Configuration Method of Resilient Distribution Network Energy Storage Devices Based on Global Sensitivity Index

Abstract

A method and system for configuring energy storage devices in resilient distribution networks based on a global sensitivity index is disclosed. This method constructs a probability distribution function of renewable energy unit output covering extreme weather scenarios based on collected basic data of the resilient distribution network and historical output data of renewable energy units, and establishes a probabilistic voltage imbalance calculation model. Then, a surrogate model is constructed using a neural network to generate a global sensitivity index of each renewable energy unit's output to the voltage imbalance of the distribution network. This index serves as the basis for configuring energy storage devices, thereby mitigating the three-phase voltage imbalance of the distribution network. This invention utilizes energy storage devices to smooth out fluctuations in renewable energy output, effectively suppressing voltage imbalance in the distribution network. This method can reduce energy storage configuration costs and improve energy storage configuration effectiveness.

Description

Technical Field

[0001] This invention relates to a technology in the field of resilient distribution network planning, specifically a method for configuring energy storage devices in resilient distribution networks based on global sensitivity indicators. Background Technology

[0002] With the rapid development of renewable energy and distributed generation technologies, large-scale single-phase renewable energy power generation devices are being connected to the distribution network, compounded by the impact of extreme and random weather scenarios, leading to voltage imbalance problems in the distribution network. Due to the inherent three-phase asymmetry in the power system, completely eliminating voltage imbalance in the distribution network is impossible. However, given the harmfulness of voltage imbalance, it is necessary to suppress it to minimize its impact on the grid. Traditional voltage imbalance suppression methods consider voltage imbalance caused by uneven load distribution, suppressing it by evenly distributing single-phase loads across the three-phase lines of the distribution network. For example, static transfer switches can be used to dynamically switch residential loads between different phases, thereby dynamically balancing the three-phase load and suppressing voltage imbalance. In addition, there is the method of balancing loads by connecting reactive power components in parallel. Variable loads, as reactive power components, can be corrected using thyristor-controlled static VAR compensators, but this has the disadvantage of injecting harmonics into the grid. With the large-scale integration of renewable energy into the distribution network, the importance of suppressing voltage imbalance in the distribution network is becoming increasingly prominent, and traditional load balancing methods are clearly insufficient to meet the suppression requirements. Summary of the Invention

[0003] This invention addresses the shortcomings of existing technologies that cannot suppress voltage imbalance based on distribution network proxy models and cannot calculate the sensitivity of voltage imbalance degree to the active power of renewable energy. It proposes a method and system for configuring resilient distribution network energy storage devices based on a global sensitivity index. By using energy storage devices to smooth out the power output fluctuations of renewable energy, the voltage imbalance of the distribution network can be effectively suppressed. This method can reduce the configuration cost of energy storage and improve the configuration effect of energy storage.

[0004] This invention is achieved through the following technical solution:

[0005] This invention relates to a method for configuring energy storage devices in resilient distribution networks based on a global sensitivity index. The method constructs a probability distribution function of renewable energy unit output covering extreme weather scenarios based on collected basic data of the resilient distribution network and historical data of renewable energy unit output, and establishes a probabilistic voltage imbalance calculation model. Then, a surrogate model is constructed using a neural network to generate a global sensitivity index of each renewable energy unit output to the voltage imbalance of the distribution network, serving as the basis for configuring energy storage devices, thereby alleviating the three-phase voltage imbalance of the distribution network.

[0006] The basic data of the resilient distribution network includes: the topology of the distribution network, the operating mode, the parameters of the lines and power equipment, the user load, and other basic operating data.

[0007] The aforementioned historical power output data of renewable energy units includes: power generation data collected during the operation of renewable energy units over a past period, and data on natural factors related to power generation such as wind speed and solar intensity.

[0008] The aforementioned probability distribution function for renewable energy unit output covering extreme weather scenarios is constructed based on historical data of renewable energy unit output. It characterizes the correlation between the outputs of different renewable energy units through a dependency structure, specifically as follows:

[0009] The aforementioned dependency structure refers to the functional relationship that characterizes the correlation between variables, including the linear correlation coefficient matrix, the normal Copula function, the t Copula function, the Gumbel Copula function, the Clayton Copula function, the Frank Copula function, etc.

[0010] The aforementioned probabilistic voltage imbalance calculation model takes the output of renewable energy units as input and the probabilistic voltage imbalance of the distribution network as output, both of which have probabilistic distribution characteristics.

[0011] The neural network, using the input and output of the probabilistic voltage imbalance calculation model for resilient distribution networks as training samples, generates a surrogate model for the probabilistic voltage imbalance calculation model, specifically including:

[0012] Step i: Generate unit output data based on the probability distribution function of renewable energy unit output and the dependent structure relationship, and use this data as input sample to calculate the corresponding output sample of distribution network voltage imbalance using the constructed resilient distribution network probabilistic voltage imbalance calculation model.

[0013] Step ii: Construct a feedforward neural network model with a single hidden layer, and use the input samples and output samples obtained in step i as the training sample set to train the neural network model;

[0014] Step iii: Use the trained neural network model as a proxy model for calculating the probabilistic voltage imbalance of resilient distribution networks. Given input samples, the corresponding distribution network voltage imbalance output samples can be quickly calculated based on the constructed proxy model.

[0015] The aforementioned global sensitivity index is calculated based on variance decomposition theory and simulation method, specifically: when In order to be in The system output function, i.e., the probabilistic voltage imbalance calculation function for resilient distribution networks, has the following input vector: In other words, the output of renewable energy units follows a continuous probability distribution function. Let be a real random variable. , where vector , , y is a subset of x; the complementary subset of y is Then the global sensitivity index is obtained. Where: D is the system output The variance; and It is based on the joint probability distribution function Two distinct random vectors generated independently; and Used to distinguish from the joint probability distribution function The generated random vectors and the conditional probability distribution The generated random vector.

[0016] The global sensitivity index Let y be the global sensitivity index for variable y. When y contains only one variable, the result of solving the global sensitivity index is the first-order sensitivity index of that variable.

[0017] The aforementioned configuration of energy storage devices refers to: arranging renewable energy units in descending order based on a global sensitivity index. The higher the global sensitivity index value, the more important the corresponding renewable energy unit. Therefore, when considering the limited number of energy storage devices to be connected, priority is given to configuring energy storage devices at renewable energy units with higher global sensitivity index values, so as to effectively alleviate the three-phase voltage imbalance of the distribution network.

[0018] This invention relates to a system for implementing the above-mentioned method, comprising: an output probability distribution unit, a sample generation unit, a model training unit, and a configuration generation unit, wherein: the output probability distribution unit calculates the output probability distribution function of renewable energy units based on basic data of resilient distribution networks and historical output data of renewable energy units; the sample generation unit calculates the probabilistic voltage imbalance based on the output probability distribution function of renewable energy units, obtaining a small number of input-output samples for neural network training; the model training unit trains a neural network surrogate model based on the small number of input-output samples, obtaining a surrogate model for quickly obtaining the probabilistic voltage imbalance; and the configuration generation unit calculates the global sensitivity index of distribution network voltage imbalance based on the large-scale input-output samples obtained from the surrogate model, obtaining the configuration basis for energy storage devices in resilient distribution networks.

[0019] Technical effect

[0020] This invention calculates the probabilistic voltage imbalance index of the distribution network and the global sensitivity index of voltage imbalance based on a surrogate model. Based on the calculation results of the global sensitivity index, energy storage devices are configured to suppress voltage imbalance in the distribution network. Compared with existing technologies, this invention significantly improves the calculation efficiency of the probabilistic voltage imbalance index and the global sensitivity index of voltage imbalance. Furthermore, the energy storage configuration based on the global sensitivity index enhances the application effect of the energy storage devices. Attached Figure Description

[0021] Figure 1 This is a flowchart of the present invention;

[0022] Figure 2 This is a circuit diagram for an IEEE-123 system that includes wind turbine (WT) and photovoltaic (PV) units;

[0023] Figure 3 The cumulative probability distribution of voltage unbalance at node 108 of the IEEE-123 system under different conditions;

[0024] Figure 4 A comparison chart showing the voltage imbalance suppression effect of energy storage devices configured based on global sensitivity index calculation results. Detailed Implementation

[0025] like Figure 1 As shown in the figure, this embodiment relates to a method for configuring a resilient distribution network energy storage device based on a global sensitivity index, including:

[0026] Step S1: Obtain basic data of the resilient distribution network and historical data of renewable energy unit output.

[0027] Step S2: Based on historical data of renewable energy unit output, extract the uncertainty of renewable energy unit output, determine the probability model of renewable energy output, and the correlation coefficient matrix between different outputs. Considering wind turbines and photovoltaic generators in the distribution network, their uncertain output is related to the environment in which the units are located. Closely related to the output of renewable energy units are the wind speed or solar intensity of their environment. Wind speed and solar intensity follow Weibull and Beta distributions, respectively. When fitting their probability distributions using historical data, the corresponding probability distribution parameters need to be calculated. The correlation between variables is characterized using a linear correlation coefficient matrix.

[0028] Step S3: Establish a probabilistic voltage imbalance calculation model for resilient distribution networks. This model takes the output of renewable energy units as input and the probabilistic voltage imbalance of the distribution network as output. Specifically, it uses the uncertain output of renewable energy units as input to the system model and calculates the output of the system model, i.e., the voltage imbalance of the distribution network, based on simulation methods. Where: VUF is the voltage unbalance; V2 is the negative sequence voltage; V1 is the positive sequence voltage, which is the line voltage V through the three-phase unbalanced line. ab V bc V ca Or it can be derived from the phase voltage, i.e. , , where: a =1∠120°, a 2 =1∠240°.

[0029] The line voltage of the aforementioned three-phase unbalanced line is obtained through three-phase power flow calculations of the distribution network. For a three-phase voltage unbalanced power system, the active and reactive power equations for each node are as follows: Where: the equation coefficients satisfy the equality conditions , and These are the mutual differentiation and mutual admittance matrices, respectively; and These are the self-derivative and self-integration matrices, respectively; and These represent the real and imaginary parts of the node voltage, respectively; the subscripts i and j represent the number of nodes; a, b, and c represent the three phases of the power system; and the superscripts α and β represent the phases.

[0030] The three-phase line power flow equations in the aforementioned three-phase power flow calculation are as follows: Where: the equation coefficients satisfy the equality conditions .

[0031] In summary, considering the uncertain output of renewable energy units in the distribution network, the objective of probabilistic voltage imbalance calculation is to express the voltage imbalance of the power system as a function of uncertain input variables. Where X is an uncertain input variable, and represents the uncertain output of distributed renewable energy. Due to the uncertainty of X, the voltage imbalance solution also exhibits probabilistic characteristics. The simulation method utilizes a large amount of random sampled data as system input to solve for the probabilistic results of the system output. The sample values ​​of the system input basically reflect the distribution characteristics of the input variables; therefore, the probabilistic voltage imbalance solution based on the simulation method can also well reflect the distribution of the system output.

[0032] Step S4: Based on the input and output samples of the probabilistic voltage imbalance calculation model for resilient distribution networks, a surrogate model for the probabilistic voltage imbalance calculation model for distribution networks is constructed using a neural network model. Specifically, this includes:

[0033] Step S41: Generate unit output data based on the probability distribution function of renewable energy unit output and the dependent structure relationship obtained in step S2, and use this data as input sample to calculate the corresponding output sample of distribution network voltage imbalance using the constructed resilient distribution network probability voltage imbalance calculation model.

[0034] Step S42: Construct a feedforward neural network model with a single hidden layer, and use the input samples and output samples obtained in step S41 as the training sample set to train the neural network model.

[0035] Step S43: Use the trained neural network model as a proxy model for calculating the probabilistic voltage imbalance of resilient distribution networks. Given input samples, the corresponding distribution network voltage imbalance output samples can be quickly calculated based on the constructed proxy model.

[0036] Step S5: Based on the constructed surrogate model, obtain the output sample under the given input sample, and use the obtained output sample to calculate the global sensitivity index of the output of each renewable energy unit to the voltage imbalance of the distribution network, and evaluate the impact of the output fluctuation of renewable energy units on the voltage imbalance of the distribution network.

[0037] Preferably, in step S5, the calculation of the global sensitivity index is based on variance decomposition theory and simulation method. When In order to be in The system output function on the above, with input vector as It follows a continuous probability distribution function. Let be a real random variable. , where vector , , is a subset of x; the complementary subset of y is The global sensitivity index is: ,in: Let be the global sensitivity index for variable y. When y contains only one variable, the result of this formula is the first-order sensitivity index of that variable. This index can be used to quantify the influence of a single variable on the system output response; the larger the value, the greater the influence on the system output response. D represents the system output. The variance; and It is based on the joint probability distribution function Two distinct random vectors generated independently; and Used to distinguish from the joint probability distribution function The generated random vectors and the conditional probability distribution The generated random vector.

[0038] Step S6: Configure energy storage devices based on the global sensitivity index calculation results to alleviate the three-phase voltage imbalance of the distribution network.

[0039] Preferably, in step S6, the energy storage device is configured according to the global sensitivity index calculation result. Specifically, based on the global sensitivity index calculation result of the output of each renewable energy unit obtained in step S5, the renewable energy units are ranked by importance. The larger the global sensitivity index value, the more important the corresponding renewable energy unit. Therefore, when considering the limited number of energy storage devices connected, the energy storage devices are preferentially configured at the renewable energy units with larger global sensitivity index values ​​to effectively alleviate the three-phase voltage imbalance of the distribution network.

[0040] After configuring energy storage devices at renewable energy units, in order to effectively mitigate power output fluctuations of the renewable energy units, this embodiment further implements an optimal mitigation strategy for the energy storage devices. Specifically, considering that the renewable energy units have T time-period power output data sampling during the observation period, an optimization model is constructed for the energy storage devices during this observation period, taking into account both energy storage power configuration and energy storage configuration. That is, during the observation period, considering the energy storage device configured for the i-th renewable energy unit, the energy storage configuration strategy mainly satisfies the requirement of minimizing the variance of the node injection power at time T after configuring the energy storage device.

[0041] The operational constraints of the optimization model include: ,in: To determine the node-injected power of the i-th renewable energy unit at time t after configuring the energy storage device; The average node-injected power of renewable energy units during the observation period; , These represent the minimum and maximum allowable power injected into the node, respectively. The output of the renewable energy unit at time t before the energy storage device is configured; Let be the state of charge of the energy storage device at time t; , These are the charging and discharging efficiencies of the energy storage device, respectively. , These represent the charging and discharging power of the energy storage device at time t, respectively. , These are the minimum and maximum values ​​of the state of charge allowed for the energy storage device, respectively. , These are Boolean values ​​representing the charging and discharging states of the energy storage device at time t; , These are the maximum allowable charging and discharging power of the energy storage device, respectively. This represents the initial state of charge of the energy storage device during the observation period.

[0042] Based on specific practical experiments, the above method was implemented on the following IEEE-123 node system: Five renewable energy units were connected, with wind turbines connected at nodes 32, 66, and 107, and photovoltaic units connected at nodes 73 and 99, as follows. Figure 2 As shown. The three wind turbine units are connected to phases c, c, and b, respectively, and the two photovoltaic units are connected to phases c and b, respectively. The wind speed and solar irradiance data were generated based on the Weibull distribution function and the Beta distribution function, respectively, with reasonable parameter settings, while also considering the correlation between wind speed and solar irradiance variables.

[0043] Set the total sample size to According to the calculation formula, obtaining the global sensitivity index for each uncertainty requires three sets of variable samples: two sets of variable samples independently generated based on the joint probability distribution and another set of variable samples generated based on the conditional probability distribution. Since the two sets of variable samples independently generated based on the joint probability distribution can be used to calculate the global sensitivity index for all uncertainties, calculating the global sensitivity index of the output of five renewable energy units to the system output response actually requires seven sets of variable samples, with each set having a sample size of [missing information]. The voltage imbalance of the distribution network corresponding to all system input samples is calculated using a neural network. Considering the correlation between variables, the sensitivity coefficients of each random input / output variable to the output response are calculated using a formula. The voltage imbalance at node 108 is selected as the output response, and the specific values ​​of the global sensitivity index of each renewable energy unit to the output response are shown in Table 1.

[0044] Table 1. Global sensitivity index of input variables to voltage imbalance at node 108

[0045] The output of the two renewable energy units with larger calculated global sensitivity index values ​​is treated as random variables. This means considering only the uncertainty of the important input variables used in determining the global sensitivity index. The cumulative probability distribution of the voltage imbalance at node 108 is compared with the result when only the uncertainty of the unimportant input variables used in determining the global sensitivity index is considered. Figure 3 As shown. Figure 3 As shown, case 1 considers the combined effects of all 5 input random variables (WT1, WT2, WT3, PV1, and PV2), case 2 considers only the effects of the two input random variables (WT3 and PV1) with larger global sensitivity indices, and case 3 considers only the effects of the two input random variables (WT1 and PV2) with smaller global sensitivity indices.

[0046] A comparison of the results of cases 1, 2, and 3 shows that the impact of the uncertainty of the output of WT3 and PV1 on the voltage imbalance at node 108 of the distribution network is similar to the impact of all five input random variables (WT1, WT2, WT3, PV1, and PV2). This indicates that the global sensitivity index effectively identifies the output of key renewable energy units that affect the voltage imbalance of the distribution network.

[0047] This embodiment further sets up three energy storage configuration scenarios to illustrate the role of the global sensitivity index in guiding the energy storage configuration of resilient distribution networks. The three energy storage configuration scenarios include: Case 4: no energy storage configured; Case 5: energy storage devices configured at WT3 and PV1 (high global sensitivity index); and Case 6: energy storage devices configured at WT1, WT2, and PV2 (low global sensitivity index). The results of the three different energy storage configuration scenarios are as follows: Figure 4 As shown.

[0048] like Figure 4 As shown, in case 5, configuring energy storage devices at renewable energy access nodes (WT3 and PV1) with higher global sensitivity indicators significantly reduced voltage imbalance volatility; while in case 6, configuring energy storage devices at renewable energy access nodes (WT1, WT2, and PV2) with lower global sensitivity indicators resulted in less change in voltage imbalance volatility. This indicates that the global sensitivity indicator guides the configuration of energy storage devices in resilient distribution networks. Configuring energy storage at renewable energy access nodes with higher global sensitivity indicators can significantly reduce voltage imbalance volatility in the distribution network, preventing the voltage imbalance in resilient distribution networks from exceeding the prescribed safety standards due to excessive volatility.

[0049] Compared with existing technologies, this method uses a global sensitivity index to obtain the energy storage device configuration point, which takes into account the impact of global fluctuations in renewable energy. It can use a proxy model to quickly and efficiently calculate the global sensitivity index of each renewable energy unit in the distribution network to the voltage imbalance of the distribution network. Therefore, it prioritizes the configuration of energy storage at the units that have a greater impact on the voltage imbalance of the distribution network, thereby effectively smoothing the output fluctuations of renewable energy, suppressing the voltage imbalance of the distribution network, and giving full play to the configuration role of energy storage devices.

[0050] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.

Claims

1. A method for configuring resilient distribution network energy storage devices based on global sensitivity indices, characterized in that, Based on the collected basic data of resilient distribution networks and historical data of renewable energy unit output, a probability distribution function of renewable energy unit output covering extreme weather scenarios is constructed, and a probability voltage imbalance calculation model is established. Then, a surrogate model is constructed through a neural network to generate a global sensitivity index of the output of each renewable energy unit to the voltage imbalance of the distribution network, which serves as the basis for configuring energy storage devices, thereby alleviating the three-phase voltage imbalance of the distribution network. The basic data of the resilient distribution network includes: the topology of the distribution network, the operating mode, the parameters of the lines and power equipment, and the user load; The aforementioned historical power output data of renewable energy units includes: power generation data collected during the operation of renewable energy units over a past period, and natural factor data related to power generation, such as wind speed and solar intensity. The aforementioned probabilistic voltage imbalance calculation model takes the output of renewable energy units as input and the probabilistic voltage imbalance of the distribution network as output, both of which have probabilistic distribution characteristics.

2. The method for configuring resilient distribution network energy storage devices based on global sensitivity index according to claim 1, characterized in that, The neural network, using the input and output of the probabilistic voltage imbalance calculation model for resilient distribution networks as training samples, generates a surrogate model for the probabilistic voltage imbalance calculation model, specifically including: Step i: Generate unit output data based on the probability distribution function of renewable energy unit output and the dependent structure relationship, and use this data as input sample to calculate the corresponding output sample of distribution network voltage imbalance using the constructed resilient distribution network probabilistic voltage imbalance calculation model. Step ii: Construct a feedforward neural network model with a single hidden layer, and use the input samples and output samples obtained in step i as the training sample set to train the neural network model; Step iii: Use the trained neural network model as a proxy model for calculating the probabilistic voltage imbalance of resilient distribution networks. Given input samples, the corresponding distribution network voltage imbalance output samples can be quickly calculated based on the constructed proxy model.

3. The method for configuring resilient distribution network energy storage devices based on global sensitivity index according to claim 1, characterized in that, The aforementioned global sensitivity index is calculated based on variance decomposition theory and simulation method, specifically: when In order to be in The system output function, i.e., the probabilistic voltage imbalance calculation function for resilient distribution networks, has the following input vector: In other words, the output of renewable energy units follows a continuous probability distribution function. Let be a real random variable; , where vector , , y is a subset of x; the complementary subset of y is Then the global sensitivity index is obtained. Where: D is the system output The variance; and It is based on the joint probability distribution function Two distinct random vectors generated independently; and Used to distinguish from the joint probability distribution function The generated random vectors and the conditional probability distribution The generated random vector.

4. The method for configuring resilient distribution network energy storage devices based on global sensitivity index according to claim 1, characterized in that, The aforementioned configuration of energy storage devices refers to: arranging renewable energy units in descending order based on a global sensitivity index. The higher the global sensitivity index value, the more important the corresponding renewable energy unit. Therefore, when considering the limited number of energy storage devices to be connected, priority is given to configuring energy storage devices at renewable energy units with higher global sensitivity index values, so as to effectively alleviate the three-phase voltage imbalance of the distribution network.

5. The method for configuring resilient distribution network energy storage devices based on global sensitivity indices according to any one of claims 1-4, characterized in that, specifically... include: Step S1: Obtain basic data of the resilient distribution network and historical data of renewable energy unit output; Step S2: Based on historical data of renewable energy unit output, extract the uncertainty of renewable energy unit output, determine the probability model of renewable energy output and the correlation coefficient matrix between different outputs; consider wind turbines and photovoltaic generators in the distribution network, whose uncertain output is related to the environment in which the units are located; closely related to the output of renewable energy units are the wind speed or light intensity of their environment, which follow Weibull and Beta distributions respectively. When fitting their probability distributions through historical data, it is necessary to calculate the corresponding probability distribution parameters; the correlation between variables is characterized by a linear correlation coefficient matrix; Step S3: Establish a probabilistic voltage imbalance calculation model for resilient distribution networks. This model takes the output of renewable energy units as input and the probabilistic voltage imbalance of the distribution network as output. Specifically, it uses the uncertain output of renewable energy units as input to the system model and calculates the output of the system model, i.e., the voltage imbalance of the distribution network, based on simulation methods. Where: VUF is the voltage unbalance; V2 is the negative sequence voltage; V1 is the positive sequence voltage, which is the line voltage V through the three-phase unbalanced line. ab V bc V ca Or it can be derived from the phase voltage, i.e. , , where: a = 1∠120°, a 2 = 1∠240°; Step S4: Based on the input and output samples of the probabilistic voltage imbalance calculation model for resilient distribution networks, a surrogate model for the probabilistic voltage imbalance calculation model for distribution networks is constructed using a neural network model. Specifically, this includes: Step S41: Generate unit output data based on the probability distribution function of renewable energy unit output and the dependent structure relationship obtained in step S2, and use this data as input sample to calculate the corresponding output sample of distribution network voltage imbalance using the constructed resilient distribution network probability voltage imbalance calculation model. Step S42: Construct a feedforward neural network model with a single hidden layer, and use the input samples and output samples obtained in step S41 as the training sample set to train the neural network model; Step S43: Use the trained neural network model as a proxy model for calculating the probabilistic voltage imbalance of resilient distribution networks. Under given input samples, the corresponding distribution network voltage imbalance output samples can be quickly calculated based on the constructed proxy model. Step S5: Based on the constructed surrogate model, obtain the output sample under the given input sample, and use the obtained output sample to calculate the global sensitivity index of the output of each renewable energy unit to the voltage imbalance of the distribution network, and evaluate the impact of the output fluctuation of renewable energy units on the voltage imbalance of the distribution network. Step S6: Configure energy storage devices based on the global sensitivity index calculation results to alleviate the three-phase voltage imbalance of the distribution network.

6. The method for configuring resilient distribution network energy storage devices based on global sensitivity index according to claim 5, characterized in that, The line voltage of the aforementioned three-phase unbalanced line is obtained through three-phase power flow calculations of the distribution network. For a three-phase voltage unbalanced power system, the active and reactive power equations for each node are as follows: Where: the equation coefficients satisfy the condition of the equation. , and These are the mutual differentiation and mutual admittance matrices, respectively; and These are the self-derivative and self-integration matrices, respectively; and These represent the real and imaginary parts of the node voltage, respectively; the subscripts i and j indicate the number of nodes; a, b, and c represent the three phases of the power system; and the superscripts α and β indicate the phases. The three-phase line power flow equations in the aforementioned three-phase power flow calculation are as follows: Where: the equation coefficients satisfy the condition of the equation. .

7. The method for configuring resilient distribution network energy storage devices based on global sensitivity index according to claim 5, characterized in that, In step S5, the calculation of the global sensitivity index is based on variance decomposition theory and simulation method; when In order to be in The system output function on the above, with input vector as It follows a continuous probability distribution function. Let be a real random variable; , where vector , , is a subset of x; the complementary subset of y is Global sensitivity index ,in: Let be the global sensitivity index for variable y. When y contains only one variable, the result of this formula is the first-order sensitivity index of that variable. This index can be used to quantify the influence of a single variable on the system output response; the larger the value, the greater the influence on the system output response. D represents the system output. The variance; and It is based on the joint probability distribution function Two distinct random vectors generated independently; and Used to distinguish from the joint probability distribution function The generated random vectors and the conditional probability distribution The generated random vector.

8. The method for configuring resilient distribution network energy storage devices based on global sensitivity index according to claim 5, characterized in that, In step S6, energy storage devices are configured according to the global sensitivity index calculation results. Specifically, based on the global sensitivity index calculation results of the output of each renewable energy unit obtained in step S5, the renewable energy units are ranked by importance. The larger the global sensitivity index value, the more important the corresponding renewable energy unit. Therefore, when considering the limited number of energy storage devices connected, the energy storage devices are preferentially configured at the renewable energy units with larger global sensitivity index values ​​to effectively alleviate the three-phase voltage imbalance of the distribution network. After configuring energy storage devices at renewable energy units, in order to effectively mitigate power output fluctuations, an optimal mitigation strategy is further developed for the energy storage devices. Specifically, considering that the renewable energy units have T power output data samples during the observation period, an optimization model is constructed for the energy storage devices during this observation period, taking into account both energy storage power configuration and energy storage configuration. That is, during the observation period, considering the energy storage device configured for the i-th renewable energy unit, the energy storage configuration strategy mainly satisfies the requirement of minimizing the variance of the node injection power at time T after configuring the energy storage device.

9. The method for configuring resilient distribution network energy storage devices based on global sensitivity index according to claim 8, characterized in that, The operational constraints of the optimization model include: ,in: To determine the node-injected power of the i-th renewable energy unit at time t after configuring the energy storage device; The average node-injected power of renewable energy units during the observation period; , These represent the minimum and maximum allowable power injected into the node, respectively. The output of the renewable energy unit at time t before the energy storage device is configured; Let be the state of charge of the energy storage device at time t; , These are the charging and discharging efficiencies of the energy storage device, respectively. , These represent the charging and discharging power of the energy storage device at time t, respectively. , These are the minimum and maximum values ​​of the state of charge allowed for the energy storage device, respectively. , These are Boolean values ​​representing the charging and discharging states of the energy storage device at time t; , These are the maximum allowable charging and discharging power of the energy storage device, respectively. This represents the initial state of charge of the energy storage device during the observation period.

10. A system for implementing the configuration method of resilient distribution network energy storage device based on global sensitivity index as described in any one of claims 1-9, characterized in that, include: The system comprises an output probability distribution unit, a sample generation unit, a model training unit, and a configuration generation unit. Specifically: the output probability distribution unit calculates the output probability distribution function of renewable energy units based on basic data of the resilient distribution network and historical output data of renewable energy units; the sample generation unit calculates the probabilistic voltage imbalance based on the output probability distribution function of renewable energy units, obtaining a small number of input-output samples for neural network training; the model training unit trains a surrogate model of the neural network based on the small number of input-output samples, obtaining a surrogate model for quickly obtaining the probabilistic voltage imbalance; and the configuration generation unit calculates the global sensitivity index of the distribution network voltage imbalance based on the large-scale input-output samples obtained from the surrogate model, thus obtaining the configuration basis for energy storage devices in the resilient distribution network.

Images