CN-121997530-A - Interconnected power grid energy storage planning model construction and application method, system, equipment and medium

Abstract

The invention provides an interconnected power grid energy storage planning model construction and application method, system, equipment and medium, wherein the distribution characteristics of photovoltaic output and load demands can be accurately described by using a Gaussian mixture model algorithm, probability density functions of interconnected power grid voltage and branch power flow can be efficiently obtained by using an accumulation method, and finally, an energy storage double-layer planning model is constructed for energy storage site selection and volume fixing.

Inventors

• CHENG TINGTING
• JIANG XIRUI
• FAN ZHENG
• WU YANAN
• MOU YING
• Guan Dashun
• ZHANG DONGLIANG
• BAI YING

Assignees

• 中国电力科学研究院有限公司
• 国网山东省电力公司经济技术研究院
• 国家电网有限公司

Dates

Publication Date

20260508

Application Date

20251209

Claims (20)

1. 1. The method for constructing the interconnected power grid energy storage planning model is characterized by comprising the following steps of: Processing uncertainty of the photovoltaic output or load demand by adopting a Gaussian mixture model algorithm based on a historical data set of the photovoltaic output or load demand respectively to obtain a probability density function of the photovoltaic output or load demand; Based on the probability density function of the photovoltaic output and the probability density function of the load demand, carrying out probability flow calculation by using an accumulation method to obtain a probability density function of the node voltage of the partitioned power grid and a probability density function of the branch flow; Based on a probability density function of the node voltage of the partitioned power grid and a probability density function of branch power flow, constructing a double-layer energy storage planning model taking photovoltaic and load uncertainty into consideration as an interconnected power grid energy storage planning model; the double-layer energy storage planning model comprises an upper-layer optimizing model and a lower-layer optimizing model, wherein the upper-layer optimizing model receives probability density functions of node voltages of a partition power grid and probability density functions of branch power flows, aims at minimizing comprehensive efficacy coefficients for improving the performance of the power grid by installed energy storage, obtains an optimal installation position of the energy storage, and optimizes rated capacity, rated power and charge-discharge scheduling strategies of the energy storage based on the optimal installation position of the energy storage, and returns a corresponding energy storage charge-discharge power curve to the upper-layer optimizing model.
2. 2. The method for constructing the interconnected power grid energy storage planning model according to claim 1, wherein the processing uncertainty of the photovoltaic output or load demand by adopting a gaussian mixture model algorithm based on the historical data set of the photovoltaic output or load demand respectively to obtain a probability density function of the photovoltaic output or load demand comprises: aiming at photovoltaic output or load demand, a K-means clustering algorithm is used for dividing a historical data set of the photovoltaic output or load demand respectively to obtain a plurality of clusters, and each cluster corresponds to a Gaussian component in a Gaussian mixture model algorithm; respectively calculating the average value and the covariance matrix of the samples corresponding to each Gaussian component as initial values of the average value and the covariance matrix of the Gaussian components, and averagely distributing the weight of each Gaussian component as the initial value of the weight of each Gaussian component; For each Gaussian component, respectively adopting a current weight, a current mean value and a current covariance matrix, and iteratively updating the weight, the mean value and the covariance matrix of the Gaussian component through an expected maximization algorithm until the weight, the mean value and the covariance matrix are converged; calculating a probability density function of photovoltaic output or load demand based on the weight, the mean value and the covariance matrix of each Gaussian component; The number of the clusters is determined by a Bayesian information criterion method.
3. 3. The method for constructing the interconnected power grid energy storage planning model as set forth in claim 2, wherein for each gaussian component, the steps of iteratively updating the weights, the means and the covariance matrix of the gaussian component by a desired maximization algorithm until all the weights, the means and the covariance matrix converge include: For each Gaussian component, calculating posterior probability based on the weight, the mean value and the covariance matrix of the Gaussian component; re-calculating the weight, the mean value and the covariance matrix of each Gaussian component by adopting the posterior probability; Judging whether the maximum likelihood function corresponding to the weight, the mean value and the covariance matrix of each Gaussian component converges or not, if yes, outputting the weight, the mean value and the covariance matrix of each current Gaussian component and ending, otherwise, jumping to the step of calculating the posterior probability based on the weight, the mean value and the covariance matrix of each Gaussian component.
4. 4. The method for constructing an energy storage planning model of an interconnected network as set forth in claim 2, wherein the posterior probability is calculated as follows: in the formula, Representing the posterior probability of the kth gaussian component nth data, Represents the nth data in the historical dataset, k represents the kth gaussian component, Representing the weight of the last iteration of the kth gaussian component, K representing the total number of best gaussian components, Representing data Is used for the distribution of the gaussian distribution of (c), Representing the mean of the last iteration of the kth gaussian component, Representing a covariance matrix of a last iteration of the kth Gaussian component; the calculation formula for recalculating the weights of the gaussian components using posterior probability is as follows: The calculation formula for recalculating the mean value of the gaussian component using posterior probability is as follows: the covariance matrix of the gaussian component is recalculated using posterior probability as follows: in the formula, Representing the weight of the kth gaussian component, Representing the mean value of the kth gaussian component, And (3) representing a covariance matrix of the kth Gaussian component, wherein N is the number of samples in the kth Gaussian component.
5. 5. The method for constructing an interconnected network energy storage planning model as claimed in claim 2, wherein the probability density function of the photovoltaic output or load demand is calculated as follows: in the formula, Is that Is a function of the probability density of (c) in the (c), For photovoltaic output or load demand, K represents the total number of optimum gaussian components, As the weight of the kth gaussian component, Representing data Is used for the distribution of the gaussian distribution of (c), Is the mean value of the kth gaussian component, Is the covariance matrix of the kth gaussian component.
6. 6. The method for constructing the interconnected power grid energy storage planning model according to claim 1, wherein the probability flow calculation is performed by using a cumulative method based on the probability density function of the photovoltaic output and the probability density function of the load demand to obtain the probability density function of the voltage of the partitioned power grid and the probability density function of the branch flow, and the method comprises the following steps: Rewriting a supply-demand balance equation of a partitioned power grid into a matrix form, expanding a power flow equation into a Taylor series at a reference operating point by adopting a Newton-Laportson method, and ignoring higher-order terms to obtain a power flow linearization model; Performing item shifting and arrangement on the power flow linearization model, and carrying the probability density function of the photovoltaic output and the probability density function of the load demand into the power flow linearization model to obtain the accumulated quantities of each step of the voltage of the partitioned power grid and the branch power flow; And based on the accumulated quantities of each stage of the partitioned power grid voltage and the branch power flow, respectively calculating by using a C-Gram-Charlier series expansion method to obtain probability density functions of the node voltage and the branch power flow.
7. 7. The method for constructing an energy storage planning model of an interconnected network as set forth in claim 6, wherein the probability density function of the node voltage or branch current is calculated as follows: in the formula, The probability density function is node voltage or branch power flow, wherein y is node voltage or branch power flow; the expansion coefficient is the number of m-order, M is the order of the series expansion, For the maximum number of the series expansion, Is an m-order Hermite polynomial The formula of (2) is as follows: the m-order series expansion coefficient By solving a system of linear equations Obtaining; in the formula, Is that The rank vector is an intermediate variable calculated from the node voltage or the accumulated amount of each rank of the branch power flow, =0, The calculation formula of the remaining order components of (c) is as follows: Wherein: Is the standard deviation of node voltage or branch current, T is The rank vector of the column, Is the m-2 order component of T, Of the remaining order The formula of (2) is as follows: wherein q is a sum index; the m-q order cumulant is node voltage or branch power flow; the m-order cumulant is node voltage or branch power flow; q-order component of T; Is that A matrix of order symmetry, Middle element The formula of (2) is as follows: Where e 1 and e 2 are the row and column indices of the matrix, s is the sum index, Representing the s-th order component of T; As a three-parameter function, the formula is as follows: Wherein: For the Gamma function, a, b and c are three parameters of the three parameter function, respectively.
8. 8. The method for constructing an energy storage planning model of an interconnected network according to claim 1, wherein the calculation formula of the objective function of the upper layer optimization model is as follows: Wherein: In order to integrate the efficacy coefficient of the product, Net power fluctuation for a partitioned power grid Is used for the treatment of the skin cancer, Voltage out-of-limit probability for a partitioned grid Is used for the treatment of the skin cancer, Network loss for a partitioned power grid Is used for the treatment of the skin cancer, , , Respectively is , , Weight coefficient of (2); the net power fluctuation amount The formula of (2) is as follows: Wherein: in order to sample the time interval of the time, For the maximum value of net power within the sampling time interval, The minimum value of the net power in the sampling time interval is taken as Tz, and the total duration is optimized; voltage out-of-limit probability of the partitioned power grid The formula of (2) is as follows: Wherein: as a function of the voltage probability density of the partitioned grid node j, For partitioning the voltage of the grid nodes, As an upper limit of the voltage to be applied, J is the total number of nodes of the partitioned power grid; network loss of the partitioned power grid The formula of (2) is as follows: in the formula, The network loss is calculated by a probability density function of branch power flow; Coefficient of efficacy The calculation formula of (2) is as follows In the formula, The single power factor for the d-th variable, , Is the actual value of the d-th evaluation index, Is the satisfaction value of the d-th evaluation index, Is the first The intolerable value of the individual evaluation index, As a result of the first constant being a first constant, As a result of the second constant being a second constant, Is a third constant.
9. 9. The method for constructing an energy storage planning model of an interconnected network according to claim 1, wherein the objective function of the lower optimization model is calculated as follows: Wherein: for annual planning and operating costs of the interconnected electrical network, In order to be a fund recovery factor, In order to achieve the discount rate, For the life of the energy storage system, For the total number of installed energy storage, J is the total number of nodes of the partitioned grid, Is the cost per unit capacity of the energy storage, Is the cost of the stored energy per unit power, For the rated capacity of the stored energy z, For the rated power of the energy storage z, W is the voltage penalty weight, For a positive out-of-limit voltage of the partition grid node j at time t, For the negative limit value of the voltage of the partitioned grid node j at time t, The cost factor is managed for the operation of the energy storage system, For the charging power of the stored energy z at time t, The discharge power of the energy storage z at the time t; The constraint conditions of the lower optimization model are as follows: Wherein: For the ratio of the energy storage rated capacity to the energy storage rated power, Active power injected at time t for the partitioned grid node j, The load power of the partitioned grid node j at the time t, For partitioning the photovoltaic output power of the grid node j at the time t, In order to achieve the energy storage and charging efficiency, In order to achieve the energy storage and discharge efficiency, For the SOC of the stored energy z at time t, For the SOC of the stored energy z at time t +1, For the minimum SOC value of the stored energy, For the maximum SOC value of the stored energy, To store the daily start-up SOC of z, As the daily end tolerance of the SOC, The final daily SOC value of the stored energy z.
10. 10. An application method of an interconnected power grid energy storage planning model is characterized by comprising the following steps: Processing uncertainty of the photovoltaic output or load demand by adopting a Gaussian mixture model algorithm based on a historical data set of the photovoltaic output or load demand respectively to obtain a probability density function of the photovoltaic output or load demand; Based on the probability density function of the photovoltaic output and the probability density function of the load demand, carrying out probability flow calculation by using an accumulation method to obtain a probability density function of the node voltage of the partitioned power grid and a probability density function of the branch flow; Inputting the probability density function of the node voltage of the partitioned power grid and the probability density function of the branch power flow into an upper planning model of a double-layer energy storage planning model, and solving to obtain the installation position of energy storage; Transmitting the installation position of the energy storage into a lower planning model of the double-layer energy storage planning model, and solving to obtain the rated capacity, rated power, charge-discharge scheduling strategy and corresponding charge-discharge power curve of the energy storage; And returning the energy storage charging and discharging power curve to the upper planning model, recalculating the comprehensive efficiency coefficient, and iteratively updating the installation position, rated capacity, rated power and charging and discharging scheduling strategy of the energy storage until the comprehensive efficiency coefficient is not reduced any more, so as to obtain the optimal installation position, rated capacity, rated power and charging and discharging scheduling strategy of the energy storage.
11. 11. An interconnected power grid energy storage planning model construction system, which is characterized by comprising: The photovoltaic load probability density module is used for processing uncertainty of photovoltaic output or load demands by adopting a Gaussian mixture model algorithm based on historical data sets of the photovoltaic output or load demands respectively to obtain probability density functions of the photovoltaic output or load demands; the voltage power flow probability density module is used for carrying out probability power flow calculation by using a cumulative method based on the probability density function of the photovoltaic output and the probability density function of the load demand to obtain the probability density function of the node voltage of the partitioned power grid and the probability density function of the branch power flow; the modeling module is used for constructing a double-layer energy storage planning model taking photovoltaic and load uncertainty into consideration as an interconnected power grid energy storage planning model based on the probability density function of the node voltage of the partitioned power grid and the probability density function of the branch power flow; the double-layer energy storage planning model comprises an upper-layer optimizing model and a lower-layer optimizing model, wherein the upper-layer optimizing model receives probability density functions of node voltages of a partition power grid and probability density functions of branch power flows, aims at minimizing comprehensive efficacy coefficients for improving the performance of the power grid by installed energy storage, obtains an optimal installation position of the energy storage, and optimizes rated capacity, rated power and charge-discharge scheduling strategies of the energy storage based on the optimal installation position of the energy storage, and returns a corresponding energy storage charge-discharge power curve to the upper-layer optimizing model.
12. 12. The interconnected grid energy storage planning model construction system of claim 11, wherein the photovoltaic load probability density module is specifically configured to: aiming at photovoltaic output or load demand, a K-means clustering algorithm is used for dividing a historical data set of the photovoltaic output or load demand respectively to obtain a plurality of clusters, and each cluster corresponds to a Gaussian component in a Gaussian mixture model algorithm; respectively calculating the average value and the covariance matrix of the samples corresponding to each Gaussian component as initial values of the average value and the covariance matrix of the Gaussian components, and averagely distributing the weight of each Gaussian component as the initial value of the weight of each Gaussian component; For each Gaussian component, respectively adopting a current weight, a current mean value and a current covariance matrix, and iteratively updating the weight, the mean value and the covariance matrix of the Gaussian component through an expected maximization algorithm until the weight, the mean value and the covariance matrix are converged; calculating a probability density function of photovoltaic output or load demand based on the weight, the mean value and the covariance matrix of each Gaussian component; The number of the clusters is determined by a Bayesian information criterion method.
13. 13. The interconnected grid energy storage planning model construction system of claim 12, wherein the photovoltaic load probability density module iteratively updates the weights, means, and covariance matrices of the gaussian components by a desired maximization algorithm for each gaussian component, respectively, until the weights, means, and covariance matrices all converge, comprising: For each Gaussian component, calculating posterior probability based on the weight, the mean value and the covariance matrix of the Gaussian component; re-calculating the weight, the mean value and the covariance matrix of each Gaussian component by adopting the posterior probability; Judging whether the maximum likelihood function corresponding to the weight, the mean value and the covariance matrix of each Gaussian component converges or not, if yes, outputting the weight, the mean value and the covariance matrix of each current Gaussian component and ending, otherwise, jumping to the step of calculating the posterior probability based on the weight, the mean value and the covariance matrix of each Gaussian component.
14. 14. The interconnected grid energy storage planning model construction system of claim 12, wherein the posterior probability in the photovoltaic load probability density module is calculated as follows: in the formula, Representing the posterior probability of the kth gaussian component nth data, Represents the nth data in the historical dataset, k represents the kth gaussian component, Representing the weight of the last iteration of the kth gaussian component, K representing the total number of best gaussian components, Representing data Is used for the distribution of the gaussian distribution of (c), Representing the mean of the last iteration of the kth gaussian component, Representing a covariance matrix of a last iteration of the kth Gaussian component; the calculation formula for recalculating the weights of the gaussian components using posterior probability is as follows: The calculation formula for recalculating the mean value of the gaussian component using posterior probability is as follows: the covariance matrix of the gaussian component is recalculated using posterior probability as follows: in the formula, Representing the weight of the kth gaussian component, Representing the mean value of the kth gaussian component, And (3) representing a covariance matrix of the kth Gaussian component, wherein N is the number of samples in the kth Gaussian component.
15. 15. The interconnected grid energy storage planning model construction system of claim 12, wherein the probability density function of the photovoltaic output or load demand in the photovoltaic load probability density module is calculated as follows: in the formula, Is that Is a function of the probability density of (c) in the (c), For photovoltaic output or load demand, K represents the total number of optimum gaussian components, As the weight of the kth gaussian component, Representing data Is used for the distribution of the gaussian distribution of (c), Is the mean value of the kth gaussian component, Is the covariance matrix of the kth gaussian component.
16. 16. The interconnected grid energy storage planning model construction system of claim 11, wherein the voltage power flow probability density module is specifically configured to: Rewriting a supply-demand balance equation of a partitioned power grid into a matrix form, expanding a power flow equation into a Taylor series at a reference operating point by adopting a Newton-Laportson method, and ignoring higher-order terms to obtain a power flow linearization model; Performing item shifting and arrangement on the power flow linearization model, and carrying the probability density function of the photovoltaic output and the probability density function of the load demand into the power flow linearization model to obtain the accumulated quantities of each step of the voltage of the partitioned power grid and the branch power flow; And based on the accumulated quantities of each stage of the partitioned power grid voltage and the branch power flow, respectively calculating by using a C-Gram-Charlier series expansion method to obtain probability density functions of the node voltage and the branch power flow.
17. 17. The interconnected network energy storage planning model construction system of claim 16, wherein the probability density function of the node voltage or branch current in the voltage current probability density module is calculated as follows: in the formula, The probability density function is node voltage or branch power flow, wherein y is node voltage or branch power flow; the expansion coefficient is the number of m-order, M is the order of the series expansion, For the maximum number of the series expansion, Is an m-order Hermite polynomial The formula of (2) is as follows: the m-order series expansion coefficient By solving a system of linear equations Obtaining; in the formula, Is that The rank vector is an intermediate variable calculated from the node voltage or the accumulated amount of each rank of the branch power flow, =0, The calculation formula of the remaining order components of (c) is as follows: Wherein: Is the standard deviation of node voltage or branch current, T is The rank vector of the column, Is the m-2 order component of T, Of the remaining order The formula of (2) is as follows: wherein q is a sum index; the m-q order cumulant is node voltage or branch power flow; the m-order cumulant is node voltage or branch power flow; q-order component of T; Is that A matrix of order symmetry, Middle element The formula of (2) is as follows: Where e 1 and e 2 are the row and column indices of the matrix, s is the sum index, Representing the s-th order component of T; As a three-parameter function, the formula is as follows: Wherein: For the Gamma function, a, b and c are three parameters of the three parameter function, respectively.
18. 18. The interconnected grid energy storage planning model as set forth in claim 11, wherein the objective function of the upper layer optimization model in the modeling module is calculated as follows: Wherein: In order to integrate the efficacy coefficient of the product, Net power fluctuation for a partitioned power grid Is used for the treatment of the skin cancer, Voltage out-of-limit probability for a partitioned grid Is used for the treatment of the skin cancer, Network loss for a partitioned power grid Is used for the treatment of the skin cancer, , , Respectively is , , Weight coefficient of (2); the net power fluctuation amount The formula of (2) is as follows: Wherein: in order to sample the time interval of the time, For the maximum value of net power within the sampling time interval, The minimum value of the net power in the sampling time interval is taken as Tz, and the total duration is optimized; voltage out-of-limit probability of the partitioned power grid The formula of (2) is as follows: Wherein: as a function of the voltage probability density of the partitioned grid node j, For partitioning the voltage of the grid nodes, As an upper limit of the voltage to be applied, J is the total number of nodes of the partitioned power grid; network loss of the partitioned power grid The formula of (2) is as follows: in the formula, The network loss is calculated by a probability density function of branch power flow; Coefficient of efficacy The calculation formula of (2) is as follows In the formula, The single power factor for the d-th variable, , Is the actual value of the d-th evaluation index, Is the satisfaction value of the d-th evaluation index, Is the first The intolerable value of the individual evaluation index, As a result of the first constant being a first constant, As a result of the second constant being a second constant, Is a third constant.
19. 19. The interconnected grid energy storage planning model construction system of claim 11, wherein the objective function of the lower optimization model in the modeling module is calculated as follows: Wherein: for annual planning and operating costs of the interconnected electrical network, In order to be a fund recovery factor, In order to achieve the discount rate, For the life of the energy storage system, For the total number of installed energy storage, J is the total number of nodes of the partitioned grid, Is the cost per unit capacity of the energy storage, Is the cost of the stored energy per unit power, For the rated capacity of the stored energy z, For the rated power of the energy storage z, W is the voltage penalty weight, For a positive out-of-limit voltage of the partition grid node j at time t, For the negative limit value of the voltage of the partitioned grid node j at time t, The cost factor is managed for the operation of the energy storage system, For the charging power of the stored energy z at time t, The discharge power of the energy storage z at the time t; The constraint conditions of the lower optimization model are as follows: Wherein: For the ratio of the energy storage rated capacity to the energy storage rated power, Active power injected at time t for the partitioned grid node j, The load power of the partitioned grid node j at the time t, For partitioning the photovoltaic output power of the grid node j at the time t, In order to achieve the energy storage and charging efficiency, In order to achieve the energy storage and discharge efficiency, For the SOC of the stored energy z at time t, For the SOC of the stored energy z at time t +1, For the minimum SOC value of the stored energy, For the maximum SOC value of the stored energy, To store the daily start-up SOC of z, As the daily end tolerance of the SOC, The final daily SOC value of the stored energy z.
20. 20. An interconnected power grid energy storage planning model application system, comprising: The data input module is used for processing uncertainty of the photovoltaic output or load demand by adopting a Gaussian mixture model algorithm based on historical data sets of the photovoltaic output or load demand respectively to obtain a probability density function of the photovoltaic output or load demand; the voltage and power flow calculation module is used for carrying out probability power flow calculation by using a cumulative method based on the probability density function of the photovoltaic output and the probability density function of the load demand to obtain the probability density function of the node voltage of the partitioned power grid and the probability density function of the branch power flow; the position calculation module is used for inputting the probability density function of the node voltage of the partitioned power grid and the probability density function of the branch power flow into an upper layer planning model of the double-layer energy storage planning model, and solving to obtain the installation position of the energy storage; The capacity strategy module is used for transmitting the installation position of the energy storage into a lower planning model of the double-layer energy storage planning model, and solving the installation position to obtain rated capacity, rated power, charge and discharge scheduling strategy and corresponding charge and discharge power curve of the energy storage; And the iteration updating module is used for returning the energy storage charge-discharge power curve to the upper planning model, recalculating the comprehensive efficiency coefficient, and iteratively updating the installation position, the rated capacity, the rated power and the charge-discharge scheduling strategy of the energy storage until the comprehensive efficiency coefficient is not reduced any more, so as to obtain the optimal installation position, the rated capacity, the rated power and the charge-discharge scheduling strategy of the energy storage.

Description

Interconnected power grid energy storage planning model construction and application method, system, equipment and medium Technical Field The invention relates to the technical field of power system planning, in particular to a method, a system, equipment and a medium for constructing and applying an energy storage planning model of an interconnected power grid. Background With the shift of global energy structures to clean and low-carbonization, renewable energy is widely applied due to the environmental protection, and development and utilization of renewable energy are important points of energy strategy of various countries. Therefore, renewable energy sources (such as photovoltaic) are widely connected to the power system, and become an important component of the power system. However, the photovoltaic power generation is difficult to predict accurately, has strong randomness and volatility, and seriously threatens the supply and demand balance of the power system. Meanwhile, after an emerging load such as an electric automobile is connected to a demand side, the fluctuation of the load demand is further increased, and a more serious challenge is presented to a supply and demand balance method of an electric power system. Especially in extreme environments, the net load demand of the power system may rapidly increase or decrease, and the risk of unbalance of supply and demand increases significantly, thereby threatening the economy, reliability and stability of the power system. Traditional stochastic programming methods often rely on parameterized distribution assumptions, such as assuming that the photovoltaic output or load requirements follow normal distribution, and it is difficult to accurately describe the nonlinear and stochastic characteristics of the photovoltaic output or load requirements themselves. The influence of uncertainty of photovoltaic power generation or load demand on supply and demand balance of a power system cannot be effectively solved by the traditional stochastic programming method. On the other hand, the energy storage system can realize peak clipping and valley filling through the energy time shifting characteristic, so that the supply and demand balance of the power system is ensured, however, the wide deployment of the energy storage system is influenced by investment and construction cost. In recent years, research on energy storage planning problems has been advanced, and many scholars and engineering technicians have proposed various methods. However, the prior related technology cannot fully consider the influence of uncertainty such as photovoltaic power generation fluctuation, load fluctuation and the like when energy storage planning is performed, so that the difference between an optimization result and an actual scheduling result occurs, and the effectiveness of supply and demand balance is further influenced. In addition, the prior art does not consider the influence of energy storage operation on the supply and demand balance of the power system, so that the risk of unbalance of the supply and demand of the power system is increased. Disclosure of Invention In order to solve the problems that the influence of uncertainty such as photovoltaic power generation fluctuation, load fluctuation and the like is not fully considered when the energy storage planning is carried out in the prior art, so that the optimization result is different from the actual scheduling result, and the effectiveness of supply-demand balance is influenced, and the influence of energy storage operation on the supply-demand balance of a power system is not considered, so that the unbalance risk of the supply-demand of the power system is increased, the invention provides an interconnected power grid energy storage planning model construction method, which comprises the following steps: Processing uncertainty of the photovoltaic output or load demand by adopting a Gaussian mixture model algorithm based on a historical data set of the photovoltaic output or load demand respectively to obtain a probability density function of the photovoltaic output or load demand; Based on the probability density function of the photovoltaic output and the probability density function of the load demand, carrying out probability flow calculation by using an accumulation method to obtain a probability density function of the node voltage of the partitioned power grid and a probability density function of the branch flow; Based on a probability density function of the node voltage of the partitioned power grid and a probability density function of branch power flow, constructing a double-layer energy storage planning model taking photovoltaic and load uncertainty into consideration as an interconnected power grid energy storage planning model; the double-layer energy storage planning model comprises an upper-layer optimizing model and a lower-layer optimizing model, wherein the upper-layer optimizing model rec