Scheduling optimization method for hydrogen-electricity coupling system based on source load prediction

Abstract

The invention discloses a scheduling optimization method of a hydrogen-electric coupling system for source load prediction, and relates to the technical field of electric power, and the method comprises the steps: S1, building a state space model based on the operation data of the hydrogen-electric coupling system, building a prediction control model through calculating the state variable observation value of the system, and building the operation constraint of the prediction control model; s2, establishing a first objective function based on the model variable of the predictive control model and the intra-day real-time scheduling expectation of the hydrogen-electricity coupling system; s3, carrying out quadratic programming conversion on the first target function, calculating a real-time source load prediction value through a prediction control model, updating the converted first target function, and obtaining a second target function containing a linear relation; s4, iteratively solving the second objective function to obtain an optimal system scheduling scheme; the calculation efficiency is remarkably improved, the scheduling efficiency is further improved, the real-time dynamic response of the system is reflected by establishing the state space model, and reliable scheduling of the system is achieved.

Description

Technical Field

[0001] The present invention relates to the field of electric power technology, and in particular to a method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction. Background Art

[0002] With the development of renewable energy power generation industry, the proportion of non-fossil energy in primary energy consumption has gradually increased. As a bridge between renewable energy and traditional coal power generation, the hydrogen-electric coupling system couples renewable energy with coal resources through the stabilization of hydrogen energy storage, which is not only conducive to solving the problem of high carbon emissions in the coal conversion process and achieving the dual goals, but also can absorb wind / power abandonment and improve the utilization rate of renewable energy. However, during the actual operation of the hydrogen-electric coupling system, it is difficult to effectively allocate the power of hydrogen energy storage and electrochemical energy storage in the face of source-load fluctuations, resulting in problems such as unreliable system scheduling schemes, which in turn affects the safety, reliability and economic operation of the system. In the prior art, the system energy input and load demand fluctuations are usually balanced according to multiple time scales. However, with the increase in system scale and complexity, the time complexity of the calculation amount of the scheduling algorithm is also increasing, thereby reducing the scheduling efficiency, and it is difficult to ensure the safety of the highly coupled system during the scheduling process. Therefore, how to improve the real-time and response speed of scheduling while ensuring the accuracy of calculations is an important challenge.

[0003] Chinese patent, publication number: CN118783550A, publication date: October 15, 2024, discloses a multi-time scale optimization scheduling method and system for hydrogen-electricity coupled microgrids. The invention establishes a source-storage-load multi-time scale coordinated optimization scheduling model based on the topological structure of the hydrogen-electricity coupled microgrid, which is composed of seasonal hydrogen storage optimization, day-ahead economic scheduling and intraday dynamic optimization with gradually shortened time scales; on the basis of summarizing and analyzing historical optimization results, expert experience and best practices are introduced to establish an optimization strategy knowledge base to form the constraints of the optimization scheduling model; a variable neighborhood search solution algorithm based on a dynamic window method is used to solve the optimization scheduling model to obtain the optimal energy procurement and equipment scheduling plan for the optimal operation of the hydrogen-electricity coupled microgrid. However, the multi-time scale in this scheme is at least an hour as a period and a period as a scale, which cannot achieve the real-time monitoring of the dynamic response characteristics of the system to optimize the short-term operation of the microgrid in real time, and has poor applicability to the multivariable, time-varying and nonlinear characteristics of hydrogen-electricity energy conversion. Summary of the invention

[0004] The purpose of the present invention is to address the problem that it is difficult for the prior art to take into account both the high-coupling computing efficiency of the system and the real-time dynamic response of the system, resulting in low real-time scheduling efficiency and poor effect of the system; a scheduling optimization method for a hydrogen-electricity coupling system based on source-load prediction is proposed, by establishing a predictive control model of the hydrogen-electricity coupling system, establishing a first objective function and operating constraints for real-time scheduling of the system within a day, performing a quadratic programming function conversion in the first objective function to obtain a second objective function, and obtaining an optimal scheduling result by iteratively solving the second objective function, thereby effectively processing a constrained nonlinear system under limited computing resources, significantly improving the computing efficiency, and further improving the scheduling efficiency; and by establishing a state space model to reflect the real-time dynamic response of the system, reliable scheduling of the system is achieved, thereby overcoming the problem that it is difficult for the prior art to take into account both the high-coupling computing efficiency of the system and the real-time dynamic response of the system, resulting in low real-time scheduling efficiency and poor effect of the system.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction, comprising the following steps: S1. Establishing a state space model based on the operating data of the hydrogen-electric coupling system, establishing a predictive control model by calculating the observed values ​​of the state variables of the system, and establishing operating constraints of the predictive control model; S2, establishing a first objective function based on the model variables of the predictive control model and the intraday real-time scheduling expectation of the hydrogen-electric coupling system; S3, performing a quadratic programming conversion on the first objective function, and updating the converted first objective function by calculating the real-time source-load prediction value through the predictive control model to obtain a second objective function containing a linear relationship; S4. Iteratively solve the second objective function to obtain an optimal system scheduling solution.

[0006] In this scheme, by establishing a discrete-time state-space model of the hydrogen-electric energy conversion device, the impact of different empty column parameters on the real-time scheduling scheme can be dynamically reflected, which is helpful for the subsequent establishment of the objective function, and the optimization of the short-term operation of the microgrid can be achieved based on the power characteristics and mechanisms of each system unit from a time scale of seconds; by establishing operating constraints, the power, energy storage state and control variables of the predictive control model can be constrained, and ultimately the maximum control range of the power can be limited to ensure the stable, safe and reliable operation of the hydrogen-electric coupling system; by constructing a predictive control model, nonlinear control systems such as the hydrogen-electric coupling system can be efficiently processed. For multivariable systems, by converting the control law for solving the optimal control offline into solving the constraint problem of the online numerical value, that is, predicting the future behavior of the system based on the observation value at the current moment, the computational complexity is significantly reduced and the computational efficiency of the optimal scheduling method is improved. At the same time, the predictive control model can adjust the charging and discharging strategy of the energy storage device in real time according to the current source and load conditions and system status to ensure that the system operates in a safe, reliable and economical operating state; by converting the first objective function into a quadratic programming function, so as to quickly and accurately find the optimal solution, the function structure of the optimization problem is simplified, and then the calculation process is simplified, so as to achieve efficient scheduling of the system.

[0007] Preferably, in S1, a state space model is established based on the operating data of the hydrogen-electricity coupling system, and a predictive control model is established by calculating the observed values ​​of the state variables of the system, including: Obtain status data, operating power data and system output data of the hydrogen-electric coupling system, and pre-process the data to obtain target system parameters; Determining a state variable, an output variable, a control variable, and a disturbance variable based on the target system parameters, and establishing a discrete-time state-space model of a hydrogen-to-electricity energy conversion device according to the state variable, the output variable, the control variable, and the disturbance variable; Iteratively calculating the incremental state space model based on the control time domain and the prediction time domain to obtain the state variable observation value of the state space model in the prediction time domain; The output variable prediction value is extracted based on the state observation value, and the predictive control model is constructed with the output variable prediction value.

[0008] Preferably, the state space model is as follows: In the formula, x(k), y(k), u(k) and d(k) are the state variable, output variable, control variable and external measurable disturbance variable of the system at time k respectively; A, B u , B d and C care the correlation matrices of the state-space equation, x(k+1) is the independent variable of the state-space equation, y c (k) is the dependent variable of the state space equation.

[0009] Preferably, constructing the predictive control model based on the output variable predicted value includes: Constructing a prediction output vector for the prediction time domain and a control vector for the control time domain; Substituting the predicted output vector and the control vector into the calculation formula of the output variable predicted value for functional integration to obtain a predictive control model; Wherein, the predictive control model is expressed as follows: Y p (k+1|k)=S x Δx(k)+S y y c (k)+S d Δd(k)+S u ΔU(k) Where Y p (k+1|k) is the predicted output vector of the system at time k predicting time k+i, ΔU(k) ​​is the control vector of the system at time k, S x , S y , S u , S d are the coefficient matrices of the state-space model respectively, Δx(k) is the state variable increment of the system at time k, and Δd(k) is the disturbance variable increment of the system at time k.

[0010] Preferably, in S1, establishing the operation constraints of the predictive control model includes: Establish corresponding power constraints and energy storage state constraints based on the basic configuration properties of the hydrogen-electric coupling system; Simultaneously, an increment constraint is established based on the control variable to limit the control range of power.

[0011] Preferably, S2 comprises the following sub-steps: Determining a first weighting factor of each of the output variables and a second weighting factor of each of the control variables based on the intra-day real-time scheduling expectation of the hydrogen-electricity coupling system; The predicted output vector and the control vector of the predictive control model are combined with the first weighting factor and the second weighting factor to establish the first objective function.

[0012] Preferably, the first objective function is expressed as follows: Wherein, J represents the first objective function, p represents the prediction time domain of the predictive control model variables, m represents the control time domain of the predictive control model, ΔU(k+i-1) is the control vector of the system at time k+i-1, r(k) represents the reference sequence of the output variables at time k, w is the first weighting factor, and Q is the second weighting factor.

[0013] Preferably, S3 includes the following sub-steps: Converting the first objective function into a quadratic programming form based on a standard quadratic programming function; Calculating a real-time source load prediction value based on the prediction control model, subtracting the real-time source load prediction value from the system day-ahead prediction value, and using the difference as a variable observation value of the prediction control model; The transformed first objective function parameters are updated with the variable observation values ​​to obtain the second objective function.

[0014] Preferably, the second objective function is expressed as follows: J ％ =ΔU(k) T HΔU(k)-G(k+1|k) T U(k); In the formula, J ％ represents the second objective function, H is the weighting factor coefficient matrix, G(k+1|k) is the function parameter of the second objective function, ΔU(k) ​​is the control vector of the system at time k, ΔU(k) T is the control matrix of the system at time k, U(k) is the control variable of the system at time k, and T represents the transposed matrix.

[0015] Preferably, S4 includes the following sub-steps: Iteratively solving the second objective function based on the operation constraint to obtain an objective function value, wherein the objective function value at least includes a control increment of a control variable; The control increment obtained in each round of iteration is applied to the hydrogen-electric coupling system, and the coefficient matrix of the state space model is updated until the set prediction time domain and control time domain are reached. The iteration is stopped, and the objective function value of the last round of iteration is output to obtain the optimal state variable observation value and disturbance observation value, and obtain the optimal system scheduling plan.

[0016] Beneficial effects of the present invention: 1. By establishing a spatial state model and a predictive control model, it is possible to more effectively handle the constrained time-varying, nonlinear and multivariable hydrogen-electricity coupling system, quickly and accurately capture source-load disturbances, and achieve more precise real-time scheduling of the hydrogen-electricity coupling system; 2. By dynamically adjusting the weighting factors, the power of different energy storage units can be flexibly allocated according to specific application requirements, achieving more efficient and reliable power control performance; 3. By converting the first objective function into a standard quadratic programming form, the scheduling objective of the hydrogen-electric coupling system can be converted into a simple quadratic optimization problem. The objective function can be minimized by iteratively updating the solution to quickly obtain the optimal control variable power regulation strategy, which has high applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Other features, objects and advantages of the present invention will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings. The drawings are only for the purpose of illustrating preferred embodiments and are not to be considered as limiting the present invention. Also, the same reference symbols are used throughout the drawings to represent the same parts.

[0018] Figure 1 This is a flow chart of scheduling optimization of a hydrogen-electricity coupling system based on source-load prediction in this embodiment.

[0019] Figure 2 This is a flow chart of a system real-time scheduling algorithm of this embodiment.

[0020] Figure 3 This is a schematic diagram of the impact of different prediction time domain and control time domain parameters on the output of this embodiment.

[0021] Figure 4 This is a schematic diagram of the influence of weight factors of different control variables on power regulation in this embodiment.

[0022] Figure 5 This is a source load data change trend diagram of this embodiment.

[0023] Figure 6 This is a schematic diagram of a daily power generation plan for a hydrogen-to-electricity conversion device of this embodiment.

[0024] Figure 7 This is a schematic diagram of a power generation plan of a hydrogen-electricity coupling system after real-time scheduling in this embodiment.

[0025] Figure 8 This is a schematic diagram of a system tie line fluctuation situation in this embodiment.

[0026] Fig. 9 A schematic diagram for comparing the energy storage status tracking effect of this embodiment. DETAILED DESCRIPTION

[0027] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation method described herein is only an optimal embodiment of the present invention, which is only used to explain the present invention and does not limit the scope of protection of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0028] Example 1: Figure 1 and Figure 2 As shown, a method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction includes steps S1-S4, wherein: S1. Establish a state space model based on the operating data of the hydrogen-electric coupling system, establish a predictive control model by calculating the observed values ​​of the state variables of the system, and establish the operating constraints of the predictive control model.

[0029] Specifically, in S1, a state space model is established based on the operating data of the hydrogen-electric coupling system, and a predictive control model is established by calculating the observed values ​​of the state variables of the system, including: Obtain status data, operating power data and system output data of the hydrogen-electric coupling system, and pre-process the data to obtain target system parameters; Determine state variables, output variables, control variables and disturbance variables based on target system parameters, and establish a discrete time state space model of the hydrogen-electric energy conversion device according to the state variables, output variables, control variables and disturbance variables; The state space model is incrementally iterated based on the control time domain and the prediction time domain to obtain the state variable observation value of the state space model in the prediction time domain; The output variable prediction value is extracted based on the state observation value, and the predictive control model is constructed with the output variable prediction value.

[0030] Specifically, the state space model is as follows: In the formula, x(k), y(k), u(k) and d(k) are the state variable, output variable, control variable and external measurable disturbance variable of the system at time k respectively; A, B u , B d and C c are the correlation matrices of the state-space equation, x(k+1) is the independent variable of the state-space equation, y c (k) is the dependent variable of the state space equation.

[0031] Among them, the variables are: Where P BT(k) is the power of the energy storage device at time k, P EL (k) is the power of the electrolyzer at time k, P FC (k) is the power of the fuel cell at time k, P GRID (k) is the power output of the power grid at time k, SOC(k) is the state of charge of the system at time k, SOH(k) is the health state of the system at time k, ΔP BT (k) is the change in the energy storage device at time k, ΔP EL (k) is the equipment power adjustment amount for converting electrical energy into hydrogen energy by the electrolyzer at time k, ΔP FC (k) is the equipment power adjustment amount for converting hydrogen energy into electrical energy by the fuel cell at time k, ΔP LD (k) is the change in the system load demand at time k, ΔP PV (k) is the change in the photovoltaic power generation at time k, ΔP WT (k) is the change in the wind power generation at time k, T represents the transpose matrix.

[0032] Correlation matrices A, B u 、B d and C c are all sparse matrices, respectively: Among them, T s is the sampling time.

[0033] Furthermore, for the convenience of implementing the predictive control model, the formula (1) of the state space model is converted into an incremental expression: Among them, Δx(k), Δy(k), Δu(k) and Δd(k) are the increments of the state variable, output variable, control variable and disturbance variable at time k, respectively; Set the control time domain as m, the prediction time domain as p, and m < p. Iteratively calculate the incremental formula, that is, formula (4), to obtain the state observation values within the prediction time domain as: In the formula, k+1|k means predicting the value at k+1 at time k, Δx(k) is the increment of the state variable of the system at time k, Δu(k) is the control increment of the system at time k, Δd(k) is the increment of the disturbance variable of the system at time k, M is the length of the control time domain, and L is the Lth moment in the prediction time domain or control time domain; Δx(k+1|k) means predicting the state increment at time k+1 at time k, Δd(k+1|k) means predicting the disturbance increment at time k+1 at time k; Δu(k+1|k) means predicting the state increment at time k The control increment at time k+1, Δx(k+m-1|k) represents the state increment of the m-1th control time domain, Δd(k+m-1|k) represents the disturbance increment of the m-1th control time domain, and Δu(k+m-1|k) represents the control increment of the m-1th control time domain; Δx(k+p-1|k) represents the state increment of the p-1th prediction time domain, Δd(k+p-1|k) represents the disturbance increment of the p-1th prediction time domain, and Δu(k+p-1|k) represents the control increment of the p-1th prediction time domain.

[0034] Then the output variable prediction value from time k+1 to k+p is: Construct the prediction output vectors of p prediction time domains and the control vectors of m control time domains as follows: Y p (k+1|k)=[y c (k+1|k),y c (k+2|k),L,y c (k+p|k)] T (7) ΔU(k)=[Δu(k),Δu(k+1),L,Δu(k+m-1)] T (8) Among them, Y p (k+1|k) is the predicted output vector of the system at time k+i predicted by time k, and ΔU(k) ​​is the control vector of the system at time k.

[0035] Specifically, a predictive control model is constructed based on the predicted value of the output variable, including: Constructing a prediction output vector for the prediction time domain and a control vector for the control time domain; Substitute the predicted output vector and the control vector into the calculation formula of the output variable prediction value for functional integration to obtain the predictive control model; wherein the predictive control model is expressed as follows: Y p (k+1|k)=S x Δx(k)+S y y c (k)+S d Δd(k)+Su ΔU(k)(9) Where Y p (k+1|k) is the predicted output vector of the system at time k predicting time k+i, ΔU(k) ​​is the control vector of the system at time k, S x , S y , the coefficient matrix of the state-space model, Δx(k) is the increment of the state variable of the system at time k, and Δd(k) is the increment of the disturbance variable of the system at time k.

[0036] Among them, the coefficient matrices of the state space model are as follows: Among them, S u The expression presents a lower triangle form, which reflects that the optimal control sequence of the open-loop optimization problem under the current observation value only acts on the current moment and will not affect the control sequence at the next moment, that is, the future state variable observation value in the prediction time domain will not affect the output variable at the current moment.

[0037] Specifically, in S1, the operation constraints of the predictive control model are established, including: Establish corresponding power constraints and energy storage state constraints based on the basic configuration properties of the hydrogen-electric coupling system; Simultaneously, an incremental constraint is established based on the control variable to limit the control range of the power.

[0038] Specifically, the operation constraints are expressed as follows: Where p is the number of prediction time domains, m is the number of control time domains, L is the Lth moment of the prediction time domain or control time domain, and y c (k+i) is the state observation value of each device in the system under the operating state corresponding to the time k+i. The state observation value of each device represents the power and energy storage state of the system at the current moment. c (k+i) min is the minimum state observation value of the system under the operating state at time k+i, y c (k+i) max is the maximum state observation value of the system under the operating state at time k+i, u(k+i) is the control variable value of the system at time k+i, u(k+i) min is the minimum control variable value of the system at time k+i, u(k+i) max is the maximum control variable value of the system at time k+i, Δu(k+i) is the increment of the control variable of the system at time k+i, Δu(k+i) min is the minimum increment of the control variable of the system at time k+i, Δu(k+i) maxis the maximum increment of the control variable of the system at time k+i.

[0039] It is understandable that the hydrogen-electric energy conversion device contains a variety of energy storage devices, so it is a multivariable system. Among them, the output characteristics of hydrogen energy components such as electrolyzers and fuel cells have nonlinear characteristics, and the system parameters change over time, so they can be classified as time-varying systems and nonlinear systems. Since model predictive control has the characteristics of online optimization and processing constraints, it can handle time-varying, nonlinear and multivariable systems with constraints, and has excellent robustness and control performance. Therefore, in the presence of source and load uncertainty and disturbances, it is a very appropriate choice to apply the model predictive control method to solve the real-time scheduling problem of hydrogen-electric coupling systems.

[0040] In this embodiment, by establishing a discrete-time state-space model of the hydrogen-electric energy conversion device, the impact of different empty column parameters on the real-time scheduling plan can be dynamically reflected, which is helpful for the subsequent establishment of the objective function, and the optimization and control of the short-term operation of the microgrid can be achieved based on the power characteristics and mechanisms of each system unit from a time scale of seconds; by establishing operating constraints, the power, energy storage state and control variables of the predictive control model can be constrained, and ultimately the maximum control range of the power can be limited to ensure the stable, safe and reliable operation of the hydrogen-electric coupling system; by constructing a predictive control model, nonlinear control systems and multivariable systems such as the hydrogen-electric coupling system can be efficiently processed.

[0041] S2. Establishing a first objective function based on the model variables of the predictive control model and the intra-day real-time scheduling expectations of the hydrogen-electricity coupling system.

[0042] Specifically, S2 includes the following sub-steps: Determine a first weighting factor of each output variable and a second weighting factor of each control variable based on the intraday real-time dispatch expectation of the hydrogen-electricity coupling system; The predicted output vector and the control vector of the predictive control model are combined with the first weighting factor and the second weighting factor to establish a first objective function.

[0043] Specifically, the first objective function is expressed as follows: Wherein, J represents the first objective function, p represents the prediction time domain of the predictive control model variables, m represents the control time domain of the predictive control model, ΔU(k+i-1) is the control vector of the system at time k+i-1, r(k) represents the reference sequence of the output variables at time k, w is the first weighting factor, and Q is the second weighting factor.

[0044] The first weighting factor w is the weighting factor of each output variable y(k), w=diag(w BT ,wEL ,w FC ,w GRID ,w SOC ,w SOH ); the second weighting factor Q is the weighting factor of each control variable u(k), Q = diag(Q BT ,Q EL ,Q FC ).

[0045] In this embodiment, the intraday real-time scheduling expectation of the hydrogen-electric coupling system is that during the intraday real-time scheduling process of the system, it is expected that the power of each energy storage unit will not deviate too much from the original power plan after power redistribution, and it is also expected that the action changes of each energy storage unit will not be too large; in this way, a first objective function is established according to each variable of the predictive control model to ensure that the final system real-time scheduling plan is more in line with the actual operation requirements of the system, while improving the economic benefits of the system.

[0046] S3, performing a quadratic programming conversion on the first objective function, and updating the converted first objective function by calculating the real-time source-load prediction value through the predictive control model to obtain a second objective function containing a linear relationship.

[0047] Specifically, S3 includes the following sub-steps: The first objective function is converted into a quadratic programming form based on a standard quadratic programming function; Calculate the real-time source load forecast value based on the predictive control model, subtract the real-time source load forecast value from the system day-ahead forecast value, and use the difference as the variable observation value of the predictive control model; The transformed first objective function parameters are updated with the variable observation values ​​to obtain the second objective function.

[0048] Specifically, the variables for transforming the first objective function into a quadratic programming form are defined according to the increments of each variable of the predictive control model: E p (k+1|k)=r(k+1)-S x Δx(k)-S y y c (k)-S d Δd(k)(13) Among them, r(k+1) represents the reference sequence of output variables at time k+1.

[0049] Then the objective function is converted to: Due to E p (k+1|k) T w T w p(k+1|k) is independent of the control variable ΔU(k), so the first objective function is ultimately equivalent to the quadratic programming form of the second objective function as shown below.

[0050] Specifically, the second objective function is expressed as follows: J ％ =ΔU(k) T HΔU(k)-G(k+1|k) T U(k)(15) In the formula, J ％ represents the second objective function, H is the weighting factor coefficient matrix, G(k+1|k) is the function parameter of the second objective function, ΔU(k) ​​is the control vector of the system at time k, ΔU(k) T is the control matrix of the system at time k, U(k) is the control variable of the system at time k, and T represents the transposed matrix.

[0051] Finally, the real-time scheduling strategy problem of the hydrogen-electric coupling system is transformed into the following (Quadratic Programming, QP) problem:

[0052] S4. Iteratively solve the second objective function to obtain the optimal system scheduling solution.

[0053] Specifically, S4 includes the following sub-steps: Iteratively solving the second objective function based on the operation constraint to obtain an objective function value, wherein the objective function value at least includes a control increment of a control variable; The control increment obtained in each round of iteration is applied to the hydrogen-electric coupling system, and the coefficient matrix of the state space model is updated until the set prediction time domain and control time domain are reached. The iteration is stopped, and the objective function value of the last round of iteration is output to obtain the optimal state variable observation value and disturbance observation value, and obtain the optimal system scheduling plan.

[0054] In this embodiment, the control law for solving the optimal control offline is converted into a constraint problem for solving online numerical values, that is, predicting the future behavior of the system based on the observation value at the current moment, which significantly reduces the computational complexity and improves the computational efficiency of the optimal scheduling method. At the same time, the predictive control model can adjust the charging and discharging strategy of the energy storage device in real time according to the current source load situation and system status to ensure that the system operates in a safe, reliable and economical operating state; by converting the first objective function into a quadratic programming function, so as to quickly and accurately find the optimal solution, the function structure of the optimization problem is simplified, and then the calculation process is simplified.

[0055] Example 2, set the sampling time of the predictive control model to 0.1 seconds, and obtain the optimal scheduling solution according to different control variables and corresponding weight factors. First, analyze the tracking performance of different prediction time domains p and control time domains m on the reference trajectory; Figure 3 As shown in the figure, taking the output of the energy storage battery as an example, at the starting moment, when the prediction time domain increases from 5 to 10, the response speed of the system to the initial value is improved. However, in the power change step signal applied at 2 seconds, obvious overshoot and adjustment time delay appear. When the prediction time domain is fixed at 10, increasing the control time domain from 2 to 4 will reduce the system's response performance to the initial value. However, it can be adjusted to the step signal in time, and the overshoot and adjustment time are the smallest among all parameters. Therefore, for model predictive control, only increasing the prediction time domain will lead to an increase in the system's response overshoot and adjustment time to the step signal. To obtain better dynamic response performance, both the prediction and control time domains must be increased at the same time.

[0056] Secondly, the influence of the change of the weight factor in the second objective function on the power regulation of the hydrogen-electric coupling system is analyzed. Figure 4 As shown in the figure, in order to more intuitively compare the weight factors between different energy storage units, all output variables and control variables are standardized. The processed weights are set according to the following adjustment rules: 0.05 is low priority, 0.2 is below average priority, 1 is average priority, 5 is above average priority, and 20 is high priority. When the weights of the control variables: energy storage battery, electrolyzer and fuel cell are all set to 1, when there is a 100kW source load disturbance at 2s, the power of the three has changed by 33.3kW, that is, the power of the source load change is evenly distributed to different energy storage units for power adjustment. When the weight of the energy storage battery is set to 2, due to the low priority of the electrolyzer and fuel cell, their control increments are increased to 44kW. When the weight of the energy storage battery is set to 5, the priority of the energy storage battery has reached a high level, and the system no longer adjusts the power of the energy storage battery. At this time, all source load fluctuations are applied to the electrolyzer and fuel cell. Therefore, by adjusting the weight factor, users can flexibly allocate the power of different energy storage units according to specific application requirements to achieve better power control performance.

[0057] It is understandable that in order to verify the feasibility of the dispatch optimization method, the local source and load data of the Ningbo Cixi Hydrogen-Electricity Coupled DC Microgrid Demonstration Project were analyzed. Figure 5 As shown, it includes annual light intensity data, wind speed data at 40m and load data. The simulation step is 1h, with a total of 8760 sets of measurement data, among which the daily average light intensity is 2.21kWh / m2, the average wind speed is 8.9m / s, and the daily average power load is 180kW.

[0058] Select the coordinated optimization scheduling plan for a certain day in the above scenario as a case (such as Figure 6 As shown in Figure 2, after analyzing the impact of different control parameters and weight factors on system performance, the control parameters with a prediction time domain of 10 and a control time domain of 4 were selected to verify the real-time scheduling of the system. The output of the energy storage unit after real-time adjustment is as follows: Figure 7 As shown in the figure, during the operation of the system, it is expected that the tie line will keep tracking the day-ahead planned value to achieve a smooth and controllable access of the system to the main grid. During the design process, it is expected that the hydrogen energy storage unit will not fluctuate violently, so the weights of the electrolyzer and fuel cell are set to 3, while the weight of the energy storage battery is set to 1, which mainly bears the uncertainty fluctuations of the source and load. In the adjusted intraday real-time scheduling plan, in order to cope with the source and load fluctuations, the output of the energy storage unit has been corrected to a certain extent, and all operate within the set range, verifying the effectiveness of the predictive control model for power adjustment.

[0059] Furthermore, the fluctuation of the contact line after real-time adjustment is analyzed, such as Figure 8 As shown in the figure, if the output of the unit is not adjusted, the violent fluctuation of the source load will directly act on the interconnection line, thereby affecting the controllable dispatch of the main grid to the microgrid and placing a serious burden on the power converter. After MPC optimization, the real-time output adjustment of the energy storage unit makes the interconnection line basically consistent with the planned value, and there is only a short adjustment process even in dynamic changes.

[0060] Furthermore, the influence of weights on the actual operating status and planned values ​​under different energy storage states is compared, such as Fig. 9 As shown in the figure, by comparing the control effects under different weights, it is found that whether it is SOC or SOH, increasing the weight will improve the tracking accuracy of the day-ahead plan value. However, increasing the weight factor will also bring about the problem of short-term state switching of the energy storage unit, such as the frequent charging and discharging process of the energy storage battery, and the continuous start-stop switching of the hydrogen energy storage unit at low power. Therefore, when setting the controller, it is necessary to balance the control accuracy and energy storage life.

[0061] Beneficial effects of the embodiment: By establishing a predictive control model, the charging and discharging strategy of the energy storage device can be adjusted in real time according to the current source load situation and system status to ensure that the system operates in a safe, reliable and economical operating state. In the scheduling calculation process, by simultaneously increasing the prediction and control time domains, it is possible to avoid large overshoots in the system during the dynamic process to obtain better dynamic response performance; by adjusting the weight factors of the energy storage units, the power of different energy storage units can be flexibly allocated according to specific application requirements to achieve better power control performance.

[0062] The above specific embodiments are preferred embodiments of the present invention, and are not intended to limit the specific implementation scope of the present invention. The scope of the present invention includes but is not limited to the present specific embodiments. All equivalent changes made in accordance with the shape, structure, and method of the present invention are within the protection scope of the present invention.

Claims

1. A method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction, characterized in that: The steps include: S1. Establishing a state space model based on the operating data of the hydrogen-electric coupling system, establishing a predictive control model by calculating the observed values ​​of the state variables of the system, and establishing operating constraints of the predictive control model; S2, establishing a first objective function based on the model variables of the predictive control model and the intraday real-time scheduling expectation of the hydrogen-electric coupling system; S3, performing a quadratic programming conversion on the first objective function, and updating the converted first objective function by calculating the real-time source-load prediction value through the predictive control model to obtain a second objective function containing a linear relationship; S4. Iteratively solve the second objective function to obtain an optimal system scheduling solution.

2. The method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction according to claim 1 is characterized in that: In S1, a state space model is established based on the operating data of the hydrogen-electric coupling system, and a predictive control model is established by calculating the observed values ​​of the state variables of the system, including: Obtain status data, operating power data and system output data of the hydrogen-electric coupling system, and pre-process the data to obtain target system parameters; Determining a state variable, an output variable, a control variable, and a disturbance variable based on the target system parameters, and establishing a discrete-time state-space model of a hydrogen-to-electricity energy conversion device according to the state variable, the output variable, the control variable, and the disturbance variable; Iteratively calculating the incremental state space model based on the control time domain and the prediction time domain to obtain the state variable observation value of the state space model in the prediction time domain; The output variable prediction value is extracted based on the state observation value, and the predictive control model is constructed with the output variable prediction value.

3. The method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction according to claim 2 is characterized in that: The state space model is as follows: In the formula, x(k), y(k), u(k) and d(k) are the state variable, output variable, control variable and external measurable disturbance variable of the system at time k respectively; A, B u , B d and C c are the correlation matrices of the state-space equation, x(k+1) is the independent variable of the state-space equation, y c (k) is the dependent variable of the state space equation.

4. The method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction according to claim 3 is characterized in that: The step of constructing the predictive control model based on the output variable predicted value comprises: Constructing a prediction output vector for the prediction time domain and a control vector for the control time domain; Substituting the predicted output vector and the control vector into the calculation formula of the output variable predicted value for functional integration to obtain a predictive control model; Wherein, the predictive control model is expressed as follows: Y p (k+1|k)=S x Δx(k)+S y y c (k)+S d Δd(k)+S u ΔU(k) Where Y p (k+1|k) is the predicted output vector of the system at time k predicting time k+i, ΔU(k) ​​is the control vector of the system at time k, S x , S y , S u , S d are the coefficient matrices of the state-space model respectively, Δx(k) is the state variable increment of the system at time k, and Δd(k) is the disturbance variable increment of the system at time k.

5. The method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction according to claim 2 is characterized in that: In S1, establishing the operation constraints of the predictive control model includes: Establish corresponding power constraints and energy storage state constraints based on the basic configuration properties of the hydrogen-electric coupling system; Simultaneously, an increment constraint is established based on the control variable to limit the control range of power.

6. The method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction according to claim 2 is characterized in that: The S2 comprises the following sub-steps: Determining a first weighting factor of each of the output variables and a second weighting factor of each of the control variables based on the intra-day real-time scheduling expectation of the hydrogen-electricity coupling system; The predicted output vector and the control vector of the predictive control model are combined with the first weighting factor and the second weighting factor to establish the first objective function.

7. The method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction according to claim 6 is characterized in that: The first objective function is expressed as follows: Wherein, J represents the first objective function, p represents the prediction time domain of the predictive control model variables, m represents the control time domain of the predictive control model, ΔU(k+i-1) is the control vector of the system at time k+i-1, r(k) represents the reference sequence of the output variables at time k, w is the first weighting factor, and Q is the second weighting factor.

8. The method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction according to claim 7 is characterized in that: The S3 comprises the following sub-steps: Converting the first objective function into a quadratic programming form based on a standard quadratic programming function; Calculating a real-time source load prediction value based on the prediction control model, subtracting the real-time source load prediction value from the system day-ahead prediction value, and using the difference as a variable observation value of the prediction control model; The transformed first objective function parameters are updated with the variable observation values ​​to obtain the second objective function.

9. The method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction according to claim 8 is characterized in that: The second objective function is expressed as follows: J ％ =ΔU(k) T HΔU(k)-G(k+1|k) T U(k); In the formula, J ％ represents the second objective function, H is the weighting factor coefficient matrix, G(k+1|k) is the function parameter of the second objective function, ΔU(k) ​​is the control vector of the system at time k, ΔU(k) T is the control matrix of the system at time k, U(k) is the control variable of the system at time k, and T represents the transposed matrix.

10. The method for optimizing the scheduling of a hydrogen-electricity coupling system based on source-load prediction according to claim 8, characterized in that: The S4 comprises the following sub-steps: Iteratively solving the second objective function based on the operation constraint to obtain an objective function value, wherein the objective function value at least includes a control increment of a control variable; The control increment obtained in each round of iteration is applied to the hydrogen-electric coupling system, and the coefficient matrix of the state space model is updated until the set prediction time domain and control time domain are reached. The iteration is stopped, and the objective function value of the last round of iteration is output to obtain the optimal state variable observation value and disturbance observation value, and obtain the optimal system scheduling plan.

Images