A method and system for early warning of power system equipment failure

Abstract

The present application relates to the field of electric power technology, and discloses a method and system for early warning of power system equipment failures. The method comprises obtaining fault information of power equipment in a provincial power grid system as regional sample data; performing data preprocessing on the regional sample data to eliminate interference data in the regional sample data; training a teacher model using the regional sample data to obtain a trained teacher model; constructing relationship constraints between the teacher model and the student model, and constructing a local student model based on the relationship constraints; obtaining local sample data, and training the local student model based on the regional sample data and the local sample data; and performing early warning of failures on local power system equipment based on the trained local student model. The method uses data from the provincial power grid system to train a teacher network to perform model teacher model training. A student network is used in the regional power grid, so that the system is more suitable for the local environment and can identify special faults.

Description

Technical Field

[0001] The present invention relates to the field of electric power technology, and in particular to a method and system for early warning of power system equipment failure. Background Art

[0002] my country's power system adopts a hierarchical management model. Provincial power grids, as a crucial component of the national grid, are responsible for regional power resource allocation and backbone grid construction. Provincial power grid companies utilize main grids with voltage levels of 220 kV and above to optimize the allocation of power resources within their province and form interconnected regional power grids with neighboring provinces. Regional power grids span multiple provinces, linking them with ultra-high voltage grids of 500 kV and above to establish interprovincial power transmission channels, enabling greater scale in adjusting power surpluses and shortages and accommodating clean energy consumption.

[0003] During the implementation of the embodiments of the present application, it was found that at least the following problems exist in the related art:

[0004] Power grids are distributed across various regions. Weather conditions and grid age vary from region to region. Consequently, the fault profiles of power cables, transformers, and other electrical equipment vary significantly depending on the environment. Applying the same fault analysis method to all regions is inefficient and can easily overlook specific faults. Designing a separate system for each region would also be time-consuming.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of this application, and therefore may include information that does not constitute prior art known to ordinary technicians in this field. Summary of the Invention

[0006] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical elements or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0007] The embodiments of the present application provide a method and system for early warning of power system equipment failures to improve the accuracy of power grid system failure prediction and its adaptability to the environment.

[0008] In some embodiments, the power system equipment fault warning method includes: obtaining fault information of power equipment in a provincial power grid system as regional sample data; performing data preprocessing on the regional sample data to eliminate interference data in the regional sample data; using the regional sample data to train a teacher model to obtain a trained teacher model; constructing relationship constraints between the teacher model and the student model, and constructing a local student model based on the relationship constraints; obtaining local sample data, and training the local student model based on the regional sample data and the local sample data; and performing fault warning on local power system equipment based on the trained local student model.

[0009] Optionally, performing data preprocessing on the regional sample data includes: performing SMOTE oversampling on the regional sample data; and performing filtering processing on the oversampled regional sample data.

[0010] Optionally, SMOTE oversampling is performed on the regional sample data, including: calculating the k-nearest neighbor density Wi of each minority class sample; allocating sampling probability according to the inverse of the density; and adjusting weight sensitivity according to an exponential decay factor.

[0011] Optionally, the regional teacher model includes a main loss function and a slave loss function, the main loss function is a cross entropy loss function, and the slave loss function is dynamically selected from a loss function library.

[0012] Optionally, the method of selecting the loss function includes: selecting an initial loss function from the loss function library according to preset rules; calculating the weighted value of the initial loss function and the main loss function, and when the convergence value of the weighted value is less than the preset convergence for N consecutive times, selecting other loss functions from the loss function library for retraining.

[0013] Optionally, obtaining local sample data includes: obtaining local initial samples; calculating the similarity of each initial sample in the regional samples, and deleting the initial sample if the similarity is greater than a similarity threshold; and determining the deleted local initial sample as the local sample data.

[0014] Optionally, it also includes: using a feature extraction network to encode local sample data and sample data from other regions in the region to generate high-dimensional feature vectors; calculating the degree of matching between the feature vectors, and when the matching degree is higher than the matching threshold, directly using the student model of the region as the local student model.

[0015] In some embodiments, the power system equipment fault warning system includes: an acquisition module, used to obtain fault information of power equipment in the provincial power grid system as regional sample data; a preprocessing module, used to perform data preprocessing on the regional sample data to eliminate interference data in the regional sample data; a first training module, used to train the teacher model using the regional sample data to obtain a trained teacher model; a construction module, used to construct relationship constraints between the teacher model and the student model, and construct a local student model based on the relationship constraints; a second training module, used to obtain local sample data, and train the local student model based on the regional sample data and the local sample data; and a warning module, used to perform fault warning on local power system equipment based on the trained local student model.

[0016] In some embodiments, an electronic device is also provided, including a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the above-mentioned power system equipment fault early warning method.

[0017] In some embodiments, a computer-readable storage medium is also provided, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the above-mentioned power system equipment fault warning method.

[0018] The fault warning method and system for power system equipment provided in the embodiments of the present application can achieve the following technical effects:

[0019] The teacher network is trained using data from the provincial power grid system. Fault data from power system equipment across the province is collected and used to train the teacher model. The student network is then deployed in regional power grids to quickly identify basic faults. The student network model is then continuously refined based on local fault information. This ensures that the system can identify basic faults while also being more adaptable to local environments and capable of identifying specific faults.

[0020] The above general description and the following description are exemplary and explanatory only and are not intended to limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] One or more embodiments are exemplarily described by corresponding drawings. These exemplary descriptions and drawings do not limit the embodiments. Elements with the same reference numerals in the drawings are shown as similar elements. The drawings do not constitute a scale limitation. In addition,

[0022] Figure 1 This is a flow chart of a method for early warning of power system equipment failure according to an embodiment of the present application;

[0023] Figure 2 Schematic diagram of a power system equipment fault warning system according to an embodiment of the present application;

[0024] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to be able to understand the features and technical contents of the embodiments of the present application in more detail, the implementation of the embodiments of the present application is described in detail below in conjunction with the accompanying drawings. The accompanying drawings are for reference only and are not used to limit the embodiments of the present application. In the following technical description, for the sake of convenience of explanation, a full understanding of the disclosed embodiments is provided through multiple details. However, one or more embodiments can still be implemented without these details. In other cases, to simplify the drawings, well-known structures and devices can be simplified for display.

[0026] In the description and claims of the embodiments of the present application and the accompanying drawings, the terms "first," "second," and the like are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate for the purposes of describing the embodiments of the present application. In addition, the terms "including," "having," and any variations thereof are intended to cover non-exclusive inclusions.

[0027] Unless otherwise stated, the term "plurality" means two or more.

[0028] In the embodiments of the present application, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B means: A or B.

[0029] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0030] Combine Figure 1 FIG. 1 is a flow chart of a method for early warning of power system equipment failure provided by an embodiment of the present application, the method comprising the following steps:

[0031] S101: Acquire fault information of power equipment in a provincial power grid system as regional sample data;

[0032] S102: performing data preprocessing on the regional sample data to eliminate interference data in the regional sample data;

[0033] S103: Using the regional sample data to train the teacher model to obtain a trained teacher model;

[0034] S104: Constructing relationship constraints between the teacher model and the student model, and constructing a local student model based on the relationship constraints;

[0035] S105: Acquire local sample data, and train the local student model based on the regional sample data and the local sample data;

[0036] S106: Providing fault warning for local power system equipment based on the trained local student model.

[0037] The power system equipment fault warning method and system proposed in the present application utilizes data from the provincial power grid system to train a teacher network. This network collects fault data from power system equipment in all regions of the province and trains the teacher model. Using a student network within the regional power grid allows for rapid identification of basic faults and continuous improvement of the student network model based on local fault information. This ensures that the system can identify basic faults while also making it more adaptable to the local environment and capable of identifying specific faults.

[0038] Specifically, it is necessary to set up a main server in the provincial power grid system to collect fault information of all power equipment in the provincial power grid area. Fault information includes fault type, occurrence time, weather conditions before and after the occurrence, regional characteristics, solutions, etc. The power system equipment in the embodiment of the present application includes but is not limited to power cables and transformers. In order to comprehensively obtain regional power grid fault sample data, it is first necessary to establish a unified data acquisition platform in the provincial power grid system. The platform uses the provincial main server as the core node and interconnects with the monitoring systems of local power supply companies through standardized interfaces. The main server will collect fault information of various power facilities such as substations, transmission lines, and distribution equipment within the jurisdiction in real time, including basic data such as fault type (such as short circuit, grounding, overload, etc.), fault occurrence time accurate to milliseconds, and abnormal records of equipment operating parameters. At the same time, it will automatically link the real-time weather data of the meteorological department to record environmental parameters such as temperature, humidity, wind speed, and rainfall before and after the fault occurs.

[0039] During data collection, the system prioritizes the integration and correlation of multidimensional information. In addition to device fault information, it also collects geographically relevant data, such as the fault's geographic coordinates, altitude, and surrounding vegetation coverage, for spatial overlay analysis with the GIS (Geographic Information System). For each fault resolution, operations and maintenance personnel use mobile devices to record a detailed description of the fault, on-site inspection results, solutions implemented (such as equipment replacement, parameter adjustments, and isolation), and the final time power was restored, creating a complete, closed-loop record of fault resolution. The system also automatically links historical maintenance records, family defect information, and other archival data for the device, providing context for subsequent analysis.

[0040] To ensure data quality, the provincial master server will deploy a data verification mechanism to cleanse and standardize collected information through technical means such as logical rule judgment and data integrity checks. The system will establish a fault sample database, categorized and stored according to dimensions such as voltage level, equipment type, and fault nature, and regularly generate data quality assessment reports. At the same time, the platform supports data exchange with power grid systems in neighboring provinces, enabling the sharing and complementarity of fault samples at the regional power grid level, providing a more comprehensive sample foundation for subsequent big data analysis. All data collection and processing processes adhere to power industry data security standards to ensure information security and privacy protection.

[0041] Furthermore, for transformers, fault information includes core features based on dissolved gas analysis (DGA) data (including the concentrations of five gases, including H2, CH4, and C2H2). For cables, fault information includes time-domain and frequency-domain features such as three-phase current, voltage harmonics, and partial discharge signals.

[0042] In some embodiments, the data preprocessing of the regional sample data in step S202 includes: performing SMOTE oversampling on the regional sample data; and performing filtering on the oversampled regional sample data.

[0043] SMOTE (Synthetic Minority Over-sampling Technique) is a classic algorithm for dealing with class imbalance. Its core idea is to generate synthetic samples by interpolating in the feature space, rather than simply copying minority class samples. This embodiment of the application uses an improved SMOTE algorithm to oversample minority class samples. Specifically, first, the k-nearest neighbor density Wi of each minority class sample is calculated; then, the following sampling probabilities are assigned according to the inverse of the density.

[0044]

[0045] Finally, adjust the weight sensitivity according to an exponential decay factor:

[0046]

[0047] The parameter α > 0 controls the intensity of the influence of density on probability (a larger α is, the more emphasis is placed on sparse regions). This embodiment of the application improves the SMOTE algorithm by dynamically adjusting the sampling weights based on the local density of samples, thereby generating more samples in sparse regions. This improves sample diversity.

[0048] Furthermore, due to the particularity of the power industry, the interference signal is relatively strong, so the collected initial data is subjected to anti-interference processing, specifically, applying Kalman filtering to the current signal.

[0049] In some instances, the method provided in the embodiment of the present application adopts a dynamic master-slave loss function to construct a teacher model. Among them, the master loss function adopts the cross entropy loss function, and the slave loss function is dynamically selected from the loss function library. In this way, the loss function can be dynamically adjusted.

[0050] The specific selection method is to select another loss function from the loss function library and retrain N times if the loss value does not converge significantly after N consecutive times. The loss function library may include but is not limited to: contrast loss, which is used to enhance the aggregation of similar fault features, or focal loss, which is used to alleviate the imbalance problem of difficult and easy samples.

[0051] The above-mentioned embodiment adopts a dynamic master-slave loss function mechanism, ensuring basic convergence through the cross-entropy master loss function. It also dynamically selects auxiliary objectives from a loss function library (such as Focal Loss to address class imbalance or KL divergence for knowledge distillation). This allows for adaptive adjustment of optimization directions, significantly improving the model's generalization capabilities under different data distributions. This design not only enhances training stability (such as balancing accuracy and robustness) through multi-objective collaborative optimization, but also reduces reliance on manual parameter adjustment. In particular, within the teacher-student framework, it allows for flexible matching of feature differences, effectively improving knowledge transfer efficiency.

[0052] In some instances, step S204 in the above embodiment may specifically include: constructing a relationship constraint between the teacher network and the student network, and determining an initial local student network based on the relationship constraint.

[0053] First, construct the knowledge distillation loss function, which can be expressed as:

[0054]

[0055] Among them, Ft is the feature set of the teacher model, Fs is the feature set of the student model, and is a similarity function. Lm is the correlation function between the teacher model and the student model. The student model is constructed based on the knowledge distillation loss function. This knowledge distillation loss function establishes an identity relationship between the teacher network and the student network, allowing the student network to continuously learn relevant knowledge from the teacher network.

[0056] In some instances, the above method may further include: obtaining local initial samples; calculating the similarity of each initial sample in the regional samples, and deleting the initial sample if the similarity is greater than a similarity threshold; and determining the deleted local initial sample as local sample data.

[0057] Specifically, a feature extraction network can be used to encode local sample data and sample data from other regions in the region to generate high-dimensional feature vectors; the degree of matching between the feature vectors is calculated, and when the matching degree is higher than the matching threshold, the student model of the region is directly used as the local student model.

[0058] The above example can achieve cross-region model migration through feature similarity measurement, for example, using the following cosine similarity formula for feature matching:

[0059]

[0060] in, represents the feature extraction network, S and T are local samples and regional samples, is the threshold.

[0061] This formula eliminates feature scale differences through normalization and can effectively evaluate distribution similarity. It can be further extended to weighted multi-level feature matching.

[0062] In some feasible embodiments, the above method may also include: using a feature extraction network to encode local sample data and sample data from other regions in the region to generate high-dimensional feature vectors; calculating the degree of matching between the feature vectors, and when the degree of matching is higher than the matching threshold, directly using the student model of the region as the local student model.

[0063] Specifically, the matching degree calculation can be performed using a feature extraction network (such as a Siamese network or contrastive learning model) to encode local and other regional fault data to generate a high-dimensional feature vector. The matching degree of fault features in different regions is measured using cosine similarity or Euclidean distance to quantify their distribution similarity.

[0064] Furthermore, if the matching degree between a region's fault characteristics and the local model exceeds a preset threshold (e.g., 0.85), indicating that the two fault patterns are highly similar, the pre-trained student network model for that region can be directly used as the initial local student model, avoiding training from scratch. Otherwise, it is necessary to independently initialize or select the regional model with the highest matching degree for fine-tuning.

[0065] Based on the above method embodiment, the present invention also provides a power system equipment failure early warning system, see Figure 2 As shown, the system 200 includes:

[0066] An acquisition module 201 is used to acquire fault information of power equipment in a provincial power grid system as regional sample data;

[0067] A preprocessing module 202 is configured to perform data preprocessing on the regional sample data to eliminate interference data in the regional sample data;

[0068] A first training module 203 is used to train the teacher model using the regional sample data to obtain a trained teacher model;

[0069] A construction module 204 is used to construct relationship constraints between the teacher model and the student model, and to construct a local student model based on the relationship constraints;

[0070] A second training module 205 is configured to obtain local sample data and train the local student model based on the regional sample data and the local sample data;

[0071] The early warning module 206 is used to provide fault early warning for local power system equipment based on the trained local student model.

[0072] The power system equipment fault warning system proposed in this embodiment utilizes data from the provincial power grid system to train a teacher network. This network collects fault data from power system equipment in all regions of the province and trains the teacher model. Using a student network within the regional power grid allows for rapid identification of basic faults and continuous improvement of the student network model based on local fault information. This ensures that the system can identify basic faults while also making it more adaptable to the local environment and capable of identifying specific faults.

[0073] The embodiment of the present invention further provides an electronic device, such as Figure 3 As shown, it is a structural diagram of the electronic device, wherein the electronic device includes a processor 301 and a memory 302, the memory 302 stores computer executable instructions that can be executed by the processor 301, and the processor 301 executes the computer executable instructions to implement the above-mentioned neural network training method and target recognition method.

[0074] exist Figure 3 In the illustrated embodiment, the electronic device further includes a bus 303 and a communication interface 304 , wherein the processor 301 , the communication interface 304 and the memory 302 are connected via the bus 303 .

[0075] Among them, the memory 302 may include high-speed random access memory (RAM), and may also include non-volatile memory (non-volatile memory), such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 304 (which can be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. can be used. The bus 303 can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 403 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0076] The processor 301 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in the processor 301 or by software instructions. The above processor 301 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present invention can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor 301 reads the information in the memory and combines its hardware to complete the steps of the neural network training method and the target recognition method of the aforementioned embodiment.

[0077] An embodiment of the present invention also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the above-mentioned neural network training method and target recognition method. The specific implementation can be found in the aforementioned method embodiment, which will not be repeated here.

[0078] The above description and the accompanying drawings sufficiently illustrate the embodiments of the present application to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible variations. Unless expressly required, individual components and functions are optional, and the order of operations may vary. Portions and features of some embodiments may be included in or replace portions and features of other embodiments. Moreover, the terms used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and claims, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. Similarly, the term "and / or" as used in this application means any and all possible combinations of one or more of the associated listings. In addition, when used in this application, the term "comprise" and its variations "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, an element defined by the sentence "comprising a..." does not exclude the presence of other identical elements in the process, method or device that includes the element. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments can be referenced to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can be referred to the description of the method part.

[0079] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present application. The technicians will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0080] In the embodiments disclosed herein, the disclosed methods and products (including but not limited to devices, equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units may be merely a logical functional division. In actual implementation, other division methods may be used, such as multiple units or components being combined or integrated into another system, or some features being ignored or not implemented. In addition, the coupling or direct coupling or communication connection shown or discussed between each other may be through some interface, indirect coupling or communication connection between devices or units, and may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected to implement the present embodiments according to actual needs. In addition, the functional units in the embodiments of the present application may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit.

[0081] The flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to the embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the part of the module, program segment or code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in an order different from that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified function or action, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A power system equipment failure early warning method, characterized in that: Methods include: Obtain fault information of power equipment in the provincial power grid system as regional sample data; Performing data preprocessing on the regional sample data to eliminate interference data in the regional sample data; Using the regional sample data to train the teacher model to obtain a trained teacher model; Constructing a relationship constraint between the teacher model and the student model, and constructing a local student model based on the relationship constraint; Acquire local sample data, and train the local student model based on the regional sample data and the local sample data; Provide fault warning for local power system equipment based on the trained local student model; The teacher model includes a main loss function and a slave loss function, the main loss function is a cross entropy loss function, and the slave loss function is dynamically selected from a loss function library; The selection method of the loss function includes: According to the preset rules, select an initial loss function from the loss function library; Calculate the weighted values ​​of the initial loss function and the main loss function. When the convergence value of the weighted value is not less than the preset convergence value for N consecutive times, select other loss functions from the loss function library for retraining.

2. The power system equipment failure early warning method according to claim 1, characterized in that: Performing data preprocessing on the regional sample data, including: Performing SMOTE oversampling on the sample data of the region; The oversampled region sample data is filtered.

3. The power system equipment failure early warning method according to claim 2, characterized in that: Performing SMOTE oversampling on the sample data of the region includes: Calculate the K nearest neighbor density of each minority class sample; Assigning sampling probability according to the inverse of the K nearest neighbor density of each minority class sample; The weight sensitivity is adjusted according to the sampling probability and the exponential decay factor.

4. The power system equipment failure early warning method according to claim 1, characterized in that: Get local sample data, including: Get local initial samples; Calculating the similarity of each initial sample in the regional samples, and deleting the initial sample if the similarity is greater than a similarity threshold; The deleted local initial sample is determined as the local sample data.

5. The power system equipment failure early warning method according to claim 1, characterized in that: Also includes: Using a feature extraction network to encode the local sample data and sample data from other regions within the region to generate a high-dimensional feature vector; The matching degree between the high-dimensional feature vectors is calculated, and when the matching degree is higher than a matching threshold, the student model of the region is directly used as the local student model.

6. A power system equipment failure early warning system, characterized in that: include: An acquisition module, used to acquire fault information of power equipment in a provincial power grid system as regional sample data; A preprocessing module, configured to perform data preprocessing on the regional sample data to eliminate interference data in the regional sample data; A first training module is used to train the teacher model using the regional sample data to obtain a trained teacher model; A construction module, configured to construct relationship constraints between a teacher model and a student model, and to construct a local student model based on the relationship constraints; A second training module is configured to obtain local sample data and train the local student model based on the regional sample data and the local sample data; An early warning module is used to provide fault warnings for local power system equipment based on the trained local student model; The teacher model includes a main loss function and a slave loss function, the main loss function is a cross entropy loss function, and the slave loss function is dynamically selected from a loss function library; The selection method of the loss function includes: According to the preset rules, select an initial loss function from the loss function library; Calculate the weighted values ​​of the initial loss function and the main loss function. When the convergence value of the weighted value is not less than the preset convergence value for N consecutive times, select other loss functions from the loss function library for retraining.

7. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method according to any one of claims 1 to 5.

Images