Grid fault rapid diagnosis method based on artificial intelligence technology

Abstract

The invention discloses a power grid fault rapid diagnosis method based on an artificial intelligence technology, and relates to the field of power grid fault diagnosis, and the method comprises the steps: S1, carrying out the feature extraction of a power grid fault based on quantum calculation; s2, constructing a power grid knowledge graph; s3, mapping entity nodes in the power grid knowledge graph to a quantum state according to key attribute information of the entity nodes, and designing quantum gate operation for encoding according to a relationship among the entity nodes in the power grid knowledge graph; s4, performing fault diagnosis; and S5, performing feedback and optimization based on the power grid knowledge graph. According to the rapid power grid fault diagnosis method based on the artificial intelligence technology, the high efficiency of the quantum calculation in the aspects of data processing and feature extraction and the advantages of the knowledge graph in the aspects of knowledge representation and reasoning are brought into full play through the power grid fault diagnosis method based on fusion of the quantum calculation and the knowledge graph; rapid and accurate diagnosis of power grid faults can be realized, and a guarantee is provided for safe and stable operation of a power grid.

Description

Technical Field

[0001] The present invention relates to the field of power grid fault diagnosis, and in particular to a power grid fault rapid diagnosis method based on artificial intelligence technology. Background Art

[0002] The whole composed of substations of various voltages and transmission and distribution lines in the power system is called the power grid, which includes three units: substation, transmission and distribution. The task of the power grid is to transmit and distribute electric energy and change the voltage.

[0003] When diagnosing a power grid fault, only the currently faulty device can be identified. However, faults have multiple causal relationships, and there may be multiple factors that jointly cause the fault. Therefore, when diagnosing the cause of the fault, a detailed fault conclusion cannot be diagnosed, resulting in the inability to guarantee its accuracy and reliability.

[0004] Therefore, it is necessary to propose a rapid diagnosis method for power grid faults based on artificial intelligence technology to solve the above problems. Summary of the invention

[0005] The main purpose of the present invention is to provide a method for rapid diagnosis of power grid faults based on artificial intelligence technology, which can effectively solve the problems in the background technology.

[0006] To achieve the above object, the technical solution adopted by the present invention is: A method for rapid diagnosis of power grid faults based on artificial intelligence technology includes the following steps: S1: Extract grid fault features based on quantum computing, map the voltage, current, power, and device status information in the grid to quantum bits according to specific coding rules to extract the features of grid faults; S2: Construction of power grid knowledge graph: collect multi-source data from the power grid, integrate and process it, perform entity recognition and relationship extraction on it, and build the power grid knowledge graph based on it; S3: Map the entity nodes in the power grid knowledge graph to quantum states based on their key attribute information, and design quantum gate operations to encode the relationships between the entity nodes in the power grid knowledge graph; S4: Fault diagnosis. When a power grid fault occurs, the Grover algorithm is used to search for nodes matching the fault characteristics in the quantum state space of the nodes encoding the power grid knowledge graph. The fault characteristics are converted into quantum state constraints. Based on the iterative operation of the Grover algorithm on the quantum state, the fault node is found. After the faulty node is identified, the quantum annealing algorithm is used to find the relevant fault propagation path and fault root cause in the power grid knowledge graph to diagnose the power grid fault. S5: Feedback and optimization based on the power grid knowledge graph, using the structural information and entity relationships in the power grid knowledge graph to provide optimization basis for the number of quantum bits and connection methods in the S4 algorithm; A large number of historical fault cases and fault handling experience in the power grid knowledge graph are used to update the algorithm parameters in the S4 algorithm.

[0007] Preferably, in S1, the extraction of power grid fault features based on quantum computing specifically includes the following operation steps: S101: Quantum state construction and data encoding, determine the entities and relationships used to describe the state of the power grid, including the voltage amplitude, current amplitude, and phase angle of each node, record the number of power grid monitoring points as N, each monitoring point has M parameters, divide the amplitude range of voltage amplitude, current amplitude, and phase angle into K intervals, set it as the encoding rule, use quantum bits to represent the value state of each parameter, and each parameter uses log2K quantum bits to encode its value. For all monitoring point parameters of the entire power grid, N*M*log2K quantum bits are required to construct the quantum state, and each quantum state corresponds to a parameter combination state of the power grid; S102: Quantum gate operation and feature mapping, using Hadamard gate to initialize quantum bits to put them in a superposition state, designing a special quantum gate combination to simulate the impact mechanism of power grid faults on parameters, introducing entanglement gate operation, entangling quantum bits corresponding to different monitoring points and different parameters, to capture the correlation and coupling between power grid parameters, and extract comprehensive fault feature information; S103: Quantum measurement and feature extraction, measuring the quantum state, and selecting the basis vector for measurement is based on the fault feature to be extracted.

[0008] Preferably, in S101, the parameter combination state of the power grid includes a voltage-related parameter combination state, a current-related parameter combination state, and a power-related parameter combination state, wherein the voltage-related parameter combination state includes a voltage amplitude combination, a voltage phase combination, and a frequency combination; the current-related parameter combination state includes a current amplitude combination and a current phase combination; and the power-related parameter combination state includes an active power combination and a reactive power combination.

[0009] Preferably, in S102, the quantum gate combination is based on the relationship between the entity nodes in the power grid knowledge graph, including a quantum gate combination for simulating voltage phase angle fault, a quantum gate combination for simulating current mutation fault, and a quantum gate combination for simulating the influence of compound fault on power parameters, wherein the quantum gate combination for simulating voltage phase angle fault: a phase rotation gate is used to rotate the phase of the quantum bit, and the quantum state of the quantum bit is recorded as: ; The phase rotation gate is: ;in is the quantum state of the qubit, is the angle of the phase rotation door, and is plural and satisfies , Indicates the angle, and is the exponential form of a complex number, is an imaginary unit, and ,When a phase angle offset fault occurs in the power grid, multiple phase rotation gates are applied to simulate the change of the phase angle, and the characteristics of the phase angle change are extracted and recorded; Quantum gate combination for simulating current mutation fault: one qubit is used as a control bit to indicate whether a fault occurs, and Indicates no fault, The qubit represents the fault, and the other qubit represents the current amplitude bit, whose states include occurrence and non-occurrence. The qubit representing the current amplitude is acted on by the CNOT gate, and the single-bit rotation gate is combined to change the state of the current amplitude qubit to simulate the mutation of the current amplitude. The characteristics related to the current mutation, including the increase of the current amplitude and the stable state after the mutation, are extracted and recorded as features; the quantum gate combination used to simulate the influence of compound faults on power parameters: two qubits are used as control bits to indicate whether different types of faults occur at the same time, and one fault corresponds to one control bit. Indicates no fault. , , It represents different fault combinations, and another quantum bit is used to represent the state of the power parameter. When a specific fault combination occurs, the Toffoli gate acts on the quantum bit representing the power parameter, and then combines with the S phase gate to change the state of the power parameter quantum bit, simulating the impact of the composite fault on the power, extracting the change of the power factor, the fluctuation characteristics of the active power and the reactive power and recording them. Preferably, in S2, the construction of the power grid knowledge graph specifically includes the following steps: S201: Collect data from multiple business systems of the power grid, including the data acquisition and monitoring control system of the power grid to obtain the real-time voltage, current and power operation data of the power grid; the energy management system to collect the topology of the power grid and the operation status information of the equipment; the power outage management system to obtain the data related to the power outage event, including the time, scope and cause of the power outage; the equipment asset information and maintenance records in the equipment maintenance management system; integrate the external environment data of the power grid, including meteorological data and geographic information data, and clean and standardize the collected data; S202: Classify and label the cleaned and standardized data based on entity and relationship types; S203: Use a bidirectional long short-term memory network combined with a conditional random field model to identify entities in the power grid data; extract the connection relationship in the power grid topology structure through predefined rules, and use a convolutional neural network to extract the causal relationship between faults and equipment; S204: Design of the power grid knowledge graph architecture, designing the architecture of the power grid knowledge graph, including defining entity types, relationship types, and attribute types. Entity types cover various equipment, locations, personnel in the power grid and their names, locations, and capacity attributes; relationship types include topological connection relationships, electrical parameter association relationships, operation and maintenance relationships, and fault causal relationships; attribute types correspond to various feature descriptions of entities; S205: Construction and update of the power grid knowledge graph, based on the results of entity recognition and relationship extraction, use the graph database to construct the power grid knowledge graph, use the identified entities as nodes, the extracted relationships as edges, and the attributes of the entities as the attribute information of the nodes to store in the graph database. During the operation of the power grid, as new data is continuously generated, the power grid knowledge graph is updated in real time.

[0010] Preferably, in S203, the predefined rules include topological structure connection relationship rules, electrical parameter association relationship rules, equipment operation and maintenance relationship rules, and fault association relationship rules, wherein the topological structure connection relationship rules include substation and transmission line connection rules: when the text contains the statement that substation [name] is connected to substation [name] through transmission line [name], three entities and the connection relationship between them are directly extracted; substation and transformer connection rules: when there are sentences such as substation [name] contains transformer [name] and transformer [name] is located in substation [name], the two entities of substation and transformer and the inclusion relationship between them are extracted, based on which the relevant entities and relationships can be extracted; transmission line and tower connection rules: when the text mentions that transmission line [name] is erected on tower [name] and tower [ [name] supports the transmission line [name], and the entities of the transmission line and the tower as well as the relationship between the erection and the support can be extracted, and the power grid knowledge graph relationship can be constructed based on this; the electrical parameter association relationship rules include the voltage, current, and resistance relationship rules based on Ohm's law: in the data description, when there is a description that according to Ohm's law, when the resistance [resistance value] remains unchanged, the voltage [voltage value] is proportional to the current [current value], the three entities of voltage, current, and resistance are extracted, and their parameters are also regarded as entities and the relationship between them according to Ohm's law; the power, voltage, and current relationship rules based on the power formula: when there is a data description that power [power value] is equal to voltage [voltage value] multiplied by current [current value], the three entities of power, voltage, and current and the relationship between them according to the power formula can be extracted, and the power grid knowledge graph relationship can be constructed based on this; Equipment operation and maintenance relationship rules include equipment and maintenance plan relationship rules: when the text contains the content that equipment [name] is planned to be maintained at [time], maintenance plan [number] involves equipment [name], and the maintenance time is [time], the two entities of equipment and maintenance plan and the planned maintenance relationship between them are extracted; equipment and maintenance personnel relationship rules: when there is a description that equipment [name] is maintained by maintenance personnel [name] and maintenance personnel [name] performs maintenance operations on equipment [name], the two entities of equipment and maintenance personnel and the responsible maintenance relationship between them are extracted, and the power grid knowledge graph relationship is constructed based on this; Fault association relationship rules include direct fault and equipment association rules: when there is content in the data that the device [name] has a [fault type] fault, the two entities of fault and device and the occurrence relationship between them are extracted; indirect fault and equipment association rules: when there is a statement that the device [name 1] fault causes the device [name 2] to have a [fault type] fault, the three entities of two devices and fault and the causing relationship between them are extracted, and the power grid knowledge graph relationship is constructed based on this.

[0011] Preferably, in S4, based on the iterative operation of the Grover algorithm on the quantum state, finding the faulty node includes the following operation steps: S401: Determine the total number of nodes in the power grid knowledge graph , according to the number of nodes ,need qubits to represent all possible node states, and all qubits are initialized to a uniform superposition state; S402: Determine node matching conditions based on detected fault characteristics and design Oracle functions , assuming is a base state and corresponds to a number in the power grid knowledge graph. If the fault feature matching condition is met, If not satisfied, then After operating the entire quantum state space, the Oracle function marks the quantum states corresponding to all nodes that meet the fault characteristics; S403: According to the principle of Grover algorithm, determine the number of iterations , the calculation formula is: ;in is the number of iterations, Indicates a round-down operation; S404: Constructing the Grover diffusion operator , apply Oracle functions to the initialized quantum state in turn and diffusion operator ,conduct iterations. After each iteration, the quantum state will change, and the probability amplitude of the quantum state corresponding to the node matching the fault feature will gradually increase. S405: After After iterations, the final quantum state is measured, and the measurement operation causes the quantum state to collapse to a ground state. The node corresponding to this base state is the node with the highest probability of matching the fault characteristics searched by the Grover algorithm. After obtaining the node with the highest probability, the node corresponding to the base state is converted into an actual node in the power grid knowledge graph. The actual node is the node related to the fault characteristics.

[0012] Preferably, in said S4, searching for relevant fault propagation paths and fault roots in the power grid knowledge graph using the quantum annealing algorithm comprises the following steps: S401: assigning a quantum bit to each node in the power grid knowledge graph to represent its state, assigning the quantum bit Indicates that the node is normal, and the quantum bit Indicates that the node is in a faulty or affected state; S402: Setting the objective function, including the energy item of the node failure state and the energy item corresponding to the connection relationship between nodes, wherein the energy item of the node failure state is , according to its prior probability of failure Set capability items, when the node In fault state , and its energy contribution formula is: ;in is the energy contribution of the current node, It is a coefficient related to the importance of the node and the reliability of fault detection; Energy items corresponding to the connection relationship between nodes: For nodes with connection relationship, The nodes with connection relationship are recorded as , according to the conditional probability of fault propagation between them Set the energy term, when the node and If both are in a fault state, the energy contribution formula of the connection relationship is: ;in is the energy term, It represents a node and The coupling coefficient of the tightness of the connection and the ease of fault propagation; The overall objective function is: ; is the overall objective function; Find the combination of quantum bit states that makes the energy of this objective function the lowest. The node state and connection relationship corresponding to this combination represent the fault propagation path and the root cause of the fault.

[0013] Compared with the prior art, the present invention provides a method for rapid diagnosis of power grid faults based on artificial intelligence technology, which has the following beneficial effects: 1. The method for rapid diagnosis of power grid faults based on artificial intelligence technology, a method for rapid diagnosis of power grid faults based on the fusion of quantum computing and knowledge graphs, gives full play to the high efficiency of quantum computing in data processing and feature extraction and the advantages of knowledge graphs in knowledge representation and reasoning, and can realize rapid and accurate diagnosis of power grid faults, providing strong guarantee for the safe and stable operation of the power grid.

[0014] 2. This rapid diagnosis method for power grid faults based on artificial intelligence technology can comprehensively and accurately integrate various knowledge information in power grid operation through the established power grid knowledge graph, and use the topological connection relationship and electrical parameter association relationship in the power grid knowledge graph to further screen out nodes closely related to fault characteristics. It can effectively find the fault propagation path and fault root cause in the power grid knowledge graph, and can quickly diagnose and locate the current power grid fault. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0016] In order to make the technical means, creative features, objectives and effects of the present invention easy to understand, the present invention is further described below in conjunction with specific implementation methods. Figure 1 As shown, a method for rapid diagnosis of power grid faults based on artificial intelligence technology includes the following steps: S1: Extract the characteristics of power grid faults based on quantum computing. Map the voltage, current, power, and device status information in the power grid to quantum bits according to specific coding rules to extract the characteristics of power grid faults. The specific steps include the following: S101: Quantum state construction and data encoding, determine the entities and relationships used to describe the state of the power grid, including the voltage amplitude, current amplitude, and phase angle of each node, record the number of power grid monitoring points as N, each monitoring point has M parameters, divide the amplitude range of the voltage amplitude, current amplitude, and phase angle into K intervals, set it as the encoding rule, use quantum bits to represent the value state of each parameter, and each parameter uses log2K quantum bits to encode its value. For all monitoring point parameters of the entire power grid, N*M*log2K quantum bits are required to construct the quantum state, and each quantum state corresponds to a parameter combination state of the power grid; the parameter combination state of the power grid includes voltage-related parameter combination state, current-related parameter combination state, and power-related parameter combination state, among which the voltage-related parameter combination state includes voltage amplitude combination, voltage phase combination, and frequency combination; the current-related parameter combination state includes current amplitude combination and current phase combination; the power-related parameter combination state includes active power combination and reactive power combination; S102: Quantum gate operation and feature mapping. Use Hadamard gate to initialize quantum bits to put them in a superposition state. Design a special quantum gate combination to simulate the impact mechanism of power grid faults on parameters. Introduce entanglement gate operation to entangle quantum bits corresponding to different monitoring points and different parameters to capture the correlation and coupling between power grid parameters, which is used to extract comprehensive fault feature information. The quantum gate combination is based on the relationship between the entity nodes in the power grid knowledge graph, including the quantum gate combination for simulating voltage phase angle fault, the quantum gate combination for simulating current mutation fault, and the quantum gate combination for simulating the influence of compound fault on power parameters. The quantum gate combination for simulating voltage phase angle fault: the phase rotation gate is used to rotate the phase of the quantum bit, and the quantum state of the quantum bit is recorded as: ; The phase rotation gate is: ;in is the quantum state of the qubit, is the angle of the phase rotation door, and is plural and satisfies , Indicates the angle, and is the exponential form of a complex number, is an imaginary unit, and When a phase angle offset fault occurs in the power grid, multiple phase rotation gates are used to simulate the change of phase angle, and the characteristics of the phase angle change are extracted and recorded; a quantum gate combination for simulating current mutation faults: a quantum bit is used as a control bit to indicate whether a fault occurs, and Indicates no fault, The qubit represents the fault, and the other qubit represents the current amplitude bit, whose states include occurrence and non-occurrence. The qubit representing the current amplitude is acted on by the CNOT gate, and the single-bit rotation gate is combined to change the state of the current amplitude qubit to simulate the mutation of the current amplitude. The characteristics related to the current mutation, including the increase of the current amplitude and the stable state after the mutation, are extracted and recorded as features; the quantum gate combination used to simulate the influence of compound faults on power parameters: two qubits are used as control bits to indicate whether different types of faults occur at the same time, and one fault corresponds to one control bit. Indicates no fault. , , It represents different fault combinations, and another quantum bit is used to represent the state of the power parameter. When a specific fault combination occurs, the Toffoli gate acts on the quantum bit representing the power parameter, and then combines with the S phase gate to change the state of the power parameter quantum bit, simulate the impact of the composite fault on the power, extract the change of the power factor, the fluctuation characteristics of the active power and the reactive power and record them; S103: quantum measurement and feature extraction, measure the quantum state, and the measurement basis vector selection is based on the fault feature to be extracted. When the abnormal change feature of the voltage amplitude is concerned, the measurement basis vector corresponding to the voltage amplitude encoding quantum bit is selected. The measurement result will obtain a set of probability distributions. These probability distributions reflect the possibility of different fault features. When the probability corresponding to a certain voltage amplitude interval is high, it means that there is a fault in the power grid that causes the voltage amplitude change. Through multiple measurements and statistical analysis, the key characteristic indicators are determined, including the probability threshold of the occurrence of specific fault features and the correlation coefficient between different fault features. The above indicators are used as the subsequent fault diagnosis The basis for fault diagnosis and classification; S2: Construction of power grid knowledge graph, collecting multi-source data from the power grid, integrating and processing it, performing entity recognition and relationship extraction on it, and constructing the power grid knowledge graph based on it, and performing fault reasoning based on the constructed power grid knowledge graph. When the power grid fault characteristics are detected, the power grid knowledge graph is used for fault reasoning. According to the fault characteristics, all the equipment and line nodes involved are matched in the power grid knowledge graph. Along the edges in the power grid knowledge graph, according to the fault propagation relationship and probability information, the root cause and impact range of the fault are gradually inferred. When an abnormal voltage drop is detected in a substation, the power grid knowledge graph is used to query the relevant information of the transmission line, upper substation and in-station equipment connected to the substation. Combined with the fault cause analysis of similar voltage drops in historical fault cases, it is inferred whether it is a transmission line fault, an upper substation fault or an in-station equipment fault, and further determines the specific location of the fault and the repair plan. This fault reasoning method based on the power grid knowledge graph can make full use of the prior knowledge of the power grid and improve the accuracy and reliability of fault diagnosis. The construction of the power grid knowledge graph specifically includes the following steps: S201: Collect data from multiple business systems of the power grid, including but not limited to the data acquisition and monitoring control system of the power grid to obtain the real-time voltage, current, and power operation data of the power grid; the energy management system to collect the power grid topology structure and equipment operation status information; the power outage management system to obtain power outage event-related data, including but not limited to power outage time, scope, and cause; equipment asset information and maintenance records in the equipment maintenance management system, and also integrate the external environment data of the power grid, including meteorological data and geographic information data, to clean and standardize the collected data; S202: Classify and label the cleaned and standardized data based on entity and relationship types, label the substation, transmission line, transformer and other equipment information as entities, label the connection relationship, electrical parameter relationship, and fault association relationship between equipment as relationships, and for historical fault data, label the time, location, type, and equipment and lines involved in detail; S203: Use a bidirectional long short-term memory network combined with a conditional random field model to identify entities in power grid data, use power grid equipment names and location names as entity categories, input the preprocessed data into the model for training and recognition, and when in a text data describing the operating status of the power grid, the model can The entity data of "XX substation" and "XX transmission line" are accurately identified, the contextual semantic information of the bidirectional long short-term memory network learning data, and the conditional random field model is used to constrain and optimize the output of the bidirectional long short-term memory network; the connection relationship in the power grid topology is extracted through predefined rules, and the causal relationship between the fault and the equipment is extracted by using the convolutional neural network. When "XX substation is connected to XX substation through XX transmission line" appears in the text data, the connection relationship between the three entities can be accurately extracted according to the pre-set connection relationship extraction rules. For some complex and semantically ambiguous relationships, such as the causal relationship between faults and equipment, the convolutional neural network is used. The CNN model is trained with a large amount of labeled data to automatically learn the potential relationship patterns between faults and equipment from text data, including extracting relationship information such as "XX equipment failure causes XX line to trip" from the fault report text; the predefined rules include topological structure connection relationship rules, electrical parameter association relationship rules, equipment operation and maintenance relationship rules, and fault association relationship rules. The topological structure connection relationship rules include substation and transmission line connection rules: when the text contains the statement that substation [name] is connected to substation [name] through transmission line [name], the three entities and the connection relationship between them are directly extracted. When A The substation is connected to the B substation through the No. 1 transmission line. According to this rule, the three entities and the connection relationship between them can be clarified; substation and transformer connection rule: when there is a sentence that the substation [name] contains the transformer [name] and the transformer [name] is located in the substation [name], the two entities of the substation and the transformer and the inclusion relationship between them are extracted, based on which the relevant entities and relationships can be extracted; transmission line and tower connection rule: when the text mentions that the transmission line [name] is erected on the tower [name] and the tower [name] supports the transmission line [name], the entities of the transmission line and the tower and the erection and support relationship can be extracted, based on which the power grid knowledge graph relationship is constructed; The electrical parameter association relationship rules include the voltage, current, and resistance relationship rules based on Ohm's law: in the data description, when there is a description that according to Ohm's law, when the resistance [resistance value] remains unchanged, the voltage [voltage value] is proportional to the current [current value], the three entities of voltage, current, and resistance are extracted, and their parameters are also regarded as entities and the association relationship between them according to Ohm's law; it is known that the line resistance is 5 ohms, when the voltage is 100V, the current can be calculated to be 20A according to Ohm's law, and the entities and relationships can be extracted according to the voltage, current, and resistance relationship rules based on Ohm's law; power, voltage, and current relationship rules based on the power formula: when there is a data description that power [power value] is equal to voltage [voltage value] multiplied by current [current value], the three entities of power, voltage, and current and the association relationship between them according to the power formula can be extracted. The power of the device is 500W and the voltage is 220V. According to the power formula, the current is about 2.27A, and the power grid knowledge graph relationship is constructed based on this; Equipment operation and maintenance relationship rules include equipment and maintenance plan relationship rules: when the text contains the content that equipment [name] is planned to be maintained at [time], maintenance plan [number] involves equipment [name], and the maintenance time is [time], the two entities of equipment and maintenance plan and the planned maintenance relationship between them are extracted; equipment and maintenance personnel relationship rules: when there is a description that equipment [name] is maintained by maintenance personnel [name] and maintenance personnel [name] performs maintenance operations on equipment [name], the two entities of equipment and maintenance personnel and the responsible maintenance relationship between them are extracted, and the power grid knowledge graph relationship is constructed based on this; fault association relationship rules include direct fault and equipment association rules: when there is content that equipment [name] has a [fault type] fault in the data, the two entities of fault and equipment and the occurrence relationship between them are extracted; indirect fault and equipment association rules: when there is a statement that the fault of equipment [name 1] causes the fault of [fault type] in equipment [name 2], the three entities of two equipment and fault and the cause relationship between them are extracted, and the power grid knowledge graph relationship is constructed based on this; S204: Design of the architecture of the power grid knowledge graph. Design the architecture of the power grid knowledge graph, including defining entity types, relationship types, and attribute types. Entity types cover various equipment, locations, personnel in the power grid and their names, locations, and capacity attributes; relationship types include topological connection relationships, electrical parameter association relationships, operation and maintenance relationships, and fault causal relationships; attribute types correspond to various feature descriptions of entities; S205: Construction and update of the power grid knowledge graph. According to the results of entity recognition and relationship extraction, a graph database is used to construct a power grid knowledge graph. The identified entities are taken as nodes, the extracted relationships are taken as edges, and the attributes of the entities are stored as the attribute information of the nodes in the graph database. During the operation of the power grid, as new data is continuously generated, the power grid knowledge graph is updated in real time. When a new substation is built and connected to the power grid, its related information is added as new entities and relationships to the power grid knowledge graph. When a new fault occurs and is analyzed, the fault is linked to related equipment, lines, and other related information. The relationship between the two roads and the cause of the fault are updated to the fault knowledge part of the power grid knowledge graph, so that the power grid knowledge graph always keeps the synchronization with the actual operation status and knowledge of the power grid; S3: Map the entity nodes in the power grid knowledge graph to quantum states according to their key attribute information. For substation nodes, they are encoded according to their voltage level, capacity, and geographical location. Assuming that the voltage level has three categories: high, medium, and low, the capacity range is divided into several intervals, and the geographical location is represented by discrete values ​​of longitude and latitude. The high, medium, and low voltage levels are respectively encoded as different basis state combinations of quantum bits. The capacity interval and geographical location information are also encoded in a similar way to other quantum bits. A unique quantum state representation is constructed for each substation node. Based on this, the complex attribute information of a substation node can be encoded in the quantum state. At the same time, the quantum bit has the superposition state characteristics, which can simultaneously represent multiple possible attribute combination states; quantum gate operations are designed to encode the relationship between the entity nodes in the power grid knowledge graph; S4: Fault diagnosis. When a power grid fails, the Grover algorithm is used to search for nodes that match the fault characteristics in the quantum state space of the nodes encoding the power grid knowledge graph, and the fault characteristics are converted into quantum state constraints. The fault node is found based on the iterative operation of the Grover algorithm on the quantum state. When the fault characteristic is a voltage drop in a certain area, the quantum states of the substations and transmission line nodes related to the area will be marked in the quantum state space due to the change in the quantum bit state encoded by the voltage attribute. The Grover algorithm quickly locates the faulty node by quickly searching and amplifying these quantum states. Compared with traditional traversal or index-based search methods, its time complexity is reduced from exponential to polynomial, greatly improving the efficiency of fault node locating. After determining the faulty node, the quantum annealing algorithm is used to find the relevant fault propagation path and fault root cause in the power grid knowledge graph to diagnose the power grid fault. Diagnosis, the quantum annealing algorithm simulates the evolution of quantum systems between different energy states, and searches for the global optimal solution in the solution space composed of nodes and relationships in the power grid knowledge graph. The fault propagation path in the power grid knowledge graph is regarded as an optimization problem. The node connection relationship and the fault propagation probability on the path are used as constraints of the optimization target. Starting from a suspected fault node, the quantum annealing algorithm evolves the quantum states representing different propagation paths, considering the fault probability of each node and the connection relationship between nodes, until the most likely path from the fault node to the root node is found. This path corresponds to the quantum state with the lowest energy. Compared with the traditional fault reasoning path search method based on greedy algorithm or dynamic programming, the quantum annealing algorithm can avoid falling into the local optimal solution and find the root cause of the fault more accurately, especially in the case of complex power grid knowledge graph structure and multiple fault sources, with better performance; Based on the iterative operation of the Grover algorithm on the quantum state, finding the faulty node includes the following steps: S401: Determine the total number of nodes in the power grid knowledge graph , according to the number of nodes ,need qubits to represent all possible node states, and initialize all qubits to a uniform superposition state; S402: Determine the node matching conditions based on the monitored fault characteristics and design the Oracle function , assuming is a base state and corresponds to a number in the power grid knowledge graph. If the fault feature matching condition is met, If not satisfied, then After operating the entire quantum state space, the Oracle function marks the quantum states corresponding to all nodes that meet the fault characteristics. When the fault characteristic is voltage drop, in the power grid knowledge graph, the voltage-related nodes, including the attributes in the substation and transformer, such as the output voltage is lower than the normal threshold, constitute the matching condition; S403: According to the principle of Grover algorithm, determine the number of iterations , the calculation formula is: ;in is the number of iterations, Indicates a round-down operation; S404: Constructing the Grover diffusion operator , apply Oracle functions to the initialized quantum state in turn and diffusion operator ,conduct iterations. After each iteration, the quantum state will change, and the probability amplitude of the quantum state corresponding to the node matching the fault feature will gradually increase. S405: After After iterations, the final quantum state is measured, and the measurement operation causes the quantum state to collapse to a ground state. The node corresponding to this base state is the node with the highest probability of matching the fault characteristics searched by the Grover algorithm. After obtaining the node with the highest probability, the node corresponding to the base state is converted into an actual node in the power grid knowledge graph. The actual node is the node related to the fault characteristics. When the searched node is a substation, the equipment operation status and voltage output parameters of the substation can be checked in detail to determine the specific cause and location of the fault.

[0017] S5: Feedback and optimization are performed based on the power grid knowledge graph. The structural information and entity relationships in the power grid knowledge graph provide an optimization basis for the number of quantum bits and connection methods in the quantum computing model. When the power grid structure in an area in the power grid knowledge graph is relatively complex and the devices are closely related, the number of quantum bits representing the nodes and relationships in the area is increased accordingly during quantum state encoding, and the entanglement relationship between quantum bits is adjusted to better capture the complex characteristics of the power grid in the area. Conversely, when the power grid structure in a certain area is simple, the number of quantum bits is appropriately reduced to improve quantum computing efficiency, so that the structural information of the power grid knowledge graph can dynamically adjust the parameters of the quantum computing model to make it more in line with the actual situation of the power grid; a large number of historical fault cases and fault handling experience in the power grid knowledge graph are used as the basis for updating the quantum computing model. The algorithm parameters in the model, the number of iterations in the quantum search algorithm and the annealing time in the quantum annealing algorithm are adjusted according to the statistical information of similar fault cases in the power grid knowledge graph. When the historical cases of a certain fault type in the power grid knowledge graph show that its fault propagation path is relatively simple and obvious, the annealing time in the quantum annealing algorithm is shortened to speed up the search speed; when the fault type is complex and changeable, the number of iterations is increased and the intensity of quantum gate operations is adjusted to improve the quantum algorithm's ability to handle complex faults. By continuously using the fault case information in the power grid knowledge graph to update the quantum algorithm parameters, the quantum computing model can adapt to the changes in the power grid fault mode and improve the accuracy and reliability of fault diagnosis; using the quantum annealing algorithm to find relevant fault propagation paths and fault roots in the power grid knowledge graph includes the following steps: S401: Assign a quantum bit to each node in the power grid knowledge graph to represent its state. Indicates that the node is normal, and the quantum bit Indicates that the node is in a state of failure or being affected by the failure; S402: Setting the objective function, including the energy item of the node failure state and the energy item corresponding to the connection relationship between nodes, wherein the energy item of the node failure state, for each node , according to its prior probability of failure Set capability items, when the node In fault state , and its energy contribution formula is: ;in is the energy contribution of the current node, is a coefficient related to the importance of the node and the reliability of fault detection; the energy term corresponding to the connection relationship between nodes: for nodes with connection relationships, The nodes with connection relationship are recorded as , according to the conditional probability of fault propagation between them Set the energy term, when the node and If both are in a fault state, the energy contribution formula of the connection relationship is: ;in is the energy term, It represents a node and The coupling coefficient of the tightness of the connection and the ease of fault propagation; The overall objective function is: ; is the overall objective function; Find the combination of quantum bit states that makes the energy of this objective function the lowest. The node state and connection relationship corresponding to this combination represent the fault propagation path and the root cause of the fault.

[0018] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for rapid diagnosis of power grid faults based on artificial intelligence technology, characterized in that: The steps include: S1: Extract grid fault features based on quantum computing, map the voltage, current, power, and device status information in the grid to quantum bits according to specific coding rules to extract the features of grid faults; S2: Construction of power grid knowledge graph: collect multi-source data from the power grid, integrate and process it, perform entity recognition and relationship extraction on it, and build the power grid knowledge graph based on it; S3: Map the entity nodes in the power grid knowledge graph to quantum states based on their key attribute information, and design quantum gate operations to encode the relationships between the entity nodes in the power grid knowledge graph; S4: Fault diagnosis. When a power grid fault occurs, the Grover algorithm is used to search for nodes matching the fault characteristics in the quantum state space of the nodes encoding the power grid knowledge graph. The fault characteristics are converted into quantum state constraints. Based on the iterative operation of the Grover algorithm on the quantum state, the fault node is found. After the faulty node is identified, the quantum annealing algorithm is used to find the relevant fault propagation path and fault root cause in the power grid knowledge graph to diagnose the power grid fault. S5: Feedback and optimization based on the power grid knowledge graph, using the structural information and entity relationships in the power grid knowledge graph to provide optimization basis for the number of quantum bits and connection methods in the S4 algorithm; A large number of historical fault cases and fault handling experience in the power grid knowledge graph are used to update the algorithm parameters in the S4 algorithm.

2. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 1 is characterized in that: In S1, the extraction of power grid fault features based on quantum computing specifically includes the following steps: S101: Quantum state construction and data encoding, determine the entities and relationships used to describe the state of the power grid, including the voltage amplitude, current amplitude, and phase angle of each node, record the number of power grid monitoring points as N, each monitoring point has M parameters, divide the amplitude range of voltage amplitude, current amplitude, and phase angle into K intervals, set it as the encoding rule, use quantum bits to represent the value state of each parameter, and each parameter uses log2K quantum bits to encode its value. For all monitoring point parameters of the entire power grid, N*M*log2K quantum bits are required to construct the quantum state, and each quantum state corresponds to a parameter combination state of the power grid; S102: Quantum gate operation and feature mapping, using Hadamard gate to initialize quantum bits to put them in a superposition state, designing a special quantum gate combination to simulate the impact mechanism of power grid faults on parameters, introducing entanglement gate operation, entangling quantum bits corresponding to different monitoring points and different parameters, to capture the correlation and coupling between power grid parameters, and extract comprehensive fault feature information; S103: Quantum measurement and feature extraction, measuring the quantum state, and selecting the basis vector for measurement is based on the fault feature to be extracted.

3. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 2 is characterized in that: In S101, the parameter combination state of the power grid includes a voltage-related parameter combination state, a current-related parameter combination state, and a power-related parameter combination state, wherein the voltage-related parameter combination state includes a voltage amplitude combination, a voltage phase combination, and a frequency combination; the current-related parameter combination state includes a current amplitude combination and a current phase combination; and the power-related parameter combination state includes an active power combination and a reactive power combination.

4. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 2 is characterized in that: In S102, the quantum gate combination is based on the relationship between the entity nodes in the power grid knowledge graph, including a quantum gate combination for simulating voltage phase angle fault, a quantum gate combination for simulating current mutation fault, and a quantum gate combination for simulating the influence of compound fault on power parameters. The quantum gate combination for simulating voltage phase angle fault: a phase rotation gate is used to rotate the phase of the quantum bit, and the quantum state of the quantum bit is recorded as: ; The phase rotation gate is: ； in is the quantum state of the qubit, is the angle of the phase rotation door, and is plural and satisfies , Indicates the angle, and is the exponential form of a complex number, is an imaginary unit, and ,When a phase angle offset fault occurs in the power grid, multiple phase rotation gates are applied to simulate the ,change of the phase angle, and the characteristics of the phase angle change are ,extracted and recorded; Quantum gate combination for simulating current mutation fault: one qubit is used as a control bit to indicate whether a fault occurs, and Indicates no fault, The qubit represents the fault, and the other qubit represents the current amplitude bit, whose states include occurrence and non-occurrence. The qubit representing the current amplitude is acted on by the CNOT gate, and the single-bit rotation gate is combined to change the state of the current amplitude qubit to simulate the mutation of the current amplitude. The characteristics related to the current mutation, including the increase of the current amplitude and the stable state after the mutation, are extracted and recorded as features; the quantum gate combination used to simulate the influence of compound faults on power parameters: two qubits are used as control bits to indicate whether different types of faults occur at the same time, and one fault corresponds to one control bit. Indicates no fault. , , It represents different fault combinations, and the other quantum bit is used to represent the state of the power parameter. When a specific fault combination occurs, the Toffoli gate acts on the quantum bit representing the power parameter, and then combines with the S phase gate to change the state of the power parameter quantum bit, simulating the impact of compound faults on power, extracting the change in power factor, and the fluctuation characteristics of active power and reactive power and recording them.

5. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 1 is characterized in that: In S2, the construction of the power grid knowledge graph specifically includes the following steps: S201: Collect data from multiple business systems of the power grid, including the data acquisition and monitoring control system of the power grid to obtain the real-time voltage, current and power operation data of the power grid; the energy management system to collect the topology of the power grid and the operation status information of the equipment; the power outage management system to obtain the data related to the power outage event, including the time, scope and cause of the power outage; the equipment asset information and maintenance records in the equipment maintenance management system; integrate the external environment data of the power grid, including meteorological data and geographic information data, and clean and standardize the collected data; S202: classify and label the cleaned and standardized data based on the entity and relationship types; S203: using a bidirectional long short-term memory network combined with a conditional random field model to identify entities in the power grid data; extracting the connection relationship in the power grid topology structure through predefined rules, and using a convolutional neural network to extract the causal relationship between the fault and the equipment; S204: Design of the knowledge graph architecture of the power grid, including the definition of entity types, relationship types, and attribute types. Entity types cover various equipment, locations, personnel in the power grid and their names, locations, and capacity attributes; relationship types include topological connection relationships, electrical parameter association relationships, operation and maintenance relationships, and fault causal relationships; attribute types correspond to various feature descriptions of entities; S205: Construction and update of power grid knowledge graph. Based on the results of entity recognition and relationship extraction, a graph database is used to construct a power grid knowledge graph. The identified entities are used as nodes, the extracted relationships are used as edges, and the attributes of the entities are stored as the attribute information of the nodes in the graph database. During the operation of the power grid, as new data is continuously generated, the power grid knowledge graph is updated in real time.

6. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 5 is characterized in that: In the S203, the predefined rules include topological structure connection relationship rules, electrical parameter association relationship rules, equipment operation and maintenance relationship rules, and fault association relationship rules, wherein the topological structure connection relationship rules include substation and transmission line connection rules: when the text contains the statement that substation [name] is connected to substation [name] through transmission line [name], three entities and the connection relationship between them are directly extracted; substation and transformer connection rules: when there are sentences such as substation [name] contains transformer [name] and transformer [name] is located in substation [name], the two entities of substation and transformer and the inclusion relationship between them are extracted, based on which related entities and relationships can be extracted; transmission line and pole tower connection rules: when the text mentions that transmission line [name] is erected on pole tower [name] and pole tower [name] supports transmission line [name], the entities of transmission line and pole tower and the erection and support relationship can be extracted, based on which the power grid knowledge graph relationship is constructed; The electrical parameter association relationship rules include the voltage, current, and resistance relationship rules based on Ohm's law: in the data description, when there is a description that according to Ohm's law, when the resistance [resistance value] remains unchanged, the voltage [voltage value] is proportional to the current [current value], the three entities of voltage, current, and resistance are extracted, and their parameters are also regarded as entities and the association relationship between them according to Ohm's law; the power, voltage, and current relationship rules based on the power formula: when there is a data description that power [power value] is equal to voltage [voltage value] multiplied by current [current value], the three entities of power, voltage, and current and the association relationship between them according to the power formula can be extracted. The power grid knowledge graph relationship is constructed here; the equipment operation and maintenance relationship rules include the equipment and maintenance plan relationship rules: when the text contains the content that the equipment [name] is planned to be repaired at [time], the maintenance plan [number] involves the equipment [name], and the maintenance time is [time], the two entities of equipment and maintenance plan and the planned maintenance relationship between them are extracted; the equipment and maintenance personnel relationship rules: when there is a description that the equipment [name] is maintained by the maintenance personnel [name] and the maintenance personnel [name] performs maintenance operations on the equipment [name], the two entities of equipment and maintenance personnel and the responsible maintenance relationship between them are extracted, and the power grid knowledge graph relationship is constructed based on this; Fault association rules include direct fault and device association rules: when the data contains content that the device [name] has a [fault type] fault, the two entities of fault and device and the occurrence relationship between them are extracted; Indirect fault and equipment association rules: When a statement appears that the fault of device [name 1] causes the fault of [fault type] in device [name 2], extract the three entities of two devices and faults and the causing relationship between them, and build the power grid knowledge graph relationship based on this.

7. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 1 is characterized in that: In S4, based on the iterative operation of the Grover algorithm on the quantum state, finding the faulty node includes the following operation steps: S401: Determine the total number of nodes in the power grid knowledge graph , according to the number of nodes ,need qubits to represent all possible node states, and all qubits are initialized to a uniform superposition state; S402: Determine node matching conditions based on detected fault characteristics and design Oracle functions , assuming is a base state and corresponds to a number in the power grid knowledge graph. If the fault feature matching condition is met, If not satisfied, then ,After operating on the entire quantum state space, the Oracle function marks the quantum states corresponding to all nodes that meet the fault characteristics; S403: Determine the number of iterations according to the principle of Grover algorithm , the calculation formula is: ;in is the number of iterations, Indicates a round-down operation; S404: Constructing the Grover diffusion operator , apply Oracle functions to the initialized quantum state in turn and diffusion operator ,conduct Iterations, after each iteration, the quantum state will change, and the probability amplitude of the quantum state corresponding to the node matching the fault feature will gradually increase; S405: Passing After iterations, the final quantum state is measured, and the measurement operation causes the quantum state to collapse to a ground state. The node corresponding to this base state is the node with the highest probability of matching the fault characteristics searched by the Grover algorithm. After obtaining the node with the highest probability, the node corresponding to the base state is converted into an actual node in the power grid knowledge graph. The actual node is the node related to the fault characteristics.

8. The method for rapid diagnosis of power grid faults based on artificial intelligence technology according to claim 1 is characterized in that: In said S4, using the quantum annealing algorithm to find relevant fault propagation paths and fault roots in the power grid knowledge graph includes the following operation steps: S401: assigning a quantum bit to each node in the power grid knowledge graph to represent its state, and assigning the quantum bit Indicates that the node is normal, and the quantum bit Indicates that the node is in a faulty or affected state; S402: Setting the objective function, including the energy item of the node failure state and the energy item corresponding to the connection relationship between nodes, wherein the energy item of the node failure state is , according to its prior probability of failure Set capability items, when the node In fault state , and its energy contribution formula is: ;in is the energy contribution of the current node, is a coefficient related to the importance of the node and the reliability of fault detection; the energy term corresponding to the connection relationship between nodes: for nodes with connection relationships, The nodes with connection relationship are recorded as , according to the conditional probability of fault propagation between them Set the energy term, when the node and If both are in a fault state, the energy contribution formula of the connection relationship is: ;in is the energy term, It represents a node and The coupling coefficient of the tightness of connection and the ease of fault propagation; the overall objective function is: ; is the overall objective function; Find the combination of quantum bit states that makes the energy of this objective function the lowest. The node state and connection relationship corresponding to this combination represent the fault propagation path and the root cause of the fault.

Images