] In one possible implementation, S3 specifically includes: S31. The parent generation non-dominated optimal solution, the quantum layer offspring MPPT control parameters, and the slime mold layer offspring capacity configuration parameters are merged to form a joint population; each individual in the joint population is an optimization scheme containing a complete set of MPPT control parameters and system capacity configuration parameters; S32. Perform non-dominant ranking on the joint group, dividing each individual in the joint group into different levels based on Pareto undominant relations. ; The crowding distance is calculated for individuals within the same level, using the following expression: ; In the formula, CD e Represented as the first The crowding distance of each individual; Indicates the index of the objective function; Indicates the number of objective functions; and They represent the first time. Under one objective function, and The objective function values ​​of the preceding and following individuals of an individual; and These represent the individual at the current level. The maximum and minimum values ​​of the objective function; S33. Based on the non-dominated sorting results and the crowding distance screening results, select the Pareto front solution set in the joint population to form a new generation of population for population update in the next iteration; return to S2 for iteration; after the iteration ends, output the current Pareto front solution set as the Pareto optimal solution set.

[0015] In one possible implementation, in S3: after several iterations, determine the improvement amount of the current Pareto front solution set. With set threshold The relationship between them; if If the condition is met, return to S2 and perform optimization operations by updating the part through the quantum rotation gate; otherwise, terminate the iteration and output the current Pareto front solution set as the Pareto optimal solution set.

[0016] Beneficial effects: This invention employs the aforementioned MPPT and capacity matching design optimization method for a wind power-to-hydrogen system. It overcomes the limitations of traditional capacity configuration methods that rely solely on static or ideal power curves. By deeply coupling dynamic MPPT control with static capacity planning through a two-layer architecture, it improves the realism and reliability of the system design. Utilizing the full-space search capability of upper-layer quantum evolution, it avoids MPPT from getting trapped in local optima. Combined with the fine-grained search advantage of the lower-layer slime mold algorithm in continuous space, it significantly improves the convergence and uniformity of the multi-objective Pareto front, thereby enhancing the overall energy efficiency of the system and enabling efficient coordinated operation between the system control and operation layers. Attached Figure Description

[0017] Figure 1 This is an overall flowchart of the MPPT and capacity matching design optimization method for a wind power hydrogen production system according to the present invention; Figure 2 The energy efficiency and overall system efficiency of components in five different regions are presented to illustrate the implementation of this invention; where (a) is region A and (b) is region B. Figure 3 The figures represent the power generation and hydrogen production in different regions under Case 5 of this invention; where (a) is region A and (b) is region B. Detailed Implementation

[0018] The following detailed description of embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0019] Please see Figure 1 A method for optimizing MPPT and capacity matching design in a wind power hydrogen production system includes the following steps:

[0020] S1. Collect meteorological data and operating parameters of each component of the wind power hydrogen production system in the target area to establish mathematical models of each component of the wind power hydrogen production system, and establish the optimization target system and operating boundary conditions of the wind power hydrogen production system.

[0021] In some embodiments, for a target research area, meteorological data, including wind speed, wind direction, ambient temperature, humidity, and atmospheric pressure, can be measured using devices such as thermocouples, anemometers, and hygrometers.

[0022] Based on thermodynamic and electrochemical principles, mathematical models of each component of a wind power-to-hydrogen system are constructed. These components include a wind turbine system, an energy storage system, and an electrolyzer. For the wind turbine system, models such as turbine power curves, wind energy utilization rate, and output power / torque can be built. For the electrolyzer, a PEM (Proton Exchange Membrane) electrolyzer characteristic model incorporating ohmic polarization, activation polarization, and concentration polarization, as well as a nonlinear model of hydrogen production rate, can be constructed. For the energy storage system, models of energy storage charge / discharge efficiency and hydrogen storage pressure dynamics can be established. For the hydrogen storage system, a compressor energy consumption power model can be established.

[0023] By setting the system's operational boundary conditions and optimization target system (evaluation index system), quantitative standards are provided for subsequent multi-objective optimization.

[0024] S2. Based on the mathematical models of each component of the wind power hydrogen production system, the optimization target system and the operating boundary conditions, a two-layer collaborative algorithm consisting of an upper quantum layer and a lower slime mold layer is adopted to optimize the MPPT control parameters and system capacity configuration parameters respectively, and to perform two-layer collaborative feedback updates.

[0025] This step provides a robust search mechanism for multi-region MPPT control and system capacity design optimization under different scenarios.

[0026] In some embodiments, this step involves constructing a two-layer collaborative algorithm, which forms a basic framework for independent solution and information exchange between the two-layer algorithms through a two-layer variable mapping and interaction mechanism.

[0027] The two-layer collaborative algorithm is constructed, specifically including:

[0028] First, based on the optimization objective system and system boundary, the decision variables of the optimization problem are defined. And the hierarchical logic of decision variables, which breaks down decision variables into upper-level control variables. and lower-level capacity configuration variables And strictly controlled within their respective feasible regions; the upper-level decision variables correspond to the key parameters of the wind turbine MPPT controller, including PID gain. or optimal tip speed ratio , Indicates proportional gain. This represents the integral gain; the lower-level capacity configuration variable corresponds to the capacity design optimization variable of the wind power hydrogen production system.

[0029] Secondly, an inter-layer interaction protocol is defined: the upper quantum layer transmits power flow data to the lower slime mold layer, and the lower slime mold layer performs multi-dimensional evaluation in conjunction with the optimization objective system and feeds back the fitness value to the upper quantum layer; the inter-layer interaction protocol enables the two sub-algorithms to collaboratively solve the optimization objective system through bidirectional information transmission when processing different optimization sub-objectives, ensuring that the joint search process is carried out within the system's operational boundary conditions; a total decision space is set. For the upper space With the lower space The Cartesian product is expressed as: ; In the formula, This represents the first optimization decision variable in the upper quantum control layer; This represents the second optimization decision variable in the upper quantum control layer; Indicates the upper quantum control layer. One optimal decision variable, The qubit encoding length (i.e., the total number of MPPT control parameters); This represents the discrete optimization decision variables in the lower slime mold layer. The number of discrete optimization decision variables. This represents the continuous optimization decision variables in the lower slime mold layer. To continuously optimize the number of decision variables, Each element corresponds to a capacity configuration dimension to be optimized; in some embodiments, the discrete optimization decision variable may include the number of wind turbines. Number of electrolytic cells connected in series Number of electrolytic cells in parallel configuration and the number of hydrogen storage tanks Continuous optimization decision variables may include the rated capacity of the energy storage system. .

[0030] The expression for the two-level interaction function is: ; In the formula, This indicates the optimal power input from the upper layer to the lower layer; express The power of the fan at all times; This represents the dynamic function of the wind turbine after MPPT control; express The wind speed entering the fan at all times.

[0031] In some embodiments, this step may further include: creating an external archive for storing non-dominated solutions, the external archive serving as a shared bootstrapping library between the upper quantum layer and the lower slime mold layer.

[0032] In some embodiments, this step may further include: defining basic control parameters, including population size. Maximum number of iterations QIE (Quantum Evolutionary Algorithm) and MOSMA (Multi-Target Slime Mold Algorithm) algorithm parameters and qubit encoding. and external file size.

[0033] The upper quantum layer optimizes the binary variables of the MPPT control parameters, including: inputting real-time wind speed data from meteorological data of the target area, combining it with the wind turbine power characteristic model, using a quantum evolution algorithm, determining the rotation direction of the quantum rotating gate through a direction judgment function, and updating the probability amplitude of the qubits; performing a qubit collapse operation based on the updated probability amplitude to generate optimized MPPT control parameters, and finally obtaining a wind turbine power curve that takes into account the actual operating characteristics of the wind turbine and can be directly used for lower-level optimization.

[0034] In some embodiments, the upper quantum layer optimizes the binary variables of the MPPT control parameters, specifically including: The upper quantum layer addresses the discrete binary characteristics in MPPT control through qubits. Constructing a quantum population enhances global exploration capabilities, and the MPPT control parameter mapping must meet the operating power characteristics of the system's fan system and electrolytic cell.

[0035] The state of a qubit is described by a pair of complex probability amplitudes, expressed as: ; In the formula, The state vector representing a single qubit; Indicates the first Among the individuals, the first The probability magnitude of a quantum bit collapsing to a state of 0; Indicates the first Among the individuals, the first The probability amplitude of a qubit collapsing to a state of 1; and These are the two orthogonal ground states of a quantum bit, corresponding to the 0 and 1 states of classical binary, respectively, and in this application, they correspond to the lower and upper limits of the MPPT control parameters.

[0036] and The expression satisfies the normalization condition and is: ; No. The encoding matrix of each quantum individual for The complex matrix is ​​expressed as: ; In the formula, The length of the quantum bit encoding is equal to the total number of MPPT control parameters. Indicates the first The probability amplitude that the first qubit in an individual is in a collapsed state of 0; Indicates the first The probability amplitude that the second qubit in an individual is in a collapsed state of 0; Indicates the first Among the individuals, the first The probability magnitude of a quantum bit collapsing to a state of 0; Indicates the first The probability amplitude that the first qubit in each individual is in a collapsed state of 1; Indicates the first The probability amplitude that the second qubit in an individual is in a collapsed state of 1; Indicates the first Among the individuals, the first The probability amplitude of a qubit collapsing to a state of 1.

[0037] Based on the current quantum probability amplitude, a set of defined MPPT control parameters are generated through observation. The specific observation mechanism is as follows: for the first... The first individual One qubit, producing one Uniformly distributed random numbers within the interval ,like If the state of the qubit is 1, then the qubit collapses to state 1; otherwise, the qubit collapses to state 0, resulting in a binary sequence. A linear mapping is used to decode the binary sequence into MPPT control parameters (continuous real-field physical control parameters).

[0038] By dynamically updating the probability amplitude (phase) of qubits through quantum rotation gates, the population can be driven to evolve towards a better control strategy. This helps maintain population diversity and avoid premature convergence. The expression is as follows: ; ; In the formula, This indicates the number of times the quantum rotating gate has been updated. Among the individuals, the first The new probability magnitude of a qubit collapsing to a state of 0; This indicates the number of times the quantum rotating gate has been updated. Among the individuals, the first The new probability magnitude of a qubit collapsing to a state of 1; Indicates the first Among the individuals, the first The rotation angle of each qubit includes a forward rotation angle and a reverse rotation angle. The forward rotation angle increases the probability of the qubit collapsing to the "1" state, guiding the quantum population towards better MPPT control parameters. The reverse rotation angle increases the probability of the qubit collapsing to the "0" state. This is determined by a direction determination function. Determine the direction of rotation (forward or reverse) of the quantum rotating gate; The step size represents the rotation angle, and its sign and magnitude are determined by the position difference and fitness difference between the current qubit solution and the reference target solution. This represents the direction determination function, ensuring that the probability amplitude converges towards the optimal solution region.

[0039] After iteration to convergence, the quantum probability amplitude corresponding to the optimal control strategy approaches 1 or 0. The binary variables of the MPPT control parameters are obtained through quantum collapse and used as the optimized MPPT control parameters to drive the pitch system or power converter, thereby obtaining the optimal power output sequence optimization curve of the wind power system. The optimal power output sequence of the wind power system is loaded into the wind turbine simulation model as a power tracking target; real-time wind speed data is input to make the wind turbine simulation model run under the constraint of the optimal power command, so as to obtain the wind turbine output power time series considering the actual operating characteristics of the wind turbine, thereby generating the wind turbine power curve.

[0040] Among them, based on real-time wind speed data and the aerodynamic principles of wind power generation, the output power of the wind turbine is calculated, and the expression is: ; In the formula, Indicates the output power of the wind turbine; Indicates air density; This represents the swept cross-sectional area of ​​the blades in a wind turbine generator system. The wind energy utilization coefficient is the ratio of the mechanical power output by the wind turbine to the wind energy power input into the rotor surface. The tip speed ratio represents the ratio of the linear velocity at the tip of the fan blade to the wind speed. Indicates the blade pitch angle; Indicates wind speed; Indicates the radius of the blade wheel; The wind energy utilization coefficient is expressed as follows: ; In the formula, All represent the fitted parameters; This indicates the correction of blade starting characteristic parameters; and and satisfy: ; ; In the formula, n represents the angular velocity of the wind turbine blades. w This indicates the rotational speed of the wind turbine generator.

[0041] The above process simulates the dynamic capture capability of the wind turbine under actual operating wind conditions. The generated wind turbine power curve contains the real control losses and dynamic characteristics, providing a high-precision input source for subsequent capacity matching.

[0042] Lower slime mold layer processing capacity configuration variables: The lower slime mold layer uses the wind turbine power curve as the input source, calls the mathematical model and optimization target system, and optimizes the capacity of the wind power hydrogen production system through the slime mold predation mechanism to generate optimized capacity configuration parameters, providing an initial basis for subsequent evolutionary screening.

[0043] In some embodiments, the lower slime mold layer optimizes the system capacity configuration parameters, specifically including: The wind turbine power curve output from the upper quantum layer is obtained and time-matched with the system load demand to calculate the energy flow and balance state of the wind power hydrogen production system; providing data support for quantitative evaluation of optimization objectives and constraints on the system's operating boundary conditions.

[0044] Construct the objective function and constraint function for optimizing the capacity configuration of a wind power-to-hydrogen system.

[0045] The constraints are constructed by considering the numerical characteristics of capacity configuration, using a group of real numbers. The search space is set based on the system's operational boundary conditions in S1. , This indicates the lower bound of the search boundary for capacity configuration variables. This represents the upper limit of the search boundary for the capacity configuration variable and initializes the position vector of the slime mold population.

[0046] The constraints include real-time power balance constraints, hydrogen storage capacity and energy storage capacity constraints, and upper and lower limits on the number of devices. The rules for co-satisfying the constraints adopt the constraint dominance principle: during the evolution process, the real-time power balance constraint is a hard constraint that must be strictly satisfied. If the real-time power balance constraint is violated, the individual is directly determined to be infeasible. For soft constraint boundaries such as the number of devices, when a candidate solution exceeds the boundary, it is pulled back to the nearest boundary point. During non-dominated sorting, the degree of constraint violation of individuals is compared first. For individuals with zero degree of constraint violation, the merits of the objective function are compared, thereby guiding the population to converge to the feasible region.

[0047] The objective vector form of a multi-objective optimization problem is: ; In the formula, Represents the target vector. The first step represents the optimization of system capacity configuration. Sub-objective functions, , The number of sub-objective functions. Sub-objective functions of the system capacity configuration optimization objective function generally refer to economic indicators, energy efficiency indicators, environmental indicators, social benefit indicators, engineering and technical indicators, and efficiency indicators. For example, sub-objective functions can be selected as the system's total life cycle cost, the system's power shortage rate, the system's carbon emissions, the reciprocal of the system's hydrogen production efficiency, and the reciprocal of the wind energy utilization rate.

[0048] The expression for the real-time power balance constraint is: ; ; In the formula, express Real-time power of the electrolytic cell at any given moment; This represents the optimized output of the upper quantum layer. The maximum power of the fan at any given time is the power at the tracking point. express The output power of the fan system at all times; express Discharge power of the energy storage system at any time; express The charging power of an energy storage system at any given time is such that the energy storage system can only be in one of the charging or discharging states at any given time. express The power consumed by the compressor at all times.

[0049] The expressions for hydrogen storage capacity and energy storage capacity constraints are as follows: ; In the formula, Indicates the lower limit of the energy storage system capacity; Indicates the capacity of the energy storage system; Indicates the upper limit of the energy storage system capacity; This indicates the lower limit of the hydrogen storage system capacity (minimum safe capacity). Indicates the capacity of the hydrogen storage system; This indicates the upper limit of the hydrogen storage system capacity (maximum design capacity).

[0050] The expressions for the upper and lower limits of the equipment quantity constraints are as follows: ; In the formula, Indicates the lower limit of the number of wind turbines; Indicates the number of wind turbines; Indicates the maximum number of wind turbines; Indicates the lower limit of the number of electrolytic cells connected in series; Indicates the number of electrolytic cells connected in series; Indicates the maximum number of electrolytic cells that can be connected in series; Indicates the lower limit of the number of electrolytic cells that can be connected in parallel; Indicates the number of electrolytic cells connected in parallel; Indicates the maximum number of electrolytic cells that can be connected in parallel; This indicates the minimum number of hydrogen storage tanks required. Indicates the number of hydrogen storage tanks configured; This indicates the maximum number of hydrogen storage tanks that can be configured.

[0051] Initialize the slime mold population location vectors, with each location vector corresponding to a set of capacity configuration parameters.

[0052] Capacity configuration parameters may include the number of wind turbines, the number of electrolyzers connected in series, the number of electrolyzers connected in parallel, the number of hydrogen storage tanks, and the rated capacity of the energy storage system.

[0053] Based on the energy flow and balance state of the wind power hydrogen production system, and using the objective function and constraint function, the fitness value of each individual in the slime mold population is calculated, and the fitness values ​​of all individuals are sorted in ascending order to obtain a fitness value sequence. ; This represents a sequence of fitness values ​​arranged in ascending order. This represents a sorting operator function that sorts the set of fitness values ​​of a population in ascending order.

[0054] D4. Based on the step fitness ranking results, calculate the adaptive weight of each slime mold individual (used to simulate the changes in the thickness of slime mold channels). The specific calculation formula is as follows: ; In the formula, Let r represent the adaptive weight of the i-th slime mold individual after sorting. Random numbers within the interval; This represents the i-th individual in a slime mold population (each individual corresponds to a set of decision variables). The fitness value (of which the value is taken); Indicates the first The fitness value of the best solution in the next iteration of the population; Indicates the first The fitness value of the worst solution in the next iteration of the population.

[0055] Among them, the top 50% of high-quality individuals in the population refer to those located in the latter half of the sequence after ascending order and with higher fitness values, while the remaining individuals are those in the first half of the sequence and with lower fitness values, determined through adaptive weighting. The thickness variation of slime mold channels is simulated, and the algorithm is adaptively adjusted according to fitness ranking through a positive-negative feedback mechanism. That is, individuals with higher ranking are given stronger guidance strength, guiding the algorithm to converge toward the optimal capacity configuration solution.

[0056] Simulating the foraging behavior of slime molds, based on adaptive weights and combined with the slime mold location update formula, the slime mold population location is updated within the capacity configuration target space. The capacity configuration parameters are optimized round by round and fed back to the upper quantum layer.

[0057] The expression for slime mold location update is: ; ; ; ; ; In the formula, This represents the new position vector of the generated slime mold individual after the update; Indicates the upper boundary of the search range; Indicates the lower boundary of the search range; This indicates that the slime mold generated in the lower layer is... A random number that is uniformly distributed within an interval; This represents the probability threshold for initializing a slime mold individual at a random location, and is typically set to 0.03. This represents the dynamic probability threshold controlling the selection of different search mechanisms by slime mold individuals; Indicates until the The position vector of the slime mold individual with the best fitness in the next iteration; The biological oscillation weights of slime molds; Indicates the first A slime mold individual randomly selected in the next iteration; Indicates the first Another slime mold individual randomly selected in the next iteration; express A vector of oscillation control parameters randomly generated within the interval; Indicates in The shrinkage control parameter vector decreases within the interval, and the two oscillation parameters gradually approach 0 as the number of iterations increases; Indicates the first The location of the slime mold individual in the next iteration; This represents the optimal fitness value obtained in all iterations so far. Represents the boundary control parameters that decrease non-linearly with the number of iterations; This represents the maximum number of iterations.

[0058] The slime mold algorithm dynamically adjusts the search step size through biological oscillation weights, focusing on global exploration in the early stages of the search and local development in the later stages.

[0059] During the solution process, a two-layer collaborative feedback update is adopted: The upper quantum layer transmits the optimized MPPT control parameters as guidance information to the lower slime mold layer to correct the capacity optimization direction.

[0060] If the MPPT control parameters output by the upper quantum layer make the capacity matching cost of the lower slime mold layer lower than the cost threshold and the efficiency higher than the efficiency threshold, then the positive rotation angle is increased to enhance the probability amplitude of the MPPT control parameters in the upper quantum layer population; otherwise, it indicates capacity configuration redundancy and suppresses the probability amplitude of the MPPT control parameters in the upper quantum layer population. This mechanism realizes a two-way synergy of "optimizing control through design and improving design through control".

[0061] The dynamic correction for rotation angle compensation, i.e., increasing the positive rotation angle, is expressed as: ; In the formula, Indicates the corrected rotation angle; Indicates the rotation angle before correction; Indicates the cooperative gain coefficient; This represents the Pareto front hypervolume increment of the lower slime mold layer (the difference between the hypervolume of the Pareto front solution set in the current iteration and the hypervolume of the previous iteration, used to quantify the distribution and diversity changes of the multi-objective optimization solution set, and as a criterion for judging the convergence of the slime mold algorithm).

[0062] The lower slime mold layer feeds back the optimized capacity configuration parameters to the upper quantum layer to correct the update direction of the MPPT control parameters.

[0063] Thus, the collaborative update of the results of the two-layer algorithm is completed.

[0064] S3. Merge the non-dominated optimal solutions of the parent generation, the MPPT control parameters of the offspring in the quantum layer, and the capacity configuration parameters of the offspring in the slime mold layer to form a joint population. Filter the Pareto front solution set (non-dominated and uniformly distributed individuals, i.e., high-quality individuals) through non-dominated sorting and crowding distance calculation, and return to S2 for iteration. After the iteration is completed, output the current Pareto front solution set as the Pareto optimal solution set.

[0065] In some embodiments, this step specifically includes: The non-dominated optimal solution (parent generation) retained from the previous iteration. The newly generated MPPT control signal sequence (quantum layer offspring) by the upper-level quantum algorithm. The newly generated capacity configuration parameters (slime mold layer offspring) from the lower slime mold algorithm. Merging to form a joint group The expression is: ; Each individual in the joint group is an optimized scheme containing a complete set of MPPT control parameters and system capacity configuration parameters.

[0066] A non-dominated ranking is performed on the joint population, dividing each individual in the joint population (a complete set of wind power to hydrogen production system optimization schemes) into different levels based on the Pareto solution dominance relationship. .

[0067] The dominance relationship is: if the individual Their performance on the objective function is no worse than that of the individual. And is strictly superior to at least one objective. Then it is called Dominate The solution set that is not controlled by any individual is classified as the first level. The non-dominated solution set remaining after removing the first level is divided into the second level. And so on.

[0068] The crowding distance is calculated for individuals within the same level, using the following expression: ; In the formula, CD e Represented as the first The crowding distance of each individual; Indicates the index of the objective function; Indicates the number of objective functions; and They represent the first time. Under one objective function, and The preceding and following individuals of an individual (in order of the first individual) The objective function values ​​in the sorted sequence of objective function values; and These represent the individual at the current level. The maximum and minimum values ​​of each objective function.

[0069] The uniformity of solution distribution is evaluated by crowding distance. Individuals with sparse distribution and large spacing are retained first, while individuals with dense distribution and high redundancy are removed, so as to ensure the diversity and uniformity of Pareto front solution set.

[0070] Based on the non-dominated ordination results and the crowding distance selection results, Pareto front solutions (e.g., 40%–50% of the joint population) are selected from the joint population to form a new generation of population. Update and maintain the external archive, store the Pareto front solution set in the external archive for population update in the next iteration; return to S2 for iteration; after the iteration is completed, output the current Pareto front solution set as the Pareto optimal solution set.

[0071] During algorithm iteration, the external archive will be continuously updated, and the solution set stored therein will serve as a guiding basis for the design of the next generation of quantum rotating gates. Simultaneously, an independent global archive maintained during optimization will be dynamically updated to preserve excellent solutions obtained in previous iterations. Furthermore, solutions with high density will be removed, thereby effectively maintaining the diversity and quality of the archive.

[0072] After several iterations, determine the improvement of the current Pareto front solution set. With set threshold The relationship between them; if If the condition is met, return to S2 and perform optimization operations by updating the part through the quantum rotation gate; otherwise, terminate the iteration and output the current Pareto front solution set as the Pareto optimal solution set.

[0073] S4. Based on the Pareto optimal solution set, generate matching MPPT control parameters and system capacity configuration scheme.

[0074] This application enables efficient coordination between the system control layer and the operation layer; each solution in the Pareto optimal solution set corresponds to a complete scheme, including a specific set of upper quantum layer MPPT control parameters and a set of lower slime mold layer system capacity configuration parameters.

[0075] The method of this invention was applied to regions A and B. The first five Pareto optimal solution sets for the multi-objective optimization in regions A and B are shown in Tables 1 and 2. Optimization results show that, under different cases (Case 1-5), the number of wind turbines... Number of electrolytic cells connected in series and parallel and Number of hydrogen storage tanks and energy storage capacity Significant differences exist, reflecting capacity optimization strategies tailored to local wind resources and load demands. The optimal solution sets for both regions exhibit a typical Pareto competition relationship among the three core indicators: Levelized Cost of Hydrogen (LCOH), Levelized Cost of Electricity (LCOE), and Probability of Power Shortage (LPSP), fully validating the effectiveness of the two-layer collaborative algorithm of this invention in solving multi-objective trade-offs. Taking region A as an example, Case 5 corresponds to the lowest LCOH, at only $2.9245 / kg, with an LCOE as low as $0.02491 / kWh and an LPSP of 2.942%, demonstrating the best overall economic efficiency; while Case 3's energy storage capacity... The largest wind turbine / electrolyzer configuration corresponds to a larger wind turbine / electrolyzer configuration, demonstrating strong resource absorption capacity. In region B, the overall LCOH is higher (lowest at $3.5817 / kg), while the LPSP decreases from 4.358% to 2.816%, reflecting the significant impact of differences in wind resource endowment between the two regions on the economics and reliability of hydrogen production. The above multi-case Pareto solution set solutions demonstrate that the MPPT-capacity matching bilayer optimization framework proposed in this invention can stably output high-quality non-dominated solution fronts under different geographical and meteorological conditions, exhibiting strong regional adaptability and engineering application value.

[0076] Table 1. The top 5 Pareto optimal solutions in multi-objective optimization in region A. ;

[0077] Table 2. The top 5 Pareto optimal solutions in the multi-objective optimization of region B. ; Please refer to Tables 3 and 4 for a deep quantitative assessment of the emission reduction potential of the optimized scheme of this invention from the perspectives of environmental benefits and the economics of carbon trading. By comparing the replacement of traditional oil-fired power plants with natural gas power plants, the optimized wind power to hydrogen production system demonstrates significant CO2 emission reductions and corresponding carbon credit gains. Taking alternatives to oil-fired power plants as an example, Scheme 3 in Region A can achieve emission reductions of up to 151.3736 tons, equivalent to a carbon trading credit of $6054.94. Compared to natural gas power plants, its emission reductions still reach 109.295 tons, with a carbon revenue of $4371.8001, significantly better than the other schemes. This indicates that when the capacity configuration and operating parameters are well matched, the system not only has strong economic competitiveness but can also significantly enhance environmental benefits. In Region B, the overall emission reduction is lower than in Region A, but there are still significant differences between different cases. Cases 1 and 2 can also achieve emission reductions of 62.564 tons and economic benefits of $2502.56, while Case 4 reduces to 25.4785 tons and $1019.1384. The results demonstrate that the multi-objective optimization not only takes into account economic efficiency and energy efficiency, but also quantitatively identifies environmental indicators under different configuration schemes, providing a quantitative basis for the feasibility of wind power hydrogen production systems in different regions, highlighting the comprehensiveness and scientific rationality of this invention in multi-dimensional and multi-objective collaborative design.

[0078] Table 3 CO2 emission reductions and carbon credits for Region A ;

[0079] Table 4 CO2 emission reductions and carbon credits for Region B ;

[0080] Please see Figure 2 and Figure 3 The paper provides an in-depth analysis of the thermodynamic energy efficiency (energy / efficiency) and actual output capacity of wind power hydrogen production systems under different configuration cases. Figure 2 The results show that, thanks to the precise MPPT dynamic tracking control of the upper quantum layer and the perfect matching of the lower layer's capacity, the overall energy efficiency of all core components and the system remains at a high level. For example, in Case 3 of Region A, the overall energy efficiency of the system exceeds 10%, and the wind efficiency approaches 8%, demonstrating the stability of the dual-layer optimization strategy in terms of energy quality utilization. In Region B, the energy efficiency and wind efficiency fluctuation ranges of the systems in each case are similar, but due to differences in wind resource characteristics, the overall efficiency is slightly lower than that of Region A. Figure 3 The power generation and hydrogen production further validated the above conclusions. Case 3 in region A achieved the highest power generation (over 500 MWh) while ensuring a hydrogen production of up to 1416.15 t. Case 1 in region B had the highest hydrogen production (488.162 t), corresponding to a power generation of approximately 215 MWh. The hydrogen production efficiency per unit of electricity was better than that in region A, demonstrating the positive support of the wind energy resource structure in this region for the operating efficiency of the hydrogen production system.

[0081] In summary, this invention achieves full-chain optimization of "maximizing wind energy capture - high-efficiency conversion - rational capacity configuration" through a two-layer collaborative evolution mechanism, ensuring that the wind power hydrogen production system has excellent dynamic response and optimal comprehensive output benefits under different complex operating conditions, and has strong industrial application value and engineering reliability.

[0082] Therefore, this invention adopts the above-mentioned MPPT and capacity matching design optimization method for wind power hydrogen production systems, which breaks through the limitations of using only static or ideal power curves in traditional capacity configuration. By deeply coupling dynamic MPPT control with static capacity planning through a two-layer architecture, the realism and reliability of system design are improved. The full-space search capability of upper-layer quantum evolution is used to avoid MPPT from getting trapped in local optima. Combined with the fine search advantage of the lower-layer slime mold algorithm in continuous space, the convergence and distribution uniformity of multi-objective Pareto fronts are significantly improved, thereby improving the overall energy efficiency of the system.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for optimizing MPPT and capacity matching design in a wind power hydrogen production system, characterized in that, Includes the following steps: S1. Collect meteorological data and operating parameters of each component of the wind power hydrogen production system in the target area to establish mathematical models of each component of the wind power hydrogen production system, and establish the optimization target system and operating boundary conditions of the wind power hydrogen production system. S2. Based on the mathematical models of each component of the wind power hydrogen production system, the optimization target system and the operating boundary conditions, a two-layer collaborative algorithm consisting of an upper quantum layer and a lower slime mold layer is adopted to optimize the MPPT control parameters and system capacity configuration parameters respectively, and to perform two-layer collaborative feedback updates. S3. Merge the parent generation's non-dominated optimal solution, the quantum layer's offspring MPPT control parameters, and the slime mold layer's offspring capacity configuration parameters to form a joint population. Filter the Pareto front solution set through non-dominated sorting and crowding distance calculation, and return to S2 for iteration. After the iteration is complete, output the current Pareto front solution set as the Pareto optimal solution set. S4. Based on the Pareto optimal solution set, generate matching MPPT control parameters and system capacity configuration scheme.

2. The method according to claim 1, characterized in that, In S2, the optimized MPPT control parameters of the upper quantum layer include: U1. Constructing a quantum population using qubits; the quantum population includes multiple quantum individuals, each corresponding to a set of MPPT control parameters; Among them, the The encoding matrix of each quantum individual for The complex matrix is ​​expressed as: ； ； In the formula, Indicates the first Among the individuals, the first The probability magnitude of a quantum bit collapsing to a state of 0; Indicates the first Among the individuals, the first The probability amplitude of a qubit collapsing to a state of 1; , The length of the quantum bit encoding is equal to the total number of MPPT control parameters; where the collapse states of 0 and 1 correspond to the lower and upper limits of the MPPT control parameters, respectively. U2. Based on the current quantum probability amplitude, a set of determined MPPT control parameters are generated through observation. The specific observation mechanism is as follows: for the first... The first individual One qubit, producing one Uniformly distributed random numbers within the interval ,like If the state of the qubit is 1, then the qubit collapses to 1; otherwise, the qubit collapses to 0, resulting in a binary sequence. The binary sequence is then decoded into MPPT control parameters using a linear mapping to obtain the corresponding MPPT control parameters. U3. Dynamically update the probability amplitude of the qubit through the quantum rotation gate to optimize the generation of MPPT control parameters; U4. The optimized MPPT control parameters are converted into a fan power curve that takes into account the actual operating characteristics of the fan, and then output to the lower slime mold layer.

3. The method according to claim 2, characterized in that, The expression for the probability magnitude of updating the qubit is: ； ； In the formula, and These represent the values ​​after the quantum rotation gate update. Among the individuals, the first The new probability magnitudes of each qubit collapsing into 0 and 1 states; Indicates the first Among the individuals, the first The rotation angle of each qubit; Indicates the step size of the rotation angle; This represents the direction determination function, ensuring that the probability amplitude converges towards the optimal solution region.

4. The method according to claim 3, characterized in that, In S2, the lower slime mold layer optimizes the system capacity configuration parameters, including: D1. Obtain the wind turbine power curve output from the upper quantum layer, perform time-series matching with the system load demand, and calculate the energy flow and balance state of the wind power hydrogen production system; construct the objective function and constraint function for optimizing the capacity configuration of the wind power hydrogen production system. D2. Initialize the slime mold population location vectors, with each location vector corresponding to a set of capacity configuration parameters; D3. Based on the energy flow and balance state of the wind power hydrogen production system, and based on the objective function and constraint function, calculate the fitness value of each individual in the slime mold population, and sort the fitness values ​​of all individuals in ascending order to obtain the fitness value sequence. D4. Based on the step fitness ranking results, calculate the adaptive weight of each slime mold individual; D5. Simulate the foraging behavior of slime molds. Based on adaptive weights and combined with the slime mold position update formula, update the slime mold population position within the capacity configuration target space, optimize the capacity configuration parameters round by round, and feed them back to the upper quantum layer.

5. The method according to claim 3, characterized in that, The adaptive weight of slime mold individuals is calculated using the following formula: ； In the formula, This represents the adaptive weight of the i-th slime mold individual after sorting. express Random numbers within the interval; This represents the fitness value of the i-th slime mold individual in the slime mold population; Indicates the first The fitness value of the best solution in the next iteration of the population; Indicates the first The fitness value of the worst solution in the next iteration of the population.

6. The method according to claim 5, characterized in that, The two-layer collaborative feedback update is as follows: the upper quantum layer transmits the MPPT control parameters to the lower layer to correct the direction of slime mold population capacity optimization; the lower slime mold layer feeds back the capacity configuration parameters to the upper layer to correct the direction of quantum population rotation angle update, thereby realizing bidirectional collaborative iteration of control and capacity.

7. The method according to claim 6, characterized in that, The two-layer collaborative feedback update also includes: if the MPPT control parameters output by the upper quantum layer make the capacity matching cost of the lower slime mold layer lower than the cost threshold and the efficiency higher than the efficiency threshold, then the positive rotation angle is increased to enhance the probability amplitude of the MPPT control parameters in the upper quantum layer population; otherwise, it indicates that the capacity configuration is redundant and the probability amplitude of the MPPT control parameters in the upper quantum layer population is suppressed.

8. The method according to claim 7, characterized in that, The dynamic correction for rotation angle compensation is expressed as: ； In the formula, Indicates the corrected rotation angle; Indicates the rotation angle before correction; Indicates the cooperative gain coefficient; This indicates the hypervolume increment of the Pareto front in the lower slime mold layer.

9. The method according to any one of claims 1 to 8, characterized in that, S3 specifically includes: S31. The parent generation non-dominated optimal solution, the quantum layer offspring MPPT control parameters, and the slime mold layer offspring capacity configuration parameters are merged to form a joint population; each individual in the joint population is an optimization scheme containing a complete set of MPPT control parameters and system capacity configuration parameters; S32. Perform non-dominant ranking on the joint group, dividing each individual in the joint group into different levels based on Pareto undominant relations. ; The crowding distance is calculated for individuals within the same level, using the following expression: ； In the formula, CD e Represented as the first The crowding distance of each individual; Indicates the index of the objective function; Indicates the number of objective functions; and They represent the first time. Under one objective function, and The objective function values ​​of the preceding and following individuals of an individual; and These represent the individual at the current level. The maximum and minimum values ​​of the objective function; S33. Based on the non-dominated sorting results and the crowding distance screening results, select the Pareto front solution set in the joint population to form a new generation of population for population update in the next iteration; return to S2 for iteration; after the iteration ends, output the current Pareto front solution set as the Pareto optimal solution set.

10. The method according to claim 9, characterized in that, In S3: After several iterations, determine the improvement of the current Pareto front solution set. With set threshold The relationship between them; if Then return to S2 and perform optimization operations through the quantum rotation gate update part; Otherwise, terminate the iteration and output the current Pareto front solution set as the Pareto optimal solution set.

Images