Adaptive identification method and system for disaster loss of multi-source load of distribution network under strong typhoon

Abstract

This invention proposes an adaptive identification method and system for multi-source load disaster damage in distribution networks under strong typhoons. The method includes: fault judgment of node communication in the distribution network under strong typhoons; if no communication fault exists, multi-source load distribution network topology identification is performed using the collected node electrical quantity information to obtain the fault type and location, and then the node power loss probability is obtained through disaster damage identification correction; if a communication fault exists, multi-dimensional raw meteorological information and node information are obtained, and the wind speed in the multi-dimensional raw meteorological information is corrected considering ground roughness and relative height to obtain a corrected data sequence. The corrected data sequence is input into a deep learning disaster damage prediction model to obtain the post-disaster situation and node component damage situation over a certain period of time, and then the independent power loss probability of a single node of the multi-source load in the distribution network is obtained using distribution network topology identification.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network disaster identification technology, and particularly relates to an adaptive identification method and system for multi-source load disasters in power distribution networks under strong typhoons. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Frequent typhoon disasters and the resulting large-scale power outages pose a significant challenge to the resilience of distribution networks, which are characterized by multiple voltage levels, complex network structures, and diverse equipment types, severely hindering their crucial task of providing electricity to various users. Current research on fault recovery strategies under extreme disasters typically focuses on fault location and isolation, using different optimization algorithms to find the optimal fault recovery strategy. Therefore, by employing disaster loss modeling and deep learning algorithms to perform topology identification and fault location for high-risk areas and ultra-short-term power outages during disasters, it is possible to maximize load recovery in the shortest possible time, reducing casualties and economic losses.

[0004] (1) Chen Ying et al. from the State Key Laboratory of Power Systems, Department of Electrical Engineering, Tsinghua University, proposed a method for predicting the probability of large-scale power outages in distribution networks under extreme weather conditions. This method proposes a modeling approach for distribution network fault probabilities, which includes: a Bayesian network model of distribution network equipment outage events under typhoon disasters, and node outage probability prediction based on the Bayesian network of distribution network equipment outage events. This method overcomes the shortcomings of previous studies in that it is difficult to characterize the dynamic characteristics of meteorology, and fully characterizes the causal dependencies and uncertainties in distribution network outage events using Bayesian networks.

[0005] This method is based solely on data-driven approaches, taking into account the spatiotemporal correlation of post-disaster equipment outage events. It utilizes historical disaster damage records and disaster numerical simulation data to construct a disaster-temporal Bayesian network model, and then rapidly infers the scope and probability of power outages in the distribution network based on the disaster situation.

[0006] (2) Wang Zengping et al. from the State Key Laboratory of New Energy Power Systems, North China Electric Power University, proposed an assessment method for 110kV line tower collapse and line break accidents under typhoon and rainstorm disasters. This method includes: a typhoon and rainstorm model and an analysis of the impact of typhoons and rainstorms on 110kV conductors and towers. Finite element models are established for different transmission tower types of 110kV lines, considering the correlation of wind and rain loads, and the dynamic response of lines and towers under wind and rain loads is analyzed. Based on structural reliability theory, the probability expressions for line breakage and tower collapse are derived, and the mechanism influence of different tower types, uses, wind direction angles, and other factors on tower collapse is studied. Weak links in the power grid are analyzed, and the 110kV line tower collapse and line break accidents under typhoon and rainstorm disasters are assessed.

[0007] This method is based solely on mechanistic analysis and structural reliability theory. It derives the component failure probability expression and analyzes the weak points of the power grid.

[0008] (3) Jianfeng Fu from Delft University of Technology proposed a real-time drone routing strategy for monitoring and inspecting power distribution networks after disaster recovery. This method includes monitoring power distribution networks after disaster recovery and a real-time drone routing strategy for coordination of inspections. Through the proposed real-time drone routing strategy, drones can inspect damage for disaster recovery. Furthermore, transmission lines can be monitored to identify potential hazards, and road infrastructure can be monitored to provide real-time information on traffic conditions, allowing maintenance personnel to choose the optimal route to reach the damaged area.

[0009] This method combines data-driven and mechanism-based approaches, using real-time drone routing strategies to monitor damage and facilitate post-disaster recovery.

[0010] In summary, method (1) only considers data-driven prediction, but the quality of data-driven post-disaster recovery strategies for distribution networks depends on the quality of the prediction data, which can easily lead to low prediction accuracy. Method (2) only relies on mechanistic analysis for prediction, and although it has less dependence on meteorological and electrical data, it is not suitable for distribution networks with multiple voltage levels, complex network structures, and diverse equipment types. Method (3) combines data-driven and mechanistic analysis methods, but such methods are only suitable for robust communication conditions with complete communication networks. Once the communication network is damaged, when the entire system is under weak communication conditions, the monitored electrical quantities cannot characterize the fault, and the relevant faults cannot be transmitted back to the dispatch center in a timely manner through the communication network. Summary of the Invention

[0011] To overcome the shortcomings of the existing technology, this invention provides an adaptive identification method for multi-source load disaster damage in power distribution networks under strong typhoons, which can achieve topology identification and fault component location under various communication integrity levels.

[0012] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0013] Firstly, an adaptive identification method for multi-source load disaster damage in power distribution networks under strong typhoons is disclosed, including:

[0014] Fault diagnosis of node communication in power distribution networks under strong typhoon conditions;

[0015] If there is no communication failure, the collected electrical quantity information of the distribution network nodes is used to identify the topology of the multi-source load distribution network, obtain the fault type and fault location, and then obtain the node power loss probability through disaster loss identification and correction.

[0016] If a communication failure occurs, multidimensional raw meteorological information and node information are obtained. The wind speed in the multidimensional raw meteorological information is corrected by considering ground roughness and relative height to obtain a corrected data sequence. The corrected data sequence is then input into a deep learning disaster loss prediction model to obtain the post-disaster situation and node component damage situation in the future. Finally, the independent power failure probability of a single node of the multi-source load in the distribution network is obtained by using distribution network topology identification.

[0017] Further technical solutions, including multi-source load distribution network topology identification, specifically include:

[0018] Real-time acquisition of voltage, current, and load size at each node in the distribution network;

[0019] Define a distribution network node state matrix S, which is a 1×n matrix. Before a disaster occurs, all n elements are 1, indicating that all n nodes are in a normal power supply state. After a disaster occurs, the values ​​in matrix S corresponding to the load loss nodes are updated to 0 according to the load loss status.

[0020] Each node in the distribution network is numbered, and a node connection matrix is ​​defined; a fault matrix F is defined to locate faults and analyze their causes.

[0021] Further technical solutions include defining a fault matrix F, performing fault location and cause analysis, including scenario one:

[0022] When assigning values ​​to the fault matrix F:

[0023] The power is lost at node k, while nodes k-1 and k+1 are normal, indicating that the transformer at node k is faulty.

[0024] By using the connection relationship matrix to find the parent node number and the child node number of k, the value of the fault matrix F located at the (k, k) coordinate is obtained.

[0025] Further technical solutions include defining a fault matrix F, performing fault location and cause analysis, and also including scenario two:

[0026] Node k loses power, node k-1 is normal, and node k+1 loses power.

[0027] If all the lower-level nodes of the branch where k is located are lost, then a disconnection fault has occurred between k and k-1.

[0028] If the number of normal lower-level nodes is not 0, it means that the transformers at k and k+1 have failed simultaneously.

[0029] Further technical solutions include defining a fault matrix F, performing fault location and cause analysis, and also including scenario three:

[0030] When node k loses power, nodes k-1 and k+1 also lose power.

[0031] If the failure occurs in the parent node of k-1, we need to jump to that node for specific analysis.

[0032] Then, the system iterates through cases one and two until all nodes corresponding to the fault values ​​in the distribution network node state matrix S have been traversed once. Finally, the fault location and fault type are reflected based on the values ​​in matrix F.

[0033] Further technical solutions involve obtaining the node power outage probability through disaster damage identification and correction, specifically including:

[0034] Considering the pole collapse and wire breakage models, let their probabilities be p1 and p2 respectively. Then the independent power loss probability of a single node of a multi-source load in a distribution network is:

[0035] p = 1 - (1 - p1)(1 - p2).

[0036] A further technical solution involves training a deep learning network based on historical meteorological and disaster damage data to obtain a deep learning disaster damage prediction model that takes meteorological data as input and outputs disaster damage data as output.

[0037] A further technical solution considers ground roughness and relative height to correct wind speed in the multidimensional raw meteorological information, specifically as follows:

[0038] Let the reference wind speed measured by the nearby meteorological station be v0, the surface roughness be z0, and the uniform wind measurement height be h meters; the corrected wind speed at the node be v1, and the surface roughness be z1.

[0039] Wind speed change caused by relative height variation: Let the relative height between the node and the reference station be Δh, and the wind speed change caused by topography be Δv. h ;

[0040] Wind speed change caused by changes in ground roughness: Let the wind speed change caused by changes in roughness be Δv. z ;

[0041] Therefore, the wind speed v1 after nodal correction can be expressed by the following formula:

[0042] v1=v0+Δv h +Δv z .

[0043] Secondly, an adaptive identification system for multi-source load disaster damage in power distribution networks under strong typhoons was disclosed, including:

[0044] The communication fault diagnosis module is configured to: diagnose communication faults in the nodes of the power distribution network under strong typhoon conditions;

[0045] The first node power failure probability calculation module is configured to: if there is no communication failure, use the collected node electrical quantity information of the distribution network to perform multi-source load distribution network topology identification, obtain the fault type and fault location, and then obtain the node power failure probability through disaster loss identification and correction.

[0046] The second node power failure probability calculation module is configured to: if a communication failure exists, obtain multi-dimensional raw meteorological information and node information, consider ground roughness and relative height to correct the wind speed in the multi-dimensional raw meteorological information, obtain the corrected data sequence, input the corrected data sequence into the deep learning disaster loss prediction model, obtain the post-disaster situation and node component damage situation in the future for a certain period of time, and then use the distribution network topology identification to obtain the independent power failure probability of a single node of the multi-source load in the distribution network.

[0047] The above one or more technical solutions have the following beneficial effects:

[0048] This invention proposes an adaptive identification and correction method for multi-source load disaster damage in power distribution networks under strong typhoons, taking into account the availability of post-disaster communication. When a typhoon strikes, real-time meteorological data is input, and the model can continuously predict the disaster damage to the power distribution network for a period of time until the typhoon ends. This is of practical significance for guiding the emergency repair of components after the disaster, enabling disaster prevention and mitigation commanders and power grid dispatch departments to make timely decisions and deployments, minimizing economic losses and casualties. Furthermore, this invention remains universally applicable even under conditions of weak communication.

[0049] This invention only requires the distribution network topology information and historical meteorological data to use deep learning to obtain a disaster damage model, predict the multi-source load disaster damage of the distribution network for a period of time until the end of the typhoon, and has a guiding role in improving the speed of distribution network repair.

[0050] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0051] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0052] Figure 1 This is a flowchart of a method according to an embodiment of the present invention. Detailed Implementation

[0053] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0054] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0055] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0056] Example 1

[0057] See appendix Figure 1 As shown in the figure, this embodiment discloses an adaptive identification method for multi-source load disaster damage in power distribution networks under strong typhoons. It can achieve topology identification and fault component location under various communication integrity levels. By analyzing fault propagation and power loss mechanisms, it can guide post-disaster emergency repair work from top to bottom.

[0058] This embodiment first proposes a method for identifying the post-disaster topology of a distribution network under conditions of sound communication.

[0059] Secondly, considering fault identification under conditions of strong typhoons and weak communication, the wind speed at representative meteorological stations is corrected based on ground roughness and relative height. Then, a deep learning algorithm is introduced to obtain the input meteorological information and output a disaster damage prediction model based on real-time meteorological information and historical disaster damage data. This model can identify faulty components under weak communication conditions and guide post-disaster emergency repairs.

[0060] The aforementioned deep learning algorithm, through an offline training process, ultimately achieves online input of real-time meteorological data such as wind speed and rainfall intensity measured by meteorological stations in the area where the distribution network is located, and outputs the probability of power outage at each node of the distribution network.

[0061] Failure mechanism of power distribution network components under strong typhoon:

[0062] a. Tower

[0063] At this time, the gust speed is v;

[0064] air column mass

[0065] m=ρSl

[0066] Air column length

[0067] l = vt

[0068] According to the momentum theorem

[0069] Ft = mv

[0070] Wind resistance on the outside of the pole

[0071] F=ρSv 2

[0072] If we consider the point near the ground at the bottom of the pole as the fulcrum of a lever, then that point will experience a bending moment. Magnitude of the bending moment:

[0073] M = FL

[0074] In the formula, the lever arm L = h1 / 2. Substituting (4) into (5) yields the bending moment on the pole.

[0075] The bending moment of the tower is expressed as:

[0076]

[0077] In the formula: k is the column correction coefficient, k<1, indicating that under the same wind-receiving area, the wind load on the side of the cylinder is smaller than that on the cuboid.

[0078] Determination of the ultimate design bending moment M0 of the pole:

[0079]

[0080] In the formula: a n For reinforcement ratio, R g The tensile design strength of the steel reinforcement is expressed in kgf / cm². 2 ;F g This refers to the cross-sectional area of ​​the reinforcing steel bars, in cm². 2 ;r g R is the radius of the circle containing the reinforcing bar, in cm. w The design strength of concrete for flexural compressive strength, in kgf / cm². 2 ;F s The concrete cross-sectional area is expressed in cm². 2 .

[0081] Formula for pole breakage conditions of 10kV distribution network ring reinforced concrete poles in strong wind conditions

[0082] M>M0

[0083]

[0084] In the formula: P1 is the probability of a 10kV power pole collapsing in a strong typhoon environment.

[0085] b. Route

[0086] Power line wind load calculation model:

[0087] W = 24.4311CSv 2 DLg

[0088] In the formula: W is the wind load on the power line (N); C is the wind load constant (taken as 0.003); v is the gust wind speed (m / s); g is the gravity constant; D is the diameter of the power line (m); L is the length of the power line (m); S is the shape coefficient of the power line, and

[0089]

[0090] In the formula, if v < 2.2352, then S = 1.

[0091] With average wind speed v m Relationship

[0092] v t =1.29v m +2.5928

[0093] Let W0 be the maximum wind load that the power line can withstand, then the maximum wind speed v that the power line can withstand can be calculated. max :

[0094]

[0095] Considering that wind load caused by extreme wind speeds is an extreme climate event, and that the GEV distribution theory has been widely used in climate analysis and climate change research, the generalized extreme value distribution is used to fit the extreme value distribution of wind load data caused by wind speed.

[0096] Probabilistic analysis was performed on the gust wind speeds of power lines to construct the GEV distribution of gust wind speeds.

[0097]

[0098] In the formula: μ is the position parameter, σ is the shape parameter, This is the tail exponent. When ξ = 0, F(v) t ) is a Gumbel distribution; when ξ < 0, F(v) t ) follows a Weibull distribution; when ξ>0, F(v) t ) is a Fréchet distribution.

[0099] The GEV distribution parameters can be estimated using the maximum likelihood method, where the maximum likelihood function is:

[0100]

[0101] In the formula: n is the number of sample data. The estimated values ​​of the GEV parameters are obtained by maximum likelihood estimation. and The probability density function for obtaining the gust wind speed is:

[0102]

[0103] Constructing a probability calculation model for power line outages under storm disasters

[0104]

[0105] In the formula: P2 is the probability of line breakage at time t.

[0106] Methods for identifying the post-disaster topology of distribution networks:

[0107] Assuming that each distribution transformer on a 10kV line is equipped with an automatic electrical quantity monitoring and signal transmission device at its low-voltage side outlet, the voltage, current and load of each node can be fed back in real time, and the values ​​of these electrical quantities can reflect the working status of the node.

[0108] Each node in the distribution network is numbered, and a node connection matrix C is defined.

[0109]

[0110] In the formula: n is the total number of distribution network nodes, and matrix C is an n×n sparse square matrix.

[0111] Define a distribution network node state matrix S, which is a 1×n matrix. Before a disaster, all n elements are 1, indicating that all n nodes are in a normal power supply state. After a disaster, the load power loss status is determined based on the values ​​returned by the information acquisition device, and the values ​​in matrix S corresponding to the load power loss nodes are updated to 0. The following section will focus on fault location and cause analysis. Let the power loss node be numbered k, where k-1 represents its parent node and k+1 represents its child node. Define the fault matrix F:

[0112]

[0113] In the formula: n is the total number of distribution network nodes, and matrix F is an n×n sparse square matrix.

[0114] The fault matrix F is assigned values ​​according to the following three cases:

[0115] Scenario 1: The k-th node loses power, while the k-1 and k+1 nodes are both normal, indicating that the transformer at the k-th node is faulty;

[0116] Use the connection matrix C to find the parent node number of k, let

[0117] j = k

[0118] Then the parent node of k

[0119] k-1 = i, if and only if C ik =1

[0120] Similarly, find the next-level node of k, let

[0121] i = k

[0122] k+1 = i if and only if C kj =1

[0123] At this moment, the value at coordinate (k, k) in F is

[0124] F kk =S k-1 S k+1

[0125] Scenario 2: Node k loses power, node k-1 is normal, and node k+1 loses power. Iterate through all lower-level nodes in the branch containing k. If all nodes are powerless, a break in the connection between k and k-1 has occurred; if the number of normal lower-level nodes is not zero, it indicates that the transformers at k and k+1 have failed simultaneously.

[0126] The method for finding the parent and child node numbers of k is the same as above. When a disconnection fault occurs between k and k-1...

[0127] F k-1,k =1, if and only if N(S) k+1 ,S k+2 ,…,S k+e =0)=e

[0128] Where: N(S) k+1 ,S k+2 ,…,S k+e =0)=e indicates that all lower-level nodes in the branch where k is located are de-energized, where e is the number of lower-level nodes. When the transformers at k and k+1 fail simultaneously...

[0129] F k,k =1, when N(S) k+1 ,S k+2 ,…,S k+e =0)>0

[0130] F k+1,k+1 =1, when N(S) k+1 ,S k+2 ,…,S k+e =0)>0

[0131] Scenario 3: Node k loses power, and nodes k-1 and k+1 also lose power. Since the probability of multiple transformers failing simultaneously is very low, it can be assumed that the fault lies with the upstream node of k-1. Therefore, we need to jump to that node for specific analysis. Let...

[0132] k = k-1

[0133] Then, the execution statuses one and two are checked in a loop until all the nodes corresponding to the fault values ​​(equal to 0) in matrix S can be traversed once. Finally, based on the fault location and fault type reflected by the values ​​in matrix F, guidance can be provided for emergency repairs after the disaster.

[0134] Improve the methods for identifying disaster damage in power distribution networks under communication infrastructure:

[0135] While topology identification based on electrical measurement data under sound communication conditions can locate and determine most faults, some limitations remain. For example, in a unidirectional distribution line, this method can trace the fault back to the originating node, but it cannot determine whether subsequent nodes have also failed. Furthermore, in the absence of a short circuit, a fallen pole will not affect downstream nodes. Therefore, it is necessary to introduce a disaster damage identification and correction method based on meteorological data and component failure mechanisms.

[0136] Considering both pole collapse and wire breakage models, and assuming their probabilities are p1 and p2 respectively, then the independent power loss probability of a single node in a multi-source load distribution network is...

[0137] p = 1 - (1 - p1)(1 - p2)

[0138] Methods for identifying disaster damage in distribution networks under weak communication conditions: The basic idea of ​​disaster damage assessment under weak communication conditions is as follows: When a strong typhoon destroys the communication equipment on the distribution network nodes, the master station cannot accurately obtain the load power loss situation. Within a week before and after the strong typhoon makes landfall, the sliding time window method is used to predict the component damage situation in the next three hours, and the post-disaster component damage prediction data is obtained. Finally, this data is input into the distribution network topology, and the topology identification method is used to obtain the disaster damage prediction result of load power loss.

[0139] In strong wind conditions, it is impossible to measure the actual wind speed of every transformer, tower, and conductor in the distribution area. Furthermore, the wind speed obtained from nearby meteorological stations cannot reflect the actual situation at a particular node. Therefore, it is necessary to correct the actual wind speed at the node.

[0140] Let the reference wind speed measured by the nearby meteorological station be v0, the surface roughness be z0, and the uniform wind measurement height be h = 10 meters; let the corrected wind speed at the node be v1, and the surface roughness be z1.

[0141] Wind speed change caused by relative height variation: Let the relative height between the node and the reference station be Δh, and the wind speed change caused by topography be Δv. h :

[0142] Δv h =Δsv0

[0143]

[0144]

[0145] In the formula: A and B are terrain parameters.

[0146] Wind speed changes caused by variations in ground roughness:

[0147] Let the wind speed change caused by the change in roughness be Δv. z Calculation formula:

[0148]

[0149] Therefore, the wind speed v1 after nodal correction can be expressed by the following formula:

[0150] v1=v0+Δv h +Δv z

[0151] Based on historical meteorological and disaster damage data, a deep learning network is trained to obtain a disaster damage prediction model with input meteorological data and output disaster damage (probability of power loss of loads at each node of the distribution network). The current meteorological data is corrected at the meteorological station, thereby achieving the goal of identifying faulty components under weak communication conditions and obtaining the independent power loss probability p of a single node of the multi-source load in the distribution network, which in turn guides post-disaster emergency repairs.

[0152] The meteorological data here refers to the real-time wind speed and rainfall intensity of the weather station where the distribution network is located. Therefore, under weak communication conditions, a deep learning network is trained offline, and the real-time wind speed and rainfall intensity of the weather station where the distribution network is located are input online to output the probability p of load loss at each node.

[0153] In practical operation: This invention needs to provide basic information about the power distribution network, including its topology, the importance of loads at each node, and power outage losses per unit time; it also needs to provide historical meteorological data, including the highest gust wind speed, etc. The specific work content is as follows:

[0154] Step 1: Obtain and organize comprehensive data on electrical quantities of the power distribution network and meteorological stations in the area. Meteorological data includes multi-dimensional data such as gust wind speed, air pressure, and rainfall intensity; electrical quantities include node voltage and current.

[0155] First, assess the post-disaster communication conditions of the power distribution network. If the communication is intact, proceed to step two. If the communication of the power distribution network itself is inadequate or damaged after the disaster (both situations are referred to as "weak communication conditions"), proceed to step three.

[0156] Step 2: Under the condition of sound communication, use the above-mentioned method for identifying the post-disaster topology of the distribution network to identify the damage status of nodes and output the power-out nodes;

[0157] Specifically, improve communication: determine the load loss status of a node based on the electrical quantities measured at each node; if some nodes do not have electrical quantity measurement terminals, determine the load loss status of a node based on the meteorological data measured at that node and in conjunction with the stress analysis model.

[0158] Step 3: Under weak communication conditions, combining historical meteorological data and topology, a deep learning network is used to predict post-disaster losses and output the power outage nodes;

[0159] Specifically, under weak communication conditions: real-time meteorological data such as wind speed and rainfall intensity from the meteorological station where the distribution network is located are input into the trained deep learning network, and the probability of load loss at each node is output.

[0160] Step 4: Present the final disaster damage situation of the multi-source loads in the form of a risk level early warning diagram to guide post-disaster emergency repairs.

[0161] This invention proposes a failure mechanism model for distribution network components under strong typhoons, a method for identifying the topology of the distribution network after a disaster, a method for identifying the damage caused by multiple load sources in the distribution network under sound communication conditions, and a method for identifying the damage caused by multiple load sources in the distribution network under weak communication conditions. When a typhoon strikes, by inputting real-time meteorological data, the established model (using a mechanism model based on stress analysis (P1, 2 first paragraph formulas) under sound communication conditions, and using a trained deep learning network under weak communication conditions) can predict the damage to the distribution network for a period of time until the end of the typhoon. This has practical significance for guiding the emergency repair of components after a disaster, enabling disaster prevention and mitigation command personnel and power grid dispatch departments to make timely decisions and deployments, and minimize economic losses and casualties.

[0162] This invention can not only identify power distribution network damage caused by typhoons under conditions of good communication, but also correct current meteorological data by using a deep learning model that inputs meteorological data and outputs damage data under conditions of weak communication, thereby identifying faulty components and achieving damage identification.

[0163] Specifically, this invention is a strategy for identifying and restoring power distribution network damage under conditions of inadequate communication during extreme disasters, and it has strong practicality.

[0164] Example 2

[0165] The purpose of this embodiment is to provide a computing device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method.

[0166] Example 3

[0167] The purpose of this embodiment is to provide a computer-readable storage medium.

[0168] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the above method.

[0169] Example 4

[0170] The purpose of this embodiment is to provide an adaptive identification system for multi-source load disaster damage in power distribution networks under strong typhoons, including:

[0171] The communication fault diagnosis module is configured to: diagnose communication faults in the nodes of the power distribution network under strong typhoon conditions;

[0172] The first node power failure probability calculation module is configured to: if there is no communication failure, use the collected node electrical quantity information of the distribution network to perform multi-source load distribution network topology identification, obtain the fault type and fault location, and then obtain the node power failure probability through disaster loss identification and correction.

[0173] The second node power failure probability calculation module is configured to: if a communication failure exists, obtain multi-dimensional raw meteorological information and node information, consider ground roughness and relative height to correct the wind speed in the multi-dimensional raw meteorological information, obtain the corrected data sequence, input the corrected data sequence into the deep learning disaster loss prediction model, obtain the post-disaster situation and node component damage situation in the future for a certain period of time, and then use the distribution network topology identification to obtain the independent power failure probability of a single node of the multi-source load in the distribution network.

[0174] This invention addresses fault identification under conditions of strong typhoons and weak communication. It corrects the wind speed at representative meteorological stations based on ground roughness and relative height, then introduces a deep learning algorithm. Based on real-time meteorological information and historical disaster damage data, it obtains the input meteorological information and outputs a disaster damage prediction model. Under weak communication conditions, it identifies multi-source load fault components in the power distribution network and guides post-disaster emergency repairs.

[0175] The steps and methods involved in the apparatuses of Embodiments 2, 3, and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0176] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0177] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. An adaptive identification method for multi-source load disaster damage in power distribution networks under strong typhoons, characterized by: include: Fault diagnosis of node communication in power distribution networks under strong typhoon conditions; If there is no communication failure, the collected electrical quantity information of the distribution network nodes is used to identify the topology of the multi-source load distribution network, obtain the fault type and fault location, and then obtain the node power loss probability through disaster loss identification and correction. If a communication failure occurs, multidimensional raw meteorological information and node information are obtained. The wind speed in the multidimensional raw meteorological information is corrected by considering ground roughness and relative height to obtain a corrected data sequence. The corrected data sequence is then input into a deep learning disaster loss prediction model to obtain the post-disaster situation and node component damage situation in the future. Then, the independent power failure probability of a single node of the multi-source load in the distribution network is obtained by using the distribution network topology identification. Specifically, the wind speed in the multidimensional raw meteorological information is corrected by considering ground roughness and relative height, as follows: Let the reference wind speed measured by the nearby weather station be... Surface roughness is The wind measurement is performed at a uniform height of h meters; the corrected wind speed at each node. for Surface roughness is ; Wind speed change caused by relative height change: Let the relative height between this node and the reference station be . Wind speed caused by terrain Change to : In the formula: A and B are terrain parameters, and L is the length of the power line. This represents the wind speed relative to the reference speed caused by the change in terrain height Δh. The proportion of wind speed change; Wind speed change caused by changes in ground roughness: Let the wind speed change caused by changes in roughness be... Calculation formula: ,in, It is the height of the atmospheric boundary layer, also known as the top height of the surface roughness layer; Therefore, the wind speed after node correction Expressed by the following formula: 。 2. The adaptive identification method for multi-source load disaster damage in power distribution networks under strong typhoons as described in claim 1, characterized in that, Multi-source load distribution network topology identification, specifically including: Real-time acquisition of voltage, current, and load size at each node in the distribution network; Define a distribution network node state matrix S, which is a 1×n matrix. Before a disaster occurs, all n elements are 1, indicating that all n nodes are in a normal power supply state. After a disaster occurs, the values ​​in matrix S corresponding to the load loss nodes are updated to 0 according to the load loss status. Each node in the distribution network is numbered, and a node connection matrix is ​​defined; a fault matrix F is defined to locate faults and analyze their causes.

3. The adaptive identification method for multi-source load disaster damage in power distribution networks under strong typhoons as described in claim 1, characterized in that, Define a fault matrix F, perform fault location and cause analysis, including scenario one: When assigning values ​​to the fault matrix F: The power outage at node k, while nodes k-1 and k+1 are both normal, indicates a transformer failure at node k. By using the connection matrix, we can find the parent node number and the child node number of k, and then obtain the value of the (k, k)th coordinate in the fault matrix F. Define a fault matrix F, perform fault location and cause analysis, and also include case two: Node k loses power, node k-1 is normal, and node k+1 loses power. Traverse all the subordinate nodes of the branch where k is located. If all of them are de-energized, then a disconnection fault has occurred between k and k-1. If the number of normal subordinate nodes is not 0, then the transformers at k and k+1 have failed simultaneously.

4. The adaptive identification method for multi-source load disaster damage in power distribution networks under strong typhoons as described in claim 3, characterized in that, Define a fault matrix F, perform fault location and cause analysis, and also include case three: When node k loses power, nodes k-1 and k+1 also lose power. Assuming the failure occurs in the parent node of k-1, we need to jump to that node for specific analysis; Then, the system iterates through cases one and two until all fault values ​​in the distribution network node state matrix S have been traversed once. Finally, the fault location and fault type are reflected based on the values ​​in matrix F.

5. The adaptive identification method for multi-source load disaster damage in power distribution networks under strong typhoons as described in claim 1, characterized in that, The probability of node power outage is obtained through disaster damage identification and correction, specifically including: Consider the models of the falling rod and the broken line, and let their probabilities be respectively... and The independent power outage probability of a single node in a multi-source load distribution network is: 。 6. The adaptive identification method for multi-source load disaster damage in power distribution networks under strong typhoons as described in claim 1, characterized in that, Based on historical meteorological and disaster damage data, a deep learning network is trained to obtain a deep learning disaster damage prediction model that takes meteorological data as input and outputs disaster damage data as output.

7. An adaptive identification system for multi-source load disaster damage in power distribution networks under strong typhoons, characterized by: include: The communication fault diagnosis module is configured to: diagnose communication faults in the nodes of the power distribution network under strong typhoon conditions; The first node power failure probability calculation module is configured to: if there is no communication failure, use the collected node electrical quantity information of the distribution network to perform multi-source load distribution network topology identification, obtain the fault type and fault location, and then obtain the node power failure probability through disaster loss identification and correction. The second node power outage probability calculation module is configured to: if a communication failure occurs, obtain multi-dimensional raw meteorological information and node information; correct the wind speed in the multi-dimensional raw meteorological information by considering ground roughness and relative height to obtain a corrected data sequence; input the corrected data sequence into a deep learning disaster damage prediction model to obtain the post-disaster situation and node component damage situation over a certain period of time; and then use distribution network topology identification to obtain the independent power outage probability of a single node of the multi-source load in the distribution network; wherein, the correction of the wind speed in the multi-dimensional raw meteorological information by considering ground roughness and relative height is specifically as follows: Let the reference wind speed measured by the nearby weather station be... Surface roughness is The wind measurement height is h meters; the corrected wind speed at the node is... Surface roughness is ; Wind speed change caused by relative height change: Let the relative height between this node and the reference station be . Wind speed caused by terrain Change to : In the formula: A and B are terrain parameters, and L is the length of the power line. This represents the wind speed relative to the reference speed caused by the change in terrain height Δh. The proportion of wind speed change; Wind speed change caused by changes in ground roughness: Let the wind speed change caused by changes in roughness be... Calculation formula: ,in, It is the height of the atmospheric boundary layer, also known as the top height of the surface roughness layer; Therefore, the wind speed after node correction Expressed by the following formula: 。 8. A computing device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it performs the steps of the method described in any of claims 1-6 above.

Images