Electro-hydrogen fusion time sequence production simulation method

Abstract

The invention provides an electricity-hydrogen fusion time sequence production simulation method, which comprises the following steps of: constructing an energy supply framework model of a new energy station, decomposing an original hydrogen load sequence through a variational mode decomposition method, and establishing an electricity-hydrogen fusion operation model of which the target is operation cost through the energy supply framework model, sparse dimensionality reduction and linear programming solving are carried out on the time sequence production simulation result, and a conjugate gradient asynchronous parallel computing method is used for solving and outputting the time sequence production simulation result. According to the method, an electricity-hydrogen fusion time sequence production simulation model of a mixed time scale is constructed, and the self-adaptive step length of a low-frequency component planned hydrogen separation energy storage scheduling period solved by decomposing a hydrogen load is utilized, so that the stability of external output power of a long-time-scale upper field station is guaranteed through a synergistic effect of the self-adaptive step length and real-time balanced electric energy; the operation economy is improved; and then aiming at electro-hydrogen large-scale fusion matrix sparse dimensionality reduction, parallel calculation of a conjugate gradient method is utilized, so that the solving speed is increased, and the accuracy of a simulation result can be ensured.

Description

Technical Field

[0001] The present invention relates to the field of new energy technology, and in particular to a time-sequential production simulation method for electric-hydrogen fusion. Background Art

[0002] The new energy station that integrates electricity and hydrogen does not only use hydrogen production electrolyzers and hydrogen fuel cells to achieve the coupling between electricity and hydrogen energy. Its deeper connotation is to achieve the flexible interactive conversion and synergistic complementary storage of electricity and hydrogen through the rational operation of the electric-hydrogen multi-energy storage system composed of electrolyzers, fuel cells, hydrogen storage devices and electrochemical energy storage, and effectively solve the problem of unstable energy supply of wind and solar power generation in new energy stations over a long period of time.

[0003] At present, the existing technology is to study the optimized operation of new energy stations from the aspects of technical and economic feasibility analysis, reducing carbon emissions and improving new energy consumption. New energy stations can achieve energy transfer through long-term hydrogen energy storage to solve the problem of long-term power shortages ranging from several days to several seasons. However, hydrogen storage tanks are limited by high costs caused by frequent scheduling and fluctuations in hydrogen production. Short-term energy storage, such as electrochemical energy storage, is good at alleviating short-term power fluctuations, but is not suitable for long-term energy transmission. The integration of electricity and hydrogen in new energy stations combines the advantages of long-term and short-term energy storage systems to solve the mismatch between actual output and power generation plans, and hydrogen energy storage does not require real-time balancing.

[0004] Existing time-series production simulation studies simulate electricity and hydrogen on the same time scale, which cannot fully reflect the operating characteristics of hydrogen as a long-term energy storage solution. Furthermore, although the introduction of hydrogen energy has expanded the simulation of single electricity production to the simulation of the joint production of multiple energy flows of electricity and hydrogen, the energy flow matrix of the fusion of electricity and hydrogen is often a large sparse matrix, most of which are large band matrices with narrow bandwidth. A band matrix refers to a matrix in which only the elements near the diagonal can be non-zero, while the elements at other positions are all zero. The characteristic of this matrix structure is that it has fewer non-zero elements, so it takes up fewer resources when storing and calculating, which makes it have great advantages in solving practical problems. Although the storage and calculation required to solve the band matrix linear equation system are smaller than those of the general matrix linear equation system, when solving the band matrix linear equation system, the current mainstream direct solution methods, such as the LU decomposition algorithm and the QR decomposition algorithm, are often inefficient due to the need to deal with a large number of zero elements. Furthermore, the time span of the time-series production simulation is relatively long, such as the annual operating time of 8760h, which makes the problem difficult to solve. Therefore, the existing literature on the solution techniques for the time dimension of production simulation is mostly divided into sequential solution and rolling solution. Sequential solution refers to splitting the production simulation cycle into multiple shorter time segments in the time dimension, and then sequentially solving the production simulation problems of each sub-segment. Among them, the operating state of the last time period of the previous sub-segment is used as the initial operating state of the next sub-segment, thereby satisfying the coupling constraints between the sub-problems and forming a complete and feasible production simulation result. Rolling solution means that when calculating the optimization problem of a certain time segment, several subsequent time steps can be appropriately taken into account, but in order to meet the coupling constraints between the time segments, the rolling solution requires rolling calculations from time to time, and its serial calculation structure limits the calculation efficiency. In addition, there is also the choice of directly studying the typical curves within the time period or reducing the dimension through time series clustering, instead of a complete medium- and long-term simulation.

[0005] Therefore, at the existing technical level, there are several main problems: 1. The operation model of electric hydrogen is simple: Most existing studies model and solve the heterogeneous energy flows of electric hydrogen on the same time scale, which cannot reflect the advantages of long-term storage and scheduling of hydrogen energy. Unlike the real-time balanced electric load, the hydrogen load does not need to meet the hourly balance, and the research on the mixed time scale operation of new energy stations based on time-series production simulation is still insufficient; 2. The electric-hydrogen fusion matrix model is complex: the introduction of hydrogen energy expands the single power production simulation to the electric-hydrogen multi-energy flow joint production simulation. There is a technical gap in the research on how to perform sparse transformation and dimensionality reduction on the electric-hydrogen fusion energy flow matrix; 3. There is a contradiction between computational efficiency and accuracy: Due to the strong timing characteristics of renewable energy generation, the commonly used rolling solution and the selection of typical scenarios for production simulation methods often introduce significant errors in the results.

[0006] In summary, these problems put forward higher requirements on the electric-hydrogen fusion model and its solution algorithm of new energy stations, and it is also necessary to optimize and improve the simulation method of electric-hydrogen fusion scenarios of new energy stations. Summary of the invention

[0007] In view of the problems existing in the prior art, the present invention provides a time-series-based electric-hydrogen fusion new energy station simulation method, which effectively improves the speed of the algorithm while ensuring the solution accuracy.

[0008] To achieve the above purpose, the technical solution adopted by the present invention is as follows: The present invention provides a method for simulating the sequential production of electric hydrogen fusion, which comprises the following steps: Construct an energy supply framework model for new energy stations based on electric-hydrogen fusion technology; Collecting hydrogen load data of the new energy station, and establishing an original hydrogen load sequence based on time series through the hydrogen load data; The original hydrogen load sequence is decomposed by the variational mode decomposition method, and the formula is: ; In the formula, is the original hydrogen loading sequence, k is the mode number, with a value of 1- K ; t is the time index, subscript h It stands for Hydrogen Logo; is the set of hydrogen load subsequences after variational mode decomposition, is the center frequency of the subsequence component, is the gradient calculation, is the impulse function, Compute symbols for convolution; After decomposition, a hydrogen load subsequence set including a high-frequency component subsequence and a low-frequency component subsequence is obtained; An adaptive step size for dividing the hydrogen energy storage scheduling period of the new energy station based on the low-frequency component subsequence; An electric-hydrogen fusion operation model with an operation cost as a target is established through the energy supply framework model, and the hydrogen load subsequence set and the adaptive step size of the hydrogen energy storage scheduling period of the new energy station are input to obtain an electric-hydrogen fusion matrix of the electric-hydrogen fusion operation model; Simplifying the electric-hydrogen fusion matrix by a sparse dimensionality reduction method and obtaining a simplified matrix equation group; The simplified matrix equation group is linearized by a linear programming solution method, and the linearized simplified matrix equation group is solved by a conjugate gradient asynchronous parallel computing method to output a sequential production simulation result.

[0009] Optionally, the original hydrogen load sequence is decomposed by a variational mode decomposition method, and the formula is: ; In the formula, is the original hydrogen loading sequence, k is the mode number, with a value of 1- K ; t is the time index, subscript h It stands for Hydrogen Logo; is the set of hydrogen load subsequences after variational mode decomposition, is the center frequency of the subsequence component, is the gradient calculation, is the impulse function, Computes the sign for the convolution.

[0010] Optionally, the energy supply framework model of the new energy station includes renewable energy power generation equipment, hydrogen-to-electricity conversion equipment, electrical energy storage equipment and hydrogen storage equipment.

[0011] Optionally, the objective function of the electric-hydrogen fusion operation model is: ; In the formula, is the annual comprehensive cost of the new energy station, For operation and maintenance costs, Penalty costs for renewable energy generation cuts, Penalty costs for insufficient hydrogen supply, is the lifetime degradation cost of the electrolyzer stack, is the lifetime degradation cost of the hydrogen fuel cell stack, It is the lifetime degradation cost of the electric energy storage equipment.

[0012] Optionally, the objective function of the electric-hydrogen fusion operation model further includes: ; In the formula, To be friendly to netizens, is the output power smoothness of the new energy station, It is the alignment between the output of renewable energy sites and the planned power generation.

[0013] Optionally, constraining conditions are constructed based on the objective function, and the constraints and the objective function are combined to form the electric-hydrogen fusion operation model; The constraints include hydrogen storage constraints, electricity storage constraints, capacity constraints, renewable energy constraints, operation constraints and reliability constraints; The renewable energy constraint is a limited reduction amount of wind energy and solar energy; The reliability constraint includes a ratio range of the output power of the new energy station to the required power.

[0014] Optionally, the electric-hydrogen fusion matrix is ​​a discrete state space matrix; The sparse dimensionality reduction method simplification includes the following steps: The linearized electric-hydrogen fusion matrix is ​​transformed into a band matrix by linear transformation, and the band matrix is ​​divided into block tridiagonal matrices; The block tridiagonal matrix is ​​decomposed by a matrix decomposition algorithm, and the simplified matrix equation group is extracted from the decomposition result.

[0015] Optionally, the matrix decomposition algorithm includes a recursive banded angularization matrix decomposition algorithm.

[0016] Optionally, the linear programming solution method is the big M method, and the electric-hydrogen fusion matrix of the electric-hydrogen fusion operation model is linearized by introducing a variable method.

[0017] Optionally, the conjugate gradient asynchronous parallel computing method comprises the following steps: Dividing the time series data of a preset time into a plurality of continuous sub-periods, and each of the sub-periods includes a preset coupling period; Initializing the residual vector, search direction vector and penalty factor of each sub-period; Perform conjugate gradient iterations in parallel for each initialized sub-period, calculate the current step size and update the state variables; Synchronizing device states of two adjacent sub-periods in the coupling period; Adaptively adjust the penalty factor based on the residual norm and update the search direction vector; The residual norms of all sub-periods are iteratively calculated through an iterative method until they reach the preset convergence tolerance.

[0018] Optionally, the step of adaptively adjusting the penalty factor based on the residual norm comprises the following steps: The penalty factor is dynamically updated according to the residual norm ratio of two adjacent iterations.

[0019] Compared with the prior art, the present invention has the following beneficial effects: The present invention constructs a hybrid time-scale electric-hydrogen fusion sequential production simulation model, and uses the low-frequency component obtained by decomposing the hydrogen load to plan the adaptive step size of the hydrogen energy storage scheduling period, so that it can work synergistically with the real-time balanced electric energy to ensure the stability of the station's external output power on a long time scale and improve the operating economy; then, based on the full-year sequential production simulation, a sparse dimensionality reduction method for the large-scale electric-hydrogen fusion matrix is ​​proposed, which is beneficial to reducing the difficulty of solving the model; finally, a parallel computing method based on the conjugate gradient method is proposed, which decomposes a long period into multiple sub-periods for distributed parallel solution, which not only speeds up the solution speed but also ensures the accuracy of the simulation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0021] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0024] It is worth noting that the methods used in the present invention are all conventional methods unless otherwise specified; the raw materials and devices used are all conventional commercially available products, and their sources are not specifically limited unless otherwise specified.

[0025] It should also be noted that in order to facilitate understanding in the specific embodiments of the present invention, the method steps are described in a certain order, but those skilled in the art can change the order of the steps according to actual needs, and therefore this cannot be used as a restriction. It should be further noted that in the description of the following specific embodiments, the min and max in the superscripts and subscripts of each parameter should be understood as representing the minimum and maximum values ​​of the parameter corresponding to the time or a period of time unless otherwise specified.

[0026] like Figure 1 As shown, this embodiment provides a method for simulating the sequential production of electric hydrogen fusion, comprising the following steps: 1. Construct an energy supply framework model for new energy stations based on electric-hydrogen fusion technology; First, the operation framework of the new energy station is constructed according to the system composition and connection mode of the new energy station. Renewable energy generation is integrated in the system of the new energy station of this embodiment, while utilizing the actual conversion and storage between electricity and hydropower energy carriers to meet the power load and hydropower demand of the chemical industry or transportation sector. Specifically, at the input source port, renewable energy such as wind and solar energy are collected and regulated by wind turbines (WTs) and photovoltaic panels (PV) to generate electricity. This electricity powers the water electrolysis process within the electrolyzer stack to produce green hydrogen. Given the limited hydrogen production capacity of a single electrolyzer, a stack of multiple electrolyzers in series or parallel is assembled to meet the required production capacity. The generated hydrogen is pressurized by a compressor and stored in a hydrogen storage tank (HST) for long-term use. During periods of renewable energy shortage, such as no wind or cloudy days, it can be converted back to electricity through hydrogen FC. During system operation, due to the relatively low efficiency of the electricity-hydrogen-electricity conversion path, short-term power fluctuations are mainly managed by BES rather than FC. This integration of electricity-hydrogen coupling and conversion paths increases operational flexibility and ensures a reliable multi-energy supply for renewable energy power plants.

[0027] Collect hydrogen load data from new energy stations, and use the hydrogen load data to establish a time-series-based original hydrogen load sequence, and then decompose the original hydrogen load sequence using the variational mode decomposition method. Among them, variational mode decomposition (VMD) is an adaptive, completely non-recursive modal variation and signal processing method that decomposes complex signals into multiple harmonic signals by performing frequency domain analysis on non-stationary signals. The overall framework of variational mode decomposition is a variational problem that minimizes the sum of the estimated bandwidths of each mode, where it is assumed that each mode is a finite bandwidth with a different center frequency. Use VMD to decompose the original hydrogen load sequence. After decomposition, we can get the high-frequency component subsequence and low frequency component subsequence Among them, the high-frequency component can capture the high-frequency noise or rapid changes in the hydrogen load data; the low-frequency component can retain the trend of the data and reduce the noise interference of individual data points. The adaptive step size of the hydrogen storage scheduling period of the new energy station can be divided based on the low-frequency component subsequence, and the synergy with the real-time balanced electric energy can ensure the stability of the station's external output power on a long time scale and improve the operation economy. The one-sided spectrum of the subsequence is obtained by Hilbert transform and compared with the operator Multiply them and adjust the center band of the subsequence to the corresponding baseband. Use the norm of the demodulation gradient to estimate the bandwidth, and adaptively divide the adjustment period step of the subsequence according to the different bandwidths, as follows: (1) In the formula, is the set of hydrogen load subsequences after VMD decomposition, is the low-frequency component subsequence, is the high frequency component subsequence; is the center frequency of the subsequence component, is the low frequency center frequency, is the high frequency center frequency; is the gradient calculation, is the impulse function, Computes the sign for the convolution.

[0028] 2. Establish an electric-hydrogen fusion operation model with the goal of balancing operation costs and grid connection friendliness through the energy supply framework model; The renewable energy power station (REPS) electric-hydrogen fusion operation model based on full-year time-series production simulation aims to minimize the annual comprehensive cost of the renewable energy power station. , and improve its friendliness to netizens , specifically as follows, (2) (3) In the formula, is the annual comprehensive cost of the new energy station, For operation and maintenance costs, Penalty costs for renewable energy generation cuts, Penalty costs for insufficient hydrogen supply, is the lifetime degradation cost of the electrolyzer stack, is the lifetime degradation cost of the hydrogen fuel cell stack, The lifetime degradation cost of the electric energy storage equipment; It consists of two key aspects: REPS output power smoothness and the degree of alignment between REPS output and REPS planned generation In order to improve the goodness of the network, it is necessary to reduce the power fluctuation and power deviation. Therefore, the smaller the calculated value of the goodness of the network in the objective function, the better, so the minimum value is also taken.

[0029] Specifically, Including the operation and maintenance costs of battery energy storage (BES) and hydrogen energy storage (HES), the formula is as follows: (4) (5) (6) In the formula, and represent the operation and maintenance costs of BES and HES respectively; / / / is the operation and maintenance cost coefficient of BES / HST / electrolyzer / hydrogen fuel cell (FC); is the rated capacity of BES; / / / Represents the investment cost per unit capacity of BES / HST / electrolyzer / FC; Indicates the rated capacity of the HST; are the rated powers of the electrolyzer stack and FC respectively.

[0030] If the electricity generated by wind and solar energy cannot be fully consumed, it will lead to energy waste. Therefore, penalty measures are taken for the curtailment of wind and solar energy. The penalty cost of such energy curtailment can be calculated as follows: (7) In the formula, is the penalty cost per unit of renewable energy forgone; Indicates at time t The amount of wind and solar power that is discarded by REPS.

[0031] If the REPS cannot fully meet the hydrogen load demand, it needs to be reduced, and the penalty cost of this energy shortfall can be expressed as follows: (8) In the formula, is the penalty cost per unit of hydrogen load reduction; Represents REPS at time t The reduction in hydrogen loading.

[0032] Determined by two factors: peak-to-valley difference and the standard deviation of the output power curve , as shown below: (9) (10) (11) (12) (13) In the formula, For the field station time t Corresponding time period T The average output power value within For time t Corresponding time period T The output power within For time t Corresponding time period T Planned power generation within.

[0033] 2.1 Add constraints; Constraints are constructed based on the objective function, and the constraints and the objective function are combined into an electric-hydrogen fusion operation model; among them, the constraints include hydrogen storage constraints, electricity storage constraints, capacity constraints, renewable energy constraints, operation constraints and reliability constraints.

[0034] 2.1.1 Hydrogen storage constraints: The general hydrogen storage system includes hydrogen production, storage and utilization stages. The operation model of the electrolyzer stack, HST, FC and electric compressor, taking into account the electricity-hydrogen conversion, can be expressed as follows: (14) (15) (16) (17) (18) (19) (20) (twenty one) (twenty two) (twenty three) (twenty four) (25) In the formula, It's in time t The electrolytic power of the electrolyzer, It's in time t The electrolysis power of the electrolyzer stack, is the number of electrolytic cells in the electrolytic cell stack; Indicates the hydrogen storage tank (HST) at time t State of energy (SOE); is the internal pressure of HST, For in time t The internal pressure of the HST; is the rated capacity of the HST; represents the molecular weight of hydrogen; and is a binary variable, representing HST at time t The charging and discharging states of the HST are shown in Figure 1. Specifically, 1 indicates that the HST is in the charging or discharging state, and 0 indicates that the HST is not in the working state. The bandwidth is estimated by the norm of the demodulation gradient, and the low-frequency component sequence is adaptively divided according to the bandwidth. The adjustment period step. and are binary variables, representing the hydrogen storage and release behaviors of HST; The status of the scheduling cycle Consistency; represents the unbalanced state during the dth dispatch period, where negative values ​​indicate power shortages and HST is used for energy consumption; D represents a set of dispatch periods. Equations (22) and (23) limit HST to only one state within the same dispatch period. Equation (24) describes the power required to compress hydrogen into HST. Equation (25) shows that the outlet pressure of the compressor is limited to the safe operating boundary and Usually, the inlet and outlet pressures of the compressor are set manually, respectively. and ; / is the hydrogen storage / release rate of HST; is the operating efficiency of the compressor; is the compression power of the compressor; is the specific heat ratio; is the compressor inlet temperature; is the specific heat capacity of hydrogen; / is the hydrogen storage / output efficiency of HST.

[0035] 2.1.2 Storage Constraints: Considering the state of charge (SOC) and charge and discharge rates, the storage dynamics of BES can be formulated as follows: (26) (27) (28) (29) (30) (31) In the formula, Indicates that BES is at time t energy state; is the self-discharge rate of BES; , Θ and Γ are fitting parameters based on experimental data; and is a binary variable, indicating the working status of BES; / is the charging / discharging power of BES; / is the charge / discharge efficiency of BES;.

[0036] 2.1.3 Renewable energy constraints: Renewable energy constraints are the limited reduction of wind and solar energy. The reduction of wind and solar energy should comply with the following constraints: (32) In the formula, In the range of [0,1], it means wind energy and solar energy The curtailment factor of power generation.

[0037] 2.1.4 Operation constraints: Both the electrolyzer and FC are key electricity-hydrogen coupling devices in REPS. Their input / output operating power cannot exceed their rated power and the two devices cannot be in operation at the same time. (33) (34) (35) In the formula, and are binary variables representing time t The working status of the electrolyzer stack and FC. Specifically, 0 means the device is in shutdown state, and 1 means the device is in operation state; are the rated powers of the electrolyzer stack and FC, respectively; is the output electrical power of FC.

[0038] 2.1.5 Reliability constraints: Based on the energy supply reliability requirements of renewable energy power plants, including the ratio range of the output power of new energy stations to the required power, the hydrogen load reduction in each time period should be limited as follows: (36) (37) In the formula, The reliability coefficient of hydrogen supply is expressed in the range of [0,1].

[0039] Step 2: Electric-hydrogen fusion matrix modeling and sparse dimension reduction method 2.2 Modeling of the electric-hydrogen fusion matrix of new energy stations; The hydrogen load subsequence set and the adaptive step size of the hydrogen energy storage scheduling period of the new energy station are input to obtain the electric-hydrogen fusion matrix of the electric-hydrogen fusion operation model.

[0040] First, a discrete state space matrix is ​​constructed M , which encapsulates the interconnected topology and energy conversion paths. M The incident matrix from the input port to the branch X , output port to branch incident matrix Y and the internal energy conversion matrix Z Composition. Z middle, is the hydrogen-to-electricity conversion efficiency of the FC. In addition, the state vector V t represents the steady-state energy input and output of the REPS device in each operating cycle. V t middle / It's in time t Output power of WTs / PV at time ; and It's time t Input electrolysis power and hydrogen production rate of the electrolyzer stack at 1 : is the input hydrogen rate of FC; is the hydrogen of REPS. Therefore, the input-output vector L t Meet REPS requirements for wind and solar power input ports, as well as output power and hydrogen load The output port requirements of the above matrix are simplified as follows: (38) (39) (40) (41) (42) (43) 3. Solve the model; 3.1 Model preprocessing, linearize the electric-hydrogen fusion matrix of the electric-hydrogen fusion operation model through the linear programming solution method, and obtain the linearized electric-hydrogen fusion matrix. Among them, the linear programming solution method is the big M method, and the electric-hydrogen fusion matrix of the electric-hydrogen fusion operation model is linearized by introducing the variable method.

[0041] Specifically, the REPS electric-hydrogen energy storage coordinated optimization configuration model is classified as a mixed integer nonlinear programming problem. The nonlinearity mainly comes from equations (15), (19), (20), (26), (29), (30), (38) and (39), which significantly increase the complexity of solving the configuration model. Specifically, the coupling between the planning and operating variables introduces nonlinearity in equations (15) and (26). In order to achieve linearization, two variables are introduced and To reformulate equations (15) and (26), the two variables represent the mass of hydrogen stored in the HST and the electricity stored in the BES at time t. In addition, the nonlinearity in equations (19), (20), (29), (30), (38), and (39) is due to the interaction between the planning variables and the operational binary variables. To address this issue, the large M method is applied, introducing a sufficiently large positive number M to manage the nonlinearity, as shown below: (44) (45) (46) (47) (48) (49) (50) (51) (52) (53) (54) (55) (56) (57) (58) (59) 3.2 Model dimensionality reduction processing, the linearized electric-hydrogen fusion matrix is ​​simplified by sparse dimensionality reduction method, and a simplified matrix equation group is obtained.

[0042] Specifically, the linearized electric-hydrogen fusion matrix is ​​transformed into a banded matrix through a linear transformation, and the banded matrix is ​​divided into block tridiagonal matrices; the discrete state space matrix is ​​decomposed in block form, and a recursive banded angular matrix decomposition algorithm (DS decomposition algorithm) is introduced to simplify the decomposed tridiagonal matrix region, and the simplified matrix equations are extracted for iterative solution, which helps to reduce the difficulty of solving the model. Among them, the banded matrix is ​​a special type of matrix whose non-zero elements are concentrated in the banded region near the main diagonal. For the decomposition of the banded matrix, the recursive banded angular matrix decomposition algorithm is a method of angularly decomposing the banded matrix using a recursive strategy. Based on the matrix sparse dimensionality reduction method of DS decomposition, the discrete state space matrix established above is M It can be observed that it can be transformed into a band matrix through the linear transformation of the determinant, which can be divided into a block tridiagonal matrix. Assuming that the number of diagonal blocks on the partition diagonal is p, then each band diagonal block The order of ( j=1,…,p ) is (or roughly n / p). For the partitions that have already been divided, ( j = 1,…,p-1 ) and ( j = 2,…,p ) is a block matrix on the sub-diagonal coupled to the diagonal blocks on the diagonal, specifically: (60) In the formula, and The order of , , and the matrix and Contains a large number of zero elements, as shown below: (61) In the formula, and is a k-order matrix.

[0043] Based on the above block mode, the discrete state space matrix M Perform DS decomposition, the specific decomposition is as follows: (62) In the formula, Represents a matrix of order The identity matrix and ,and and The columns where the non-zero elements of form a The matrix multiplication of the above DS decomposition can be obtained as and The expression is as follows (63) and It can be obtained by solving the matrix equation, (64) In the formula, and The state vectors are V t and the input and output vectors L t The jth partition; and They are and Sub-matrix.

[0044] The block tridiagonal matrix is ​​decomposed by a matrix decomposition algorithm, and a simplified matrix equation system is extracted from the decomposition result. Specifically, in the extraction of the simplified matrix equation system, the discrete state space matrix M After DS decomposition, it can be reduced to two steps to solve equation (63), namely: and . Solution The vector obtained by the linear equation system in Will be used as a solution The right-hand side vector of the linear equation system. When solving a linear system of equations, we use the matrix The special diagonal block form of It can be transformed into a linear equation system with p smaller coefficient matrices to solve, which will greatly reduce the complexity of the calculation.

[0045] Will solve the linear system of equations Further simplification is to solve a linear equation system with a smaller coefficient matrix order, as follows, (65) In the formula, By matrix Each block consists of k rows directly above and below it.

[0046] In fact, the spike matrix and It can also be divided into: (66) (67) In the formula, , , and , , They are and The top k rows, the middle Similar to the bottom k rows, corresponding to and have: (68) from The simplified linear equation system (65) is extracted from , , and The top and bottom k rows are the coefficient matrices of this simplified linear system of equations. Contains p-1 diagonal blocks, and its i-th diagonal block is expressed as , the corresponding sub-diagonal block is expressed as , The state vector block and the right-hand side vector block in the reduced linear equation system are expressed as .

[0047] 3.3 Model solution: Solve the simplified matrix equations based on the conjugate gradient asynchronous parallel computing method and output the time series production simulation results.

[0048] The conjugate gradient asynchronous parallel computing method includes the following steps: Dividing the time series data of a preset time into a plurality of continuous sub-periods, and each sub-period includes a preset coupling period; Specifically, a parallel computing algorithm of the conjugate gradient method is proposed based on the parallel computing method of the conjugate gradient method, which only requires limited information interaction. The entire period of the time series production simulation is divided into multiple smaller continuous sub-periods, and the concept of associated periods is introduced. The sub-period problems after solution do not lose the time series. The iteration direction of the variables SOC and SOE is taken as the conjugate gradient direction for iteration, and the penalty factor with dynamic adaptive adjustment is used to accelerate the iterative interaction to reach a consensus, and the status of the electric energy storage SOC and the hydrogen storage tank SOE is obtained.

[0049] To decompose the time-series production problem into multiple smaller continuous sub-period problems, we first introduce auxiliary variables related to the electric energy storage SOC and hydrogen energy storage SOE. , , and To decouple the original SOC and SOE values ​​between the sub-periods i and j of the new energy station, the consistency constraint during the day can be reshaped as: (69) In the formula, is a collection of time series; is the set of coupling time points.

[0050] In order to ensure the timing between sub-periods i and j of the new energy station, the local optimization operation objective function can be rewritten as (70) in, and They are the auxiliary vectors of SOC and SOE of the new energy station between sub-periods i and j; are the penalty factors associated with the consistency constraints of electrical energy storage and hydrogen storage at the kth decoupling point, respectively.

[0051] exist In the +1 iteration, the new energy station uses The obtained and , select a set of search directions = , = -A , where A is a positive definite matrix, and is a conjugate vector group about A. If the direction Already carried out Next one-dimensional search, we find , the next step is to determine Direction can Faster with Reach consensus in After determination, the descent algorithm of equations (71) and (72) is used to obtain .

[0052] Thus, the steps also include initializing the residual vector, search direction vector and penalty factor of each sub-period; Perform conjugate gradient iterations in parallel for each initialized sub-period, calculate the current step size and update the state variables; Synchronizing the device states of two adjacent sub-periods in a coupling period; Adaptively adjust the penalty factor based on the residual norm and update the search direction vector; optionally, dynamically update the penalty factor based on the residual norm ratio of two adjacent iterations; The residual norms of all sub-periods are iteratively calculated through an iterative method until they reach the preset convergence tolerance.

[0053] The specific iteration process is as follows: Initialization: Electric hydrogen energy storage auxiliary vector for new energy stations , , calculate the initial residual = -A , convergence deviation value , number of iterations =1 Iteration: Calculate the search direction: Use the formula Calculate the search direction coefficient , using the formula Calculate the search direction ; Calculate step length: Use the formula calculate; renew : ; Update residual : ; Convergence judgment: judge whether it satisfies ; Update iterations: = + 1; Output: , , ; renew and ; (71) (72) (73) (74) (75) In the formula, when ( , A ) = 0, ,and( , ) = ( , )=0, that is .because = are mutually orthogonal, so in There is at least one zero vector in + =0, which means convergence is achieved.

[0054] During the iteration process, the asynchronous parallel computing method based on conjugate gradients can overcome the problem of unstable fluctuations in the search direction when it is close to convergence. The orthogonality between conjugate search directions can accelerate the convergence speed of the iteration. While using the coupling period to ensure the operation timing, parallel computing can be used to speed up the solution speed.

[0055] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.

Claims

1. A method for simulating the sequential production of electric hydrogen fusion, characterized in that: The steps include: Construct an energy supply framework model for new energy stations based on electric-hydrogen fusion technology; Collecting hydrogen load data of the new energy station, and establishing an original hydrogen load sequence based on time series through the hydrogen load data; The original hydrogen load sequence is decomposed by the variational mode decomposition method, and the formula is: ； In the formula, is the original hydrogen loading sequence, k is the mode number, with a value of 1- K ; t is the time index, subscript h It stands for Hydrogen Logo; is the set of hydrogen load subsequences after variational mode decomposition, is the center frequency of the subsequence component, is the gradient calculation, is the impulse function, Compute symbols for convolution; After decomposition, a hydrogen load subsequence set including a high-frequency component subsequence and a low-frequency component subsequence is obtained; An adaptive step size for dividing the hydrogen energy storage scheduling period of the new energy station based on the low-frequency component subsequence; An electric-hydrogen fusion operation model with an operation cost as a target is established through the energy supply framework model, and the hydrogen load subsequence set and the adaptive step size of the hydrogen energy storage scheduling period of the new energy station are input to obtain an electric-hydrogen fusion matrix of the electric-hydrogen fusion operation model; Simplifying the electric-hydrogen fusion matrix by a sparse dimensionality reduction method and obtaining a simplified matrix equation group; The simplified matrix equation group is linearized by a linear programming solution method, and the linearized simplified matrix equation group is solved by a conjugate gradient asynchronous parallel computing method to output a sequential production simulation result.

2. The method for simulating the sequential production of electric hydrogen fusion according to claim 1, characterized in that: The energy supply framework model of the new energy station includes renewable energy power generation equipment, hydrogen-to-electricity conversion equipment, electrical energy storage equipment and hydrogen storage equipment.

3. The method for simulating the sequential production of electric hydrogen fusion according to claim 2, characterized in that: The objective function of the electric-hydrogen fusion operation model is: ； In the formula, is the annual comprehensive cost of the new energy station, For operation and maintenance costs, Penalty costs for renewable energy generation cuts, Penalty costs for insufficient hydrogen supply, is the lifetime degradation cost of the electrolyzer stack, is the lifetime degradation cost of the hydrogen fuel cell stack, It is the lifetime degradation cost of the electric energy storage equipment.

4. The method for simulating the sequential production of electric hydrogen fusion according to claim 3, characterized in that: The objective function of the electric-hydrogen fusion operation model also includes: ； In the formula, To be friendly to netizens, is the output power smoothness of the new energy station, It is the alignment between the output of renewable energy sites and the planned power generation.

5. The method for simulating the sequential production of electric hydrogen fusion according to claim 4, characterized in that: Constructing constraint conditions based on the objective function, and combining the constraint conditions and the objective function to form the electric-hydrogen fusion operation model; The constraints include hydrogen storage constraints, electricity storage constraints, capacity constraints, renewable energy constraints, operation constraints and reliability constraints; The renewable energy constraint is a limited reduction amount of wind energy and solar energy; The reliability constraint includes a ratio range of the output power of the new energy station to the required power.

6. The method for simulating the sequential production of electric hydrogen fusion according to claim 1, characterized in that: The electric-hydrogen fusion matrix is ​​a discrete state space matrix; The sparse dimensionality reduction method simplification includes the following steps: The linearized electric-hydrogen fusion matrix is ​​transformed into a band matrix by linear transformation, and the band matrix is ​​divided into block tridiagonal matrices; The block tridiagonal matrix is ​​decomposed by a matrix decomposition algorithm, and the simplified matrix equation group is extracted from the decomposition result.

7. The method for simulating the sequential production of electric hydrogen fusion according to claim 6, characterized in that: The matrix decomposition algorithm includes a recursive banded angularization matrix decomposition algorithm.

8. The method for simulating the sequential production of electric hydrogen fusion according to claim 1, characterized in that: The linear programming solution method is the big M method, and the electric-hydrogen fusion matrix of the electric-hydrogen fusion operation model is linearized by introducing a variable method.

9. The method for simulating the sequential production of electric hydrogen fusion according to claim 1, characterized in that: The conjugate gradient asynchronous parallel computing method comprises the following steps: Dividing the time series data of a preset time into a plurality of continuous sub-periods, and each of the sub-periods includes a preset coupling period; Initializing the residual vector, search direction vector and penalty factor of each sub-period; Perform conjugate gradient iterations in parallel for each initialized sub-period, calculate the current step size and update the state variables; Synchronizing device states of two adjacent sub-periods in the coupling period; Adaptively adjust the penalty factor based on the residual norm and update the search direction vector; The residual norms of all sub-periods are iteratively calculated through an iterative method until they reach the preset convergence tolerance.

10. The method for simulating the sequential production of electric hydrogen fusion according to claim 9, characterized in that: The method of adaptively adjusting the penalty factor based on the residual norm comprises the following steps: The penalty factor is dynamically updated according to the residual norm ratio of two adjacent iterations.

Images