Power distribution network tower fault prediction method for extreme meteorological disasters

Abstract

The invention relates to a power distribution network tower fault prediction method for extreme meteorological disasters. The method comprises the following steps: constructing a comprehensive database comprising a meteorological database, a power grid database and a geographic database; meteorological data, power grid data and geographic data are respectively collected, and the data are preprocessed and then are respectively stored in corresponding databases; constructing a tower fault prediction model based on the CNN-LSTM hybrid model, and training the tower fault prediction model through the data in the integrated database to obtain a trained tower fault prediction model; and carrying out tower fault prediction through the trained tower fault prediction model.

Description

Technical Field

[0001] The invention relates to a distribution network tower fault prediction method facing extreme meteorological disasters, and belongs to the technical field of power network fault prediction. Background Art

[0002] Typhoons frequently occur in coastal areas of my country. The normal operation of the power grid is closely related to meteorological conditions. Typhoon disasters not only cause damage to power equipment, but also cause secondary floods and waterlogging disasters due to the precipitation they bring, endangering the safety of power facilities. Therefore, meteorological disaster warnings are of great significance for the power sector to make advance arrangements for responding to meteorological disasters.

[0003] As the support of high-voltage transmission lines in power grids, the safety of transmission towers is directly related to the safe and stable operation of power grids and the reliability of regional power supply. In order to improve the ability of distribution networks to avoid or resist natural disasters, the risk assessment and disaster identification of regional disasters in towers have received the focus of attention of the power sector, aiming to assess the disaster risk in the areas where typhoons pass through through more reliable and accurate methods, guide the scientific layout of future transmission lines, and minimize the impact of meteorological disasters on power grid operations.

[0004] The invention patent with the publication number "CN108427856B" discloses a method for fitting the failure probability curve of a 10 kV tower in a distribution network. With the help of the real damage data of historical typhoons and the meteorological data provided by the meteorological department, the fitting index F corresponding to different K values ​​in the failure probability curve is calculated. appro The optimal K value is obtained, and then the optimal distribution network 10 kV tower failure probability curve matching the actual disaster loss results is fitted. The present invention can simply, quickly and effectively fit the distribution line failure probability curve suitable for strong typhoon environments, and can be applied to the safety risk assessment of distribution towers in distribution networks under strong typhoon environments.

[0005] However, the path and intensity of a typhoon are highly uncertain, and a single failure probability curve may not fully reflect the actual situation. It needs to be dynamically adjusted in combination with real-time data. In addition to typhoons, tower failures may also be affected by other environmental factors (such as corrosion, aging, etc.), which may not be fully considered in the above invention. Summary of the invention

[0006] In order to solve the above problems existing in the prior art, the present invention proposes a distribution network tower failure prediction method for extreme meteorological disasters.

[0007] The technical solution of the present invention is as follows:

[0008] On the one hand, the present invention provides a distribution network tower failure prediction method for extreme meteorological disasters, comprising the following steps:

[0009] Build a comprehensive database including meteorological database, power grid database and geographic database;

[0010] Collect meteorological data, power grid data and geographic data respectively, and save them in corresponding databases after preprocessing;

[0011] A tower fault prediction model is constructed based on the CNN-LSTM hybrid model. The tower fault prediction model is trained by using the data in the comprehensive database to obtain a trained tower fault prediction model.

[0012] Pole tower failure prediction is performed through the trained pole tower failure prediction model.

[0013] As a preferred embodiment of the present invention, the collected meteorological data, power grid data and geographic data are normalized and preprocessed, as shown in the following formula:

[0014]

[0015] Where: X represents the original data sequence; X * represents the normalized data sequence; X max Indicates the maximum value in the original data sequence; X min Represents the minimum value in the original data sequence.

[0016] As a preferred embodiment of the present invention, the CNN-LSTM hybrid model is composed of a convolutional neural network and a long short-term memory neural network;

[0017] The convolutional neural network includes a plurality of continuous convolutional layers and a pooling layer at the end, and the data in the comprehensive database is input into the convolutional neural network to extract the spatial features of the data samples;

[0018] The long short-term memory neural network is composed of multiple layers of continuous long short-term memory neural units, and the spatial features of the data samples are input into the long short-term memory neural network to obtain the spatial-temporal features of the data samples;

[0019] Based on the spatial-temporal characteristics of data samples, a fully connected layer is used for fault prediction.

[0020] As a preferred embodiment of the present invention, the convolution layer is used to extract the spatial features of data samples, as shown in the following formula:

[0021]

[0022] in: M represents the spatial features of the data sample extracted by neuron j of convolutional layer l; j represents the input space; Represents the spatial features of the data sample extracted by neuron i in convolutional layer l-1; Represents the weights of neuron i and neuron j; represents the bias of neuron j; f represents the activation function.

[0023] As a preferred embodiment of the present invention, the pooling layer is used to downsample the data sample features extracted by the convolution layer, as shown in the following formula:

[0024]

[0025] in: represents the spatial features of sample data output by neuron j after convolution layer l+1 pooling; f down Indicates downsampling.

[0026] As a preferred embodiment of the present invention, each long short-term memory neural unit includes an input gate, a forget gate, an output gate and a unit state, as shown in the following formula:

[0027] I t =δ(U i x t +W i h t-1 +b i )

[0028] F t =δ(U f x t +W f h t-1 +b f )

[0029] O t =δ(U o x t +W o h t-1 +b o )

[0030]

[0031] h t =O t σ(C t )

[0032] Where: x t represents the input value of the t-th long short-term memory neural unit; h t represents the output value of the t-th long short-term memory neural unit; h t-1 Represents the output value of the t-1th long short-term memory neural unit; I t 、F t , O tRespectively represent the output values ​​of the input gate, forget gate and output gate of the t-th long short-term memory neural unit; C t The unit state of the t-th LSTM neural unit; C t-1 represents the unit state of the t-1th long short-term memory neural unit; represents the temporary unit state of the t-th LSTM neural unit; U i , U f , U o and U c are the weight matrices of the input gate, forget gate, output gate and cell state respectively; b i , b f and b o is the bias of the input gate, forget gate, and output gate; δ represents the sigmiod function; σ represents the tanh function.

[0033] As a preferred embodiment of the present invention, the use of a fully connected layer for fault prediction is specifically shown in the following formula:

[0034] Y=δ(W c H+b c )

[0035] Where: Y represents the failure probability; W c represents the weight of the fully connected layer; b c represents the bias of the fully connected layer; H represents the output of the long short-term memory neural network.

[0036] On the other hand, the present invention also provides a distribution network tower fault prediction system for extreme meteorological disasters, including a database module, a data acquisition module, a model training module and a fault prediction module;

[0037] The database module is composed of a comprehensive database including a meteorological database, a power grid database and a geographic database;

[0038] The data acquisition module is used to collect meteorological data, power grid data and geographic data, and save the data in corresponding databases after pre-processing;

[0039] The model training module is used to build a pole tower fault prediction model based on the CNN-LSTM hybrid model, and train the pole tower fault prediction model through the data in the comprehensive database to obtain a trained pole tower fault prediction model;

[0040] The fault prediction module is used to predict tower faults through a trained tower fault prediction model.

[0041] On the other hand, the present invention further 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 implements the method described in any embodiment of the present invention when executing the program.

[0042] In yet another aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any embodiment of the present invention.

[0043] The present invention has the following beneficial effects:

[0044] 1. The present invention is based on a method for predicting tower failures under extreme meteorological disasters based on a CNN-LSTM hybrid model. The meteorological information, tower body information, and geographic information distribution network body parameter information when historical tower failures occurred are matched, and the CNN-LSTM hybrid model is trained to obtain a tower failure prediction model. Finally, according to the prediction model obtained by training, the probability of tower failures in the event of extreme meteorological disasters in the future can be predicted. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0046] 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 only 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 are within the scope of protection of the present invention.

[0047] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0048] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0049] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0050] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0051] Embodiment 1:

[0052] See also Figure 1 , a distribution network tower failure prediction method for extreme meteorological disasters, comprising the following steps:

[0053] Build a comprehensive database including meteorological database, power grid database and geographic database;

[0054] Collect meteorological data, power grid data and geographic data respectively, and save them in corresponding databases after preprocessing;

[0055] A tower fault prediction model is constructed based on the CNN-LSTM hybrid model. The tower fault prediction model is trained by using the data in the comprehensive database to obtain a trained tower fault prediction model.

[0056] Pole tower failure prediction is performed through the trained pole tower failure prediction model.

[0057] In this embodiment, the meteorological data includes typhoon disaster data, specifically including wind speed, wind direction, temperature, humidity, typhoon path, wind circle level and typhoon level;

[0058] The power grid data includes pole tower ledger data and pole tower fault data;

[0059] The power grid data includes pole tower ledger data and pole tower fault data;

[0060] The tower ledger data includes sample data of towers that did not fail in the distribution network area when the meteorological disaster occurred, specifically including tower material, pole height, pole weight, main pole burial depth, fixing method, wind resistance level, designed maximum wind speed and tower structure;

[0061] The tower failure data includes sample data of tower failures in the distribution network area when meteorological disasters occur, specifically including tower material, pole height, pole weight, main pole burial depth, fixing method, wind resistance level, designed maximum wind speed and tower structure;

[0062] The inherent characteristics of the sample data of the towers that have not failed are completely consistent with those of the sample data of the towers that have failed. Whether the tower is faulty is determined based on whether the tower is powered off.

[0063] As a preferred implementation of this embodiment, irrelevant data, duplicate data, and erroneous data in the collected meteorological data, power grid data, and geographic data are eliminated through a threshold value set in advance to eliminate data that exceeds the range;

[0064] For missing data in the collected meteorological data, power grid data and geographic data, such as missing meteorological data, the nearest neighbor interpolation method is used to fill the missing data with existing adjacent data. For missing parameters of the tower itself, the missing attributes can be replaced with a constant, and the fixed value method can be used to eliminate the error caused by it to the model.

[0065] In order to facilitate CNN neural network processing, the data should be normalized. The min-max normalization method is used to reduce the range of all data to between 0 and 1. The collected meteorological data, power grid data and geographic data are normalized and preprocessed, as shown in the following formula:

[0066]

[0067] Where: X represents the original data sequence; X * represents the normalized data sequence; X max Indicates the maximum value in the original data sequence; X min Represents the minimum value in the original data sequence.

[0068] The normalized data is constructed as a training sample set Z:

[0069] Z={(X1,y 11 ,y 12 ,y 13 ),(X2,y 21 ,y 22 ,y 23 ),…(X n ,y n1 ,y n2 ,y n3 )}

[0070] X n =(x n1 ,x n2 ,…x nm )

[0071] Where: X n represents the input features of the nth sample, i.e., the meteorological, power grid, and geographical parameters under the nth extreme meteorological condition; x nm is the mth input feature in the nth sample, y n1 ,y n2 ,y n3 They represent the tower fault status (i.e., power outage or no power outage), power outage duration, and fault type corresponding to the nth sample;

[0072] The sample data set is stored in a grid format using longitude and latitude;

[0073] As a preferred implementation of this embodiment, the CNN-LSTM hybrid model is composed of a convolutional neural network and a long short-term memory neural network;

[0074] The convolutional neural network includes several continuous convolutional layers and a pooling layer at the end. The data in the comprehensive database is input into the convolutional neural network to extract the spatial features of the data samples. The convolutional layer is used to extract the features of the grid of the tower and meteorological composition image, wherein the convolution kernel is a small window containing a coefficient matrix. This window slides on the sample data set grid and performs a convolution operation, thereby extracting the spatial features of the sample data set.

[0075] The long short-term memory neural network is composed of multiple layers of continuous long short-term memory neural units. The spatial characteristics of the data samples are input into the long short-term memory neural network to obtain the spatial-temporal characteristics of the data samples, focusing on the trend and periodicity of the tower sequence itself.

[0076] Based on the spatial-temporal characteristics of data samples, a fully connected layer is used for fault prediction.

[0077] As a preferred implementation of this embodiment, the convolution layer is used to extract the spatial features of data samples, as shown in the following formula:

[0078]

[0079] in: M represents the spatial features of the data sample extracted by neuron j of convolutional layer l; j represents the input space; Represents the spatial features of the data sample extracted by neuron i in convolutional layer l-1; Represents the weights of neuron i and neuron j; represents the bias of neuron j; f represents the activation function. In this embodiment, the ELU function is used as the activation function, which is specifically shown in the following formula:

[0080]

[0081] As a preferred implementation of this embodiment, in order to compress the convolution layer features, the pooling layer can downsample the high-dimensional spatial correlation features of the sample data set and convert them into spatial correlation features extracted by the CNN layer. In this embodiment, the maximum pooling method is adopted, as shown in the following formula:

[0082]

[0083] in: represents the spatial features of sample data output by neuron j after convolution layer l+1 pooling; f down Indicates downsampling.

[0084] As a preferred implementation of this embodiment, each long short-term memory neural unit includes an input gate, a forget gate, an output gate and a unit state, as shown in the following formula:

[0085] I t =δ(U i x t +W i h t-1 +b i )

[0086] F t =δ(U f x t +W f h t-1 +b f )

[0087] O t =δ(U o x t +W o h t-1 +b o )

[0088]

[0089] h t =O t σ(C t )

[0090] Where: x t represents the input value of the t-th LSTM neural unit, i.e., the spatial correlation feature of the CNN layer output; h t represents the output value of the t-th long short-term memory neural unit; h t-1 Represents the output value of the t-1th long short-term memory neural unit; I t 、F t , O t Respectively represent the output values ​​of the input gate, forget gate and output gate of the t-th long short-term memory neural unit; C t The unit state of the t-th LSTM neural unit; C t-1 represents the unit state of the t-1th long short-term memory neural unit; represents the temporary unit state of the t-th LSTM neural unit; U i , U f , U o and U c are the weight matrices of the input gate, forget gate, output gate and cell state respectively; b i , b f and b oare the biases of the input gate, forget gate, and output gate; δ represents the sigmoid function; σ represents the tanh function.

[0091] The input gate I t determines which data features to retain for extracting useful information at the current time node;

[0092] The forget gate F t functions to filter h t-1 to select useful memories from previous time nodes and act on C t-1 ;

[0093] The output gate O t determines which data features to transmit and integrates the information at the current time node and the cell state to complete the output;

[0094] As a preferred implementation manner of this embodiment, using a fully connected layer for fault prediction is specifically shown as the following formula:

[0095] Y = δ(W c H + B c )

[0096] where: Y represents the fault probability; W c represents the weight of the fully connected layer; b c represents the bias of the fully connected layer; H represents the output of the long short-term memory neural network.

[0097] As a preferred implementation manner of this embodiment, based on the population, GDP, and terrain elevation data in the geographic dataset, the typhoon disaster risk index R is calculated. When the typhoon disaster risk index is below 20%, 20% - 40%, 40% - 60%, 60% - 80%, and above 80% of the overall distribution of the typhoon disaster risk index, it corresponds to the first-level risk, second-level risk, third-level risk, fourth-level risk, and fifth-level risk respectively.

[0098] For the first-level disaster risk level, 0 < R ≤ 0.2, conduct regular inspections of the poles and towers, check the damage of the pole and tower bodies, discover and repair the defects and damages in the pole and tower structures to ensure that the poles and towers are in good condition.

[0099] For the second-level disaster risk level, 0.2 < R ≤ 0.4, strengthen daily maintenance and upkeep, evaluate the surrounding geological risks, strengthen the management of the vegetation around the poles and towers to prevent the influence of vegetation growth on the poles and towers, and ensure their safety.

[0100] For the third-level disaster risk level, 0.4 < R ≤ 0.6, list it as a key inspection area and conduct round-the-clock patrol and值守 during disasters.

[0101] Fourth-level disaster risk level, 0.6 < R ≤ 0.8. Set windbreak forests or flood control dikes around, regularly perform anti-corrosion treatment on the poles and towers to extend their service life, and use a more stable foundation construction design, such as buried foundation or special protection design, to improve the stability of the poles and towers.

[0102] Fifth-level disaster risk level, 0.8 < R ≤ 1. Consider upgrading the pole and tower materials, adopting wind-resistant design standards and more durable materials to improve the wind resistance of the poles and towers and reduce the impact of wind load on the poles and towers; or optimize the pole and tower structure, adopt collapsible poles and towers or anti-collapse tower designs to reduce the risk of pole and tower collapse caused by typhoons.

[0103] Example Two:

[0104] A distribution network pole and tower fault prediction system for extreme meteorological disasters, including a database module, a data acquisition module, a model training module, and a fault prediction module;

[0105] The database module is composed of a comprehensive database including a meteorological database, a power grid database, and a geographical database;

[0106] The data acquisition module is used to collect meteorological data, power grid data, and geographical data, and after preprocessing the data, save them in the corresponding databases respectively;

[0107] The model training module is used to build a pole and tower fault prediction model based on the CNN-LSTM hybrid model, and train the pole and tower fault prediction model through the data in the comprehensive database to obtain a trained pole and tower fault prediction model;

[0108] The fault prediction module is used to perform pole and tower fault prediction through the trained pole and tower fault prediction model.

[0109] This system is used to implement the functions in Example One, which will not be elaborated here.

[0110] Example Three:

[0111] This example proposes an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the method described in any embodiment of the present invention.

[0112] Example Four:

[0113] This example proposes a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method described in any embodiment of the present invention.

[0114] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0115] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can 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 this application.

[0116] Those skilled in the art can 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.

[0117] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0118] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A distribution network tower failure prediction method for extreme meteorological disasters. It is characterized in that The following steps are involved: Build a comprehensive database including meteorological database, power grid database and geographic database; Collect meteorological data, power grid data and geographic data respectively, and save them in corresponding databases after preprocessing; A tower fault prediction model is constructed based on the CNN-LSTM hybrid model. The tower fault prediction model is trained by using the data in the comprehensive database to obtain a trained tower fault prediction model. Pole tower failure prediction is performed through the trained pole tower failure prediction model.

2. A distribution network tower failure prediction method for extreme weather disasters according to claim 1, It is characterized in that The collected meteorological data, power grid data and geographic data are normalized and preprocessed, as shown in the following formula: Where: X represents the original data sequence; X * represents the normalized data sequence; X max Indicates the maximum value in the original data sequence; X min Represents the minimum value in the original data sequence.

3. A distribution network tower failure prediction method for extreme weather disasters according to claim 1, It is characterized in that The CNN-LSTM hybrid model is composed of a convolutional neural network and a long short-term memory neural network; The convolutional neural network includes a plurality of continuous convolutional layers and a pooling layer at the end, and the data in the comprehensive database is input into the convolutional neural network to extract the spatial features of the data samples; The long short-term memory neural network is composed of multiple layers of continuous long short-term memory neural units, and the spatial features of the data samples are input into the long short-term memory neural network to obtain the spatial-temporal features of the data samples; Based on the spatial-temporal characteristics of data samples, a fully connected layer is used for fault prediction.

4. A distribution network tower failure prediction method for extreme weather disasters according to claim 3, It is characterized in that The convolutional layer is used to extract the spatial features of data samples, as shown in the following formula: in: M represents the spatial features of the data sample extracted by neuron j of convolutional layer l; j represents the input space; Represents the spatial features of the data sample extracted by neuron i in convolutional layer l-1; Represents the weights of neuron i and neuron j; represents the bias of neuron j; f represents the activation function.

5. A distribution network tower failure prediction method for extreme weather disasters according to claim 3, It is characterized in that The pooling layer is used to downsample the data sample features extracted by the convolutional layer, as shown in the following formula: in: represents the spatial features of sample data output by neuron j after convolution layer l+1 pooling; f down Indicates downsampling.

6. A distribution network tower failure prediction method for extreme weather disasters according to claim 3, It is characterized in that Each of the long short-term memory neural units includes an input gate, a forget gate, an output gate, and a unit state, as shown in the following formula: I t =δ(U i x t +W i h t-1 +b i ) F t =δ(U f x t +W f h t-1 +b f ) O t =δ(U o x t +W o h t-1 +b o ) h t =O t σ(C t ) Where: x t represents the input value of the t-th long short-term memory neural unit; h t represents the output value of the t-th long short-term memory neural unit; h t-1 Represents the output value of the t-1th long short-term memory neural unit; I t 、F t , O t Respectively represent the output values ​​of the input gate, forget gate and output gate of the t-th long short-term memory neural unit; C t The unit state of the t-th LSTM neural unit; C t-1 represents the unit state of the t-1th long short-term memory neural unit; represents the temporary unit state of the t-th LSTM neural unit; U i , U f , U o and U c are the weight matrices of the input gate, forget gate, output gate and cell state respectively; b i , b f and b o is the bias of the input gate, forget gate, and output gate; δ represents the sigmiod function; σ represents the tanh function.

7. A distribution network tower failure prediction method for extreme weather disasters according to claim 3, It is characterized in that The use of a fully connected layer for fault prediction is specifically shown in the following formula: Y=δ(W c H+b c ) Where: Y represents the failure probability; W c represents the weight of the fully connected layer; b c represents the bias of the fully connected layer; H represents the output of the long short-term memory neural network.

8. A distribution network tower failure prediction system for extreme weather disasters. It is characterized in that It includes database module, data acquisition module, model training module and fault prediction module; The database module is composed of a comprehensive database including a meteorological database, a power grid database and a geographic database; The data acquisition module is used to collect meteorological data, power grid data and geographic data, and save the data in corresponding databases after preprocessing; The model training module is used to build a pole tower fault prediction model based on the CNN-LSTM hybrid model, and train the pole tower fault prediction model through the data in the comprehensive database to obtain a trained pole tower fault prediction model; The fault prediction module is used to predict tower faults through a trained tower fault prediction model.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images