Distributed energy storage system scheduling method and system

Abstract

The invention discloses a distributed energy storage system scheduling method and system, and the method is used for achieving quantifiable dynamic balance between peak load shifting and power supply reliability improvement, and comprises the steps: obtaining an equivalent continuous load curve based on random production simulation, determining rated power and rated capacity of the energy storage system by taking a target reliability index threshold value of the power shortage probability or the power shortage expected value as a constraint; constructing a dynamic balance optimization model considering load stabilization, time-of-use electricity price cost and reliability risk at the same time by taking the energy storage charging and discharging power of each time period as a decision variable; and performing global search on the dynamic balance optimization model by adopting a reliability-driven adaptive particle swarm algorithm, performing local refinement on a globally optimal solution obtained by the particle swarm by adopting an interior point method, and outputting an optimal charging and discharging strategy. According to the invention, dynamic balance of distributed energy storage in peak load shifting and power supply reliability improvement can be realized.

Description

Technical Field

[0001] This invention relates to the field of power system energy storage optimization technology, specifically to a method and system for scheduling distributed energy storage systems. Background Technology

[0002] With the fluctuation of power grid load and the increase in renewable energy penetration, the traditional method of relying on capacity expansion is difficult to balance economic efficiency and power supply quality. Battery energy storage has become an important means of peak shaving and valley filling due to its fast charging and discharging capabilities. However, existing methods have the following problems: (1) They only focus on economic efficiency or load smoothing, without modeling reliability indicators and peak shaving and valley filling objectives in a unified manner, which makes the scheme prone to failure during faults or periods of tight supply and demand; (2) Although some studies have introduced intelligent optimization algorithms such as particle swarm optimization, the main improvements are concentrated on linear reduction of inertial weights or weight adjustment based on fuzzy / neural networks, lacking a mechanism to directly use reliability indicators as algorithm parameters to adjust the driving signal, and are prone to generating infeasible solutions under strong SOC constraints; (3) Single heuristic algorithms often have insufficient convergence accuracy under high-dimensional continuous variables, making it difficult to simultaneously guarantee "feasibility + optimality". Summary of the Invention

[0003] The technical problem to be solved by this invention is that current energy storage system scheduling does not consider the problem of achieving a quantifiable dynamic balance between peak shaving and valley filling and improving power supply reliability.

[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A distributed energy storage system scheduling method is used to achieve a quantifiable dynamic balance between peak shaving and valley filling and improving power supply reliability, including: The equivalent continuous load curve is obtained based on random production simulation, and the rated power and rated capacity of the energy storage system are determined by the target reliability index threshold of the power shortage probability or the expected value of power shortage as a constraint. Using the energy storage charging and discharging power in each time period as the decision variable, a dynamic balance optimization model is constructed that simultaneously considers load smoothing, time-of-use electricity price costs, and reliability risks. A reliability-driven adaptive particle swarm optimization algorithm is used to perform a global search on the dynamic equilibrium optimization model. Then, the interior point method is used to refine the global optimal solution obtained by the particle swarm optimization and output the optimal charging and discharging strategy.

[0005] In this embodiment, the rated power and rated capacity of the energy storage system are determined using the following formula: ; ,or ; In the formula, Let the objective function be the scale of the energy storage system. , These are two trade-off coefficients, The rated power of the energy storage system, For the rated capacity of the energy storage system, In configuration The probability of insufficient power at that time Its threshold; For configuration The battery level was lower than expected. Its threshold.

[0006] In this embodiment, the dynamic equilibrium optimization model is represented by the following formula: ; In the formula, For dynamic equilibrium optimization model, , , These are the multi-objective weighting coefficients. For the load variance, for Net load during the cycle, for Equivalent charge / discharge power during the cycle, For time-of-use electricity costs, This poses a reliability risk.

[0007] In this embodiment, reliability risks include the summary penalty for insufficient power, the expected penalty for insufficient power, or the conditional risk value of the expected power. Charge and discharge power constraints, SOC dynamic constraints, SOC upper and lower limit constraints, and SOC regression constraints at the end of the cycle are set: SOC(T)=SOC(0); where T is the scheduling cycle.

[0008] In this embodiment, a reliability-driven adaptive particle swarm optimization algorithm is used to perform a global search on the dynamic equilibrium optimization model, including: Encode a particle as an entire scheduling sequence and obtain the particle's velocity and position; Define particle population diversity and reliability default level; The inertia weights and learning factors of the adaptive particle swarm optimization algorithm are jointly adjusted based on particle swarm diversity and reliability default levels.

[0009] In this embodiment, particle population diversity and reliability breach degree are represented by the following formula: ; ; In the formula, For the first Particle population diversity characterization during the next iteration. For the total particle population, For the first During the nth iteration, the 1st The entire scheduling sequence of each particle. For the first The entire scheduling sequence during the next iteration. It is a 2-norm; For the first Degree of reliability breach during the next iteration For the first Summary of power shortage during the next iteration The threshold for the reliability index is based on a summary of power shortages. For the first The battery level in the next iteration was lower than expected. The threshold for the reliability index is set with the expected value of insufficient power.

[0010] In this embodiment, the inertia weight and learning factor of the adaptive particle swarm optimization algorithm are jointly adjusted based on particle population diversity and reliability default degree, as expressed by the following formula: ; ; ; In the formula, For the first The inertia weight of the next iteration, , These are the minimum and maximum inertia weights, respectively. Let be the base of the natural logarithm function. The inertial weighting exponential decay coefficient is... The maximum number of iterations, For reliability default feedback weighting coefficient, For the first The degree of reliability failure in the next iteration. , These are two learning factors in the adaptive particle swarm optimization algorithm. , Learning factors The maximum and minimum values, , Learning factors The maximum and minimum values, For the first Characterization of particle population diversity during the next iteration.

[0011] In this embodiment, the adaptive particle swarm optimization algorithm performs feasible region projection repair on the particles after each position update to ensure that the power and SOC satisfy the constraints throughout the process; wherein, feasible region projection repair is expressed by the following formula: ; In the formula, For the period The corrected value after applying upper and lower limit constraints to the state of charge at that time. The charge state is obtained from the particle position update before repair. The upper and lower limits of the charged state range. This refers to the period during which feasibility repairs are required.

[0012] In this embodiment, the global optimal solution obtained by particle swarm optimization is locally refined using the interior-point method, including: rewriting the constraints of the dynamic equilibrium optimization model as... The form involves using the global optimal solution obtained from the particle swarm optimization as the initial point, constructing a barrier function, and iteratively updating the barrier factor to obtain a locally optimal feasible solution; where the barrier function is expressed by the following formula: ; In the formula, For the barrier function, The position vector of the particle The corresponding original objective function value, As a barrier factor, The position vector of the particle The One constraint It is the natural logarithm.

[0013] This invention also provides a system applying the above-described distributed energy storage system scheduling method, used to achieve a quantifiable dynamic balance between peak shaving and valley filling and improving power supply reliability, comprising: The reliability-driven energy storage rated parameter determination module is used to obtain the equivalent continuous load curve based on random production simulation, and determine the rated power and rated capacity of the energy storage system by using the target reliability index threshold of power shortage probability or power shortage expectation value as a constraint. The dynamic balance optimization module is used to construct a dynamic balance optimization model that simultaneously considers load smoothing, time-of-use electricity price costs, and reliability risks, using the energy storage charging and discharging power of each time period as the decision variable. The optimal charging and discharging strategy module is used to perform a global search on the dynamic equilibrium optimization model using a reliability-driven adaptive particle swarm optimization algorithm, and then refine the global optimal solution obtained by the particle swarm optimization using an interior point method to output the optimal charging and discharging strategy.

[0014] Compared with the prior art, the beneficial effects of the present invention are: Unified modeling of reliability and peak shaving: LOLP / EENS is explicitly introduced into the objective function or constraints to ensure the reliability of the charging and discharging strategy during periods of supply and demand tension.

[0015] Feasible region projection repair: ensures that the SOC trajectory satisfies the constraints throughout the entire process and meets the end-of-cycle regression, reducing infeasible or "boundary jittering" solutions caused by the penalty function.

[0016] Global-local collaborative optimization: Particle swarm optimization is responsible for global search, while interior point optimization is responsible for local refinement, so that the final solution takes into account both feasibility and accuracy, and improves the feasibility of engineering implementation.

[0017] To obtain stable and feasible solutions under high-dimensional and multi-constraint conditions, this invention proposes a hybrid solution strategy consisting of "reliability-driven adaptive particle swarm optimization global search + feasible region projection repair + interior-point method local refinement": the inertia weight and learning factor of the particle swarm optimization algorithm are jointly adjusted by population diversity and reliability default degree, and SOC trajectory projection repair is performed on particles after each iteration to ensure constraints throughout the process; the current global optimal solution is further optimized locally using the interior-point method to improve convergence accuracy. This invention can reduce load peak-valley difference and reduce electricity costs while satisfying reliability constraints, and is suitable for day-ahead / rolling scheduling scenarios of distributed energy storage systems. Attached Figure Description

[0018] Figure 1 This is a flowchart of a distributed energy storage system scheduling method according to an embodiment of the present invention.

[0019] Figure 2 This is a block diagram of a distributed energy storage system scheduling system according to an embodiment of the present invention. Detailed Implementation

[0020] To facilitate understanding of the technical solution of the present invention by those skilled in the art, the technical solution of the present invention will now be further described in conjunction with the accompanying drawings.

[0021] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0022] Please see Figure 1 As shown, this invention provides a distributed energy storage system scheduling method for achieving a quantifiable dynamic balance between peak shaving and valley filling and improving power supply reliability, comprising: S10: Based on the equivalent continuous load curve obtained from random production simulation, and constrained by the target reliability index threshold of power shortage probability or power shortage expectation value, determine the rated power and rated capacity of the energy storage system.

[0023] In one embodiment of the present invention, reliability-driven energy storage rating parameters are determined by inputting generator availability, load forecast curves, and renewable output scenarios to obtain the equivalent continuous load curve (ELDC) for each scenario. A target reliability threshold is then used. or To constrain the rated power of the energy storage system, traverse the system's power rating. With rated capacity Given the combination of , solve for: ,(1); ,or (2); In the formula, Let the objective function be the scale of the energy storage system. , These are two trade-off coefficients, The rated power of the energy storage system, For the rated capacity of the energy storage system, Configured as The probability of insufficient power at that time Its threshold; For configuration The battery level was lower than expected. Its threshold.

[0024] Furthermore, the amount of reliability improvement can be used as a constraint and can be used as an indicator: (3); (4); In the formula, The improvement amount for the power shortage summary, This represents the probability of power shortage under the baseline operating condition (when no energy storage system is configured or energy storage is not involved in dispatch). To achieve a rated energy storage capacity of Rated capacity is The probability of insufficient power under the configuration; This represents the expected improvement in battery capacity. This represents the expected power shortage under the baseline operating condition (when no energy storage system is configured or energy storage is not involved in dispatch). To achieve a rated energy storage capacity of Rated capacity is The battery capacity under this configuration is less than expected.

[0025] S20 uses the energy storage charging and discharging power of each time period as the decision variable to construct a dynamic balance optimization model that simultaneously considers load smoothing, time-of-use electricity price costs, and reliability risks.

[0026] In one embodiment of the present invention, the dynamic equilibrium optimization model is expressed by the following formula: (5); In the formula, For dynamic equilibrium optimization model, , , These are the multi-objective weighting coefficients. For the load variance, for Net load during the cycle, for Equivalent charge / discharge power during the cycle, For time-of-use electricity costs, This poses a reliability risk.

[0027] In this embodiment, the parameters in the dynamic equilibrium optimization model are obtained in the following way.

[0028] 1. Net load and equivalent load after energy storage: (6); (7); In the formula, Net load, To predict load, For renewable energy output, The equivalent load after the energy storage system operates. The equivalent charge / discharge power is given by, where, (t)= (t)- (t), (t) represents the discharge power of the energy storage system. (t) represents the charging power of the energy storage system.

[0029] 2. Peak shaving and valley filling indicators can be expressed in the form of variance or peak-valley difference. Taking variance as an example: (8); in, ; In the formula, Let T be the load variance and T be the scheduling period.

[0030] 3. Time-of-use electricity cost.

[0031] (9); in, .

[0032] In the formula, π(t) represents the time-of-use electricity price. (t) represents the purchased power, and the dispatch period is discrete as t=1,2,… T， Step size is .

[0033] 4. Reliability Risk: Reliability risk includes the conditional risk value of insufficient power summary over-limit penalty, insufficient power expected over-term penalty, or insufficient power expected, and sets charging and discharging power constraints, SOC dynamic constraints, SOC upper and lower limit constraints and period end SOC regression constraints SOC(T)=SOC(0).

[0034] In this embodiment, for a given set of fault / output scenarios s =1,2,…, S (Obtained from random generation simulation), let the scenario probability be... The power shortage in the scenario is The penalty for insufficient power is as follows: (10); ; In the formula, For the scene Below, in the cycle The lack of power supply For the scene Below, in the cycle The system has available power supply.

[0035] The penalty for insufficient battery power and exceeding the expected time limit is: ,(11) In the formula, As an indicator variable (0-1 variable), it is used to represent the scene Check if a "power shortage / loss of load" event has occurred.

[0036] To enhance the mitigation of extreme supply shortages, the conditional value of risk of EENS can be introduced, allowing the random variable to... Confidence level is (e.g., 0.9~0.99), introduce auxiliary variables. With slack variables The conditional value at risk for insufficient power is then: (12); ; In the reliability risk penalty term, substitute one of the formulas (10) to (12) into the dynamic balance optimization model. Taking the conditional risk value of insufficient power as an example, the specific dynamic balance optimization model is as follows: (13); 5. The constraints on the energy storage system in the dynamic equilibrium optimization model are: 0≤ (t)≤ , 0≤ (t)≤ ,(14) (15); ≤SOC(t)≤ , and SOC(T)=SOC(0), (16); In the formula, SOC(t) represents the state of charge of the energy storage system. , These are the upper and lower limits of the state of charge of the energy storage system, respectively. These represent the charge / discharge efficiency, respectively.

[0037] The energy storage constraints of the dynamic equilibrium optimization model also include: to avoid simultaneous charging and discharging, mutual exclusion constraints can be used. (t)· (t)=0 or the implementation layer adopts the rule of "first determine the charge / discharge state and then assign a value".

[0038] S30 employs a reliability-driven adaptive particle swarm optimization algorithm for global search of the dynamic equilibrium optimization model, and then uses the interior point method to refine the global optimal solution obtained by the particle swarm optimization, outputting the optimal charging and discharging strategy.

[0039] In one embodiment of the invention, a reliability-driven adaptive particle swarm optimization algorithm is used to perform a global search on the dynamic equilibrium optimization model, including: S311 encodes a particle as a whole scheduling sequence and obtains the particle's velocity and position.

[0040] In this embodiment, particle encoding and updating: a particle is represented as an entire scheduling sequence. Speed ​​and position updated as follows: (17); (18); In the formula, This represents the number of iterations in the particle swarm optimization algorithm. ; For the first The particle in the first The velocity (vector) at the next iteration; For the first The particle in the first The position (vector) at the next iteration is encoded as a complete scheduling sequence; For the first The inertia weight of the next iteration; , The first Two learning factors in the next iteration; , for Random numbers; This represents the optimal position for an individual particle. This is the global optimal position for the particle.

[0041] 312, defines particle population diversity and reliability default degree.

[0042] In this embodiment, to reflect "reliability-driven" characteristics, particle population diversity and reliability default levels are represented by the following formula: ,(19) (20); In the formula, For the first Particle population diversity characterization during the next iteration. For the total particle population, For the first During the nth iteration, the 1st The entire scheduling sequence of each particle. For the first The entire scheduling sequence during the next iteration. It is a 2-norm; For the first Degree of reliability breach during the next iteration For the first Summary of power shortage during the next iteration The threshold for the reliability index is based on a summary of power shortages. For the first The battery level in the next iteration was lower than expected. The threshold for the reliability index is set with the expected value of insufficient power.

[0043] S313, the inertia weight and learning factor of the adaptive particle swarm algorithm are jointly adjusted based on the diversity of particle population and the degree of reliability default.

[0044] In this embodiment, adaptive setting: when When the value is large, it improves the overall exploration ability. Decline and When ≈0, the convergence speed is increased, for example: ,(twenty one); ,(twenty two); ,(twenty three); In the formula, For the first The inertia weight of the next iteration, , These are the minimum and maximum inertia weights, respectively. Let be the base of the natural logarithm function. The inertial weighting exponential decay coefficient is... The maximum number of iterations, For reliability default feedback weighting coefficient, For the first The degree of reliability failure in the next iteration. , These are two learning factors in the adaptive particle swarm optimization algorithm. , Learning factors The maximum and minimum values, , Learning factors The maximum and minimum values, For the first Characterization of particle population diversity during the next iteration.

[0045] In this embodiment, At least including and When V(k) increases, the global search strength is increased to prioritize reliability recovery. Approaching 0 and Decrease And increase the convergence factor to speed up convergence.

[0046] In one embodiment of the present invention, SOC trajectory projection repair is performed, and SOC(t) is recursively calculated for each particle according to formula (15): The adaptive particle swarm algorithm performs feasible region projection repair on the particles after each position update to ensure that the power and SOC satisfy the constraints throughout the process; wherein, feasible region projection repair is expressed by the following formula: ,(twenty four); In the formula, For the period The repaired value after applying upper and lower limit constraints to the state of charge. The charge state is obtained from the particle position update before repair. The upper and lower limits of the charged state range. This refers to the period during which feasibility repairs are required.

[0047] The energy difference caused by this projection Feedback to subsequent periods The above (e.g., prioritizing charging during off-peak hours or discharging during peak hours) is corrected according to the principle of minimum disturbance: (25); This ensures that the modified SOC satisfies equation (16) and the end-of-cycle constraint SOC(T) = SOC(0). Where, For period For energy storage power sequence The minimum perturbation correction.

[0048] In this embodiment, the SOC trajectory is calculated sequentially over time after particle updates. When an SOC out-of-bounds error occurs, subsequent time periods are analyzed. Perform minimal modifications to the projection or mirror repair to allow the SOC to fall back into place. At the end of the scheduling period, the energy balance of the entire sequence is corrected to satisfy SOC(T)=SOC(0).

[0049] In one embodiment of the present invention, the global optimal solution obtained by particle swarm optimization is locally refined using the interior-point method, including: rewriting the energy storage constraints of the dynamic equilibrium optimization model as... The form involves using the global optimal solution obtained from the particle swarm optimization as the initial point, constructing a barrier function, and iteratively updating the barrier factor to obtain a locally optimal feasible solution; where the barrier function is expressed by the following formula: ,(26; In the formula, This is a barrier function used to incorporate inequality constraints into the objective function. The position vector of the particle The corresponding original objective function value, As a barrier factor, The position vector of the particle The One constraint It is the natural logarithm.

[0050] along with By gradually reducing the iterative solution, a locally optimal solution that satisfies the constraints can be obtained, thereby improving the accuracy and stability of the final solution.

[0051] In this embodiment, the energy storage charging and discharging power sequence for each time period within the output scheduling cycle is provided. And the corresponding peak shaving and valley filling effects and reliability indicators.

[0052] Please see Figure 2 As shown, the present invention also provides a system applying the above-described distributed energy storage system scheduling method, used to achieve a quantifiable dynamic balance between peak shaving and valley filling and improving power supply reliability, comprising: The reliability-driven energy storage rated parameter determination module is used to obtain the equivalent continuous load curve based on random production simulation, and determine the rated power and rated capacity of the energy storage system by using the target reliability index threshold of power shortage probability or power shortage expectation value as a constraint.

[0053] The dynamic balance optimization module is used to construct a dynamic balance optimization model that simultaneously considers load smoothing, time-of-use electricity price costs, and reliability risks, using the energy storage charging and discharging power of each time period as the decision variable.

[0054] The optimal charging and discharging strategy module is used to perform a global search on the dynamic equilibrium optimization model using a reliability-driven adaptive particle swarm optimization algorithm, and then refine the global optimal solution obtained by the particle swarm optimization using an interior point method to output the optimal charging and discharging strategy.

[0055] This invention also provides an example: by Taking day-ahead scheduling as an example. Inputs include: load forecast curve. Renewable Energy Prediction Time-of-use electricity pricing Energy storage parameters and reliability threshold ( or ) and confidence level .

[0056] The output includes: energy storage charging and discharging power sequence , Trajectory, peak shaving and valley filling effect indicators (such as) or peak-valley difference Electricity cost and reliability indicators .

[0057] The peak-valley difference is calculated as follows: (27); If it is necessary to characterize the relative improvement, it can be calculated as follows: , , (28); in For indicators under no energy storage or benchmark strategies, , , These are the weighting coefficients for the peak shaving objective, the economic cost objective, and the reliability objective, respectively, used to adjust the relative importance of each objective in the overall objective function.

[0058] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention, and no reference numerals in the claims should be construed as limiting the scope of the claims.

[0059] The above embodiments are merely examples of implementation methods of the invention. The scope of protection of the present invention is not limited to the above embodiments. For those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all fall within the scope of protection of the present invention.

Claims

1. A method for scheduling a distributed energy storage system, characterized in that, This is used to achieve a quantifiable dynamic balance between peak shaving and valley filling and improving power supply reliability, including: The equivalent continuous load curve is obtained based on random production simulation, and the rated power and rated capacity of the energy storage system are determined by the target reliability index threshold of the power shortage probability or the expected value of power shortage as a constraint. Using the energy storage charging and discharging power in each time period as the decision variable, a dynamic balance optimization model is constructed that simultaneously considers load smoothing, time-of-use electricity price costs, and reliability risks. A reliability-driven adaptive particle swarm optimization algorithm is used to perform a global search on the dynamic equilibrium optimization model. Then, the interior point method is used to refine the global optimal solution obtained by the particle swarm optimization and output the optimal charging and discharging strategy.

2. The distributed energy storage system scheduling method according to claim 1, characterized in that, The rated power and rated capacity of an energy storage system are determined using the following formulas: ； ,or ; In the formula, Let the objective function be the scale of the energy storage system. , These are two trade-off coefficients, The rated power of the energy storage system, For the rated capacity of the energy storage system, In configuration The probability of insufficient power at that time Its threshold; For configuration The battery level was lower than expected. Its threshold.

3. The distributed energy storage system scheduling method according to claim 1, characterized in that, The dynamic equilibrium optimization model is expressed by the following formula: ； In the formula, For dynamic equilibrium optimization model, , , These are the multi-objective weighting coefficients. For the load variance, for Net load during the cycle, for Equivalent charge / discharge power during the cycle, For time-of-use electricity costs, This poses a reliability risk.

4. The distributed energy storage system scheduling method according to claim 1, characterized in that, Reliability risks include the summary penalty for insufficient power, the expected penalty for insufficient power, or the conditional risk value of insufficient power. Charge and discharge power constraints, SOC dynamic constraints, SOC upper and lower limit constraints, and SOC regression constraints at the end of the cycle are set: SOC(T)=SOC(0); where T is the scheduling cycle.

5. The distributed energy storage system scheduling method according to claim 1, characterized in that, A reliability-driven adaptive particle swarm optimization algorithm is used for global search of the dynamic equilibrium optimization model, including: Encode a particle as an entire scheduling sequence and obtain the particle's velocity and position; Define particle population diversity and reliability default level; The inertia weights and learning factors of the adaptive particle swarm optimization algorithm are jointly adjusted based on particle swarm diversity and reliability default levels.

6. The distributed energy storage system scheduling method according to claim 5, characterized in that, The degree of default on particle population diversity and reliability is expressed by the following formula: ； ； In the formula, For the first Particle population diversity characterization during the next iteration. For the total particle population, For the first During the nth iteration, the 1st The entire scheduling sequence of each particle. For the first The entire scheduling sequence during the next iteration. It is a 2-norm; For the first Degree of reliability breach at the next iteration For the first Summary of power shortage during the next iteration The threshold for the reliability index is based on a summary of power shortages. For the first The battery level in the next iteration was lower than expected. The threshold for the reliability index is set with the expected value of insufficient power.

7. The distributed energy storage system scheduling method according to claim 6, characterized in that, The inertia weights and learning factors of the adaptive particle swarm optimization algorithm are jointly adjusted based on particle population diversity and reliability default levels, as expressed by the following formula: ； ； ； In the formula, For the first The inertia weight of the next iteration, , These are the minimum and maximum inertia weights, respectively. Let be the base of the natural logarithm function. The inertial weighting exponential decay coefficient is... The maximum number of iterations, For reliability default feedback weighting coefficient, For the first The degree of reliability failure in the next iteration. , These are two learning factors in the adaptive particle swarm optimization algorithm. , Learning factors The maximum and minimum values, , Learning factors The maximum and minimum values, For the first Characterization of particle population diversity during the next iteration.

8. The distributed energy storage system scheduling method according to claim 1, characterized in that, The adaptive particle swarm optimization algorithm performs feasible region projection repair on particles after each position update to ensure that power and SOC satisfy the constraints throughout the process; the feasible region projection repair is expressed by the following formula: ； In the formula, For the period The corrected value after applying upper and lower limit constraints to the state of charge at that time. The charge state is obtained from particle position updates before repair. The upper and lower limits of the charged state range. This refers to the period during which feasibility repairs are required.

9. The distributed energy storage system scheduling method according to claim 1, characterized in that, The global optimal solution obtained from particle swarm optimization is locally refined using the interior-point method, including: rewriting the constraints of the dynamic equilibrium optimization model as... In this approach, the global optimal solution obtained from the particle swarm optimization is used as the initial point to construct a barrier function. The local optimal feasible solution is obtained by iteratively updating the barrier factor. The barrier function is expressed by the following formula: ； In the formula, For the barrier function, The position vector of the particle The corresponding original objective function value, As a barrier factor, The position vector of the particle The One constraint It is the natural logarithm.

10. A system applying the distributed energy storage system scheduling method according to any one of claims 1-9, characterized in that, This is used to achieve a quantifiable dynamic balance between peak shaving and valley filling and improving power supply reliability, including: The reliability-driven energy storage rated parameter determination module is used to obtain the equivalent continuous load curve based on random production simulation, and determine the rated power and rated capacity of the energy storage system by using the target reliability index threshold of power shortage probability or power shortage expectation value as a constraint. The dynamic balance optimization module is used to construct a dynamic balance optimization model that simultaneously considers load smoothing, time-of-use electricity price costs, and reliability risks, using the energy storage charging and discharging power of each time period as the decision variable. The optimal charging and discharging strategy module is used to perform a global search on the dynamic equilibrium optimization model using a reliability-driven adaptive particle swarm optimization algorithm, and then refine the global optimal solution obtained by the particle swarm optimization using an interior point method to output the optimal charging and discharging strategy.

Images