Acceleration algorithm and device for photovoltaic hydrogen production capacity configuration optimization system

Abstract

The invention discloses an acceleration algorithm and device for a photovoltaic hydrogen production capacity configuration optimization system, and the method comprises the following steps: building a double-layer optimization model of a photovoltaic hydrogen production system, and enabling an upper-layer model of the double-layer optimization model to take the minimum unit hydrogen production cost of a full life cycle as an upper-layer objective function, the lower-layer model takes the minimum electricity abandoning rate as a lower-layer objective function; the photovoltaic hydrogen production system at least comprises a photovoltaic power generation system, a proton exchange membrane PEM electrolytic hydrogen production system, an alkaline electrolytic hydrogen production system, an energy storage battery and a hydrogen storage tank; solving the double-layer optimization model to obtain the optimal capacity configuration of the photovoltaic hydrogen production system; wherein the upper-layer model is solved by adopting a differential particle swarm hybrid optimization acceleration algorithm, and the lower-layer model is solved by adopting a linear programming problem solving method. Through a series of acceleration algorithms on the basis of the original algorithm, the algorithm operation time of the photovoltaic hydrogen production capacity configuration optimization system can be effectively shortened, and the capacity configuration optimization efficiency is improved.

Description

Technical Field

[0001] The present application relates to the technical field of photovoltaic hydrogen production, and in particular to an acceleration algorithm and device for a photovoltaic hydrogen production capacity configuration optimization system. Background Art

[0002] As global energy demand grows and environmental pressure increases, the promotion of clean energy becomes increasingly important. Photovoltaic hydrogen production technology, which combines photovoltaic power generation technology with hydrogen production technology, can provide clean energy with high energy density. However, due to the instability of solar and wind energy, photovoltaic power generation is intermittent and volatile. In order to ensure the economy and reliability of photovoltaic hydrogen production, it is necessary to optimize the system capacity configuration.

[0003] However, traditional photovoltaic hydrogen production optimization methods (such as linear programming, nonlinear programming and dynamic programming) are difficult to meet actual needs due to high computational complexity and long solution time. Most of the existing acceleration algorithms in the field are also small-scale optimizations based on the original algorithms, such as using crossover, mutation, prediction and other methods, which cannot produce acceleration algorithms that are more suitable for the current scenario. Summary of the invention

[0004] In view of this, an embodiment of the present application provides an acceleration algorithm and device for a photovoltaic hydrogen production capacity configuration optimization system, which improves the efficiency of capacity configuration optimization by optimizing the algorithm during the capacity configuration optimization process.

[0005] The technical solutions provided by the embodiments of this application are as follows:

[0006] A photovoltaic hydrogen production capacity configuration optimization method, comprising:

[0007] A two-layer optimization model for a photovoltaic hydrogen production system is established, wherein the upper model of the two-layer optimization model takes the minimum unit hydrogen production cost over the entire life cycle as the upper objective function, and the lower model takes the minimum power abandonment rate as the lower objective function; the photovoltaic hydrogen production system at least includes a photovoltaic power generation system, a proton exchange membrane PEM electrolysis hydrogen production system, an alkaline electrolysis hydrogen production system, an energy storage battery and a hydrogen storage tank;

[0008] Solving the two-layer optimization model to obtain the optimal capacity configuration of the photovoltaic hydrogen production system; wherein the upper layer model is solved by a differential particle swarm hybrid optimization algorithm, and the lower layer model is solved by a linear programming problem solving method; using a differential particle swarm hybrid optimization algorithm to solve, including using one of the following three acceleration algorithms to solve:

[0009] Acceleration algorithm 1: An improved differential particle swarm hybrid optimization algorithm that reduces iterative graph output, pauses briefly to update the graph, uses vectorized operations, and simplifies position and velocity constraint processing;

[0010] Acceleration algorithm 2: Based on acceleration algorithm 1, it combines GPU parallel optimization algorithm;

[0011] Acceleration algorithm three: An optimization algorithm that combines parallel computing based on acceleration algorithm one.

[0012] An acceleration device for a photovoltaic hydrogen production capacity configuration optimization system, comprising:

[0013] A model building module is used to establish a double-layer optimization model for a photovoltaic hydrogen production system, wherein the upper model of the double-layer optimization model takes the minimum unit hydrogen production cost over the entire life cycle as the upper layer objective function, and the lower model takes the minimum power abandonment rate as the lower layer objective function; the photovoltaic hydrogen production system at least includes a photovoltaic power generation system, a proton exchange membrane PEM electrolysis hydrogen production system, an alkaline electrolysis hydrogen production system, an energy storage battery and a hydrogen storage tank;

[0014] A model solving module is used to solve the double-layer optimization model to obtain the optimal capacity configuration of the photovoltaic hydrogen production system; wherein the upper layer model is solved by using a differential particle swarm hybrid optimization algorithm, and the lower layer model is solved by using a linear programming problem solving method; the differential particle swarm hybrid optimization algorithm is used to solve, including using one of the following three acceleration algorithms to solve:

[0015] Acceleration algorithm 1: An improved differential particle swarm hybrid optimization algorithm that reduces iterative graph output, pauses briefly to update the graph, uses vectorized operations, and simplifies position and velocity constraint processing;

[0016] Acceleration algorithm 2: Based on acceleration algorithm 1, it combines GPU parallel optimization algorithm;

[0017] Acceleration algorithm three: an optimization algorithm that combines parallel computing based on acceleration algorithm one.

[0018] It can be seen from the above scheme that this application establishes a two-layer optimization model for the photovoltaic hydrogen production system, designs the objective functions of the upper model and the lower model in the two-layer optimization model, and respectively uses the differential particle swarm hybrid optimization acceleration algorithm and the linear programming problem solving method to solve the upper model and the lower model, and finally obtains the optimal capacity configuration of the photovoltaic hydrogen production system. It can be seen that this application performs algorithm optimization in the process of capacity configuration optimization, effectively reducing the algorithm running time, thereby improving the efficiency of capacity configuration optimization; and by adopting an improved differential particle swarm hybrid optimization algorithm that reduces iterative graph output, briefly pauses to update graphics, vectorized forms, and simplifies position and speed restriction processing, and combines parallel computing, it is not simply limited to the small point optimization of the heuristic algorithm, making the optimization results more extensive and representative. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1This is a flow chart of a photovoltaic hydrogen production capacity configuration optimization method provided in an embodiment of the present application;

[0020] Figure 2 It is a solution flow chart of a double-layer optimization model of a photovoltaic hydrogen production system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical scheme and advantages of the present application clearer, the specific embodiments of the present application are described in detail below in conjunction with the accompanying drawings. Examples of these preferred embodiments are illustrated in the accompanying drawings. The embodiments of the present application shown in the accompanying drawings and described according to the accompanying drawings are merely exemplary, and the present application is not limited to these embodiments.

[0022] It should also be noted here that in order to avoid obscuring the focus of the present application due to unnecessary details, the accompanying drawings mainly show structures and / or processing steps that are closely related to the scheme according to the present application, while omitting some other details that are not very relevant to the present application.

[0023] The technical solutions in the embodiments of the present application are described in detail below.

[0024] The present application embodiment provides an acceleration algorithm for a photovoltaic hydrogen production capacity configuration optimization system, such as Figure 1 As shown, the acceleration algorithm includes the following steps:

[0025] Step 101, establish a two-layer optimization model for a photovoltaic hydrogen production system, wherein the upper layer model of the two-layer optimization model takes minimizing the unit hydrogen production cost over the entire life cycle as the upper layer objective function, and the lower layer model takes minimizing the power abandonment rate as the lower layer objective function.

[0026] In one embodiment, the photovoltaic hydrogen production system includes at least a photovoltaic power generation system, a proton exchange membrane (PEM) electrolysis hydrogen production system, an alkaline electrolysis hydrogen production system, an energy storage battery and a hydrogen storage tank. Among them, the photovoltaic power generation system may include a photovoltaic array, the PEM electrolysis hydrogen production system may include a PEM electrolyzer, and the alkaline electrolysis hydrogen production system may include an alkaline electrolyzer.

[0027] In one embodiment, the photovoltaic output data of the photovoltaic hydrogen production system within a preset time can be clustered, and a photovoltaic output scenario can be obtained according to each type of photovoltaic output data, so as to perform corresponding system optimization for different photovoltaic output scenarios.

[0028] Photovoltaic output data generally refers to the electric energy data generated by a photovoltaic power generation system within a certain period of time, including but not limited to irradiance, ambient temperature, wind speed, humidity, cloud cover, sunshine time, theoretical power and actual power. In one example, the photovoltaic output data can be collected by a preset collection device, and then the collected data can be summarized and processed for clustering.

[0029] In one example, the output power of the photovoltaic hydrogen production system can be calculated based on the temperature and solar irradiance, and the output power can be further used as photovoltaic output data. Specifically, the output power of the photovoltaic hydrogen production system can be calculated by formula (1):

[0030]

[0031] Among them, P PV Represents the output power of the photovoltaic power generation system, in kW; P PV,rated represents the rated power of the photovoltaic power generation system, in kW; f represents the power generation efficiency of the photovoltaic power generation system, which is selected according to the actual situation. In this embodiment, it is 0.9; δ represents the temperature coefficient, which is generally -0.47% / K; G and Gs represent the actual solar irradiance and the reference irradiance, in W / m 2 ; T and Ts represent the actual working temperature and reference temperature, the unit is K.

[0032] In one example, data clustering can be performed using a clustering algorithm available in the art, such as a K-means algorithm.

[0033] In one example, clustering can be performed based on the photovoltaic output data throughout the year, and the number of days each photovoltaic scene appears can be obtained. For example, if the K-means algorithm is used to cluster the photovoltaic output data, six clustering results corresponding to six photovoltaic output scenes can be obtained, and the number of days each photovoltaic scene appears can be further counted. In a specific example, the number of days that the six photovoltaic scenes appear throughout the year are: 127 days, 86 days, 32 days, 32 days, 66 days, and 22 days.

[0034] Before establishing the double-layer optimization model of the photovoltaic hydrogen production system, in step 101, in one embodiment, a hydrogen production model of the photovoltaic hydrogen production system is first established based on the physical structure of the photovoltaic hydrogen production system and combined with the efficiency characteristics. Specifically, the hydrogen production rate formula of the photovoltaic hydrogen production system is as follows:

[0035]

[0036]

[0037]

[0038] in, Represents the unit power consumption of hydrogen, the unit is kWh / Nm 3 ; η el,ALK , η el,PEM , η el,Eb Respectively represent the electrolysis efficiency of alkaline electrolysis hydrogen production system, PEM electrolysis hydrogen production system and energy storage battery; P ALK , P PEM , P Eb They represent the input power of the alkaline electrolysis hydrogen production system, the PEM electrolysis hydrogen production system and the energy storage battery, respectively, in kW; They represent the hydrogen production rates of the alkaline electrolysis hydrogen production system, PEM electrolysis hydrogen production system and energy storage battery respectively.

[0039] Regarding the two-layer optimization model established in step 101, in one embodiment, the upper layer objective function formula is as follows:

[0040]

[0041] Among them, C ALK represents the total life cycle cost of the alkaline electrolysis hydrogen production system, in yuan; C PEM represents the total life cycle cost of the PEM electrolysis hydrogen production system, in yuan; C Eb Represents the total life cycle cost of the energy storage battery, in yuan; C Eh Represents the total life cycle cost of the hydrogen storage tank, in yuan; C LOSE represents the penalty cost for power abandonment, in yuan; Q H2 Represents the total hydrogen production of the system over its entire life cycle, in Nm 3 .

[0042] In one embodiment, the formula of the lower layer objective function in step 102 is as follows:

[0043]

[0044] Among them, P ALK (t) represents the input power of the alkaline electrolysis hydrogen production system at time t, in kW; P PEM (t) represents the input power of the PEM electrolysis hydrogen production system at time t, in kW; P PV (t) represents the output power of the photovoltaic power generation system at time t, in kW.

[0045] It should be noted that the specific upper-level objective function and lower-level objective function formulas given above are only examples. For those skilled in the art, several changes can be made without departing from the principles of this application, and all of them should be regarded as within the scope of protection of this application.

[0046] Step 102, solving the double-layer optimization model to obtain the optimal capacity configuration of the photovoltaic hydrogen production system; wherein the upper layer model is solved by a differential particle swarm hybrid optimization algorithm, and the lower layer model is solved by a linear programming problem solving method.

[0047] In one embodiment, when solving the double-layer optimization model in step 102, the upper layer model is solved using a differential particle swarm hybrid optimization algorithm, where the differential particle swarm hybrid optimization algorithm is a hybrid algorithm of a differential evolution algorithm and a particle swarm algorithm.

[0048] In some embodiments, when using the differential particle swarm hybrid optimization algorithm to solve, the following three acceleration algorithms can be used to increase the computing speed:

[0049] Acceleration algorithm 1: An improved differential particle swarm hybrid optimization algorithm that reduces iterative graph output, pauses briefly to update the graph, uses vectorized operations, and simplifies position and speed limit processing. This optimization algorithm can reduce the number of loops and reduce the time consumption caused by drawing operations.

[0050] The above-mentioned acceleration algorithm 1 includes four designs: "reducing the output of the iteration graph", "pausing briefly to update the graph", "using vectorized operations" and "simplifying the position and speed restriction processing". Each part can reduce the computing time consumption and improve the computing speed. Specifically, the following methods can be used to reduce the running time: adding a command to output the iteration graph once every multiple iterations (such as five times) to the algorithm to reduce the output of the iteration graph; replacing the original stop to update the graph with a short pause can effectively reduce the time consumption; using vectorized operations to effectively reduce the number of loops; using max and min functions to process the speed and position at one time to simplify the position and speed restriction processing.

[0051] Acceleration algorithm 2: Based on acceleration algorithm 1, a GPU parallel optimization algorithm is combined. This optimization algorithm can minimize the time consumption of the iterative task process.

[0052] Specifically, acceleration algorithm 2 can package all data and transmit it to the GPU in the initial stage to perform population updates, fitness calculations, etc. It tries to start the kernel only once to process all iterative tasks. Before and after the algorithm is parallelized, only one data memory copy is required. When drawing the fitness change graph, it is moved from the GPU to the CPU, which minimizes the time consumption of the process.

[0053] Acceleration algorithm 3: An optimization algorithm that combines parallel computing with acceleration algorithm 1. This optimization algorithm can further speed up the calculation.

[0054] Specifically, acceleration algorithm three can divide the computing task into multiple parts and execute them on multiple processors or computing cores at the same time to speed up the computing speed. At the same time, after clearing the environment, it is necessary to check whether the parallel computing pool has been turned on, and then use parallelism to calculate the fitness value.

[0055] In one embodiment, the above three acceleration algorithms can be compared in advance, and finally the acceleration algorithm with the shortest running time, smaller fluctuation, faster and more stable convergence speed is selected as the optimal algorithm for capacity configuration of the photovoltaic hydrogen production system.

[0056] In a specific example, after comparing the running time, the acceleration algorithm 3 was finally selected for capacity configuration. By adopting the improved differential particle swarm hybrid optimization algorithm that reduces the iterative graph output, pauses briefly to update the graph, vectorizes the form, simplifies the position and speed limit processing, and combines parallel computing, it is not limited to the small point optimization of the heuristic algorithm, making the optimization results more extensive and representative.

[0057] In one embodiment, when solving the two-layer optimization model in step 102, a linear programming problem solving method can be used for solving. Specifically, a mathematical optimization solver capable of solving linear programming problems can be used for solving. In one example, the lower layer model belongs to mixed integer linear programming, and can be solved by an existing mixed integer linear programming optimization solution method. For example, the CPLEX solver can be called by the YALMIP toolkit of MATLAB for optimization and solution.

[0058] In one embodiment, the upper-level constraint condition of the upper-level model includes a device capacity constraint, and the device capacity constraint formula is as follows:

[0059] P ALK,rated +P PEM,rated +P Eb,rated ≤P PV,rated (7)

[0060] P ALK,rated ≥0(8)

[0061] P PEM,rated ≥0(9)

[0062] P Eb,rated ≥0(10)

[0063] Among them, P ALK,rated Represents the capacity of the alkaline electrolysis hydrogen production system, in kW; P PEM,rated Represents the capacity of the PEM electrolysis hydrogen production system, in kW; P Eb,rated Represents the capacity of the energy storage battery, in kW.

[0064] In one embodiment, the lower-level constraint of the lower-level model includes a system power balance constraint, and the system power balance constraint formula is as follows:

[0065] P ALK +P PEM +P l =P PV (11)

[0066] P ALK ≥0(12)

[0067] P PEM ≥0(13)

[0068] Among them, P l Represents the abandoned power.

[0069] In one embodiment, the lower-level constraint of the lower-level model includes an operating load range constraint, and the operating load range constraint formula is as follows:

[0070]

[0071]

[0072]

[0073]

[0074] in, Represents the lowest operable load of the alkaline electrolysis hydrogen production system; Represents the maximum operable load of the alkaline electrolysis hydrogen production system; Represents the minimum operating load of the PEM electrolysis hydrogen production system; Represents the maximum operational load of the PEM electrolysis hydrogen production system; Represents the lowest operational load of the energy storage battery; Represents the maximum operational load of the energy storage battery; Represents the lowest operable load of the hydrogen storage tank; Represents the maximum operational load of the hydrogen storage tank; P Eh Represents the input power of the hydrogen storage tank, in kW; P Eh,rated Represents the capacity of the hydrogen storage tank in kW.

[0075] In one embodiment, the lower-level constraint of the lower-level model includes an electrolytic cell climbing constraint, and the electrolytic cell climbing constraint formula is as follows:

[0076] -d ALK ×P ALK,rated ≤P ALK (t)-P ALK (t-1)≤uALK ×P ALK,rated (18)

[0077] -d PEM ×P PEM,rated ≤P PEM (t)-P PEM (t-1)≤u PEM ×P PEM,rated (19)

[0078] Among them, d ALK Represents the maximum load reduction rate allowed for the alkaline electrolysis hydrogen production system; u ALK represents the maximum loading rate allowed by the alkaline electrolysis hydrogen production system; d PEM Represents the maximum load reduction rate allowed by the PEM electrolysis hydrogen production system; u PEM Represents the maximum loading rate allowed by the PEM electrolysis hydrogen production system.

[0079] In one embodiment, the lower-level constraint of the lower-level model includes an electrolyzer hydrogen production efficiency constraint, and the electrolyzer hydrogen production efficiency constraint formula is as follows:

[0080]

[0081]

[0082] In one embodiment, the lower-level constraints of the lower-level model include energy storage battery and hydrogen storage tank constraints, and the energy storage battery and hydrogen storage tank constraint formulas are as follows:

[0083]

[0084]

[0085] P Eb ≥0(24)

[0086] P Eh ≥0(25)

[0087] Among them, d Eb Represents the maximum load reduction rate allowed by the energy storage battery; u Eb Represents the maximum loading rate allowed for the energy storage battery; d Eh Represents the maximum load reduction rate allowed for the hydrogen storage tank; u Eh Represents the maximum loading rate allowed for the hydrogen storage tank; P Eb,ch Represents the charging power of the energy storage battery, in kW; P Eb,dis Represents the discharge power of the energy storage battery, in kW; P Eh,ch Represents the charging power of the hydrogen storage tank, in kW; P Eh,dis Represents the discharge power of the hydrogen storage tank, in kW.

[0088] In one embodiment, the lower-level constraints of the lower-level model include at least one of the system power balance constraints, operating load range constraints, electrolyzer ramp constraints, electrolyzer hydrogen production efficiency constraints, energy storage batteries, and hydrogen storage tank constraints described above.

[0089] It should be noted that the specific constraint formulas of the upper model and the lower model given above are only examples. For those skilled in the art, several changes can be made without departing from the principles of the present application, and all of them should be regarded as within the scope of protection of the present application.

[0090] So far, the Figure 1 Description of the process shown. Figure 1 The process shown can be used to optimize the algorithm during the capacity configuration optimization process, effectively reduce the algorithm running time, and improve the efficiency of capacity configuration optimization. In addition, in some embodiments, by adopting a differential particle swarm hybrid optimization algorithm that reduces iterative graph output, briefly pauses to update the graph, vectorizes the form, simplifies the position and speed limit processing, and combines parallel computing, it is not simply limited to the small point optimization of the heuristic algorithm, so that the optimization results are more extensive and representative.

[0091] Combine the following Figure 2 A detailed explanation is given on how to solve the two-layer optimization model of the photovoltaic hydrogen production system to obtain the optimal capacity configuration of the photovoltaic hydrogen production system.

[0092] Figure 2 An embodiment of the present application provides a flow chart for solving a double-layer optimization model for a photovoltaic hydrogen production system.

[0093] like Figure 2 As shown, the process includes the following steps:

[0094] Step 201, randomly initialize the position and velocity of each particle in the population.

[0095] Specifically, this step 201 includes setting the equipment capacity of the photovoltaic hydrogen production system under the current calculation, and transferring the equipment capacity parameters to the lower model.

[0096] Step 202, setting parameters for the lower layer model.

[0097] Specifically, the parameter settings here at least include the settings of equipment capacity parameters and photovoltaic typical daily output data. The photovoltaic typical daily output data can be determined by referring to the photovoltaic output scenario.

[0098] Step 203, setting lower-level constraints for the lower-level model.

[0099] The lower-level model here has been pre-built, and the objective function of the lower-level model is to minimize the power abandonment rate.

[0100] The lower-level constraints may be at least one of the system power balance constraints, operating load range constraints, electrolyzer ramp constraints, electrolyzer hydrogen production efficiency constraints, energy storage battery and hydrogen storage tank constraints exemplified above.

[0101] Step 204, calling CPLEX to optimize and solve the lower model, obtain the optimal operating power of the alkaline electrolysis hydrogen production system and the PEM electrolysis hydrogen production system at each moment, and return the optimal operating power to the upper model.

[0102] Step 205: The upper model takes the minimum unit hydrogen production cost over the entire life cycle as the upper objective function, calculates the individual fitness values ​​of the population particles, and updates the population particles.

[0103] Step 206 , when the maximum number of iterations is reached, the iteration is stopped and the optimal capacity configuration of the photovoltaic hydrogen production system is output; otherwise, the process returns to step 202 .

[0104] For ease of understanding Figure 2 The process shown is briefly described below for the differential particle swarm hybrid optimization algorithm used in this application. The differential particle swarm hybrid optimization algorithm (DPSO) is based on the particle swarm algorithm and introduces the mutation and crossover operations in the differential evolution algorithm to update the speed and position of the particles. Initialize it as a group of random particles, and then find the optimal solution through iteration. In each iteration, the particles update themselves by tracking the individual optimal point of the particles themselves and the global optimal point of the population. Assume that in a D-dimensional target search space, there are N particles forming a community, where the position of the i-th particle is represented as a D-dimensional vector:

[0105] X i =(x i1 ,x i2 ,…,x iD ),i=1,2,…,N

[0106] The velocity of the i-th particle is also a D-dimensional vector:

[0107] V i =(v i1 ,v i2 ,…,v iD ),i=1,2,…,N

[0108] The optimal position searched by the i-th particle so far is called the individual extreme value:

[0109] P best =(p i1 ,p i2 ,…,p iD ),i=1,2,…,N

[0110] The optimal position searched by the entire population so far is the global extreme value:

[0111] g best =(p g1 ,p g2 ,...,p gD )

[0112] For each particle, its d-dimensional velocity and position values ​​are updated according to the following formula:

[0113]

[0114]

[0115] Among them, ω represents the inertia weight; c1 and c2 represent learning factors; r1 and r2 represent random uniform numbers in the range of [0,1].

[0116] In one embodiment, the inertia weight ω is expressed as follows:

[0117]

[0118] ω=ω max , f>f avg (29)

[0119] Among them, ω min ,ω max represents the minimum and maximum values ​​of the inertia weight; f represents the real-time fitness value of the particle; f avg , f min Represents the average fitness value and minimum fitness value of all current particles.

[0120] In one embodiment, the particle position update formula is:

[0121] u ij =x r1,j +F(x r2,j -x r3,j ),rand<C R (30)

[0122] u ij =x r1,j ,rand>C R (31)

[0123] Among them, x r1,j , x r2,j , x r3,j represents three random individuals in the current population; F represents the scaling factor; C R represents the crossover probability.

[0124] In some embodiments, when selecting an optimization acceleration algorithm, three different optimization acceleration algorithms can be used, namely: using an improved differential particle swarm hybrid optimization algorithm that reduces iterative graph output, pauses briefly to update the graph, uses vectorized operations, and simplifies the position and speed restriction processing for iterative tour solving, thereby reducing the number of cycles and reducing the time consumption caused by drawing operations; using an optimization algorithm that combines GPU parallelism on the basis of the first optimization to minimize the time consumption of the iterative task process; using an optimization algorithm that combines parallel computing on the basis of the first optimization to speed up the calculation speed. By comparing the three optimization acceleration algorithms, the optimization acceleration algorithm with the least running time, smaller fluctuations, faster and more stable convergence speed is finally selected as the optimal algorithm for capacity configuration. After comparison, the third algorithm is finally selected for capacity configuration in this application. By adopting an improved differential particle swarm hybrid optimization algorithm that reduces iterative graph output, pauses briefly to update the graph, uses vectorized form, simplifies the position and speed restriction processing, and combines parallel computing, it is not simply limited to the small point optimization of the heuristic algorithm, making the optimization results more extensive and representative.

[0125] In some embodiments, after the iteration of step 206 is completed, the following may also be performed: sensitivity and economic analysis of the photovoltaic hydrogen production system optimization configuration scheme is performed to provide theoretical scientific guidance for the configuration, construction and development of the renewable energy photovoltaic hydrogen production system.

[0126] So far, the Figure 2 Description of the process shown.

[0127] pass Figure 2 Compared with the existing technology, the process shown can effectively improve the solution speed of the optimization model, thereby improving the capacity configuration efficiency.

[0128] In one embodiment, an acceleration device for a photovoltaic hydrogen production capacity configuration optimization system is provided, comprising:

[0129] A model building module is used to establish a double-layer optimization model for a photovoltaic hydrogen production system, wherein the upper model of the double-layer optimization model takes the minimum unit hydrogen production cost over the entire life cycle as the upper layer objective function, and the lower model takes the minimum power abandonment rate as the lower layer objective function; the photovoltaic hydrogen production system at least includes a photovoltaic power generation system, a proton exchange membrane PEM electrolysis hydrogen production system, an alkaline electrolysis hydrogen production system, an energy storage battery and a hydrogen storage tank;

[0130] A model solving module is used to solve the double-layer optimization model to obtain the optimal capacity configuration of the photovoltaic hydrogen production system; wherein the upper layer model is solved by using a differential particle swarm hybrid optimization algorithm, and the lower layer model is solved by using a linear programming problem solving method; the differential particle swarm hybrid optimization algorithm is used to solve, including using one of the following three acceleration algorithms to solve:

[0131] Acceleration algorithm 1: An improved differential particle swarm hybrid optimization algorithm that reduces iterative graph output, pauses briefly to update the graph, uses vectorized operations, and simplifies position and velocity constraint processing;

[0132] Acceleration algorithm 2: Based on acceleration algorithm 1, it combines GPU parallel optimization algorithm;

[0133] Acceleration algorithm three: an optimization algorithm that combines parallel computing based on acceleration algorithm one.

[0134] The implementation process of the functions and effects of each module in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, which will not be repeated here.

[0135] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0136] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. An acceleration algorithm for photovoltaic hydrogen production capacity configuration optimization system, characterized in that: include: A two-layer optimization model for a photovoltaic hydrogen production system is established, wherein the upper model of the two-layer optimization model takes the minimum unit hydrogen production cost over the entire life cycle as the upper objective function, and the lower model takes the minimum power abandonment rate as the lower objective function; the photovoltaic hydrogen production system at least includes a photovoltaic power generation system, a proton exchange membrane PEM electrolysis hydrogen production system, an alkaline electrolysis hydrogen production system, an energy storage battery and a hydrogen storage tank; Solving the two-layer optimization model to obtain the optimal capacity configuration of the photovoltaic hydrogen production system; wherein the upper layer model is solved by a differential particle swarm hybrid optimization algorithm, and the lower layer model is solved by a linear programming problem solving method; the differential particle swarm hybrid optimization algorithm is used to solve the problem, including using one of the following three acceleration algorithms to solve the problem: Acceleration algorithm 1: An improved differential particle swarm hybrid optimization algorithm that reduces iterative graph output, pauses briefly to update the graph, uses vectorized operations, and simplifies position and velocity constraint processing; Acceleration algorithm 2: Based on the acceleration algorithm 1, a GPU parallel optimization algorithm is combined; Acceleration algorithm three: an optimization algorithm combining parallel computing with the acceleration algorithm one.

2. The method according to claim 1, characterized in that The running times of the acceleration algorithm 1, the acceleration algorithm 2 and the acceleration algorithm 3 are compared in advance, and finally an acceleration algorithm with the shortest running time is adopted.

3. The method according to claim 1, characterized in that The upper objective function formula is as follows: Among them, C ALK represents the total life cycle cost of the alkaline electrolysis hydrogen production system, in yuan; C PEM represents the total life cycle cost of the PEM electrolysis hydrogen production system, in yuan; C Eb Represents the total life cycle cost of the energy storage battery, in yuan; C Eh Represents the total life cycle cost of the hydrogen storage tank, in yuan; C LOSE represents the penalty cost for power abandonment, in yuan; Q H2 Represents the total hydrogen production of the system over its entire life cycle, in Nm 3 .

4. The method according to claim 1, characterized in that: The lower layer objective function formula is as follows: Among them, P PV Represents the output power of the photovoltaic power generation system, in kW; P PV,rated represents the rated power of the photovoltaic power generation system, in kW; f represents the power generation efficiency of the photovoltaic power generation system; δ represents the temperature coefficient; G and Gs represent the actual solar irradiance and reference irradiance, in W / m 2 ; T and Ts represent the actual working temperature and reference temperature, the unit is K.

5. The method according to claim 1, characterized in that The upper-level constraint condition of the upper-level model includes equipment capacity constraint, and the equipment capacity constraint formula is as follows: P ALK,rated +P PEM,rated +P Eb,rated ≤P PV,rated P ALK,rated ≥0 P PEM,rated ≥0 P Eb,rated ≥0 Among them, P ALK,rated Represents the capacity of the alkaline electrolysis hydrogen production system, in kW; P PEM,rated Represents the capacity of the PEM electrolysis hydrogen production system, in kW; P Eb,rated Represents the capacity of the energy storage battery, in kW.

6. The method according to claim 1, characterized in that The step of solving the double-layer optimization model to obtain the optimal capacity configuration of the photovoltaic hydrogen production system includes: Setting equipment capacity parameters of the photovoltaic hydrogen production system under current calculation, and transferring the equipment capacity parameters into the lower model; Setting parameters for the lower model, wherein the parameters at least include equipment capacity parameters and photovoltaic typical daily output data; Calling CPLEX to optimize and solve the lower model, obtain the optimal operating power of the alkaline electrolysis hydrogen production system and the PEM electrolysis hydrogen production system at each moment, and return the optimal operating power to the upper model; The upper model takes the minimum unit hydrogen production cost over the entire life cycle as the upper objective function, calculates the individual fitness value of the population particles, and updates the population particles; When the maximum number of iterations is reached, the iteration is stopped and the optimal capacity configuration of the photovoltaic hydrogen production system is output; otherwise, the process returns to execute parameter setting of the lower model.

7. The method according to claim 1, characterized in that The method further includes: establishing a hydrogen production model for the photovoltaic hydrogen production system, the hydrogen production model at least including: in, Represents the unit power consumption of hydrogen, the unit is kWh / Nm 3 ; η el,ALK , η el,PEM , η el,Eb Respectively represent the electrolysis efficiency of alkaline electrolysis hydrogen production system, PEM electrolysis hydrogen production system and energy storage battery; P ALK , P PEM , P Eb They represent the input power of the alkaline electrolysis hydrogen production system, the PEM electrolysis hydrogen production system and the energy storage battery, respectively, in kW; They represent the hydrogen production rates of the alkaline electrolysis hydrogen production system, PEM electrolysis hydrogen production system and energy storage battery respectively.

8. The method according to claim 1, characterized in that: The photovoltaic output data of the photovoltaic hydrogen production system within a preset time is clustered, and a photovoltaic output scenario is obtained according to each type of photovoltaic output data obtained by clustering.

9. An acceleration device for photovoltaic hydrogen production capacity configuration optimization system, characterized in that: include: A model building module is used to establish a double-layer optimization model for a photovoltaic hydrogen production system, wherein the upper model of the double-layer optimization model takes the minimum unit hydrogen production cost over the entire life cycle as the upper layer objective function, and the lower model takes the minimum power abandonment rate as the lower layer objective function; the photovoltaic hydrogen production system at least includes a photovoltaic power generation system, a proton exchange membrane PEM electrolysis hydrogen production system, an alkaline electrolysis hydrogen production system, an energy storage battery and a hydrogen storage tank; A model solving module is used to solve the double-layer optimization model to obtain the optimal capacity configuration of the photovoltaic hydrogen production system; wherein the upper layer model is solved by using a differential particle swarm hybrid optimization algorithm, and the lower layer model is solved by using a linear programming problem solving method; the differential particle swarm hybrid optimization algorithm is used to solve, including using one of the following three acceleration algorithms to solve: Acceleration algorithm 1: An improved differential particle swarm hybrid optimization algorithm that reduces iterative graph output, pauses briefly to update the graph, uses vectorized operations, and simplifies position and velocity constraint processing; Acceleration algorithm 2: Based on the acceleration algorithm 1, a GPU parallel optimization algorithm is combined; Acceleration algorithm three: an optimization algorithm combining parallel computing with the acceleration algorithm one.

10. The acceleration device according to claim 9, characterized in that: The acceleration algorithm 1, the acceleration algorithm 2 and the acceleration algorithm 3 are compared in advance, and finally an acceleration algorithm with the shortest running time, smaller fluctuation, faster and more stable convergence speed is adopted.

Images