A distribution network whole-process operation state early warning method

Abstract

The present application relates to a kind of distribution network whole process operating state early warning method, comprising the following steps: collecting the operating characteristic data of each power grid node in distribution network whole process;Set evaluation time range, select multiple time nodes in evaluation time range to obtain the operating characteristic data of each power grid node;The operating characteristic data of each power grid node obtained from each time node is formed into feature vector;Based on each feature vector, the operating state information entropy value of each time node is calculated, and the operating state feature time sequence is synthesized by the operating state information entropy value of each time node calculated;Operating state feature time sequence is used as the input of neural network model pre-trained, and corresponding early warning score is output, and the early warning score indicates the probability of failure of distribution network in the given length of prediction time range after evaluation time range.

Description

TECHNICAL FIELD

[0001] The present application relates to a power distribution network full-process operation state early warning method, belonging to the technical field of power distribution network state early warning. BACKGROUND

[0002] As an important component of the power system, the purpose of the power distribution network state early warning system is to discover and warn potential safety hazards in advance through real-time monitoring and analysis of the operation state of the power distribution network, so as to ensure the safe and stable operation of the power system. However, there are still some obvious defects in the application process of the existing power distribution network state early warning system, which not only affects the accuracy and reliability of the early warning system, but also may pose a potential threat to the safe operation of the power system.

[0003] Firstly, there are limitations in data acquisition and processing of the early warning system. Power distribution network state early warning requires real-time acquisition and processing of a large amount of power grid data, including voltage, current, power factor and other key parameters. However, due to the limited coverage of data acquisition equipment or the limitations of data transmission technology, data in some areas may not be accurately and timely acquired. In addition, the complexity and real-time requirements of data processing algorithms also pose challenges to the performance of the early warning system.

[0004] Secondly, the fault location accuracy of the early warning system needs to be improved. In the power distribution network, faults can occur anywhere and the fault types are diverse. However, due to the fact that the algorithms and models of the early warning system may not fully adapt to all situations, it may not be able to accurately locate the fault location in some cases. This not only increases the difficulty and time cost of fault repair, but also may pose a potential threat to the stable operation of the power distribution network.

[0005] Thirdly, the real-time response and decision-making ability of the early warning system is limited. Power distribution network state early warning requires timely response and decision-making to quickly take measures for repair when a fault occurs. However, due to the processing speed and complexity of the decision-making algorithm of the early warning system, it may not be able to make correct decisions in time in emergency situations.

[0006] In addition, the intelligent level of the early warning system needs to be improved. At present, many power distribution network state early warning systems still rely on manual monitoring and judgment, lacking sufficient intelligence and automation capabilities. This not only increases the labor cost, but also may lead to misjudgment or omission due to human factors. SUMMARY

[0007] In order to solve the problems existing in the prior art, the present application proposes a power distribution network full-process operation state early warning method.

[0008] The technical solution of the present application is as follows:

[0009] In one aspect, the present application provides a power distribution network full-process running state early warning method, comprising the following steps:

[0010] Collecting running characteristic data of each power grid node in the power distribution network full process;

[0011] Setting an evaluation time range, selecting multiple time nodes within the evaluation time range to obtain the running characteristic data of each power grid node; and forming a feature vector from the running characteristic data of each power grid node obtained from each time node;

[0012] Calculating the running state information entropy value of each time node based on each feature vector, and synthesizing a running state feature time sequence from the calculated running state information entropy values of each time node;

[0013] Using the running state feature time sequence as the input of a pre-trained neural network model, and outputting a corresponding early warning score, the early warning score indicating the probability of a power distribution network failure in a given length of a prediction time range after the evaluation time range.

[0014] As a preferred embodiment, the step of calculating the running state information entropy value of each time node based on each feature vector is specifically:

[0015] Setting that the feature vector and the node running failure probability conform to a Gaussian distribution:

[0016] P(X / Qi)=G(Qi,σ 2 );

[0017] Wherein, X represents a target power distribution network, Qi is the i-th feature vector, P(X / Qi) represents the failure probability of the power distribution network when the feature vector is Qi; G(·,σ 2 ) represents a Gaussian distribution function;

[0018] Calculating the corresponding information entropy value based on the feature vector:

[0019]

[0020] Wherein, m is the number of feature vectors, corresponding to the number of time nodes selected within the evaluation time range.

[0021] As a preferred embodiment, in the step of using the running state feature time sequence as the input of a pre-trained neural network model and outputting a corresponding early warning score:

[0022] The pre-trained neural network model comprises a first encoder network, a second encoder network, a first decoder network, a second decoder network and a recurrent neural network;

[0023] The first encoder network performs a downsampling operation on the input operation state feature time sequence to generate an initial input with a first scale; the first decoder network decodes the initial input to generate an initial output;

[0024] The second decoder network performs an upsampling operation on the initial output to generate a secondary input with a second scale higher than the first scale; and the second encoder network encodes the secondary input to obtain a secondary output.

[0025] The recurrent neural network takes the secondary output as input and outputs a predicted early warning score.

[0026] As a preferred embodiment, the operation feature data of each power grid node in the whole process of the power distribution network includes voltage, current, power factor, load rate, temperature, humidity, rainfall, and equipment age.

[0027] In another aspect, the present application also provides a power distribution network whole-process operation state early warning system, comprising:

[0028] A data acquisition module for acquiring operation feature data of each power grid node in the whole process of the power distribution network;

[0029] A feature data construction module for setting an evaluation time range, selecting a plurality of time nodes to obtain operation feature data of each power grid node within the evaluation time range, and combining the operation feature data of each power grid node obtained from each time node into a feature vector;

[0030] A feature sequence construction module for calculating an operation state information entropy value of each time node based on each feature vector, and synthesizing an operation state feature time sequence from the calculated operation state information entropy values of each time node;

[0031] An early warning analysis module for using the operation state feature time sequence as input of a pre-trained neural network model, and outputting a corresponding early warning score, the early warning score indicating a probability of failure of the power distribution network within a given length of a prediction time range after the evaluation time range.

[0032] As a preferred embodiment, the step of calculating the operation state information entropy value of each time node based on each feature vector is specifically:

[0033] The feature vector and the node operation failure probability are set to conform to a Gaussian distribution:

[0034] P(X / Qi)=G(Qi,σ 2 );

[0035] Wherein, X represents a target power distribution network, Qi is the i-th feature vector, and P(X / Qi) represents a failure probability of the power distribution network when the feature vector is Qi; G(·,σ 2) represents a Gaussian distribution function;

[0036] The corresponding information entropy value is calculated based on the feature vector:

[0037]

[0038] Wherein, m is the number of feature vectors, corresponding to the number of time nodes selected in the evaluation time range.

[0039] As a preferred embodiment, in the step of using the running state feature time sequence as the input of the pre-trained neural network model, outputting the corresponding early warning score:

[0040] The pre-trained neural network model comprises a first encoder network, a second encoder network, a first decoder network, a second decoder network and a recurrent neural network.

[0041] Wherein, the first encoder network performs downsampling operation on the input running state feature time sequence to generate an initial input with a first scale; the first decoder network decodes the initial input to generate an initial output;

[0042] The second decoder network performs upsampling operation on the initial output to generate a secondary input with a second scale, the second scale being higher than the first scale; the second encoder network encodes the secondary input to obtain a secondary output;

[0043] The recurrent neural network takes the secondary output as the input and outputs the predicted early warning score.

[0044] As a preferred embodiment, the running feature data of each power grid node in the whole process of the power distribution network includes voltage, current, power factor, load rate, temperature, humidity, rainfall, and equipment age.

[0045] In another aspect, the present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to realize the power distribution network whole process running state early warning method according to any one of the embodiments of the present application.

[0046] In another aspect, the present application also provides a computer readable storage medium, which stores a computer program, wherein the program is executed by a processor to realize the power distribution network whole process running state early warning method according to any one of the embodiments of the present application.

[0047] The present application has the following advantages:

[0048] The application discloses a distribution network whole-process operation state early warning method, which comprises the following steps: acquiring operation characteristic data of each time node in an evaluation time range to form a characteristic vector; calculating an information entropy value based on the characteristic vector to indicate the importance of the characteristic vector to a fault, and synthesizing an operation state characteristic time sequence based on the calculated information entropy value, and inputting the operation state characteristic time sequence into a neural network model to obtain a corresponding early warning score, so that the time and feature importance are comprehensively considered to accurately early warn a distribution network operation fault. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 A method flowchart of the embodiment one of the application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the application will be clearly and completely described in connection with the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0051] It should be understood that the step numbers used herein are only for the convenience of description, and are not limited to the execution sequence of the steps.

[0052] It should be understood that the terms used in the specification of the application are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in the specification and the appended claims of the application, the singular forms "a", "an" and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0053] The terms "comprise" and "include" indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not exclude one or more other features, integers, steps, operations, elements, components, and / or sets thereof.

[0054] The term "and / or" means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0055] Embodiment one:

[0056] Referring to Figure 1 The embodiment provides a distribution network whole-process operation state early warning method, which comprises the following steps:

[0057] S100, collecting operation characteristic data of each power grid node in a distribution network whole process;

[0058] S200, set an evaluation time range, select a plurality of time nodes to obtain the operation characteristic data of each power grid node in the evaluation time range, the selection of the time nodes should follow a certain rule, such as division according to time units of hours, days, months, etc., so that we can comprehensively and carefully understand the operation state of the power grid under different time scales, and at the same time, some specific time nodes can be selected according to the actual operation of the power grid, such as the load peak period, the fault occurrence period, etc., to obtain more targeted data; the operation characteristic data of each power grid node obtained from each time node is normalized, and the normalized operation characteristic data is composed into a feature vector; the plurality of operation characteristic data are integrated together through the feature vector, which is convenient for subsequent analysis and processing.

[0059] S300, calculate the operation state information entropy value of each time node based on each feature vector, the information entropy is a concept in information theory, which is used to measure the uncertainty or chaos degree of information. The operation state information entropy value of each time node is calculated, and the operation state feature time sequence is further synthesized; this time sequence can intuitively show the operation state change of the distribution network at different time nodes, and provide support for subsequent fault early warning, etc.

[0060] S400, use the operation state feature time sequence as the input of the pre-trained neural network model, and output the corresponding early warning score, the early warning score indicates the probability of the fault of the distribution network in a given length of the prediction time range after the evaluation time range.

[0061] As a preferred embodiment of the present embodiment, the step of calculating the operation state information entropy value of each time node based on each feature vector is specifically:

[0062] The feature vector and the node operation fault probability conform to the Gaussian distribution:

[0063] P(X / Qi)=G(Qi,σ 2 );

[0064] Wherein, X represents the target distribution network, Qi is the i-th feature vector, P(X / Qi) represents the fault probability of the distribution network when the feature vector is Qi; G(·,σ 2 ) represents the Gaussian distribution function;

[0065] The Gaussian distribution describes the probability distribution of a random variable deviating from its mean value. In the present embodiment, the feature vector Qi is taken as a random variable, and the corresponding fault probability P(X / Qi) obeys the Gaussian distribution G(Qi,σ 2 ). It means that for each feature vector Qi, its corresponding distribution of the distribution network fault probability has a mean and a variance, so as to describe its distribution form and dispersion degree.

[0066] The corresponding information entropy value is calculated based on the feature vector:

[0067]

[0068] where m is the number of feature vectors, corresponding to the number of time nodes selected within the evaluation time range. By calculating the information entropy, the importance of each feature vector to the fault probability of the power distribution network can be understood, thereby screening out the feature vectors that have a significant impact on fault prediction.

[0069] As a preferred embodiment of the present embodiment, in the step of using the operation state feature time sequence as the input of the pre-trained neural network model and outputting the corresponding early warning score:

[0070] The pre-trained neural network model is a fusion model that integrates multiple components, aiming to achieve deep processing and analysis of input data. This model includes a first encoder network, a second encoder network, a first decoder network, a second decoder network, and a recurrent neural network, which work together to complete the prediction of the operation state feature time sequence.

[0071] Specifically, the first encoder network is the entrance of the model, responsible for downsampling the input operation state feature time sequence. The main purpose of this step is to reduce the dimensionality of the data and extract key features to generate an initial input with a first scale. Through downsampling, the model can remove redundant information and retain features that are crucial for subsequent processing, thereby improving the efficiency and accuracy of the model.

[0072] Subsequently, the first decoder network decodes the initial input to generate an initial output. The role of the decoder is to reconstruct the encoded data into a sequence form for subsequent processing. In this stage, the model uses learned knowledge and experience to analyze and reconstruct the initial input, generating more rich output information.

[0073] Next, the second decoder network performs upsampling on the initial output to generate a secondary input with a second scale. Through upsampling, the model can further enrich the output information and improve the prediction accuracy of the model.

[0074] Subsequently, the second encoder network encodes the secondary input to obtain a secondary output. This step is similar to the first encoder network, aiming to further extract deep features of the data to support subsequent prediction tasks.

[0075] Finally, the recurrent neural network takes the secondary output as input and outputs the predicted early warning score. The recurrent neural network has a memory function that can capture the time sequence dependency in the data, thereby achieving accurate prediction of future states. Through the processing of the recurrent neural network, the model can output accurate early warning scores.

[0076] As a preferred embodiment of the present embodiment, the operation characteristic data of each power grid node in the whole process of the power distribution network includes voltage, current, power factor, load rate, temperature, humidity, rainfall, and equipment age.

[0077] The electrical parameters are the core part of the operation characteristic data of the power distribution network. The voltage reflects the potential difference between the power grid nodes, and is of great significance to ensure the power quality. The current is the manifestation of the flow of electric charge in the power grid, and its size directly reflects the load condition of the power grid. The power factor is an index for measuring the proportional relationship between useful work and useless work in the power grid. The load rate is the ratio between the actual load and the rated load of the power grid node, and reflects the load capacity of the power grid node.

[0078] In addition to the electrical parameters, environmental factors are also important factors affecting the operation of the power distribution network. Temperature has a significant impact on the operating performance and service life of power grid equipment. Too high or too low temperature can cause equipment failure. Humidity is closely related to the insulation performance of power grid equipment. Excessive humidity can cause the insulation performance of the equipment to decline, thereby causing electrical failure. Rainfall can affect the insulation performance and grounding performance of the power grid.

[0079] In addition, the age of the equipment is also a kind of operation characteristic. With the increase of the use time of the equipment, its performance will gradually decline, and the failure rate will gradually rise.

[0080] Embodiment Two:

[0081] The present embodiment provides a power distribution network whole-process operation state early warning system, comprising:

[0082] A data acquisition module is configured to acquire operation characteristic data of each power grid node in the whole process of the power distribution network. The module is configured to realize the function of step S100 in embodiment one, and will not be described here again.

[0083] A characteristic data construction module is configured to set an evaluation time range, select a plurality of time nodes within the evaluation time range to obtain the operation characteristic data of each power grid node, and form a characteristic vector from the operation characteristic data of each power grid node obtained from each time node. The module is configured to realize the function of step S200 in embodiment one, and will not be described here again.

[0084] A characteristic sequence construction module is configured to calculate the operation state information entropy value of each time node based on each characteristic vector, and synthesize the operation state characteristic time sequence through the calculated operation state information entropy values of each time node. The module is configured to realize the function of step S300 in embodiment one, and will not be described here again.

[0085] The early warning analysis module uses the running state feature time sequence as the input of a pre-trained neural network model, and outputs a corresponding early warning score, which indicates the probability of failure of the power distribution network in a given length of a prediction time range after an evaluation time range.

[0086] As a preferred embodiment of the present embodiment, the step of calculating the running state information entropy value of each time node based on each feature vector is specifically:

[0087] The feature vector and the node running failure probability conform to a Gaussian distribution:

[0088] P(X / Qi)=G(Qi,σ 2 );

[0089] Wherein, X represents the target power distribution network, Qi is the i-th feature vector, P(X / Qi) represents the failure probability of the power distribution network when the feature vector is Qi; G(·,σ 2 ) represents a Gaussian distribution function;

[0090] Based on the feature vector, the corresponding information entropy value is calculated:

[0091]

[0092] Wherein, m is the number of feature vectors, corresponding to the number of time nodes selected in the evaluation time range.

[0093] As a preferred embodiment of the present embodiment, in the step of using the running state feature time sequence as the input of a pre-trained neural network model and outputting a corresponding early warning score:

[0094] The pre-trained neural network model includes a first encoder network, a second encoder network, a first decoder network, a second decoder network and a recurrent neural network;

[0095] Wherein, the first encoder network performs a downsampling operation on the input running state feature time sequence to generate an initial input with a first scale; the first decoder network decodes the initial input to generate an initial output;

[0096] The second decoder network performs an upsampling operation on the initial output to generate a secondary input with a second scale, the second scale being higher than the first scale; the second encoder network encodes the secondary input to obtain a secondary output;

[0097] The recurrent neural network takes the secondary output as input and outputs a predicted early warning score.

[0098] As a preferred embodiment of the present embodiment, the operation characteristic data of each power grid node in the whole process of the power distribution network includes voltage, current, power factor, load rate, temperature, humidity, rainfall, and equipment age.

[0099] Embodiment three:

[0100] The present embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the power distribution network whole process operation state early warning method according to any embodiment of the present application when executing the program.

[0101] Embodiment four:

[0102] The present embodiment provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the power distribution network whole process operation state early warning method according to any embodiment of the present application.

[0103] In the embodiments of the present application, "at least one" means one or more, and "multiple" means two or more. The "and / or" describes the association relationship between the associated objects, which means that there can be three kinds of relationships, for example, A and / or B, which can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Wherein A and B can be singular or plural. The character " / " generally represents an "or" relationship between the front and rear associated objects. "At least one of the following" and the like means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, wherein a, b, and c can be single or multiple.

[0104] Those of ordinary skill in the art can realize that the units and algorithm steps described in the embodiments disclosed herein can be realized by electronic hardware, computer software and a combination of electronic hardware and computer software. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0105] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0106] In several embodiments provided in the present application, any function, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the parts of the technical solutions that essentially contribute to the prior art or the parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0107] The above description is only some embodiments of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation, or direct or indirect application in other related technical fields, based on the content of the specification and drawings of the present application, are also included in the patent protection scope of the present application.

Claims

1. A distribution network full-process operation status early warning method, characterized in that: The following steps are involved: Collect operational characteristic data of each grid node throughout the entire distribution network process; Set the evaluation time range, and select multiple time nodes within the evaluation time range to obtain the operating characteristic data of each grid node; The operation characteristic data of each grid node obtained from each time node are combined into a feature vector; Calculate the running state information entropy value of each time node based on each feature vector, and synthesize the running state feature time series through the calculated running state information entropy value of each time node; Using the operating status feature time series as input to a pre-trained neural network model, outputting a corresponding early warning score indicating the probability of a distribution network failure within a forecast time range of a given length after the evaluation time range; The step of calculating the running state information entropy value of each time node based on each feature vector is specifically as follows: Assume that the eigenvector and node failure probability conform to Gaussian distribution: P(X / Qi)=G(Qi,σ 2 ); Where X represents the target distribution network, Qi is the i-th eigenvector, P(X / Qi) represents the failure probability of the distribution network when the eigenvector is Qi; G(·,σ 2 ) represents the Gaussian distribution function; Calculate the corresponding information entropy value based on the eigenvector: Where m is the number of feature vectors, corresponding to the number of time nodes selected within the evaluation time range; Wherein, in the step of using the operating status feature time series as the input of the pre-trained neural network model to output the corresponding warning score: A pre-trained neural network model, comprising a first encoder network, a second encoder network, a first decoder network, a second decoder network, and a recurrent neural network; The first encoder network performs a downsampling operation on the input running state feature time series to generate an initial input with a first scale; the first decoder network decodes the initial input to generate an initial output; The second decoder network performs an upsampling operation on the initial output to generate a secondary input having a second scale, wherein the second scale is higher than the first scale; the second encoder network encodes the secondary input to obtain a secondary output; The recurrent neural network takes the secondary output as input and outputs a predicted warning score.

2. A distribution network full-process operation status early warning method according to claim 1, characterized in that: The operating characteristic data of each grid node in the entire distribution network process include voltage, current, power factor, load rate, temperature, humidity, rainfall and equipment age.

3. A distribution network full-process operation status early warning system, characterized in that: include: Data acquisition module, used to collect operational characteristic data of each grid node in the entire distribution network process; The characteristic data construction module is used to set the evaluation time range and select multiple time nodes within the evaluation time range to obtain the operating characteristic data of each power grid node; The operation characteristic data of each grid node obtained from each time node are combined into a feature vector; The feature sequence construction module calculates the running state information entropy value of each time node based on each feature vector, and synthesizes the running state feature time series through the calculated running state information entropy value of each time node; an early warning analysis module that uses the operating status feature time series as input to a pre-trained neural network model and outputs a corresponding early warning score indicating the probability of a distribution network failure within a forecast time range of a given length following the evaluation time range; The step of calculating the running state information entropy value of each time node based on each feature vector is specifically as follows: Assume that the eigenvector and node failure probability conform to Gaussian distribution: P(X / Qi)=G(Qi,σ 2 ); Where X represents the target distribution network, Qi is the i-th eigenvector, P(X / Qi) represents the failure probability of the distribution network when the eigenvector is Qi; G(·,σ 2 ) represents the Gaussian distribution function; Calculate the corresponding information entropy value based on the eigenvector: Where m is the number of feature vectors, corresponding to the number of time nodes selected within the evaluation time range; Wherein, in the step of using the operating status feature time series as the input of the pre-trained neural network model to output the corresponding warning score: A pre-trained neural network model, comprising a first encoder network, a second encoder network, a first decoder network, a second decoder network, and a recurrent neural network; The first encoder network performs a downsampling operation on the input running state feature time series to generate an initial input with a first scale; the first decoder network decodes the initial input to generate an initial output; The second decoder network performs an upsampling operation on the initial output to generate a secondary input having a second scale, wherein the second scale is higher than the first scale; the second encoder network encodes the secondary input to obtain a secondary output; The recurrent neural network takes the secondary output as input and outputs a predicted warning score.

4. A distribution network full-process operation status early warning system according to claim 3, characterized in that: The operating characteristic data of each grid node in the entire distribution network process include voltage, current, power factor, load rate, temperature, humidity, rainfall and equipment age.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the distribution network full-process operation status early warning method as described in any one of claims 1 to 2 is implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the distribution network full-process operation status early warning method as described in any one of claims 1 to 2 is implemented.

Images