Method and system for identifying element states during power system fault restoration

Abstract

The application belongs to the technical field of power system fault diagnosis and recovery, and proposes a component state identification method and system during power system fault recovery, determines multiple candidate attribute subsets according to a component disaster-causing condition attribute set, and reduces the component disaster-causing condition attribute set according to a knowledge granularity; based on information entropy, information filling is performed on a missing attribute interval of the reduced disaster-causing condition attribute set to obtain a complete disaster-causing condition attribute set; the component state correlation degree in the complete disaster-causing condition attribute set is calculated, a reasoning network correlation matrix is constructed based on the component state correlation degree, and a state unascertained component damage confidence is determined according to the reasoning network correlation matrix; the application can effectively utilize limited available information under the condition of information loss, realizes online reasoning on the unascertained component fault confidence by constructing a disaster-causing attribute set, calculating a state correlation degree and establishing a reasoning network, and improves the safety and efficiency of recovery decision.

Description

Technical Field

[0001] This invention relates to the field of power system fault diagnosis and recovery technology, specifically to a method and system for identifying the status of components during power system fault recovery. 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] Under the influence of extreme events (such as typhoons, extreme cold, earthquakes, etc.), the Cyber-Physical Power System (CPPS) can suffer severe damage, resulting in the inability to obtain status information of some power components (such as transmission lines, transformers, generators, etc.) through conventional monitoring methods (such as SCADA, PMU), leaving them in an unobservable state. This incomplete information poses a significant challenge to post-disaster recovery decision-making for the power system. The recovery control center struggles to accurately grasp the real-time topology and component status of the power grid, making it impossible to formulate safe and efficient recovery strategies. Furthermore, misjudging the status may even trigger secondary faults, thus prolonging power outages and increasing economic losses.

[0004] In existing solutions, obtaining the status of components after a disaster mainly relies on on-site inspection reports and limited observable point information, which is inefficient and has limited coverage. Although there are some fault diagnosis methods based on probability or artificial intelligence, their accuracy and robustness are insufficient in scenarios with large-scale information loss, sparse data, and complex correlations after extreme events. This is especially true in information-isolated areas where the dispatch center can hardly obtain any real-time information, making it difficult to effectively infer the status of components within the area. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for identifying the status of components during power system fault recovery. This method effectively utilizes limited available information even in the absence of other information. By constructing a set of disaster-causing attributes, calculating status correlations, and establishing an inference network, it enables online inference of the confidence level of unknown component faults, thereby improving the safety and efficiency of recovery decisions.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a method for identifying the status of components during power system fault recovery.

[0008] A method for identifying the status of components during power system fault recovery includes the following steps:

[0009] Obtain the dominant non-electrical factors that cause component damage under disaster conditions, and establish a set of disaster-causing condition attributes for the components based on the dominant non-electrical factors;

[0010] Multiple candidate attribute subsets are determined based on the component disaster-causing condition attribute set. The component disaster-causing condition attribute set is reduced based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset.

[0011] Information entropy is used to fill in the missing attribute intervals of the reduced disaster-causing condition attribute set to obtain the complete disaster-causing condition attribute set.

[0012] Calculate the correlation degree of component states in the complete disaster-causing condition attribute set;

[0013] Based on the correlation degree of component state, an inference network correlation matrix is ​​constructed, and the confidence degree of damage of components with unknown state is determined according to the inference network correlation matrix.

[0014] In one implementation of the first aspect of the present invention, the dominant non-electrical factors include the component's own properties and external environmental properties: the component's own properties include the insulator string's own gravity load, the conductor's own gravity load, the tower steel strength, the conductor's outer diameter, and the conductor's length; the external environmental properties include wind speed, precipitation rate, temperature, humidity, icing thickness, altitude, power flow, and voltage.

[0015] In one implementation of the first aspect of the present invention, multiple candidate attribute subsets are determined based on a set of disaster-causing conditions for components, and the set of disaster-causing conditions for components is reduced based on the knowledge granularity of the set of disaster-causing conditions for components and each candidate attribute subset, including:

[0016] Select multiple candidate attribute subsets from the disaster-causing condition attribute set;

[0017] Calculate the knowledge granularity of the disaster-causing condition attribute set and each candidate attribute subset separately;

[0018] Calculate the difference between the knowledge granularity of each candidate attribute subset and the knowledge granularity of the entire disaster-causing condition attribute set;

[0019] A subset of candidate attributes with a knowledge granularity difference less than a set threshold is selected as the reduced set of disaster-causing attribute conditions for the components.

[0020] In one implementation of the first aspect of the present invention, information is filled into the missing attribute intervals of the reduced disaster-causing condition attribute set based on information entropy to obtain a complete disaster-causing condition attribute set, including:

[0021] The reduced disaster-causing condition attribute set is split into multiple sub-unit sets containing only a single condition attribute;

[0022] Calculate the original information entropy of each sub-unit set;

[0023] Insert an object into each sub-unit set, gradually increasing the length of the missing attribute interval;

[0024] Calculate the information entropy after each interval expansion;

[0025] Select the interval corresponding to the information entropy closest to the original information entropy as the optimal filling interval;

[0026] Fill each subset unit according to its optimal filling interval, and then combine them to form a complete disaster-causing condition attribute set.

[0027] In one implementation of the first aspect of the present invention, calculating the component state correlation degree of the complete disaster-causing condition attribute set includes:

[0028] Based on the complete set of disaster-causing condition attributes, an attribute sequence is formed for each component;

[0029] The state correlation degree between components is calculated using Deng's grey relational degree formula.

[0030] As a further limitation of the first aspect of the present invention, for interval value attributes, the upper bound gray correlation degree and the lower bound gray correlation degree of the interval value attributes are calculated respectively, and the average value of the upper bound gray correlation degree and the lower bound gray correlation degree is taken as the final state correlation degree.

[0031] In one implementation of the first aspect of the present invention, a reasoning network correlation matrix is ​​constructed based on the component state correlation degree, and the confidence degree of damage of components with unknown states is determined according to the reasoning network correlation matrix, including:

[0032] Using the calculated component state correlation degree as elements, construct the correlation matrix between components with unknown state and components with known state;

[0033] The correlation matrix is ​​decomposed into multiple sub-correlation matrices based on electrical distance and component type.

[0034] Based on the sub-correlation matrices obtained from the decomposition, Petri network models based on the path of a state-known damaged component and the path of a state-known intact component are constructed respectively.

[0035] Calculate confidence propagation in the Petri network model;

[0036] The reasoning results of the integrated state-known damaged component path and the state-known intact component path are normalized to obtain the damage confidence of the unknown component.

[0037] Secondly, the present invention provides a component status identification system during power system fault recovery.

[0038] A component status identification system during power system fault recovery includes:

[0039] The attribute set construction unit is configured to: acquire the dominant non-electrical factors that cause component damage under disaster conditions, and establish a component disaster-causing condition attribute set based on the dominant non-electrical factors;

[0040] The attribute set reduction element is configured as follows: determine multiple candidate attribute subsets based on the component disaster-causing condition attribute set, and reduce the component disaster-causing condition attribute set based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset.

[0041] The missing information filling unit is configured to fill the missing attribute intervals of the reduced disaster-causing condition attribute set with information entropy to obtain the complete disaster-causing condition attribute set.

[0042] The correlation calculation unit is configured to calculate the correlation degree of the component states in the complete disaster-causing condition attribute set;

[0043] The confidence inference unit is configured to: construct an inference network association matrix based on the component state association degree, and determine the confidence degree of damage of components with unknown state based on the inference network association matrix.

[0044] In one implementation of the second aspect of the present invention, the attribute set construction unit is dominated by non-electrical factors, including component-specific attributes and external environmental attributes:

[0045] The inherent properties of the components include the gravity load of the insulator string itself, the gravity load of the conductor itself, the strength of the tower steel, the outer diameter of the conductor, and the length of the conductor; the external environmental properties include wind speed, precipitation rate, temperature, humidity, icing thickness, altitude, power flow, and voltage.

[0046] In one implementation of the second aspect of the present invention, in the attribute set reduction element, multiple candidate attribute subsets are determined based on the component disaster-causing condition attribute set, and the component disaster-causing condition attribute set is reduced based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset, including:

[0047] Select multiple candidate attribute subsets from the disaster-causing condition attribute set;

[0048] Calculate the knowledge granularity of the disaster-causing condition attribute set and each candidate attribute subset separately;

[0049] Calculate the difference between the knowledge granularity of each candidate attribute subset and the knowledge granularity of the entire disaster-causing condition attribute set;

[0050] A subset of candidate attributes with a knowledge granularity difference less than a set threshold is selected as the reduced set of disaster-causing attribute conditions for the components.

[0051] In one implementation of the second aspect of the present invention, the missing information filling unit fills in the missing attribute intervals of the reduced disaster-causing condition attribute set based on information entropy to obtain a complete disaster-causing condition attribute set, including:

[0052] The reduced disaster-causing condition attribute set is split into multiple sub-unit sets containing only a single condition attribute;

[0053] Calculate the original information entropy of each sub-unit set;

[0054] Insert an object into each sub-unit set, gradually increasing the length of the missing attribute interval;

[0055] Calculate the information entropy after each interval expansion;

[0056] Select the interval corresponding to the information entropy closest to the original information entropy as the optimal filling interval;

[0057] Fill each subset unit according to its optimal filling interval, and then combine them to form a complete disaster-causing condition attribute set.

[0058] In one implementation of the second aspect of the present invention, the correlation calculation unit calculates the correlation degree of the component states in the complete disaster-causing condition attribute set, including:

[0059] Based on the complete set of disaster-causing condition attributes, an attribute sequence is formed for each component;

[0060] The state correlation degree between components is calculated using Deng's grey relational degree formula.

[0061] For interval-valued attributes, calculate the upper bound grey relational degree and the lower bound grey relational degree of the interval-valued attributes respectively, and take the average of the upper bound grey relational degree and the lower bound grey relational degree as the final state relational degree.

[0062] In one implementation of the second aspect of the present invention, the confidence inference unit constructs an inference network correlation matrix based on the component state correlation degree, and determines the confidence degree of damage of components with unknown states based on the inference network correlation matrix, including:

[0063] Using the calculated component state correlation degree as elements, construct the correlation matrix between components with unknown state and components with known state;

[0064] The correlation matrix is ​​decomposed into multiple sub-correlation matrices based on electrical distance and component type.

[0065] Based on the sub-correlation matrices obtained from the decomposition, Petri network models based on the path of a state-known damaged component and the path of a state-known intact component are constructed respectively.

[0066] Calculate confidence propagation in the Petri network model;

[0067] The reasoning results of the integrated state-known damaged component path and the state-known intact component path are normalized to obtain the damage confidence of the unknown component.

[0068] Thirdly, the present invention provides a computer device, comprising: a processor and a computer-readable storage medium;

[0069] A processor, adapted to execute computer programs;

[0070] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the component status identification method during power system fault recovery as described in the first aspect of this invention.

[0071] Fourthly, the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed as described in the first aspect of the present invention, a method for identifying the status of components during power system fault recovery.

[0072] Fifthly, the present invention provides a computer program product comprising a computer program that, when executed by a processor, implements the component status identification method during power system fault recovery as described in the first aspect of the present invention.

[0073] Compared with the prior art, the beneficial effects of the present invention are:

[0074] This invention effectively integrates information on dominant non-electrical factors by constructing a set of component-induced disaster condition attributes, overcoming the limitations of relying solely on electrical monitoring data. It reduces data processing complexity and improves reasoning efficiency by using knowledge granularity for attribute set reduction. Information filling technology based on information entropy ensures data integrity and enhances the accuracy of state identification. By calculating component state correlation and constructing a reasoning network, it achieves online reasoning for the confidence level of unknown component faults, providing key state information for the recovery control center. This effectively solves the problem of component state identification under conditions of missing information, improving the safety and efficiency of recovery decision-making.

[0075] This invention does not rely on complete real-time monitoring data, but can effectively use limited and potentially sparse historical data, environmental data and some observables for inference. It is particularly suitable for recovery scenarios where information is severely lacking after extreme events. By constructing an attribute set that reflects the disaster-causing mechanism of extreme events, combining grey relational analysis to quantify the correlation between component states, and using Petri networks for confidence propagation, it can more accurately infer the fault state of unknown components.

[0076] The component damage confidence level output by this invention provides critical status information for the recovery control center, which helps to identify potential fault points, assess recovery path risks, optimize emergency repair resource allocation, and formulate safer and more reliable recovery strategies, thereby accelerating the system recovery process and reducing power outage losses. The data processing dimensions are reduced through attribute reduction, and data integrity is ensured through information filling methods. The calculation of grey relational analysis and Petri network is more efficient, meeting the timeliness requirements of online decision-making during recovery.

[0077] The dominant non-electrical factors of this invention include the component's own properties (such as the gravity load of the insulator string) and external environmental properties (such as wind speed and temperature). This solves the problem that existing solutions rely solely on conventional electrical monitoring methods and have a single source of information. By comprehensively considering the component's own disaster resistance capability and the influence of the external environment, the constructed component disaster-causing condition attribute set more comprehensively and accurately reflects the actual damage probability of the component. In scenarios where information is missing after extreme events, it can more accurately capture the key factors that lead to component damage, providing a reliable foundation for subsequent attribute reduction, correlation calculation, etc., improving the adaptability of the entire component state identification method to complex disaster situations, and effectively assisting in the formulation of recovery decisions.

[0078] This invention filters the reduced attribute set by selecting multiple candidate attribute subsets and calculating the knowledge granularity difference. By using knowledge granularity to measure the importance of attributes, the invention selects candidate subsets with small differences in knowledge granularity from the whole set. This can accurately eliminate irrelevant and redundant attributes, reduce the dimensionality of data processing, highlight key disaster-causing factors, and improve the efficiency and accuracy of subsequent steps such as information filling and correlation calculation. This makes state recognition more efficient and reliable, and provides stronger data support for recovery decisions.

[0079] This invention utilizes information entropy to fill in missing attribute intervals in a reduced attribute set. In extreme events leading to severe information loss, incomplete data can significantly impact component state assessment. Information entropy measures data uncertainty. By applying it to fill in missing attribute intervals, using the original information entropy as a reference, the interval is gradually expanded and new information entropy is calculated. The interval closest to the original is selected for filling, maximizing the use of existing information and ensuring the rationality and completeness of the filled data. This makes subsequent calculations based on the complete attribute set more accurate, improves the effectiveness of component state identification methods in information-deficient scenarios, and provides a reliable basis for recovery decisions.

[0080] This invention, based on a complete set of disaster-causing condition attributes, uses Deng's grey relational degree formula to calculate the correlation degree of component states. After extreme events in the power system, the states of components are interconnected but difficult to quantify directly. By comparing component attribute sequences and considering factors such as the degree of influence of disaster-causing condition attributes and the resolution coefficient, the correlation between the states of components can be accurately quantified. This solves the problem of difficulty in directly judging the mutual influence of components when information is missing, and provides key data for constructing the correlation matrix of the inference network. This enables subsequent inference to more accurately reflect the actual state relationship between components and improves the accuracy of confidence inference for unknown component faults.

[0081] 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

[0082] 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.

[0083] Figure 1 A flowchart illustrating a method for identifying the status of components during power system fault recovery, provided as an exemplary embodiment of the present invention;

[0084] Figure 2 A schematic diagram of a rigid straight rod model of a suspension insulator string provided as an exemplary embodiment of the present invention;

[0085] Figure 3 A failure rate curve of a transmission line tower under strong winds is provided as an exemplary embodiment of the present invention.

[0086] Figure 4 A schematic diagram of a Petri network for component failure confidence inference provided as an exemplary embodiment of the present invention;

[0087] Figure 5 A schematic diagram of a component status identification system during power system fault recovery provided as an exemplary embodiment of the present invention;

[0088] Figure 6 A schematic diagram of a computer device provided for an exemplary embodiment of the present invention. Detailed Implementation

[0089] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0090] 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.

[0091] This implementation proposes a method for component state identification during power system fault recovery. It constructs a component attribute set based on the disaster-causing mechanism of extreme disasters, reduces attributes using historical data, fills missing intervals using the information entropy principle, quantifies state correlation using grey relational analysis, and finally calculates damage confidence by integrating dual-path results through Petri networks (a graphical mathematical modeling and analysis tool). The proposed method meets the needs of state inference under incomplete information conditions, effectively supports fault point identification and risk assessment in recovery decision-making, and significantly improves the accuracy and timeliness of component state inference. Figure 1 As shown, the mechanism of extreme disasters is analyzed, then a set of disaster-causing condition attributes for components is established, and then the attribute set is reduced. After that, it is determined whether there are missing attributes. If there are, the missing attribute intervals are filled with information. Then, the component state correlation degree is calculated together with the case where there are no missing attributes. After the calculation is completed, an uncertain reasoning Petri network is established, and then the confidence propagation of the path of the known damaged component and the confidence propagation of the path of the known intact component are carried out simultaneously. Then, the reasoning results are integrated, and finally the component damage confidence degree is output.

[0092] More specifically, it includes the following processes:

[0093] Step S101: Analyze the disaster-causing mechanism of components under extreme disasters, obtain the causes of component damage, and establish a set of disaster-causing condition attributes for components;

[0094] Step S102: Perform attribute reduction on the disaster-causing condition attribute set based on historical cumulative data;

[0095] Step S103: Fill in the missing attribute intervals of the disaster-causing condition attribute set based on information entropy to obtain a complete set of component disaster-causing condition attributes;

[0096] Step S104: Calculate the state correlation degree of the components using grey relational analysis theory;

[0097] Step S105: Construct the inference network association matrix based on the component state correlation degree, establish the uncertain inference Petri network, and calculate the confidence degree of component damage with uncertain state.

[0098] In step S101 of this implementation method, the disaster-causing mechanism of components under extreme disasters is analyzed, and based on this mechanism, the causes of component damage are obtained, and a set of component disaster-causing condition attributes is established, specifically including:

[0099] Step S101-1: Analyze the mechanism of component damage during typhoon weather.

[0100] In typhoon disasters, wind speed is the dominant non-electrical factor leading to successive failures of power grid equipment. The key to assessing the impact of typhoon disasters on the power grid lies in establishing a mapping relationship between wind speed and the probability of power equipment failure, i.e., constructing a spatiotemporal failure probability model for the equipment. For wind-induced tripping caused by typhoons, the event triggering mechanism is that the wind deflection angle under strong winds exceeds a critical threshold, leading to air breakdown discharge and causing a tripping accident.

[0101] like Figure 2 As shown, for suspension insulator strings and jumpers with jumper insulator strings installed, the wind deflection angle of the insulator string can be calculated using a rigid straight rod model, expressed as:

[0102] (1);

[0103] in, i I Indicates the wind deflection angle of the insulator string; or The pulsation coefficient; G and W These represent the self-gravity loads of the insulator string and the conductor, respectively. G H and W H These represent the wind loads acting horizontally on the insulator string and the conductor, respectively. W Z The gravity load of the bottom suspension weight. Indicates the horizontal offset distance. Represents the distance in the vertical direction.

[0104] For jumpers without jumper insulator strings, wind deflection angle i 2 is represented as:

[0105] (2);

[0106] in, g 1 and g 2 represents the conductor's self-weight load and wind pressure load, respectively.

[0107] Compared to wind-induced tripping events, which are essentially electrical phenomena caused by insulation breakdown, tower collapses and line breaks occur because the wind load on the towers and lines under strong winds exceeds the design strength of the materials. From the perspective of the structural strength of the materials used in the lines and towers, the failure probability of the lines and towers is expressed as:

[0108] (3);

[0109] (4);

[0110] in, PL and P T These represent the failure probabilities of the line and the tower, respectively. s g and s t These represent the stresses on the line cross section and the angle steel of the tower, respectively. m L and d L These represent the mean and standard deviation of the conductor's strength, respectively. m T and d T These represent the mean and standard deviation of the strength of the tower steel, respectively. Represents stress.

[0111] Whether it's tower collapse, line breakage, or wind-induced tripping events, the models are all constructed by analyzing the mechanisms of these events under strong winds; these are called mechanistic analysis models. In addition, there are classic failure probability models that are widely accepted and used. These models are not derived from the mechanistic level but are empirical models fitted from large amounts of data; they are called statistical analysis models. For example, exponential models are generally used to characterize the failure rate of transmission lines and towers under strong winds during typhoons, such as... Figure 3 As shown (design wind speed is 30m / s, model parameters) The value of is 0.2), and the failure rate calculation formula is expressed as follows:

[0112] (5);

[0113] (6);

[0114] in, l ( t )express t Failure rate at any time (including safe operating zone, damage alarm zone and equipment failure zone); P ( t ) indicates time t Failure probability within; v ( t )express t Wind speed at any given time; v d,tower Indicates the design wind speed of the tower (can be replaced with) v d,line (representing the design wind speed of the line). k These are model parameters; represent The efficiency of failure at any moment.

[0115] Step S101-2: Analyze the disaster mechanism of components under extreme cold weather.

[0116] The dominant non-electrical factor causing power grid damage in extreme cold weather is the thickness of ice cover. The ways in which ice and snow disasters affect power grid safety can be summarized as ice galloping of transmission lines, ice flashover of insulators, and overload damage to power equipment.

[0117] Driven by wind, transmission lines with uneven ice accumulation are prone to self-excited vibration. This vibration is characterized by low frequency and high amplitude, and is known as transmission line ice galloping. Galloping usually occurs in winter and is characterized by high destructiveness and long duration. It may cause damage to hardware and grounding wires, and even lead to serious accidents such as line breaks, seriously affecting the safe and stable operation of the power grid.

[0118] Den Hartog's vertical dance mechanism can be represented as follows:

[0119] (7);

[0120] in, C L and C D These are the aerodynamic lift and drag coefficients of the conductor, respectively. α The angle of attack of the conductor is eccentrically iced and facing the wind.

[0121] The expression for the Nigol torsional dance mechanism is:

[0122] (8);

[0123] in, i k and oh k They are wires k The amplitude and angular frequency of the first-order vibration; v The horizontal wind speed is perpendicular to the direction of the line.

[0124] If equation (7) or equation (8) holds true in a certain simulation, it is considered that a line galloping event has occurred. N total In this experiment, the number of line galloping events was obtained as follows: N The statistical probability of line galloping can then be expressed as:

[0125] (9);

[0126] Ice accretion on the surface of insulators contains a large amount of conductive impurities. When the temperature rises and the ice melts, a highly conductive water film forms on the surface, reducing the flashover voltage of the insulator and making it highly susceptible to ice-induced flashover events. The key to analyzing the mechanism of insulator ice flashover lies in establishing a model relating flashover voltage to icing conditions, thereby obtaining a probabilistic model for insulator ice flashover discharge. After comprehensively considering relevant information such as insulator icing conditions, insulator type, and meteorological conditions, the formula for calculating the ice flashover voltage can be expressed as follows:

[0127] (10);

[0128] in: U i This is the flashover voltage for icing. r SDD For insulator salt density; C and d These are constants related to the insulator type and the number of discs, respectively. b The pollution characteristic coefficient; H ins Indicates altitude; W ins Indicates the weight of the ice layer; n ins and r These are the influence coefficients of altitude and icing amount, respectively.

[0129] The icing load in equation (10) is calculated using the following formula:

[0130] (11);

[0131] in: r i This refers to the density of the ice layer. g It is the acceleration due to gravity; R This refers to the thickness of the ice layer. D The outer diameter of the conductor; L This represents the length of the conductor.

[0132] When the actual operating voltage U < U i When the insulator is iced, the flashover probability is constant; while when U ≥ U i At that time, the probability of insulator flashover due to icing can be expressed by an exponential function as follows:

[0133] (12);

[0134] in: k and g These are the model coefficients.

[0135] Ice accumulation increases the load on transmission lines and towers, leading to increased stress on the tower and line structures. When the ice weight reaches a critical point, it can cause tower collapse and line breakage. The forced outage rate of components under icing loads is generally represented by an exponential function:

[0136] (13);

[0137] in, a These are the model coefficients; I ( t )for t The ice load on the line at any given time; ζ is the damping coefficient.

[0138] t The amount of ice carried by a line at any given time can be considered as the icing rate. v i ( t The integral of ) and the icing rate are related to the precipitation rate in the area where the line is located. r ( t Related to ) therefore I ( t This can be represented as follows:

[0139] (14);

[0140] in, r This refers to the density of the ice layer. k These are the model coefficients; v ( t )for t Wind speed along the route at any given time; for t The precipitation rate at the location of the route at any given time; for t The rate at which ice forms.

[0141] Step S101-3: Analysis of the disaster-causing mechanisms of components under typhoons and extreme cold weather reveals that component damage under extreme weather conditions is primarily influenced by both the component's own characteristics and external environmental factors. From the component's perspective, the construction intensity of transmission lines, towers, and substation outgoing lines affects the component's disaster resistance, while electrical factors such as power flow and voltage affect the component's failure probability. From the perspective of the external environment, factors such as wind speed, temperature, and humidity directly affect the intensity of the disaster. Based on this, a set of component disaster-causing attributes is established as shown in Table 1.

[0142] Table 1: Set of Disaster-Causing Attributes for Components

[0143]

[0144] In step S102 of this implementation, attribute reduction is performed on the component disaster-causing condition attribute set based on historical accumulated data. Specifically, this includes:

[0145] Step S102-1: Let the disaster-causing condition attribute set be... C Decision-makers select a subset of representative attributes based on their experience. C 1, C 2, …, C N .

[0146] Step S102-2: Calculate the knowledge granularity of the disaster-causing condition attribute set and attribute subset, and calculate the knowledge granularity of the disaster-causing condition attribute set and attribute subset.

[0147] Step S102-3: If there exists an attribute subset with a difference less than the threshold value, select the attribute subset with the smallest difference as the reduced disaster-causing attribute set; otherwise, return to step S102-1 to reselect a representative attribute subset, represented as:

[0148] (15);

[0149] in, GP U ( C ) is an attribute set C Knowledge granularity; U The domain is the set of all objects and their properties. U | indicates the number of objects in the universe of discourse; X i For knowledge, i.e., attribute set C The Knowledge is generally obtained by classifying it according to conditions and attributes. N C For attribute set C The number of knowledge points obtained through classification; e This is the threshold value for attribute reduction; Subset of attributes C Knowledge granularity of 1; For attribute subset C Number of knowledge points obtained from 1 category; Subset of attributes C 1 of One piece of knowledge; Representing the The number of objects of knowledge. Representing the The number of objects containing knowledge.

[0150] In step S103 of this implementation, information entropy is used to fill in the missing attribute interval values ​​of the disaster-causing condition attribute set to obtain a complete set of component disaster-causing condition attributes, supporting the calculation of the state correlation degree of power grid components. To reduce the length of the filled attribute intervals, an information filling method based on information entropy is adopted. First, the original system is split into multiple independent subsystems with only a single condition attribute, and the information entropy of each subsystem is calculated. Then, an object with a small interval is added to each subsystem, and the interval length is gradually increased, and the system information entropy is calculated until it is close to the original information entropy of the subsystem.

[0151] In step S104 of this implementation, the correlation degree of component states is calculated based on grey relational analysis theory. Since components are affected by the same extreme disaster and are located in the same power grid, their states are correlated. The disaster-causing condition attribute of a component is a mapping quantity of its state, and the relationship between these mapping quantities reflects the relationship between the component states. Based on the axiom of grey relational analysis, the Deng's grey relational degree between two components is calculated. Assume the attribute sequence of a certain component state is... A i =( a i (0),…, a i ( k ),…, a i ( n For the calculation of component state correlation degree, the improved Deng's grey correlation degree calculation formula is shown in equations (16) and (17).

[0152] The disaster-causing condition attributes of components are represented by interval values. The Deng's grey relational degree of the upper and lower bounds of the interval values ​​is calculated separately, and the average value is taken as the component state relational degree.

[0153] (16);

[0154] (17);

[0155] in, for A i and A j Grey relational degree; for A i and A j In attributes k The correlation coefficient; Indicates the attributes of disaster-causing conditions k The extent of its impact is obtained through mechanistic analysis; The resolution coefficient; Ai The attribute sequence representing the i-th element. A j The attribute sequence representing the j-th element. Representing the The element in the first The specific values ​​of each disaster-causing condition attribute. Representing the The element in the first The specific values ​​of each disaster-causing condition attribute. The total number of attributes representing disaster-causing conditions; This represents the scope of all components and all disaster-causing condition attributes, the first The first component and the second The element in the first absolute difference of each attribute The global minimum value; This represents the scope of all components and all disaster-causing condition attributes, the first The first component and the second The element in the first absolute difference of each attribute The global maximum value.

[0156] In step S105 of this implementation, a reasoning network correlation matrix is ​​constructed based on the component state correlation degree, an uncertain reasoning Petri network is established, and the confidence of component damage with uncertain reasoning state is determined. Specifically, this includes:

[0157] S105-1: Constructing the inference network association matrix based on component state correlation. The component state correlation is used as an element of the association matrix, which is further decomposed into association matrices between components with unknown states and known damaged components, and between components with unknown states and known intact components. Based on the electrical distance between different components and the component type, the association matrix between components with unknown states and known damaged components is further decomposed into... R ns , R nn , R fs as well as R fn Four categories, among which R ns This represents an association matrix that represents components that are electrically close and of the same type. R nn This represents an association matrix that represents components that are electrically close but different in type. R fs This represents an association matrix that represents components that are electrically distant but of the same type. R fnThis represents the correlation matrix for components with large electrical distances and different component types. Similarly, the correlation matrix between components with unknown states and components known to be in good condition is decomposed into... R ns ′，R nn ′，R fs ′ as well as R fn ′ Four categories.

[0158] S105-2: A Petri network for inference based on uncertainties is established using the correlation matrix obtained from decomposition, and the confidence level of damage to components with uncertain states is calculated. To fully utilize the known state information of damaged and intact components, the Petri network is divided into two parts: inference based on damaged components and inference based on intact components. The results from both parts are then integrated, as follows: Figure 4 As shown, transitions T1 to T5 represent the calculation process of damage confidence. Matrix calculations can quickly obtain the damage confidence of each unknown component in each state. In the upper path, the state of the known damaged component is projected onto the unknown component and iteratively propagated to form damage evidence. In the lower path, the state of the known intact component is projected and propagated to form intact evidence. The two paths are finally merged at the end of the library, and a weighted normalization formula is used to obtain the damage confidence of the unknown component, realizing parallel reasoning based on damaged and intact components.

[0159] More specifically, let the total number of system components be N The component states are divided into known damage sets. D (size is) n d ), known complete set H (size is) n h ) and the set of unknown states U (size is) n d The overall correlation matrix obtained based on grey relational analysis. A ∈ R N×N For A, set by row and column U , D , H Rearranged to:

[0160] (18);

[0161] in, A ud Indicates an unknown set of components U With known damaged component set D The correlation strength submatrix between them, the matrix in which the first... iLine number j Column elements a ( uh ) ij Reflects unknown components i With known damaged components j The degree of state correlation between them is represented by the correlation coefficient obtained from grey relational analysis and normalized to ensure that the influence between different components is on the same scale. A uh Indicates an unknown set of components U With known perfect sets H The correlation strength submatrix between them, the matrix in which the first... i Line number j Column elements a ( ud ) ij Characterized unknown components i and intact components j The degree of similarity between them. A ud This indicates that the interrelationships between components are unknown, and its elements... a ( oh ) ij Characterizing unknown components i With known damaged components j The degree of coupling between them in terms of state propagation. This represents the submatrix representing the correlation strength within the set of unknown components U; This represents the correlation strength submatrix within the known set of damaged components D; This represents the correlation strength submatrix within the known intact set H; Represents the set of known damaged components. D With unknown component set U The correlation strength submatrix between them; Represents the set of known damaged components D The correlation strength submatrix between the known intact set H; The submatrix representing the correlation strength between the known intact set H and the unknown set of components U; This represents the correlation strength submatrix between the known intact set H and the known damaged set D.

[0162] To take into account the four types of sub-matrices decomposed by electrical distance and component type in S105-1 R ns ，R nn ，R fs and R fn , A ud It can be expressed in weighted sum form as:

[0163] (19);

[0164] Among them, weight α x ≥0 and You x =1, where x represents the different types of the submatrix. The weights of the correlation matrix representing unknown state components and known damaged components that are electrically close and of the same type. The weights of the correlation matrix representing unknown state components and known damaged components that are electrically close and of different component types; Time represents the weight of the correlation matrix between unknown components and known damaged components that are electrically distant and of the same type; The weights of the correlation matrix representing the relationship between unknown components and known damaged components that are electrically distant and of different types; , , , Each is a set of unknown components. U With known damaged component set D The correlation strength submatrix between A ud The decomposition results.

[0165] Similarly, the correlation submatrix A uh It can also be expressed in a weighted form as follows:

[0166] (20);

[0167] in, , , , Each is a set of unknown components. U The submatrix of correlation strength between the known intact set H and the known intact set H A uh The decomposition results.

[0168] Based on this, establish such Figure 4 The Petri network shown is for uncertain reasoning. In the Petri network, "places" are used to represent the state information of different categories of components, transitions are used to represent the state propagation and reasoning calculation process, and the arc weights are determined by the elements of the aforementioned submatrix. This Petri network contains two parallel reasoning paths: a reasoning path based on damaged components and a reasoning path based on intact components. Finally, the damage confidence of uncertain components is calculated by fusing transitions.

[0169] Figure 4 In the middle, the warehouseP 1 and P 2. Input a known set of damaged components and their corresponding sub-incident matrices, and then perform a transition... T 1. Project to warehouse P 6. Forming initial evidence of damage to unknown components. :

[0170] (twenty one);

[0171] in, d d The state vector of the known damaged component.

[0172] Subsequently, the warehouse P 5 and P 6 Input to Transition T 3. Evidence of damage spreads in unidentified component subnets:

[0173] (twenty two);

[0174] in, l ∈(0,1), where ∈ is the propagation attenuation coefficient, and the propagation result is... Storage location P 9. To form evidence of accumulated damage. Represents the identity matrix.

[0175] warehouse P 3 and P 4. Input a known set of intact components and its corresponding sub-relational matrix, and then perform transitions. T 2. Projected to warehouse P 7. Obtain initial intact evidence :

[0176] (twenty three);

[0177] in, d h The state vector of the known damaged component.

[0178] Subsequently, the warehouse P 7, P 8 Enter T 4. Enable the propagation of intact evidence in unknown component subnets:

[0179] (twenty four);

[0180] Dissemination results Storage location P 10 This forms a complete set of accumulated evidence.

[0181] Ultimately, the warehouseP 9 and P 10 Evidence input to change T 5. In the fusion library P 11 The failure confidence level of the unknown component is calculated. :

[0182] (25);

[0183] in, oh d , oh h The weights for damaged evidence and intact evidence are respectively. e To prevent small positive numbers with a denominator of zero, the element-wise form is:

[0184] (26);

[0185] in, Representing the Confidence of failure for an unknown component Representing the Strength of evidence of damage to an unknown component No. The strength of evidence of damage corresponding to an unknown component, when C u,i When the value is close to 1, it indicates that the component is unknown. The probability of damage is relatively high; when C u,i A value close to 0 indicates a greater likelihood that the substance remains intact.

[0186] Finally, a judgment of "damaged / intact / uncertain" is given based on the threshold:

[0187] (27);

[0188] in, and These represent the set threshold values.

[0189] Figure 5 A component status identification system during power system fault recovery is shown, comprising:

[0190] The attribute set construction unit 501 is configured to: acquire the dominant non-electrical factors that cause component damage under disaster conditions, and establish a component disaster-causing condition attribute set based on the dominant non-electrical factors;

[0191] The attribute set reduction element 502 is configured to: determine multiple candidate attribute subsets based on the component disaster-causing condition attribute set, and reduce the component disaster-causing condition attribute set based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset.

[0192] The missing information filling unit 503 is configured to: fill the missing attribute intervals of the reduced disaster-causing condition attribute set with information entropy to obtain a complete disaster-causing condition attribute set.

[0193] The correlation calculation unit 504 is configured to: calculate the correlation degree of the component state in the complete disaster-causing condition attribute set;

[0194] The confidence inference unit 505 is configured to: construct an inference network association matrix based on the component state association degree, and determine the confidence degree of damage of components with unknown state based on the inference network association matrix.

[0195] It is understood that the aforementioned units can be individually or entirely merged into one or more other units, or some of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effects of the embodiments of the present invention. The aforementioned units are based on logical functional division. In practical applications, the function of one unit can be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of the present invention, the system may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented collaboratively by multiple units.

[0196] According to another embodiment of the present invention, the system of this embodiment can be constructed by running a computer program (including program code) capable of performing the steps involved in the corresponding method of the present invention on a general-purpose computing device, such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, loaded into the aforementioned computing device through the computer-readable recording medium, and run therein.

[0197] Figure 6 A computer device is shown, which includes a processor 601, a communication interface 602, and a computer-readable storage medium 603. The processor 601, communication interface 602, and computer-readable storage medium 603 can be connected via a bus or other means.

[0198] The communication interface 602 is used to receive and send data. The computer-readable storage medium 603 can be stored in the memory of the electronic device. The computer-readable storage medium 603 is used to store computer programs, which include program instructions. The processor 601 is used to execute the program instructions stored in the computer-readable storage medium 603.

[0199] The processor 601 is the computing and control core of an electronic device. It is suitable for implementing one or more instructions, specifically for loading and executing one or more instructions to achieve the corresponding method flow or corresponding function.

[0200] Processor 601 is configured to perform the following procedure:

[0201] Obtain the dominant non-electrical factors that cause component damage under disaster conditions, and establish a set of disaster-causing condition attributes for the components based on the dominant non-electrical factors;

[0202] Multiple candidate attribute subsets are determined based on the component disaster-causing condition attribute set. The component disaster-causing condition attribute set is reduced based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset.

[0203] Information entropy is used to fill in the missing attribute intervals of the reduced disaster-causing condition attribute set to obtain the complete disaster-causing condition attribute set.

[0204] Calculate the correlation degree of component states in the complete disaster-causing condition attribute set;

[0205] Based on the correlation degree of component state, an inference network correlation matrix is ​​constructed, and the confidence degree of damage of components with unknown state is determined according to the inference network correlation matrix.

[0206] This invention also provides a computer-readable storage medium, which is a memory device in an electronic device for storing programs and data. It is understood that the computer-readable storage medium here may include both built-in storage media in the electronic device and extended storage media supported by the electronic device. The computer-readable storage medium provides storage space for storing the processing system of the electronic device.

[0207] Furthermore, this storage space also contains one or more instructions suitable for loading and execution by the processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM memory or unstable memory, such as at least one disk storage device; optionally, it can also be at least one computer-readable storage medium located remotely from the aforementioned processor.

[0208] In one embodiment, the computer-readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer-readable storage medium to perform the following process:

[0209] Obtain the dominant non-electrical factors that cause component damage under disaster conditions, and establish a set of disaster-causing condition attributes for the components based on the dominant non-electrical factors;

[0210] Multiple candidate attribute subsets are determined based on the component disaster-causing condition attribute set. The component disaster-causing condition attribute set is reduced based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset.

[0211] Information entropy is used to fill in the missing attribute intervals of the reduced disaster-causing condition attribute set to obtain the complete disaster-causing condition attribute set.

[0212] Calculate the correlation degree of component states in the complete disaster-causing condition attribute set;

[0213] Based on the correlation degree of component state, an inference network correlation matrix is ​​constructed, and the confidence degree of damage of components with unknown state is determined according to the inference network correlation matrix.

[0214] The present invention also provides a computer program product or computer program comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the following process:

[0215] Obtain the dominant non-electrical factors that cause component damage under disaster conditions, and establish a set of disaster-causing condition attributes for the components based on the dominant non-electrical factors;

[0216] Multiple candidate attribute subsets are determined based on the component disaster-causing condition attribute set. The component disaster-causing condition attribute set is reduced based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset.

[0217] Information entropy is used to fill in the missing attribute intervals of the reduced disaster-causing condition attribute set to obtain the complete disaster-causing condition attribute set.

[0218] Calculate the correlation degree of component states in the complete disaster-causing condition attribute set;

[0219] Based on the correlation degree of component state, an inference network correlation matrix is ​​constructed, and the confidence degree of damage of components with unknown state is determined according to the inference network correlation matrix.

[0220] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can implement the described functions using different methods for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0221] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic cable, digital cable) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0222] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method of identifying element states during power system fault restoration, characterized by, Includes the following processes: Obtain the dominant non-electrical factors that cause component damage under disaster conditions, and establish a component disaster-causing condition attribute set based on the dominant non-electrical factors; Multiple candidate attribute subsets are determined based on the component disaster-causing condition attribute set. The component disaster-causing condition attribute set is reduced based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset. Information entropy is used to fill in the missing attribute intervals of the reduced disaster-causing condition attribute set to obtain the complete disaster-causing condition attribute set. Calculate the component state correlation degree in the complete disaster-causing condition attribute set, construct the inference network correlation matrix based on the component state correlation degree, and determine the confidence degree of damage of components with unknown state based on the inference network correlation matrix; Based on the component disaster-causing condition attribute set, multiple candidate attribute subsets are determined. Then, based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset, the component disaster-causing condition attribute set is reduced, including: Select a subset of multiple candidate attributes from the disaster-causing condition attribute set; Calculate the knowledge granularity of the disaster-causing condition attribute set and each candidate attribute subset respectively; Calculate the difference between the knowledge granularity of each candidate attribute subset and the knowledge granularity of the entire disaster-causing condition attribute set; Select a subset of candidate attributes whose knowledge granularity difference is less than a set threshold as the reduced set of disaster-causing attribute properties for the components; Calculate the difference between the knowledge granularity of the candidate attribute subset and the knowledge granularity of the entire disaster-causing condition attribute set, including: ； in, GP U ( C ) is an attribute set C Knowledge granularity; U For the domain, | U | indicates the number of objects in the universe of discourse; X i For attribute set C The One piece of knowledge; N C For attribute set C The number of knowledge points obtained through classification; Subset of attributes C Knowledge granularity of 1; For attribute subset C Number of knowledge points obtained from 1 category; Subset of attributes C 1 of j One piece of knowledge; Representing the The number of objects containing knowledge; Representing the j The number of objects containing knowledge; Based on the component state correlation, an inference network correlation matrix is ​​constructed. The confidence level of component damage with an uncertain state is determined based on the inference network correlation matrix, including: Using the calculated component state correlation degree as elements, construct an correlation matrix between components with unknown states and components with known states; The correlation matrix is ​​decomposed into multiple sub-correlation matrices based on electrical distance and component type. Based on the sub-correlation matrix obtained from the decomposition, Petri network models based on the path of a state-known damaged component and the path of a state-known intact component are constructed respectively. Confidence propagation calculations are performed in the Petri network model. The reasoning results of the known damaged component path and the known intact component path are integrated and normalized to obtain the damage confidence of the component whose state is unknown.

2. The component status identification method during power system fault recovery as described in claim 1, characterized in that, The dominant non-electrical factors include both component intrinsic properties and external environmental properties: The intrinsic properties of the components include the insulator string's own gravity load, the conductor's own gravity load, the tower steel strength, the conductor's outer diameter, and the conductor's length; the extrinsic environmental properties include wind speed, precipitation rate, temperature, humidity, icing thickness, altitude, power flow, and voltage.

3. The component status identification method during power system fault recovery as described in claim 1, characterized in that, Information entropy is used to fill in the missing attribute intervals of the reduced disaster-causing condition attribute set to obtain the complete disaster-causing condition attribute set, including: The reduced disaster-causing condition attribute set is split into multiple sub-unit sets containing only a single condition attribute; Calculate the original information entropy of each of the aforementioned sub-unit sets; An object is inserted into each of the said sub-unit sets, gradually increasing the length of the missing attribute interval; Calculate the information entropy after each interval expansion; Select the interval corresponding to the information entropy closest to the original information entropy as the optimal filling interval; Fill each subset unit according to its optimal filling interval, and then combine them to form a complete disaster-causing condition attribute set.

4. The component status identification method during power system fault recovery as described in claim 1, characterized in that, Calculate the component state correlation degree in the complete disaster-causing condition attribute set, including: Based on the complete set of disaster-causing condition attributes, an attribute sequence is formed for each component; The state correlation degree between components is calculated using Deng's grey relational degree formula.

5. The component status identification method during power system fault recovery as described in claim 4, characterized in that, For interval-valued attributes, calculate the upper bound grey relational degree and the lower bound grey relational degree of the interval-valued attributes respectively, and take the average of the upper bound grey relational degree and the lower bound grey relational degree as the final state relational degree.

6. The component status identification method during power system fault recovery as described in claim 4, characterized in that, Deng's grey relational formula is: ； ； wherein, is A i and A j the grey correlation degree of is A i and A j the correlation coefficient of attribute k represents the influence degree of disaster-causing condition attribute k is the resolution coefficient; A i represents the attribute sequence of the th element, A j represents the attribute sequence of the j th element, represents the specific value of the th element on the th disaster-causing condition attribute, represents the specific value of the j th element on the th disaster-causing condition attribute, represents the total number of disaster-causing condition attributes; represents the global minimum value of the absolute difference of the j th element and the th element on the th attribute in the range of all elements and all disaster-causing condition attributes; represents the global maximum value of the absolute difference of the j th element and the th element on the th attribute in the range of all elements and all disaster-causing condition attributes.​​ 7. An electrical power system fault restoration period component state identification system characterized by, include: The attribute set construction unit is configured to: acquire the dominant non-electrical factors that cause component damage under disaster conditions, and establish a component disaster-causing condition attribute set based on the dominant non-electrical factors; The attribute set reduction element is configured as follows: determine multiple candidate attribute subsets based on the component disaster-causing condition attribute set, and reduce the component disaster-causing condition attribute set based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset. The missing information filling unit is configured to fill the missing attribute intervals of the reduced disaster-causing condition attribute set with information entropy to obtain the complete disaster-causing condition attribute set. The correlation calculation unit is configured to calculate the correlation degree of the component states in the complete disaster-causing condition attribute set; The confidence inference unit is configured to: construct an inference network correlation matrix based on the component state correlation, and determine the confidence level of damage to components with unknown states based on the inference network correlation matrix; In the attribute set reduction simplification element, multiple candidate attribute subsets are determined based on the component disaster-causing condition attribute set. The component disaster-causing condition attribute set is then reduced based on the knowledge granularity of the component disaster-causing condition attribute set and each candidate attribute subset, including: Select a subset of multiple candidate attributes from the disaster-causing condition attribute set; Calculate the knowledge granularity of the disaster-causing condition attribute set and each candidate attribute subset respectively; Calculate the difference between the knowledge granularity of each candidate attribute subset and the knowledge granularity of the entire disaster-causing condition attribute set; Select a subset of candidate attributes whose knowledge granularity difference is less than a set threshold as the reduced set of disaster-causing attribute properties for the components; Calculate the difference between the knowledge granularity of the candidate attribute subset and the knowledge granularity of the entire disaster-causing condition attribute set, including: ； in, GP U ( C ) is an attribute set C Knowledge granularity; U For the domain, | U | indicates the number of objects in the universe of discourse; X i For attribute set C The One piece of knowledge; N C For attribute set C The number of knowledge points obtained through classification; Subset of attributes C Knowledge granularity of 1; For attribute subset C Number of knowledge points obtained from 1 category; Subset of attributes C 1 of j One piece of knowledge; Representing the The number of objects containing knowledge; Representing the j The number of objects containing knowledge; In the confidence inference unit, an inference network association matrix is ​​constructed based on the component state correlation degree. The confidence degree of damage to components with unknown states is determined based on the inference network association matrix, including: Using the calculated component state correlation degree as elements, construct an correlation matrix between components with unknown states and components with known states; The correlation matrix is ​​decomposed into multiple sub-correlation matrices based on electrical distance and component type. Based on the sub-correlation matrix obtained from the decomposition, Petri network models based on the path of a state-known damaged component and the path of a state-known intact component are constructed respectively. Confidence propagation calculations are performed in the Petri network model. The reasoning results of the known damaged component path and the known intact component path are integrated and normalized to obtain the damage confidence of the component whose state is unknown.

8. The component status identification system during power system fault recovery as described in claim 7, characterized in that, In the attribute set construction unit, the dominant non-electrical factors include both component-specific attributes and external environmental attributes: The intrinsic properties of the components include the insulator string's own gravity load, the conductor's own gravity load, the tower steel strength, the conductor's outer diameter, and the conductor's length; the extrinsic environmental properties include wind speed, precipitation rate, temperature, humidity, icing thickness, altitude, power flow, and voltage.

9. The component status identification system during power system fault recovery as described in claim 7, characterized in that, In the missing information filling unit, information is filled into the missing attribute intervals of the reduced disaster-causing condition attribute set based on information entropy to obtain the complete disaster-causing condition attribute set, including: The reduced disaster-causing condition attribute set is split into multiple sub-unit sets containing only a single condition attribute; Calculate the original information entropy of each sub-unit set; Insert an object into each sub-unit set, gradually increasing the length of the missing attribute interval; Calculate the information entropy after each interval expansion; Select the interval corresponding to the information entropy closest to the original information entropy as the optimal filling interval; Fill each subset unit according to its optimal filling interval, and then combine them to form a complete disaster-causing condition attribute set.

10. The component status identification system during power system fault recovery as described in claim 7, characterized in that, In the correlation calculation unit, the correlation degree of component states in the complete disaster-causing condition attribute set is calculated, including: Based on the complete set of disaster-causing condition attributes, an attribute sequence is formed for each component; The state correlation degree between components is calculated using Deng's grey relational degree formula. For interval-valued attributes, calculate the upper bound grey relational degree and the lower bound grey relational degree of the interval-valued attributes respectively, and take the average of the upper bound grey relational degree and the lower bound grey relational degree as the final state relational degree.

11. A computer device, characterized in that, include: Processor and computer-readable storage media; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the component status identification method during power system fault recovery as described in any one of claims 1 to 6.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1 to 6, the method for identifying the status of components during power system fault recovery.

13. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the component status identification method during power system fault recovery as described in any one of claims 1 to 6.

Images