A wind-storage scheduling simulation method and system based on pattern decoupling and continuous embedding
By employing a mode decoupling and continuous embedding wind and energy storage scheduling simulation method, the conflict between modeling and solving efficiency of battery nonlinear degradation costs in wind and energy storage scheduling simulation is resolved, achieving efficient and accurate wind and energy storage capacity optimization configuration.
Patent Information
- Application Number
- CN202611269403.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-20
- Publication Date
- 2026-11-17
AI Technical Summary
Existing wind and energy storage scheduling simulation methods cannot simultaneously balance the accuracy of battery nonlinear degradation cost modeling and the efficiency of solution, leading to systematic biases in capacity allocation decisions.
By employing a pattern decoupling and continuous embedding approach, the scheduling cycle is divided into continuous time periods. A nonlinear degradation cost function is constructed, and backward recursion is performed by embedding immediate rewards through dynamic programming to generate a pattern sequence. This sequence is then transformed into pattern equality constraints of a linear programming model to solve for the optimal continuous charging and discharging power.
It achieves high-precision battery degradation cost modeling, reduces solution time, improves the efficiency and accuracy of scheduling simulation, and provides a reliable wind and storage capacity optimization configuration scheme.
Smart Images

Figure CN123121749A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of wind farm energy storage capacity optimization and configuration technology, specifically involving a wind-storage scheduling simulation method and system based on mode decoupling and continuous embedding. Background Technology
[0002] In the capacity optimization configuration of wind-storage combined power generation systems, the power and capacity specifications of the energy storage system ( Energy storage (ESD) is a key decision variable affecting the economic viability of a project throughout its entire lifecycle. To evaluate the economic feasibility of a candidate configuration, engineering practice typically involves simulating the optimal charge-discharge scheduling plan for that configuration on a time-period (usually half-hour or one-hour) scale throughout the year, and calculating the internal rate of return (IRR) or net present value (NPV) based on the scheduling results. This serves as the basis for comparing different configuration options. Due to the increasingly complex settlement rules in the electricity spot market, involving multiple dimensions such as day-ahead market tariffs, real-time market deviation settlement, government-authorized contract price difference settlement, day-ahead-real-time deviation revenue recovery, and power forecast deviation assessment, the accuracy of scheduling simulation directly determines the reliability of capacity allocation decisions. Therefore, constructing a scheduling simulation method that accurately reflects energy storage operation constraints and market settlement rules while possessing high solution efficiency has become one of the core technical requirements in the field of wind-storage capacity optimization.
[0003] Currently, linear programming (LP) is the most mainstream technique in wind and energy storage scheduling simulation. Its advantages lie in the fact that LP models guarantee global optimality, and solvers (such as HiGHS and Gurobi) possess mature and efficient processing capabilities for large-scale sparse LP problems, capable of completing monthly-scale (e.g., N=1488 time periods) scheduling optimization calculations within seconds. However, LP models require that both the objective function and constraints be linear functions of the decision variables. This inherent limitation necessitates that the marginal degradation cost of energy storage batteries be simplified to a constant independent of the state of charge (SOC) in the LP model. However, numerous experimental studies have shown a strong nonlinear positive correlation between the degradation rate of lithium-ion batteries and the depth of charge / discharge (DOD). Taking standard cycle aging test data of lithium iron phosphate (LFP) batteries as an example, the equivalent full cycle life under shallow cycling (DOD≈20%) can reach more than 15,000 cycles, corresponding to a degradation cost of approximately 23.8 yuan / MWh per unit energy throughput; while the cycle life under deep cycling (DOD≈80%) is only 3,000~4,000 cycles, corresponding to a degradation cost of approximately 118.5 yuan / MWh, a difference of about 5 times. LP uses a fixed degradation cost coefficient (usually taking an intermediate value of about 50 yuan / MWh) for modeling, which leads to excessive penalty for shallow cycling arbitrage behavior on the cost side and insufficient penalty for deep cycling behavior. This causes the optimization model to deviate from the true global optimum in the degradation cost dimension, ultimately affecting the calculation accuracy of NPV and IRR, and causing a systematic bias in capacity configuration decisions.
[0004] To overcome the linearity limitations of dynamic programming (LP), dynamic programming (DP) can accurately characterize the nonlinear characteristics of degradation costs by embedding different degradation costs according to the state of charge (SOC) in the immediate reward of the value function. However, there is a direct conflict between the modeling accuracy and solution efficiency of DP for degradation costs: if the discretization error of degradation costs is to be controlled within an acceptable range, the number of SOC states K usually needs to reach more than 100. With an action discretization level of M ≥ 10, the monthly recursive complexity O(N×K×2M) can reach millions of times, and the monthly solution time increases sharply from 3-5 seconds in LP to 20-30 seconds. In scenarios where thousands of candidate configuration schemes need to be evaluated in the outer layer of wind storage capacity optimization, the computational burden of pure DP makes it difficult to apply in engineering. Summary of the Invention
[0005] The purpose of this invention is to overcome the above-mentioned problems and provide a wind and energy storage scheduling simulation method and system based on mode decoupling and continuous embedding, so as to solve the technical problem that the existing wind and energy storage scheduling simulation methods are difficult to simultaneously take into account the modeling accuracy and solution efficiency of battery nonlinear degradation cost.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a wind-storage scheduling simulation method based on pattern decoupling and continuous embedding, comprising the following steps: The scheduling cycle is divided into continuous time periods, and a nonlinear degradation cost function is constructed based on the battery cycle aging test data. A dynamic programming model is constructed, and the nonlinear degradation cost function is embedded into the immediate reward for backward recursion to solve the charging and discharging actions in each time period. A pattern sequence is generated based on the action type of the charging and discharging actions. The pattern sequence is transformed into pattern equality constraints of a linear programming model. The linear programming model is then solved under the pattern equality constraints to obtain the optimal continuous charging and discharging power values for each time period, thus completing the wind-storage scheduling simulation.
[0007] A further improvement of this invention is that the nonlinear degradation cost function is constructed as follows: Cyclic aging tests were conducted on the battery at different depths of charge and discharge, and the equivalent number of full cycles at each depth of charge and discharge level was recorded. (DOD); An exponential-power-law hybrid model was used for fitting, and the exponential-power-law hybrid model is as follows:
[0008] In the formula, Depth of charge / discharge, These are the fitting parameters; Substituting the fitting result into the following equation yields the nonlinear degradation cost function:
[0009] In the formula, For the depth of charge and discharge is Marginal degradation cost at time Cost per unit capacity of energy storage.
[0010] A further improvement of this invention is that the dynamic programming model is constructed as follows: Discretize the continuous interval of the state of charge as Each state node discretizes the charging and discharging power into... Each charging level and Each discharge level; For each state node, the marginal degradation cost is obtained from the nonlinear degradation cost function based on the corresponding charge / discharge depth, and the marginal degradation cost is included in the immediate reward. Based on the Bellman optimality principle, a backward recursive equation is constructed. All charging levels, discharging levels, and idle actions are traversed at each state node in each time period. The charging and discharging actions in each time period are selected based on maximizing the cumulative revenue.
[0011] A further improvement of this invention is that the continuous interval of the state of charge is discretized into... Each equidistant state node, the charging and discharging power is discretized as follows: Each charging level and For each discharge level, the complexity of the backward recursion is:
[0012] In the formula, The total number of scheduling periods. The discrete number of the state of charge. This refers to the action level.
[0013] A further improvement of the present invention is that the instant reward Calculated using the following formula:
[0014] in, For time period index, For state of charge index, For action indexing, For action The corresponding charging power, For action The corresponding discharge power, For time period Real-time electricity price For time step, In the state of charge The corresponding marginal degradation cost, To predict the point-by-point marginal penalty rate for deviation, and Time periods Positive deviation assessment auxiliary variables and negative deviation assessment auxiliary variables, and Time periods The over-issue recovery factor and the under-issue recovery factor, and Time periods Over-collected variables and under-collected variables.
[0015] A further improvement of the present invention is that the pattern sequence The elements in the array can take values of +1, 1, or 0, representing charging mode, discharging mode, and idle mode, respectively. The specific method for transforming the pattern sequence into pattern equality constraints of a linear programming model is as follows: when At that time, generate constraints Only retain the charging power variable. ; when At that time, generate constraints Only the discharge power variable is retained. ; when At the same time, constraints are generated. and ; In the formula, For time period The charging power is a continuous decision variable. For time period The discharge power is a continuous decision variable.
[0016] A further improvement of the present invention is that the objective function of the linear programming model is to minimize the total cost, and the marginal degradation cost of energy storage in the objective function adopts a fixed coefficient; The decision variables of the linear programming model include: charging power decision variables for each time period. Discharge power decision variables for each time period State of charge decision variables at different time periods And various auxiliary variables for assessment; The constraints of the linear programming model include at least the following: dynamic equation constraints on the state of charge at each time period, upper and lower limit constraints on charging power and discharging power, upper and lower limit constraints on the state of charge, and non-negative constraints on grid-connected power.
[0017] A further improvement of this invention lies in constructing the constraint matrix of the linear programming model using a compressed sparse row format, wherein the total number of non-zero elements in the constraint matrix is:
[0018] In the formula, Given the total number of scheduling periods, the constructed constraint matrix is passed to the solver for solving.
[0019] A further improvement of this invention is that, after solving the linear programming model to obtain the optimal continuous charge and discharge power values for each time period, a post-evaluation step is also included: The optimal continuous charging and discharging power values for each time period are used as the final dispatch plan. The comprehensive incremental revenue is calculated based on the settlement rules of the electricity spot market. The comprehensive incremental revenue includes three dimensions: day-ahead-real-time price arbitrage revenue, revenue from forecast deviation assessment reduction, and revenue from day-ahead-real-time deviation recovery reduction.
[0020] Secondly, the present invention also provides a wind-storage scheduling simulation system based on pattern decoupling and continuous embedding, comprising the following modules: The degradation cost function construction module is used to divide the scheduling cycle into continuous time periods and construct a nonlinear degradation cost function based on the battery's cyclic aging experimental data. The pattern sequence solving module is used to construct a dynamic programming model, embed the nonlinear degradation cost function into the immediate reward for backward recursion, solve the charging and discharging actions in each time period, and generate a pattern sequence according to the action type of the charging and discharging actions. The continuous power optimization module is used to transform the pattern sequence into pattern equality constraints of a linear programming model, solve the linear programming model under the pattern equality constraints, obtain the optimal continuous charging and discharging power values for each time period, and complete the wind-storage scheduling simulation.
[0021] Compared with the prior art, the present invention has the following beneficial effects: This invention provides a wind-storage scheduling simulation method based on pattern decoupling and continuous embedding. By decomposing the wind-storage charge-discharge scheduling problem into two decoupled stages—dynamic programming pattern recognition and linear programming continuous optimization—it effectively solves the technical problem of simultaneously achieving high accuracy and efficient solution in existing technologies for modeling nonlinear degradation costs of batteries. Compared to existing pure linear programming methods that use fixed degradation cost coefficients, leading to excessive suppression of shallow-cycle arbitrage and insufficient penalties for deep-cycle operations, this application first constructs a nonlinear degradation cost function based on an exponential-power-law hybrid model using cyclic aging experimental data. This accurately characterizes the real physical characteristic of lithium-ion battery degradation costs increasing nonlinearly with charge-discharge depth. Then, it embeds this nonlinear degradation cost function into the immediate reward through coarse-grained dynamic programming. K =20、 M The lightweight configuration with a power density of 5 solves for the optimal charging and discharging mode sequence for each time period, achieving mode decision-making in only about 1.55 times the solution time of pure linear programming (4.8 seconds vs. 3.1 seconds). Finally, the mode sequence is transformed into mode equality constraints within linear programming, and continuous power values are solved under these constraints. This ensures global optimality of power decisions while avoiding the exponential computational burden of pure dynamic programming. Experimental results show that the degradation cost deviation rate of this method is only 4.5%, a reduction of approximately 75.4% compared to 18.3% for pure linear programming; the incremental benefit reaches 614,820 yuan, only 0.38% different from pure dynamic programming (617,150 yuan), but the solution time is only 22.3% of that of pure dynamic programming. This approach achieves efficient scheduling simulation while maintaining high-precision degradation cost modeling, providing a reliable scheduling simulation engine for wind and storage capacity optimization. Attached Figure Description
[0022] The accompanying drawings described herein are for illustrative purposes only and are not intended to limit the scope of the invention in any way. Furthermore, the shapes and proportions of the components in the drawings are merely illustrative to aid in understanding the invention and do not specifically limit the shapes and proportions of the components of the invention.
[0023] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system flowchart of the present invention. Detailed Implementation
[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0025] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0026] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0027] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0028] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.
[0029] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.
[0030] The present invention will now be described in further detail with reference to the accompanying drawings: like Figure 1 As shown, this invention provides a wind-storage scheduling simulation method based on pattern decoupling and continuous embedding, comprising the following steps: S1 divides the scheduling cycle into continuous time periods and constructs a nonlinear degradation cost function based on the battery's cyclic aging experimental data.
[0031] The nonlinear degradation cost function is constructed as follows: based on the degradation-cycle life relationship fitting of the cyclic aging test data, constant current and constant voltage charge-discharge cyclic aging tests are conducted on the same type of lithium iron phosphate (LFP) battery to be used at 5 standard depth of charge / discharge levels (DOD=20%, 40%, 60%, 80%, 100%), and the equivalent number of full cycles when the battery capacity decays to 80% of the initial capacity at each DOD level is recorded. (DOD). Based on experimental data, an exponential-power-law hybrid model was used for fitting:
[0032] In the formula, Depth of charge / discharge, These are the fitting parameters. Taking a 280Ah lithium iron phosphate battery as an example, the calibration results are as follows: A =42000 times B =2.87、 C =180 times, goodness of fit R 2 =0.983. (Exponential term) This describes the exponential decay of cycle lifetime caused by accelerated SEI film growth during deep discharge, with the reciprocal term... The physical trend of a sharp increase in cycle lifetime with decreasing DOD during shallow discharge is described. The fitting results are then analyzed. Substitute (DOD) into the marginal degradation cost of energy storage. This yields the degenerate cost function in closed analytic form:
[0033] In the formula, For the depth of charge and discharge is Marginal degradation cost at time Cost per unit capacity of energy storage.
[0034] Calculated based on typical parameters of lithium iron phosphate batteries, =600,000 yuan / MWh, when DOD=20%, =23.8 yuan / MWh (shallow cycle, low penalty), when DOD=80%, =118.5 yuan / MWh (high penalty for deep cycling). This function is continuously differentiable and monotonically increasing throughout the DOD interval, meaning that deeper discharge generates higher marginal degradation costs, providing an accurate mathematical basis for calculating degradation costs in the DP stage. This step is a preliminary preparation for the method, and after one calibration, it is applicable to all scheduling simulation calculations for this battery model.
[0035] S2, construct a dynamic programming model, embed the nonlinear degradation cost function into the immediate reward for backward recursion, solve the charging and discharging actions in each time period, and generate a pattern sequence based on the action type of the charging and discharging actions.
[0036] The dynamic programming model is constructed by dividing the continuous interval of the energy storage's state of charge. Discretize into There are equidistant state nodes, as follows:
[0037]
[0038] Taking a 20MW / 40MWh energy storage configuration as an example, the state of charge of the energy storage is discretized into 20 equidistant states, with a discretization step size of... =(36.0-4.0) / (20-1)=1.684MWh, with each SOC range spanning approximately 4.2% of the rated capacity. The action space discretization method is as follows: the charging power is discretized into M+1 levels: Discretize the discharge power into M levels (excluding zero): The charging and discharging actions are mutually exclusive. The total action space size is 2M+1 (M charging levels + M discharging levels + 1 idle level, typically 5). Based on the above rules, the discretized charging and discharging actions are defined. a : (idle); (Charging), action ; (Discharge), action , For each state node, the marginal degradation cost is obtained from the nonlinear degradation cost function based on the corresponding charge / discharge depth, and this marginal degradation cost is included in the immediate reward. The value function is defined and the final value is initialized as follows: Define From time Start, SOC status is When executing the optimal strategy, the cutoff time is... Maximum cumulative return. Terminal value condition. Reflection period The economic value of remaining electricity (after the last period at the end of the month) can be assessed using the following formula: ,in, This represents the average real-time electricity price for the month. This refers to the energy storage discharge efficiency. The economic meaning of this definition is: if the final value... Higher than the initial value This indicates that energy storage net stores electricity that can be monetized in the future; if the future value... Below the initial value This indicates that energy storage overdrawn its capacity to generate higher revenue for the month, and the reset cost of the overdrawn capacity must be deducted. This definition ensures that DP does not favor extreme strategies of "depleting energy storage" or "hoarding energy storage," maintaining the sustainability of the scheduling plan. The way to construct degradation-aware instant rewards is: for each discrete SOC state... Calculate the depth of charge / discharge corresponding to this state. Substituting the aforementioned Constructing a degradation-aware instant reward function:
[0039] in, For time period index, For state of charge index, For action indexing, For action The corresponding charging power, For action The corresponding discharge power, For time period Real-time electricity price For time step, In the state of charge The corresponding marginal degradation cost, To predict the point-by-point marginal penalty rate for deviation, and Time periods Positive deviation assessment auxiliary variables and negative deviation assessment auxiliary variables, and Time periods The over-issue recovery factor and the under-issue recovery factor, and Time periods The variables for over-issuance and under-issuance recovery. The first term is the charging / discharging revenue. Indicates time period The first item is the real-time electricity price; the second item is the marginal degradation cost of energy storage. Shallow cycles incur low penalties, while deep cycles incur high penalties; the third item is a penalty for prediction deviation. This represents the point-by-point marginal penalty rate, which is the penalty amount payable for exceeding the allowable threshold by 1 MW for one consecutive time period, expressed in yuan / MW / time period, with a typical value of 4.17 yuan / MW / time period; the fourth item is the day-ahead-to-real-time deviation revenue recovery. Indicates time period Over-issuance recovery factor, Indicates time period Reduce the number of recovery coefficients. Based on the Bellman optimality principle, construct a backward recursive equation to establish the Bellman optimality equation: Then, the time loop is iterated in reverse order. From Decrease sequentially until In each time period and at each state node, all charging levels, discharging levels, and idle actions are traversed to select the charging / discharging action for each time period based on maximizing cumulative revenue. The following three nested loops are executed in each time period. Outer loop: Traverses the SOC state. For each discrete SOC state k (k=0,1,...,K-1) at time t, "assuming the energy storage is currently in..." The state is defined, and the optimal action is found for this hypothetical state. The middle loop iterates through the available actions. (Regarding the state...) Iterate through all valid actions that have passed the pre-filter. (Charging from 0 to Level, discharge from 1 to Level, free time, approximately (each effective action) is evaluated one by one: "If the action is performed..." How much total revenue will be obtained? Inner operations: For each action Perform the following three calculations. Step 1 (State Transition): Calculate the action to be performed. The subsequent continuous SOC value. ,in, = = , The round-trip efficiency for energy storage charge / discharge (typically 90%). The second step (state mapping): [This refers to the process of mapping continuous energy storage to a specific state]. Mapping to the nearest of the K discrete states. Since subsequent recursion only occurs when... Defined on a discrete state This requires assigning continuous values to their nearest discrete grid points. A binary search is used, based on a pre-stored array of K increasing SOC values. Search in The insertion position is determined by taking the nearest state index. Step 3 (Value Comparison and Update): Read the pre-stored value for the next time period. (exist (Time has been calculated and stored), with instant rewards. Add them together to get the total value Take the maximum value among all valid actions. Value as The updated value is recorded, along with the action index that reaches that maximum value. Wherein, the continuous interval of the state of charge is discretized as: Each equidistant state node, the charging and discharging power is discretized as follows: Each charging level and For each discharge level, the complexity of the backward recursion is:
[0040] In the formula, The total number of scheduling periods. The discrete number of the state of charge. This refers to the action level. The method for extracting the charging / discharging mode sequence is as follows: starting from the initial SOC, proceeding along the optimal strategy time by time, and marking the mode according to the extracted optimal action type: charging action marker. Discharge action marker Free space marker Output a ternary pattern sequence of length N. The pattern sequence The element in the value is +1. 1 or 0 represents charging mode, discharging mode, and idle mode, respectively.
[0041] S3, the pattern sequence is transformed into pattern equality constraints of a linear programming model, and the linear programming model is solved under the pattern equality constraints to obtain the optimal continuous charging and discharging power values for each time period, thus completing the wind-storage scheduling simulation.
[0042] The linear programming model is constructed as follows: an LP model containing 7 types of decision variables, 4 objective components, and 8 types of constraints. The 7 decision variables include: energy storage charging power. Energy storage and discharge power Energy storage state of charge Auxiliary variables for predicting positive deviation assessment penalties Auxiliary variable for predicting negative deviation assessment penalties Over-issuance and recycling of variables Fewer variables to be recycled t=0,1,...,N-1, where N is the number of time periods. The time step is 0.5 hours (typically). The objective function is to minimize the total cost.
[0043] In this formula, the marginal degradation cost of energy storage λ A fixed coefficient (typically 50 yuan / MWh) is used. The pattern sequence is transformed into linear equality constraints for linear programming. The pattern sequence is then traversed. For each time period Generate LP equality constraints according to the following rules: If (Charging mode), generate constraints ;like (Discharge mode), generating constraints ;like (Idle mode), while generating constraints and Unconstrained power variables under each mode remain continuous decision variables, with values ranging from [value range missing]. The decision variables of the linear programming model include: the charging power decision variables for each time period. Discharge power decision variables for each time period State of charge decision variables at different time periods and various assessment auxiliary variables ( , , , The constraints of the linear programming model include at least the following: dynamic equation constraints on the state of charge between time periods (N-1 equality constraints): ( ,in , For energy storage charge / discharge round-trip efficiency, a typical value is 90%), initial SOC constraint (1 equality constraint): ( , (For the initial energy storage state, a typical value of 0.5), upper and lower limits constraints on charging and discharging power (2N inequality constraints): ( , ), upper and lower limits of charge state constraints (2N inequality constraints): ( ), To provide the minimum amount of electricity required for energy storage operation, The maximum operating power of the energy storage; non-negative grid-connected power constraint (N inequality constraints): , for Real-time wind power output during the time period; Terminal SOC sustainability constraints (1 inequality constraint): Prediction deviation assessment constraints (2N inequality constraints): , , , .in, For real-time grid-connected power of wind and energy storage, ; The limit for wind power prediction deviation assessment. , Power prediction yaw assessment target (typical value 83%). Rated power of the wind farm; Deviation revenue recovery constraints (2N inequality constraints): , , , ,in, For the day-ahead market in electricity trading Power over a given period.
[0044] The constraint matrix of the linear programming model is constructed using a compressed sparse row format, which significantly reduces memory usage and improves solution speed. For N time periods, the total number of variables is 7N (…). N, N, N, N, N, N, The total number of constraints is 7N+1: N-1 SOC recursive equation constraints, 1 initial SOC equation constraint, 1 terminal SOC inequality constraint, N grid-connected power non-negativity constraints, 2N prediction deviation assessment constraints, and 2N deviation revenue recovery constraints. The total number of non-zero elements is 19N-2, distributed as follows: 1) SOC dynamic equations: 4 non-zero elements per equation, totaling (N-1)×4; 2) Initial SOC constraints: 1 non-zero element; 3) Terminal SOC constraints: 1 non-zero element; 4) Grid-connected non-negativity constraints: 2 non-zero elements per equation, totaling 2N; 5) Prediction deviation constraints: 3 non-zero elements per equation, totaling 6N; 6) Deviation recovery constraints: 3 non-zero elements per equation, totaling 6N; 7) Mode equation constraints: 1 non-zero element per equation, totaling N. Taking N=1,488 as an example, the total number of non-zero elements is 19N-2=28270, and the total number of constraints is 7N+1=10417. The total number of variables is approximately 5.46N=8124. The matrix sparsity is 1-28270 / (10417). (8124)≈99.97%. The matrix is constructed by adding triples row by row, and finally converted into CSR format and passed to HiGHS for solving. The constructed constraint matrix is then passed to the solver for solving.
[0045] After solving the linear programming model to obtain the optimal continuous charging and discharging power values for each time period, a post-evaluation step is also included: using the optimal continuous charging and discharging power values for each time period as the final scheduling plan, a precise daily-scale penalty formula is used for post-evaluation, the recovery fee is calculated according to the electricity spot market deviation revenue recovery formula, and the comprehensive incremental revenue is calculated based on the electricity spot market settlement rules. This post-evaluation uses the exact same formula for pure LP, pure DP, and the method of this invention, ensuring the fairness of the comparison. The "incremental revenue" in the post-evaluation result is defined as the total revenue increment with energy storage relative to the baseline without energy storage, including three dimensions: day-ahead-real-time price arbitrage revenue, forecast deviation assessment reduction revenue, and day-ahead-real-time deviation revenue recovery reduction revenue. Using data from the Cangnan No. 2 offshore wind farm in January 2025 (N=1488 half-hour time periods, energy storage configuration 20MW / 40MWh), As an example, the performance of the three methods was tested under the same hardware environment. The comparison data of the three methods are as follows: The solution time of the pure LP method (no DP configuration, LP configuration λ=50) is 3.1s, the incremental benefit is 599,405 yuan, and the degradation cost deviation rate is 18.3%; the solution time of the pure DP method (DP configuration K=100, M=10, λ(DOD) nonlinearity, no LP configuration) is 21.5s, the incremental benefit is 617,150 yuan, and the degradation cost deviation rate is 3.8%; the solution time of the method of this invention (DP configuration K=20, M=5, λ(DOD) nonlinear output mode sequence, LP configuration λ=50 mode equality constraint) is 4.8s, the incremental benefit is 614,820 yuan, and the degradation cost deviation rate is 4.5%. The method of this invention reduces the degradation cost deviation rate from 18.3% to 4.5% with a solution time of about 1.55 times that of pure LP, with incremental benefits differing from pure DP by only 0.38%, while the solution time is only about 22.3% of that of pure DP.
[0046] like Figure 2 As shown, the present invention also provides a wind-storage scheduling simulation system based on pattern decoupling and continuous embedding, comprising the following modules: The degradation cost function construction module is used to divide the scheduling cycle into continuous time periods and construct a nonlinear degradation cost function based on the battery's cyclic aging experimental data. The pattern sequence solving module is used to construct a dynamic programming model, embed the nonlinear degradation cost function into the immediate reward for backward recursion, solve the charging and discharging actions in each time period, and generate a pattern sequence according to the action type of the charging and discharging actions. The continuous power optimization module is used to transform the pattern sequence into pattern equality constraints of a linear programming model, solve the linear programming model under the pattern equality constraints, obtain the optimal continuous charging and discharging power values for each time period, and complete the wind-storage scheduling simulation.
[0047] In one embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions from the computer storage medium to achieve a corresponding method flow or function. The processor described in this embodiment of the present invention can be used in the operation of a wind turbine blade lightning damage detection method based on infrared-electromagnetic coupling.
[0048] This invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the wind turbine blade lightning damage detection method based on infrared-electromagnetic coupling in the above embodiments.
[0049] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0050] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0051] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0052] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0053] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.
Claims
1. A wind-storage scheduling simulation method based on pattern decoupling and continuous embedding, characterized in that, Includes the following steps: The scheduling cycle is divided into continuous time periods, and a nonlinear degradation cost function is constructed based on the battery cycle aging test data. A dynamic programming model is constructed, and the nonlinear degradation cost function is embedded into the immediate reward for backward recursion to solve the charging and discharging actions in each time period. A pattern sequence is generated based on the action type of the charging and discharging actions. The pattern sequence is transformed into pattern equality constraints of a linear programming model. The linear programming model is then solved under the pattern equality constraints to obtain the optimal continuous charging and discharging power values for each time period, thus completing the wind-storage scheduling simulation.
2. The wind-storage scheduling simulation method based on pattern decoupling and continuous embedding according to claim 1, characterized in that, The nonlinear degradation cost function is constructed as follows: Cyclic aging tests were conducted on the battery at different depths of charge and discharge, and the equivalent number of full cycles at each depth of charge and discharge level was recorded. (DOD); An exponential-power-law hybrid model was used for fitting, and the exponential-power-law hybrid model is as follows: In the formula, Depth of charge / discharge, These are the fitting parameters; Substituting the fitting result into the following equation yields the nonlinear degradation cost function: In the formula, For the depth of charge and discharge is Marginal degradation cost at time Cost per unit capacity of energy storage.
3. The wind-storage scheduling simulation method based on pattern decoupling and continuous embedding according to claim 1, characterized in that, The dynamic programming model is constructed as follows: Discretize the continuous interval of the state of charge as Each state node discretizes the charging and discharging power into... Each charging level and Each discharge level; For each state node, the marginal degradation cost is obtained from the nonlinear degradation cost function based on the corresponding charge / discharge depth, and the marginal degradation cost is included in the immediate reward. Based on the Bellman optimality principle, a backward recursive equation is constructed. All charging levels, discharging levels, and idle actions are traversed at each state node in each time period. The charging and discharging actions in each time period are selected based on maximizing the cumulative revenue.
4. The wind-storage scheduling simulation method based on pattern decoupling and continuous embedding according to claim 3, characterized in that, The continuous interval of the charge state is discretized as follows: Each equidistant state node, the charging and discharging power is discretized as follows: Each charging level and For each discharge level, the complexity of the backward recursion is: In the formula, The total number of scheduling periods. The discrete number of the state of charge. This refers to the action level.
5. The wind-storage scheduling simulation method based on pattern decoupling and continuous embedding according to claim 1, characterized in that, The instant reward Calculated using the following formula: in, For time period index, For state of charge index, For action indexing, For action The corresponding charging power, For action The corresponding discharge power, For time period Real-time electricity price For time step, In the state of charge The corresponding marginal degradation cost, To predict the point-by-point marginal penalty rate for deviation, and Time periods Positive deviation assessment auxiliary variables and negative deviation assessment auxiliary variables, and Time periods The over-issue recovery factor and the under-issue recovery factor, and Time periods Over-collected variables and under-collected variables.
6. The wind-storage scheduling simulation method based on pattern decoupling and continuous embedding according to claim 1, characterized in that, The pattern sequence The elements in the array can take values of +1, 1, or 0, representing charging mode, discharging mode, and idle mode, respectively. The specific method for transforming the pattern sequence into pattern equality constraints of a linear programming model is as follows: when At that time, generate constraints Only retain the charging power variable. ; when At that time, generate constraints Only the discharge power variable is retained. ; when At the same time, constraints are generated. and ; In the formula, For time period The charging power is a continuous decision variable. For time period The discharge power is a continuous decision variable.
7. The wind-storage scheduling simulation method based on pattern decoupling and continuous embedding according to claim 6, characterized in that, The objective function of the linear programming model is to minimize the total cost, and the marginal degradation cost of energy storage in the objective function adopts a fixed coefficient. The decision variables of the linear programming model include: charging power decision variables for each time period. Discharge power decision variables for each time period State of charge decision variables at different time periods And various auxiliary variables for assessment; The constraints of the linear programming model include at least the following: dynamic equation constraints on the state of charge at each time period, upper and lower limit constraints on charging power and discharging power, upper and lower limit constraints on the state of charge, and non-negative constraints on grid-connected power.
8. The wind-storage scheduling simulation method based on pattern decoupling and continuous embedding according to claim 7, characterized in that, The constraint matrix of the linear programming model is constructed using a compressed sparse row format. The total number of non-zero elements in the constraint matrix is: In the formula, Given the total number of scheduling periods, the constructed constraint matrix is passed to the solver for solving.
9. The wind-storage scheduling simulation method based on pattern decoupling and continuous embedding according to claim 1, characterized in that, After solving the linear programming model to obtain the optimal continuous charge and discharge power values for each time period, a post-evaluation step is also included: The optimal continuous charging and discharging power values for each time period are used as the final dispatch plan. The comprehensive incremental revenue is calculated based on the settlement rules of the electricity spot market. The comprehensive incremental revenue includes three dimensions: day-ahead-real-time price arbitrage revenue, revenue from forecast deviation assessment reduction, and revenue from day-ahead-real-time deviation recovery reduction.
10. A wind-storage scheduling simulation system based on pattern decoupling and continuous embedding, characterized in that, Includes the following modules: The degradation cost function construction module is used to divide the scheduling cycle into continuous time periods and construct a nonlinear degradation cost function based on the battery's cyclic aging experimental data. The pattern sequence solving module is used to construct a dynamic programming model, embed the nonlinear degradation cost function into the immediate reward for backward recursion, solve the charging and discharging actions in each time period, and generate a pattern sequence according to the action type of the charging and discharging actions. The continuous power optimization module is used to transform the pattern sequence into pattern equality constraints of a linear programming model, solve the linear programming model under the pattern equality constraints, obtain the optimal continuous charging and discharging power values for each time period, and complete the wind-storage scheduling simulation.
