An energy storage optimization scheduling method considering operating costs and reverse heavy overload relief

Abstract

The present invention proposes an energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief, comprising the following steps: S1: predicting the photovoltaic node power generation power and user load node power of the second day based on the K-means-XGBoost-Informer model and data set, and formulating and executing the day-ahead decision based on the constraint conditions and the objective function; S2: obtaining the operation data of the key nodes in the substation area, and judging whether there is a reverse overload according to the operation data of the key nodes, if so, executing step S3, if not, skipping step S3 and executing step S4; S3: suspending the execution of the day-ahead decision, formulating and executing the real-time decision, while executing the real-time decision, updating the day-ahead decision and obtaining the operation data of the transformer energy storage node in real time and judging the flow direction until the flow direction of the transformer energy storage node becomes positive; S4: judging whether the day-ahead decision is fully executed, if not fully executed, continuing to execute the day-ahead decision and returning to step S2, if fully executed, exiting.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic power regulation, and specifically relates to an energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief. Background Art

[0002] With the increasing attention paid to and widespread application of clean energy, photovoltaic grid-connected power generation systems, as a key green power supply channel, are receiving widespread attention and praise from the industry.

[0003] It cannot be ignored that photovoltaic power generation is constrained by external factors such as natural conditions, such as weather changes. The access of photovoltaic power generation systems may bring many power quality challenges to the distribution network. In particular, when photovoltaic power generation is large, it is easy to cause reverse heavy overload in the substation area, which will pose a threat to the safe and stable operation of the power grid.

[0004] In order to effectively deal with the problem of reverse overload in the substation, the traditional solution is to add transformers and reconstruct the lines. However, the traditional solution is difficult to promote effectively due to the complex terrain and high construction costs. With the emergence of distributed energy storage equipment, it has become a new way to solve reverse overload due to its flexibility and easy deployment. How to use distributed energy storage equipment to adjust the optimal operation of energy storage equipment, ensure the stability and reliability of power supply in the substation, and achieve the goal of alleviating reverse overload in the substation and the economy of energy storage operation is a technical problem that needs to be solved. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, a method for optimizing energy storage scheduling taking into account operating costs and reverse heavy overload relief is provided.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] The present invention proposes an energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief, comprising the following steps:

[0008] S1: Based on the K-means-XGBoost-Informer model and data set, the photovoltaic node power generation and user load node power of the next day are predicted, and the day-ahead decision is made and executed based on the constraints and objective function;

[0009] S2: During the execution of the day-ahead decision, the operation data of the key nodes in the substation area are obtained in real time, and whether there is a reverse overload is determined based on the operation data of the key nodes. If so, step S3 is executed; if not, step S3 is skipped and step S4 is executed;

[0010] S3: Suspend the execution of the day-ahead decision, make and execute the real-time decision based on the current SOC value of the energy storage device with the goal of minimizing the reverse power transmission, and while executing the real-time decision, update the day-ahead decision and obtain the operating data of the transformer energy storage node in real time and determine the power flow direction until the power flow direction of the transformer energy storage node becomes positive, thereby eliminating the reverse overload;

[0011] S4: Determine whether the day-ahead decision has been fully executed. If not, continue to execute the day-ahead decision and return to step S2. If all have been executed, exit.

[0012] Furthermore, in S1, predicting the photovoltaic node power generation and user load node power on the second day includes the following steps:

[0013] Preprocess the data in the data set, remove abnormal data points, and replace the abnormal data with the average of adjacent normal values;

[0014] The preprocessed data is normalized to eliminate the dimension difference. The formula is as follows:

[0015] (1)

[0016] In the formula, x 0 is the initial data set, μ is the mean of the data set, σ is the standard deviation of the data set, x is the standardized data set;

[0017] The weather data in the dataset is clustered using the K-means algorithm according to the three weather conditions of sunny, rainy, and cloudy days to obtain photovoltaic data under different scenarios;

[0018] Considering the impact of environmental factors, the XGBoost algorithm is used to sort environmental variables and select environmental variables that have a great impact on the power generation of photovoltaic nodes and the power of user load nodes;

[0019] The Informer model is trained using the data set, and the Informer model is used to perform day-ahead predictions to obtain the photovoltaic node power generation and user load node power data for the next day.

[0020] Furthermore, in S1, the constraints include SOC value constraints and charge and discharge power constraints; the SOC value constraint formula is as follows:

[0021] (2)

[0022] In the formula, SOC t express tThe state of charge of the energy storage device at all times, SOC min Indicates the minimum SOC value allowed by the energy storage device. SOC max Indicates the maximum SOC value allowed by the energy storage device;

[0023] The charging and discharging power constraint formula is as follows:

[0024] (3)

[0025] In the formula, express t The charging and discharging power of the energy storage device at all times, When it is positive, it indicates charging power, and when it is negative, it indicates discharging power. Indicates the maximum discharge power of the energy storage device. Indicates the maximum charging power of the energy storage device.

[0026] Furthermore, SOC min The value range is , SOC max The value range is .

[0027] Furthermore, SOC min The values ​​of include 0.1, SOC max The values ​​of include 0.9, and The values ​​include 100kW.

[0028] Furthermore, in S1, the formula of the objective function is as follows:

[0029] (4)

[0030] In the formula, f 1 represents the variance of the power on the power supply side of the distribution network, f 2 Represents the total cost of 24-hour operation of the energy storage equipment, λ 1 express f 1 The weight of λ 2 express f 2 The weight of express f 1 The initial value of express f 2The initial value of It represents the cost of energy storage cycle charging at the lowest electricity price and discharging at the highest electricity price. μ t Represents the penalty coefficient, which is related to the load rate β t Related, P 1,t express t The active power on the power supply side of the distribution network at all times, Q 1,t express t The reactive power on the power supply side of the distribution network at all times, S N Indicates the rated capacity of the distribution network transformer;

[0031] Load factor β t The calculation formula is as follows:

[0032] (5)

[0033] Penalty coefficient μ t The calculation formula is as follows:

[0034] (6)

[0035] In the formula, Sgn ( P 1,t ) is the symbol function, P 1,t express t The active power on the power supply side of the distribution network at all times,

[0036] Q 1,t express t The reactive power on the power supply side of the distribution network at all times, S N Indicates the rated capacity of the distribution network transformer, β t A positive value indicates that the power flow direction is positive, i.e., the distribution transformer supplies power to the substation; β t A negative value means the power flow direction is negative, that is, the substation supplies power back to the distribution transformer.

[0037] Furthermore, λ 1 The values ​​of include 0.9, λ 2 The value of includes 0.1.

[0038] Furthermore, the variance of the power on the power supply side of the distribution network f 1Total cost of 24-hour operation of energy storage equipment f 2 The calculation of includes the following steps:

[0039] Define the power supply side of the distribution network as node 1, divide the 24 hours of a day into 24 time points, and calculate the active power and reactive power of the power supply side of the distribution network based on the predicted data. The formula is as follows:

[0040] (7)

[0041] In the formula, P 1,t express t The active power on the power supply side of the distribution network at all times, Q 1,t express t The reactive power on the power supply side of the distribution network at all times, U 1,t express t The voltage at node 1 at time t, U j,t express t time j The voltage at the node, N Indicates the size of the node, G 1j Indicates the power supply side and nodes of the distribution network in the line j The conductance between B 1j Indicates the power supply side and nodes of the distribution network in the line j The electrical susceptance between θ 1j,t express t time j Phase angle of the node;

[0042] Summarize the active power on the power supply side of the distribution network within 24 hours and calculate the average power and variance of the power on the power supply side of the distribution network f 1 , the formula is as follows:

[0043] (8)

[0044] (9)

[0045] In the formula, P 1,t express t The active power on the power supply side of the distribution network at all times, Indicates the average power of the distribution network within 24 hours;

[0046] Calculate the total cost of 24-hour operation of energy storage equipment f 2 , the formula is as follows:

[0047] (10)

[0048] In the formula, Indicates the charging and discharging power of the energy storage device. B(t) Indicates the electricity price.

[0049] Furthermore, in S2, key nodes include transformer energy storage nodes, distributed photovoltaic power generation nodes, and load side nodes, and operating data include voltage, current, active power, and reactive power.

[0050] Furthermore, in S2, the determination of reverse overload includes the following steps:

[0051] The power flow direction of the transformer energy storage node is obtained based on the operating data of the key nodes;

[0052] Determine the direction of the power flow. When the power flow direction is reversed and the reverse power transmission power is greater than or equal to 60% of the distribution transformer capacity, it is considered that there is a risk of reverse overload.

[0053] Furthermore, in S3, the real-time decision making includes the following steps:

[0054] Calculate the charging power of the energy storage system based on operating data P , the formula is as follows:

[0055] (11)

[0056] In the formula, P GE Indicates the power value of the transformer energy storage node;

[0057] Calculate the allowable power of the energy storage system P 1 ,judge P Relationship with the maximum charging power, if the charging power P Less than the maximum charging power, P 1 equal P ;otherwise, P 1 Equal to the maximum charging power;

[0058] Issue corresponding control commands to adjust the charging power of the energy storage system to P 1 , the overload power is partially or fully absorbed by the energy storage system, thereby reducing or eliminating the reverse overload.

[0059] Compared with the prior art, the present invention has the following advantages:

[0060] The present invention obtains the forecast data of photovoltaic power generation and load in the substation. Through the forecast data, it can identify the time periods of possible overload and high line loss in advance, and formulate and execute the day-ahead decision as the predetermined strategy for energy storage operation. Before the period with high reverse overload rate, the remaining power of the energy storage device is controlled to be at a low level. When entering the period with high reverse overload rate, the charging power of the energy storage device is increased to effectively absorb the overload power. In addition, during the execution of the day-ahead decision, the operating data of the key nodes in the substation is obtained in real time. When it is detected that the substation has a sudden abnormal situation of excessive overload rate or excessive line loss according to the operating data of the key nodes in the substation, the real-time strategy can be formulated and executed in time, and the charging and discharging power of the energy storage device can be adjusted immediately, so as to respond to and solve the problem as quickly as possible and ensure the stability and safety of the substation operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0062] Figure 1 It is the overall flow chart of the present invention;

[0063] Figure 2 It is a flow chart of K-means algorithm clustering analysis in the present invention;

[0064] Figure 3 This is a flow chart of the day-ahead prediction performed by the Informer algorithm in the present invention;

[0065] Figure 4 It is a node power change prediction curve diagram that has not been optimized by day-ahead decision in the present invention;

[0066] Figure 5 It is a node power variation curve diagram after day-ahead decision optimization during the specific implementation of the present invention;

[0067] Figure 6 It is a curve diagram of energy storage operation cost change before optimization in the present invention;

[0068] Figure 7 It is a curve diagram of energy storage operation cost change after day-ahead decision optimization during the specific implementation of the present invention;

[0069] Figure 8 It is a power variation curve diagram of the present invention when the photovoltaic power simulated at the ninth time point is greatly different from the predicted power;

[0070] Fig. 9 The present invention is a change curve diagram of the decision of the remaining time points after adjusting the energy storage power in real time at the ninth time point;

[0071] Fig.10 It is a system architecture diagram when the present invention is specifically implemented;

[0072] Fig.11 It is a flow chart of day-ahead decision and real-time decision in the present invention. DETAILED DESCRIPTION

[0073] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0074] like Figure 1-11 As shown, this embodiment proposes an energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief, including the following steps:

[0075] S1: Based on the K-means-XGBoost-Informer model and data set, the photovoltaic node power generation and user load node power of the next day are predicted, and the day-ahead decision is made and executed based on the constraints and objective function;

[0076] S2: During the execution of the day-ahead decision, the operation data of the key nodes in the substation area are obtained in real time, and whether there is a reverse overload is determined based on the operation data of the key nodes. If so, step S3 is executed; if not, step S3 is skipped and step S4 is executed;

[0077] S3: Suspend the execution of the day-ahead decision, make and execute the real-time decision based on the current SOC value of the energy storage device with the goal of minimizing the reverse power transmission, and while executing the real-time decision, update the day-ahead decision and obtain the operating data of the transformer energy storage node in real time and determine the power flow direction until the power flow direction of the transformer energy storage node becomes positive, thereby eliminating the reverse overload;

[0078] S4: Determine whether the day-ahead decision has been fully executed. If not, continue to execute the day-ahead decision and return to step S2. If all have been executed, exit.

[0079] In S1, predicting the photovoltaic node power generation and user load node power on the second day includes the following steps:

[0080] Preprocess the data in the data set, remove abnormal data points, and replace the abnormal data with the average of adjacent normal values;

[0081] The preprocessed data is normalized to eliminate the dimension difference. The formula is as follows:

[0082] (1)

[0083] In the formula, x0 is the initial data set, μ is the mean of the data set, σ is the standard deviation of the data set, x is the standardized data set;

[0084] It should be specially noted that abnormal data refers to data that is all zero or data that obviously exceeds other data by a large amount and has a very large difference from the average value of other data. In this embodiment, data that is all zero, data that exceeds other data by ten times or more, and data that is ten times or more of the average value of other data are all listed as abnormal data.

[0085] like Figure 2 As shown in the figure, the weather data in the data set are clustered using the K-means algorithm according to the three weather conditions of sunny, rainy and cloudy days to obtain photovoltaic data under different scenarios;

[0086] Considering the impact of environmental factors, the XGBoost algorithm is used to sort environmental variables and select environmental variables that have a great impact on the power generation of photovoltaic nodes and the power of user load nodes;

[0087] It should be specially noted that, in this embodiment, since photovoltaic power generation and electricity load are affected by various environmental factors, the XGBoost algorithm is used to sort the environmental variables to screen out the top five groups of variables that have a greater impact on photovoltaic power generation and electricity load.

[0088] During the construction process, the XGBoost algorithm intelligently determines which feature should be used as a split point by evaluating the contribution of each feature to the structural score gain. The importance of a feature is directly reflected in its cumulative number of appearances in all trees in the final integrated model, that is, the more frequently the feature is used to build a decision tree, the greater its importance.

[0089] In order to quantify this importance, a weighted sum operation is performed on the results of a feature in all boosted trees, and then the average is calculated to obtain a specific importance score. In this embodiment, the top five features and time power data form a training data set.

[0090] like Figure 3 As shown in the figure, the Informer model is trained using the data set, and the Informer model is used to perform day-ahead prediction to obtain the photovoltaic node power generation power and user load node power data for the second day.

[0091] In this embodiment, the dataset used is the open source Australian Alice Spring photovoltaic data on the Internet. There is no set error, and the prediction accuracy is reflected by MAE (mean absolute error), RMSE (root mean square error), and R² (coefficient of determination). The prediction accuracy of this prediction method can reach MAE=6.361, RMSE=15.315, and R²=0.982).

[0092] In S1, the constraints include SOC value constraints and charge and discharge power constraints; wherein the SOC value constraint formula is as follows:

[0093] (2)

[0094] In the formula, SOC t express t The state of charge of the energy storage device at all times, SOC min Indicates the minimum SOC value allowed by the energy storage device. SOC max Indicates the maximum SOC value allowed by the energy storage device; in this embodiment, SOC min The value range is , SOC max The value range is , SOC min The values ​​of include 0.1, SOC max The value of includes 0.9.

[0095] It should be noted that in this embodiment, the energy storage device includes an energy storage battery, etc. SOC t To measure the amount of electricity stored in the energy storage device itself, SOC t It is the ratio of the remaining capacity of the energy storage device to its fully charged capacity. The formula is as follows:

[0096] (3)

[0097] In the formula, express t The remaining power of the energy storage device at all times, Indicates the rated capacity of the energy storage device. In actual operation, .

[0098] In addition, the charging and discharging power constraint formula is as follows:

[0099] (4)

[0100] In the formula, express t The charging and discharging power of the energy storage device at all times, When it is positive, it indicates charging power, and when it is negative, it indicates discharging power. Indicates the maximum discharge power of the energy storage device. Indicates the maximum value of the charging power of the energy storage device. In this embodiment, and The values ​​include 100kW.

[0101] The charge and discharge power of the energy storage device directly affects the power of the energy storage device. In this embodiment, the relationship between the power of the energy storage battery and the charge and discharge power is as follows:

[0102] (5)

[0103] In the formula, express t The remaining power of the energy storage device at all times, express t- The remaining power of the energy storage device at 1 moment, η Represents the charging and discharging efficiency of the energy storage device. In this embodiment, η The values ​​of include 0.95, Δt express t Moment and t- 1 The time interval between moments.

[0104] In S1, the formula of the objective function is as follows:

[0105] (6)

[0106] In the formula, f 1 represents the variance of the power on the power supply side of the distribution network, f 2 Represents the total cost of 24-hour operation of the energy storage equipment, λ 1 express f 1 The weight of λ 2 express f 2 The weight of express f 1 The initial value of express f 2 The initial value of It represents the cost of energy storage cycle charging at the lowest electricity price and discharging at the highest electricity price. μ t Represents the penalty coefficient, which is related to the load rateβ t Related, P 1,t express t The active power on the power supply side of the distribution network at all times, Q 1,t express t The reactive power on the power supply side of the distribution network at all times, S N Indicates the rated capacity of the distribution network transformer;

[0107] It should be noted that the objective function formula is the sum of the power variance, operating cost, and load rate penalty on the power supply side. Ultimately, the optimal value is to minimize the function value.

[0108] In this embodiment, the calculation When the power supply side of the distribution network is used, the variance of the power supply side of the distribution network within 24 hours per day is calculated based on the historical data of the past year. Take the maximum value among them, ,in, It represents the cost of energy storage when it is charged at the highest electricity price and discharged at the lowest electricity price within a day. It represents the cost of energy storage cycle charging at the lowest electricity price and discharging at the highest electricity price. λ 1 The values ​​of include 0.9, λ 2 The value of includes 0.1.

[0109] Load factor β t The calculation formula is as follows:

[0110] (7)

[0111] Penalty coefficient μ t The calculation formula is as follows:

[0112] (8)

[0113] In the formula, Sgn ( P 1,t ) is the symbol function, P 1,t express t The active power on the power supply side of the distribution network at all times,

[0114] Q 1,t express t The reactive power on the power supply side of the distribution network at all times, S N Indicates the rated capacity of the distribution network transformer,β t A positive value indicates that the power flow direction is positive, i.e., the distribution transformer supplies power to the substation; β t A negative value means the power flow direction is negative, that is, the substation supplies power back to the distribution transformer.

[0115] It should be noted that in this embodiment, the variance of the power on the power supply side of the distribution network is f 1 Total cost of 24-hour operation of energy storage equipment f 2 The calculation of includes the following steps:

[0116] Define the power supply side of the distribution network as node 1, divide the 24 hours of a day into 24 time points, and calculate the active power and reactive power of the power supply side of the distribution network based on the predicted data. The formula is as follows:

[0117] (9)

[0118] In the formula, P 1,t express t The active power on the power supply side of the distribution network at all times, Q 1,t express t The reactive power on the power supply side of the distribution network at all times, U 1,t express t The voltage at node 1 at time t, U j,t express t time j The voltage at the node, N Indicates the size of the node, G 1j Indicates the power supply side and nodes of the distribution network in the line j The conductance between B 1j Indicates the power supply side and nodes of the distribution network in the line j The electrical susceptance between θ 1j,t express t time j Phase angle of the node;

[0119] In this embodiment, the forecast data includes the photovoltaic node power generation power, user load node power, etc. on the second day.

[0120] It should be noted that, when the application scenario of this embodiment has been determined, the conductance and admittance data of this photovoltaic area can be obtained as a fixed value; the above equation is a power flow equation, and the unknowns are P, Q, U and the phase angle. When P and Q are known, the voltage of each node can be calculated.

[0121] Summarize the active power on the power supply side of the distribution network within 24 hours and calculate the average power and variance of the power on the power supply side of the distribution network f 1 , the formula is as follows:

[0122] (10)

[0123] (11)

[0124] In the formula, P 1,t express t The active power on the power supply side of the distribution network at all times, Indicates the average power of the distribution network within 24 hours;

[0125] Calculate the total cost of 24-hour operation of energy storage equipment f 2 , the formula is as follows:

[0126] (12)

[0127] In the formula, Indicates the charging and discharging power of the energy storage device. B(t) represents the electricity price. In this embodiment, the electricity price is obtained from the previous electricity price.

[0128] In S2, key nodes include transformer energy storage nodes, distributed photovoltaic power generation nodes, and load side nodes, and operating data include voltage, current, active power, and reactive power.

[0129] In addition, in S2, the judgment of reverse overload includes the following steps:

[0130] The power flow direction of the transformer energy storage node is obtained based on the operating data of the key nodes;

[0131] Determine the direction of the flow. When the flow direction is reversed and the photovoltaic power generation power is greater than the user load power, it is considered that there is a risk of reverse overload.

[0132] In this embodiment, in order to ensure the stability of system operation, when the reverse power transmission is greater than or equal to 60% of the distribution transformer capacity, it is considered that there is a risk of reverse overload; when the reverse power transmission is less than 60% of the distribution transformer capacity, it is considered that the reverse overload has been eliminated.

[0133] In S3, real-time decision making includes the following steps:

[0134] Calculate the charging power of the energy storage system based on operating data P , the formula is as follows:

[0135] (13)

[0136] In the formula, P GE Indicates the power value of the transformer energy storage node;

[0137] Calculate the allowable power of the energy storage system P 1 ,judge P Relationship with the maximum charging power, if the charging power P Less than the maximum charging power, P 1 equal P ;otherwise, P 1 Equal to the maximum charging power;

[0138] Issue corresponding control commands to adjust the charging power of the energy storage system to P 1 , the overload power is partially or fully absorbed by the energy storage system, thereby reducing or eliminating the reverse overload.

[0139] In this embodiment, the purpose of the day-ahead optimization decision operation is mainly to optimize the overall charging and discharging power of the energy storage equipment in the substation within 24 hours, consider the power flow balance of the distribution network, take the substation overload rate and energy storage operation cost as the optimization targets, and construct the optimization operation constraints and objective function.

[0140] like Fig.11 As shown, in this embodiment, the formulation of the day-ahead decision includes the following steps: obtaining the forecast data for the next day, the forecast data including photovoltaic power generation and user load data; obtaining the time period when the photovoltaic power generation is much greater than the user load power demand according to the forecast data, and this time period has the risk of heavy overload, formulating constraints and objective functions, optimizing the energy storage charging and discharging power at 24 moments of the next day through the particle swarm algorithm, and formulating an optimization strategy; storing the optimization strategy in the cloud platform as the preset operation strategy of the energy storage system.

[0141] When the day-ahead decision is executed, the 24 moments are divided into 23 time periods, and one decision is executed in each time period. During the execution process, if the current decision is interrupted by a real-time decision, the cloud platform will re-make decisions for the remaining time periods based on the collected real-time SOC of the energy storage and replace the day-ahead decision.

[0142] The purpose of the previous day's optimization operation is mainly to optimize the overall charging and discharging power of the energy storage equipment in the substation within 24 hours, taking into account the balance of the distribution network flow. The optimization target is the reverse overload rate and the operating economy of the energy storage; according to the constraints and objective functions determined by the reverse overload rate and the operating economy of the energy storage, it is divided into 24 time periods and decision commands for 24 time periods are generated and issued to the energy storage equipment for execution, so as to achieve the goal of alleviating the reverse heavy overload in the substation and the economic operation of the energy storage.

[0143] The purpose of real-time operation decision-making is mainly to optimize the charging and discharging power of the energy storage equipment in the substation in case of sudden excessive reverse overload rate.

[0144] By obtaining the forecast data of photovoltaic and load in the substation, the possible overload and high line loss time periods can be identified in advance through the forecast data, and the day-ahead decision can be made as a predetermined strategy for energy storage operation and executed. Before the period with high reverse overload rate, the remaining power of the energy storage device is controlled to be at a low level. When entering the period with high reverse overload rate, the charging power of the energy storage device is increased to effectively absorb the overload power. In addition, during the execution of the day-ahead decision, the operating data of the key nodes in the substation is obtained in real time. When it is detected that the substation has a sudden overload rate or high line loss according to the operating data of the key nodes in the substation, the real-time strategy can be formulated and executed in time, and the charging and discharging power of the energy storage device can be adjusted immediately, so as to respond to and solve the problem as quickly as possible and ensure the stability and safety of the substation operation.

[0145] like Fig.10 As shown, the present invention also provides an energy storage optimization scheduling system taking into account operating costs and reverse heavy overload relief, the system implements an energy storage optimization scheduling method taking into account operating costs and reverse heavy overload relief, specifically, the system includes an energy storage system and a cloud computing center.

[0146] The energy storage system uses existing technologies to achieve interconnection with the cloud computing center. It can not only execute the preset operation logic, but also receive and execute the control commands of the cloud computing center to achieve integration with the distribution network of the substation area. It has operational management functions such as fast response and precise control, providing reactive power support, smoothing load fluctuations, and comprehensive operation monitoring. By cleverly using these functions, the energy storage system can ultimately achieve multiple goals such as peak shaving and valley filling, improving power quality, and alleviating reverse heavy overloads.

[0147] The cloud computing center consists of a data access management module, an algorithm model library module, a data resource pool module, a service registration management module, an intelligent analysis and processing module, an energy storage dispatch management module, and a cloud computing center. Data is uniformly accessed through the data access management module, and then enters the intelligent analysis and processing module for in-depth processing. The intelligent decision-making system consists of a unified perception access and cloud intelligent decision-making module, which interacts with the distribution network in real time through intelligent sensors to ensure the accuracy and timeliness of decisions. The algorithm model library module provides a variety of algorithm model support, while the data resource pool module is responsible for storing and managing data resources within the system. The service registration management module ensures the effective registration and management of services and provides stable service support for the entire system. The entire system architecture achieves efficient data access, processing, analysis, and decision-making through close collaboration between modules, providing strong technical support for the cloud computing center.

[0148] The energy storage optimization scheduling system of the present invention that takes into account the operation cost and the relief of reverse heavy overload is a "cloud-edge-end" collaborative energy storage optimization operation system architecture that takes into account the heavy overload of the substation and the operation cost of energy storage. The system is designed and optimized, and can adjust the charging and discharging of the energy storage located at the head end of the substation in real time according to the operation status of the substation distribution network. Its core functions are day-ahead decision-making and real-time decision-making. Day-ahead decision-making optimizes the 24 hours before the day through the forecast data of the substation, and the optimization targets are the reverse overload rate and the operation economy of the energy storage; when making real-time decisions, the data is collected through the intelligent sensors of the substation, and the cloud platform judges and makes decisions, and sends instructions to the energy storage equipment, which is executed by the energy storage equipment. Through day-ahead decision-making and implementation of decisions, the goals of alleviating the reverse heavy overload of the substation and the operation economy of energy storage are achieved.

[0149] When the day-ahead decision optimization system is implemented, it can determine the time period when heavy overload is likely to occur based on the forecast data of photovoltaic and load, and lower the SOC of the energy storage by discharging the energy storage before that. When a heavy overload occurs, it absorbs power for charging to alleviate the reverse heavy overload and ensure the stability and reliability of the power supply to the substation.

[0150] When the real-time decision optimization system is implemented, it can monitor the current, voltage, power and other parameters of the substation in real time, automatically judge the operation status of the substation and make optimization decisions on the energy storage working status. The cloud platform judges and makes decisions and sends instructions to the energy storage equipment for execution. Through day-ahead decision-making and implementation, the goal of alleviating reverse overload in the substation and the economic operation of energy storage is achieved, ensuring the stability and reliability of power supply in the substation.

[0151] The energy storage optimization scheduling method of the present invention taking into account the operation cost and the reverse heavy overload relief has significant innovation and practicality, which can not only greatly alleviate the reverse heavy overload of the substation area and reduce the energy storage operation cost, but also improve the stability and reliability of the power supply of the substation area. In particular, for some remote areas or areas with a large number of photovoltaic access, the energy storage optimization system of the present invention can play an important role.

[0152] Please refer to the attached Figure 4-9 In the specific implementation, the particle swarm algorithm is used to solve the day-ahead optimization decision. Using IEEE33 nodes, photovoltaics are connected to nodes 10, 15, and 30, and energy storage is connected to node 2. The photovoltaic operation data, node data, time-of-use electricity price and other data are obtained. It is assumed that the original node data does not change within 24 hours. The power of photovoltaics, energy storage, and distribution networks is unified, and the maximum charging and discharging power of energy storage is set to 100kW, the energy storage capacity is 300kWh, the initial SOC is set to 0.5, and the distribution transformer capacity is set to 315kVA. When β t Between 0 and 100%, μ Take 0; when it exceeds 100% or is lower than -60%, μ Take 1000.

[0153] Among them, the total operating cost of energy storage is the accumulation of the operating costs of all nodes, that is, the 24th data is the total cost of energy storage operation in one day (explanation: there is a negative cost, because when energy storage supplies electricity to the grid, it is considered that the grid buys electricity from the energy storage, which is considered as income, and negative cost represents income).

[0154] During simulation, please refer to Figure 4 At around the 10th time point, due to the large photovoltaic power generation, reverse power transmission occurs. At this time, the reverse power transmission power exceeds the limit. Please refer to Figure 5 , under the influence of the day-ahead decision, after the 10th time point, the injected power of the distribution network did not become negative, which shows that the reverse overload situation was alleviated. Please refer to Figure 6 and Figure 7 , under the influence of day-ahead decision-making, negative costs represent benefits, and the operating costs of energy storage are greatly reduced. Please refer to Figure 8 and Figure 4 , Figure 8 At around the ninth time point, the actual photovoltaic power generation was higher than predicted. Fig. 9 , at this time, the system detects the reverse flow in real time, stops the day-ahead decision, makes a real-time decision, and mobilizes energy storage for charging. In addition, the day-ahead decision is completed again at the ninth to tenth time points based on the implementation decision and forecast data. When the implementation decision is completed, the new day-ahead decision for the corresponding time node is implemented. It can be seen from this that this management method can effectively alleviate the reverse overload situation in the photovoltaic area and significantly reduce the operating cost of energy storage.

[0155] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for optimizing energy storage scheduling taking into account operating costs and reverse heavy overload relief, characterized in that: The following steps are involved: S1: Based on the K-means-XGBoost-Informer model and data set, the photovoltaic node power generation and user load node power of the next day are predicted, and the day-ahead decision is made and executed based on the constraints and objective function; S2: During the execution of the day-ahead decision, the operation data of the key nodes in the substation area are obtained in real time, and whether there is a reverse overload is determined based on the operation data of the key nodes. If so, step S3 is executed; if not, step S3 is skipped and step S4 is executed; S3: Suspend the execution of the day-ahead decision, make and execute the real-time decision based on the current SOC value of the energy storage device with the goal of minimizing the reverse power transmission, and while executing the real-time decision, update the day-ahead decision and obtain the operating data of the transformer energy storage node in real time and determine the power flow direction until the power flow direction of the transformer energy storage node becomes positive, thereby eliminating the reverse overload; S4: Determine whether the day-ahead decision has been fully executed. If not, continue to execute the day-ahead decision and return to step S2. If all have been executed, exit. Predicting the photovoltaic node power generation and user load node power on the second day includes the following steps: Preprocess the data in the data set, remove abnormal data points, and replace the abnormal data with the average of adjacent normal values; The preprocessed data is normalized to eliminate the dimension difference. The formula is as follows: （1） In the formula, x 0 is the initial data set, μ is the mean of the data set, σ is the standard deviation of the data set, x is the standardized data set; The weather data in the dataset is clustered using the K-means algorithm according to the three weather conditions of sunny, rainy, and cloudy days to obtain photovoltaic data under different scenarios; Considering the impact of environmental factors, the XGBoost algorithm is used to sort environmental variables and select environmental variables that have a great impact on the power generation of photovoltaic nodes and the power of user load nodes; Use the data set to train the Informer model, and use the Informer model to perform day-ahead forecasts to obtain the photovoltaic node power generation and user load node power data for the next day; The objective function formula is as follows: （2） In the formula, f 1 represents the variance of the power on the power supply side of the distribution network, f 2 represents the total cost of 24-hour operation of the energy storage equipment. λ 1 means f The weight is 1, λ 2 means f The weight is 2, express f The initial value is 1, express f The initial value is 2, It represents the cost of energy storage cycle charging at the lowest electricity price and discharging at the highest electricity price. μ t Represents the penalty coefficient, which is related to the load rate β t Related, P 1,t express t The active power on the power supply side of the distribution network at all times, Q 1,t express t The reactive power on the power supply side of the distribution network at all times, S N Indicates the rated capacity of the distribution network transformer; Load factor β t The calculation formula is as follows: （3） Penalty coefficient μ t The calculation formula is as follows: （4） In the formula, Sgn ( P 1,t ) is the symbol function, P 1,t express t The active power on the power supply side of the distribution network at all times, Q 1,t express t The reactive power on the power supply side of the distribution network at all times, S N Indicates the rated capacity of the distribution network transformer, β t A positive value indicates that the power flow direction is positive, i.e., the distribution transformer supplies power to the substation; β t A negative value indicates that the power flow direction is negative, that is, the substation sends power back to the distribution transformer; Real-time decision making involves the following steps: Calculate the charging power of the energy storage system based on operating data P , the formula is as follows: （5） In the formula, P GE Indicates the power value of the transformer energy storage node; Calculate the allowable power of the energy storage system P 1. Judgment P Relationship with the maximum charging power, if the charging power P Less than the maximum charging power, P 1 is equal to P ;otherwise, P 1 equals the maximum charging power; Issue corresponding control commands to adjust the charging power of the energy storage system to P 1. The overload power is partially or completely absorbed by the energy storage system, thereby reducing or eliminating the reverse overload.

2. The energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief according to claim 1 is characterized in that: In S1, the constraints include SOC value constraints and charge and discharge power constraints; the SOC value constraint formula is as follows: （6） In the formula, SOC t express t The state of charge of the energy storage device at all times, SOC min Indicates the minimum SOC value allowed by the energy storage device. SOC max Indicates the maximum SOC value allowed by the energy storage device; The charging and discharging power constraint formula is as follows: （7） In the formula, represents the charging and discharging power of the energy storage device at time t, where: When it is positive, it indicates charging power, and when it is negative, it indicates discharging power; Indicates the maximum discharge power of the energy storage device. Indicates the maximum charging power of the energy storage device.

3. The energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief according to claim 2 is characterized in that: SOC min The values ​​of include 0.1, SOC max The values ​​of include 0.9, and The values ​​include 100kW.

4. The energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief according to claim 1 is characterized in that: λ The value of 1 includes 0.9, λ The value of 2 includes 0.

1.

5. The energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief according to claim 1 is characterized in that: Variance of power on the power supply side of the distribution network f 1. Total cost of 24-hour operation of energy storage equipment f The calculation of 2 includes the following steps: Define the power supply side of the distribution network as node 1, divide the 24 hours of a day into 24 time points, and calculate the active power and reactive power of the power supply side of the distribution network based on the predicted data. The formula is as follows: （8） In the formula, P 1,t express t The active power on the power supply side of the distribution network at all times, Q 1,t express t The reactive power on the power supply side of the distribution network at all times, U 1,t express t The voltage at node 1 at time t, U j,t express t time j The voltage at the node, N Indicates the size of the node, G 1j Indicates the power supply side and nodes of the distribution network in the line j The conductance between B 1j Indicates the power supply side and nodes of the distribution network in the line j The electrical susceptance between θ 1j,t express t time j Phase angle of the node; Summarize the active power on the power supply side of the distribution network within 24 hours and calculate the average power and variance of the power on the power supply side of the distribution network f 1. The formula is as follows: （9） （10） In the formula, P 1,t express t The active power on the power supply side of the distribution network at all times, Indicates the average power of the distribution network within 24 hours; Calculate the total cost of 24-hour operation of energy storage equipment f 2. The formula is as follows: （11） In the formula, Indicates the charging and discharging power of the energy storage device. B ( t ) represents the electricity price.

6. The energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief according to claim 1 is characterized in that: In S2, key nodes include transformer energy storage nodes, distributed photovoltaic power generation nodes, and load side nodes, and operating data include voltage, current, active power, and reactive power.

7. The energy storage optimization scheduling method taking into account operation cost and reverse heavy overload relief according to claim 6 is characterized in that: In S2, the determination of reverse overload includes the following steps: The power flow direction of the transformer energy storage node is obtained based on the operating data of the key nodes; Determine the direction of the power flow. When the power flow direction is reversed and the reverse power transmission power is greater than or equal to 60% of the distribution transformer capacity, it is considered that there is a risk of reverse overload.

Images