Power distribution network self-healing strategy generation method and system

Abstract

The invention discloses a power distribution network self-healing strategy generation method and system which are used for solving the technical problem that when an existing power distribution network self-healing strategy generation method faces a fault, simple label matching is carried out only according to a fault type and a line basic structure, and the rapid power restoration quality of a power distribution network is poor. The method comprises the following steps: acquiring surface temperature distribution data of power distribution network equipment and abrupt change high-frequency transient signals of power distribution line nodes; constructing a dynamic topological graph of the power distribution network, and constructing a dynamic three-dimensional matrix according to the dynamic topological graph of the power distribution network, the surface temperature distribution data of the power distribution network equipment and the abrupt change high-frequency transient signals of the power distribution line nodes; generating a plurality of priority self-healing strategies according to a preset basic strategy library, the power distribution network optimization function and the power distribution network constraint conditions; and based on a fuzzy analytic hierarchy process, screening each priority self-healing strategy by adopting a power distribution network optimization function according to the dynamic three-dimensional matrix, and determining a target power distribution network self-healing strategy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power systems, and in particular to a power distribution network self-healing strategy generation method and system. BACKGROUND

[0002] With the continuous expansion of the power distribution network in China, 10kV distribution lines, as a key link in power transmission, have the characteristics of complex network structure, dense branch nodes, and multiple load types. The line operation environment is often disturbed by multiple factors throughout the year: on the climate level, thunderstorms can easily cause line insulation breakdown, and icing can cause conductor overload; on the terrain level, dense urban building groups increase the difficulty of line maintenance, and complex suburban topography can easily cause line tree barrier short circuits. These factors make the fault causes complex and the fault types often hidden (such as weak initial transient signals of single-phase ground fault, which are difficult to capture by conventional monitoring methods).

[0003] To address the above problems and improve power supply reliability, distribution automation switches have been widely deployed in 10kV lines, and the voltage-time coordinated self-healing power supply mode has become the mainstream solution. This mode can achieve preliminary self-healing after a fault occurs by pre-setting the switch action timing logic, which to some extent shortens the manual intervention period. In recent years, however, with the significant increase in load density of the power distribution network (such as the grid connection of distributed new energy stations and the popularization of electric vehicle charging piles), the frequency and complexity of faults have also increased: when the main station self-healing system fails due to communication link interruption, data transmission delay, terminal device failure, etc., the self-healing strategy can only fall back to the "manual dispatching + traditional fault location and isolation mode"; especially in scenarios where the fault is not obvious (such as intermittent ground fault) or multiple lines fail at the same time, dispatchers need to rely solely on their individual skill levels and past experience to determine the fault section, which not only takes a long time to make decisions, but also easily leads to an expanded fault isolation range due to experience bias, and in extreme cases, even causes the entire line to be out of service, seriously affecting normal life and industrial continuous production, and posing a serious challenge to the core needs of the power distribution network for "fast restoration and reliable power supply".

[0004] Existing power distribution network self-healing strategy generation methods only rely on simple label matching based on fault types (such as single-phase ground fault, line break) and line basic structure (such as tree-like radial network, loop network) when facing faults, which can easily lead to "misjudgment" and "omission", resulting in a long time for fault isolation operations, further expanding the power outage impact range, prolonging user outage time, and leading to poor quality of power distribution network fast restoration. SUMMARY

[0005] The application provides a power distribution network self-healing strategy generation method and system, which is used to solve the technical problem that the existing power distribution network self-healing strategy generation method only performs simple label matching according to fault types and line basic structures when facing faults, resulting in poor power distribution network fast power restoration quality.

[0006] The application provides a power distribution network self-healing strategy generation method, which comprises the following steps:

[0007] Obtain power distribution network equipment surface temperature distribution data and sudden high-frequency transient signal of a power distribution line node;

[0008] Construct a power distribution network dynamic topology graph, and construct a dynamic three-dimensional matrix according to the power distribution network dynamic topology graph, the power distribution network equipment surface temperature distribution data and the sudden high-frequency transient signal of the power distribution line node.

[0009] Generate a plurality of priority self-healing strategies according to a preset basic strategy library, a power distribution network optimization function and power distribution network constraint conditions.

[0010] According to the dynamic three-dimensional matrix, each priority self-healing strategy is screened by using the power distribution network optimization function based on a fuzzy analytic hierarchy process, and a target power distribution network self-healing strategy is determined.

[0011] Optionally, the dynamic three-dimensional matrix is constructed according to the power distribution network dynamic topology graph, the power distribution network equipment surface temperature distribution data and the sudden high-frequency transient signal of the power distribution line node.

[0012] Determine a topology matrix, a temperature matrix and a time-frequency matrix based on the power distribution network dynamic topology graph, the power distribution network equipment surface temperature distribution data and the sudden high-frequency transient signal of the power distribution line node.

[0013] Perform affine transformation on the topology nodes of the topology matrix to generate a plurality of two-dimensional plane coordinates.

[0014] Select key nodes from the plurality of two-dimensional plane coordinates.

[0015] Take the key nodes as a reference, nonlinearly map each two-dimensional plane coordinate to an X-Y plane of a three-dimensional space through homogeneous coordinate transformation, and stack the temperature matrix along a Z axis to generate an initial three-dimensional temperature field.

[0016] Generate a sinusoidal signal based on the time-frequency matrix.

[0017] Fuse the initial three-dimensional temperature field and the sinusoidal signal to output a composite three-dimensional matrix, and perform normalization processing on the composite three-dimensional matrix to generate a dynamic three-dimensional matrix.

[0018] Optionally, the determining, based on the power distribution network dynamic topology graph, the power distribution network equipment surface temperature distribution data, and the abrupt high-frequency transient signal of the power distribution line node, a topology matrix, a temperature matrix, and a time-frequency matrix comprises:

[0019] The temperature matrix is generated according to the power distribution network equipment surface temperature distribution data.

[0020] The time-frequency matrix is output by performing matrix transformation on the abrupt high-frequency transient signal of the power distribution line node.

[0021] The topology matrix is generated by extracting topology features from the power distribution network dynamic topology graph and performing graph embedding on the topology features.

[0022] Optionally, the generating a plurality of priority self-healing strategies according to a preset basic strategy library, a power distribution network optimization function, and a power distribution network constraint condition comprises:

[0023] The plurality of initial candidate self-healing strategies are determined based on the preset basic strategy library.

[0024] The plurality of intermediate candidate self-healing strategies are determined by screening the initial candidate self-healing strategies based on the power distribution network constraint condition.

[0025] The objective function values of the intermediate candidate self-healing strategies are output by quantitatively evaluating the intermediate candidate self-healing strategies using the power distribution network optimization function.

[0026] The intermediate candidate self-healing strategies are sorted in ascending order according to the objective function values, and the plurality of priority self-healing strategies are determined.

[0027] Optionally, the determining a target power distribution network self-healing strategy by screening the priority self-healing strategies according to the dynamic three-dimensional matrix using the power distribution network optimization function based on the fuzzy analytic hierarchy process comprises:

[0028] The criterion weight vector is determined by converting the fixed weight of the power distribution network optimization function using the fuzzy analytic hierarchy process.

[0029] The original index values of the priority self-healing strategies are determined based on the dynamic three-dimensional matrix.

[0030] The normalized index values corresponding to the original index values are determined by normalizing the original index values.

[0031] The comprehensive scores of the priority self-healing strategies are determined according to the normalized index values and the criterion weight vector.

[0032] The priority self-healing strategy corresponding to the maximum comprehensive score is selected as the initial power distribution network self-healing strategy.

[0033] verify the initial power distribution network self-healing strategy to determine a target power distribution network self-healing strategy.

[0034] Optionally, the composite three-dimensional matrix is specifically:

[0035]

[0036] wherein, is a composite three-dimensional matrix; , is a weight coefficient; is an initial three-dimensional temperature field; is a sinusoidal signal; is a horizontal coordinate of a power distribution network topology; is a vertical coordinate of a power distribution network topology; is a time / signal evolution dimension.

[0037] The second aspect of the present application provides a power distribution network self-healing strategy generation system, comprising:

[0038] An acquisition module is configured to acquire power distribution network equipment surface temperature distribution data and sudden high-frequency transient signal of a power distribution line node;

[0039] A construction module is configured to construct a power distribution network dynamic topology graph, and construct a dynamic three-dimensional matrix according to the power distribution network dynamic topology graph, the power distribution network equipment surface temperature distribution data and the sudden high-frequency transient signal of the power distribution line node;

[0040] A generation module is configured to generate a plurality of priority self-healing strategies according to a preset basic strategy library, a power distribution network optimization function and a power distribution network constraint condition;

[0041] A determination module is configured to filter each of the priority self-healing strategies according to the dynamic three-dimensional matrix based on a fuzzy analytic hierarchy process and the power distribution network optimization function, and determine a target power distribution network self-healing strategy.

[0042] The third aspect of the present application provides a computer device, comprising a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the power distribution network self-healing strategy generation method according to any one of the above aspects.

[0043] The fourth aspect of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed to implement the steps of the power distribution network self-healing strategy generation method according to any one of the above aspects.

[0044] ​The fifth aspect of the present application provides a computer program product, the computer program product comprises a computer program stored on a non-transitory computer-readable storage medium, the computer program comprises program instructions, wherein when the program instructions are executed by a computer, the computer executes the steps of the power distribution network self-healing strategy generation method according to any one of the above.

[0045] From the above technical solutions, the present application has the following advantages:

[0046] The above technical solutions of the present application provide a power distribution network self-healing strategy generation method, obtain power distribution network equipment surface temperature distribution data and power distribution line node mutation high-frequency transient signal; construct a power distribution network dynamic topology graph, and construct a dynamic three-dimensional matrix according to the power distribution network dynamic topology graph, the power distribution network equipment surface temperature distribution data, and the power distribution line node mutation high-frequency transient signal; generate multiple priority self-healing strategies according to a preset basic strategy library, a power distribution network optimization function, and a power distribution network constraint condition; based on fuzzy analytic hierarchy process, use the power distribution network optimization function to screen each priority self-healing strategy according to the dynamic three-dimensional matrix to determine a target power distribution network self-healing strategy; based on the above scheme, the present application fuses power distribution network topology, equipment temperature, and high-frequency transient multi-class data with the help of a dynamic three-dimensional matrix, can comprehensively depict the spatial position, thermal evolution trend, and electromagnetic characteristics of a fault, at the same time, generates multiple priority strategies based on a basic strategy library and an optimization function, and the screening process combines real-time fault data of the dynamic three-dimensional matrix, can flexibly adapt to complex scenarios such as intermittent grounding and multi-line concurrent fault, can realize rapid fault isolation and accurate load recovery, and effectively improves the power distribution network rapid power restoration quality. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0048] Figure 1 A step flow chart of a power distribution network self-healing strategy generation method provided for the first embodiment of the present application;

[0049] Figure 2 A flowchart of a power distribution network self-healing strategy generation method provided for the first embodiment of the present application;

[0050] Figure 3 A structural block diagram of a power distribution network self-healing strategy generation system provided for the second embodiment of the present application. DETAILED DESCRIPTION

[0051] The embodiment of the present application provides a power distribution network self-healing strategy generation method and system, and aims at solving the technical problem that the existing power distribution network self-healing strategy generation method only performs simple label matching according to fault types and line basic structures when facing faults, and thus leads to poor power distribution network fast power recovery quality.

[0052] In order to make the invention purposes, features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the following described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0053] Please refer to Figure 1 , Figure 1 A step flow chart of a power distribution network self-healing strategy generation method provided by the embodiment one of the present application.

[0054] The power distribution network self-healing strategy generation method provided by the present application comprises the following steps.

[0055] In step 101, power distribution network equipment surface temperature distribution data and mutation high-frequency transient signal of power distribution line nodes are acquired.

[0056] It should be noted that a non-contact infrared temperature measurement sensor (such as a thermistor or a DS18B20 digital temperature sensor) or a contact sensor (such as an ATE400 / ATE200 series wireless temperature measurement sensor) is used to deploy a multi-point sensor on an integrated intelligent switch; a high-frequency acquisition unit (such as a transient recording device) is deployed, and the sampling rate needs to reach MHz (megahertz) level to capture microsecond-level electric wave changes. Through PLC (Power Line Communication, power line carrier communication), sensor network (such as temperature and current probes) to collect equipment state and connection relationship, integrate SCADA (Supervisory Control And Data Acquisition, data acquisition and monitoring control system), GIS (Geographic Information System, geographic information system) system data, and establish a device-line-user association model.

[0057] Further, the surface temperature distribution data of the equipment (surface temperature distribution data of the power distribution network equipment) and the sudden high-frequency transient signal of the power distribution line node (sudden high-frequency transient signal of the power distribution line node) are collected in real time; wherein the surface temperature distribution data of the equipment is collected by a non-contact (such as an infrared temperature sensor) or a contact (such as a wireless temperature sensor) device at multiple monitoring points on the surface of the key equipment (such as an integrated intelligent switch, a transformer, a cable joint, a circuit breaker, etc.) of the power distribution network, and is a data set reflecting the spatial distribution and time variation of the surface temperature of the equipment; the sudden high-frequency transient signal of the power distribution line node refers to the electromagnetic transient signal generated when the current and voltage suddenly change due to the fault (such as single-phase grounding, phase-to-phase short circuit, broken line, etc.) or operation (such as switch opening and closing) of the power distribution network at the power distribution line node (such as a line branch point, an equipment access node, a bus connection point, etc.), and the frequency is kHz to MHz (much higher than the power frequency of 50 / 60 Hz), and the duration is only microseconds to milliseconds.

[0058] Step 102, constructing a dynamic topology diagram of the power distribution network, and constructing a dynamic three-dimensional matrix according to the dynamic topology diagram of the power distribution network, the surface temperature distribution data of the power distribution network equipment, and the sudden high-frequency transient signal of the power distribution line node.

[0059] It should be noted that when constructing the dynamic topology graph of the power distribution network, first, core data is collected through multiple sources: the line direction, tower position, and geographic coordinates and physical parameters of equipment (such as intelligent switches and transformers) are obtained with the aid of GIS (Geographic Information System), the running state (opening and closing signals, on-off state) and line on-off state of the equipment are collected using PLC (Power Line Communication) and sensor networks (current probes, switch state sensors), real-time remote signaling and telemetry data are extracted from SCADA (Supervisory Control And Data Acquisition) at the same time, and the user access information of the marketing system is integrated to determine the "equipment-line-user" association; then, the collected data is preprocessed to eliminate abnormal data such as false positives and missing data, the time series data is completed and converted into standardized formats such as JSON (JavaScript Object Notation) and SVG (Scalable Vector Graphics), and the data is aligned based on timestamps to ensure consistency; then, using a graph theory model, the equipment is abstracted as "topological nodes" with labeled IDs, types, and coordinates, and the connecting lines are abstracted as "topological edges" with labeled parameters, the initial topology structure is drawn based on the GIS (Geographic Information System) line relationship and the SCADA (Supervisory Control And Data Acquisition) initial state, and a "node-edge" adjacency matrix is generated; then, dynamic updating is realized through "event triggering (such as adjusting the topology elements immediately when the switch opens and closes, fault isolation) + periodic verification (collecting data at 100 ms / second to correct deviations)", ensuring that the topology is synchronized with the actual power grid state; finally, the accuracy is cross-verified with current and voltage data and equipment temperature data, the equipment state (normal / fault), line on-off, and load distribution are intuitively marked on the graph, and finally the dynamic topology graph of the power distribution network is obtained, which can reflect the connection relationship and running state of the power distribution network in real time and support historical backtracking.

[0060] Further, step 102 can include the following sub-steps S21-S26:

[0061] Step S21, based on the dynamic topology graph of the power distribution network, the surface temperature distribution data of the power distribution network equipment, and the sudden high-frequency transient signal of the power distribution line node, determine the topology matrix, the temperature matrix, and the time-frequency matrix.

[0062] Specifically, step S21 can include the following sub-steps S211-S213:

[0063] Step S211, generating a temperature matrix according to the power distribution network equipment surface temperature distribution data;

[0064] Step S212, performing matrix transformation on the sudden high-frequency transient signal of the power distribution line node, and outputting a time-frequency matrix;

[0065] Step S213, extracting topological features in the power distribution network dynamic topology graph, and performing graph embedding on the topological features to generate a topological matrix.

[0066] It should be noted that when the temperature matrix is obtained by processing the power distribution network equipment surface temperature distribution data, the collected power distribution network equipment surface temperature distribution original data, i.e., the power distribution network equipment surface temperature distribution data (collected by non-contact infrared temperature measurement sensors, contactless wireless temperature measurement sensors, etc. deployed on integrated intelligent switches and other equipment, containing real-time temperature values of each monitoring point, collection time stamps) and equipment topological position information (from the X-Y plane coordinates corresponding to the equipment in the power distribution network dynamic topology graph) are used as basic data. Then, the original temperature data is preprocessed, the threshold is used to judge and eliminate the abnormal high / low temperature values reported by the sensor, the linear interpolation or Kriging interpolation method is used to complete the missing data caused by communication interruption, and the data is grouped according to the fixed time step (such as 1 minute / step) according to the collection time stamp, to ensure the time sequence consistency of the data. Then, the X-Y plane coordinates of the equipment topological position are used as the matrix row and column indexes, and the temperature values of each equipment monitoring point in the same time step after preprocessing are one-to-one mapped to the corresponding coordinate positions. If there is no direct monitoring point at a certain coordinate, it is filled by neighborhood temperature weighted calculation. Finally, after data regularization (such as unifying the temperature unit to Celsius, converting the data format to matrix form), the temperature matrix is obtained, with the row corresponding to the X-axis coordinate, the column corresponding to the Y-axis coordinate, and the element being the temperature value at the corresponding position (if containing time dimension, it is a three-dimensional matrix, and the third dimension corresponds to the time step).

[0067] Further, the sudden high-frequency transient signal is processed by S transform (Stockwell Transform, adaptive time-frequency analysis method) or HHT transform (Hilbert-Huang Transform, Hilbert-Huang Transform) to obtain a time-frequency matrix. The topological features are processed by graph embedding to obtain a topological matrix. Among them, the graph structure is converted into a low-dimensional vector representation (such as GraphSAGE, GCN), the dimensions of the topological features are unified, the graph structure is simplified by topological sorting, the key nodes and edges are identified, and the data redundancy is reduced.

[0068] Step S22, performing affine transformation on the topological nodes of the topological matrix to generate a plurality of two-dimensional plane coordinates;

[0069] Step S23, selecting key nodes in the plurality of two-dimensional plane coordinates;

[0070] Step S24: Using the key nodes as a reference, the coordinates of each two-dimensional plane are nonlinearly mapped to the XY plane of three-dimensional space through homogeneous coordinate transformation, and the temperature matrix is ​​stacked along the Z-axis to generate the initial three-dimensional temperature field.

[0071] Step S25: Generate a sine wave signal based on the time-frequency matrix;

[0072] Step S26: Fuse the initial three-dimensional temperature field with the sinusoidal signal to output a composite three-dimensional matrix, and normalize the composite three-dimensional matrix to generate a dynamic three-dimensional matrix.

[0073] It should be noted that the temperature matrix, time-frequency matrix, and topology matrix are normalized, and a dynamic Bayesian network structure is established based on the normalized temperature matrix, time-frequency matrix, and topology matrix; a time-slicing mechanism is introduced, with each time step corresponding to a set of weight parameters W. t =( , , ),in, The weights of the topological matrix G are... The weights of the temperature matrix T are... The weights are denoted as T for the time-frequency matrix F. A joint tree inference engine is used to calculate the posterior probability distribution, and the weights are output through a dynamic Bayesian network structure. Specifically, the temperature matrix, time-frequency matrix, and topology matrix are linearly normalized to the [0,1] interval, denoted as T. t F t G t Where t represents the time step; hidden state node H t : Represents the potential state of the system's dynamic evolution; weight parameter node W t : Dynamically adjust the influence of different matrices on the hidden state; the dependencies between time slices are connected by directed edges (such as H t →H t+1 ).

[0074] Furthermore, the conditional probability of the hidden state is jointly determined by the weight parameters and the observation matrix:

[0075] P(H t |H t-1 W t = Softmax ( T t + F t + G t );

[0076] Among them, P(H t |H t-1 W t) is the conditional probability of hidden state; Softmax is the Softmax activation function; W is the weight parameter t Subject to Dirichlet prior distribution, reflecting the time dynamics.

[0077] Further, the weight is updated by Bayesian online learning:

[0078] ;

[0079] wherein, is a time decay factor, L is a loss function, and is optimized by gradient descent.

[0080] Further, the DBN (Dynamic Bayesian Network) is converted into a static joint tree structure, and the marginal probability is calculated by the message passing algorithm:

[0081] ;

[0082] wherein, is a node potential function; is a transition potential function; is a posterior probability; is a time series of hidden states of the power distribution network; is a sequence of temperature distribution data from time 1 to t; is a sequence of high-frequency transient signals from time 1 to t; is a sequence of topological feature data from time 1 to t; is a normalization constant; is the hidden state of the power distribution network at the i-th time step; is the hidden state of the power distribution network at the i+1-th time step; is the temperature distribution data sequence at the i-th time step; is the high-frequency transient signal sequence at the i-th time step; is the topological feature data sequence at the i-th time step.

[0083] Further, the final posterior distribution of the weight parameter is:

[0084] ;

[0085] wherein, is the posterior distribution of the weight parameter; is a time series of weight parameters of the model; is the value of the weight W that maximizes the subsequent expression; is a joint probability.

[0086] Finally, the dynamic weights of the topological features, temperature distribution data and high-frequency transient signals are solved by using Maximum A Posteriori Estimation (MAP) or Monte Carlo sampling.

[0087] Based on the above, the topological nodes of the topological matrix are mapped to the X-Y plane coordinates of the dynamic three-dimensional matrix; the affine transformation is performed on the topological node coordinates by using the adjacency matrix normalization method in graph theory. The formula is expressed as: , wherein is the coordinate matrix of the transformed topological nodes in the X-Y plane of the dynamic three-dimensional matrix, which is composed of a plurality of two-dimensional plane coordinates, is the weight matrix, is the adjacency matrix, is the transpose of the original coordinate matrix of the topological nodes (composed of the coordinates of a plurality of topological nodes), is the translation vector. Then, based on the plane-three-dimensional coordinate transformation principle, three topological key nodes are selected as the reference coordinate system (such as the centroid, the maximum degree node and the boundary point), and the nonlinear mapping is realized by using the homogeneous coordinate transformation to stack the temperature distribution data of the temperature matrix and the high-frequency periodic data of the time-frequency matrix along the Z axis of the dynamic three-dimensional matrix corresponding to each time step; the time-frequency matrix contains the main frequency component and the amplitude , for each position (x, y), a sinusoidal signal is generated along the Z axis: , wherein is the sinusoidal signal, is the signal amplitude corresponding to the time t and the position (x, y), which is obtained from the amplitude component of the position in the time-frequency matrix, reflecting the intensity of the high-frequency transient signal, is the signal frequency corresponding to the time t and the position (x, y), which is from the main frequency component of the position in the time-frequency matrix, embodying the oscillation frequency characteristics of the high-frequency transient signal, z is the vertical axis coordinate, z ∈ [1, Z max ], Z max is the maximum coordinate value in the Z axis direction.

[0088] Further, is updated according to the time step t and is stacked synchronously with the temperature data; the temperature distribution and the sinusoidal wave data are fused according to the weights to generate a composite dynamic three-dimensional matrix (composite three-dimensional matrix):

[0089] ;

[0090] , wherein is the composite three-dimensional matrix; , is the weight coefficient, This is used to balance the contributions of temperature and frequency domain information; This represents the initial three-dimensional temperature field; It is a sine wave signal; The x-axis represents the distribution network topology. The vertical axis represents the distribution network topology. For time / signal evolution dimension.

[0091] Finally, for Normalization is performed to prevent numerical overflow, resulting in a dynamic three-dimensional matrix.

[0092] In this system, the temperature matrix at each time step is stacked sequentially onto the Z-axis, forming a spatiotemporally continuous three-dimensional volumetric data. This structure is similar to the representation of volumetric objects in GIS, intuitively reflecting the temporal evolution of the temperature field. The high-frequency components (periodic vibration signals) of the time-frequency matrix are extracted using Fourier transform and encoded as sine waves along the Z-axis. For example, amplitude reflects signal strength, frequency corresponds to period length, and phase shift represents time delay. The XY plane preserves the topological structure, while the Z-axis simultaneously encodes the temporal evolution of temperature (static trend) and the periodic fluctuations of high-frequency signals (dynamic details), facilitating the reveal of hidden correlations using 3D visualization tools (such as contour lines in MATLAB or Surfer). Furthermore, in power equipment monitoring, the temperature field reflects the heat dissipation status, and high-frequency vibration signals may correspond to mechanical wear. The superposition of these two can establish a thermo-mechanical coupling model to assess the overall health status of the equipment. Sine perturbations in the temperature distribution may correspond to periodic heating during equipment failure; after superposition, abnormal nodes can be quickly located through waveform characteristics.

[0093] Specifically, an affine transformation is performed on the topological nodes of the topological matrix to map the irregularly distributed nodes to regular two-dimensional plane coordinates. Then, key nodes (including the centroid, the largest node, and boundary points) are selected from all the regular two-dimensional plane coordinates. Using these three key nodes as references, a homogeneous coordinate transformation is used to nonlinearly map the two-dimensional plane coordinates to the XY plane in three-dimensional space. Next, the time-series temperature matrix data is stacked along the Z-axis to form an initial three-dimensional temperature field D3D(x,y,z). Simultaneously, a sinusoidal signal reflecting frequency characteristics is generated using the time-frequency matrix data. Finally, the temperature field and the frequency domain signal are fused according to the weighting formula and normalized to obtain a dynamic three-dimensional matrix that can characterize both the spatiotemporal distribution of temperature and include frequency domain features. The entire process realizes a systematic transformation from raw topological data to a multi-dimensional dynamic feature expression.

[0094] In this embodiment, topological features represent the system's structural correlations (such as the connection relationships between power grid nodes), reflecting the logical constraints between devices. Temperature distribution data monitors the thermal state of equipment (such as localized overheating of transformers) through infrared thermal imaging or sensors. High-frequency transient signals capture rapid electromagnetic fluctuations (such as arc fault characteristics). The three-dimensional dynamic matrix achieves multi-dimensional coupling of spatial topology, thermodynamics, and electromagnetic transients through dynamic weight allocation, overcoming the limitations of a single data source. Simultaneously, the three-dimensional dynamic matrix can express spatial structure, thermodynamic state, and temporal dynamic characteristics, capturing transient changes in the power grid. The fusion of these three aspects improves the real-time performance and accuracy of fault prediction. The three-dimensional matrix associates fault points with spatial coordinates, enabling precise location of abnormal equipment positions and rapid differentiation of fault types. Furthermore, the weights of topological features, temperature distribution data, and high-frequency transient signals are dynamic. Through weight allocation (e.g., high weight for temperature data, low weight for sine waves), redundant information can be reduced. For example, coarse-grained sampling is used for low-frequency temperature changes, while only key periodic components are retained for high-frequency signals, thereby reducing storage and computational overhead.

[0095] Step 103: Generate multiple priority self-healing strategies based on the preset basic strategy library, distribution network optimization function, and distribution network constraints.

[0096] It should be noted that, based on the dynamic three-dimensional matrix to obtain fault location information and minimum isolation area information, the distribution network optimization function is: Min( In the given condition, w1 > w2 > w3. Preferably, w1 is 0.7, w2 is 0.2, and w3 is 0.1. Distribution network constraints include topology constraints, power balance constraints, and temperature constraints. Topology constraints: Network connectivity requirements (e.g., tree-like radial networks). Power balance constraints: Node-injected power equals load demand (P...). 注入 =P 负荷 Temperature constraint: The operating temperature of the equipment shall not exceed the rated value (T). 设备 ≤Tmax).

[0097] Specifically, step 103 may include the following sub-steps S31-S34:

[0098] Step S31: Based on the preset basic strategy library, determine multiple initial candidate self-healing strategies;

[0099] Step S32: Based on the distribution network constraints, screen each initial candidate self-healing strategy to determine multiple intermediate candidate self-healing strategies;

[0100] Step S33: Use the distribution network optimization function to quantitatively evaluate each intermediate candidate self-healing strategy and output the objective function value of each intermediate candidate self-healing strategy.

[0101] Step S34, sort the intermediate candidate self-healing strategies in ascending order according to the target function values, and determine a plurality of priority self-healing strategies.

[0102] The preset basic strategy library is a preset typical recovery strategy (such as segment switch priority, important load priority) such as a circuit breaker tripping sequence, which contains 10 types of standard operation processes (SOP).

[0103] The historical strategy library is an optimized scheme based on historical fault scenes, which generates a mapping table by mining historical fault handling records through LSTM.

[0104] It should be noted that the application generates a candidate strategy set through a multi-objective optimization algorithm (such as NSGA-II (Non-dominated Sorting Genetic Algorithm II, Non-dominated Sorting Genetic Algorithm II), ADMM2 (Alternating Direction Method of Multipliers 2, Alternating Direction Method of Multipliers 2)) to ensure that all constraint conditions are met. Specifically, first, when preprocessing the two strategy libraries, the basic strategy library itself contains 10 different standard operation processes (such as segment switch priority, important load priority, and different circuit breaker tripping sequences), and the historical strategy library will also form multiple optimized schemes matched with different fault scenes after mining historical fault records through LSTM (Long Short-Term Memory, Long Short-Term Memory Network). Therefore, when matching the current fault scene (such as line structure, load distribution, fault type), multiple strategies with the same label (such as 2-3 standard strategies matching the "single-phase grounding" and "tree line" labels) will be selected from the basic strategy library, and multiple historical optimization schemes with a similarity of ≥80% will be matched from the historical strategy library, not a single strategy; then enter the candidate strategy generation stage, and the multiple strategies matched from the two libraries are respectively fused with "template + experience" (such as supplementing different switch action time sequences and energy storage capacity parameters for different strategies), which will form multiple initial candidate strategies. After verifying the topological constraints, power balance constraints, and temperature constraints, even if some infeasible strategies are removed, multiple strategies that meet the constraints will still be retained. When further optimized through a multi-objective optimization algorithm, 3-5 feasible final candidate strategy sets are generated, ensuring that the number of strategies meets the multi-priority requirement; then perform optimization function quantitative evaluation, calculate the number of households affected by each strategy (count the number of users affected by different strategies according to the isolation range), the recovery time (statistical difference in execution time of different strategies), and the device loss (estimate the energy loss value of different strategies) for the 3-5 candidate strategies, and then substitute them into the optimization function Min (N+T+D) to obtain the optimal strategy set. (w1=0.7, w2=0.2, w3=0.1), to obtain the target function value of each candidate strategy (such as 31.9 for strategy A, 34.2 for strategy B, and 44.6 for strategy C); when performing priority ranking, the multiple candidate strategies are ranked in ascending order of the target function value (such as strategy A being ranked 1st, strategy B being ranked 2nd, and strategy C being ranked 3rd), and if there are strategies with a target function value difference of less than 5% (such as strategy D being 32.5, which is close to strategy A), the ranking is further fine-tuned in combination with the anti-interference capability (such as strategy A relying on 2 communication nodes and strategy D relying on 3 communication nodes), and finally a group of multiple strategies with a clear priority order is determined (such as 1st strategy A, 2nd strategy D, 3rd strategy B, and 4th strategy C), rather than a single strategy, and these ranked strategies collectively constitute a priority strategy set (i.e., multiple priority self-healing strategies), which can be flexibly enabled according to the actual fault handling situation (such as communication node failure when the 1st strategy is executed), to ensure the reliability of the self-healing process.

[0105] Step 104: Based on the fuzzy analytic hierarchy process, the power distribution network optimization function is used to screen each priority self-healing strategy according to the dynamic three-dimensional matrix, and a target power distribution network self-healing strategy is determined.

[0106] It should be noted that, in use, the number of households affected by power outage in the optimization function is quantified by the product of the fault isolation range and the recovery time; the recovery time is based on the spatiotemporal scheduling constraints of mobile energy storage to minimize the key load recovery time; and the equipment loss takes into account the charging and discharging efficiency of mobile energy storage and line loss.

[0107] Further, a hierarchical structure is established: target layer: optimal recovery scheme. Criterion layer: number of households affected by power outage, recovery time, and equipment loss. Scheme layer: candidate strategy set. Fuzzy judgment matrix construction: through expert scoring or data statistics, a fuzzy complementary judgment matrix between criteria is constructed, for example, using triangular fuzzy numbers to represent weight relationships. Scheme scoring: the two-library strategies of the basic strategy library and the historical strategy library are fused through fuzzy logic to generate a priority strategy set; the index values of each candidate strategy are normalized, and the comprehensive score is calculated by combining the fuzzy comprehensive evaluation model. In the simulation platform, real line parameters are reproduced to verify whether it is the optimal self-healing strategy.

[0108] Specifically, step 104 can include the following sub-steps S41-S46:

[0109] Step S41: The fixed weights of the power distribution network optimization function are converted using the fuzzy analytic hierarchy process to determine the criterion weight vector;

[0110] Step S42: Based on the dynamic three-dimensional matrix, the original index values of each priority self-healing strategy are determined;

[0111] Step S43, normalize each original index value to determine a normalized index value corresponding to each original index value;

[0112] Step S44, determine a comprehensive score of each priority self-healing strategy according to each normalized index value and a criterion weight vector;

[0113] Step S45, select a priority self-healing strategy corresponding to the maximum comprehensive score as an initial power distribution network self-healing strategy;

[0114] Step S46, verify the initial power distribution network self-healing strategy to determine a target power distribution network self-healing strategy.

[0115] It should be noted that a clear hierarchy is first constructed, and the target layer, the criterion layer and the scheme layer are determined, wherein the target layer is "selecting the optimal self-healing strategy under 10kV single-phase grounding fault"; the criterion layer is closely related to the core target of the optimization function, and is set as "household number during power outage", "recovery time" and "device loss", which are three key evaluation indexes (consistent with the evaluation dimensions corresponding to w1, w2 and w3 in the optimization function), and the basic data of the three indexes are all derived from the dynamic three-dimensional matrix, the fault isolation area is determined through the X-Y plane topology node mapping of the dynamic three-dimensional matrix, then the "household number during power outage" is counted, the "recovery time" is determined by judging the fault processing timing efficiency through the Z-axis superimposed high-frequency transient signal cycle data, and the "device loss" is determined by estimating the device heat loss through the temperature distribution data of the Z-axis; the scheme layer is the priority strategy set (such as candidate strategies A, B and C) obtained after the screening and optimization function preliminary evaluation of the basic strategy library and the historical strategy library, and the fault positioning information and the minimum isolation area information required in the strategy screening process are also provided by the dynamic three-dimensional matrix after the fusion of the topological characteristics, temperature data and high-frequency signals.

[0116] Then the fuzzy judgment matrix is constructed. For the criterion layer, the relative importance of the three criteria is quantitatively assigned by inviting grid operation experts to combine the actual operation and maintenance experience of 10kV distribution network, and referring to the fault severity data (such as high-frequency transient signal energy change, temperature abnormal amplitude) output by the dynamic three-dimensional matrix in real time, and using triangular fuzzy numbers (such as "(1, 2, 3)" representing "slightly important" and "(3, 4, 5)" representing "obviously important"). For example, when the dynamic three-dimensional matrix shows that the temperature in the fault area is abnormally severe (the risk of equipment loss is high), the importance weight of the "equipment loss" criterion can be appropriately increased. If the dynamic three-dimensional matrix shows that the user density in the fault area is large (the range of power outage impact is wide), the importance of "household number during power outage" is strengthened. Combined with the core weight relationship w1> w2> w3 in the optimization function, the importance of "household number during power outage" to "restoration time" is determined to be "obviously important" (corresponding to triangular fuzzy number "(3, 4, 5)"), and the importance of "restoration time" to "equipment loss" is "slightly important" (corresponding to triangular fuzzy number "(1, 2, 3)"), forming the fuzzy complementary judgment matrix of the criterion layer. For the scheme layer, based on expert experience, historical data and the simulation execution data of each strategy provided by the dynamic three-dimensional matrix (such as the temperature change and signal fluctuation in the isolation area when the simulation strategy A is executed), the performance of each candidate strategy under the three criteria is evaluated. For example, under the "household number during power outage" criterion, the dynamic three-dimensional matrix shows that strategy A covers only 20 households and strategy B covers 50 households, so strategy A is given a "strongly important" fuzzy evaluation of strategy B (corresponding to triangular fuzzy number "(5, 6, 7)"), and then the fuzzy judgment matrix of the scheme layer under each criterion is formed.

[0117] Then the weight calculation and consistency check are performed. First, the fuzzy judgment matrix of the criterion layer is processed by "fuzzy number de-fuzzification" (such as using the barycentric method to calculate the mean value of triangular fuzzy numbers), which is converted into a clear judgment matrix. Then, the weight calculation method of traditional AHP such as sum method and root method is used to solve the weight vector of the three criteria. In this process, the real-time fault characteristics (such as whether the fault involves important loads and whether the current temperature of the equipment is close to the rated value) fed back by the dynamic three-dimensional matrix are combined to ensure that the weight results not only meet the core logic w1=0.7, w2=0.2, w3=0.1 in the optimization function, but also adapt to the dynamic needs of the current fault (such as adjusting the "restoration time" weight to 0.25 when the fault involves important loads). If the weight deviation is large, the fuzzy judgment matrix needs to be adjusted again. At the same time, the consistency index CI, the random consistency index RI and the consistency ratio CR are calculated to check the logical consistency of the fuzzy judgment matrix. The requirement is CR<0.1. If it does not meet the requirement, the expert evaluation assignment is revised based on the real-time data of the dynamic three-dimensional matrix until the matrix passes the consistency check.

[0118] After that, the program fuzzy scoring is carried out, first, the original index values of each candidate strategy under three criteria are obtained based on the dynamic three-dimensional matrix, for example, the "household number during power failure" of strategy A is the isolated area user number (20 households) counted by the dynamic three-dimensional matrix, the "restoration time" is the switch action time sequence and signal response time length (15 minutes) fed back by the matrix, and the "equipment loss" is the energy loss corresponding to the temperature rise calculated by the matrix ); the original index values are normalized (for example, the household number during power failure is mapped to the [0, 1] interval according to the principle of "the smaller the better", and the total number of users in the network provided by the dynamic three-dimensional matrix) to eliminate the dimensional difference; then, combined with the determined criterion weight vector, the "weighted average type" fuzzy comprehensive evaluation model is used to multiply and sum the normalized scores of each strategy under each criterion and the corresponding criterion weight to obtain the comprehensive score of each candidate strategy (the higher the score, the better the strategy), and during the scoring process, if the dynamic three-dimensional matrix feedbacks that there is a potential temperature over-temperature risk for a strategy, the score of the "equipment loss" dimension of the strategy can be appropriately deducted.

[0119] Finally, the optimal scheme is determined, all candidate strategies are sorted according to the comprehensive score from high to low, and the strategy with the highest score is selected as the preliminary optimal scheme (initial distribution network self-healing strategy); in order to ensure the feasibility and adaptability of the scheme, the real line parameters (such as line length, transformer capacity, wire diameter, etc.) of the 10kV distribution network are reproduced in the simulation platform, and the device temperature, high-frequency signal and topological structure data collected by the dynamic three-dimensional matrix are imported to simulate the execution process of the preliminary optimal scheme, and to verify whether the execution topology feedback by the dynamic three-dimensional matrix meets the connectivity (topology constraint), whether the power distribution is balanced (power balance constraint), and whether the device temperature is controlled within the rated value (temperature constraint); if the verification is passed, the optimal self-healing scheme (target distribution network self-healing strategy) is finally determined, if the verification is not passed (for example, the dynamic three-dimensional matrix shows that the temperature of a node after execution is over-temperature), the evaluation assignment of the fuzzy judgment matrix is adjusted again combined with the matrix data, and the above scoring process is repeated until the optimal scheme that passes the verification is selected.

[0120] Further, after obtaining the optimal self-healing strategy (target power distribution network self-healing strategy), the dynamic three-dimensional matrix is projected into a two-dimensional strategy space by using SVD (singular value decomposition), the principal components are extracted to generate a fixed value matrix M, and the parameters of each node, such as overcurrent protection threshold and reclosing time, are included. The fixed value matrix M is converted into a CID (configuration instance description) file according to the IEC 61850 protocol, and is adapted to the LD (logic device) model of the intelligent switch. The key fields include: overcurrent setting based on wire diameter, adaptive reclosing sequence driven by time-frequency characteristics, and dynamic threshold constrained by transformer capacity. The DL / T 860-9-2 protocol is used to download data (1 remote control interface compatibility) through optical fiber or 5G network. AES-256 encryption and CRC-32 check are used to ensure the integrity of transmission. The intelligent switch has a built-in lightweight decision module that updates the fixed value based on real-time data, and downloads the fixed value matrix to the corresponding integrated intelligent switch, thereby completing the optimal self-healing strategy for the corresponding line.

[0121] As a comparison of technical effects, reference can be made in combination with the prior art. The distribution line has a network structure, numerous branches and complex line conditions. With the wide application of distribution automation switches in 10 kV distribution lines, the power supply reliability is greatly improved after the implementation of the master station and the voltage-time cooperative self-healing power supply. However, after the failure of the master station self-healing, only the original fault location and isolation mode can be used to judge the isolation fault, which puts higher requirements on the skill level and practical experience of the dispatching operator. When the fault phenomenon is not obvious or multiple lines fail at the same time, it is difficult to quickly isolate the fault section, thereby causing the expansion of the power distribution line fault outage range or the complete outage of the entire line.

[0122] In view of the above problems, the present application provides a power distribution network self-healing strategy generation method, please refer to Figure 2 , real-time acquisition of device surface temperature distribution data; real-time acquisition of sudden high-frequency transient signals of the power distribution line nodes; construction of a dynamic topology graph and extraction of topology features; fusion of the topology features, temperature distribution data and high-frequency transient signals, and dynamic calculation of the weights of the topology features, temperature distribution data and high-frequency transient signals according to a Bayesian network evaluation module to obtain a dynamic three-dimensional matrix; establishment of an optimization function based on the dynamic three-dimensional matrix; obtaining a priority strategy according to a preset basic strategy library, a historical strategy library and the optimization function; comprehensive scoring of the priority strategy by using a fuzzy analytic hierarchy process to obtain an optimal scheme. The present application can capture transient changes in the power grid, and the dynamic weights can respond to environmental changes in real time, which is more suitable for complex power grid environments.

[0123] Further, the application fuses the topological features, temperature distribution data and high-frequency transient signals, and dynamically calculates the weights of the topological features, temperature distribution data and high-frequency transient signals according to a Bayesian network evaluation module to obtain a dynamic three-dimensional matrix. The Bayesian network dynamically adjusts the weights through probability reasoning and can cope with system state mutations. For example, at the initial stage of power grid failure, the weight of the high-frequency transient signal automatically increases to quickly locate the disturbance; when in steady state, the weight of the temperature distribution dominates to monitor the equipment health status. This dynamic adjustment mechanism is superior to the fixed weight fusion method. Moreover, the Bayesian network can handle uncertainty problems such as sensor noise and data missing. When a sensor fails and causes abnormal temperature data, the network will automatically correct the weight according to the topological correlation (such as the temperature correlation of adjacent nodes) and historical data to avoid misjudgment. In the power system fault diagnosis scenario, the dynamic matrix can not only narrow down the fault range and provide positioning accuracy through topological correlation, but also improve the accuracy of fault type recognition through transient waveform feature and temperature anomaly pattern matching.

[0124] In summary, the dynamic three-dimensional matrix constructed by the application can express spatial structure, thermodynamic state and time domain dynamic characteristics at the same time, can capture transient changes of the power grid, and the fusion of the three can improve the real-time performance and accuracy of fault monitoring. Moreover, the weights of the fusion of the three are dynamic weights, which can respond to environmental changes in real time, are more suitable for complex power grid environments, and improve the accuracy and efficiency of self-healing strategy acquisition. At the same time, the application takes the practical difficulties encountered in the construction and operation of 10kV distribution networks as the starting point, realizes the pre-rehearsal of different self-healing strategies in the multi-source complex distribution network scenario, simulates and predicts the outage range, researches and compares multiple types of self-healing strategies, selects the strategy with the minimum number of households from different self-healing strategies, and generates a constant matrix of the self-healing strategy based on the basic parameters such as the line model, transformer capacity distribution and line diameter. The constant matrix is loaded into the corresponding integrated intelligent switch to complete the optimal self-healing strategy for the corresponding line. In addition, the application effectively solves the problems of self-healing line constant calculation difficulty, implementation difficulty, and system comparison between different strategies, and quickly disposes the 10kV line grounding or broken line fault, improves power supply reliability and reduces the number of households during power outage. It empowers and increases the efficiency of the safe and reliable operation of the distribution network, and effectively guarantees the quality of the fast power restoration of the distribution network.

[0125] In the embodiment of the present application, the present application provides a power distribution network self-healing strategy generation method, obtains power distribution network equipment surface temperature distribution data and mutation high-frequency transient signal of power distribution line nodes; constructs a power distribution network dynamic topology graph, and constructs a dynamic three-dimensional matrix according to the power distribution network dynamic topology graph, the power distribution network equipment surface temperature distribution data and the mutation high-frequency transient signal of the power distribution line nodes; generates a plurality of priority self-healing strategies according to a preset basic strategy library, a power distribution network optimization function and power distribution network constraint conditions; and determines a target power distribution network self-healing strategy by screening each priority self-healing strategy according to the dynamic three-dimensional matrix based on the power distribution network optimization function by using a fuzzy analytic hierarchy process. Based on the above scheme, the present application can comprehensively depict the spatial position, thermal evolution trend and electromagnetic characteristics of the fault by fusing the power distribution network topology, equipment temperature and high-frequency transient multi-class data with the aid of the dynamic three-dimensional matrix. Meanwhile, a plurality of priority strategies are generated based on the basic strategy library and the optimization function, and the screening process combines real-time fault data of the dynamic three-dimensional matrix, so that the complex scenes such as intermittent grounding and multi-line concurrent fault can be flexibly adapted, the fault can be quickly isolated and the load can be accurately restored, and the quality of power distribution network rapid restoration is effectively improved.

[0126] Please refer to Figure 3 , Figure 3 The structure block diagram of a power distribution network self-healing strategy generation system provided for the second embodiment of the present application is shown in FIG. 2.

[0127] The power distribution network self-healing strategy generation system provided by the present application comprises:

[0128] The obtaining module 301 is configured to obtain power distribution network equipment surface temperature distribution data and mutation high-frequency transient signal of power distribution line nodes.

[0129] The construction module 302 is configured to construct a power distribution network dynamic topology graph, and construct a dynamic three-dimensional matrix according to the power distribution network dynamic topology graph, the power distribution network equipment surface temperature distribution data and the mutation high-frequency transient signal of the power distribution line nodes.

[0130] The generation module 303 is configured to generate a plurality of priority self-healing strategies according to a preset basic strategy library, a power distribution network optimization function and power distribution network constraint conditions.

[0131] The determination module 304 is configured to determine a target power distribution network self-healing strategy by screening each priority self-healing strategy according to the dynamic three-dimensional matrix based on the power distribution network optimization function by using a fuzzy analytic hierarchy process.

[0132] Further, the construction module 302 comprises:

[0133] The first sub-module is configured to determine a topology matrix, a temperature matrix and a time-frequency matrix based on the power distribution network dynamic topology graph, the power distribution network equipment surface temperature distribution data and the mutation high-frequency transient signal of the power distribution line nodes.

[0134] The second sub-module is configured to perform affine transformation on the topology nodes of the topology matrix to generate a plurality of two-dimensional plane coordinates;

[0135] The third sub-module is configured to select key nodes from the plurality of two-dimensional plane coordinates;

[0136] The fourth sub-module is configured to take the key nodes as a reference, perform homogeneous coordinate transformation on each two-dimensional plane coordinate to nonlinearly map the two-dimensional plane coordinate to an X-Y plane of a three-dimensional space, and stack the temperature matrix along a Z axis to generate an initial three-dimensional temperature field;

[0137] The fifth sub-module is configured to generate a sinusoidal signal based on the time-frequency matrix;

[0138] The sixth sub-module is configured to fuse the initial three-dimensional temperature field and the sinusoidal signal to output a composite three-dimensional matrix, and perform normalization processing on the composite three-dimensional matrix to generate a dynamic three-dimensional matrix.

[0139] Further, the first sub-module is specifically configured to:

[0140] generate a temperature matrix based on the surface temperature distribution data of the power distribution network equipment;

[0141] perform matrix transformation on the sudden high-frequency transient signal of the power distribution line node to output a time-frequency matrix;

[0142] extract topology features in the dynamic topology graph of the power distribution network, and perform graph embedding on the topology features to generate a topology matrix.

[0143] Further, the generation module 303 is specifically configured to:

[0144] determine a plurality of initial candidate self-healing strategies based on a preset basic strategy library;

[0145] screen each initial candidate self-healing strategy based on a constraint condition of the power distribution network to determine a plurality of intermediate candidate self-healing strategies;

[0146] quantitatively evaluate each intermediate candidate self-healing strategy by using a power distribution network optimization function to output a target function value of each intermediate candidate self-healing strategy;

[0147] sort each intermediate candidate self-healing strategy in ascending order according to each target function value to determine a plurality of priority self-healing strategies.

[0148] Further, the determination module 304 is specifically configured to:

[0149] convert a fixed weight of the power distribution network optimization function by using a fuzzy analytic hierarchy process to determine a criterion weight vector;

[0150] determine an original index value of each priority self-healing strategy based on the dynamic three-dimensional matrix;

[0151] normalizing each original index value to determine a normalized index value corresponding to each original index value;

[0152] determining a comprehensive score of each priority self-healing strategy according to each normalized index value and a criterion weight vector;

[0153] selecting a priority self-healing strategy corresponding to the maximum comprehensive score as an initial power distribution network self-healing strategy;

[0154] verifying the initial power distribution network self-healing strategy to determine a target power distribution network self-healing strategy.

[0155] Further, the composite three-dimensional matrix is specifically:

[0156]

[0157] wherein, is the composite three-dimensional matrix; , is a weight coefficient; is an initial three-dimensional temperature field; is a sinusoidal signal; is a horizontal coordinate of the power distribution network topology; is a vertical coordinate of the power distribution network topology; is a time / signal evolution dimension.

[0158] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, modules and sub-modules can refer to the corresponding processes in the foregoing method embodiments, which will not be described herein.

[0159] The embodiment of the present application also provides a computer device, which comprises a memory and a processor, and the memory stores a computer program; the computer program is executed by the processor to enable the processor to execute the steps of the power distribution network self-healing strategy generation method according to any one of the foregoing embodiments.

[0160] The embodiment of the present application also provides a computer readable storage medium, which stores a computer program / instruction; the computer program / instruction is executed by a processor to implement the steps of the power distribution network self-healing strategy generation method according to any one of the foregoing embodiments.

[0161] The embodiment of the present application also provides a computer program product, which comprises a computer program / instruction; the computer program / instruction is executed by a processor to implement the steps of the power distribution network self-healing strategy generation method according to any one of the foregoing embodiments.

[0162] ​In several embodiments provided in the present application, it should be understood that the disclosed system and method can be implemented in other manners. For example, the division of the units is only a logical function division, and there can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0163] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, can be located in one place, or can be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0164] The above description and the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can still be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for power distribution network self-healing strategy generation, characterized in that, The method comprises the following steps: obtaining power distribution network equipment surface temperature distribution data and sudden high-frequency transient signal of power distribution line nodes; constructing a dynamic topology graph of the power distribution network, and constructing a dynamic three-dimensional matrix according to the dynamic topology graph of the power distribution network, the power distribution network equipment surface temperature distribution data and the sudden high-frequency transient signal of the power distribution line nodes; generating a plurality of priority self-healing strategies according to a preset basic strategy library, a power distribution network optimization function and power distribution network constraint conditions; screening each of the priority self-healing strategies according to the dynamic three-dimensional matrix by using the power distribution network optimization function based on a fuzzy analytic hierarchy process, and determining a target power distribution network self-healing strategy.

2. The power distribution grid self-healing strategy generation method of claim 1, wherein, The method of constructing a dynamic three-dimensional matrix according to the dynamic topology graph of the power distribution network, the power distribution network equipment surface temperature distribution data and the sudden high-frequency transient signal of the power distribution line nodes comprises the following steps: determining a topology matrix, a temperature matrix and a time-frequency matrix based on the dynamic topology graph of the power distribution network, the power distribution network equipment surface temperature distribution data and the sudden high-frequency transient signal of the power distribution line nodes; performing affine transformation on the topology nodes of the topology matrix to generate a plurality of two-dimensional plane coordinates; selecting key nodes from the plurality of two-dimensional plane coordinates; taking the key nodes as a reference, performing homogeneous coordinate transformation to nonlinearly map each of the two-dimensional plane coordinates to an X-Y plane of a three-dimensional space, and stacking the temperature matrix along a Z axis to generate an initial three-dimensional temperature field; generating a sinusoidal signal based on the time-frequency matrix; fusing the initial three-dimensional temperature field and the sinusoidal signal to output a composite three-dimensional matrix, and performing normalization processing on the composite three-dimensional matrix to generate a dynamic three-dimensional matrix.

3. The power distribution grid self-healing strategy generation method of claim 2, wherein, The method of determining a topology matrix, a temperature matrix and a time-frequency matrix based on the dynamic topology graph of the power distribution network, the power distribution network equipment surface temperature distribution data and the sudden high-frequency transient signal of the power distribution line nodes comprises the following steps: generating a temperature matrix according to the power distribution network equipment surface temperature distribution data; performing matrix transformation on the sudden high-frequency transient signal of the power distribution line nodes to output a time-frequency matrix; extracting topology features in the dynamic topology graph of the power distribution network, and performing graph embedding on the topology features to generate a topology matrix.

4. The method of claim 1, wherein, The method of generating a plurality of priority self-healing strategies according to a preset basic strategy library, a power distribution network optimization function and power distribution network constraint conditions comprises the following steps: determining a plurality of initial candidate self-healing strategies based on the preset basic strategy library; screening each of the initial candidate self-healing strategies based on the power distribution network constraint conditions to determine a plurality of intermediate candidate self-healing strategies; quantitatively evaluating each of the intermediate candidate self-healing strategies by using the power distribution network optimization function to output a target function value of each of the intermediate candidate self-healing strategies; sorting each of the intermediate candidate self-healing strategies in ascending order according to the target function value to determine a plurality of priority self-healing strategies.

5. The method of claim 1, wherein, The method of screening each of the priority self-healing strategies according to the dynamic three-dimensional matrix by using the power distribution network optimization function based on a fuzzy analytic hierarchy process, and determining a target power distribution network self-healing strategy comprises the following steps: transforming a fixed weight of the power distribution network optimization function by using the fuzzy analytic hierarchy process to determine a criterion weight vector. Determine original index values of each of the priority self-healing strategies based on the dynamic three-dimensional matrix; Normalize each of the original index values to determine normalized index values corresponding to each of the original index values; Determine comprehensive scores of each of the priority self-healing strategies according to the normalized index values and a criterion weight vector; Select a priority self-healing strategy corresponding to the maximum comprehensive score as an initial power distribution network self-healing strategy; Verify the initial power distribution network self-healing strategy to determine a target power distribution network self-healing strategy.

6. The power distribution grid self-healing strategy generation method of claim 2, wherein, The composite three-dimensional matrix specifically comprises: ； wherein, is a complex three-dimensional matrix; , is a weight coefficient; is an initial three-dimensional temperature field; is a sinusoidal signal; is a horizontal coordinate of the power distribution network topology; is a vertical coordinate of the power distribution network topology; is a time / signal evolution dimension.

7. A power distribution grid self-healing strategy generation system, characterized by, comprises: An acquisition module configured to acquire surface temperature distribution data of power distribution network equipment and a sudden high-frequency transient signal of a power distribution line node; A construction module configured to construct a dynamic topology graph of the power distribution network, and construct a dynamic three-dimensional matrix according to the dynamic topology graph of the power distribution network, the surface temperature distribution data of the power distribution network equipment, and the sudden high-frequency transient signal of the power distribution line node; A generation module configured to generate a plurality of priority self-healing strategies according to a preset basic strategy library, a power distribution network optimization function, and power distribution network constraint conditions; A determination module configured to filter each of the priority self-healing strategies according to the dynamic three-dimensional matrix by using the power distribution network optimization function based on a fuzzy analytic hierarchy process, and determine a target power distribution network self-healing strategy.

8. A computer device, comprising: The computer program is executed to implement the power distribution network self-healing strategy generation method according to any one of claims 1-6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed to implement the power distribution network self-healing strategy generation method according to any one of claims 1-6.

10. A computer program product, characterised in that, The computer program product comprises a computer program stored on a non-transitory computer-readable storage medium, and the computer program comprises program instructions, wherein when the program instructions are executed by a computer, the computer executes the power distribution network self-healing strategy generation method according to any one of claims 1-6.

Images