A method for optimizing the configuration of energy storage devices in active distribution networks for improving power supply reliability

Abstract

A method for optimizing the configuration of energy storage devices in an active distribution network for improving power supply reliability belongs to the field of control. The relevant operating data of the distribution network including each node is obtained; a CART regression decision tree is generated to estimate the reliability demand value of each block of the distribution network, and each block is sorted according to the risk of load loss, and the node set D to be connected to the energy storage device is obtained. ess ; Assuming that the corresponding line of the distribution network fails, the restoration plan is obtained through simulation and recorded, and the block with the largest restored load in the island is obtained, and the block access evaluation value is obtained; for D ess The evaluation values ​​of each access block in the node are sorted from small to large, and the access point with the highest evaluation value is selected as the optimal access point. It selects the site for energy storage devices with the goal of maximizing power supply reliability, minimizing energy storage investment costs and operating costs, and maximizing reliability prompts. A two-layer optimization algorithm is used to optimize the system's energy storage site selection and scale. It is suitable for the planning and design of power grid energy storage devices.

Description

Technical Field

[0001] The present invention belongs to the field of power control, and in particular relates to an optimization configuration method for an active distribution network energy storage device. Background Art

[0002] As society pays more attention to the environment, traditional non-renewable energy generation has been replaced by green renewable energy, and a large number of distributed power sources such as photovoltaic and wind power have been connected.

[0003] Due to the intermittent and random characteristics of the landscape, a large number of distributed power sources pose a huge challenge to the safe and stable operation of the power grid.

[0004] Energy Storage System (ESS) is a distributed power system that reduces the discarding and taming of wind power. When ESS is connected to the distribution network, it can effectively improve the safety and reliability of the distribution network operation. The self-healing ability of the distribution network in this area is relatively weak, and the imbalance between supply and demand leads to low power supply reliability and increased operation risks. ESS provides the possibility of isolated power supply after a distribution network failure, making up for the short-term unstable island caused by wind and solar output and load uncertainty.

[0005] How to effectively reduce the investment and operating costs of energy storage units and improve the safety and reliability of distribution network operation is an urgent problem that needs to be solved in energy storage configuration planning.

[0006] Existing literature has studied the sizing and siting of ESS in transmission and distribution networks. In this literature, the main objectives are to reduce the cost of grid operation losses, improve resilience, and enhance voltage and frequency stability.

[0007] For example, in the transmission network, the paper "Robust Energy Storage System Siting Strategy Considering Physical Attacks on Transmission Lines" (K. Lai et al. in Proc. North Amer. Power Symp., 2018, pp. 1–6.) proposed a three-level optimization model to determine the size and site selection of ESS in the transmission system to prevent extreme weather or physical events from causing damage to transmission lines.

[0008] In the paper "Optimizing Storage Siting, Sizing, and Technology Combination in Transmission-Constrained Networks" (S.Wogrin andD.F.Gayme,IEEE Trans.Power Syst.,vol.30,no.6,pp.3304–3313,Nov.2015.), an improved strategy based on ESS is proposed to solve the constraint problem in the transmission network and select the size and location of energy storage capacity according to the severity of the constraint.

[0009] In addition, in the invention patent with the authorization announcement date of August 11, 2020 and the authorization announcement number CN 107508303 B, a "modular energy storage device optimization configuration and control method for microgrids" is disclosed, which includes 1) configuring energy storage power; 2) calculating the capacity of the energy storage battery; 3) selecting the energy storage battery according to the actual energy storage battery selection principle; 4) during the operation of the microgrid system, the charge and discharge state and charge and discharge values ​​of the energy storage system are established according to the output value of each distributed power source and the SOC and SOH of the energy storage battery; 5) when switching between on-grid and off-grid is required, a seamless switching method is adopted, that is, off-grid to on-grid, planned off-grid and unplanned off-grid all achieve seamless switching.

[0010] The invention patent with the authorization announcement date of June 17, 2022 and the authorization announcement number of CN 111969603 B discloses "a micro-energy grid system and its coordinated optimization operation control method". By reasonably configuring the specific equipment contained in the energy supply side and the energy conversion device, the micro-energy grid system not only has a high proportion of new energy access, but also can optimize the configuration of the energy storage device, meet the energy demand of users through the power grid, heat network and gas network, and achieve the basic balance between local energy production and user energy load. The coordinated optimization operation control method is used to carry out rolling optimization of the micro-energy grid system, and the power output value of the controllable energy supply equipment in the micro-energy grid system is continuously adjusted. Under the premise of meeting the real-time balance of supply and demand, the micro-energy grid system is operated economically, thereby achieving the purpose of energy saving. Unify and coordinate the dispatch of energy consumption, energy supply and energy storage, realize the complementarity of multiple energy supplies in the micro-energy grid system, improve the comprehensive utilization efficiency and step-by-step utilization of multiple energy supplies in the micro-energy grid, reduce emissions, and realize local energy consumption.

[0011] However, the above methods or technical solutions do not involve or consider the impact of energy storage configuration nodes on the power supply reliability of the distribution network.

[0012] How to select the installation location of the energy storage device to maximize the power supply reliability of the distribution network is the focus of the technical solution research of the present invention. Summary of the invention

[0013] The technical problem to be solved by the present invention is to provide an optimization configuration method for active distribution network energy storage devices for improving power supply reliability. The method takes energy storage devices as the research object, selects the site with the goal of maximizing power supply reliability, minimizes energy storage investment costs and operating costs, and maximizes reliability prompts, and uses a double-layer optimization algorithm to optimize the system's energy storage site selection and scale.

[0014] The technical solution of the present invention is to provide an active distribution network energy storage device optimization configuration method for improving power supply reliability, including the selection of the energy storage device access position, which is characterized in that the selection of the energy storage device access position includes the following steps:

[0015] Step 1: Obtain relevant data of the distribution network including load fault data of each node, wind turbine and photovoltaic output data, and distribution network topology structure;

[0016] Step 2: Generate a CART regression decision tree, estimate the reliability demand value of each block in the distribution network, and sort each block according to the risk of load loss to obtain the set of nodes to be connected to the energy storage device D ess ;

[0017] Step 3: Assuming that the corresponding line of the distribution network fails, the restoration plan is obtained through simulation and recorded, and the block with the largest restored load in the island is obtained, and the block access evaluation value is obtained;

[0018] Step 4: Sort the evaluation values ​​of each access block in the Dess node from small to large, and select the access point with the highest evaluation value as the optimal access point.

[0019] Specifically, the active distribution network energy storage device optimization configuration method takes the energy storage device as the research object, selects the site with the goal of maximizing power supply reliability, minimizes the energy storage investment cost and operating cost, and maximizes the reliability prompt. A two-layer optimization algorithm is used to optimize the energy storage site selection and scale of the system. The upper layer uses the node reliability requirements to determine the site selection of the energy storage device, and the lower layer uses the particle swarm optimization algorithm to obtain the size and power of the energy storage device at the optimal device site selection.

[0020] Furthermore, the establishment process of the CART regression decision tree is as follows:

[0021] 1) Construct a training sample data set, which contains the fault impact feature vector X and the reliability requirement value Y corresponding to the feature vector, and calculate the mean and variance of Y in the data set;

[0022] 2) Use a heuristic method to divide the feature space into sub-nodes; investigate all values ​​of all features in each set in each partition; select the optimal cutoff point based on the minimum square error criterion of the reliability requirement value Y of each sub-node to segment the data set;

[0023] 3) After finding the optimal dividing point of each child node, the input space is divided into two regions in turn, and then the above division process is repeated for each region until the Y variance of each child node is small enough and meets the stopping condition;

[0024] 4) Generate a CART regression decision tree; the output value of the child node of the regression tree is the average value of Y in the child node data set, which is the predicted value of the regression tree obtained through training.

[0025] The CART regression decision tree is generated using the training samples and CART generation rules. Then, the reliability requirement value is predicted for the event to be predicted whose fault feature vector X is known but the reliability requirement value Y is unknown according to the decision rules.

[0026] Specifically, during the distribution network fault period, the current maximum power supply capacity of the energy storage and the output of photovoltaic and gas turbines during the fault period are used as the maximum recovery capacity of the island. The energy storage is used as the power supply center of the island, and load nodes with as many capacities as possible are searched for in each branch connected to the energy storage until an island with the largest range that can be supported by the output of distributed power sources in the island is formed. The search result is used as the island division result.

[0027] During feeder partition search, if a faulty section is detected, the search in that feeder direction is stopped, and the load nodes in the fault isolation section cannot be included in the island range.

[0028] Furthermore, the charging and discharging model of the energy storage device is:

[0029] ΔP(t)=P PV (t)+P WT (t)-P L (t)

[0030] Where: ΔP(t) represents the energy storage charging and discharging power at time t; P PV (t) represents the photovoltaic output at time t; P WT (t) represents the fan output at time t; P L (t) represents the load size at time t;

[0031] The sequential charging and discharging model of the energy storage device is as follows:

[0032] When the energy storage device is charging, ΔP(t)>0, and the state of charge model at time t+1 is:

[0033] S(t+1)=(1-ηΔt)S(t)+P C Δtη c / Q

[0034] When the energy storage device is discharged, ΔP(t)<0, and the state of charge model at time t is:

[0035] S(t+1)=(1-ηΔt)S(t)-P D Δtη c / η d Q

[0036] Where: S(t) is the state of charge SOC of the energy storage device at time t; η is the energy storage self-discharge rate; P i c and P i dRespectively represent the charging and discharging power of energy storage i; η c and η d It represents the charging and discharging efficiency of energy storage; C is the rated capacity of the energy storage device;

[0037] When the energy storage device supplies power to the load in the island, the corresponding maximum discharge power at a certain charge state is:

[0038] P D，m (t) = min ((S(t) - SOC min )η d / Δt,P D，max )

[0039] Where: SOC min Indicates the minimum state of charge of energy storage; P D，max Indicates the maximum discharge power allowed by energy storage;

[0040] The set of states that define a successful island operation is as follows:

[0041] M T ={(i, j)|ΔP(t) i ≤P D，m (t) j}

[0042] Therefore, the probability that the distribution network enters an island state after a failure at a certain moment and can successfully operate for t hours is:

[0043]

[0044] In the formula, p SOC，i (t) is the probability distribution of the SOC of the energy storage during the fault period t, assuming that the distribution follows a normal distribution.

[0045] Specifically, the active distribution network energy storage device optimization configuration method quantitatively indicates the economic loss caused by the load in a block after power failure through the "block power failure loss amount" indicator, wherein the greater the power failure loss amount in a block, the higher the importance of the load in the block, and when selecting the energy storage access point, it is necessary to give priority to improving the power supply reliability of the load in the block; the block power failure loss amount indicator calculation expression is as follows:

[0046]

[0047] Where: P im Indicates the active power demand of the load in the i-node in a certain block; ω k represents the economic loss caused by the loss of unit power of load k, where k∈[1,3]; T B Indicates the average power outage time after a block failure.

[0048] Furthermore, the failure probability of a block in the distribution network is the sum of the failure probabilities of each component in the block, so the load loss risk of each block is calculated as follows:

[0049]

[0050] Where: K j represents the load loss risk of block j.

[0051] Specifically, when planning and configuring energy storage devices, the economic feasibility of the investment is considered, and the objective function is to maximize the economic benefits of the energy storage device throughout its life cycle:

[0052] max P=F / C

[0053] Where: P is the net profit of the energy storage device; F is the reliability benefit generated by configuring the energy storage device; C is the cost of energy storage configuration.

[0054] Furthermore, the reliability benefit of configuring energy storage is defined as the amount of load loss reduced after a failure occurs in a block of the distribution network before and after configuring energy storage.

[0055]

[0056] F=F La -F Lb

[0057] Where: M is the number of blocks, K i is the number of power outages in block i, P ik is the load value of block i at the time of the kth power outage, ω ik is the average power outage loss cost per unit of electricity during the kth power outage in block i, which is determined by the load type and power outage time. Lb 、F La They respectively represent the economic losses caused by the same failure before and after the energy storage configuration.

[0058] The method for optimizing the configuration of energy storage devices in an active distribution network described in the technical solution of the present invention adopts a double-layer optimization algorithm to solve the problems of the access location and configuration capacity of the energy storage device, wherein the selection problem of the installation location of the upper-layer energy storage device is determined according to the fault loss degree of a node in the distribution network and the network loss sensitivity, while the optimal configuration capacity problem of the lower-layer energy storage is determined by using a particle swarm algorithm.

[0059] Compared with the prior art, the advantages of the present invention are:

[0060] 1. The reliability demand value prediction method based on regression decision tree theory described in the technical solution of the present invention selects the site with the goal of maximizing power supply reliability, minimizes energy storage investment cost and operating cost, and maximizes reliability prompts; after the energy storage equipment is configured, once a failure occurs, part of the load can restore power supply through island operation, shortening the power outage time, greatly reducing system power shortages, and significantly improving system reliability;

[0061] 2. The technical solution of the present invention aims at the self-healing and restoration problem of active distribution network power supply, proposes the optimal site selection and scale determination scheme for the distribution network to improve the power supply reliability after energy storage access, establishes a distribution network fault block model, and estimates the reliability demand value based on the regression decision tree method, thereby improving the power supply reliability of nodes with low reliability in the distribution network and enhancing the self-healing power supply restoration capability;

[0062] 3. The technical solution of the present invention establishes a model of energy storage cost and reliability benefit. Based on the energy storage cost benefit, the impact of energy storage construction on the reliability of the distribution network is studied. A two-layer solution model based on the particle swarm algorithm is established to evaluate and calculate the economic benefits brought by improving the reliability of power supply. The optimal location of the energy storage system in the distribution network and the energy storage scale and power of each location can be determined to meet economic rationality and minimize the total cost of the entire distribution network system. The reliability of power supply is improved to the maximum extent, the power outage loss in the power outage area after the distribution network failure is reduced, and the reliability benefit is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a block diagram of the probability calculation process of continuous island sustainable power supply of the present invention;

[0064] Figure 2 It is a schematic diagram of the topological structure of the distribution network;

[0065] Figure 3 It is a schematic diagram of the modules of the distribution network affected by the fault;

[0066] Figure 4 It is a schematic diagram of the double-layer algorithm flow block of the present invention;

[0067] Figure 5 This is a schematic diagram of the topological structure of the active distribution network test example system;

[0068] Figure 6 is a typical daily power curve diagram of photovoltaic and wind turbine in the embodiment;

[0069] Figure 7 It is a simulation diagram of node reliability requirement value and island operation success rate;

[0070] Figure 8a and Figure 8bIt is a schematic diagram of the impact of energy storage capacity and maximum charge and discharge power on system reliability;

[0071] Fig. 9 It is a schematic diagram of the relationship between configuration cost, energy storage capacity and charging and discharging power;

[0072] Fig.10 This is an example diagram of fault island division in distribution network;

[0073] Fig.11 It is the SOC curve and charging and discharging power curve after energy storage configuration. DETAILED DESCRIPTION

[0074] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0075] The application of ESS can improve the consumption of renewable energy, provide system auxiliary services such as system backup, frequency regulation, peak regulation, and improve the reliability, stability, and safety of the power grid. The technical solution of the present invention takes the energy storage device as the research object, selects the site with the goal of maximizing the reliability of power supply, minimizes the investment cost and operating cost of energy storage, and maximizes the reliability prompt. A two-layer optimization algorithm is used to optimize the energy storage site selection and scale of the system. The upper layer uses the node reliability requirements to determine the site selection of the energy storage device, and the lower layer uses the particle swarm optimization algorithm to obtain the size and power of the energy storage device at the optimal device site selection.

[0076] (1) Island division strategy based on the probability of continuous power supply of energy storage

[0077] 1) Island division and power balancing strategy:

[0078] When the distribution network operates normally, power interaction is carried out with the upper-level power grid through the 10kV bus to ensure the reliability and economy of load operation in the area.

[0079] When a distribution network fails, its internal fault types are mainly divided into internal distributed power supply failure and non-power supply component failure. When a distributed power supply fails, the faulty power supply can be cut off through the control device, while other power supplies can supply power to the load normally.

[0080] When a non-power supply component fails, the fault can be located and isolated through relay protection and distribution automation equipment, and some non-fault loads can be restored. For the de-energized load downstream of the faulty component, if the load can be transferred by closing the tie switch, the tie switch will be operated to restore power supply; if the load cannot be transferred at the end of the line, the downstream load of the fault will be de-listed from the system to form an island, and the power supply will be restored through the distributed power supply in the island.

[0081] Before analyzing the division of islands, we must first solve the strategy of islands to deal with fluctuations in renewable energy output and load demand. When dividing islands, in order to reduce power outage losses, the islands should include loads with the maximum output that can be provided as much as possible. However, due to the intermittent and volatile output of distributed power sources, the power output and load in the island after a fault do not match, causing the load nodes to be in a power outage and power supply state, which will lead to a decrease in the reliability of the load point. Therefore, the technical solution of this patent assumes that if the distributed power source cannot meet the load demand during island operation, the island operation will stop until the power supply is restored after the fault is repaired.

[0082] The island division strategy adopted by the technical solution of this patent is: during the distribution network fault period, the current maximum power supply capacity of the energy storage and the photovoltaic and gas turbine output during the fault period are used as the maximum recovery capacity of the island, and the energy storage is used as the power supply center of the island. The branches connected to the energy storage are searched for load nodes with as much capacity as possible until an island with the largest range that can be supported by the output of distributed power sources in the island is formed, and the search result is used as the island division result.

[0083] During feeder partition search, if a faulty section is detected, the search in that feeder direction is stopped, and the load nodes in the fault isolation section cannot be included in the island range.

[0084] 2) Analysis of island continuous power supply based on timing probability:

[0085] Energy storage devices can effectively smooth out output fluctuations of distributed power sources and improve system stability and reliability.

[0086] When a distribution network failure occurs, the energy storage device, as the main power source of the island system, works together with the DG to supply power to the power-lost load in the island, reducing the system load loss, reducing regional economic losses, and improving the island's operating capacity.

[0087] The charging and discharging power of the energy storage device is mainly determined by the DG output and load power in the island. There are two main charging and discharging strategies: when the total output of DG resources in the island is less than the power of the total power loss load in the island, the energy storage device releases energy to participate in power supply restoration; if the total output of DG resources in the island is greater than the power of the total power loss load in the island, the energy storage device uses the remaining output of distributed power sources to store energy, so as to reduce wind and solar power abandonment and make full use of renewable energy.

[0088] When the island is restored, the energy storage device is an important energy supply point, and its sequential charging and discharging strategy will directly affect the supply of electrical loads. If the sum of the output of wind turbines and photovoltaics is greater than the load at that moment, the energy storage device is charged; if it is less than the load, the energy storage is discharged outward. This operation mode can make full use of wind and solar energy resources. Assuming that the DG resources in the medium-voltage distribution network are mainly photovoltaics and wind turbines, the charging and discharging model of the energy storage device can be established as follows:

[0089] ΔP(t)=P PV (t)+P WT (t)-P L (t) (1)

[0090] Where: ΔP(t) represents the energy storage charging and discharging power at time t; P PV (t) represents the photovoltaic output at time t; P WT (t) represents the fan output at time t; P L (t) represents the load size at time t.

[0091] When operating on an isolated island, the charging and discharging power of the energy storage is affected by many uncertain factors such as the fluctuation of photovoltaic and wind turbine output and the time-varying nature of the load, and can be regarded as a random variable. Therefore, the charge state of the energy storage also has a certain degree of randomness.

[0092] The state of charge in the current period is closely related to the size of the previous period, and has a strong temporal coupling relationship. Therefore, in order to determine whether the current energy storage device can fill the gap between renewable energy output and load, and reliably maintain island operation until the fault is repaired, it is necessary to analyze the temporal change process of the state of charge.

[0093] When the energy storage device is charging and discharging, the state of charge of the battery at time t+1 is determined by the SOC (State Of Charge, also known as the remaining power) at time t, capacity, power, charge and discharge efficiency, and self-discharge rate. The sequential charge and discharge model of the energy storage device is as follows:

[0094] When the energy storage device is charging, ΔP(t)>0, and the state of charge model at time t+1 is:

[0095] S(t+1)=(1-ηΔt)S(t)+P C Δtη c / Q (2)

[0096] When the energy storage device is discharged, ΔP(t)<0, and the state of charge model at time t is:

[0097] S(t+1)=(1-ηΔt)S(t)-P D Δtη c / η d Q (3)

[0098] Where: S(t) is the state of charge SOC of the energy storage device at time t; η is the energy storage self-discharge rate; P i c and P i d Respectively represent the charging and discharging power of energy storage i; η cand η d They respectively represent the charging and discharging efficiency of energy storage; C is the rated capacity of the energy storage device.

[0099] During the island operation period, the charge state of the energy storage is related to the difference between the DG output and the load demand in the island, so the change of the charge state is a time-series change process. When the energy storage supplies power to the load in the island, the corresponding maximum discharge power at a certain charge state is:

[0100] P D，m (t) = min ((S(t) - SOC min )η d / Δt,P D，max ) (4)

[0101] Where: SOC min Indicates the minimum state of charge of energy storage; P D，max Indicates the maximum discharge power allowed by the energy storage.

[0102] When an island is operating, its internal power balance is the key to maintaining the stability of the island. When the maximum power supply corresponding to the charge state of the energy storage in a certain period of time is greater than or equal to the supply and demand gap ΔP(t) within the island, the island can operate sustainably.

[0103] The set of states that define a successful island operation is as follows:

[0104] M T ={(i, j)|ΔP(t) i ≤P D，m (t) j} (5)

[0105] Therefore, the probability that the distribution network enters an island state after a failure at a certain moment and can successfully operate for t hours is

[0106]

[0107] Where: p soC，i (t) is the probability distribution of the SOC of the energy storage during the fault period t, assuming that the distribution follows a normal distribution.

[0108] Assume that the continuous operation time of the island after the fault is T off , then the calculation process of the probability of sustainable power supply of the timing island is as follows: Figure 1 as shown in .

[0109] (2) Energy storage device site selection strategy:

[0110] 1) Fault impact block analysis and reliability requirement modeling:

[0111] There are many components in the distribution network, which requires a lot of work to analyze the failure of a single component. At the same time, the access of distributed power sources to the distribution network increases the complexity of fault analysis. Therefore, before analyzing the reliability of energy storage, the topology of the distribution network is first analyzed to simplify the topology of the network. Figure 2 As shown in the figure, it is a simple active distribution network schematic diagram. Figure 2 It can be seen that the location and type of the switch are closely related to the impact range of the distribution network after the fault, and have an important impact on the reliability of load power supply.

[0112] In the present technical solution, the connecting elements between the elements not affected by the switching action are referred to as blocks, because from the perspective of load power supply, the impact of the upstream switching action on the failure of the elements in the block is the same.

[0113] Through Figure 2 According to the specific analysis, the power supply status of Line 4 and Line 5 is consistent. If Line 2 or Line 3 fails, switches CB1 and S1 work together to isolate the fault section, so Line 2 and Line 3 can form a block. Before the fault is repaired, the power outage area is Line 2, Line 3 and Load Point 1. Line 1 restores power by isolating the fault section, while Line 4, Line 5, Line 6 and Load Points 2 and 3 form an island recovery through DG power supply.

[0114] After box analysis, Figure 2 The distribution network structure diagram shown can be simplified. The simplified specific structure is as follows Figure 3 as shown in .

[0115] Compared with repeatedly analyzing the failure of a single component, analyzing the impact of the failure on the blocking unit can reduce a lot of analysis and improve the efficiency of the failure analysis of the distribution network components. Since the reliability of each module can be calculated by the series reliability model, the failure rate and fault repair time of each module in the distribution network are calculated as follows.

[0116]

[0117]

[0118] Where: r i ,t i Respectively represent the failure probability and repair time of a single component; r j ,t j They respectively represent the failure probability and mean repair time of a block composed of several components.

[0119] When a line, switch element or transformer in the distribution network fails, the circuit breaker closest to the upstream of the faulty element will operate to remove the faulty element, and then the section switch closest to the upstream of the faulty element will be opened to isolate the faulty element, and the power supply to the upstream non-fault area will be restored. Therefore, the affected upstream power outage load will accumulate one power outage, and due to the existence of automation devices in the distribution network, the fault location and isolation time are not considered. Therefore, the power outage time in the upstream area is short and can be ignored. The downstream external load of the island affected by the fault will accumulate one power outage, and the fault power outage time will be accumulated and counted according to the equipment maintenance time. The above are all processing methods for the external load of the island. For the internal load of the island, the power outage number and power outage time are counted according to whether the island can be seamlessly switched and the duration of the island operation. When the power supply in the island cannot meet the needs of all loads in the island, the loads are sorted according to their importance, and some loads are optimized for power supply.

[0120] In order to select the best energy storage installation location to maximize the power supply reliability of the distribution network after the energy storage device is connected, this technical solution defines an indicator called "block power loss amount", which is used to quantitatively indicate the economic loss caused by the load in a block after power failure. The larger the power loss amount in a block, the more important the load in the block is. When selecting the energy storage access point, it is necessary to give priority to improving the power supply reliability of the load in the block. The calculation expression of this indicator is as follows:

[0121]

[0122] Where: P im Indicates the active power demand of the load in the i-node in a certain block; ω k represents the economic loss caused by the loss of unit power of load k, where k∈[1,3]; T B Indicates the average power outage time after a block failure.

[0123] In order to select the access point of the energy storage device and install it at the best location to maximize the improvement of power supply reliability and minimize the total economic loss caused by the failure of the distribution network, the probability of component failure in the distribution network should be analyzed. From the above-mentioned fault block failure analysis, it can be concluded that the failure probability of a block in the distribution network is the sum of the failure probabilities of each component in the block, so the load loss risk of each block is calculated as follows:

[0124]

[0125] Where: K j represents the load loss risk of block j.

[0126] 2) Reliability requirement estimation based on classification and regression tree:

[0127] When a component of the distribution network fails, energy storage is used as the main power source to supply power to the loads on the island to improve power supply reliability. When selecting the energy storage configuration access point, it is necessary to consider various influencing factors before and after the failure to ensure that the reliability benefits obtained through the energy storage configuration are not limited to the case of a single fault segment. Before the failure occurs, the main influencing factors considered by this technical solution are the failure probability of components in the distribution network and their repair time, as well as the power loss in the module. Taking into account the electrical distance from the reliable DG, whether the load can be transferred after the failure, and the maintenance cost, and comprehensively considering the above-mentioned influencing factors for the selection of energy storage access nodes, this technical solution uses a regression decision tree to evaluate the energy storage access point.

[0128] Classification and regression tree (CART) is a decision tree used for classification and prediction. After learning historical data, events can be classified and predicted based on a set of disordered and irregular given features. The CART decision tree uses known multivariate data to construct prediction criteria, and then predicts one variable based on the values ​​of other variables. The CART decision tree consists of two parts: classification and regression. The former can determine the category of an object by measuring the object and using certain classification criteria; the latter is used to predict a certain value of the object. The CART prediction criterion is constructed based on a systematic analysis of a known data set. In the prediction of reliability demand values, the data set consists of many features of the fault event and its corresponding degree of demand. This technical solution uses the influencing factors of the above-mentioned components on reliability before and after the fault as the degree of influence of the reliability demand value, and forms a fault feature vector to predict the reliability demand value. The core issue of establishing a regression decision tree is the determination of the dividing point and the output value.

[0129] The process of establishing a regression decision tree based on CART is as follows:

[0130] 1) Construct a training sample data set, which contains the fault impact feature vector X and the reliability requirement value Y corresponding to the feature vector, and calculate the mean and variance of Y in the data set;

[0131] 2) Use a heuristic method to divide the feature space into sub-nodes; investigate all values ​​of all features in each set in each partition; select the optimal cutoff point based on the minimum square error criterion of the reliability requirement value Y of each sub-node to segment the data set;

[0132] 3) After finding the optimal dividing point of each child node, the input space is divided into two regions in turn, and then the above division process is repeated for each region until the Y variance of each child node is small enough and meets the stopping condition;

[0133] 4) Generate a CART regression decision tree; the output value of the child node of the regression tree is the average value of Y in the child node data set, which is the predicted value of the regression tree obtained through training.

[0134] The CART regression decision tree is generated using the training samples and CART generation rules. Then, the reliability requirement value is predicted for the event to be predicted whose fault feature vector X is known but the reliability requirement value Y is unknown according to the decision rules.

[0135] 3) Energy storage device access location selection process:

[0136] In order to improve the reliability of the distribution network after the energy storage device is connected, the downstream of the component with the highest component failure loss in the distribution network is preferentially connected to the energy storage device; in order to maximize its power supply recovery capability, it is necessary to analyze the reliability requirements of each block of the distribution network and the failure probability of the components in the block, and obtain the set of blocks to be connected to the energy storage device D ess , giving priority to powering blocks with lower reliability and more important loads to reduce fault load losses, thereby evaluating the maximum power supply capacity that can be provided by the access point after configuring the energy storage device.

[0137] The specific process of selecting the energy storage access location is as follows:

[0138] Step 1: Obtain distribution network related data, including load fault data of each node, wind turbine and photovoltaic output data, and distribution network topology structure information;

[0139] Step 2: Generate a CART regression decision tree based on the above method, estimate the reliability demand value of each block of the distribution network, and sort each block according to the risk of load loss to obtain the set of nodes to be connected to the energy storage device D ess ;

[0140] Step 3: Assuming that the corresponding line of the distribution network fails, the restoration plan is obtained through simulation and recorded, and the block with the largest restored load in the island is obtained, and the block access evaluation value is obtained;

[0141] Step 4: D ess The evaluation values ​​of each access block in the node are sorted from small to large, and the access point with the highest evaluation value is selected as the optimal access point.

[0142] (3) Energy storage device optimization configuration model

[0143] 1) Objective function:

[0144] In actual projects, it is common to encounter the practical problem of configuring energy storage to increase the benefits of the built charging station after it has been in operation for a period of time. On the one hand, the charging station has accumulated actual operation data, which provides a basis for optimizing the configuration of energy storage. On the other hand, the rapid decline in battery costs and the increase in investment enthusiasm provide an opportunity for configuring energy storage in charging stations. Since the initial investment in energy storage devices is large and the capital recovery period is long, the economic feasibility of the investment needs to be considered when planning the configuration of energy storage devices. The objective function is to maximize the economic benefits of the energy storage device throughout its life cycle, as shown in formula (11).

[0145] max P=F / C (11)

[0146] Where: P is the net profit of the energy storage device; F is the reliability benefit generated by configuring the energy storage device; C is the cost of energy storage configuration.

[0147] Among them, since this technical solution mainly considers the effect of improving reliability before and after configuring energy storage, this technical solution does not consider the peak-valley electricity price benefits obtained by installing energy storage, and the capacity benefits generated by reducing wind and solar power abandonment, but mainly considers the reliability benefits of improving power supply reliability.

[0148] Considering that configuring energy storage can improve the power supply reliability of the distribution network, this technical solution defines the reliability benefit of configuring energy storage as the amount of load loss reduced after a fault occurs in a block of the distribution network before and after configuring energy storage.

[0149]

[0150] F=F La -F Lb (13)

[0151] Where: M is the number of blocks, K i is the number of power outages in block i, P ik is the load value of block i at the time of the kth power outage, ω ik is the average power outage loss cost per unit of electricity during the kth power outage in block i, which is determined by the load type and power outage time. Lb 、F La They respectively represent the economic losses caused by the same failure before and after the energy storage configuration.

[0152] The cost C of configuring the energy storage device is mainly composed of the initial construction cost C1 and the later operation and maintenance cost C2. The initial construction cost C1 is mainly composed of the installed capacity E of the energy storage device. B，i And the charge and discharge power P B，i Jointly decide. The initial construction cost C1 is expressed as follows:

[0153]

[0154] Where: k E represents the construction cost of energy storage unit capacity; k P is the charge and discharge power cost.

[0155] The later operation and maintenance cost C2 refers to the capital investment incurred during the life cycle of the energy storage device due to equipment failure, aging, etc., and the cost is mainly determined by the annual charge and discharge volume of the energy storage device. This part of the cost changes continuously during the life cycle of the energy storage device, and usually includes the cost of maintenance and commissioning of the energy storage device. The annual operation and maintenance cost of the energy storage device is related to its operation status, and this part of the cost can be expressed as:

[0156]

[0157] Where: C d is the dynamic cost per unit of electricity, and its value fluctuates with the years; s represents the discount rate.

[0158] Constraints:

[0159] The main constraints for optimizing the configuration of energy storage devices are: First, the equipment characteristic constraints, that is, the power constraints on the energy storage device and the distribution network system, and the prevention of equipment damage caused by excessive charging and discharging of the energy storage device. Its state of charge SOC and the upper and lower limits of energy storage charging and discharging power need to meet certain constraints. Second, the constraints are system operation constraints, that is, the power balance constraints that the system must meet at every moment of the operation cycle, and the state of charge SOC of the energy storage device must be consistent at the initial and end moments of the operation cycle. The specific constraints are shown in the following formula:

[0160] 1. System operation constraints:

[0161]

[0162]

[0163] P ij =V i V j (G ij cosθ ij +B ij sinθ ij )-V i 2 G ij (18)

[0164]

[0165] Where: P i , Q i Respectively represent the active and reactive power injected into node i; V i 、V j Respectively represent the voltage amplitude of node i and node j; θ ij is the voltage phase angle difference between node i and node j; G ij , Bij Respectively represent the conductance and susceptance of the line between node i and node j. V imin 、V imax Respectively represent the minimum and maximum voltage amplitudes of node i; θ imin ,θ imax They represent the minimum and maximum values ​​of the voltage phase angle of node i respectively; P ij is the line flow between nodes i and j, is the maximum line transmission capacity between node i and node j.

[0166] 2. Equipment characteristic constraints:

[0167]

[0168] SOC min ≤SOC t ≤SOC max (twenty one)

[0169] E B ≤E max (twenty two)

[0170]

[0171] Where: P C , P D Respectively represent the maximum charging and discharging power of the energy storage device; α and β represent the minimum operating coefficient of the charging and discharging power of the energy storage device. min , SOC max Respectively represent the minimum and maximum discharge capacity allowed by the energy storage device; SOC t E represents the state of charge of the energy storage device at time t, where t∈[1,24]. max The maximum energy storage capacity that can be installed is affected by its investment cost and installation site. max Indicates the maximum number of energy storage devices allowed to be connected to the distribution network.

[0172] 2) Optimization model solution:

[0173] like Figure 4 As shown in , this technical solution adopts a two-layer optimization algorithm to solve the access location and configuration capacity problems of the energy storage device, wherein the selection of the installation location of the upper energy storage device is determined according to the fault loss degree of a node in the distribution network and the network loss sensitivity, while the optimal configuration capacity problem of the lower energy storage is determined by the particle swarm algorithm.

[0174] Example:

[0175] 1) Example description:

[0176] Using system topology such as Figure 5 The 41-node test system shown in the figure is simulated and analyzed, where nodes 8, 9, 24, 26, and 28 are connected to different types of distributed power sources, including MT1 units with independent frequency and voltage regulation capabilities, which together form an active distribution network system. The typical daily output curves of photovoltaic and wind turbines are shown in Figure 6 shown.

[0177] According to the characteristics of the load curve, 0-6 o'clock is the load valley period, and 8-11 o'clock and 17-21 o'clock are the load peak periods. The specific time-of-use electricity price is shown in Table 1.

[0178] Table 1 Grid time-of-use electricity price

[0179]

[0180] Assume that the maximum number of connections allowed for the energy storage system at each node is 1 and the maximum installed capacity is 4 MW·h. The calculation example uses a typical day to represent the entire year.

[0181] Since the selection of sample data will not affect the optimization method of energy storage site selection and size determination, a typical day is selected to represent the whole year to verify the effectiveness of the proposed method.

[0182] 2) Simulation results:

[0183] Through simulation verification of the method described in this technical solution, the relationship between the reliability requirements of each node and the success rate of island operation after configuring energy storage is as follows: Figure 7 shown.

[0184] Among them, the reliability requirement of node 11 in the example test system is the highest and the success rate of island operation is higher, so the energy storage system is connected to node 11.

[0185] After selecting the energy storage access point, in order to study the impact of the energy storage capacity and charging and discharging power of the access point on improving power supply reliability, the simulation results are as follows: Figure 8a and Figure 8b as shown in .

[0186] from Figure 8a and Figure 8b From the curve, we can see that with the increase of energy storage capacity and maximum charge and discharge power, the average power outage time of the system gradually decreases, the average power supply availability of users gradually increases, and the reliability level of the system significantly improves. However, with the gradual increase of capacity and charge and discharge power, the improvement of energy storage reliability level gradually decreases.

[0187] For the test system shown, the reliability of the distribution network power supply is improved by energy storage capacity and charging and discharging power. Fig. 9As shown in the figure, when the energy storage access capacity is 2.5MW·h and the charging and discharging power is 0.8MW, the power grid of the entire system is reliable. The performance improvement is the greatest, the operating cost is the smallest, and the total investment cost is 6.3642 million yuan.

[0188] Assuming that the transformer fails at 10 am, the island partitioning scheme is obtained according to the branch and bound algorithm, such as Fig.10 shown.

[0189] At this time, the restorable area of ​​the distribution network is divided into three islands, where the ESS is the main power source for frequency and voltage regulation. The installed capacity of the island wind turbine unit DG2 is higher than the rated capacity of the photovoltaic unit DG1. Fig.10 It can be seen that in the initial stage of island operation, the peak power reaches 1.52MW. At this time, the recovery capacity of island I accounts for 38.1% of the total power loss load capacity. The main power supply of island II is a micro gas turbine. The two photovoltaic generators DG3 and DG5 on the island support the load on the island, and the recovery capacity accounts for 26.4% of the total power loss load. The load in island III is transmitted through the feeder and powered by the tie line. It has a relatively stable support capacity, and the recovery capacity accounts for 20.6% of the 10-point power loss load capacity.

[0190] When the energy storage configuration scheme proposed in this technical solution is adopted, by comparing the non-selective energy storage configuration method, when the distribution network is operating on the island, the recovery power of the island where the ESS is located is 798.5kW·h higher, as shown in Table 2. This is because after considering the energy storage access method with reliability requirements, the energy storage can support the needs of most loads in the event of a fault. It can be seen from Table 2 that for the non-selective scheme, the recovery rate and reliability benefit cost of the entire network are higher.

[0191] Table 2 Comparison of results of different ESS configuration schemes

[0192]

[0193] The charging and discharging power curve and SOC curve of the above energy storage system are as follows: Fig.11 As shown. It can be seen that the energy storage system is charged during the low period, that is, from 0 am to 7 am, the energy storage SOC gradually increases during this period, and the energy storage system is discharged during the peak period (8-11) and (17-21). During this period, the energy storage SOC gradually decreases. At the same time, it can be seen that in addition to automatic charging during the valley period and automatic discharging during the peak period, the energy storage system can also reasonably automatically charge and discharge during the flat valley period.

[0194] The reliability indexes of the above embodiment are calculated in the following three cases:

[0195] 1) No energy storage is configured;

[0196] 2) The distribution network does not consider power supply reliability when configuring energy storage;

[0197] 3) The reliability of power supply in the distribution network when the energy storage configuration scheme proposed in this patent is adopted.

[0198] The method described in this technical solution is used for calculation and simulation, and the calculation results are shown in Table 3.

[0199] Four reliability indicators were selected: average power supply availability ASAI, average system outage frequency SAIFI, average system outage duration SAIDI, expected power supply ENS, and avoided failure loss ΔF. L .

[0200] Table 3 Impact of energy storage configuration on system reliability

[0201]

[0202] As can be seen from Table 3, after the energy storage equipment is configured, once a failure occurs, part of the load can be restored to power through island operation, which shortens the power outage time, greatly reduces system power shortage, and significantly improves system reliability. After adopting the energy storage configuration scheme proposed in this patent, the system reliability is further improved.

[0203] The simulation example uses a 41-node system as an example for simulation verification. The results of the example show that the patented method can ultimately determine the optimal location of the energy storage system in the distribution network and the energy storage scale and power of each location, meet economic rationality, and minimize the total cost of the entire distribution network system. It also maximizes the power supply reliability, reduces the power outage losses in the power outage area after the distribution network fails, and improves the reliability benefits.

[0204] Energy storage has the characteristics of fast response speed and strong flexibility. It can provide a variety of auxiliary services for the power grid. Absorbing a large amount of renewable energy is an important flexible resource. The technical solution of this patent, aimed at the problem of self-healing and restoration of power supply in active distribution networks, proposes the optimal site selection and scale determination scheme for distribution networks to improve the reliability of power supply after energy storage access, establishes a distribution network fault block model, establishes a two-layer solution model based on particle swarm algorithm, and estimates the reliability demand value based on the regression decision tree method, thereby improving the power supply reliability of nodes with low reliability in the distribution network and enhancing the self-healing power supply restoration capability. On this basis, an energy storage cost and reliability benefit model is established. Based on the cost-effectiveness of energy storage, the impact of energy storage construction on the reliability of the distribution network is studied, and the economic benefits brought by improving the reliability of power supply are evaluated and calculated.

[0205] The present invention can be widely used in the field of planning and designing of active distribution network energy storage devices.

Claims

1. A method for optimizing the configuration of energy storage devices in an active distribution network for improving power supply reliability, comprising selecting a connection location for the energy storage device, wherein the method is characterized in that The selection of the energy storage device access location includes the following steps: Step 1: Obtain relevant data of the distribution network including load fault data of each node, wind turbine and photovoltaic output data, and distribution network topology structure; Step 2: Generate a CART regression decision tree, estimate the reliability demand value of each block in the distribution network, and sort each block according to the risk of load loss to obtain the set of nodes to be connected to the energy storage device D ess ; Step 3: Assuming that the corresponding line of the distribution network fails, the restoration plan is obtained through simulation and recorded, and the block with the largest restored load in the island is obtained, and the block access evaluation value is obtained; Step 4: Sort the evaluation values ​​of each access block in the Dess node from small to large, and select the access point with the highest evaluation value as the optimal access point; Among them, during the distribution network fault period, the current maximum power supply capacity of the energy storage and the output of photovoltaic and gas turbines during the fault period are used as the maximum recovery capacity of the island. The energy storage is used as the power supply center of the island, and load nodes with as many capacities as possible are searched for in each branch connected to the energy storage until an island with the largest range that can be supported by the output of distributed power sources in the island is formed. The search result is used as the island division result; During feeder partition search, if a faulty section is detected, the search in that feeder direction will be stopped, and the load nodes in the fault isolation section cannot be included in the island range; The method for optimizing the configuration of energy storage devices in an active distribution network for improving power supply reliability uses the "block power loss amount" indicator to quantitatively indicate the economic loss caused by a load in a block after a power failure. The greater the power loss amount in a block, the more important the load in the block is. When selecting energy storage access points, it is necessary to give priority to improving the power supply reliability of the load in the block. The block power loss index calculation expression is as follows: Where: P i Indicates the active power demand of the load in the i-node in a certain block; ω k represents the economic loss caused by the loss of unit power of load k, where k∈[1,3]; T B represents the average power outage time after a block failure; The method for optimizing the configuration of energy storage devices in an active distribution network for improving power supply reliability takes energy storage devices as the research object, selects a site with the goal of maximizing power supply reliability, minimizes energy storage investment costs and operating costs, and maximizes reliability prompts; a two-layer optimization algorithm is used to optimize the energy storage site and scale of the system; the upper layer uses node reliability requirements to determine the site selection of the energy storage device, and the lower layer uses a particle swarm optimization algorithm to obtain the size and power of the energy storage device at the optimal device site selection.

2. The method for optimizing the configuration of active distribution network energy storage devices for improving power supply reliability according to claim 1, characterized in that The establishment process of the CART regression decision tree is as follows: 1) Construct a training sample data set, which contains the fault impact feature vector X and the reliability requirement value Y corresponding to the feature vector, and calculate the mean and variance of Y in the data set; 2) Use a heuristic method to divide the feature space into sub-nodes; investigate all values ​​of all features in each set in each partition; select the optimal cutoff point based on the minimum square error criterion of the reliability requirement value Y of each sub-node to segment the data set; 3) After finding the optimal dividing point of each child node, the input space is divided into two regions in turn, and then the above division process is repeated for each region until the Y variance of each child node is small enough and meets the stopping condition; 4) Generate a CART regression decision tree; the output value of the child node of the regression tree is the average value of Y in the child node data set, which is the predicted value of the regression tree obtained through training; The CART regression decision tree is generated using the training samples and CART generation rules. Then, the reliability requirement value is predicted for the event to be predicted whose fault feature vector X is known but the reliability requirement value Y is unknown according to the decision rules.

3. The method for optimizing the configuration of energy storage devices in an active distribution network for improving power supply reliability according to claim 1, characterized in that Assuming that the DG resources in the medium voltage distribution network are mainly photovoltaic and wind turbines, the charging and discharging model of the energy storage device is: ΔP(t)=P PV (t)+P WT (t)-P L (t) Where: ΔP(t) represents the energy storage charging and discharging power at time t; P PV (t) represents the photovoltaic output at time t; P WT (t) represents the fan output at time t; P L (t) represents the load size at time t; The sequential charging and discharging model of the energy storage device is as follows: When the energy storage device is charging, ΔP(t)>0, and the state of charge model at time t+1 is: S(t+1)=(1-ηΔt)S(t)+P C Δtη c / Q When the energy storage device is discharged, ΔP(t)<0, and the state of charge model at time t is: S(t+1)=(1-ηΔt)S(t)-P D Dtη c / or d Q Where: S(t) is the state of charge SOC of the energy storage device at time t; η is the energy storage self-discharge rate; P C and P D Respectively represent the charging and discharging power of i energy storage; η c and η d It represents the charging and discharging efficiency of energy storage; C is the rated capacity of the energy storage device; When the energy storage device supplies power to the load in the island, the corresponding maximum discharge power at a certain charge state is: P D,m (t)=min((S(t)-SOC min )η d / Δt,P D,max ) Where: SOC min Indicates the minimum state of charge of energy storage; P D,max Indicates the maximum discharge power allowed by energy storage; The set of states that define a successful island operation is as follows: M T ={(i,j)|ΔP(t) i ≤P D,m (t) j } Therefore, the probability that the distribution network enters an island state after a failure at a certain moment and can successfully operate for t hours is: In the formula, p SOC,i (t) is the probability distribution of the SOC of the energy storage during the fault period t, assuming that the distribution follows a normal distribution.

4. The method for optimizing the configuration of energy storage devices in an active distribution network for improving power supply reliability according to claim 1, characterized in that In order to select the access point of the energy storage device and install it at the best location to maximize the improvement of power supply reliability and minimize the total economic loss caused by a distribution network failure, the probability of component failure in the distribution network should be analyzed; The failure probability of a block in the distribution network is the sum of the failure probabilities of each component in the block, so the load loss risk of each block is calculated as follows: Where: K j represents the load loss risk of block j.

5. The method for optimizing the configuration of energy storage devices in an active distribution network for improving power supply reliability according to claim 1, characterized in that When planning and configuring energy storage devices, the economic feasibility of the investment is considered, and the objective function is to maximize the economic benefits of the energy storage device throughout its life cycle: maxP=F / C Where: P is the net profit of the energy storage device; F is the reliability benefit generated by configuring the energy storage device; C is the cost of energy storage configuration.

6. The method for optimizing the configuration of energy storage devices in an active distribution network for improving power supply reliability according to claim 5, characterized in that The reliability benefit F generated by configuring energy storage is the amount of load loss reduced after a failure occurs in a block of the distribution network before and after configuring energy storage; F=F La -F Lb Where: M is the number of blocks, K i is the number of power outages in block i, P ik is the load value of block i at the time of the kth power outage, ω ik is the average power outage loss cost per unit of electricity during the kth power outage in block i, which is determined by the load type and power outage time. Lb 、F La They respectively represent the economic losses caused by the same failure before and after the energy storage configuration.

7. The method for optimizing the configuration of energy storage devices in an active distribution network for improving power supply reliability according to claim 1, characterized in that The active distribution network energy storage device optimization configuration method adopts a double-layer optimization algorithm to solve the access location and configuration capacity problems of the energy storage device, wherein the selection problem of the installation location of the upper-layer energy storage device is determined according to the fault loss degree of a node in the distribution network and the network loss sensitivity, while the optimal configuration capacity problem of the lower-layer energy storage is determined by using a particle swarm algorithm.

Images