Power grid side energy storage two-stage planning method under sea wind access scene based on temperature-tide safety domain mapping

By employing a two-stage planning method based on temperature-power flow safety domain mapping, combined with LSTM, TCN, and Gaussian mixture models, a source-load coupling scenario is constructed to optimize energy storage site selection and capacity determination. This addresses the issue of unconsidered meteorological conditions in energy storage configuration, achieving more efficient energy storage configuration and voltage stability.

CN121769961APending Publication Date: 2026-03-31FUZHOU UNIV
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202512042331.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-31
Publication Date
2026-03-31

AI Technical Summary

Technical Problem

Existing studies on energy storage site selection and capacity determination do not fully consider changes in meteorological conditions, resulting in inaccurate assessment of changes in cable current carrying capacity, power flow safety domain of the grid, and node voltage levels. Furthermore, the lack of overall modeling of the source-load-environment coupling randomness of wind power, load, and temperature makes it difficult for energy storage configuration results to reflect actual operating characteristics.

Method used

A two-stage planning method based on temperature-power flow safety domain mapping is adopted. A temperature trend prediction model combining LSTM and TCN is used, and a Gaussian mixture model is used for temperature clustering. A source-load spatiotemporal coupled stochastic optimization scenario is constructed. The DistFlow multi-period optimal power flow model and second-order cone relaxation SOCP are combined, and the target cascade algorithm ATC is used for energy storage site selection and capacity optimization.

Benefits of technology

It enables more efficient use of cable network transmission capacity within the power flow safety domain, improves the adaptability and effectiveness of energy storage configuration, suppresses voltage fluctuations, reduces network losses, and provides a better energy storage configuration solution.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121769961A_ABST
    Figure CN121769961A_ABST
Patent Text Reader

Abstract

The invention relates to a power grid side energy storage two-stage locating and sizing method in a sea wind access scene considering the influence of temperature on a tide safety domain. The method comprises the following steps: adopting a temperature trend prediction model combining LSTM and TCN; a Gaussian mixture model is adopted to calculate a temperature variable capacitance expectation coefficient, and a power flow safety domain is reconstructed; establishing a random optimization scene set of source-load space-time coupling by adopting a typical source-load daily trajectory clustering method; establishing a power distribution network energy storage locating and sizing planning layer model taking the maximum comprehensive utility as a target; a power distribution network energy storage locating and sizing operation layer model with the minimum voltage fluctuation quadratic sum as the target is established; constructing a DistFlow multi-period optimal power flow model under the sea wind access scene; and solving the planning layer by adopting second-order cone relaxation, and carrying out secondary optimization on a charge-discharge and capacity configuration strategy on the operation layer by adopting a target cascade algorithm on the basis of anchoring a locating and sizing decision variable of the planning layer. According to the method, planning and operation targets can be more effectively coordinated, and comprehensive optimal configuration of power grid side energy storage is realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of power system planning and operation optimization technology, specifically to a two-stage planning method for grid-side energy storage in a sea breeze access scenario based on temperature-current safety domain mapping. Background Technology

[0002] Currently, China's offshore wind power industry is developing towards large-scale and clustered development. Compared to onshore wind power, offshore wind power output exhibits more significant intermittency, randomness, and low predictability. Therefore, supporting flexibility resources have become a key factor in ensuring system stability. New energy storage is an important support for improving system flexibility and renewable energy absorption capacity. To mitigate the uncertainty impact of offshore wind power integration, the grid side is accelerating the construction of energy storage power stations. On the one hand, this can enhance the self-regulation capability of wind farms against volatility; on the other hand, it can reduce grid absorption pressure and operational risks.

[0003] Current research on grid-side energy storage configuration mainly focuses on capacity configuration. Existing studies include a proposed long-term and short-term energy storage joint planning method that considers flexibility and supply-demand balance, and the establishment of a planning model with the minimum annualized comprehensive cost of the system as the optimization objective; another study constructs a new energy output-energy storage collaborative planning and configuration model based on the cascading of planning and operation objectives to solve for the minimum intraday comprehensive cost; and yet another study introduces system inertia and frequency security models to conduct energy storage configuration research with system security as the objective.

[0004] Some research has already been conducted on considering the coupling of severe weather conditions when configuring energy storage. Current research includes studies that propose a two-stage resilience enhancement planning model for distribution networks based on component failure rate models under the influence of compound disasters; studies that propose a photovoltaic-energy storage optimization configuration method based on an islanding model under extreme disasters; and studies that propose a distribution network cable fault repair and dual-timescale energy storage planning enhancement model under extreme high temperatures to optimize energy storage configuration strategies.

[0005] Existing research on energy storage site selection and capacity determination still suffers from several shortcomings in terms of methodology and scenario construction. First, most studies fail to adequately incorporate meteorological changes into distribution network planning models, neglecting the impact of intraday and seasonal temperature fluctuations on cable current-carrying capacity, grid power flow safety domain, and node voltage levels. This makes it impossible to accurately assess the operating conditions of the same node under different ambient temperature conditions. Second, the marginal contribution and strategies of energy storage in ancillary services such as peak shaving, frequency regulation, and voltage support largely depend on the grid's power flow feasible domain. However, in traditional planning, the differences in energy storage contributions under different meteorological scenarios are often simplified, making it difficult for site selection and capacity determination results to accurately reflect actual operating characteristics. Furthermore, existing research is mostly based on static typical days or simplified scenarios for optimization, lacking comprehensive modeling of the stochastic coupling of source-load-environment factors such as wind power, load, and temperature. When addressing the coupling relationship between planning and operation, most current methods still employ single-layer models or nested solutions, failing to achieve small-range tolerance-based iterative coordination between planning strategies and operational feasibility. This is even more inadequate when considering changes in power flow safety domain and voltage stability constraints. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a two-stage planning method for grid-side energy storage in the scenario of offshore wind access based on temperature-current safety domain mapping, so as to more effectively coordinate planning and operation objectives and achieve the comprehensive optimal configuration of grid-side energy storage.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a two-stage planning method for grid-side energy storage in a sea breeze access scenario based on temperature-current safety domain mapping, comprising the following steps:

[0008] (1) Obtain historical ambient temperature data and use a temperature trend prediction model that combines a long short-term memory network (LSTM) and a temporal convolutional network (TCN) to predict temperature data during the planning period.

[0009] (2) Based on the predicted temperature data, the Gaussian Mixture Model (GMM) is used for temperature clustering, the expected coefficient of temperature variation capacity of each cluster is calculated, and the power flow safety domain of the distribution network is reconstructed.

[0010] (3) A set of random optimization scenarios with spatiotemporal coupling of source and load is established by using the typical source-load daily trajectory clustering method;

[0011] (4) Establish a distribution network energy storage site selection and capacity planning layer model with the goal of maximizing comprehensive utility;

[0012] (5) Establish a distribution network energy storage location and capacity-controlled operation layer model with the goal of minimizing the sum of squares of voltage fluctuations;

[0013] (6) Construct a multi-period optimal power flow model of the distribution network DistFlow under the scenario of sea breeze access;

[0014] (7) The second-order cone relaxation SOCP is used to solve the planning layer, and the target cascade algorithm ATC is used to optimize the charging and discharging and capacity configuration strategy of the operation layer based on the anchoring of the site selection and capacity determination decision variables of the planning layer.

[0015] Furthermore, in step (1), the method for constructing the temperature trend prediction model combining the Long Short-Term Memory (LSTM) network and the Temporal Convolutional Network (TCN) is as follows:

[0016] set up , Given the predicted values ​​of LSTM and TCN respectively, the predicted value of the temperature trend prediction model is:

[0017] (1)

[0018] In the formula, , , These represent the predicted values ​​for time point t by the LSTM, TCN, and temperature trend prediction models, respectively. These are the weighting coefficients;

[0019] To make the predicted value as close as possible to the true value, let:

[0020] (2)

[0021] In the formula, Represents the actual value; This means summing up the prediction errors at all time points to find the optimal weighting coefficient that minimizes the overall prediction error and falls within the range of [0,1].

[0022] To solve for the optimal weight coefficients, the model prediction error is defined. and residual Two variables are used, and equation (2) is rewritten as follows:

[0023] (3)

[0024] Substituting equation (3) into equation (2), equation (2) is transformed into a second-order least squares problem, and by differentiation, we obtain... :

[0025] (4)

[0026] In the formula, The optimal values ​​of the weight coefficients are obtained through second-order least squares optimization;

[0027] Equation (4) is passed through covariance and variance This indicates that the weighting coefficients are constrained to an interval. The final expression for the weighting coefficients is as follows:

[0028] (5)

[0029] In the formula, Cov represents covariance calculation, and Var represents variance calculation. This represents the optimal value of the weighting coefficients that may exceed the boundary. Cut off from The final weighting coefficients.

[0030] Furthermore, the implementation method of step (2) is as follows:

[0031] Let the set T of temperature observation samples be:

[0032] (6)

[0033] In the formula, T i Let represent the i-th temperature observation sample, and M represent the number of temperature observation samples;

[0034] Temperature observation sample T i The probability density function is derived from The weighted superposition of Gaussian distributions:

[0035] (7)

[0036] In the formula, Indicates temperature observation sample T i The probability of occurrence; For the first The mixed weights of temperature clusters, ; Let the probability density function of the k-th Gaussian distribution in time T be... i The value at; and These represent the mean and variance of the clusters, respectively; model parameter set. After transforming equation (7) into a log-likelihood function, the maximum value is obtained using the expectation-maximization algorithm in equation (8):

[0037] (8)

[0038] In the formula, It is the log-likelihood function;

[0039] To balance the accuracy of multi-peak fitting in temperature prediction models with the avoidance of overfitting, the number of clusters... The Bayesian Information Criterion (BIC) is adaptively determined by minimizing the BIC, and its expression is as follows:

[0040] (9)

[0041] In the formula, These are the model parameters; This represents the optimal parameter set obtained by calculation using equation (8);

[0042] After obtaining the temperature at the center of each cluster, the correction factor for the current carrying capacity of the main power grid cable is calculated. When the outdoor temperature is The corrected formula is as follows:

[0043] (10)

[0044] In the formula, Indicates the maximum operating temperature of the cable;

[0045] Obtain the temperature of each cluster center and the corresponding probability Subsequently, the uncertainty in temperature distribution was further mapped to the cable current carrying capacity correction factor. Above; to characterize the active power transmission capacity of backbone cables under temperature variations, a temperature variation expectation coefficient is defined. as follows:

[0046] (11)

[0047] in, To determine the optimal number of clusters based on the Bayesian information criterion, This indicates that the ambient temperature is the temperature at the cluster center. The current carrying capacity correction factor at that time; The probability weights and cable current carrying capacity correction effects of various typical temperature scenarios were taken into account.

[0048] Furthermore, the implementation method of step (3) includes:

[0049] 1) Constructing a daily trajectory sample set: For each natural day Extract the 24-hour time-of-use active and reactive power data of all N nodes in the distribution network to form a high-dimensional daily trajectory vector. :

[0050] (12)

[0051] In the formula, , They represent the first The active and reactive power values ​​of the Nth node of the distribution network at 24:00 on a given natural day; , representing the intraday power distribution of nodes contained in each sample; vector It also preserves the spatiotemporal correlation between nodes and the 24-hour time series characteristics;

[0052] 2) Standardization and Typical Clustering: The samples at each node are standardized to zero mean and variance, then K-means clustering is used to minimize the within-cluster squared error.

[0053] (13)

[0054] In the formula, These are the standardized eigenvalues. For the first One natural day at node eigenvalues ​​on, , For nodes Mean and standard deviation of all samples; Cluster The sample set, The center vector is used; the number of cluster centers is preset. After iterating through and calculating the profile coefficients, the one with the largest average profile coefficient is selected. The optimal number of clusters is used to ensure intra-class similarity and inter-class separability.

[0055] 3) Typical Scene Extraction: Within each cluster, the closest real sample to the cluster center is selected as the representative trajectory, resulting in... A typical daily scene Simultaneously calculate the probability of occurrence of typical daily scenarios:

[0056] (14)

[0057] In the formula, For clusters Mid-distance from center The most recent real sample For the standardized first One sample variable, For the first The probability of occurrence of a typical daily scenario Cluster The number of samples in the middle; This represents the sum of the number of samples in all clusters, i.e., the total number of all valid samples;

[0058] 4) Calculation of expected power in typical daily scenarios: A finite set of scenarios is obtained by calculation using equation (14). Each node at time The expected power is obtained by weighted average:

[0059] (15)

[0060] In the formula, This represents the weighted average power under the probabilities of each typical scenario. , They are nodes At any moment The active and reactive power; , They represent the first time. Nodes in a typical scenario At any moment The active and reactive power.

[0061] Furthermore, the implementation method of step (4) is as follows:

[0062] The goal of the planning layer is to maximize the total daily utility of the system, taking into account network line losses, offshore wind curtailment penalties, economic regularization for energy absorption, and energy storage investment; the objective function is as follows:

[0063] (16)

[0064] (17)

[0065] (18)

[0066] In the formula, Represent the objective function of the planning layer; , These represent the electricity purchased by the distribution network from the superior power grid and the total network loss expenditure of the transmission lines, respectively. Time-of-use pricing for purchasing electricity from the higher-level power grid; , , Representing the system The amount of renewable energy curtailed, output, and absorption capacity during each time period; This indicates a punishment for abandoning the wind; , Representing the system Power purchased from the upper-level power grid and total grid loss during the time period; This represents the converted investment in energy storage for daily chemical products; the energy storage investment model is as follows:

[0067] (19)

[0068] In the formula, , This indicates the daily investment cost and total operation and maintenance cost of energy storage equipment. Indicates the upper limit of the energy storage investment budget; These are the unit power and capacity cost coefficients for energy storage, respectively. This represents the annual operation and maintenance cost coefficient for energy storage. This is the capital recovery factor for energy storage. The social average discount rate, For the operating life of energy storage equipment; They are nodes The rated power and capacity of the energy storage installed at the location;

[0069] The operational constraints of the planning layer include:

[0070] 1) Energy storage quantity constraint, its expression is:

[0071] (20)

[0072] In the formula, For the set of distribution network nodes; It is a 0-1 variable, indicating whether it is in the node. If energy storage is installed at a certain location, the value is 1; otherwise, the value is 0. This indicates the maximum number of energy storage devices that can be installed in a microgrid system;

[0073] 2) Energy storage charge / discharge state constraints, the expression of which is:

[0074] (twenty one)

[0075] In the formula, Representing nodes respectively The rated power and capacity of the energy storage installation. , These represent the corresponding maximum installable energy storage power and capacity, respectively. , for Time Node Discharge and charge power, For 0-1 variables, equations 3 and 4 in equation (21) are used to restrict the same node from being charged and discharged simultaneously in the same time period;

[0076] 3) Energy storage charging and discharging capacity constraints, the expression of which is:

[0077] (twenty two)

[0078] In the formula, For charge and discharge efficiency; express time The State of Charge (SOC) value of energy stored on the node; , This indicates the upper and lower limits of the energy storage SOC.

[0079] Furthermore, the implementation method of step (5) is as follows:

[0080] Given the scheduling results from the planning layer, the operation layer optimizes based on hourly power flow and voltage fluctuations, with the following objective function:

[0081] (twenty three)

[0082] (twenty four)

[0083] In the formula, Represents the objective function of the runtime layer; For the number of distribution network nodes, This is the system's nominal voltage. This represents the voltage at node i during time period t; This represents the optimal value of the objective function at the runtime layer; This represents the optimal value of the objective function at the target layer. , representing the optimal location and capacity scheme solved under the planning layer objective; To configure the capacity and charge / discharge strategy tolerance for the second-level energy storage, so as to ensure that the rated power and capacity of the energy storage configuration can be finely adjusted at the hour level without deviating from the site selection scheme of the planning layer, so as to minimize voltage fluctuations and operational safety without significantly increasing the system operating cost;

[0084] The constraints of the operation layer are the same as those of the planning layer.

[0085] Furthermore, the implementation method of step (6) is as follows:

[0086] The DistFlow multi-period optimal power flow model for the distribution network must satisfy the following constraints:

[0087] 1) Power flow equality constraint, its expression is:

[0088] (25)

[0089] In the formula, , They are nodes The set of parent and child nodes, , for Time period Node to Active and reactive power injected into the node. , for Time period The outflow of both beneficial and ineffective contributions; , They are respectively Active and reactive power injected by the time-period transformer; , for Time period nodes Energy storage discharge or charging power; for Time period nodes Reactive power changes at energy storage sites; , They are respectively Time period nodes The contribution and reactive power output of new energy sources; , They are respectively Time period nodes The active and reactive loads; , They are respectively Time period nodes Active and reactive load loss for Time period node voltage, , Corresponding route The resistance and reactance on it, For the route The current;

[0090] 2) Node voltage constraints, the expression of which is:

[0091] (26)

[0092] 3) Line transmission capacity constraint, its expression is:

[0093] (27)

[0094] In the formula, For the line The maximum apparent capacity that can be transported is This is the expected coefficient for temperature-dependent volume change;

[0095] 4) The power supply capacity and load shedding constraints of the upstream power grid are expressed as follows:

[0096] (28)

[0097] (29)

[0098] In the formula, , The upper and lower limits of active and reactive power flow that can be provided by the upper-level power grid; Defined as 0, meaning that under the dual source-side protection of a large amount of sea breeze and power supply from the upstream power source, active loads are not allowed to be disconnected except in case of a fault; The meaning of a value greater than zero is that reactive power can be fed back to the upper-level power grid to meet reactive power balance.

[0099] Furthermore, the implementation method of step (7) includes:

[0100] 1) Second-order cone relaxation of DistFlow power flow equation

[0101] For the power flow equation of the distribution network constructed by equation (25), a second-order cone relaxation form is used to convexify the power flow equation. Let... , Equation (25) can be equivalently relaxed to the following standard second-order cone form:

[0102] (30)

[0103] (31)

[0104] (32)

[0105] (33)

[0106] 2) Introduce the Objective Cascaded Algorithm (ATC) to transform the collaborative solution of objectives in the planning-running optimization model into a distributed optimization solution framework based on consistency constraints;

[0107] After solving the objective function at the planning level, it provides optimal solutions for energy storage site selection, capacity determination, and other planning-scale considerations. , These correspond to the nodes, rated power, and rated capacity of the energy storage configuration, respectively; the operation layer at a given anchor point variable Under these conditions, hourly-level voltage and power flow optimization is performed. By introducing consistency constraints and augmented Lagrange penalty functions between the upper and lower layers, i.e., between the planning layer and the operation layer, the dual-layer optimization objectives are jointly expressed in the following compact form:

[0108] (34)

[0109] In the formula, Represents lower-level decision variables. For Lagrange multipliers, The penalty factor; the consistency residual vector to eliminate differences in different physical dimensions. The normalized form is used for construction, as shown in the following equation:

[0110] (35)

[0111] In the formula, =1, The upper limit constraint of the corresponding energy storage power capacity and This is used to ensure the numerical stability of different decision variables during the consistency coordination process; in the first... In each ATC iteration, the lower-level running problem is updated by solving the optimization problem in the first equation of (35). Then, the lower-level problem is solved alternately and the multipliers are updated until the infinity norm of the consistency residual is less than the convergence threshold. If the convergence condition is not met, the loop exits; if the convergence condition is not met, the multiplier is updated and the next iteration begins, as shown in the following expression:

[0112] (36)

[0113] The ATC solution process is equivalent to approximating the constrained feasible region in the form of a penalty function within the augmented Lagrange framework, thereby achieving global coordination while maintaining decoupling between upper and lower layers. This method can achieve collaborative optimization between the planning layer and the execution layer, improving the stability and global convergence performance of the system optimization.

[0114] The present invention also provides a two-stage planning system for grid-side energy storage in a sea breeze access scenario based on temperature-current safety domain mapping, including a memory, a processor, and computer program instructions stored in the memory and capable of being executed by the processor. When the processor executes the computer program instructions, it can implement the above-mentioned method.

[0115] The present invention also provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the above-described method.

[0116] Compared with existing technologies, this invention has the following advantages: 1) Based on historical ambient temperature data, the LSTM+TCN hybrid prediction method reasonably predicts the temperature trend during the energy storage planning period and then introduces the ambient temperature expected capacity variation coefficient, thereby reconstructing the power flow constraint SOCP space. The system can more effectively utilize the transmission capacity of the cable network under the premise of satisfying the power flow safety domain, thus obtaining a better energy storage configuration scheme than the static power flow model; 2) Based on the source-load scenario construction method of typical daily trajectory clustering and random optimization scenario probability weighting, the source-load data can be more reasonably predicted during the energy storage planning period, making the energy storage planning more adaptable and effective under the uncertainty of source and load; 3) The two-stage method of dynamic capacity variation of the power flow safety domain and ATC-coordinated energy storage site selection and capacity determination is superior to the traditional static power flow boundary model in terms of voltage robustness, power purchase peak shaving and energy storage utilization efficiency. It can effectively suppress voltage fluctuations and reduce network losses, providing a feasible and valuable technical route for the energy storage site selection and capacity determination planning of existing distribution networks. Attached Figure Description

[0117] Figure 1 This is a flowchart illustrating the implementation of a two-stage planning method for grid-side energy storage in a sea breeze access scenario based on temperature-current safety domain mapping, provided by an embodiment of the present invention. Detailed Implementation

[0118] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0119] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0120] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0121] like Figure 1 As shown, this embodiment provides a two-stage planning method for grid-side energy storage in an offshore wind access scenario based on temperature-current safety domain mapping, including the following steps:

[0122] (1) Obtain historical ambient temperature data and use a temperature trend prediction model that combines Long Short-Term Memory Network (LSTM) and Temporal Convolutional Network (TCN) to predict temperature data during the planning period;

[0123] (2) Based on the predicted temperature data, the Gaussian Mixture Model (GMM) is used to perform temperature clustering, calculate the expected coefficient of temperature variation capacity of each cluster, and reconstruct the power flow safety domain of the distribution network.

[0124] (3) Obtain historical source-load data and establish a stochastic optimization framework for source-load spatiotemporal coupling using a typical source-load daily trajectory clustering method;

[0125] (4) Establish a distribution network energy storage site selection and capacity planning layer model with the goal of maximizing comprehensive utility;

[0126] (5) Establish a distribution network energy storage location and capacity-controlled operation layer model with the goal of minimizing the sum of squares of voltage fluctuations;

[0127] (6) Construct a multi-period optimal power flow model of the distribution network DistFlow under the scenario of large-scale sea wind access;

[0128] (7) The second-order cone relaxation (SOCP) is used to solve the planning layer, and the target cascade algorithm (ATC) is used to optimize the charging and discharging and capacity configuration strategy of the operation layer based on the anchoring of the site selection and capacity determination decision variables of the planning layer.

[0129] The following section provides further explanation of the relevant aspects involved in the implementation of this method.

[0130] I. Construction of a Temperature Trend Prediction Model Combining Long Short-Term Memory Network (LSTM) and Temporal Convolutional Network (TCN)

[0131] This method constructs a temperature trend prediction model that combines a Long Short-Term Memory (LSTM) network with a Temporal Convolutional Network (TCN). Due to the complementary advantages of LSTM and TCN in sequence pattern learning, the least squares method is further employed to determine the optimal fusion weights, fully leveraging the strengths of both models and thus improving prediction performance.

[0132] set up , Given the predicted values ​​of LSTM and TCN respectively, the predicted value of the temperature trend prediction model is:

[0133] (1)

[0134] In the formula, , , These represent the predicted values ​​for time point t by the LSTM, TCN, and temperature trend prediction models, respectively. These are the weighting coefficients.

[0135] To make the predicted value as close as possible to the true value, let:

[0136] (2)

[0137] In the formula, Represents the actual value; This indicates that the prediction errors at all time points are accumulated to find the optimal weighting coefficient that minimizes the overall prediction error and falls within the range of [0,1].

[0138] To facilitate solving for the optimal weight coefficients, the model prediction difference is defined. and residual Two variables are used, and equation (2) is rewritten as follows:

[0139] (3)

[0140] Substituting equation (3) into equation (2), equation (2) can be transformed into a second-order least squares problem, and by differentiation, we can obtain... :

[0141] (4)

[0142] In the formula, The optimal values ​​of the weight coefficients are obtained through second-order least squares optimization.

[0143] Equation (4) is passed through covariance and variance This indicates that the weighting coefficients are constrained to an interval. The final expression for the weighting coefficients can be obtained as follows:

[0144] (5)

[0145] In the formula, Cov represents covariance calculation, and Var represents variance calculation. This represents the optimal value of the weighting coefficients that may exceed the boundary. Cut off from The final weighting coefficients.

[0146] 2. Temperature clustering is performed using a Gaussian Mixture Model (GMM) to calculate the expected coefficient of temperature variation capacity for each cluster, thereby reconstructing the power flow safety domain of the distribution network.

[0147] Let the set T of temperature observation samples be:

[0148] (6)

[0149] In the formula, T i Let M represent the i-th temperature observation sample, and M represent the number of temperature observation samples.

[0150] Temperature observation sample T i The probability density function can be derived from The weighted superposition of Gaussian distributions:

[0151] (7)

[0152] In the formula, Indicates temperature observation sample T i The probability of occurrence; For the first Mixed weights of temperature clusters ( ); Let the probability density function of the k-th Gaussian distribution in time T be... i The value at; and These represent the mean and variance of the corresponding clusters, respectively. Model parameter set. After transforming equation (7) into a log-likelihood function, the maximum value is solved using the Expectation-Maximization (EM) algorithm in equation (8):

[0153] (8)

[0154] In the formula, Let be the log-likelihood function.

[0155] To balance the accuracy of multi-peak fitting in temperature prediction models with the avoidance of overfitting, the number of clusters... The Bayesian Information Criterion (BIC) is adaptively determined using the method of minimizing Bayesian Information Criterion (BIC), and its expression is as follows:

[0156] (9)

[0157] In the formula, These are the model parameters; This represents the optimal parameter set obtained by calculation using equation (8).

[0158] After obtaining the temperature of each cluster center, the correction factor for the current carrying capacity of the main distribution network cable is calculated according to the temperature correction formula in Appendix A of IEC 60287-1-1:2006. When the outdoor temperature is The corrected formula is as follows:

[0159] (10)

[0160] In the formula, This indicates the cable's maximum operating temperature.

[0161] Obtain the temperature of each cluster center and the corresponding probability Subsequently, the uncertainty in temperature distribution can be further mapped to the cable current carrying capacity correction factor. Above. To better characterize the active power transmission capacity of backbone cables under temperature variations, a temperature-dependent capacity expectation coefficient is defined. as follows:

[0162] (11)

[0163] in, To determine the optimal number of clusters based on the Bayesian information criterion, This indicates that the ambient temperature is the temperature at the cluster center. The current carrying capacity correction factor at that time. The probability weights and cable current carrying capacity correction effects of various typical temperature scenarios were taken into account.

[0164] III. A typical source-load solar trajectory clustering method is used to establish a spatiotemporally coupled stochastic optimization framework for source-load.

[0165] 1) Constructing a daily trajectory sample set: For each natural day Extract all nodes of the distribution network The 24-hour time-sharing active and reactive power data form a high-dimensional daily trajectory vector. :

[0166] (12)

[0167] In the formula, , They represent the first The active and reactive power values ​​of the Nth node of the distribution network at 24:00 on a given natural day; , representing the intraday power distribution of nodes contained in each sample; vector It also preserves the spatiotemporal correlation between nodes and the 24-hour time series characteristics.

[0168] 2) Standardization and Typical Clustering: To eliminate the influence of different node power levels on distance metrics, zero-mean-variance standardization is performed on the samples of each node. Then, K-means clustering is used to minimize the within-cluster squared error.

[0169] (13)

[0170] In the formula, These are the standardized eigenvalues. For the first One natural day at node eigenvalues ​​on, , For nodes Mean and standard deviation of all samples; Cluster The sample set, The center vector. Preset number of cluster centers. After iterating through and calculating the profile coefficients, the one with the largest average profile coefficient is selected. The optimal number of clusters is determined to ensure intra-class similarity and inter-class separability.

[0171] 3) Typical Scene Extraction: Within each cluster, the closest real sample to the cluster center is selected as the representative trajectory, resulting in... A typical daily scene Simultaneously calculate the probability of occurrence of typical daily scenarios:

[0172] (14)

[0173] In the formula, For clusters Mid-distance from center The most recent real sample (i.e., the representative trajectory medoid). For the standardized first One sample variable, For the first The probability of occurrence of a typical daily scenario Cluster The number of samples in the middle; This represents the sum of the number of samples in all clusters, i.e., the total number of all valid samples.

[0174] 4) Calculation of expected power in typical daily scenarios: A finite set of scenarios is obtained by calculation using equation (14). This set is considered a discrete approximation of the empirical distribution, preserving the main stochastic characteristics of the system's operation. Therefore, each node at time... The expected power can be obtained by weighted average:

[0175] (15)

[0176] In the formula, This represents the weighted average power under the probabilities of each typical scenario. , They are nodes At any moment The active and reactive power; , They represent the first time. Nodes in a typical scenario At any moment The active and reactive power.

[0177] IV. Establishing a distribution network energy storage site selection and capacity planning layer model with the goal of maximizing comprehensive utility.

[0178] The goal of the planning layer is to maximize the total daily utility of the system, taking into account network line losses, offshore wind curtailment penalties, economic regularization for energy absorption, and energy storage investment; the objective function is as follows:

[0179] (16)

[0180] (17)

[0181] (18)

[0182] In the formula, Represent the objective function of the planning layer; , These represent the electricity purchased by the distribution network from the superior power grid and the total network loss expenditure of the transmission lines, respectively. Time-of-use pricing for purchasing electricity from the higher-level power grid; , , Representing the system The amount of renewable energy curtailed, output, and absorption capacity during each time period; This indicates a punishment for abandoning the wind; , Representing the system Power purchased from the upper-level power grid and total grid loss during the time period; This represents the converted investment in energy storage for daily chemical products. The energy storage investment model is as follows:

[0183] (19)

[0184] In the formula, , This indicates the daily investment cost and total operation and maintenance cost of energy storage equipment. Indicates the upper limit of the energy storage investment budget; These are the unit power and capacity cost coefficients for energy storage, respectively. This represents the annual operation and maintenance cost coefficient for energy storage. This is the capital recovery factor for energy storage. The social average discount rate, For the operating life of energy storage equipment; They are nodes The rated power and capacity of the energy storage installed at the location.

[0185] The operational constraints of the planning layer include:

[0186] 1) Energy storage quantity constraint, its expression is:

[0187] (20)

[0188] In the formula, For the set of distribution network nodes; It is a 0-1 variable, indicating whether it is in the node. If energy storage is installed at a certain location, the value is 1; otherwise, the value is 0. This indicates the maximum number of energy storage devices that can be installed in a microgrid system.

[0189] 2) Energy storage charge / discharge state constraints, the expression of which is:

[0190] (twenty one)

[0191] In the formula, Representing nodes respectively The rated power and capacity of the energy storage installation. , These represent the corresponding maximum installable energy storage power and capacity, respectively. , for Time Node Discharge and charge power, For variables of 0-1, equations 3 and 4 in equation (21) are used to restrict the same node from being charged and discharged at the same time.

[0192] 3) Energy storage charging and discharging capacity constraints, the expression of which is:

[0193] (twenty two)

[0194] In the formula, For charge and discharge efficiency; express time The State of Charge (SOC) value of energy stored on the node; , This indicates the upper and lower limits of the energy storage SOC.

[0195] V. Establish a distribution network energy storage location and capacity-controlled operation layer model with the objective of minimizing the sum of squared voltage fluctuations.

[0196] Given the scheduling results from the planning layer, the operation layer optimizes based on hourly power flow and voltage fluctuations, with the following objective function:

[0197] (twenty three)

[0198] (twenty four)

[0199] In the formula, Represents the objective function of the runtime layer; For the number of distribution network nodes, This is the system's nominal voltage. This represents the voltage at node i during time period t; This represents the optimal value of the objective function at the runtime layer; This represents the optimal value of the objective function at the target layer. , representing the optimal location and capacity scheme solved under the planning layer objective; The capacity and charge / discharge strategy tolerances for the second-level energy storage configuration are configured to ensure that the rated power and capacity of the energy storage configuration can be finely adjusted at the hourly level without deviating from the site selection scheme of the planning layer, so as to minimize voltage fluctuations and ensure operational safety without significantly increasing the system operating costs.

[0200] The constraints of the operation layer are the same as those of the planning layer, as shown in equations (20)-(22).

[0201] VI. Constructing a multi-period optimal power flow model for the distribution network under scenarios with a large amount of offshore wind access.

[0202] The DistFlow multi-period optimal power flow model for the distribution network must satisfy the following constraints:

[0203] 1) Power flow equality constraint, its expression is:

[0204] (25)

[0205] In the formula, , They are nodes The set of parent and child nodes, , for Time period Node to Active and reactive power injected into the node. , for Time period The outflow of both beneficial and ineffective contributions; , They are respectively Active and reactive power injected by the time-period transformer; , for Time period nodes Energy storage discharge or charging power; for Time period nodes Reactive power changes at energy storage sites; , They are respectively Time period nodes The contribution and reactive power output of new energy sources; , They are respectively Time period nodes The active and reactive loads; , They are respectively Time period nodes Active and reactive load loss for Time period node voltage, , Corresponding route The resistance and reactance on it, For the route The current.

[0206] 2) Node voltage constraints, the expression of which is:

[0207] (26)

[0208] 3) Line transmission capacity constraint, its expression is:

[0209] (27)

[0210] In the formula, For the line The maximum apparent capacity that can be transported is is the expected coefficient for temperature-varying volume.

[0211] 4) The power supply capacity and load shedding constraints of the upstream power grid are expressed as follows:

[0212] (28)

[0213] (29)

[0214] In the formula, , The upper and lower limits of active and reactive power flow that the upper-level power grid can provide; in the model of this paper. Defined as 0, meaning that under the dual source-side protection of a large amount of sea breeze and power supply from the upstream power source, active loads are not allowed to be disconnected except in case of a fault; The meaning of being greater than zero is that, under certain circumstances, reactive power can be fed back to the upper-level power grid to meet reactive power balance.

[0215] VII. The second-order cone relaxation (SOCP) is used to solve the planning layer problem. For the operation layer, the target cascade algorithm (ATC) is used to perform secondary optimization of the charging / discharging and capacity configuration strategies based on the site selection and capacity determination decision variables of the planning layer.

[0216] 1) Second-order cone relaxation of DistFlow power flow equation

[0217] In the power flow equation of the distribution network constructed by equation (25), the nonlinear terms mainly include the nodal voltage square terms. and the product of apparent power and the reciprocal of the square of voltage. If the traditional AC power flow equations are used directly, the model will exhibit non-convexity, making it difficult to obtain the global optimum using conventional optimization algorithms. To ensure the convexity and solvability of the model, this paper employs Second-Order Cone Programming (SOCP) to convexize the power flow equations. , Equation (25) can be equivalently relaxed to the following standard second-order cone form:

[0218] (30)

[0219] (31)

[0220] (32)

[0221] (33)

[0222] 2) Introduce the Objective Cascaded Algorithm (ATC) to transform the collaborative solution of objectives in the planning-running optimization model into a distributed optimization solution framework based on consistency constraints.

[0223] After solving the objective function at the planning level, it provides optimal solutions for energy storage site selection, capacity determination, and other planning-scale considerations. , These correspond to the nodes, rated power, and rated capacity of the energy storage configuration, respectively; the operation layer at a given anchor point variable Under these conditions, hourly-level voltage and power flow optimization is performed. By introducing consistency constraints and augmented Lagrange penalty functions between the upper and lower layers, i.e., between the planning layer and the operation layer, the dual-layer optimization objectives are jointly expressed in the following compact form:

[0224] (34)

[0225] In the formula, Represents lower-level decision variables. For Lagrange multipliers, The penalty factor; the consistency residual vector to eliminate differences in different physical dimensions. The normalized form is used for construction, as shown in the following equation:

[0226] (35)

[0227] In the formula, =1, The upper limit constraint of the corresponding energy storage power capacity and This is used to ensure the numerical stability of different decision variables during the consistency coordination process; in the first... In each ATC iteration, the lower-level running problem is updated by solving the optimization problem in the first equation of (35). Then, the lower-level problem is solved alternately and the multipliers are updated until the infinity norm of the consistency residual is less than the convergence threshold. If the convergence condition is not met, the loop exits; if the convergence condition is not met, the multiplier is updated and the next iteration begins, as shown in the following expression:

[0228] (36)

[0229] The ATC solution process is equivalent to approximating the constrained feasible region in the form of a penalty function within the augmented Lagrange framework, thereby achieving global coordination while maintaining decoupling between upper and lower layers. This method can achieve collaborative optimization between the planning layer and the execution layer, improving the stability and global convergence performance of the system optimization.

[0230] As described above, this invention provides a two-stage planning method for grid-side energy storage in offshore wind access scenarios based on temperature-current safety domain mapping. First, this method constructs an uncertain scenario system oriented towards ambient temperature-source-load coupling. It utilizes an LSTM-TCN hybrid temperature prediction model, a fusion of Long Short-Term Memory (LSTM) and Temporal Convolutional Network (TCN), combined with a source-load stochastic scenario. K-means and medoid (cluster representative samples) are used to cluster and extract representativeness of typical daily trajectories, ensuring that typical wind turbine output and load scenarios simultaneously maintain spatiotemporal coupling characteristics. Second, this method proposes a variable capacity current flow constraint modeling method for cable networks based on temperature influence. Through stochastic planning of cable current-carrying capacity across multiple scenarios, a "variable capacity expectation coefficient" for the distribution network is formed, and the current flow safety domain is reconstructed accordingly, realizing the ambient temperature variable capacity effect and embedding it into the optimization model. Finally, a two-stage planning method for energy storage configuration is performed based on the Objective Cascaded Algorithm (ATC). Under the reconstructed power flow safety domain boundary, an energy storage site selection and capacity determination model is established and solved with the first-stage objective of minimizing the comprehensive daily operating cost and the second-stage objective of minimizing the operating voltage fluctuation, thereby realizing two-stage optimization from the planning layer to the operation layer.

[0231] This embodiment also provides a two-stage planning system for grid-side energy storage in a sea breeze access scenario based on temperature-current safety domain mapping, including a memory, a processor, and computer program instructions stored in the memory and executable by the processor. When the processor executes the computer program instructions, it can implement the above-mentioned method.

[0232] This embodiment also provides a computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the above-described method.

[0233] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0234] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0235] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0236] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0237] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A two-stage planning method for grid-side energy storage in a sea wind access scenario based on temperature-tidal current safety domain mapping, characterized in that, The method comprises the following steps: (1) obtaining historical environmental temperature data, using a temperature trend prediction model combining a long short-term memory network (LSTM) and a time series convolution network (TCN) to predict temperature data in a planning period; (2) based on the predicted temperature data, using a Gaussian mixture model (GMM) to perform temperature clustering, calculating a temperature variable-volume expectation coefficient of each cluster, and reconstructing a power flow security domain of the power distribution network; (3) using a typical source-load daily trajectory clustering method, a random optimization scenario set of source-load space-time coupling is established; (4) a planning layer model of the power distribution network energy storage site selection and capacity planning is established, with the maximum comprehensive utility as the target; (5) a running layer model of the power distribution network energy storage site selection and capacity planning is established, with the minimum square sum of voltage fluctuations as the target; (6) a power distribution network DistFlow multi-period optimal power flow model under a sea wind access scenario is constructed; (7) the planning layer is solved by using a second-order cone relaxation (SOCP), and the running layer is secondarily optimized by using an anchor target cascade (ATC) algorithm based on the site selection and capacity planning decision variables of the planning layer to obtain charging and discharging strategies and capacity configuration strategies.

2. The two-stage planning method for grid-side energy storage in a sea wind access scenario based on temperature-tidal current safety domain mapping according to claim 1, characterized in that, In step (1), the temperature trend prediction model combining the long short-term memory network (LSTM) and the time series convolution network (TCN) is constructed in the following manner: Let , are the predicted values of LSTM, TCN respectively, then the predicted value of the temperature trend prediction model is: (1) In the formula, , , respectively represent the predicted value of the LSTM, TCN and temperature trend prediction model at time point t, is a weight coefficient. In order to make the predicted value closest to the true value, let: (2) wherein represents the true value; represents the cumulative prediction error for all time points to find the optimal weight coefficient that minimizes the overall prediction error and lies in the range [0, 1]; To solve the optimal weight coefficients, define model prediction error and residual two variables, and rewrite equation (2) as follows: (3) Substitute formula (3) into formula (2), convert formula (2) into a second-order least square problem, and obtain by derivation : (4) In the formula, is the optimal value of the weight coefficient obtained by the second-order least square optimization. The formula (4) is expressed by covariance and variance and the weight coefficient is constrained to the interval The final expression of the weight coefficient is as follows: (5) where Cov denotes a covariance calculation, Var denotes a variance calculation, denotes the optimal value of the weight coefficient that can exceed the boundary truncated to the maximum weight coefficient.

3. The two-stage planning method for grid-side energy storage in a sea wind access scenario based on temperature-tidal current safety domain mapping according to claim 1, characterized in that, The implementation method of step (2) is as follows: Let the set T of temperature observation samples be: (6) In the formula, T i represents the ith temperature observation sample, and M represents the number of temperature observation samples. Temperature observation sample T i The probability density function of T is represented by a weighted superposition of K Gaussian distributions: where the parameters of the Gaussian distributions are given by (7) wherein denotes the temperature observation sample T i the probability of occurrence; is the mixing weight of the temperature cluster, ; denotes the value of the probability density function of the k-th Gaussian distribution at T i ; and and are the mean and variance of the cluster, respectively; the set of model parameters By transforming equation (7) into a log-likelihood function, the maximum is solved by the expectation-maximization algorithm of equation (8): (8) In the formula, is the log-likelihood function; To balance the multi-peak fitting accuracy of the air temperature prediction model and avoid overfitting, the cluster number k is adaptively determined by minimizing the Bayesian information criterion (BIC), which is expressed as follows: The adaptive determination is realized by minimizing the Bayesian information criterion (BIC), which is expressed as follows: (9) In the formula, is the number of model parameters; denotes the optimal parameter set calculated by formula (8). After obtaining the temperature of each clustering cluster center, the cable carrying capacity correction coefficient of the main network frame of the power distribution network is calculated The correction formula under the outdoor temperature of is as follows: (10) In the formula, represents the maximum operating temperature of the cable; After obtaining the cluster centers temperature and the corresponding probability Further, the uncertainty of temperature distribution is mapped to the cable ampacity correction factor ; in order to characterize the active power flow carrying capacity of the backbone cable under temperature change, the temperature variable capacitance expectation coefficient is defined as follows: (11) wherein, is the optimal cluster number determined according to the Bayesian information criterion, represents the current temperature of the environment as a cluster center temperature of the current temperature of the environment as a cluster center temperature The probability weight of each typical temperature scene and the cable current-carrying capacity correction effect are comprehensively considered.

4. The two-stage planning method for grid-side energy storage in a sea wind access scenario based on temperature-tidal current safety domain mapping according to claim 1, characterized in that, The implementation method of step (3) comprises the following steps: 1) Constructing daily trajectory sample set: for each natural day , extract 24h time-series active and reactive power data of all N nodes of the power distribution network, form high-dimensional daily trajectory vector : (12) wherein, , represent the active and reactive power values of the Nth node of the power distribution network at 24:00 of the Nth natural day, respectively; represents the intra-day power profile of each sample containing nodes; vector while preserving the spatio-temporal correlation among nodes and the 24h time-series characteristics.​ 2) standardization and typical clustering: the zero-mean variance standardization is performed on each node sample, and then the K-means clustering is used to minimize the intra-class square error: (13) In the formula, These are the standardized eigenvalues. For the first One natural day at node eigenvalues ​​on, , For nodes Mean and standard deviation of all samples; Cluster The sample set, The center vector is used; the number of cluster centers is preset. After iterating through and calculating the profile coefficients, the one with the largest average profile coefficient is selected. The optimal number of clusters is used to ensure intra-class similarity and inter-class separability. 3) Typical scene extraction: In each cluster, the nearest real sample to the cluster center is selected as the representative trajectory, and the typical day scene is obtained Typical day scene The typical day scene occurrence probability is also calculated. (14) wherein is the cluster is the mean distance center is the nearest real sample, is the standardized first sample variable, is the appearance probability of the first typical day scenario, denotes the number of samples in the cluster ; and denotes the sum of the number of samples in all clusters, i.e. the total number of valid samples. 4) Desired power calculation for typical day scenarios: The finite set of scenarios is obtained by calculating via equation (14) The desired power of each node at time is obtained by a weighted average: (15) wherein, denotes the weighted average power over the typical scenario probabilities, , denote the active and reactive power at the node at time ; , denote the active and reactive power at the node under the th typical scenario at time .

5. The two-stage planning method for grid-side energy storage in a sea wind access scenario based on temperature-tidal current safety domain mapping according to claim 1, characterized in that, The implementation method of step (4) is as follows: The target of the planning layer is to maximize the total utility of the system within a day, and the network line loss, the sea wind power penalty term, the economic regularization term and the energy storage investment are comprehensively considered; the objective function is as follows: (16) (17) (18) In the formula, represents the planning layer objective function; , respectively represent the power purchase from the upper-level power grid and the total line loss expenditure of the distribution network; is the time-of-use price of power purchase from the upper-level power grid; , , respectively represent the system period new energy curtailment power, output power and consumption power; represents the wind curtailment penalty; , respectively represent the system period power purchase from the upper-level power grid and the total network loss power; represents the converted daily investment of energy storage; the energy storage investment model is as follows: (19) In the formula, , represents the daily investment cost and total operation and maintenance cost of the energy storage device, represents the upper limit of the energy storage investment budget; are the unit power and capacity cost coefficients of the energy storage, respectively; is the annual operation and maintenance cost coefficient of the energy storage; is the capital recovery coefficient of the energy storage, is the average social discount rate, is the service life of the energy storage device; are the rated power and capacity of the energy storage installed at the node , respectively. The running constraints of the planning layer comprise the following: 1) the expression of the energy storage quantity constraint is as follows: (20) wherein, is a set of distribution grid nodes; is a 0-1 variable indicating whether or not to install energy storage at node , and takes 1 if installed, otherwise 0; denotes the maximum number of energy storage devices that can be installed in the microgrid system; 2) the expression of the energy storage charging and discharging state constraint is as follows: (21) wherein, respectively represent the nodes the rated power and capacity of the energy storage installed on the upper, , respectively represent the upper limit of the installable energy storage power and capacity; , is the discharge and charge power of the node at the moment, is a 0-1 variable, and the 3rd and 4th formulas in formula (21) are used to limit the same node from simultaneously charging and discharging at the same time period. 3) the expression of the energy storage charging and discharging capacity constraint is as follows: (22) In the formula, for charge and discharge efficiency; denotes time SOC value of energy storage on the node; , denotes upper and lower limit values of energy storage SOC.

6. The two-stage planning method for grid-side energy storage in a sea wind access scenario based on temperature-tidal current safety domain mapping according to claim 1, characterized in that, The implementation method of step (5) is as follows: The running layer takes the hourly power flow and voltage fluctuation state as the optimization object under the condition of the given planning layer scheduling result, and the objective function is as follows: (23) (24) wherein, represents the operation layer objective function; is the number of nodes in the distribution network, is the system nominal voltage, represents the voltage at node i at time period t; represents the optimal value of the operation layer objective function; represents the optimal value of the target layer objective function; represents the optimal sizing and siting scheme solved under the planning layer objective; is the tolerance of the two-layer energy storage configuration capacity and charging and discharging strategy, used to ensure that the operation layer makes hour-level fine tuning on the rated power and capacity of the energy storage configuration without deviating from the planning layer siting scheme, so as to realize voltage fluctuation minimization and operation safety without significantly increasing the system operation cost. The constraint conditions of the running layer are the same as those of the planning layer.

7. The two-stage planning method for grid-side energy storage in a sea wind access scenario based on temperature-tidal current safety domain mapping according to claim 1, characterized in that, The implementation method of step (6) is as follows: The power distribution network DistFlow multi-period optimal power flow model needs to satisfy the following constraint conditions: 1) the expression of the power flow equation constraint is as follows: (25) In the formula, , are the parent node and child node set of node , , is the period active and reactive power injected by the node to the node , is the period active and reactive power outflowed by the node , are the active and reactive power injected by the transformer in the period ; , is the period discharge or charge power of the energy storage at the node ; is the period reactive power change of the energy storage at the node , are the active and reactive power output of the new energy at the node in the period , are the active and reactive load of the node in the period ; , are the active and reactive load loss of the node in the period , is the voltage of the node in the period , correspond to the resistance and reactance on the line , is the current flowing through the line ; 2) the expression of the node voltage constraint is as follows: (26) 3) the expression of the line transmission capacity constraint is as follows: (27) wherein is the line the maximum apparent capacity that can be transported, is the temperature-dependent variable desired coefficient; 4) the expression of the upper-level power grid power supply and loss of load constraint is as follows: (28) (29) In the formula, , are the upper and lower limits of active and reactive power flow that the upper grid can provide; is defined as 0, that is, under the dual-source guarantee of a large amount of wind power access and upper power supply, active load is not allowed to be cut off except for faults; can be greater than zero, which means that reactive power can be sent back to the upper grid to meet the reactive power balance.

8. The two-stage planning method for grid-side energy storage in a sea wind access scenario based on temperature-tidal current safety domain mapping according to claim 1, characterized in that, The implementation method of step (7) comprises the following steps: 1) DistFlow power flow equation second-order cone relaxation For the power distribution network DistFlow power flow equation constructed by formula (25), the convexity processing is carried out on the power flow equation by adopting a second-order cone relaxation form, and formula (25) is equivalent to the following standard second-order cone form: 、 , formula (25) can be equivalent to the following standard second-order cone form: (30) (31) (32) (33) 2) the target cascade (ATC) algorithm is introduced to convert the cooperative solving of the target in the planning layer-running layer optimization model into a distributed optimization solving framework based on a consistency constraint; After the solution of the planning layer objective function, the optimal solution of energy storage siting, sizing, etc. is provided , corresponding to the node, rated power and rated capacity of energy storage configuration respectively; the operation layer carries out hour-level voltage and power flow optimization under the condition of given anchor point variables . By introducing consistency constraint items and augmented Lagrange penalty function between the upper and lower layers, i.e. between the planning layer and the operation layer, the double-layer optimization objective is jointly expressed in the following compact form: (34) wherein denotes the lower level decision variable, is the Lagrange multiplier, is the penalty factor; to eliminate the difference of different physical dimensions, a consistent residual vector is constructed in a normalized form as follows: (35) where = 1, Corresponding power capacity upper bound constraint on energy storage and , to guarantee numerical stability of different decision variables in the consistency coordination process; in the th ATC iteration, the lower-level operational problem is updated by solving the optimization problem of the first equation in (35), and then the multipliers are updated by alternately solving the lower-level problem and updating the multipliers until the infinity norm of the consistency residual is less than the convergence threshold , the loop is exited; if the convergence condition is not satisfied, the next iteration is followed by updating the multipliers, and the expression is as follows: (36) The ATC solving process is equivalent to approximating the constraint feasible region in the form of a penalty function under the augmented Lagrangian framework, so as to realize the global coordination while maintaining the decoupled solving of the upper and lower layers; this method can realize the cooperative optimization of the planning layer and the running layer, and improve the stability and global convergence performance of the system optimization.

9. A two-stage planning system for grid-side energy storage in a sea wind access scenario based on temperature-tidal current safety domain mapping, characterized in that, A computer program product comprising a storage medium to store computer program instructions, which instructions when executed by a processor, implement the method according to any one of claims 1-8.

10. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions, when executed by the processor, implement the method according to any one of claims 1-8.

Citation Information

Cited By

  • Heat dissipation control method and system for bus duct

    CN122051846A