CN116540545A - A Stochastic Optimal Scheduling Method for Photovoltaic Power Generation Hydrogen Production Cluster Based on Ito Process - Google Patents

Abstract

The invention discloses a random optimization scheduling method of a photovoltaic power generation hydrogen production cluster based on an illite process, which comprises the steps of firstly establishing a photovoltaic power generation hydrogen production system model, wherein the photovoltaic power generation hydrogen production system model comprises an illite process model of photovoltaic output, an electrolytic cell cluster scheduling model and a hydrogen storage tank model; then, taking the highest expected economic benefit and the highest energy utilization rate under the influence of uncertainty of photovoltaic output prediction errors as objective functions, and establishing a random optimization scheduling model of the electrolytic tank cluster based on a photovoltaic power generation hydrogen production system model and by utilizing a control law of an affine strategy design system; and finally, based on the track sensitivity decomposition, transforming the random optimization control model into a deterministic optimization problem, and adopting a model predictive control mode to roll and optimize the solution. The invention can dynamically adjust the running state and power distribution of each electrolytic cell, effectively realize high-efficiency electro-hydrogen production under the uncertain condition of photovoltaic output and fully absorb fluctuating photovoltaic output.

Description

A Stochastic Optimal Scheduling Method for Photovoltaic Power Generation Hydrogen Production Cluster Based on Ito Process

technical field

The invention relates to the technical field of electrolyzer cluster scheduling, in particular to a random optimal scheduling method for photovoltaic power generation hydrogen production clusters based on the Ito process.

Background technique

Hydrogen production from renewable energy such as photovoltaics is one of the important technical routes for building a clean and low-carbon society in my country. Limited by the capacity of a single machine, it is necessary to form a cluster of several to dozens of large-scale hydrogen production machines to meet the demand for hydrogen production. In addition, the uncertainty of renewable energy output will further affect the efficiency and reliability of electric hydrogen production. Therefore, research on the stochastic optimization scheduling method of electrolyzer clusters considering the uncertainty of photovoltaic output has great practical significance for its own high-efficiency economic operation and participation in the consumption of photovoltaic power generation.

In the existing electrolyzer cluster scheduling research, Shen Xiaojun et al. proposed an optimal control strategy for off-grid wind power hydrogen production alkaline electrolyzer arrays considering electrothermal characteristics. Coordinated control strategy for electrolyzer array rotation in the system. Expanding the load flexibility of industrial P2H factories: the process constraint-aware scheduling method also takes into account the hydrogen generator temperature and the accumulation effect of hydrogen impurities in oxygen, and proposes a variable load control method for hydrogen production clusters to accommodate wind and solar power generation. In the discussion of the configuration and operation rules of the multi-electrolyzer hybrid system of the large-scale alkaline water hydrogen production system, Li Y et al. proposed the electrolyzer cycle rotation strategy of the wind power hydrogen production system to balance the working time of each electrolyzer. The impact of renewable energy access on the hydrogen energy storage system and the control strategy proposed by Niu Meng et al. Based on the reaction mechanism of the electrolytic hydrogen production equipment in the hydrogen energy storage system, a modular system to alleviate the impact of renewable energy on the hydrogen energy storage system is proposed. Hydrogen Control Strategies. Yuan Tiejiang et al. proposed a day-ahead output optimization model for the hydrogen production system considering the start-stop characteristics of the electrolyzer based on the operation state transition relationship of the electrolyzer in the day-ahead output plan of the hydrogen production system considering the start-stop characteristics of the electrolyzer.

However, the above studies are all based on the deterministic optimization of electricity price or renewable energy output prediction, and there is still a lack of research on the stochastic optimal scheduling of electrolyzer clusters under the uncertainty of photovoltaic output.

Contents of the invention

In view of the above problems, the purpose of the present invention is to provide a random optimization scheduling method for electrolyzer clusters of a photovoltaic power generation hydrogen production system, which can model the uncertainty of photovoltaic output based on the Ito process, and can realize the consideration of the uncertainty of photovoltaic output. Randomly optimize the scheduling of electrolyzer clusters to achieve efficient hydrogen production by electricity and improve its ability to absorb fluctuating photovoltaics.

The technical solution is as follows:

A stochastic optimal scheduling method for a photovoltaic power generation hydrogen production cluster based on the Ito process, comprising the following steps:

Step 1: Establish a photovoltaic power generation hydrogen production system model, including the Ito process model of photovoltaic output, electrolyzer cluster scheduling model and hydrogen storage tank model;

Step 2: Taking the highest expected economic benefit and the highest energy utilization rate under the influence of photovoltaic output forecast error uncertainty as the objective function, based on the photovoltaic power generation hydrogen production system model and using the affine strategy to design the control law of the system, establish an electrolyzer cluster The stochastic optimization scheduling model of ;

Step 3: Based on the trajectory sensitivity decomposition, the stochastic optimization control model is transformed into a deterministic optimization problem, and the rolling optimization problem is solved in the form of model predictive control.

Further, the Ito process model of the photovoltaic output is:

In the formula: P _t ^PV is the actual output of photovoltaic power; P _t ^PV,pred is the predicted value of photovoltaic power; is the prediction error;

The Ito process model of the forecast error is modeled as:

In the formula: W _t represents the standard Wiener process; is the drift item; /> is the diffusion item; a is the time constant of cloud shadow change; b is the mean return target of the drift item; t _r and t _s are the sunrise and sunset times respectively; the sine function represents the sun altitude angle; c represents the uncertainty intensity; d and e represent the change interval of the irradiation intensity.

Furthermore, the electrolyzer cluster scheduling model includes:

Logical constraints for the transformation of the operating state of a single electrolyzer:

In the formula: n represents the nth electrolyzer in the electrolyzer cluster; and /> Respectively represent the three operating states of the electrolyzer: production, standby and shutdown; /> and /> Respectively for the start-up and shutdown operations of the electrolyzer, /> Indicates the transition of the electrolyzer from the standby state to the running state; the above variables are all represented by binary variables; the subscripts t-1 and t-2 respectively represent the previous scheduling period and the last scheduling period when the scheduling period is t;

The physical constraints of the electrolyser cluster itself:

In the formula: N is the total number of electrolyzers; P _n,t is the power of the nth electrolyzer at time t; P ^max and P ^min are the upper and lower limit constraints of the power of a single electrolyzer in the production state; P ^SB is the power of a single electrolyzer The constant power of the auxiliary machine in the standby state of the electrolyzer; F _n,t is the hydrogen production of the nth electrolyzer; A ₁ , A ₂ , and A ₃ are constant coefficients, which are determined by the characteristics of the electrolyzer.

Furthermore, the hydrogen storage tank model includes:

The hydrogen buffer tank storage capacity and its capacity constraints are:

In the formula: is the storage capacity of the hydrogen buffer tank at time t; F _t ^out is the outlet flow of the hydrogen buffer tank, which is supplied to the downstream application of hydrogen; /> and /> Respectively, the upper and lower limits of the available storage range of the hydrogen buffer tank; /> is the climbing rate of hydrogen exhaust; Δt is the scheduling step; /> are the upper and lower limits of the ramp rate of hydrogen emission, respectively.

Further, the step 2 is specifically:

Step 2.1: the objective function is:

In the formula: Indicates that in the initial state of the system x ₀ and the initial value of the prediction error /> Conditional expectations under; /> and C ^PV are the selling price of hydrogen and the kWh cost of electricity purchased from photovoltaic power stations; C ^SU and C ^SD are the operating costs of starting and shutting down the electrolyzer respectively; α and β are cost conversion coefficients, and the two corresponding to α The secondary term represents the maximum utilization of energy, and the secondary term corresponding to β represents that the hydrogen buffer tank storage is as close as possible to the initial value after the entire scheduling period; /> and /> are the storage capacity of the hydrogen buffer tank at the end T time and the storage capacity at the initial time, respectively; /> is the augmented state variable, where: /> Represents a state variable, /> Indicates the control variable; Q is a constant coefficient matrix; T is the scheduling period, and x _T is the state variable at the end T time;

Step 2.2: Express the inequality constraints in the photovoltaic hydrogen production system as a probability form, construct the opportunity constraints, and express them in vector form:

In the formula: γ represents the tolerance size; Pr[·] represents the probability; vector of coefficients representing inequality constraints, /> is the upper limit of inequality constraints; Ω ^C is the set of all constraints;

Step 2.3: Use the affine strategy to design the control law of the system, and parameterize the control command as an affine function of the PV output prediction error:

In the formula: is a constant term, and K _t is the gain coefficient matrix; therefore, the /> simplifies to />

Step 2.4: The objective function, constraint conditions and control function established by Lianli, and the stochastic optimization scheduling model of the electrolyzer cluster are as follows:

s.t.

In the formula: is an integer variable; φ _i is the coefficient of an integer variable; /> is the constraint upper limit; A, B, C and D are the coefficient matrices of state variables, control variables, random variables and integer variables, respectively.

Further, the step 3 is specifically:

Step 3.1: Transform the stochastic optimal scheduling model of electrolyzer clusters into deterministic optimization by augmenting the trajectory sensitivity decomposition of state variables; specifically include:

Step 3.1.1: Put the variables in the random process Decomposition into baseline trajectory variables /> and trajectory sensitivity variables /> As follows:

In the formula: and /> are the reference value trajectory variable and the trajectory sensitivity variable form of the state variable respectively, /> is the step size of the series expansion term; o(·) represents the high-order infinitesimal term; /> and /> are the baseline value trajectory variable and the trajectory sensitivity variable of the random variable, respectively;

Step 3.1.2: Split the formulas (20)-(21) in the stochastic optimal scheduling model into the reference value trajectory equation and the trajectory sensitivity equation as shown below;

Baseline trajectory equation:

In the formula: the initial condition is

Trajectory sensitivity equation:

In the formula: the initial condition is is the gain coefficient in the system control law;

Step 3.1.3: Based on the trajectory sensitivity decomposition to M _t , the objective function (19) is decomposed as follows:

J＝J ₀ +J ₁ (31)

In the formula: is the reference value trajectory variable of the state variable at time T; /> is the trajectory sensitivity variable form of the variable set; /> Represents the derivative of each variable in the function;

Step 3.1.4: In order to ensure that the optimal scheduling model is convex, do the following relaxation processing:

In the formula: is the reference value trajectory variable form of the variable set;

The chance constraint (23) expresses:

In the formula: κ _γ is a constant term, corresponding to the quantile of standard normal distribution γ;

Step 3.1.5: Formulas (19)-(24) in the stochastic optimization scheduling model are expressed as the following deterministic optimization problems:

min J＝J ₀ +J ₁ (37)

s.t.

Step 3.2: Solve the stochastic optimal scheduling model (37)-(46) rollingly in the form of model predictive control;

Step 3.2.1: Given the initial time t, the initial state of the photovoltaic power generation hydrogen production system x _t , and z _t , a given control cycle T; where, /> is the prediction error at the initial moment;

Step 3.2.2: Predict the PV output at time t+1 As the input of model predictive control, solve the stochastic optimal scheduling model shown in equations (37)-(46), and obtain the state x _t+ 1 of the system at time t+1, /> and z _t+1 and the control law

Step 3.2.3: Repeat step 3.2.2 until t>T; output the optimal scheduling results x _T , z _T and the objective function J of the photovoltaic power generation hydrogen production system.

The beneficial effects of the present invention are:

1) The present invention proposes to model the uncertainty of photovoltaic output based on the Ito process model, realize unified modeling, analysis and control with the cluster scheduling model of electrolyzers under the framework of stochastic dynamics, and establish an electrolyzer cluster considering the uncertainty of photovoltaic output Stochastic optimization scheduling model; it can dynamically adjust the operating status and power distribution of each electrolyzer, realize efficient hydrogen production under the condition of photovoltaic output uncertainty, and improve its ability to absorb fluctuating photovoltaics.

2) The present invention converts the stochastic optimal scheduling problem into deterministic optimization based on dynamic trajectory sensitivity decomposition, decomposes the objective function and chance constraints of the stochastic optimal scheduling model, and transforms the stochastic dynamic optimization problem into deterministic second-order cone programming, The model predictive control rolling solution is adopted, which effectively avoids the shortcomings of high computational complexity and low solution efficiency of the traditional method based on random sampling simulation.

Description of drawings

Fig. 1 is a schematic flow diagram of the method for randomly optimizing the scheduling of electrolyzers in the photovoltaic hydrogen production system based on the Ito process of the present invention.

Figure 2 shows the overall structure of the photovoltaic power generation hydrogen production system.

Fig. 3 is the 96-period operating state of each electrolyzer in the embodiment.

Fig. 4 shows the photovoltaic output and electrolyzer cluster power in the embodiment.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings and specific embodiments. The present invention models the uncertainty of photovoltaic output based on the Ito process, and can realize random optimal scheduling of electrolyzer clusters under the condition of taking into account the uncertainty of photovoltaic output, realize high-efficiency electric hydrogen production, and improve its ability to absorb fluctuating photovoltaics. Ability. Firstly, the system model of photovoltaic power generation hydrogen production is established, including the Ito process model of photovoltaic output, the electrolyzer cluster scheduling model and the hydrogen storage tank model; then the expected economic benefit and energy utilization rate under the influence of the uncertainty of photovoltaic output prediction error are the highest objective function, and use the affine strategy to design the control law of the system, and build a stochastic optimal scheduling model for the electrolyzer cluster based on the above model; finally, based on the trajectory sensitivity decomposition, the stochastic optimal control model is transformed into a deterministic optimization problem, and the model is used to predict Formal rolling optimization solvers for control. As shown in Figure 1, the detailed process is as follows:

Step 1: Establish a photovoltaic power generation hydrogen production system model, including the Ito process model of photovoltaic output, electrolyzer cluster scheduling model and hydrogen storage tank model.

The photovoltaic power generation hydrogen production system is composed of photovoltaic power generation, electrolyzer clusters, hydrogen buffer tanks and other units. The overall structure is shown in Figure 2.

1.1 Ito process model of photovoltaic output

In the formula: P _t ^PV is the actual output of photovoltaic power; P _t ^PV,pred is the predicted value of photovoltaic power; is the prediction error.

The Ito process model of the forecast error is modeled as:

In the formula: W _t represents the standard Wiener process; is the drift item; /> is the diffusion item; a is the time constant of cloud shadow change; b is the mean return target of the drift item; t _r and t _s are the sunrise and sunset times respectively; the sine function represents the sun altitude angle; c represents the uncertainty intensity; d and e represent the change interval of the irradiation intensity.

1.2 Electrolyzer cluster scheduling model

Logical constraints for the transformation of the operating state of a single electrolyzer:

In the formula: n represents the nth electrolyzer in the electrolyzer cluster; Three operating states of the electrolyzer: production, standby, shutdown; /> Respectively for the start-up and shutdown operations of the electrolyzer, /> Indicates the transition of the electrolyzer from the standby state to the running state; the above variables are all represented by binary variables; the subscripts t-1 and t-2 represent the previous scheduling period and the next scheduling period when the scheduling period is t, respectively.

The physical constraints of the electrolyser cluster itself:

In the formula: N is the total number of electrolyzers; P _n,t is the power of the nth electrolyzer at time t; P ^max and P ^min are the upper and lower limit constraints of the power of a single electrolyzer in the production state; P ^SB is the power of a single electrolyzer The constant power of the auxiliary machine in the standby state of the electrolyzer; F _n,t is the hydrogen production of the nth electrolyzer; A ₁ , A ₂ , and A ₃ are constant coefficients, which are determined by the characteristics of the electrolyzer.

1.3 The hydrogen buffer tank storage capacity and its capacity constraints are:

In the formula: is the storage capacity of the hydrogen buffer tank; F _t ^out is the outlet flow of the hydrogen buffer tank, which is supplied to the downstream application of hydrogen; and /> Respectively, the upper and lower limits of the available storage range of the hydrogen buffer tank; /> is the climbing rate of hydrogen exhaust; Δt is the scheduling step; /> are the upper and lower limits of the ramp rate of hydrogen emission, respectively.

Step 2: Taking the highest expected economic benefit and the highest energy utilization rate under the influence of photovoltaic output forecast error uncertainty as the objective function, based on the photovoltaic power generation hydrogen production system model and using the affine strategy to design the control law of the system, establish an electrolyzer cluster stochastic optimization scheduling model.

2.1 The objective function is to maximize the expected economic benefits and energy utilization efficiency of the electrolyzer cluster under the influence of the uncertainty of photovoltaic output prediction error:

In the formula: Indicates that in the initial state of the system x ₀ and the initial value of the prediction error /> Conditional expectations under; /> C ^PV is the selling price of hydrogen and the kWh cost of electricity purchased from photovoltaic power plants; C ^SU and C ^SD are the operating costs of starting and shutting down the electrolyzer; α and β are cost conversion coefficients, and the quadratic The term represents the maximum utilization of energy, and the quadratic term corresponding to β represents that the storage capacity of the hydrogen buffer tank is as close as possible to the initial value after the entire scheduling period; /> is the augmented state variable, where: /> Represents a state variable, /> Indicates the control variable; Q is a constant coefficient matrix.

2.2 Express the inequality constraints in the photovoltaic power generation hydrogen production system as a probability form, construct the opportunity constraints, and express them in vector form:

In the formula: γ represents the tolerance size; Pr[·] represents the probability; vector of coefficients representing inequality constraints, /> is the upper limit of inequality constraints; Ω ^C is the set of all constraints.

2.3 The control law of the system is designed using the affine strategy, and the control command is parameterized as an affine function of the photovoltaic output prediction error:

In the formula: is a constant item, and K _t is the gain coefficient matrix.

2.4 The objective function, constraint conditions and control function established by Lianli, and the stochastic optimization scheduling model of the electrolyzer cluster are as follows:

s.t.

In the formula: is an integer variable; φ _i is the coefficient of an integer variable; /> is the upper bound limit.

Step 3: Based on the trajectory sensitivity decomposition, the stochastic optimization scheduling model is transformed into a deterministic optimization problem, and the rolling solution is adopted in the form of MPC, as follows:

3.1 Transform stochastic optimization scheduling models (19)-(24) into deterministic optimization by augmenting the trajectory sensitivity decomposition of state variables

First, the variables in the random process Decomposition into baseline trajectory variables /> and trajectory sensitivity variables /> As follows:

For this reason, equations (20)-(21) in the stochastic optimal scheduling model can be split into baseline value trajectory equations (27)-(28) and trajectory sensitivity equations (29)-(30) as shown below.

The reference value trajectory equation of formula (20)-(21):

In the formula: the initial condition is

The trajectory sensitivity equations of equations (20)-(21):

In the formula: the initial condition is is the gain coefficient in the system control law.

Then, based on the trajectory sensitivity decomposition to _Mt , the objective function (19) is decomposed as follows:

J＝J ₀ +J ₁ (31)

In the formula:

In order to ensure that the optimal scheduling model is convex, the following relaxation processing is done:

And the chance constraint (23) can be expressed as:

So far, the stochastic optimization scheduling model (19)-(24) can be expressed as the following deterministic optimization problem:

min J＝J ₀ +J ₁ (37)

s.t.

3.2 Using the form of model predictive control to solve the stochastic optimal scheduling model rollingly (37)-(46)

Step 1): given the initial state x _t of the photovoltaic power generation hydrogen production system at the initial time t, z _t , a given control cycle T.

Step 2): The predicted value of photovoltaic output at time t+1 As the input of model predictive control, solve the stochastic optimal scheduling model shown in equations (37)-(46), and obtain the state x _t+ 1 of the system at time t+1, /> z _t+1 and the control law

Step 3): Repeat step 2 until t>T. Output the optimal scheduling results x _T , z _T and the objective function J of the photovoltaic power generation hydrogen production system.

4. Example analysis

In this example, a photovoltaic power station in a certain area in southwest Sichuan is selected. The hydrogen production plant is equipped with 4 alkaline electrolyzers to form a cluster, and 24-hour optimal scheduling is carried out, with a scheduling step of 15 minutes.

This example builds a stochastic optimization scheduling model of electrolyzer based on Wolfram Mathematica platform, and uses Mosek solver to solve it.

Figure 3 shows the operating status of each electrolyzer at 96 hours under the example; Figure 4 shows the photovoltaic output and the cluster power of the electrolyzer under the example.

It can be seen that the random optimization scheduling method of the electrolyzer cluster of the photovoltaic hydrogen production system based on the Ito process proposed in the present invention can dynamically adjust the operating status of each electrolyzer, and effectively realize the efficient electric hydrogen production under the condition of photovoltaic output uncertainty , and fully absorb the fluctuating photovoltaic output.

Claims (6)

1. A photovoltaic power generation hydrogen production cluster stochastic optimal scheduling method based on the Ito process, characterized in that it comprises the following steps: Step 1: Establish a photovoltaic power generation hydrogen production system model, including the Ito process model of photovoltaic output, electrolyzer cluster scheduling model and hydrogen storage tank model; Step 2: Taking the highest expected economic benefit and the highest energy utilization rate under the influence of photovoltaic output forecast error uncertainty as the objective function, based on the photovoltaic power generation hydrogen production system model and using the affine strategy to design the control law of the system, establish an electrolyzer cluster The stochastic optimization scheduling model of ; Step 3: Based on the trajectory sensitivity decomposition, the stochastic optimization control model is transformed into a deterministic optimization problem, and the rolling optimization problem is solved in the form of model predictive control. 2. The stochastic optimal scheduling method of photovoltaic power generation hydrogen production cluster based on Ito process according to claim 1, characterized in that, the Ito process model of said photovoltaic output is: In the formula: P _t ^PV is the actual output of photovoltaic power; P _t ^PV,pred is the predicted value of photovoltaic power; is the prediction error; The Ito process model of the forecast error is modeled as: In the formula: W _t represents the standard Wiener process; is the drift item; /> is the diffusion item; a is the time constant of cloud shadow change; b is the mean return target of the drift item; t _r and t _s are the sunrise and sunset times respectively; the sine function represents the sun altitude angle; c represents the uncertainty intensity; d and e represent the change interval of the irradiation intensity. 3. The random optimal scheduling method for photovoltaic power generation hydrogen production cluster based on Ito process according to claim 2, characterized in that, the electrolyzer cluster scheduling model includes: Logical constraints for the transformation of the operating state of a single electrolyzer: In the formula: n represents the nth electrolyzer in the electrolyzer cluster; and /> Respectively represent the three operating states of the electrolyzer: production, standby and shutdown; /> and /> Respectively for the start-up and shutdown operations of the electrolyzer, /> Indicates the transition of the electrolyzer from the standby state to the running state; the above variables are all represented by binary variables; the subscripts t-1 and t-2 respectively represent the previous scheduling period and the last scheduling period when the scheduling period is t; The physical constraints of the electrolyser cluster itself: In the formula: N is the total number of electrolyzers; P _n,t is the power of the nth electrolyzer at time t; P ^max and P ^min are the upper and lower limit constraints of the power of a single electrolyzer in the production state; P ^SB is the power of a single electrolyzer The constant power of the auxiliary machine in the standby state of the electrolyzer; F _n,t is the hydrogen production of the nth electrolyzer; A ₁ , A ₂ , and A ₃ are constant coefficients, which are determined by the characteristics of the electrolyzer. 4. The stochastic optimization scheduling method of photovoltaic power generation hydrogen production cluster based on Ito process according to claim 3, characterized in that, the hydrogen storage tank model includes: The hydrogen buffer tank storage capacity and its capacity constraints are: In the formula: is the hydrogen buffer tank storage capacity at time t; /> is the outlet flow rate of the hydrogen buffer tank, supplying the downstream application of hydrogen; and /> Respectively, the upper and lower limits of the available storage range of the hydrogen buffer tank; /> is the climbing rate of hydrogen exhaust; Δt is the scheduling step; /> are the upper and lower limits of the ramp rate of hydrogen emission, respectively. 5. The random optimal scheduling method of photovoltaic power generation hydrogen production cluster based on Ito process according to claim 4, characterized in that, the step 2 is specifically: Step 2.1: the objective function is: In the formula: Indicates that in the initial state of the system x ₀ and the initial value of the prediction error /> Conditional expectations under; /> and C ^PV are the selling price of hydrogen and the kWh cost of electricity purchased from photovoltaic power stations; C ^SU and C ^SD are the operating costs of starting and shutting down the electrolyzer respectively; α and β are cost conversion coefficients, and the two corresponding to α The secondary term represents the maximum utilization of energy, and the secondary term corresponding to β represents that the hydrogen buffer tank storage is as close as possible to the initial value after the entire scheduling period; /> and /> Respectively, the storage capacity of the hydrogen buffer tank at the end T time and the storage capacity of the hydrogen buffer tank at the initial time; /> is the augmented state variable, where: /> Represents a state variable, /> Indicates the control variable; Q is a constant coefficient matrix; T is the scheduling period, and x _T is the state variable at the end T time; Step 2.2: Express the inequality constraints in the photovoltaic hydrogen production system as a probability form, construct the opportunity constraints, and express them in vector form: In the formula: γ represents the tolerance size; Pr[·] represents the probability; vector of coefficients representing inequality constraints, /> is the upper limit of inequality constraints; Ω ^C is the set of all constraints; Step 2.3: Use the affine strategy to design the control law of the system, and parameterize the control command as an affine function of the PV output prediction error: In the formula: is a constant term, and K _t is the gain coefficient matrix; therefore, the /> simplifies to /> Step 2.4: The objective function, constraint conditions and control function established by Lianli, and the stochastic optimization scheduling model of the electrolyzer cluster are as follows: s.t. In the formula: is an integer variable; φ _i is the coefficient of an integer variable; /> is the constraint upper limit; A, B, C and D are the coefficient matrices of state variables, control variables, random variables and integer variables, respectively. 6. The random optimization scheduling method of photovoltaic power generation hydrogen production cluster based on Ito process according to claim 5, characterized in that, the step 3 is specifically: Step 3.1: Transform the stochastic optimal scheduling model of electrolyzer clusters into deterministic optimization by augmenting the trajectory sensitivity decomposition of state variables; specifically include: Step 3.1.1: Put the variables in the random process Decomposition into baseline trajectory variables /> and trajectory sensitivity variables /> As follows: In the formula: and /> are the reference value trajectory variable and the trajectory sensitivity variable form of the state variable respectively, /> is the step size of the series expansion term; o(·) represents the high-order infinitesimal term; /> and /> are the baseline value trajectory variable and the trajectory sensitivity variable of the random variable, respectively; Step 3.1.2: Split the formulas (20)-(21) in the stochastic optimal scheduling model into the reference value trajectory equation and the trajectory sensitivity equation as shown below; Baseline trajectory equation: ) In the formula: the initial condition is Trajectory sensitivity equation: In the formula: the initial condition is is the gain coefficient in the system control law; Step 3.1.3: Based on the trajectory sensitivity decomposition to M _t , the objective function (19) is decomposed as follows: J＝J ₀ +J ₁ (31) In the formula: is the reference value trajectory variable of the state variable at time T; /> is the trajectory sensitivity variable form of the variable set; /> Represents the derivative of each variable in the function; Step 3.1.4: In order to ensure that the optimal scheduling model is convex, do the following relaxation processing: In the formula: is the reference value trajectory variable form of the variable set; The chance constraint (23) expresses: In the formula: κ _γ is a constant term, corresponding to the quantile of standard normal distribution γ; Step 3.1.5: Formulas (19)-(24) in the stochastic optimization scheduling model are expressed as the following deterministic optimization problems: min J＝J ₀ +J ₁ (37) s.t. Step 3.2: Solve the stochastic optimal scheduling model (37)-(46) rollingly in the form of model predictive control; Step 3.2.1: Given the initial time t, the initial state of the photovoltaic power generation hydrogen production system x _t , and z _t , a given control cycle T; where, /> is the prediction error at the initial moment; Step 3.2.2: Predict the PV output at time t+1 As the input of model predictive control, solve the stochastic optimal scheduling model shown in equations (37)-(46), and obtain the state x _t+ 1 of the system at time t+1, /> and z _t+1 and the control law Step 3.2.3: Repeat step 3.2.2 until t>T; output the optimal scheduling results x _T , z _T and the objective function J of the photovoltaic power generation hydrogen production system.