Power transmission system fault identification method and system based on quantum neural network

Abstract

The invention relates to a power transmission system fault identification method based on a quantum neural network. The method comprises the steps that electrical data are collected and preprocessed; mapping the preprocessed electrical data to a quantum calculation space to obtain quantum state data; and inputting the quantum state data into the quantum neural network, and outputting a fault identification result. Compared with the prior art, the method has the advantages of high recognition accuracy, high reliability and the like.

Description

Technical Field

[0001] The present invention relates to the field of power system fault detection, and in particular to a method and system for power transmission system fault identification based on quantum neural network. Background Art

[0002] As an important part of the modern power network, the power transmission system is responsible for transmitting electricity from power plants to users, so it is crucial to maintain the stable operation of the power grid. The power transmission system includes key equipment such as power transformers, high-voltage transmission lines and distribution networks, and its stability is directly related to the safety and efficiency of the power grid. However, due to the complex and changeable external environment and the aging or design defects of the equipment itself, power transmission lines face many potential failure risks. For example, natural disasters such as lightning strikes, blizzards, floods and other extreme weather can cause direct damage to transmission equipment, resulting in power outages. Equipment aging and material fatigue will also gradually accumulate during long-term use, eventually causing equipment failure. In addition, human factors, such as operating errors, improper maintenance, and third-party construction, may also pose a threat to the stability of the power transmission system.

[0003] The occurrence of faults may not only lead to power supply interruptions, but also have a profound impact on the safety and stability of power equipment. Long-term power outages will cause great inconvenience to users and may even paralyze key industries such as medical and transportation systems. At the same time, the chain reaction caused by power failures may trigger large-scale power grid failures, leading to more serious economic losses. Especially in the context of the increasing popularity of smart grids, the degree of automation in each link of the power grid is getting higher and higher, and faults may spread at a faster rate. Therefore, quickly and accurately detecting whether there is an electrical fault, especially a fault in the power transmission line, has become a top priority to ensure the safe and stable operation of the power system. Existing electrical fault identification methods can be mainly divided into the following categories:

[0004] (1) Rule-based fault identification: Rule-based fault identification methods are widely used in power fault diagnosis. These methods usually rely on expert experience formulas, logical reasoning, and decision tree models, etc., and try to determine whether the fault is a problem through a predefined set of rules. The advantages of this type of method are clear logic and easy implementation. Especially when the scenario is simple and the data type is small, the fault judgment can be made in a short time. However, with the expansion of the scale and increase in complexity of the power system, this method has gradually become inadequate. There are many types of data in the power grid, and the feature differences of different data have become more complex, which is difficult to accurately distinguish by simple rules alone. In addition, rule-based methods usually require a lot of manual intervention and prior knowledge, and are slow to respond to fault identification in unknown or complex environments. Therefore, although this method has played an important role in the past, its limitations are becoming more and more prominent, and it is difficult to adapt to the needs of the current complex power system.

[0005] (2) Fault identification based on statistical analysis: Statistical analysis methods automatically build fault identification models by analyzing patterns and rules in historical data. Common statistical analysis methods include support vector machines (SVM), random forests (RF), and K nearest neighbors (KNN). These methods extract and model the electrical signal characteristics of the power transmission system and try to determine whether a fault has occurred through pattern matching. Statistical analysis methods can usually improve the efficiency and accuracy of fault detection to a certain extent, especially when the amount of data is large, and can improve the generalization ability of the model through a large number of sample learning. In particular, ensemble learning algorithms such as random forests can improve the accuracy by combining multiple decision trees, and have good application prospects in power fault identification tasks.

[0006] However, with the increase in data types and data dimensions in power systems, fault identification methods based on statistical analysis have gradually exposed their high computational complexity and high demand for computing resources. For high-dimensional data or multi-parameter scenarios, the training and inference processes of these methods often require a lot of computing resources. At the same time, as the amount of data increases, the training time of the model also increases significantly. In addition, the performance of statistical analysis methods in fault identification is not satisfactory when dealing with complex power system scenarios. These algorithms often perform poorly when dealing with nonlinear fault characteristics and have difficulty capturing weak but important features in electrical signals.

[0007] (3) Fault identification based on deep learning: The rise of deep learning technology has provided a new solution for electrical fault identification. Deep neural networks (DNNs) have powerful feature extraction capabilities and can automatically learn the electrical data characteristics of power systems from large-scale and complex data. They are particularly suitable for processing high-dimensional and multi-parameter electrical data. For example, convolutional neural networks (CNNs) can capture spatial features in electrical signals and are suitable for processing image or time series data; long short-term memory networks (LSTMs) can process time series data and are suitable for processing time-varying electrical signals. However, when faced with large-scale and complex data, deep learning models still face challenges such as high computing resource consumption and long training time. In addition, the "black box" nature of deep learning models makes the fault identification process less transparent and difficult to provide clear diagnostic basis.

[0008] It can be seen that with the continuous increase in data scale and complexity, the limitations of traditional fault identification methods in the power system fault identification process are becoming more and more obvious. Therefore, it is a problem that needs to be solved to provide a fault identification method that can adapt to the scale of big data and complex data environment, with high accuracy, high efficiency and low computing cost. Summary of the invention

[0009] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a method and system for fault identification of power transmission system based on quantum neural network.

[0010] The purpose of the present invention can be achieved by the following technical solutions:

[0011] According to a first aspect of the present invention, a method for fault identification of a power transmission system based on a quantum neural network is provided, the method comprising:

[0012] Collecting electrical data and preprocessing it, wherein the electrical data includes current data and voltage data;

[0013] Mapping the preprocessed electrical data to the quantum computing space to obtain quantum state data;

[0014] The quantum state data is input into a quantum neural network to output a fault identification result, wherein the quantum neural network includes an input layer, an adaptive quantum circuit layer and an output layer; the adaptive quantum circuit layer is used to evolve the quantum state data and calculate the probability of fault occurrence based on the evolved quantum state data, and includes multiple evolution modules, each of which includes two layers of gate structures, each of which includes multiple quantum gates and the previous gate structure has one more quantum gate than the next gate structure.

[0015] As a preferred technical solution, the pretreatment includes:

[0016] Extracting the characteristics of the electrical data to obtain characteristic data and screening out maximum characteristic data and minimum characteristic data in the characteristic data;

[0017] The feature data is normalized to data between [0,1], where the normalization expression is:

[0018]

[0019] x′ represents the preprocessed electrical data, x min Represents the minimum feature data, x max represents the maximum value feature data, and x represents the feature data.

[0020] As a preferred technical solution, the method for obtaining the quantum state data is:

[0021] Each preprocessed electrical data is mapped to the quantum computing space using the quantum state mapping function. The expression of the mapping process is:

[0022] Φ(x′)=cos(x′)|0>+sin(x′)|1>,

[0023] Among them, Φ(x′) represents the quantum state data, x′ represents the preprocessed electrical data, and |1> and |0> represent the two ground states of the quantum bit.

[0024] As a preferred technical solution, the method for outputting fault identification results is:

[0025] The quantum state data is transmitted to the adaptive quantum circuit layer via the input layer, and quantum evolution is performed in the adaptive quantum circuit layer to generate an output quantum state;

[0026] The probability of a fault occurring is calculated based on the output quantum state, and the probability of a fault occurring is the fault identification result.

[0027] As a preferred technical solution, the process of quantum evolution is:

[0028] Assuming that the first-layer gate structure of each evolution module includes N quantum gates, the number of quantum gates in the second-layer gate structure is N-1, the number of evolution modules is K, and each quantum gate receives its input and outputs the first data and the second data, and the first data and the second data are exactly the same;

[0029] Each quantum gate in the first layer gate structure of the first level evolution module receives the inner product of at least two random quantum state data as input;

[0030] The first quantum gate in the first-layer gate structure of the k-th (1 < k ≤ K) level evolution module receives as input the inner product of the first data output by the first quantum gate in the first layer of the (k - 1)-th level evolution module and the first data output by the first quantum gate in the second layer of the (k - 1)-th level evolution module. The n-th (1 < n < N) quantum gate receives as input the inner product of the second data output by the (n - 1)-th quantum gate in the second layer of the (k - 1)-th level evolution module and the first data output by the n-th quantum gate. The N-th quantum gate receives as input the second data output by the N-th quantum gate in the first layer of the (k - 1)-th level evolution module and the second data output by the (N - 1)-th quantum gate in the second layer of the (k - 1)-th level evolution module.

[0031] The n-th (1 ≤ n ≤ N - 1) quantum gate in the second-layer gate structure of the k-th (1 ≤ k ≤ K) level evolution module receives as input the inner product of the second data of the n-th gate structure and the first data of the (n + 1)-th gate structure in the first-layer gate structure of the same-level evolution module.

[0032] The first data output by the first quantum gate, the second data output by the N-th quantum gate, and the data output by the second-layer gate structure in the first-layer gate structure of the K-th level evolution module are used as the output quantum state.

[0033] As a preferred technical solution, the method for calculating the fault occurrence probability based on the output quantum state is as follows:

[0034]

[0035] where p represents the fault occurrence probability, represents the measurement operator, and ψ represents the output quantum state.

[0036] As a preferred technical solution, the training method of the quantum neural network is a combination of offline training and online learning. The process of offline training is as follows:

[0037] Obtain historical electrical data and perform data processing to generate a fault data set. Divide the fault data into a training set and a test set according to a preset ratio. Each data in the fault data set includes historical electrical data and a fault label, and the fault label is type 01 data. When the fault label is 0, it indicates no fault, and when the fault label is 1, it indicates a fault.

[0038] Initialize the parameters of the quantum neural network;

[0039] Use the training set as the input of the quantum neural network, and use the quantum gates in the quantum neural network to perform calculations and transformations on the training set, and output the fault occurrence probability;

[0040] Use the loss function to calculate the loss value between the fault occurrence probability and the fault label, and update the quantum gate parameters based on the loss value;

[0041] Repeat this process until the performance of the quantum neural network on the training set is stable, and use the test set to evaluate the performance of the trained quantum neural network. If the performance meets the standard, the training is terminated, otherwise the training is restarted.

[0042] As a preferred technical solution, the loss function expression is:

[0043]

[0044] L(θ) represents the loss value, y i represents the fault label of the i-th sample in the training set, p i represents the probability of failure of the i-th sample, θ represents the quantum gate parameter, and N represents the total number of samples in the training set.

[0045] As a preferred technical solution, the method for updating the quantum gate parameters is the gradient descent method, which is expressed as:

[0046]

[0047] θ′ represents the updated quantum gate parameters, θ represents the original quantum gate parameters, and η represents the learning rate.

[0048] According to a second aspect of the present invention, a system for fault identification of a power transmission system based on a quantum neural network is provided, and the system is used in the above method.

[0049] Compared with the prior art, the present invention provides a fault identification method based on quantum neural network, which converts electrical data into quantum state data, and then evolves the quantum state data into output quantum state through transparent quantum state evolution processing, clarifies the data processing process, and improves the reliability of the algorithm and the accuracy of fault identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A flow chart of a method for fault identification of a power transmission system according to the present invention;

[0051] Figure 2 It is a schematic diagram of the structure of the quantum neural network of the present invention;

[0052] Figure 3 This is a result diagram of fault identification using quantum neural network of the present invention;

[0053] Figure 4 This is a result diagram of fault identification using a fully connected neural network according to the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0055] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantitative limitation, and may represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0056] Example 1

[0057] As a new computing paradigm, quantum computing uses quantum mechanical properties such as quantum superposition and quantum entanglement to achieve exponential computing acceleration in certain specific tasks, especially in high-dimensional data processing and complex optimization problems, showing great potential to outperform classical computing. The core advantage of quantum computing lies in its parallel computing capability. Through quantum superposition, a quantum bit can represent multiple states at the same time, which enables quantum computing to process multiple data paths at the same time, thereby significantly improving computing efficiency. In the task of electrical fault identification, this feature of quantum computing can effectively deal with the computing bottleneck brought by high-dimensional data and reduce the time for data processing and model training. In addition, quantum entangled states can transmit information between multiple quantum bits through non-local quantum correlations, allowing quantum computing to simultaneously process the complex correlations between multiple fault features and improve the performance of the algorithm.

[0058] Quantum neural network (QNN), as a product of the combination of quantum computing and neural networks, uses quantum state evolution to process data. Compared with traditional neural networks, QNN processes input data through quantum gate operations, maps classical data to quantum states, and processes data features through quantum state evolution to adapt to different fault types. This technical idea provides a new theoretical basis for power system fault identification, especially showing the potential of quantum computing in high-dimensional complex data scenarios.

[0059] Based on the above situation, this embodiment provides a method for fault identification of power transmission system based on quantum neural network, which improves computing efficiency and reduces computing cost by introducing adaptive quantum circuit (ADQC) layer as the core processing layer and utilizing the parallel processing capability and efficient data representation of quantum computing. This method is not only suitable for processing complex and diverse electrical signals, but also particularly suitable for meeting the real-time fault identification requirements in power grid systems. The specific process of this method is as follows: Figure 1 As shown, the detailed steps include:

[0060] S1. Collect electrical data and pre-process them.

[0061] Specifically, electrical data includes current data and voltage data. The detailed preprocessing steps are as follows:

[0062] S11, extracting the characteristics of the electrical data, obtaining characteristic data and screening out the maximum characteristic data and the minimum characteristic data in the characteristic data;

[0063] S12, normalize the characteristic data to data between [0,1] to facilitate the subsequent quantum state mapping process, wherein the normalization expression is:

[0064]

[0065] x′ represents the preprocessed electrical data, x min Represents the minimum feature data, x max represents the maximum value feature data, and x represents the feature data.

[0066] S2. Map the preprocessed electrical data to the quantum computing space to obtain quantum state data.

[0067] The preprocessed electrical data is converted into a quantum state through a specific quantum state mapping function. Feature mapping is a key step in embedding classical data into quantum computing space. In this embodiment, cosine and sine functions are used to convert input data into a quantum state. The expression of the mapping process is:

[0068] Φ(x′)=cos(x′)|0>+sin(x′)|1>,

[0069] Among them, Φ(x′) represents quantum state data, x′ represents preprocessed electrical data, and |1> and |0> represent the two ground states of qubits.

[0070] S3. Input the quantum state data into the quantum neural network and output the fault identification result.

[0071] Among them, the structure of the quantum neural network is as Figure 3 shown, specifically including an input layer, an adaptive quantum circuit layer, and an output layer; the adaptive quantum circuit layer is used to evolve the quantum state data and calculate the fault occurrence probability based on the evolved quantum state data, including multiple evolution modules, each evolution module includes a two-layer gate structure, each layer of the gate structure includes multiple quantum gates and the previous gate structure has one more quantum gate than the latter gate structure.

[0072] The detailed process of this step is as follows:

[0073] S31. The quantum state data is transmitted to the adaptive quantum circuit layer through the input layer and undergoes quantum evolution in the adaptive quantum circuit layer to generate an output quantum state.

[0074] The detailed process of this step is described in combination with Figure 3 as follows:

[0075] S311. Assume that the first layer of the gate structure of each evolution module includes N quantum gates, then the number of quantum gates in the second layer of the gate structure is N - 1, the number of evolution modules is K, each quantum gate receives its input and outputs the first data and the second data, and the first data and the second data are exactly the same.

[0076] S312. Each quantum gate in the first layer of the gate structure of the first-level evolution module receives the inner product of at least two random quantum state data as the input.

[0077] S313. The first quantum gate in the first layer of the gate structure of the kth (1 < k ≤ K) - level evolution module receives the inner product of the first data output by the first quantum gate in the first layer of the (k - 1)th - level evolution module and the first data output by the first quantum gate in the second layer of the (k - 1)th - level evolution module as the input, the nth (1 < n < N) quantum gate receives the inner product of the second data output by the (n - 1)th quantum gate in the second layer of the (k - 1)th - level evolution module and the first data output by the nth quantum gate as the input, and the Nth quantum gate receives the second data output by the Nth quantum gate in the first layer of the (k - 1)th - level evolution module and the second data output by the (N - 1)th quantum gate in the second layer of the (k - 1)th - level evolution module as the input.

[0078] S314, the nth (1≤n≤N-1)th quantum gate in the second layer gate structure of the kth (1≤k≤K)th level evolution module receives the inner product of the second data of the nth gate structure of the first layer gate structure in the same level evolution module and the first data of the n+1th gate structure as input;

[0079] S315, the first data output by the first quantum gate in the first layer gate structure of the K-th level evolution module, the second data output by the N-th quantum gate, and the data output by the second layer gate structure are taken as the output quantum state.

[0080] S32. Output a fault identification result based on the output quantum state.

[0081] Specifically, the fault identification result is the probability of fault occurrence, which is calculated as follows: p represents the probability of failure, represents the measurement operator, and ψ represents the output quantum state.

[0082] In this embodiment, the training method of the quantum neural network is a combination of offline training and online learning, wherein the process of offline training is:

[0083] A1. Obtain historical electrical data and process the data to generate a fault data set. The fault data is divided into a training set and a test set at a ratio of 8:2. Each data in the fault data set includes historical electrical data and a fault label, and the fault label is 01 type data. When the fault label is 0, it indicates no fault, and when the fault label is 1, it indicates a fault.

[0084] A2. Initialize the parameters of the quantum neural network.

[0085] A3. Use the training set as the input of the quantum neural network, use the quantum gates in the quantum neural network to calculate and transform the training set, and output the probability of fault occurrence.

[0086] A4. Use the loss function to calculate the loss value between the fault probability and the fault label, and use the gradient descent method to update the quantum gate parameters based on the loss value. The expression of the loss function is:

[0087]

[0088] L(θ) represents the loss value, y i represents the fault label of the i-th sample in the training set, p i represents the probability of failure of the i-th sample, θ represents the quantum gate parameter, and N represents the total number of samples in the training set.

[0089] The expression of the gradient descent method is:

[0090]

[0091] θ′ represents the updated quantum gate parameters, θ represents the original quantum gate parameters, and η represents the learning rate.

[0092] A5. Repeat A3-A4 until the performance of the quantum neural network on the training set is stable, and use the test set to evaluate the performance of the trained quantum neural network. If the performance meets the standard, the training is terminated, otherwise the training is restarted.

[0093] The above is an introduction to the method embodiment. The present application also provides a system for fault identification of a power transmission system based on a quantum neural network for implementing the above method. Technical personnel in the relevant field can clearly understand that for the convenience and conciseness of description, the specific working process described can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0094] The system of the embodiment includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) or loaded from a storage unit into a random access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The CPU, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0095] Multiple components in the device are connected to the I / O interface, including: input units, such as keyboards, mice, etc.; output units, such as various types of displays, speakers, etc.; storage units, such as disks, optical disks, etc.; and communication units, such as network cards, modems, wireless communication transceivers, etc. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunication networks.

[0096] The processing unit performs the various methods and processes described above, such as methods S1 to S3 and methods A1 to A5. For example, in some embodiments, methods S1 to S3 and methods A1 to A5 may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via a ROM and / or a communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of methods S1 to S3 and methods A1 to A5 described above may be executed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S3 and methods A1 to A5 by any other appropriate means (e.g., by means of firmware).

[0097] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0098] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0099] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0100] Example 2

[0101] In order to verify the feasibility of the method provided in Example 1, in this embodiment, the method provided in the above embodiment and the method based on a fully connected neural network are used simultaneously to perform fault identification on the same power system.

[0102] Among them, the fully connected neural network (FCNN) is a classic deep learning model that contains multiple hidden layers and adjusts the weights and biases in the network through the back propagation algorithm. In this embodiment, FCNN is used as a benchmark model to compare the performance with the quantum neural network.

[0103] In the experimental process of this embodiment, the training and testing of the two method models are carried out under the same conditions. Specifically, the training set data includes 9601 data, the test set includes 2400 data, the algorithm is trained 200 times, the trainers used by the two algorithms are Adam optimizers, and the parameter updating methods are gradient updates; and the hidden layer of FCNN is set to 50 layers, and the loss function is the cross entropy loss function.

[0104] During both the training and testing processes, the model performance was evaluated through accuracy, and the confusion matrix was drawn to analyze the recognition effects of the two methods in detail.

[0105] Specifically, the accuracy results of the two methods are shown in Table 1. QNN has higher accuracy in fault identification. In order to further analyze the algorithm performance, the confusion matrices of QNN and FCNN on the test set are plotted respectively, as shown in Table 1. Figure 3 and Figure 4 As shown, compared with FCNN, which has 254 false positive results, QNN has 146 false negative data, which is smaller than FCNN.

[0106] It can be seen that although the structure of FCNN is relatively simple and mainly relies on the linear combination of multiple layers of neurons to learn data features, there are certain limitations in the accuracy of FCNN fault recognition when facing high-dimensional data. The method provided by the above embodiment is more accurate and reliable.

[0107] Table 1 Accuracy results data table

[0108] QNN FCNN Training set accuracy (%) 93.04 88.32 Test set accuracy (%) 93.92 89.42

[0109] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for fault identification of power transmission system based on quantum neural network, characterized in that: The method includes: Collecting electrical data and performing preprocessing, where the electrical data includes current data and voltage data; Mapping the preprocessed electrical data to a quantum computing space to obtain quantum state data; Inputting the quantum state data into a quantum neural network to output a fault identification result, where the quantum neural network includes an input layer, an adaptive quantum circuit layer, and an output layer; the adaptive quantum circuit layer is used to evolve the quantum state data and calculate the probability of a fault occurring based on the evolved quantum state data, and includes multiple evolution modules, each evolution module includes a two-layer gate structure, each layer of the gate structure includes multiple quantum gates, and the previous gate structure has one more quantum gate than the latter gate structure.

2. The method for fault identification of power transmission system based on quantum neural network according to claim 1 is characterized in that: The preprocessing includes: Extracting the features of the electrical data to obtain feature data and screening out the maximum feature data and minimum feature data in the feature data; Normalizing the feature data to data between [0, 1], where the expression for normalization is: x′ represents the preprocessed electrical data, x min Represents the minimum feature data, x max represents the maximum value feature data, and x represents the feature data.

3. The method for fault identification of power transmission system based on quantum neural network according to claim 1 is characterized in that: The method for obtaining the quantum state data is: Mapping each preprocessed electrical data to the quantum computing space using a quantum state mapping function, and the expression for the mapping process is: Φ(x′) = cos(x′)|0> + sin(x′)|1>, where Φ(x′) represents the quantum state data, x′ represents the preprocessed electrical data, and |1> and |0> represent the two ground states of the quantum bit.

4. The method for fault identification of power transmission system based on quantum neural network according to claim 1 is characterized in that: The method for outputting the fault identification result is: The quantum state data is transmitted through the input layer to the adaptive quantum circuit layer, and quantum evolution is performed in the adaptive quantum circuit layer to generate an output quantum state; Calculating the probability of a fault occurring based on the output quantum state, and the probability of a fault occurring is the fault identification result.

5. The method for fault identification of power transmission system based on quantum neural network according to claim 4 is characterized in that: The process of the quantum evolution is: Assume that the first layer of the gate structure of each evolution module includes N quantum gates, then the number of quantum gates in the second layer of the gate structure is N - 1, the number of evolution modules is K, each quantum gate receives its input and outputs first data and second data, and the first data and the second data are exactly the same; Each quantum gate in the first layer of the gate structure of the first-level evolution module receives the inner product of at least two random quantum state data as input; The first quantum gate in the first layer of the gate structure of the kth (1 < k ≤ K) - level evolution module receives the inner product of the first data output by the first quantum gate in the first layer of the (k - 1)th - level evolution module and the first data output by the first quantum gate in the second layer of the (k - 1)th - level evolution module as input, the nth (1 < n < N) quantum gate receives the inner product of the second data output by the (n - 1)th quantum gate in the second layer of the (k - 1)th - level evolution module and the first data output by the nth quantum gate as input, and the Nth quantum gate receives the second data output by the Nth quantum gate in the first layer of the (k - 1)th - level evolution module and the second data output by the (N - 1)th quantum gate in the second layer of the (k - 1)th - level evolution module as input; The nth (1≤n≤N-1)th quantum gate in the second layer gate structure of the kth (1≤k≤K)th level evolution module receives the inner product of the second data of the nth gate structure of the first layer gate structure in the same level evolution module and the first data of the n+1th gate structure as input; The first data output by the first quantum gate in the first-layer gate structure of the K-th level evolution module, the second data output by the N-th quantum gate, and the data output by the second-layer gate structure are used as the output quantum state.

6. The method for fault identification of power transmission system based on quantum neural network according to claim 4 is characterized in that: The method for calculating the probability of failure based on the output quantum state is: Where p represents the probability of failure. represents the measurement operator, and ψ represents the output quantum state.

7. The method for fault identification of power transmission system based on quantum neural network according to claim 1, characterized in that: The training method of the quantum neural network is a combination of offline training and online learning, wherein the process of offline training is: Acquire historical electrical data and perform data processing to generate a fault data set, and divide the fault data into a training set and a test set according to a preset ratio, wherein each data in the fault data set includes historical electrical data and a fault label, and the fault label is 01 type data, when the fault label is 0, it indicates no fault, and when the fault label is 1, it indicates a fault; Initialize the parameters of the quantum neural network; The training set is used as the input of the quantum neural network, and the quantum gates in the quantum neural network are used to calculate and transform the training set, and the probability of fault occurrence is output; The loss function is used to calculate the loss value between the fault probability and the fault label, and the quantum gate parameters are updated based on the loss value; Repeat this process until the performance of the quantum neural network on the training set is stable, and use the test set to evaluate the performance of the trained quantum neural network. If the performance meets the standard, the training is terminated, otherwise the training is restarted.

8. The method for fault identification of power transmission system based on quantum neural network according to claim 7 is characterized in that: The loss function expression is: L(θ) represents the loss value, y i represents the fault label of the i-th sample in the training set, p i represents the probability of failure of the i-th sample, θ represents the quantum gate parameter, and N represents the total number of samples in the training set.

9. The method for fault identification of power transmission system based on quantum neural network according to claim 7, characterized in that: The method for updating the quantum gate parameters is the gradient descent method, and its expression is: θ′ represents the updated quantum gate parameters, θ represents the original quantum gate parameters, and η represents the learning rate.

10. A system for fault identification of power transmission system based on quantum neural network, characterized in that: The system is used to implement the method described in any one of claims 1 to 9.

Images