A distributed energy storage data-driven predictive control method for optimizing station voltage

Abstract

The present invention provides a distributed energy storage data-driven predictive control method for optimizing substation voltage, which belongs to the technical field of power system operation and control, including: collecting distribution network energy storage output and corresponding node voltage amplitude data at minute-level time sections to obtain historical data, using the historical data to construct a data model matrix, building a data-driven voltage optimization model, monitoring the real-time voltage of the substation and solving the data model matrix to obtain the control instructions of the prediction time window, selecting the control instructions of the prediction time window at the first moment, using an energy storage device to manage the over-limit voltage and synchronously updating the historical input and output vectors of the corresponding data model, realizing real-time rolling optimization of the substation voltage, and optimizing the over-limit problem of the substation voltage only through local measurement data.

Description

Technical Field

[0001] The present invention relates to the technical field of power system operation and control, and in particular to a distributed energy storage data-driven predictive control method for optimizing station area voltage. Background Art

[0002] The data-driven predictive control (DeePC) algorithm can achieve optimal control of system performance only through the input and output data of the system when the specific structure of the system is unknown.

[0003] Generally speaking, physical models can be built for specific problems, and optimization models can be solved to resolve the problem of voltage exceeding the limit in the substation. However, the incomplete and inaccurate data such as network topology and line parameters make the construction of the model increasingly complex and difficult, making it difficult to efficiently manage the voltage in the substation. The rapid development of artificial intelligence has also provided a data-driven voltage optimization method that does not require precise modeling. However, it relies on a large amount of data for learning, and the training effect is affected by multiple factors such as model hyperparameters and reward functions. It lacks perfect theoretical guidance, and the control optimization effect is unstable.

[0004] Generally speaking, distribution network voltage optimization requires a large amount of measurement data as support. However, uneven data quality, such as data loss, will affect the accuracy of the built model, and thus affect the effect of voltage optimization. How to achieve model-free control of distribution network voltage under the condition of considerable local data is one of the important technical issues in this field.

[0005] Therefore, the present invention proposes a distributed energy storage data-driven predictive control method for optimizing station area voltage. Summary of the invention

[0006] The present invention provides a distributed energy storage data-driven predictive control method for optimizing substation voltage, which is used to solve the technical problem existing in the above-mentioned prior art, namely, it is difficult to achieve efficient substation voltage optimization through local observation data.

[0007] The present invention provides a distributed energy storage data-driven predictive control method for optimizing station area voltage, comprising:

[0008] Step 1: Collect the energy storage output of the distribution network and the voltage amplitude data of the corresponding nodes at the minute level to obtain historical data;

[0009] Step 2: constructing a data model matrix using the historical data, wherein after the number of elements in the data model matrix meets a preset condition, the data model matrix is ​​made full rank;

[0010] Step 3: Build a data-driven voltage optimization model, monitor the real-time voltage of the substation, and solve the data model matrix to obtain the control instructions of the prediction time window;

[0011] Step 4: Select the control instruction at the first moment of the prediction time window, use the energy storage device to control the over-limit voltage and synchronously update the historical input and output vectors of the corresponding data model to achieve real-time rolling optimization of the substation voltage.

[0012] Preferably, historical data is obtained, including:

[0013] Export the data in the distribution network monitoring and acquisition system, filter and integrate continuous and complete valid data, where the valid data includes: the active output of energy storage at each time section and the voltage amplitude data of the node;

[0014] The difference calculation is performed on the active output of the energy storage and the voltage amplitude data of the node respectively, so as to form historical data with the change of the active output of the energy storage as the historical input and the change of the voltage amplitude as the historical output.

[0015] Preferably, constructing a data model matrix using the historical data includes:

[0016] Selecting a vector quantity parameter T so that the data model matrix, i.e., the Hanke l matrix, is full rank;

[0017] Use historical data to build the data model input matrix and data model output matrix, and combine the historical time window T ini And the prediction time window T f The size of the data model input and data model output matrix is ​​divided into blocks to obtain the historical input matrix P p 、History output matrix V p , prediction input matrix P f And the predicted output matrix V f .

[0018] Preferably, monitoring the real-time voltage of the substation and solving the data model matrix to obtain the control instruction of the prediction time window includes:

[0019] Initializing the historical input and historical output of the data model to build a data-driven voltage optimization model;

[0020] The voltage of key nodes is monitored and judged in real time. If the voltage exceeds the limit, the energy storage output is calculated and instructions are issued through the voltage optimization model to effectively optimize the substation voltage. At the same time, the data in the data model is synchronously updated to complete the data-driven model predictive control within the rolling cycle.

[0021] Preferably, the difference calculation is performed on the active output of the energy storage and the voltage amplitude data of the node, including:

[0022] Calculate the active power output and voltage amplitude at the second moment of any two adjacent moments minus the active power output and voltage amplitude at the first moment, and get sequence;

[0023]

[0024] in, They represent the active output vector of the energy storage at different times and the voltage amplitude vector of the node at different times, and the values ​​at different times are 1, 2, 3, ..., T, T+1; They represent the active power output at the corresponding moment and the output difference between any two adjacent moments respectively; They respectively represent the voltage amplitude at the corresponding moment and the amplitude difference between any two moments.

[0025] Preferably, the vector quantity parameter T is selected so that the data model matrix, i.e., the Hanke l matrix, is full rank, including:

[0026]

[0027]

[0028] in, Respectively represent the number of rows T ini The system input and output Hanke l matrix blocks, Respectively represent the number of rows T f The system input and output Hankel matrix blocks are constructed from known historical data, and the input and output vectors contain T components: Indicates based on Sequence and make the number of matrix rows T ini +T f The Hankel matrix of Indicates based on Sequence and make the number of matrix rows T ini +T f The Hankel matrix of .

[0029] Preferably, the data in the data model is synchronously updated to complete the data-driven model predictive control within the rolling cycle, including:

[0030] Historical input to the data model And historical output Initialize;

[0031] At time t, the initial input vector is set to The initial output vector is After the optimization is completed at each moment, the initial input vector and the initial output vector need to be optimized synchronously to realize the rolling predictive control of the model.

[0032] Preferably, the voltage optimization model

[0033]

[0034] in, They represent the system input vector and system output vector at time k, that is, the results obtained by solving the voltage optimization model. To solve the active output vector of the obtained energy storage, is the node voltage amplitude vector after energy storage management.

[0035] Preferably, it also includes:

[0036] Determine the lifetime cost of the battery C bat :

[0037]

[0038] Where: Indicates the depreciation cost corresponding to the unit charge and discharge amount; and They represent the charging power and discharging power of the energy storage, respectively; Δt is the unit time interval; T is the total number of time intervals;

[0039] Determine the cost per kilowatt-hour of battery degradation

[0040]

[0041] Where: L E is the total charge and discharge capacity of the energy storage over its entire life cycle; C APV is the annual value of battery investment and operation and maintenance costs, T calendar Calendar life of the battery's stored energy;

[0042] The annual value of battery investment and operation and maintenance costs is corrected using the capital return factor:

[0043]

[0044] Where: C PV C is the energy storage investment cost obtained by multiplying the unit investment cost by the capacity; op is the annual operation and maintenance cost; r is the discount rate; C APV To correct the results;

[0045] Get the voltage optimization model considering the energy storage life decay:

[0046]

[0047] Where: T f To predict the time window size, P, Q, and R are the weight matrices in the objective function that minimize the deviation of the observation node, the change in the active output of the energy storage, and the minimum energy storage life attenuation cost. μ is the voltage reference value, and λ is ini is the penalty coefficient.

[0048] Preferably, it also includes:

[0049] Determine the charging and discharging power constraints and SOC constraints of energy storage:

[0050]

[0051] SOC min ≤SOC i,t ≤SOC max

[0052] Where: P max , P min They are the maximum and minimum charging and discharging power of energy storage, respectively. The value of energy storage charging is expressed as positive, and the value of discharging is expressed as negative. It indicates that the absolute value of all elements of the active output vector of the energy storage obtained is the largest, ensuring that the charging and discharging power of the energy storage is within a certain range, SOC min , SOC i,t-1 , SOC i,t , SOC max They represent the minimum state of charge of the energy storage system, the state of charge of the i-th energy storage system at the previous moment t-1, the state of charge of the i-th energy storage system at time t, and the maximum state of charge of the energy storage system, η c,i , η d,j Respectively represent the charging and discharging efficiency of the i-th energy storage system, They represent the charging power and discharging power of the i-th energy storage system at time t, respectively, and E ESS,i represents the capacity of the i-th energy storage system, and Δt is the unit time interval.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] There is no need to build a precise model to carry out further optimization control. Under the conditions of unknown network topology and specific line parameters, partial data can be used to build a data-driven predictive control model. In the actual control process, the voltage over-limit problem in the substation area can be optimized only through local measurement data. The factor of energy storage life attenuation can be considered at the same time, and the economy of the optimization process can be taken into account.

[0055] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0056] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0058] Figure 1 It is a flow chart of a distributed energy storage data-driven predictive control method for optimizing station area voltage in an embodiment of the present invention;

[0059] Figure 2 A solution diagram of a distributed energy storage data-driven predictive control method for optimizing station area voltage in an embodiment of the present invention;

[0060] Figure 3 It is a daily simulation curve of voltage variation of key nodes in the substation without any control optimization means in the embodiment of the present invention;

[0061] Figure 4 In the embodiment of the present invention, based on Figure 3 Daily simulation curve of voltage variation at key nodes in the substation area;

[0062] Figure 5 The model algorithm in the embodiment of the present invention is based on Figure 3 Daily simulation curve of voltage variation at key nodes in the substation area. DETAILED DESCRIPTION

[0063] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0064] The present invention provides a distributed energy storage data driven predictive control method for optimizing the voltage in the substation area. Figure 1 As shown, including:

[0065] Step 1: Collect the energy storage output of the distribution network and the voltage amplitude data of the corresponding nodes at the minute level to obtain historical data;

[0066] Step 2: constructing a data model matrix using the historical data, wherein after the number of elements in the data model matrix meets a preset condition, the data model matrix is ​​made full rank;

[0067] Step 3: Build a data-driven voltage optimization model, monitor the real-time voltage of the substation, and solve the data model matrix to obtain the control instructions of the prediction time window;

[0068] Step 4: Select the control instruction at the first moment of the prediction time window, use the energy storage device to control the over-limit voltage and synchronously update the historical input and output vectors of the corresponding data model to achieve real-time rolling optimization of the substation voltage.

[0069] Preferably, historical data is obtained, including:

[0070] Export the data in the distribution network monitoring and acquisition system, filter and integrate continuous and complete valid data, where the valid data includes: the active output of energy storage at each time section and the voltage amplitude data of the node;

[0071] The difference calculation is performed on the active output of the energy storage and the voltage amplitude data of the node respectively, so as to form historical data with the change of the active output of the energy storage as the historical input and the change of the voltage amplitude as the historical output.

[0072] Preferably, constructing a data model matrix using the historical data includes:

[0073] Selecting a vector quantity parameter T so that the data model matrix, i.e., the Hankel matrix, is full rank;

[0074] Use historical data to build the data model input matrix and data model output matrix, and combine the historical time window T ini And the prediction time window T f The size of the data model input and data model output matrix is ​​divided into blocks to obtain the historical input matrix P p 、History output matrix V p , prediction input matrix P f And the predicted output matrix V f .

[0075] Preferably, monitoring the real-time voltage of the substation and solving the data model matrix to obtain the control instruction of the prediction time window includes:

[0076] Initializing the historical input and historical output of the data model to build a data-driven voltage optimization model;

[0077] The voltage of key nodes is monitored and judged in real time. If the voltage exceeds the limit, the energy storage output is calculated and instructions are issued through the voltage optimization model to effectively optimize the substation voltage. At the same time, the data in the data model is synchronously updated to complete the data-driven model predictive control within the rolling cycle.

[0078] Preferably, the difference calculation is performed on the active output of the energy storage and the voltage amplitude data of the node, including:

[0079] Calculate the active power output and voltage amplitude at the second moment of any two adjacent moments minus the active power output and voltage amplitude at the first moment, and get sequence;

[0080]

[0081] in, They represent the active output vector of the energy storage at different times and the voltage amplitude vector of the node at different times, and the values ​​at different times are 1, 2, 3, ..., T, T+1; They represent the active power output at the corresponding moment and the output difference between any two adjacent moments respectively; They respectively represent the voltage amplitude at the corresponding moment and the amplitude difference between any two moments.

[0082] Preferably, the vector quantity parameter T is selected so that the data model matrix, i.e., the Hankel matrix, is full rank, including:

[0083]

[0084] in, Respectively represent the number of rows T ini The system input and output Hankel matrix blocks, Respectively represent the number of rows T f The system input and output Hankel matrix blocks are constructed from known historical data, and the input and output vectors contain T components: Indicates based on Sequence and make the number of matrix rows T ini +T f The Hankel matrix of Indicates based on Sequence and make the number of matrix rows T ini +T f The Hankel matrix of .

[0085] In this embodiment, are all input / output vectors constructed from historical data matrices. The brackets indicate the number of components in the vector sequence. For example, Represents the Tth input vector ini +T f Components, T ini +T f is the number of rows of the Hankel matrix, TT ini +T f +1 is the number of columns of the Hankel matrix, thereby constructing a Hankel matrix consisting of T components.

[0086] Preferably, the data in the data model is synchronously updated to complete the data-driven model predictive control within the rolling cycle, including:

[0087] Historical input to the data model And historical output Initialize;

[0088] At time t, the initial input vector is set to The initial output vector is After the optimization is completed at each moment, the initial input vector and the initial output vector need to be optimized synchronously to realize the rolling predictive control of the model.

[0089] In this embodiment, Col is an abbreviation of "column", and col() indicates that the elements therein are represented as column vectors.

[0090] Represents the initial input / output vector, taking the T closest to the current time ini data, assuming There are 200 components in total, and the Hankel matrix is ​​constructed from these 200 components. Tini = 5, t = k, that is, before the voltage optimization solution is performed at the current moment, Pick The 196th to 200th components (Tini in total) in the components need to be solved at time k by making the same difference to obtain the And record; at time t = k + 1, Will be updated to The 197th to 200th components of the components are added to the solution when t = k. (There are a total of Tini), and so on, updated gradually.

[0091] Same reason.

[0092] Preferably, the voltage optimization model

[0093]

[0094] in, They represent the system input vector and system output vector at time k, that is, the results obtained by solving the voltage optimization model. To solve the active output vector of the obtained energy storage, is the node voltage amplitude vector after energy storage management.

[0095] In this embodiment, g is a column vector (association vector) associating the historical input / output matrix with the system input / output vector. This matrix will be solved together with the unknown vector during the solution process. Through this vector, the connection between the historical input / output matrix and the historical input / output vector can be established, and the relationship between the predicted input / output matrix and the predicted input / output vector can also be associated.

[0096] Preferably, it also includes:

[0097] Determine the lifetime cost of the battery C bat :

[0098]

[0099] Where: Indicates the depreciation cost corresponding to the unit charge and discharge amount; and They represent the charging power and discharging power of the energy storage, respectively; Δt is the unit time interval; T is the total number of time intervals;

[0100] Determine the cost per kilowatt-hour of battery degradation

[0101]

[0102] Where: L E is the total charge and discharge capacity of the energy storage over its entire life cycle; C APV is the annual value of battery investment and operation and maintenance costs, T calendar Calendar life of the battery's stored energy;

[0103] The annual value of battery investment and operation and maintenance costs is corrected using the capital return factor:

[0104]

[0105] Where: C PV C is the energy storage investment cost obtained by multiplying the unit investment cost by the capacity; op is the annual operation and maintenance cost; r is the discount rate; C APV To correct the results;

[0106] Get the voltage optimization model considering the energy storage life decay:

[0107]

[0108] Where: T f To predict the time window size, P, Q, and R are the weight matrices in the objective function that minimize the deviation of the observation node, the change in the active output of the energy storage, and the minimum energy storage life attenuation cost. μ is the voltage reference value, and λ is ini is the penalty coefficient.

[0109] Preferably, it also includes:

[0110] Determine the charging and discharging power constraints and SOC constraints of energy storage:

[0111]

[0112] SOC min ≤SOC i,t ≤SOC max

[0113] Where: P max , P min They are the maximum and minimum charging and discharging power of energy storage, respectively. The value of energy storage charging is expressed as positive, and the value of discharging is expressed as negative. It indicates that the absolute value of all elements of the active output vector of the energy storage obtained is the largest, ensuring that the charging and discharging power of the energy storage is within a certain range, SOC min , SOC i,t-1 , SOC i,t , SOC max They represent the minimum state of charge of the energy storage system, the state of charge of the i-th energy storage system at the previous moment t-1, the state of charge of the i-th energy storage system at time t, and the maximum state of charge of the energy storage system, η c,i , η d,j Respectively represent the charging and discharging efficiency of the i-th energy storage system, They represent the charging power and discharging power of the i-th energy storage system at time t, respectively, and E ESS,i represents the capacity of the i-th energy storage system, and Δt is the unit time interval.

[0114] In this embodiment, the Hankel matrix is ​​defined as:

[0115]

[0116] Where: ω(·) is a column vector of the input signal, which can contain m variables. When the matrix H L When (ω) is full rank, the input signal ω can achieve continuous excitation, where the selection of the number of vectors T must satisfy T≥(m+1)L-1, where m is the number of variables contained in a single vector, and L is the minimum integer that makes the system controllable and observable.

[0117] like Figure 2As shown in the figure, it is mainly divided into the data model construction step and the rolling control optimization step. In the data model construction framework, the historical data of the key node voltage and the energy storage output must be obtained first. After the data is differenced and the parameters required for the optimization process are selected, the Hankel matrix is ​​constructed and the input and output block matrices of the distribution network data-driven model are obtained from it, and the initialization input and output vectors are initialized. On the basis of the above preparations, the real-time optimization control of the distribution network is carried out. First, the voltage at the current moment is collected and it is judged whether the voltage of each node exceeds the limit. If there is an over-limit situation, the data-driven model and the input and output vectors must be updated and solved through the rolling control optimization step to obtain the energy storage action state that optimizes the substation voltage, and the control command is issued to manage the over-limit voltage in the substation, and then it is judged whether the optimization cycle is over; if the voltage at the current moment is not over-limit, the energy storage device does not take any action until the substation voltage state is judged again at the next moment, thereby forming a complete distributed energy storage data-driven predictive control process for optimizing the substation.

[0118] Please refer to Figure 3 and Figure 4 ,from Figure 3 It can be seen that due to the access of renewable energy, especially photovoltaics, the node load curve changes show more serious fluctuations, and even overvoltage or low voltage occurs at some times. Figure 4 It can be seen that the proposed method can effectively optimize the problem of voltage over-limit in the substation area and limit the voltage of each node to a safe range.

[0119] Please refer to Figure 3 and Figure 5 .from Figure 3 It can also be seen that due to the access of renewable energy, especially photovoltaics, the node load curve changes show more serious fluctuations, and even overvoltage or low voltage phenomena occur at some times. Figure 5 It can be seen that the voltage over-limit problem in the control area has been effectively optimized.

[0120] Please refer to Figure 4 and Figure 5 . Figure 4 For Figure 3 The voltage over-limit problem is shown in the following figure: the voltage variation daily simulation curve of the key nodes in the back-end area is implemented according to the distributed energy storage data-driven predictive control method for optimizing the area voltage. Figure 4 Yes Figure 2The daily simulation curve of voltage change at the key nodes of the substation after the voltage over-limit problem is controlled by the model algorithm is shown. The two methods can achieve almost the same effect in optimizing the voltage, but the advantage of the former over the latter is that it does not require the precise construction of the distribution network model for further control optimization, but only achieves the control optimization effect equivalent to the model method through local measurement data.

[0121] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A distributed energy storage data-driven predictive control method for optimizing station area voltage, characterized in that: include: Step 1: Collect the energy storage output of the distribution network and the voltage amplitude data of the corresponding nodes at the minute level to obtain historical data; Step 2: constructing a data model matrix using the historical data, wherein after the number of elements in the data model matrix meets a preset condition, the data model matrix is ​​made full rank; Step 3: Build a data-driven voltage optimization model, monitor the real-time voltage of the substation, and solve the data model matrix to obtain the control instructions of the prediction time window; Step 4: Select the control instruction at the first moment of the prediction time window, use the energy storage device to manage the over-limit voltage and synchronously update the historical input and output vectors of the corresponding data model to achieve real-time rolling optimization of the substation voltage; Wherein, constructing a data model matrix using the historical data includes: Selecting a vector quantity parameter T so that the data model matrix, i.e., the Hankel matrix, is full rank; Use historical data to build the data model input matrix and data model output matrix, and combine the historical time window T ini And the prediction time window T f The size of the data model input and data model output matrix is ​​divided into blocks to obtain the historical input matrix P p 、History output matrix V p , prediction input matrix P f And the predicted output matrix V f ; Among them, the voltage optimization model in, They represent the system input vector and system output vector at time k, that is, the results obtained by solving the voltage optimization model. To solve the active output vector of the obtained energy storage, is the node voltage amplitude vector after energy storage management.

2. The distributed energy storage data-driven predictive control method for optimizing the area voltage according to claim 1 is characterized in that: Get historical data, including: Export the data in the distribution network monitoring and acquisition system, filter and integrate continuous and complete valid data, where the valid data includes: the active output of energy storage at each time section and the voltage amplitude data of the node; The difference calculation is performed on the active output of the energy storage and the voltage amplitude data of the node respectively, so as to form historical data with the change of the active output of the energy storage as the historical input and the change of the voltage amplitude as the historical output.

3. The distributed energy storage data-driven predictive control method for optimizing the area voltage according to claim 1 is characterized in that: Monitoring the real-time voltage of the substation area and solving the data model matrix to obtain the control instructions of the prediction time window includes: Initializing the historical input and historical output of the data model to build a data-driven voltage optimization model; The voltage of key nodes is monitored and judged in real time. If the voltage exceeds the limit, the energy storage output is calculated and instructions are issued through the voltage optimization model to effectively optimize the substation voltage. At the same time, the data in the data model is synchronously updated to complete the data-driven model predictive control within the rolling cycle.

4. The distributed energy storage data-driven predictive control method for optimizing the area voltage according to claim 2 is characterized in that: The difference calculation is performed on the active output of the energy storage and the voltage amplitude data of the node, including: Calculate the active power output and voltage amplitude at the second moment of any two adjacent moments minus the active power output and voltage amplitude at the first moment, and get sequence; in, They represent the active output vector of the energy storage at different times and the voltage amplitude vector of the node at different times, and the values ​​at different times are 1, 2, 3, ..., T, T+1; They represent the active power output at the corresponding moment and the output difference between any two adjacent moments respectively; They respectively represent the voltage amplitude at the corresponding moment and the amplitude difference between any two moments.

5. The distributed energy storage data-driven predictive control method for optimizing the area voltage according to claim 2 is characterized in that: The vector quantity parameter T is selected so that the data model matrix, i.e., the Hankel matrix, is full rank, including: in, Respectively represent the number of rows T ini The system input and output Hankel matrix blocks, Respectively represent the number of rows T f The system input and output Hankel matrix blocks are constructed from known historical data, and the input and output vectors contain T components: Indicates based on Sequence and make the number of matrix rows T ini +T f The Hankel matrix of Indicates based on Sequence and make the number of matrix rows T ini +T f The Hankel matrix of .

6. The distributed energy storage data-driven predictive control method for optimizing the area voltage according to claim 3 is characterized in that: Synchronously update the data in the data model to complete the data-driven model predictive control within the rolling cycle, including: At time t, the initial input vector is set to The initial output vector is After the optimization is completed at each moment, the initial input vector and the initial output vector need to be optimized synchronously to realize the rolling predictive control of the model.

7. The distributed energy storage data-driven predictive control method for optimizing the area voltage according to claim 1 is characterized in that: Also includes: Determine the lifetime cost of the battery C bat : Where: Indicates the depreciation cost corresponding to the unit charge and discharge amount; and They represent the charging power and discharging power of the energy storage, respectively; Δt is the unit time interval; T is the total number of time intervals; Determine the cost per kilowatt-hour of battery degradation Where: L E is the total charge and discharge capacity of the energy storage over its entire life cycle; C APV is the annual value of battery investment and operation and maintenance costs, T calendar Calendar life of the battery's stored energy; The annual value of battery investment and operation and maintenance costs is corrected using the capital return factor: Where: C PV C is the energy storage investment cost obtained by multiplying the unit investment cost by the capacity; op is the annual operation and maintenance cost; r is the discount rate; C APV To correct the results; Get the voltage optimization model considering the energy storage life decay: Where: T f To predict the time window size, P, Q, and R are the weight matrices in the objective function that minimize the deviation of the observation node, the change in the active output of the energy storage, and the minimum energy storage life attenuation cost. μ is the voltage reference value, and λ is ini is the penalty coefficient.

8. The distributed energy storage data-driven predictive control method for optimizing the area voltage according to claim 7 is characterized in that: Also includes: Determine the charging and discharging power constraints and SOC constraints of energy storage: SOC min ≤SOC i,t ≤SOC max Where: P max , P min They are the maximum and minimum charging and discharging power of energy storage, respectively. The value of energy storage charging is expressed as positive, and the value of discharging is expressed as negative. It indicates that the absolute value of all elements of the active output vector of the energy storage obtained is the largest, ensuring that the charging and discharging power of the energy storage is within a certain range, SOC min , SOC i,t-1 , SOC i,t , SOC max They represent the minimum state of charge of the energy storage system, the state of charge of the i-th energy storage system at the previous moment t-1, the state of charge of the i-th energy storage system at time t, and the maximum state of charge of the energy storage system, η c,i , η d,j Respectively represent the charging and discharging efficiency of the i-th energy storage system, They represent the charging power and discharging power of the i-th energy storage system at time t, respectively, and E ESS,i represents the capacity of the i-th energy storage system, and Δt is the unit time interval.

Images