An energy storage strategy initialization method based on divide-and-conquer algorithm

Abstract

The application belongs to the technical field of energy storage strategy initialization, and particularly relates to an energy storage strategy initialization method based on a divide-and-conquer algorithm. period =2, obtain the final P 1 feasible value range and P 2 feasible value range, and randomly select values in the respective value ranges to randomly generate an energy storage strategy satisfying the constraints under the condition of period =2; for the condition of period >2, split period into two parts by halving, split the original problem into two sub-problems, further split each sub-problem until the length of period is 2 or 1; and step 4 is to determine the range of random number selection for the decision variable corresponding to period 1. The application can meet the "feasibility" requirement of an intelligent optimization algorithm for an initial solution, and can effectively improve the calculation efficiency and convergence of the intelligent optimization algorithm.

Description

Technical Field

[0001] This invention belongs to the field of energy storage strategy initialization technology, specifically relating to an energy storage strategy initialization method based on a divide-and-conquer algorithm. Background Technology

[0002] Energy storage systems are an important component of distribution networks, microgrids, and integrated energy systems. They represent the most rational solution for addressing renewable energy consumption, enhancing grid stability, and improving the utilization efficiency of distribution systems. Introducing energy storage into a system can effectively achieve demand-side management, eliminate day-night peak-valley differences, and smooth load fluctuations. This not only allows for more efficient use of electrical equipment and reduced electricity costs but also promotes the application of renewable energy. Furthermore, it serves as an effective means to improve system operational stability, participate in frequency and voltage regulation, and compensate for load fluctuations.

[0003] Considering the complex structure of energy systems and the various energy storage application scenarios mentioned above, it is usually necessary to construct optimization models to calculate energy storage scheduling strategies. Among these, intelligent optimization algorithms have been widely used in the optimization process of energy storage strategies. Intelligent optimization algorithms simulate objective functions based on natural phenomena to find the closest optimal solution. The iterative process of intelligent optimization algorithms must include the following three steps: First, within the feasible range of the objective function, a set of initial solutions is found using a pre-defined strategy for finding optimal solutions; second, the search for optimal solutions continues within the feasible range of the objective function according to the original strategy; third, the termination condition is checked. If satisfied, the optimal solution is selected from all solutions; otherwise, the process returns to step two and continues until the termination condition is met. The first step requires finding a set of initial solutions within the "feasible range." However, for energy storage strategies, due to the complexity of their constraint models, finding a "feasible" initial solution is often difficult.

[0004] Currently, the initial solution acquisition for energy storage strategies in intelligent optimization algorithms typically ignores the existence of constraint models, randomly generating the charging and discharging power of energy storage at different times. Then, a so-called penalty function is added during the execution of the optimization algorithm to gradually correct individuals that do not meet the constraint models. The main problems with this method are: ① It significantly increases the computational load required for optimization algorithm iterations, reducing optimization efficiency; ② When the feasible region is small, the penalty-correction method often struggles to successfully correct individuals into the feasible region, leading to the optimization algorithm failing to converge. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide an energy storage strategy initialization method based on the divide-and-conquer algorithm, so as to meet the "feasibility" requirement of the intelligent optimization algorithm for the initial solution and effectively improve the computational efficiency and convergence of the intelligent optimization algorithm.

[0006] The objective of this invention is achieved as follows: a method for initializing an energy storage strategy based on a divide-and-conquer algorithm, wherein P represents the charging and discharging power of the energy storage, period represents the scheduling period of the energy storage, and the energy storage scheduling strategy is the time series [P1, P2, ..., P] of the charging and discharging power P within the scheduling period period. period It includes the following steps:

[0007] Step 1: Determine the constraints that the energy storage strategy needs to meet;

[0008] Step 2: With period=2, obtain the final feasible value ranges of P1 and P2, and randomly select values ​​within their respective value ranges to randomly generate energy storage strategies that satisfy the constraints with period=2.

[0009] Step 3: For the case where period > 2, the original problem is split into two subproblems by splitting period in half. Each subproblem is further split until the length of period is 2 or 1.

[0010] Step 4: Determine the range of random numbers to be taken for the decision variable when period is 1.

[0011] The constraints that the energy storage strategy in step 1 needs to meet include:

[0012] Energy storage difference constraint at the beginning and end:

[0013]

[0014] Energy storage at any time does not exceed the limit constraint:

[0015]

[0016] The charge / discharge power at any given time satisfies the following constraints:

[0017]

[0018] In the above formulas, P represents the energy storage charging and discharging power (positive value for discharging, negative value for charging), in kW, which is the decision variable of the optimization model; Δt represents the scheduling time interval, in hours; period represents the energy storage scheduling cycle; S represents the energy storage capacity, in kWh; subscript ini represents the initial time; subscript lower represents the lower limit of the energy storage capacity at the last time; subscript upper represents the upper limit of the energy storage capacity at the last time; and subscript lim represents the upper limit of the energy storage system capacity.

[0019] Step 2 includes:

[0020] To simplify the expression form of the subsequent derivation, assume Δt = 1 (note that this assumption does not affect the generality of the entire method); in the special case, i.e., period = 2, equations (1) to (3) can be written as:

[0021] S lower -S ini ≤P1+P2≤S upper -S ini (4)

[0022] 0≤P1≤S lim (5)

[0023] 0≤P1+P2≤S lim (6)

[0024] P lower,1 ≤P1≤P upper,1 (7)

[0025] P lower,2 ≤P2≤P upper,2 (8)

[0026] Combining equations (4) and (6), we obtain:

[0027] max(0,S lower -S ini )≤P1+P2≤min(S lim ,S upper -S ini (9)

[0028] Further simplification of equations (8) and (9) yields:

[0029] max(0,S lower -S ini )-P upper,2 ≤P1≤min(S lim ,S upper -S ini )-P lower,2 (10)

[0030] Combining equations (5), (7), and (10), we obtain the final feasible range of values ​​for P1:

[0031]

[0032] Then, random numbers are generated for P1 within this range. After the random number generation, P1 becomes a known value, denoted by P1′. Subsequently, the feasible range of values ​​for P2 can be determined. The derivation process is similar to the process described above.

[0033]

[0034] Next, random numbers are selected for P2 within this range, thus completing the random generation of energy storage strategies that satisfy the constraints when period=2.

[0035] Step 3 includes:

[0036] For the case where period>2, by rearranging and transforming equations (1) to (3) according to the aforementioned steps, we can obtain the following equations:

[0037]

[0038]

[0039]

[0040]

[0041]

[0042] In the formula This represents rounding up x. The above process splits period in half into two parts. Equivalent to P1 in equations (4) to (12) above, Equivalent to P2 in equations (4) to (12), the first half is obtained according to the aforementioned steps. The range of values ​​for is shown in equation (18):

[0043]

[0044] when Once the value of is determined, that is, it is known. You can get The range of values ​​for:

[0045]

[0046] This further confirms The values ​​of each charge and discharge power;

[0047] Therefore, the key question below lies in how to determine them separately. and The values ​​of charge and discharge power at each moment; according to equation (18), Split in half At this point, equation (18) is formally the same as equation (13), and can construct the... Expressions of the same form as (14) to (17) within the range are similarly treated. The same applies to the period, which involves splitting the original problem into two subproblems, and then further splitting each subproblem until the period has a length of 2 or 1.

[0048] Step 4 includes:

[0049] When the period length is 1, the decision variable is P. j Assuming that equations (6) and (8) do not exist and P2 = 0 in (4), P j That is, P1, at this time P j The range of random numbers to be generated is:

[0050] max(max(0,S lower -S ini ),P lower,1 )≤P1≤min(min(S lim ,S upper -S ini ),P upper,1 (20).

[0051] The beneficial effects of this invention are as follows: This invention provides an energy storage strategy initialization method based on a divide-and-conquer algorithm. Step 1: Determine the constraints that the energy storage strategy needs to satisfy. Step 2: When period = 2, obtain the feasible value ranges for P1 and P2, and randomly generate energy storage strategies that satisfy the constraints within their respective ranges. Step 3: For the case where period > 2, the original problem is divided into two subproblems by splitting the period in half, and each subproblem is further divided until the period length is 2 or 1. Step 4: Determine the range of random values ​​for the decision variables when period is 1. By recursively splitting and solving the energy storage model constraint equations layer by layer, an initial energy storage operation strategy that satisfies the energy storage operation constraints and also has random characteristics can be obtained. This method can meet the "feasibility" requirement of the initial solution for intelligent optimization algorithms and can effectively improve the computational efficiency and convergence of intelligent optimization algorithms. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating an energy storage strategy initialization method based on a divide-and-conquer algorithm according to the present invention. Detailed Implementation

[0053] The present invention will now be further described with reference to the accompanying drawings.

[0054] like Figure 1As shown, an energy storage strategy initialization method based on a divide-and-conquer algorithm is presented, where P represents the energy storage charging and discharging power, period represents the energy storage scheduling period, and the energy storage scheduling strategy is the time series [P1, P2, ..., P] of the charging and discharging power P within the scheduling period period. period It includes the following steps:

[0055] Step 1: Determine the constraints that the energy storage strategy needs to meet;

[0056] Step 2: With period=2, obtain the final feasible value ranges of P1 and P2, and randomly select values ​​within their respective value ranges to randomly generate energy storage strategies that satisfy the constraints with period=2.

[0057] Step 3: For the case where period > 2, the original problem is split into two subproblems by splitting period in half. Each subproblem is further split until the length of period is 2 or 1.

[0058] Step 4: Determine the range of random numbers to be taken for the decision variable when period is 1.

[0059] The constraints that the energy storage strategy in step 1 needs to meet include:

[0060] Energy storage difference constraint at the beginning and end:

[0061]

[0062] Energy storage at any time does not exceed the limit constraint:

[0063]

[0064] The charge / discharge power at any given time satisfies the following constraints:

[0065]

[0066] In the above formulas, P represents the energy storage charging and discharging power (positive value for discharging, negative value for charging), in kW, which is the decision variable of the optimization model; Δt represents the scheduling time interval, in hours; period represents the energy storage scheduling cycle, for example, if energy storage is scheduled at 1-hour intervals (Δt = 1) for 1 day, then period = 24; S represents the energy storage capacity, in kWh, the subscript ini represents the initial time, the subscript lower represents the allowable lower limit of the energy storage capacity at the last time, the subscript upper represents the allowable upper limit of the energy storage capacity at the last time, and the subscript lim represents the upper limit of the energy storage system capacity.

[0067] Energy storage dispatch strategy refers to the time series [P1, P2, ..., P] of the charging and discharging power P within the dispatch period. period As can be seen, the constraint model of energy storage strategy is very complex. When the number of initial "feasible solutions" required by the intelligent optimization algorithm is large and needs to meet certain random distribution rules, obtaining the initial solution becomes a very important issue.

[0068] Step 2 includes:

[0069] To simplify the expression form of the subsequent derivation, assume Δt = 1 (note that this assumption does not affect the generality of the entire method); in the special case, i.e., period = 2, equations (1) to (3) can be written as:

[0070] S lower -S ini ≤P1+P2≤S upper -S ini (4)

[0071] 0≤P1≤S lim (5)

[0072] 0≤P1+P2≤S lim (6)

[0073] P lower,1 ≤P1≤P upper,1 (7)

[0074] P lower,2 ≤P2≤P upper,2 (8)

[0075] Combining equations (4) and (6), we obtain:

[0076] max(0,S lower -S ini )≤P1+P2≤min(S lim ,S upper -S ini (9)

[0077] Further simplification of equations (8) and (9) yields:

[0078] max(0,S lower -S ini )-P upper,2 ≤P1≤min(S lim ,S upper -S ini )-P lower,2 (10)

[0079] Combining equations (5), (7), and (10), we obtain the final feasible range of values ​​for P1:

[0080]

[0081] Then, random numbers are generated for P1 within this range. After the random number generation, P1 becomes a known value, denoted by P1′. Subsequently, the feasible range of values ​​for P2 can be determined. The derivation process is similar to the process described above.

[0082]

[0083] Next, random numbers are selected for P2 within this range, thus completing the random generation of energy storage strategies that satisfy the constraints when period=2.

[0084] Step 3 includes:

[0085] For the case where period>2, by rearranging and transforming equations (1) to (3) according to the aforementioned steps, we can obtain the following equations:

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] In the formula This represents rounding up x. The above process splits period in half into two parts. Equivalent to P1 in equations (4) to (12) above, Equivalent to P2 in equations (4) to (12), the first half is obtained according to the aforementioned steps. The range of values ​​for is shown in equation (18):

[0092]

[0093] when Once the value of is determined, that is, it is known. You can get The range of values ​​for:

[0094]

[0095] This further confirms The values ​​of each charge and discharge power;

[0096] Therefore, the key question below lies in how to determine them separately. and The values ​​of charge and discharge power at each moment; according to equation (18), Split in half At this point, equation (18) is formally the same as equation (13), and can construct the... Expressions of the same form as (14) to (17) within the range are similarly treated. The same applies to the period, which involves splitting the original problem into two subproblems, and then further splitting each subproblem until the period has a length of 2 or 1.

[0097] The decision variable corresponding to a period length of 2 is P. i and P i+1 At this point, we can use the ideas from equations (4) to (12) to analyze P. i and P i+1 Random values ​​are selected.

[0098] Step 4 includes:

[0099] When the period length is 1, the decision variable is P. j Assuming that equations (6) and (8) do not exist and P2 = 0 in (4), P j That is, P1, at this time P j The range of random numbers to be generated is:

[0100] max(max(0,S l o wer -S ini ),P l o wer,1 )≤P1≤min(min(S lim ,S upper -S ini ),P upper,1 (20).

[0101] This process of solving subproblems by splitting them in half is called the divide-and-conquer algorithm. Through this algorithm, an energy storage operation strategy that satisfies all model constraints can be obtained.

[0102] In summary, this invention provides an energy storage strategy initialization method based on a divide-and-conquer algorithm. The method comprises the following steps: Step 1: Determine the constraints that the energy storage strategy must satisfy; Step 2: When period = 2, obtain the feasible value ranges for P1 and P2, and randomly generate energy storage strategies that satisfy the constraints within their respective ranges; Step 3: For the case where period > 2, the original problem is divided into two subproblems by splitting the period in half, and each subproblem is further divided until the period length is 2 or 1; Step 4: Determine the range of random values ​​for the decision variables when period = 1. By recursively splitting and solving the energy storage model constraint equations layer by layer, an initial energy storage operation strategy that satisfies the energy storage operation constraints and also possesses random characteristics can be obtained. This method can meet the "feasibility" requirement of the initial solution for intelligent optimization algorithms and can effectively improve the computational efficiency and convergence of intelligent optimization algorithms.

Claims

1. A method for initializing an energy storage strategy based on a divide-and-conquer algorithm, characterized in that, The method is used to generate an initial operation strategy for energy storage that satisfies the energy storage operation constraints, where P represents the charging and discharging power of the energy storage, period represents the scheduling period of the energy storage, and the energy storage scheduling strategy is the time series [P1, P2, ..., P] of the charging and discharging power P within the scheduling period period. period The method includes the following steps: Step 1: Determine the constraints that the energy storage strategy needs to meet. The constraints include the difference in energy storage at the beginning and end of the energy storage system, the constraint that the energy storage does not exceed the limit at any time, and the constraint of charging and discharging power at any time. Step 2: With period=2, obtain the final feasible value ranges of P1 and P2, and randomly select values ​​within their respective value ranges to randomly generate energy storage strategies that satisfy the constraints with period=2. Step 3: For the case where period > 2, the original problem is split into two subproblems by splitting period in half. Each subproblem is further split until the length of period is 2 or 1. Step 4: Determine the range of random numbers to be taken for the decision variable when period is 1.

2. The energy storage strategy initialization method based on the divide-and-conquer algorithm as described in claim 1, characterized in that, The constraints that the energy storage strategy in step 1 needs to meet include: Energy storage difference constraint at the beginning and end: （1） Energy storage at any time does not exceed the limit constraint: （2） The charge / discharge power at any given time satisfies the following constraints: （3） In the above formulas, P represents the charging and discharging power of energy storage in kW, which is the decision variable of the optimization model; ∆t represents the scheduling time interval in hours; period represents the scheduling cycle of energy storage; S represents the energy storage capacity in kWh; subscript ini represents the initial time; subscript lower represents the lower limit of the energy storage capacity at the last time; subscript upper represents the upper limit of the energy storage capacity at the last time; and subscript lim represents the upper limit of the energy storage system capacity.

3. The energy storage strategy initialization method based on the divide-and-conquer algorithm as described in claim 2, characterized in that, Step 2 includes: To simplify the expression form of the subsequent derivation, assume ∆t=1; in the special case, i.e. period=2, equations (1) to (3) are written as: （4） （5） （6） （7） （8） Combining equations (4) and (6), we obtain: （9） Further simplification of equations (8) and (9) yields: （10） Combining equations (5), (7), and (10), we obtain the final feasible range of values ​​for P1: （11） Then, within this range, a random number is generated for P1. After the value is generated, P1 becomes a known value. This means that the feasible range of values ​​for P2 is then known, and the derivation process is similar to the process described above: （12） Next, random numbers are selected for P2 within this range, thus completing the random generation of energy storage strategies that satisfy the constraints when period=2.

4. The energy storage strategy initialization method based on the divide-and-conquer algorithm as described in claim 3, characterized in that, Step 3 includes: For the case where period>2, equations (1) to (3) are rearranged and transformed according to the aforementioned steps to obtain the following equations: （13） （16） （17） In the formula This represents rounding up x. The above process splits period in half into two parts. Equivalent to P1 in equations (4) to (12) above, Equivalent to P2 in equations (4) to (12), the first half is obtained according to the aforementioned steps. The range of values ​​for is as shown in equation (18): when Once the value of is determined, that is, it is known. , then obtain The range of values ​​for: This further confirms The values ​​of each charge and discharge power; Therefore, the key question below lies in how to determine them separately. and The values ​​of charge and discharge power at each moment; according to equation (18), Split in half At this point, equation (18) is formally the same as equation (13), and can construct the structure in 1~ Expressions of the same form as (14) to (17) within the range are similarly treated. The same applies within the range of +1 to period, which means splitting the original problem into two subproblems, and then further splitting each subproblem until the length of period is 2 or 1.

5. The energy storage strategy initialization method based on the divide-and-conquer algorithm as described in claim 4, characterized in that, Step 4 includes: When the period length is 1, the decision variable is P. j Assuming that equations (6) and (8) do not exist and P2=0 in (4), P j That is, P1, at this time P j The range of random numbers to be generated is: （20）。

Images