A transmission line defect detection method and system based on deep learning

Abstract

The present application provides a deep learning-based transmission line defect detection method and system, which determines multiple first defect circuit information sets from a first line setting directory, and then determines the influence vectors between the first defect circuit information sets through the hidden vectors of each first defect circuit information set. Since the first influence vectors of each first defect circuit information set are obtained based on the hidden vectors of each first defect circuit information set, the influence vectors between the first defect circuit information sets cover the hidden information in the first part of the transmission line data. The process of optimizing the first defect description attributes using the first influence vectors is equivalent to optimizing the first defect description attributes of each first defect circuit information set based on the hidden information in the first transmission line data, thereby weakening the hidden-related influence data covered in the first part of the transmission line data caused by the hidden information, thereby improving the accuracy of data detection.

Description

Technical Field

[0001] The present application relates to the field of defect detection technology, and more specifically, to a method and system for power transmission line defect detection based on deep learning. Background Art

[0002] Transmission lines are constructed by boosting the voltage of generator power using transformers, then connecting it to the transmission line via control devices such as circuit breakers. Transmission lines are divided into overhead transmission lines and cable lines. This process ensures the proper functioning of electrical equipment, not only ensuring normal power supply in hospitals but also ensuring the normal operation of critical equipment, effectively safeguarding people's quality of life.

[0003] However, if there are defects in the transmission line, various problems may occur, but there is no quick way to detect them. As a result, it takes a certain amount of time to restore power. Therefore, there is an urgent need for a transmission line defect detection technology to overcome the above technical problems. Summary of the Invention

[0004] In order to improve the technical problems existing in the related technologies, the present application provides a transmission line defect detection method and system based on deep learning.

[0005] In a first aspect, a deep learning-based transmission line defect detection method is provided, the method comprising at least: obtaining partial knowledge fragments of a first line setting directory in transmission line data to be detected; determining multiple first defect circuit information sets from the first line setting directory using the obtained partial knowledge fragments, and selecting a first defect description attribute of each first defect circuit information set; determining a hidden core hidden in the first line setting directory, and determining a first hidden vector hidden by each first defect circuit information set using the detected hidden core; obtaining a first influence vector between each first defect circuit information set detected by a statistically obtained first hidden vector; optimizing each first defect description attribute using the obtained first influence vector, and comparing the optimized first defect description attribute with a second defect description attribute to obtain a defect detection result of the first line setting directory, wherein the second defect description attribute is: a key content obtained by optimizing the defect description attribute of each second defect circuit information set using each second influence vector, and each second defect circuit information set is: a knowledge fragment corresponding to each first defect circuit information set in the second line setting directory set in the partial transmission line data.

[0006] In the present application, the optimized first defect description attribute is compared with the second defect description attribute to obtain the defect detection result of the first line setting directory, including: determining the similarity knowledge fragments between the optimized first defect description attribute of each first defect circuit information set and the corresponding second defect description attribute, and determining them as the similarity knowledge fragments corresponding to each first defect circuit information set; based on the first hidden vector of each first defect circuit information set, determining the trust weight of the first defect description attribute of each first defect circuit information set for the characteristics of the first line setting directory; integrating the similarity knowledge fragments corresponding to each first defect circuit information set through the detected trust weight to obtain the integration processing result, and determining them as the similarity knowledge fragments between the first line setting directory and the second line setting directory; and determining the defect detection result of the first line setting directory through the obtained similarity knowledge fragments.

[0007] In the present application, the method of determining the trust weight of the first defect description attribute of each first defect circuit information set for the characteristics of the first line setting directory based on the first hidden vector of each first defect circuit information set includes: determining the trust weight of the first defect description attribute of each first defect circuit information set for the characteristics of the first line setting directory based on the first hidden vector of each first defect circuit information set and the second hidden vector of the corresponding second defect circuit information set.

[0008] In the present application, the comparison of the optimized first defect description attribute with the second defect description attribute to obtain the defect detection result of the first circuit setting directory includes: combining the first defect description attributes of each first defect circuit information set through the first hidden vector of each first defect circuit information set to obtain a first combined processing vector, and combining the second defect description attributes of each second defect circuit information set through the second hidden vector of each second defect circuit information set to obtain a second combined processing vector; determining the similarity knowledge fragments between the first combined processing vector and the second combined processing vector; and determining the defect detection result of the first circuit setting directory by determining the obtained similarity knowledge fragments.

[0009] In this application, the first impact vector is determined according to the following expression: Effect Pro = max(hidden vector effect vector, threshold) x (1-hidden vector effect vector), wherein Effect Pro is the first impact vector of the first defective circuit information set Defect Electron on the first defective circuit information set effect vector, the hidden vector effect vector represents the hidden vector of the first defective circuit information set effect vector, and threshold is the set threshold.

[0010] In the present application, the method of optimizing each first defect description attribute by obtaining the first influence vector and comparing the optimized first defect description attribute with the second defect description attribute to obtain the defect detection result of the first line setting directory includes: loading each first defect description attribute, the first influence vector between each first defect circuit information set, the defect description attribute of each second defect circuit information set, and the second influence vector between each second defect circuit information set into a pre-trained defect detection network, so that the defect detection network optimizes each first defect description attribute based on each first influence vector to obtain the optimized first defect description attribute, and optimizes the defect description attribute of each second defect circuit information set based on each second influence vector to obtain the second defect description attribute, and compares the optimized first defect description attribute with the second defect description attribute to output the defect detection result; and obtains the defect detection result output by the defect detection network.

[0011] In the present application, the method of determining multiple first defective circuit information sets from the first line setting directory through the acquired partial knowledge fragments includes: determining the knowledge fragment category to which each partial knowledge fragment belongs through the key category of the acquired partial knowledge fragment and the association relationship between the set partial knowledge fragment category and the knowledge fragment category of the first defective circuit information set; for each knowledge fragment category, based on the position information of the partial knowledge fragment belonging to the knowledge fragment category, obtaining the standard position of the defective circuit information set belonging to the knowledge fragment category, and determining the knowledge fragment difference between the standard position of the defective circuit information set belonging to the knowledge fragment category and the standard position of the adjacent defective circuit information set, and determining the knowledge fragment vector of the defective circuit information set belonging to the knowledge fragment category through the detected knowledge fragment difference; determining the first defective circuit information set to which each standard position belongs based on the acquired standard position and the determined knowledge fragment vector.

[0012] In the present application, the selection of partial descriptive content within each first defective circuit information set includes: selecting a global knowledge vector set of the first line setting directory in the transmission line data to be detected; determining the knowledge fragments corresponding to each first defective circuit information set in the global knowledge vector set based on the position of each first defective circuit information set in the first line setting directory; debugging the knowledge fragments of each first defective circuit information set according to the set knowledge vector set, and generating a vector that is a knowledge vector of the set knowledge vector set; determining the partial descriptive content corresponding to the knowledge vector of each first defective circuit information set, and determining it as the partial descriptive content within each first defective circuit information set.

[0013] In a second aspect, a deep learning-based transmission line defect detection system is provided, comprising a processor and a memory communicating with each other, wherein the processor is configured to read a computer program from the memory and execute the program to implement the above-mentioned method.

[0014] A deep learning-based transmission line defect detection method and system provided in an embodiment of the present application determines multiple first defect circuit information sets from a first line setting directory, and then determines the influence vectors between each first defect circuit information set through the hidden vectors of each first defect circuit information set. Since the first influence vectors of each first defect circuit information set are obtained based on the hidden vectors of each first defect circuit information set, the influence vectors between the first defect circuit information sets cover the hidden information in the first part of the transmission line data. The process of optimizing the first defect description attributes using the first influence vectors is equivalent to optimizing the first defect description attributes of each first defect circuit information set based on the hidden information in the first transmission line data, thereby weakening the hidden-related influence data covered in the first part of the transmission line data caused by the hidden information, thereby improving the accuracy of data detection.

[0015] Furthermore, the influence vectors between the first defective circuit information sets are determined through the hidden vectors of the first defective circuit information sets, thereby realizing the analysis of the relationship between the first defective circuit information sets, and then using the analyzed relationship between the knowledge fragments to realize data processing, further improving the accuracy of data detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1A flowchart of a power transmission line defect detection method based on deep learning provided in an embodiment of the present application.

[0018] Figure 2 This is a block diagram of a power transmission line defect detection device based on deep learning provided in an embodiment of the present application.

[0019] Figure 3 This is an architectural diagram of a deep learning-based power transmission line defect detection system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] In order to better understand the above technical solution, the technical solution of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0021] See also Figure 1 , shows a method for detecting transmission line defects based on deep learning, which may include the technical solutions described in the following steps S101-S105.

[0022] S101: Obtaining partial knowledge fragments of a first line setting directory in the power transmission line data to be detected.

[0023] S102: Determine a plurality of first defective circuit information sets from a first circuit setting directory using the acquired partial knowledge fragments, and select a first defect description attribute of each first defective circuit information set.

[0024] S103: Determine the hidden core hidden in the first circuit setting directory, and determine the first hidden vector hidden in each first defective circuit information set through the detected hidden core.

[0025] S104: Obtain a first influence vector between first defective circuit information sets detected by the first hidden vector obtained through statistics.

[0026] S105: Optimize each first defect description attribute through the obtained first influence vector, compare the optimized first defect description attribute with the second defect description attribute, and obtain the defect detection result of the first line setting directory, wherein the second defect description attribute is: the key content obtained by optimizing the defect description attribute of each second defect circuit information set through each second influence vector, and each second defect circuit information set is: the knowledge fragment corresponding to each first defect circuit information set in the second line setting directory set in the partial transmission line data.

[0027] An embodiment of the present disclosure provides a transmission line defect detection method based on deep learning, which determines multiple first defect circuit information sets from a first line setting directory, and then determines the influence vectors between the first defect circuit information sets through the hidden vectors of each first defect circuit information set. Since the first influence vectors of each first defect circuit information set are obtained based on the hidden vectors of each first defect circuit information set, the influence vectors between the first defect circuit information sets cover the hidden information in the first part of the transmission line data. The process of optimizing the first defect description attributes using the first influence vectors is equivalent to optimizing the first defect description attributes of each first defect circuit information set based on the hidden information in the first transmission line data, thereby weakening the hidden-related influence data covered in the first part of the transmission line data caused by the hidden information, thereby improving the accuracy of data detection.

[0028] Furthermore, similar knowledge fragments between each first defective circuit information set are determined through the hidden vectors of each first defective circuit information set, thereby realizing the analysis of the relationship between each first defective circuit information set, and then using the analyzed relationship between the knowledge fragments to realize data processing, further improving the accuracy of data detection.

[0029] For the second line setting directory in the partial transmission line data, the second defect circuit information set, the defect description attributes of the second defect circuit information set, the second influence vectors between each second defect circuit information set, and the second defect description attributes obtained by optimizing the defect description attributes based on the second influence vector can be detected in advance.

[0030] Optionally, the data passing through the partial transmission line is processed according to steps S101 to S104 to obtain a second defect description attribute of a second defective circuit information set of a second line setting directory in the data passing through the partial transmission line.

[0031] In one possible implementation embodiment, for each first defect circuit information set, the second defect circuit information set corresponding to the first defect circuit information set may be: a second defect circuit information set whose position in the second line setting directory is the same as the position of the first defect circuit information set in the first line setting directory, or a second defect circuit information set whose knowledge segment category is the same as the knowledge segment category of the first defect circuit information set, or a second defect circuit information set that meets both of the above requirements.

[0032] When the knowledge segment category of the first defective circuit information set is the local knowledge segment category, the second defective circuit information set corresponding to the first defective circuit information set may be: a second defective circuit information set whose knowledge segment category in the second line setting directory is the local knowledge segment category.

[0033] In one embodiment of the present disclosure, the above-mentioned step S105 can be implemented based on a pre-trained defect detection network, and specifically can include the following steps: loading each first defect description attribute, the first influence vector between each first defect circuit information set, the defect description attribute of each second defect circuit information set, and the second influence vector between each second defect circuit information set into the pre-trained defect detection network, so that the defect detection network optimizes each first defect description attribute based on each first influence vector to obtain the optimized first defect description attribute, and optimizes the defect description attribute of each second defect circuit information set based on each second influence vector to obtain the second defect description attribute, and compares the optimized first defect description attribute with the second defect description attribute to output the defect detection result; and obtains the defect detection result output by the defect detection network.

[0034] Furthermore, the first defect description attribute of each first defective circuit information set, the defect description attribute of each second defective circuit information set, and the adjacent data set recording the first influence vector between the first defective circuit information sets and the second influence vector between the second defective circuit information sets are regarded as inputs of the defect detection network.

[0035] The embodiment of the present disclosure further provides a method for detecting power transmission line defects based on deep learning, and the implementation of step S105 may specifically include the following content.

[0036] S201: Determine similarity knowledge segments between the optimized first defect description attributes of each first defective circuit information set and the corresponding second defect description attributes, and determine them as similarity knowledge segments corresponding to each first defective circuit information set.

[0037] In this step, the similarity knowledge segments between the optimized first defect description attribute and the second defect description attribute can be determined by the knowledge segment queues corresponding to the first defect description attribute and the second defect description attribute. By determining the cosine similarity knowledge segments of the knowledge segment queue corresponding to the first defect description attribute and the knowledge segment queue corresponding to the second defect description attribute, the similarity knowledge segments between the optimized first defect description attribute and the corresponding second defect description attribute are determined. Optionally, the larger the cosine similarity knowledge segment, the higher the similarity knowledge segment.

[0038] S202: Determine, based on the first hidden vector of each first defective circuit information set, a trust weight of a first defect description attribute of each first defective circuit information set with respect to a feature of the first circuit setting catalog.

[0039] In this step, the larger the first hidden vector is, the more influencing data is covered in the first defect description attribute of the first defect circuit information set, and the larger the first hidden vector is, the smaller the trust weight of the first defect description attribute of the first defect circuit information set for the feature of the first line setting catalog is.

[0040] Optionally, in one implementation, an association relationship between hidden vectors and trusted weights may be established first. After determining the first hidden vectors of each first defective circuit information set, the trusted weights corresponding to the first hidden vectors of each first defective circuit information set may be determined based on the association relationship.

[0041] Optionally, in another implementation, each defective circuit information set may be evaluated based on its first hidden vector, and the trust weight of the evaluation score of each defective circuit information set may be regarded as the trust weight of each first defective circuit information set.

[0042] Optionally, in another implementation, the trust weight of the first defect description attribute of each first defect circuit information set for the feature of the first line setting directory can be determined based on the first hidden vector of each first defect circuit information set and the second hidden vector of the corresponding second defect circuit information set.

[0043] S203: integrating similarity knowledge segments corresponding to the first defective circuit information sets using the detected trust weights to obtain integration results, which are determined as similarity knowledge segments between the first circuit setting catalog and the second circuit setting catalog.

[0044] S204: Determine a defect detection result of the first line setting catalog based on the obtained similarity knowledge fragments.

[0045] In this step, a similarity knowledge segment threshold may be set. When the obtained similarity knowledge segment is greater than the similarity knowledge segment threshold, it is determined that the user information of the first line setting directory is the same as the user information of the second line setting directory. Otherwise, they are different.

[0046] A transmission line defect detection method based on deep learning provided by an embodiment of the present disclosure is the basis for the beneficial effects of a transmission line defect detection method based on deep learning, and provides a technical solution for comparing the optimized first defect description attribute with the second defect description attribute. Since the similarity knowledge fragments of each first defect description attribute are determined separately, the workload of each detection can be reduced.

[0047] On the premise of a transmission line defect detection method based on deep learning, the embodiment of the present disclosure also provides a transmission line defect detection method based on deep learning, and the implementation of step S105 may specifically include the following content.

[0048] S301: Combine the first defect description attributes of each first defect circuit information set using the first hidden vector of each first defect circuit information set to obtain a first combined processing vector, and combine the second defect description attributes of each second defect circuit information set using the second hidden vector of each second defect circuit information set to obtain a second combined processing vector.

[0049] The second defect description attributes of each second defective circuit information set are combined with the first defect description attributes of each first defective circuit information set.

[0050] S302: Determine similarity knowledge segments between the first combined processing vector and the second combined processing vector.

[0051] S303: Determine the defect detection result of the first line setting catalog by determining the obtained similarity knowledge fragments.

[0052] The embodiment of the present disclosure provides a transmission line defect detection method based on deep learning, which is the basis for the beneficial effects of a transmission line defect detection method based on deep learning. It provides a technical solution for comparing the optimized first defect description attribute with the second defect description attribute. Since the processing vectors are combined first and then the similarity knowledge fragments of the features are determined, the determination steps can be simplified and the efficiency of the combination processing can be improved.

[0053] On the premise of the above-mentioned deep learning-based transmission line defect detection method, the embodiment of the present disclosure also provides a deep learning-based transmission line defect detection method to achieve the determination of the first defective circuit information set, which may specifically include the following steps.

[0054] S401: Determine the knowledge segment category to which each partial knowledge segment belongs based on the key categories of the acquired partial knowledge segments and the set association relationship between the partial knowledge segment category and the knowledge segment category of the first defective circuit information set.

[0055] S402: For each knowledge segment category, based on the position information of some knowledge segments belonging to the knowledge segment category, obtain the standard position of the defective circuit information set belonging to the knowledge segment category, and determine the difference between the standard position of the defective circuit information set belonging to the knowledge segment category and the standard position of the adjacent defective circuit information set. Through the detected difference between the knowledge segments, determine the knowledge segment vector of the defective circuit information set belonging to the knowledge segment category.

[0056] S403: Based on the acquired standard positions and the determined knowledge fragment vectors, determine the first defective circuit information set to which each standard position belongs.

[0057] On the premise of the above-mentioned deep learning-based transmission line defect detection method, the embodiment of the present disclosure also provides a deep learning-based transmission line defect detection method to realize the selection of the first defect description attribute, which may specifically include the following steps.

[0058] S501: Selecting a global knowledge vector set of a first line setting directory in the transmission line data to be detected.

[0059] S502: Determine, based on the position of each first defective circuit information set in the first circuit setting directory, a knowledge segment corresponding to each first defective circuit information set in the global knowledge vector set.

[0060] In this step, each pixel point in the first circuit setting directory exists at a corresponding position in the global knowledge vector set. Therefore, the knowledge fragments mapped to each first defective circuit information set on the global knowledge vector set can be determined through the association relationship between the first circuit setting directory and the global knowledge vector set.

[0061] S503: According to the set knowledge vector set, the knowledge segments of each first defective circuit information set are debugged to generate a vector which is a knowledge vector of the set knowledge vector set.

[0062] S504: Determine partial description content corresponding to the knowledge vector of each first defective circuit information set, and determine it as the partial description content in each first defective circuit information set.

[0063] In this step, the knowledge vectors acquired in step S503 may be further processed to obtain partial description content in each first defective circuit information set.

[0064] Under the above premise, please refer to Figure 2 , provides a power transmission line defect detection device 200 based on deep learning, which is applied to a power transmission line defect detection system based on deep learning, and the device includes:

[0065] The attribute selection module 210 is configured to obtain partial knowledge fragments of a first line setting directory in the transmission line data to be inspected; determine a plurality of first defective circuit information sets from the first line setting directory using the obtained partial knowledge fragments, and select a first defect description attribute for each first defective circuit information set;

[0066] a vector determination module 220 configured to determine a hidden core hidden in the first circuit setting directory, and determine a first hidden vector for hiding each first defective circuit information set based on the detected hidden core;

[0067] A vector influence module 230 is configured to obtain a first influence vector between first defective circuit information sets detected by the first hidden vector obtained through statistics;

[0068] The result detection module 240 is used to optimize each first defect description attribute using the obtained first influence vector, compare the optimized first defect description attribute with the second defect description attribute, and obtain the defect detection result of the first line setting directory, wherein the second defect description attribute is: the key content obtained by optimizing the defect description attribute of each second defect circuit information set using each second influence vector, and each second defect circuit information set is: the knowledge fragment corresponding to each first defect circuit information set in the second line setting directory set in the partial transmission line data.

[0069] Under the above premise, please refer to Figure 3 , shows a deep learning-based transmission line defect detection system 300, including a processor 310 and a memory 320 that communicate with each other, and the processor 310 is used to read and execute a computer program from the memory 320 to implement the above method.

[0070] Under the above premise, a computer-readable storage medium is also provided, on which a computer program stored implements the above method when running.

[0071] In summary, based on the above scheme, multiple first defect circuit information sets are determined from the first line setting directory, and then the influence vectors between the first defect circuit information sets are determined through the hidden vectors of each first defect circuit information set. Since the first influence vectors of each first defect circuit information set are obtained based on the hidden vectors of each first defect circuit information set, the influence vectors between the first defect circuit information sets cover the hidden information in the first part of the transmission line data. The process of optimizing the first defect description attributes using the first influence vectors is equivalent to optimizing the first defect description attributes of each first defect circuit information set based on the hidden information in the first transmission line data, thereby weakening the hidden-related influence data covered in the first part of the transmission line data caused by the hidden information, and improving the accuracy of data detection.

[0072] Furthermore, the influence vectors between the first defective circuit information sets are determined through the hidden vectors of the first defective circuit information sets, thereby realizing the analysis of the relationship between the first defective circuit information sets, and then using the analyzed relationship between the knowledge fragments to realize data processing, further improving the accuracy of data detection.

[0073] It should be understood that the system and its modules shown above can be implemented in various ways. For example, in some embodiments, the system and its modules can be implemented by hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated hardware. Those skilled in the art will understand that the above-mentioned methods and systems can be implemented using computer-executable instructions and / or contained in processor control code, for example, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the system and its modules of the present application. Not only can hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc. be implemented, they can also be implemented using software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software (for example, firmware).

[0074] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects that may be produced may be any one or a combination of the above, or any other possible beneficial effects.

[0075] The basic concepts have been described above. It will be apparent to those skilled in the art that the detailed disclosure above is intended merely as an example and does not constitute a limitation of this application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and revisions to this application. Such modifications, improvements, and revisions are suggested in this application and remain within the spirit and scope of the exemplary embodiments of this application.

[0076] At the same time, this application uses specific terms to describe the embodiments of this application. For example, "one embodiment," "an embodiment," and / or "some embodiments" refer to a certain feature, structure, or characteristic related to at least one embodiment of this application. Therefore, it should be emphasized and noted that "one embodiment," "an embodiment," or "an alternative embodiment" mentioned twice or multiple times in different locations in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures, or characteristics in one or more embodiments of this application may be appropriately combined.

[0077] In addition, it will be understood by those skilled in the art that various aspects of the present application can be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application can be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0078] A computer storage medium may include a propagated data signal embodying computer program code, for example, in baseband or as part of a carrier wave. The propagated signal may take many forms, including electromagnetic, optical, or any suitable combination thereof. A computer storage medium may be any computer-readable medium other than a computer-readable storage medium that can be connected to an instruction execution system, apparatus, or device to communicate, propagate, or transfer the program for use. The program code on a computer storage medium may be transmitted via any suitable medium, including radio, cable, fiber optic cable, RF, or similar media, or any combination of these.

[0079] The computer program code required for the operation of the various parts of this application can be written in any one or more programming languages, including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python, etc., conventional procedural programming languages ​​such as C, Visual Basic, Fortran 2003, Perl, COBOL 2002, PHP, ABAP, dynamic programming languages ​​such as Python, Ruby and Groovy, or other programming languages. The program code can be executed entirely on the user's computer, or determined to be executed on the user's computer as a separate software package, or partially executed on the user's computer and partially executed on a remote computer, or executed entirely on a remote computer or server. In the latter case, the remote computer can be connected to the user's computer via any network form, such as a local area network (LAN) or a wide area network (WAN), or connected to an external computer (e.g., via the Internet), or in a cloud computing environment, or determined to be used as a service such as software as a service (SaaS).

[0080] In addition, unless expressly stated in the claims, the order of the processing elements and sequences described in this application, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this application. Although the above disclosure discusses some of the invention embodiments currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that meet the essence and scope of the embodiments of this application. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0081] Similarly, it should be noted that, in order to simplify the presentation of this disclosure and thus facilitate understanding of one or more embodiments of the invention, the foregoing descriptions of the embodiments of this application sometimes combine multiple features into a single embodiment, figure, or description thereof. However, this disclosure method does not mean that the subject matter of this application requires more features than those recited in the claims. In fact, the features of the embodiments may be fewer than the global features of the individual embodiments disclosed above.

[0082] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise stated, "about", "approximately" or "substantially" indicate that the numbers allow adaptive changes. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which can be changed according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of the present application are approximate values, in specific embodiments, the settings of such numerical values ​​are as accurate as possible within the feasible range.

[0083] Each patent, patent application, patent application disclosure, and other materials, such as articles, books, instructions, publications, documents, etc., cited in this application is hereby incorporated by reference in its entirety. Except for application history documents that are inconsistent with or conflict with the content of this application, documents that limit the broadest scope of the claims of this application (currently or subsequently attached to this application) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the accompanying materials of this application are inconsistent or conflicting with the content of this application, the descriptions, definitions, and / or use of terms in this application shall prevail.

[0084] Finally, it should be understood that the embodiments described in this application are intended only to illustrate the principles of the embodiments of this application. Other variations may also fall within the scope of this application. Therefore, the alternative configurations of the embodiments of this application are intended to be illustrative rather than limiting, and may be considered consistent with the teachings of this application. Accordingly, the embodiments of this application are not limited to the embodiments explicitly described and illustrated in this application.

[0085] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A power transmission line defect detection method based on deep learning, characterized in that: The method at least comprises: Obtaining partial knowledge fragments of a first line setting directory in the transmission line data to be inspected; determining a plurality of first defective circuit information sets from the first line setting directory using the obtained partial knowledge fragments, and selecting a first defect description attribute of each first defective circuit information set; Determining a hidden core hidden in the first circuit setting directory, and determining a first hidden vector for hiding each first defective circuit information set based on the detected hidden core; Obtaining a first influence vector between each first defective circuit information set detected by the first hidden vector obtained by statistics; Based on the first hidden vector of each first defective circuit information set, a trust weight of the first defect description attribute of each first defective circuit information set relative to the feature of the first circuit setting catalog is determined; using the detected trust weight, similarity knowledge fragments corresponding to each first defective circuit information set are integrated to obtain an integration result, which is determined to be a similarity knowledge fragment between the first circuit setting catalog and the second circuit setting catalog; Optimizing each first defect description attribute using the obtained first influence vector, and comparing the optimized first defect description attribute with the second defect description attribute to obtain a defect detection result for the first line setting directory, wherein the second defect description attribute is a key content obtained by optimizing the defect description attribute of each second defect circuit information set using each second influence vector, and each second defect circuit information set is a knowledge segment corresponding to each first defect circuit information set in the second line setting directory set in the partial transmission line data; The step of comparing the optimized first defect description attribute with the second defect description attribute to obtain the defect detection result of the first line setting directory includes: Combining the first defect description attributes of each first defective circuit information set using the first hidden vector of each first defective circuit information set to obtain a first combined processing vector, and combining the second defect description attributes of each second defective circuit information set using the second hidden vector of each second defective circuit information set to obtain a second combined processing vector; determining similarity knowledge segments between the first combined processing vector and the second combined processing vector; Determining a defect detection result of the first line setting catalog by determining the obtained similarity knowledge fragments; Among them, the first influence vector is determined according to the following expression: Effect Pro = max(hidden vector effectvector, threshold) x (1-hidden vector effect vector) wherein Effect Pro is the first influence vector of the first defective circuit information set Defect Electron on the first defective circuit information set effect vector, the hidden vector effect vector represents the hidden vector of the first defective circuit information set effect vector, and threshold is the set threshold.

2. The method according to claim 1, characterized in that The comparing the optimized first defect description attribute with the second defect description attribute to obtain the defect detection result of the first line setting directory includes: Determine similarity knowledge segments between the optimized first defect description attributes of each first defective circuit information set and the corresponding second defect description attributes, and determine them as similarity knowledge segments corresponding to each first defective circuit information set; determining, based on the first hidden vector of each first defective circuit information set, a trust weight of a first defect description attribute of each first defective circuit information set for a feature of the first circuit setting catalog; Integrating the similarity knowledge segments corresponding to the first defective circuit information sets using the detected trust weights to obtain an integration result, which is determined to be the similarity knowledge segments between the first circuit setting directory and the second circuit setting directory; The defect detection result of the first line setting catalog is determined using the obtained similarity knowledge fragments.

3. The method according to claim 2, characterized in that The method of determining the trust weight of the first defect description attribute of each first defect circuit information set for the characteristics of the first line setting directory based on the first hidden vector of each first defect circuit information set includes: determining the trust weight of the first defect description attribute of each first defect circuit information set for the characteristics of the first line setting directory based on the first hidden vector of each first defect circuit information set and the second hidden vector of the corresponding second defect circuit information set.

4. The method according to claim 1, wherein The optimizing each first defect description attribute by using the obtained first impact vector, and comparing the optimized first defect description attribute with the second defect description attribute to obtain the defect detection result of the first line setting directory, includes: Each first defect description attribute, a first influence vector between each first defect circuit information set, a defect description attribute of each second defect circuit information set, and a second influence vector between each second defect circuit information set are loaded into a pre-trained defect detection network, so that the defect detection network optimizes each first defect description attribute based on each first influence vector to obtain an optimized first defect description attribute, and optimizes the defect description attribute of each second defect circuit information set based on each second influence vector to obtain a second defect description attribute, and compares the optimized first defect description attribute with the second defect description attribute to output a defect detection result; and obtains the defect detection result output by the defect detection network.

5. The method according to claim 1, characterized in that The determining of a plurality of first defective circuit information sets from the first circuit setting directory using the acquired partial knowledge fragments includes: Determine the knowledge segment category to which each partial knowledge segment belongs based on the key category of the obtained partial knowledge segment and the association relationship between the set partial knowledge segment category and the knowledge segment category of the first defective circuit information set; For each knowledge segment category, based on the position information of some knowledge segments belonging to the knowledge segment category, a standard position of the defective circuit information set belonging to the knowledge segment category is obtained, and an inter-knowledge segment difference between the standard position of the defective circuit information set belonging to the knowledge segment category and the standard position of an adjacent defective circuit information set is determined. Based on the detected inter-knowledge segment difference, a knowledge segment vector of the defective circuit information set belonging to the knowledge segment category is determined; Based on the acquired standard positions and the determined knowledge fragment vectors, a first defective circuit information set to which each standard position belongs is determined.

6. The method according to claim 1, characterized in that Selecting part of the description content in each first defective circuit information set, including: Selecting a global knowledge vector set of a first line setting directory in the transmission line data to be detected; Determining, based on the position of each first defective circuit information set in the first circuit setting directory, a knowledge segment corresponding to each first defective circuit information set in the global knowledge vector set; According to the set knowledge vector set, debugging the knowledge segments of each first defective circuit information set is performed to generate a vector which is a knowledge vector of the set knowledge vector set; Partial description content corresponding to the knowledge vector of each first defective circuit information set is determined as the partial description content in each first defective circuit information set.

7. A power transmission line defect detection system based on deep learning, characterized in that: The method comprises a processor and a memory communicating with each other, wherein the processor is used to read a computer program from the memory and execute the computer program to implement the method according to any one of claims 1 to 6.

Images