Cooperative scheduling method and system for distributed energy storage system

Abstract

The invention belongs to the technical field of energy storage system scheduling, and provides a distributed energy storage system cooperative scheduling method and system, and the method comprises the steps: constructing an energy storage unit refined dynamic model containing coulombic efficiency and rain flow counting method life attenuation, building an SOC evolution equation based on discrete time step length, and setting charge and discharge mutual exclusion constraint and capacity limitation. A power distribution network security constraint model is established based on a Distflow equation and a second-order cone relaxation technology, power flow and voltage constraints are processed, a multi-agent game framework is constructed in combination with a master-slave game model and a Nash bargaining model, a decision interaction process is expressed, and a risk perception mechanism is introduced to carry out income distribution. An improved ADMM and a branch and bound method are adopted to cooperatively solve a mixed integer nonlinear problem, and mixed integer scheduling optimization of a multi-node energy storage unit is carried out. According to the invention, group collaborative scheduling of the distributed energy storage system is realized, and an extensible and highly reliable scheduling solution is provided for a high-proportion renewable energy power grid.

Description

Technical Field

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

[0002] With the increasing penetration of renewable energy, energy storage systems are playing an increasingly crucial role in the power system. However, existing technologies have the following shortcomings:

[0003] 1. Coarse dynamic model: Most studies only use linear SOC accumulation model, ignoring battery life decay, charge-discharge mutual exclusion constraints and coulomb efficiency, resulting in large prediction errors and scheduling results deviating from actual operation.

[0004] 2. Simplification of power grid constraints: The distribution network is often approximated as an "ideal bus" with negligible reactive power and fixed node voltage. This cannot guarantee the voltage qualification rate under high power dispatch. The power flow and voltage constraints of the distribution network are not modeled in detail, which poses a safety hazard.

[0005] 3. Lack of multi-entity collaboration: Multi-entity energy storage clusters usually distribute benefits according to "average sharing" or "first come, first served", lacking risk compensation and differentiated preference modeling, lacking effective incentive mechanisms and distributed solution algorithms, and making it difficult to balance individual benefits with the overall system goals.

[0006] 4. Poor algorithm scalability: The nonlinear scale of mixed integers expands exponentially with the number of nodes; the solution time of centralized MILP / MINLP is longer than hours when N>50, and the Chinese method is difficult to cope with the computation and communication pressure of large-scale scenarios. Summary of the Invention

[0007] The purpose of this invention is to overcome the existing defects and provide a method and system for collaborative scheduling of distributed energy storage systems.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0009] A method for coordinated scheduling of distributed energy storage systems includes:

[0010] Refined dynamic modeling: Construct a refined dynamic model of the energy storage unit that includes Coulomb efficiency and rainflow counting method lifetime decay, calculate the dispatchable power X_i of the energy storage unit at time t, and provide the accurate state variables and cost functions required for subsequent pricing and order generation;

[0011] Power grid security constraint modeling: Based on the Distflow equation and second-order cone relaxation technique, a distribution network security constraint model is established to handle power flow and voltage constraints and generate nodal marginal prices or incentive price signals;

[0012] Multi-agent game and order generation: A multi-agent game framework is constructed by combining the master-slave game model and the Nash bargaining model to represent the decision-making interaction and revenue distribution process between the dispatch center and the energy storage unit, and a risk perception mechanism is introduced to compensate risk-averse subjects; the energy storage unit generates transaction orders containing multi-dimensional attributes based on the electricity price signal;

[0013] Order matching: Orders are matched through a multi-dimensional auction or matching mechanism to generate transaction volume and price information;

[0014] Blockchain Settlement and Evidence Storage: An improved ADMM and branch-and-bound method are used to solve mixed-integer nonlinear problems, realizing distributed collaborative scheduling optimization of multi-node energy storage units. Transaction data is stored using blockchain technology, and Merkle trees, sidechains, or sharding techniques are used to achieve system scalability and information confidentiality.

[0015] Furthermore, the refined dynamic model of the energy storage unit includes:

[0016] (1) SOC evolution equation:

[0017]

[0018] Among them, SOC i,t Let Δt be the state of charge of energy storage unit i at time t, and Δt be the discrete time step. Let represent the charging efficiency and discharging efficiency of energy storage unit i, respectively. Let represent the charging power and discharging power of energy storage unit i at time t, respectively. The rated capacity of energy storage unit i;

[0019] (2) Charge / discharge mutual exclusion constraints and capacity limitations:

[0020]

[0021] Among them, P i max,ch P i max,dis These represent the maximum allowable charging power and maximum discharging power of energy storage unit i, respectively; u i,t Indicates whether the device is currently in discharge mode, u i,t =1 indicates discharge, u i,t =0 indicates charging or idle; These represent the minimum and maximum allowable states of charge of energy storage unit i, respectively;

[0022] (3) Lifetime decay model:

[0023]

[0024] Where, N cycle The total number of cycles during the observation period, DOD k Let DOD be the depth of discharge cycle k. ref As a reference loop depth, γ i C is an index related to the properties of battery materials. ref For reference, the cost of capacity degradation Let t be the cost of battery loss due to charging and discharging operations.

[0025] Furthermore, the distribution network security constraint model includes:

[0026] (1) Distflow equation:

[0027]

[0028] Among them, P ij,t Q ij,t P represents the active power and reactive power from node i to node j at time t, respectively. ik,t Q ik,t Let r represent the active power and reactive power from node i to its parent node k at time t, respectively; ij x ij Let l represent the resistance and reactance of line (i,j) respectively; ij,t V represents the square of the current in line (i,j). i,t v j,t Let be the squares of the voltages at nodes i and j, respectively; These represent the active power and reactive power of the load at node j, respectively. Let represent the active power and reactive power injected by the distributed power source at node j, respectively; C(j) represents the set of child nodes of node j. This represents the line loss term, used to correct the voltage drop model;

[0029] (2) Second-order cone relaxation constraint:

[0030] ||2P ij,t 2Q ij,t ,l ij,t -v j,t ||2≤l ij,t +v i,t

[0031] And limit node voltage:

[0032] v min ≤v j,t ≤v max

[0033] v min vmax These represent the minimum and maximum allowable values ​​for the node voltage, respectively.

[0034] Furthermore, in the master-slave game model, the dispatch center, as the leader, issues incentive electricity prices, while the energy storage unit, as the follower, maximizes its own profits.

[0035] Dispatch Center:

[0036]

[0037] λ t The incentive electricity price or subsidy coefficient at time t is set according to the operational objectives; P grid,t Let α be the net power purchased from or sold to the grid at time t; α is the penalty factor used to suppress the target power. The deviation.

[0038] Energy storage unit:

[0039]

[0040] Furthermore, the Nash bargaining model is expressed as follows:

[0041]

[0042] Among them, ∏ i For the final benefit of energy storage unit i, For the bottom-line return of energy storage unit i, The target value for the overall system revenue;

[0043] The risk perception mechanism is represented as follows:

[0044]

[0045] β i Risk aversion coefficient; CVaR α (∏ i α represents the conditional value of risk at confidence level α, which measures the expected loss in the worst-case scenario.

[0046] Furthermore, the improved ADMM algorithm step (1) initialization:

[0047] λ (0) =0, ρ=1.2, k=0

[0048] x i Let represent the decision vector of energy storage unit or node i; λ is the dual variable; ρ is the penalty coefficient; and k is the number of iterations.

[0049] (2) Local optimization:

[0050]

[0051] Each node solves the subproblem in parallel locally, combining it with its own objective function f. i (x i ) and constrained projection (||A) i x i +…-b) items;

[0052] (3) Dual update:

[0053]

[0054] The dual variable λ is modified based on the new solutions of each subproblem in order to balance the global coupling constraints;

[0055] (4) Convergence criterion:

[0056]

[0057] If the original residual ||r (k) ||and dual residual||s (k) If all values ​​are within the given threshold, the algorithm is considered to have converged.

[0058] Furthermore, the branch and bound method includes:

[0059] Branching: In the branch and bound framework, a mutual exclusion variable is divided into two branches:

[0060] or

[0061] Bounding: Relax each subproblem to obtain a lower bound. If the lower bound is greater than the current known optimal upper bound, prune the branch in advance.

[0062] Pruning: Remove branches that have no solution or cannot lead to a better solution, reducing the size of the search tree; at the same time, solve multiple subproblems and exchange information in parallel.

[0063] Another object of the present invention is to provide a distributed energy storage system collaborative scheduling system, comprising:

[0064] The refined dynamic modeling module is used to construct a refined dynamic model of the energy storage unit that includes Coulomb efficiency and rainflow counting method lifetime decay, calculate the dispatchable power X_i of the energy storage unit at time t, and provide the accurate state quantities and cost functions required for subsequent pricing and order generation.

[0065] The power grid security constraint modeling module is used to establish a distribution network security constraint model based on the Distflow equation and second-order cone relaxation technique, handle power flow and voltage constraints, and generate nodal marginal prices or incentive price signals.

[0066] The multi-agent game and order generation module is used to construct a multi-agent game framework by combining the master-slave game model and the Nash bargaining model. It represents the decision-making interaction and revenue distribution process between the dispatch center and the energy storage unit, and introduces a risk perception mechanism to compensate risk-averse agents. The energy storage unit generates transaction orders containing multi-dimensional attributes based on the electricity price signal.

[0067] The order matching module is used to match orders through a multi-dimensional auction or matching mechanism to generate transaction volume and price information;

[0068] The blockchain settlement and evidence storage module is used to solve mixed integer nonlinear problems in collaboration with the improved ADMM and branch and bound method, realize the distributed collaborative scheduling optimization of multi-node energy storage units, and store transaction data using blockchain technology. Merkle tree, sidechain or sharding technology is used to achieve system scalability and information confidentiality.

[0069] Another object of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the distributed energy storage system collaborative scheduling method provided by the first object of the present invention.

[0070] Another object of the present invention is to provide a server comprising at least one processor and a memory communicatively connected to the processor, the memory storing instructions executable by the at least one processor, the instructions being executed by the processor to cause the at least one processor to perform the distributed energy storage system collaborative scheduling method provided in the first object of the present invention.

[0071] In combination with the above technical solutions, the beneficial effects of the present invention compared with the prior art are as follows:

[0072] Refined Dynamic Modeling: To address the problems of large prediction errors and scheduling deviations from reality caused by traditional SOC models neglecting Coulomb efficiency, lifetime decay, and charge-discharge mutual exclusion constraints, this invention introduces Coulomb efficiency, a rainflow counting-based lifetime decay model, and charge-discharge mutual exclusion constraints to establish a refined SOC evolution equation. This scheme reduces SOC prediction errors, improves the quantification accuracy of battery lifetime decay, and thus enhances the consistency between scheduling results and actual operation.

[0073] Power Grid Safety Constraint Modeling: Addressing the issue that existing dispatching methods neglect power flow and voltage constraints in distribution networks, leading to potential voltage limit exceedances, this invention, based on the Distflow power flow equations and incorporating second-order cone relaxation (SOCP) technology, transforms nonlinear voltage constraints into convex constraints and sets upper and lower limits for node voltages. This scheme significantly improves node voltage compliance rates and enhances system stability and dispatch feasibility under high-power dispatching.

[0074] Multi-Agent Game Theory Framework Construction: Addressing the issues of insufficient coordination mechanisms and unfair revenue distribution in existing distributed energy storage systems, this invention combines a leader-follower game model with a Nash bargaining model, introducing a CVaR-based risk perception mechanism. This scheme, on the one hand, enables energy storage units to autonomously respond to price signals through leader-follower game theory, improving group coordination efficiency; on the other hand, it guarantees individual bottom-line returns through bargaining and risk compensation, achieving fair and differentiated revenue distribution.

[0075] Distributed Solution: To address the high computational complexity and difficulty in real-time solution of centralized methods in large-scale energy storage systems, this invention combines an improved ADMM parallel distributed algorithm with the branch and bound method to efficiently and collaboratively solve mixed-integer nonlinear problems. This scheme effectively reduces computational complexity, improves solution accuracy and convergence speed, and significantly shortens optimization time, thus possessing good real-time performance and scalability.

[0076] In summary, this invention represents a technological leap from single-unit optimization to group-based collaborative scheduling in distributed energy storage systems. Refined battery modeling improves scheduling accuracy, grid security constraints ensure system stability, game theory mechanisms optimize multi-stakeholder collaboration efficiency and revenue distribution, and distributed algorithms address the real-time solvability of large-scale problems. Ultimately, it provides a scalable and highly reliable group-based collaborative scheduling solution for high-proportion renewable energy grids.

[0077] This invention achieves a technological leap from single-unit optimization to group collaboration in distributed energy storage systems, providing a scalable and highly reliable dispatch solution for high-proportion renewable energy grids. Attached Figure Description

[0078] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0079] Figure 1 This is a flowchart of the distributed energy storage system collaborative scheduling method provided in the embodiments of the present invention;

[0080] Figure 2 This is a schematic diagram of the structure of the distributed energy storage system collaborative scheduling system provided in the embodiment of the present invention. Detailed Implementation

[0081] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0082] Example 1:

[0083] like Figure 1The diagram shows an embodiment of the distributed energy storage system collaborative scheduling method provided by the present invention, which specifically includes the following steps:

[0084] S1: Refined Dynamic Modeling: Construct a refined dynamic model of the energy storage unit that includes Coulomb efficiency and rainflow counting method lifetime decay, calculate the dispatchable power X_i of the energy storage unit at time t, and provide the accurate state quantities and cost functions required for subsequent pricing and order generation.

[0085] S2: Power grid security constraint modeling: Based on the Distflow equation and second-order cone relaxation technique, a distribution network security constraint model is established to handle power flow and voltage constraints and generate nodal marginal price or incentive price signals.

[0086] S3: Multi-agent game and order generation: A multi-agent game framework is constructed by combining the master-slave game model and the Nash bargaining model to represent the decision-making interaction and revenue distribution process between the dispatch center and the energy storage unit, and a risk perception mechanism is introduced to compensate risk-averse subjects; the energy storage unit generates transaction orders containing multi-dimensional attributes based on the electricity price signal;

[0087] S4: Order Matching: Match orders through a multi-dimensional auction or matching mechanism to generate transaction volume and price information;

[0088] S5: Blockchain Settlement and Evidence Storage: An improved ADMM and branch-and-bound method are used to solve mixed integer nonlinear problems in a collaborative manner, realizing the distributed collaborative scheduling optimization of multi-node energy storage units. Transaction data is stored using blockchain technology, and Merkle trees, sidechains, or sharding techniques are used to achieve system scalability and information confidentiality.

[0089] The invention will be further explained below in conjunction with the model construction process of each step.

[0090] 1. Dynamic model of energy storage unit:

[0091] 1.1 SOC Evolution Equation:

[0092] To accurately describe the dynamic characteristics of energy storage units (such as batteries), charge / discharge efficiency and capacity upper limit are considered, and the following equations are established using a discrete time step Δt:

[0093]

[0094] SOC i,t : The state of charge of energy storage unit i at time t, which is usually in the range of [0,1] or can be expressed as a percentage.

[0095] Δt: Discrete time step (e.g., 15 minutes or 1 hour).

[0096] These represent the charging efficiency and discharging efficiency of energy storage unit i, respectively (typical values ​​are between 0.92 and 0.98).

[0097] The charging and discharging power of energy storage unit i at time t (unit: kW or MW).

[0098] Rated capacity of energy storage unit i (unit: kWh or MWh).

[0099] Mutual Exclusion Constraints and Capacity Limits

[0100]

[0101] P i max,ch ,P i max,dis : The maximum allowable charging power and maximum discharge power of energy storage unit i.

[0102] u i,t : A 0-1 binary variable used to indicate whether the current time is in discharge mode (u i,t =1 indicates discharging, 0 indicates charging or idle), thus ensuring that charging and discharging cannot occur simultaneously.

[0103] The minimum and maximum allowable SOC of energy storage unit i are set to avoid overcharging or over-discharging.

[0104] 1.2 Lifetime Decay Model:

[0105] Since battery performance degrades during multiple charge-discharge cycles, this degradation needs to be converted into economic costs and incorporated into scheduling optimization. Therefore, rainflow counting is used to quantify cycle losses.

[0106]

[0107] N cycle Total number of cycles during the observation period.

[0108] DOD k : Depth of Discharge (k-th discharge cycle).

[0109] DOD ref : Reference loop depth; γ i An index related to the characteristics of battery materials (e.g., 2.1 for lithium batteries).

[0110] C refReference capacity degradation cost (unit: $ / kWh or € / kWh).

[0111] The battery loss cost incurred at time t due to charging and discharging operations can be regarded as a "virtual cost" added to the scheduling objective to suppress excessively frequent or deep discharge behavior.

[0112] 2. Power Grid Security Constraint Modeling

[0113] 2.1 Distflow power flow equations:

[0114] For power flow and voltage distribution in distribution networks (which are typically tree-like or near-tree-like structures), the Distflow model can be used:

[0115]

[0116] P ij,t Q ij,t Active and reactive power from node i to node j at time t;

[0117] P ik,t Q ik,t : Active and reactive power from node i to its parent node k at time t;

[0118] r ij ,x ij Resistance and reactance of line (i,j);

[0119] l ij,t : The square of the current in line (i,j);

[0120] v i,t ,v j,t : The square of the voltage at node i and node j;

[0121] The active and reactive power of the load at node j;

[0122] The active and reactive power injected into the distributed power source at node j;

[0123] C(j): The set of child nodes of node j;

[0124] The line loss term is used to correct the voltage drop model.

[0125] 2.2 Convex Relaxation Process (SOCP):

[0126] To transform the aforementioned nonlinear constraints into a solvable convex problem, the second-order cone relaxation (SOCP) technique can be used:

[0127] ||2P ij,t 2Q ij,t ,l ij,t -v j,t ||2≤l ij,t +v i,t

[0128] At the same time, node voltages are limited:

[0129] v min ≤v j,t ≤v max

[0130] The above equation means that, within the range of physical feasibility, the originally non-convex power flow constraints are simplified into conical constraints, making them easier to solve with convex optimization tools.

[0131] v min ,v max : Specifies the minimum and maximum allowable values ​​for node voltages (typically [0.95]). 2 1.05 2 ]wait).

[0132] 3. Master-Slave Game Model:

[0133] When there is a dispatch center (leader) and multiple energy storage units (followers), the Stackelberg game can be used to describe their decision-making interaction process.

[0134] 3.1 Leader Issues (Dispatch Center):

[0135]

[0136] λ t The incentive price or subsidy coefficient at time t can be set by the dispatch center according to the operational objectives.

[0137] P grid,t Net power purchased from (or sold to) the grid at time t.

[0138] α: Penalty factor, used to suppress interference with target power. The penalty factor α is a non-negative real number. If α = 0, it means the dispatch center does not consider the deviation constraint from the target power; if α > 0, the penalty for deviation from the target power increases with the increase of the value of α. To ensure the physical rationality and numerical stability of the optimization objective, the value of α should not be less than zero, and should be related to the electricity price coefficient λ. t The magnitude of the values ​​should be kept consistent to achieve a balance between economy and operational stability. Specific values ​​can be set based on system operational requirements and experimental optimization results.

[0139] 3.2 Follower Problem (Energy Storage Unit):

[0140]

[0141] It also follows the mutual exclusion charge / discharge constraints in (1.1) and the lifetime loss model in (1.2).

[0142] Energy storage units are priced according to the leader's announcement. t It determines the charging / discharging power at each moment to maximize its own benefits (electricity price difference) minus the cost of lifespan degradation.

[0143] 4. Nash Bargaining Model:

[0144] In multi-party cooperative game theory, each energy storage unit or stakeholder may need to negotiate the distribution of overall revenue to ensure that individual revenue does not fall below their respective bottom lines. This process can be characterized using the Nash bargaining model.

[0145]

[0146] ∏ i : The final benefit of energy storage unit i.

[0147] The bottom-line revenue of energy storage unit i (which can be understood as the retained revenue of "not participating in cooperation").

[0148] The overall revenue target for the system.

[0149] Risk Sensitive Extension (CVaR): To account for revenue volatility or uncertainties in electricity prices and loads, the CVaR indicator can be introduced.

[0150]

[0151] Where β i Risk aversion coefficient; CVaR α (∏ i Let α represent the conditional value of risk at confidence level α, measuring the expected loss in the worst-case scenario. Within this bargaining framework, the objective function or constraints can be adjusted accordingly to reflect the risk preferences of different entities.

[0152] 5. Improved ADMM algorithm:

[0153] To efficiently solve the above model (including power grid constraints and mixed integer variables) in a distributed environment, an improved alternating direction multiplier method (ADMM) can be used. Algorithm steps:

[0154] (1) Initialization

[0155] λ (0)=0, ρ=1.2, k=0

[0156] x i Let represent the decision vector of energy storage unit or node i; λ is the dual variable; ρ is the penalty coefficient; and k is the number of iterations.

[0157] (2) Local optimization

[0158]

[0159] Each node solves the subproblem in parallel locally, considering its own objective function f. i (x i ) and constrained projection (||A) i x i +…-b) items.

[0160] (3) Dual update

[0161]

[0162] The dual variable λ is modified based on the new solutions of each subproblem in order to balance the global coupling constraints.

[0163] (4) Convergence determination

[0164]

[0165] If the original residual ||r (k) ||and dual residual||s (k) If all values ​​are within the given threshold, the algorithm is considered to have converged.

[0166] 6. Mixed Integer Solving Strategy:

[0167] For energy storage charging and discharging mutual exclusion constraints (i.e.) ) and 0-1 variable u i,t The complexity of mixed integer programming (MIP) needs to be addressed during the optimization process. The following steps can improve the solution efficiency:

[0168] (1) Branch: In the branch and bound framework, a mutual exclusion variable is divided into two branches:

[0169] or

[0170] Equivalent to a fixed binary variable u i,t The value of .

[0171] (2) Bound: Relax each branch subproblem (such as relaxing some binary or SOCP constraints) to obtain a lower bound; if the lower bound is greater than the current known optimal upper bound, the branch can be pruned in advance.

[0172] (3) Pruning: Remove branches that have no solution or cannot bring a better solution, reducing the size of the search tree; at the same time, multiple sub-problems can be solved and information exchanged in parallel, improving overall efficiency.

[0173] By combining with distributed methods such as ADMM, it is possible to achieve hybrid integer scheduling optimization of multi-node energy storage units in large-scale scenarios, balancing solution accuracy and scalability.

[0174] 7. The distributed energy storage system collaborative scheduling method of the present invention forms a complete "unattended energy storage supermarket" conveyor belt:

[0175] It achieves a closed-loop process from dynamic modeling, power grid constraints, game optimization to distributed solution and blockchain settlement.

[0176] The corresponding steps for each model construction are explained below:

[0177] (1) Refined dynamic modeling (corresponding to X) i calculate):

[0178] Model: The SOC evolution equation is established based on the discrete time step, considering the Coulomb efficiency η. c ,η d Capacity limitations [SOC] min SOC max In addition, charging and discharging mutual exclusion constraints are used; combined with the rainflow counting method lifetime decay model, the cycle loss cost is quantified.

[0179] Function: Calculate the dispatchable energy X of the energy storage unit at time t. i It also provides the accurate state quantities and cost functions required for subsequent pricing and order generation.

[0180] (2) Power grid security constraint modeling (corresponding to π dynamic pricing):

[0181] Model: The Distflow power flow equation is used to characterize the power and voltage distribution of the distribution network. Second-order cone relaxation (SOCP) is introduced to transform the non-convex power flow constraint into a convex constraint; allowable node voltage ranges are set [V]. min V max ].

[0182] Function: Under the premise of ensuring the safe operation of the power grid, form the nodal marginal price or incentive price π, and transmit it to the energy storage unit as a market signal.

[0183] (3) Multi-agent game framework (corresponding to writing order Ω → pairing (E) tx ,π tx )):

[0184] Model: Employing a master-slave game model (Stackelberg), with the scheduling center acting as the leader and setting the price π. t The energy storage unit, as a follower, selects the optimal charging and discharging power to maximize its own benefits; at the same time, the Nash bargaining model is introduced to allocate the overall benefits, and the risk perception mechanism (CVaR) is combined to compensate the risk-averse subjects.

[0185] Function: The energy storage unit generates an order Ω based on the price π, completes the matching under a multi-dimensional auction / matching mechanism, and outputs the traded electricity volume E. tx π of the transaction price tx .

[0186] (4) Distributed solution (corresponding to on-chain settlement → evidence storage & scaling):

[0187] Model: The improved ADMM algorithm is used for distributed parallel optimization. The steps include initialization, local subproblem solving, dual update and convergence determination. The branch and bound method is combined to handle mixed integer constraints to ensure optimality. At the same time, a blockchain notarization mechanism is introduced to write the transaction hash into the main chain and to achieve expansion through Merkle tree or side chain technology.

[0188] Function: Completes hybrid integer nonlinear scheduling optimization for multi-node energy storage groups, and stores the settlement results on the blockchain to ensure that transactions are transparent, tamper-proof, and support large-scale expansion.

[0189] As shown in the table below, the distributed energy storage system collaborative scheduling method of this invention can be abstracted as an "unattended energy storage supermarket" production line. First, the battery manager inventories the remaining capacity, the price calculator dynamically sets prices accordingly, the order writer generates multi-dimensional orders with conditions, the multi-dimensional auction completes matching and determines the winning triplet based on price / time limit / power priority, the on-chain cash register completes the synchronous settlement of electricity and funds through the atomic function `settle(·)`, and finally, the audit archive achieves traceability, scalability, and confidentiality based on the blockchain's evidence storage and expansion mechanism. This process realizes closed-loop scheduling from physical modeling, grid constraints, game theory optimization to distributed solution and trusted settlement.

[0190] A quick overview of how data flow and model work together:

[0191]

[0192] Plain explanation:

[0193] Battery Manager (7.1) first checks the inventory: How much electricity can still be sold, and how much electricity needs to be purchased in 30 minutes? It then generates the "Remaining Balance Bill" X. i .

[0194] The price calculator (7.2) calculates the total inventory balance by comparing it with the previous one. If there is more inventory, the price will be reduced; if there is less inventory, the price will be increased. It then gives today's benchmark price π.

[0195] The order writer (7.3) wrote "I want to sell 150kWh of electricity within 20 minutes using ≤40kW, at a price ≥0.62 yuan" as a single line of five-dimensional order.

[0196] Multi-dimensional auction (7.4) queues up by price, and determines who wins the auction, how much is won, and the price by considering power and time limit gates.

[0197] The on-chain cash register (7.5) deducts money from the buyer and pays the seller in one transaction and immediately prints a receipt on the chain.

[0198] The Audit Archives (7.6) packages all receipt hashes and pastes them onto the main chain, allowing external regulators to verify their authenticity in just a few steps; when the volume is large, it can be moved to the side chain, and privacy prices can be encrypted using ZK proofs.

[0199] 8. Algorithm Complexity Analysis

[0200] Method comparison:

[0201] method Time complexity Communication overhead Convergence speed Centralized MPC (O(N^3)) high slow This article ADMM (O(N)) Low quick Distributed gradient method (O(N^2)) middle middle

[0202] Scalability testing:

[0203] Number of energy storage units Solution time (s) Number of iterations 10 3.2 8 50 4.1 10 100 5.7 12

[0204] The technical effects of the present invention will be explained below with reference to specific experiments.

[0205] 1. Energy storage clusters in industrial parks:

[0206] Parameter settings:

[0207]

[0208]

[0209] Experimental results:

[0210] Cost reduction: Daily electricity costs decreased from $1,256 to $1,123 (a 10.6% reduction), improving economic efficiency.

[0211] Peak load reduction: Peak load decreased from 2.8MW to 2.3MW (a reduction of 17.9%).

[0212] Voltage qualification rate increased from 92.3% to 98.7%, improving safety.

[0213] 2. Microgrid coordinated dispatch

[0214] Nash bargaining parameters:

[0215]

[0216] Profit distribution:

[0217] Initial phase: Allocation based on capacity ratio (3:2:4).

[0218] Equilibrium solution: The payoff ratio is adjusted to 1.1:1.0:1.4, resulting in a 12% increase in total payoff.

[0219] Enhanced fairness: Risk appetite is customizable, and profit distribution meets Pareto improvement standards.

[0220] Example 2:

[0221] like Figure 2 As shown in the figure, the present invention provides a distributed energy storage system collaborative scheduling system, comprising:

[0222] The refined dynamic modeling module is used to construct a refined dynamic model of the energy storage unit that includes Coulomb efficiency and rainflow counting method lifetime decay, calculate the dispatchable power X_i of the energy storage unit at time t, and provide the accurate state quantities and cost functions required for subsequent pricing and order generation.

[0223] The power grid security constraint modeling module is used to establish a distribution network security constraint model based on the Distflow equation and second-order cone relaxation technique, handle power flow and voltage constraints, and generate nodal marginal prices or incentive price signals.

[0224] The multi-agent game and order generation module is used to construct a multi-agent game framework by combining the master-slave game model and the Nash bargaining model. It represents the decision-making interaction and revenue distribution process between the dispatch center and the energy storage unit, and introduces a risk perception mechanism to compensate risk-averse agents. The energy storage unit generates transaction orders containing multi-dimensional attributes based on the electricity price signal.

[0225] The order matching module is used to match orders through a multi-dimensional auction or matching mechanism to generate transaction volume and price information;

[0226] The blockchain settlement and evidence storage module is used to solve mixed integer nonlinear problems in collaboration with the improved ADMM and branch and bound method, realize the distributed collaborative scheduling optimization of multi-node energy storage units, and store transaction data using blockchain technology. Merkle tree, sidechain or sharding technology is used to achieve system scalability and information confidentiality.

[0227] Example 3: This embodiment of the invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements the distributed energy storage system collaborative scheduling method provided in Example 1 of the invention.

[0228] Example 4: This embodiment of the invention provides a server, including at least one processor and a memory communicatively connected to the processor. The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the processor to cause the at least one processor to execute the distributed energy storage system collaborative scheduling method provided in Example 1 of the invention.

[0229] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated in the present invention, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0230] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0231] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for collaborative scheduling of a distributed energy storage system, characterized in that, The method includes: Refined Dynamic Modeling: Construct a refined dynamic model of the energy storage unit that includes Coulomb efficiency and rainflow counting lifetime degradation, and calculate the dispatchable energy X of the energy storage unit at time t. i It also provides the accurate state quantities and cost functions required for subsequent pricing and order generation; Power grid security constraint modeling: Based on the Distflow equation and second-order cone relaxation technique, a distribution network security constraint model is established to handle power flow and voltage constraints and generate nodal marginal prices or incentive price signals; Multi-agent game and order generation: A multi-agent game framework is constructed by combining the master-slave game model and the Nash bargaining model to represent the decision-making interaction and revenue distribution process between the dispatch center and the energy storage unit, and a risk perception mechanism is introduced to compensate risk-averse subjects; the energy storage unit generates transaction orders containing multi-dimensional attributes based on the electricity price signal; Order matching: Orders are matched through a multi-dimensional auction or matching mechanism to generate transaction volume and price information; Blockchain Settlement and Evidence Storage: An improved ADMM and branch-and-bound method are used to solve mixed-integer nonlinear problems, realizing distributed collaborative scheduling optimization of multi-node energy storage units. Transaction data is stored using blockchain technology, and Merkle trees, sidechains, or sharding techniques are used to achieve system scalability and information confidentiality.

2. The distributed energy storage system collaborative scheduling method according to claim 1, characterized in that, The refined dynamic model of the energy storage unit includes: (1) SOC evolution equation: Among them, SOC i，t Let Δt be the state of charge of energy storage unit i at time t, and Δt be the discrete time step. Let represent the charging efficiency and discharging efficiency of energy storage unit i, respectively. Let represent the charging power and discharging power of energy storage unit i at time t, respectively. The rated capacity of energy storage unit i; (2) Charge / discharge mutual exclusion constraints and capacity limitations: in, These represent the maximum allowable charging power and maximum discharging power of energy storage unit i, respectively; u i，t Indicates whether the device is currently in discharge mode, u i，t =1 indicates discharge, u i，t =0 indicates charging or idle; These represent the minimum and maximum allowable states of charge of energy storage unit i, respectively; (3) Lifetime decay model: Where, N cycle The total number of cycles during the observation period, DOD k Let DOD be the depth of discharge cycle k. ref As a reference loop depth, γ i C is an index related to the properties of battery materials. ref For reference, the cost of capacity degradation Let t be the cost of battery loss due to charging and discharging operations.

3. The distributed energy storage system collaborative scheduling method according to claim 1, characterized in that, The power distribution network security constraint model includes: (1) Distflow equation: Among them, P ij，t Q ij，t P represents the active power and reactive power from node i to node j at time t, respectively. ik，t Q ik，t Let r represent the active power and reactive power from node i to its parent node k at time t, respectively; ij x ij Let l represent the resistance and reactance of line (i,j) respectively; ij，t V represents the square of the current in line (i, j). i,t v j,t Let be the squares of the voltages at nodes i and j, respectively; These represent the active power and reactive power of the load at node j, respectively. Let represent the active power and reactive power injected by the distributed power source at node j, respectively; C(j) represents the set of child nodes of node j. This represents the line loss term, used to correct the voltage drop model; (2) Second-order cone relaxation constraint: ||2P ij,t ,2Q ij,t ,l ij,t -in j,t ||2≤l ij,t +v i,t And limit node voltage: in min ≤in j,t ≤in max v min v max These represent the minimum and maximum allowable values ​​for the node voltage, respectively.

4. The distributed energy storage system collaborative scheduling method according to claim 2, characterized in that, In the master-slave game model, the dispatch center, as the leader, issues incentive electricity prices, while the energy storage unit, as the follower, maximizes its own benefits. Dispatch Center: λ t The incentive electricity price or subsidy coefficient at time t is set according to the operational objectives; P grid,t Let α be the net power purchased from or sold to the grid at time t; α is the penalty factor used to suppress the target power. Deviation; Energy storage unit:

5. The distributed energy storage system collaborative scheduling method according to claim 1, characterized in that, The Nash bargaining model is expressed as follows: Among them, ∏ i For the final benefit of energy storage unit i, For the bottom-line return of energy storage unit i, The target value for the overall system revenue; The risk perception mechanism is represented as follows: β i Risk aversion coefficient; CVaR α (∏ i α represents the conditional value of risk at confidence level α, which measures the expected loss in the worst-case scenario.

6. The distributed energy storage system collaborative scheduling method according to claim 1, characterized in that, The improved ADMM algorithm steps include: (1) Initialization: x i Let represent the decision vector of energy storage unit or node i; λ is the dual variable; ρ is the penalty coefficient; and k is the number of iterations. (2) Local optimization: Each node solves the subproblem in parallel locally, combining it with its own objective function f. i (x i ) and constrained projection (||A) i x i +…-b) items; (3) Dual update: The dual variable λ is modified based on the new solutions of each subproblem in order to balance the global coupling constraints; (4) Convergence criterion: If the original residual ||r (k) ||and dual residual||s (k) If all values ​​are within the given threshold, the algorithm is considered to have converged.

7. The distributed energy storage system collaborative scheduling method according to claim 1, characterized in that, The branch and bound method includes: Branching: In the branch and bound framework, a mutual exclusion variable is divided into two branches: or Bounding: Relax each subproblem to obtain a lower bound. If the lower bound is greater than the current known optimal upper bound, prune the branch in advance. Pruning: Remove branches that have no solution or cannot lead to a better solution, reducing the size of the search tree; at the same time, solve multiple subproblems and exchange information in parallel.

8. A distributed energy storage system collaborative scheduling system, characterized in that, The system includes: The refined dynamic modeling module is used to construct a refined dynamic model of the energy storage unit that includes Coulomb efficiency and rainflow counting method lifetime decay, calculate the dispatchable power X_i of the energy storage unit at time t, and provide the accurate state quantities and cost functions required for subsequent pricing and order generation. The power grid security constraint modeling module is used to establish a distribution network security constraint model based on the Distflow equation and second-order cone relaxation technique, handle power flow and voltage constraints, and generate nodal marginal prices or incentive price signals. The multi-agent game and order generation module is used to construct a multi-agent game framework by combining the master-slave game model and the Nash bargaining model. It represents the decision-making interaction and revenue distribution process between the dispatch center and the energy storage unit, and introduces a risk perception mechanism to compensate risk-averse agents. The energy storage unit generates transaction orders containing multi-dimensional attributes based on the electricity price signal. The order matching module is used to match orders through a multi-dimensional auction or matching mechanism to generate transaction volume and price information; The blockchain settlement and evidence storage module is used to solve mixed integer nonlinear problems in collaboration with the improved ADMM and branch and bound method, realize the distributed collaborative scheduling optimization of multi-node energy storage units, and store transaction data using blockchain technology. Merkle tree, sidechain or sharding technology is used to achieve system scalability and information confidentiality.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the distributed energy storage system collaborative scheduling method according to any one of claims 1 to 6.

10. A server, characterized in that: The system includes at least one processor and a memory communicatively connected to the processor, the memory storing instructions executable by the at least one processor, the instructions being executed by the processor to cause the at least one processor to perform the distributed energy storage system collaborative scheduling method as described in any one of claims 1 to 6.

Images