Pantograph anomaly detection method and system

Abstract

The invention provides a pantograph anomaly detection method and system, and belongs to the technical field of pantograph anomaly detection based on computer vision, and the method comprises the steps: obtaining a pantograph image; the acquired pantograph image is processed by using a pre-trained anomaly detection model, a detection result is obtained, abnormal two-dimensional rectangular coordinate frame information with relatively high reliability is obtained, and accurate detection of three types of anomalies of large sparks, structural anomalies and foreign matters of the pantograph is realized. According to the pantograph positioning method for eliminating dirt interference, the problem of low pantograph anomaly detection precision caused by complex environment interference is solved. The data set is automatically generated by means of the visual large model, a large amount of manpower and material resources are not needed, a deep learning neural network method is used in the pantograph anomaly detection implementation process, parameters do not need to be manually adjusted, and complete automation and intelligentization are achieved. And meanwhile, the judgment of the reliability of the detection result is additionally added, the credible defect caused by the AI algorithm is made up, and the detection precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pantograph anomaly detection based on computer vision, and in particular to a pantograph anomaly detection method and system. Background Art

[0002] The pantograph is an important power supply component of the electric locomotive. It is installed on the roof of the electric locomotive and is exposed to a complex and changeable open-air environment for a long time. It is easily affected by friction, foreign objects, etc., which may cause the pantograph to be abnormal, thus affecting power transmission and even causing traffic interruption or safety accidents. Therefore, abnormal detection of pantographs is an important measure to ensure the safety of train operation. With the continuous improvement of rail transit operation efficiency and quality requirements, the operating status monitoring of pantographs has also put forward high requirements for high accuracy, high efficiency and high reliability.

[0003] Pantograph detection methods mainly include traditional detection and intelligent detection based on image processing. Traditional detection methods include manual detection, warehouse detection and ground detection. Manual rooftop detection is not only inefficient and inaccurate, but also poses safety risks for high-altitude operations. The warehouse detection method is inefficient and difficult to achieve real-time detection and warning, which will also affect the normal operation of the train. Ground detection also has the problems of incomplete coverage and untimely detection. The existing method of using image processing technology to perform intelligent analysis of pantograph images requires the preset of relevant parameters to achieve abnormal state detection, and the degree of automation is not high. Summary of the invention

[0004] The object of the present invention is to provide a pantograph abnormality detection method and system to solve at least one technical problem existing in the above-mentioned background technology.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] In a first aspect, the present invention provides a pantograph abnormality detection method, comprising:

[0007] Obtain pantograph images;

[0008] The acquired pantograph image is processed using a pre-trained anomaly detection model to obtain the detection results, and the abnormal two-dimensional rectangular coordinate frame information with high reliability is obtained to achieve accurate detection of three types of anomalies in the pantograph: large sparks, structural anomalies, and foreign objects.

[0009] Furthermore, training the anomaly detection model includes: installing a camera at a fixed position on the roof of the train to monitor the pantograph in real time to obtain the original pantograph image at the current moment; using a method for accurately locating the pantograph area that excludes dirty interference to achieve pantograph positioning in a clean pantograph image, and obtain the pantograph two-dimensional rectangular coordinate frame information; using a method for automatically labeling pantograph anomalies based on a large visual model to automatically generate a pantograph anomaly data set; using the pantograph anomaly data set to train the constructed pantograph anomaly detection model to obtain three types of anomaly two-dimensional rectangular coordinate frame information of pantograph large sparks, structural anomalies, and foreign objects; using an efficient verification method for pantograph detection to achieve accurate detection of pantograph anomalies. .

[0010] Furthermore, the target detection algorithm is used to train, verify and test the dirty pantograph dataset to obtain the optimal weight, realize the binary classification of dirty pantograph and clean pantograph images, and obtain the two-dimensional rectangular coordinate frame of the clean pantograph.

[0011] Furthermore, an automatic pantograph anomaly labeling method based on a visual big model is used to automatically generate a pantograph anomaly data set, including: preparing one pantograph anomaly image of each type, and labeling them accordingly to form a one-to-one pantograph anomaly data set, which includes the original image and the labeling file; the pantograph anomaly data set of each anomaly category is input into the visual big model for data generation to obtain a complete pantograph anomaly data set.

[0012] Furthermore, a real-time target detection algorithm is used to train, verify and test the pantograph abnormality data set to obtain the optimal weight, thereby achieving preliminary detection results of three pantograph abnormality categories: large sparks, structural abnormalities and foreign objects; if there are abnormal categories, the abnormal two-dimensional rectangular coordinate frame information of each abnormal category in the preliminary detection results of the pantograph abnormality is saved in turn to form arrays.

[0013] Furthermore, an efficient verification method for pantograph anomaly detection is designed to determine the reliability of anomaly detection results by using the relative position relationship of the coordinate frame; including:

[0014] Extract the four coordinate points point1, point2, point3 and point4 of the pantograph two-dimensional rectangular coordinate frame rect1;

[0015] Sequentially extract the four coordinate points point1, point2, point3 and point4 of the abnormal two-dimensional rectangular coordinate box rect2 of each abnormal category and stack them in order;

[0016] Determine whether any point of the abnormal two-dimensional rectangular coordinate box of each abnormal category is within the two-dimensional rectangular coordinate box of the pantograph. If it exists, that is, inside = 1, it is considered that the detection result of the abnormal category belongs to the detected pantograph, and its reliability is strong; if it does not exist, that is, inside = 0, it is considered that the detection result of the abnormal category may not belong to the detected pantograph, which may be a misjudgment, and its reliability is poor, and manual screening is required.

[0017] Furthermore, for the anomaly detection results with strong reliability, the anomaly category label and the anomaly coordinate frame are drawn on the original pantograph image to achieve the output of the anomaly detection result image.

[0018] In a second aspect, the present invention provides a pantograph abnormality detection system, comprising:

[0019] An acquisition module, used for acquiring a pantograph image;

[0020] The processing module is used to process the acquired pantograph image using a pre-trained anomaly detection model to obtain the detection result, obtain the abnormal two-dimensional rectangular coordinate frame information with high reliability, and realize the accurate detection of three types of anomalies of the pantograph: large sparks, structural anomalies, and foreign objects.

[0021] In a third aspect, the present invention provides a non-transitory computer-readable storage medium, which is used to store computer instructions. When the computer instructions are executed by a processor, the pantograph abnormality detection method as described in the first aspect is implemented.

[0022] In a fourth aspect, the present invention provides a computer device comprising a memory and a processor, wherein the processor and the memory communicate with each other, the memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the pantograph abnormality detection method as described in the first aspect.

[0023] In a fifth aspect, the present invention provides an electronic device, comprising: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes instructions for implementing the pantograph abnormality detection method as described in the first aspect.

[0024] Beneficial effects of the present invention: The pantograph positioning method that eliminates interference from dirt is used to solve the problem of low accuracy in pantograph abnormality detection caused by interference from complex environments. Compared with the existing pantograph abnormality detection method based on image processing, the present invention uses a large visual model to automatically generate a data set, which no longer requires a large amount of manpower and material resources, and uses a deep learning neural network method in the process of pantograph abnormality detection. There is no need to manually adjust parameters, and full automation and intelligence are achieved. At the same time, the reliability of the detection results is additionally judged, which makes up for the credibility defects brought by the AI ​​algorithm and improves the detection accuracy. In addition, the method has no limitations on the type of pantograph abnormalities and has strong universality. The method realizes accurate, real-time and reliable detection of various types of pantograph abnormalities.

[0025] Additional advantages of the present invention will be more clearly presented in the following description or learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0027] Figure 1 This is a flow chart of the design scheme of the pantograph abnormality detection method described in an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below by the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0029] It should be understood by those skilled in the art that unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which this invention belongs.

[0030] It should also be understood that terms, such as those defined in commonly used dictionaries, should be understood to have a meaning consistent with that in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless as defined herein.

[0031] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements and / or groups thereof.

[0032] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. Different embodiments or examples described in this specification and features of different embodiments or examples may be combined and combined by those skilled in the art without contradiction.

[0033] To facilitate understanding of the present invention, the present invention is further explained below with reference to specific embodiments in conjunction with the accompanying drawings, and the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0034] Those skilled in the art should understand that the drawings are merely schematic diagrams of embodiments, and the components in the drawings are not necessarily necessary for implementing the present invention.

[0035] Example 1

[0036] In this embodiment 1, a pantograph abnormality detection system is first provided, which includes: an acquisition module for acquiring a pantograph image; a processing module for processing the acquired pantograph image using a pre-trained abnormality detection model to obtain a detection result, obtain abnormal two-dimensional rectangular coordinate frame information with high reliability, and realize accurate detection of three types of abnormalities of the pantograph, namely, large sparks, structural abnormalities, and foreign objects.

[0037] In this embodiment, the above-mentioned system is used to implement a pantograph abnormality detection method, including: obtaining a pantograph image; using a pre-trained abnormality detection model to process the obtained pantograph image to obtain a detection result, and obtaining abnormal two-dimensional rectangular coordinate frame information with high reliability, thereby achieving accurate detection of three types of abnormalities of the pantograph: large sparks, structural abnormalities, and foreign objects.

[0038] Among them, training the anomaly detection model includes: installing a camera at a fixed position on the roof of the train to monitor the pantograph in real time to obtain the original pantograph image at the current moment; using a method for accurately locating the pantograph area that eliminates interference from dirt to achieve pantograph positioning in a clean pantograph image, and obtaining the pantograph's two-dimensional rectangular coordinate frame information; using a method for automatically labeling pantograph anomalies based on a large visual model to automatically generate a pantograph anomaly data set; using the pantograph anomaly data set to train the constructed pantograph anomaly detection model to achieve accurate detection of three types of anomalies, namely, large pantograph sparks, structural anomalies, and foreign objects, and obtain the anomaly two-dimensional rectangular coordinate frame information.

[0039] The YOLOv8 target detection algorithm is used to train, verify and test the dirty pantograph dataset to obtain the optimal weight and realize the binary classification of dirty pantograph and clean pantograph images.

[0040] An automatic pantograph anomaly labeling method based on a visual big model is used to automatically generate a pantograph anomaly data set, including: preparing one pantograph anomaly image of each type and labeling them accordingly to form a one-to-one pantograph anomaly data set, which includes the original image and the labeling file; the pantograph anomaly data set of each anomaly category is input into the visual big model for data generation to obtain a complete pantograph anomaly data set.

[0041] A real-time target detection algorithm is used to train, verify and test the pantograph abnormality data set to obtain the optimal weight, and achieve preliminary detection results of three pantograph abnormality categories: large sparks, structural abnormalities, and foreign objects; if there is an abnormal category, the abnormal two-dimensional rectangular coordinate frame information of each abnormal category in the preliminary detection results of the pantograph abnormality is saved in turn to form arrays.

[0042] Design an efficient verification method for pantograph anomaly detection, and use the relative position relationship of the coordinate frame to determine the reliability of the anomaly detection results; including:

[0043] Extract the four coordinate points point1, point2, point3 and point4 of the pantograph two-dimensional rectangular coordinate frame rect1;

[0044] Sequentially extract the four coordinate points point1, point2, point3 and point4 of the abnormal two-dimensional rectangular coordinate box rect2 of each abnormal category and stack them in order;

[0045] Determine whether any point of the abnormal two-dimensional rectangular coordinate box of each abnormal category is within the two-dimensional rectangular coordinate box of the pantograph. If it exists, that is, inside = 1, it is considered that the detection result of the abnormal category belongs to the detected pantograph, and its reliability is strong; if it does not exist, that is, inside = 0, it is considered that the detection result of the abnormal category may not belong to the detected pantograph, which may be a misjudgment, and its reliability is poor, and manual screening is required.

[0046] For the anomaly detection results with strong reliability, the anomaly category label and the anomaly coordinate frame are drawn on the original pantograph image to output the anomaly detection result image.

[0047] Example 2

[0048] In this embodiment 2, a pantograph abnormality detection method is provided, including: installing a camera at a fixed position on the roof of a train to monitor the pantograph in real time to obtain an original pantograph image at the current moment; using a method for accurately locating the pantograph area that eliminates interference from dirt to achieve pantograph positioning in a clean pantograph image, and obtaining pantograph two-dimensional rectangular coordinate frame information; using a pantograph abnormality automatic labeling method based on a large visual model to automatically generate a pantograph abnormality data set; constructing a real-time pantograph abnormality detection model to achieve accurate detection of three types of abnormalities, namely, large pantograph sparks, structural abnormalities, and foreign objects, and obtain abnormal two-dimensional rectangular coordinate frame information; designing an efficient verification method for pantograph abnormality detection to determine the reliability of abnormality detection results with the help of the relative position relationship of the coordinate frame; for abnormality detection results with high reliability, drawing abnormality category labels and abnormal coordinate frames on the original pantograph image to achieve output of abnormality detection result images.

[0049] A method for accurately locating a pantograph area while eliminating dirty interference comprises: constructing a dirty pantograph data set, wherein the specific categories include dirty and clean categories; using a real-time target detection algorithm to automatically extract and classify pantograph images to achieve binary classification of dirty pantograph and clean pantograph images; and saving the two-dimensional rectangular coordinate frame information of the pantograph in the clean pantograph image.

[0050] The method for automatic labeling of pantograph anomalies based on a visual big model includes: preparing one image of each type of pantograph anomaly, and labeling it with corresponding labels to form a one-to-one pantograph anomaly data group, which includes an original image and a labeling file; inputting the pantograph anomaly data group of each abnormal category into the visual big model for data generation to obtain a complete pantograph anomaly data set.

[0051] A real-time pantograph anomaly detection model is constructed, including: using a real-time target detection algorithm to automatically extract the features of a clean pantograph image, determining whether the current image contains three types of pantograph anomalies, namely, large sparks, structural anomalies, and foreign objects, and outputting preliminary detection results of pantograph anomalies; if an abnormal category exists, the abnormal two-dimensional rectangular coordinate frame information of each category in the preliminary detection results of the pantograph anomaly is saved.

[0052] Design an efficient verification method for pantograph anomaly detection, including:

[0053] Extract the four coordinate points point1, point2, point3 and point4 of the pantograph two-dimensional rectangular coordinate frame rect1; extract the four coordinate points point1, point2, point3 and point4 of the abnormal two-dimensional rectangular coordinate frame rect2 of each abnormal category in turn; judge whether any point of the abnormal two-dimensional rectangular coordinate frame of each category is inside the pantograph two-dimensional rectangular coordinate frame. If it exists, that is, inside=1, it is considered that the detection result of the abnormal category is highly reliable; if it does not exist, that is, inside=0, it is considered that the detection result of the abnormal category is less reliable and needs manual screening.

[0054] Example 3

[0055] In this embodiment 3, a pantograph abnormality detection method is provided. In response to the demand for real-time, accurate and reliable detection of pantographs, a train roof camera is used to collect pantograph images, and a pantograph area accurate positioning method is used to accurately position the pantograph on the basis of eliminating the influence of dirt. Then, an automatic pantograph abnormality labeling method based on a large visual model is used to generate an abnormal data set, a pantograph abnormality detection model is constructed to obtain preliminary detection results of abnormal states, and a pantograph abnormality detection efficient verification method is used to obtain a final highly reliable pantograph abnormality detection result, thereby achieving real-time and accurate pantograph abnormality detection. The pantograph abnormality detection method includes the following steps:

[0056] Step 1: A camera is installed at a fixed position on the train roof to monitor the pantograph in real time. The camera's upward monitoring angle of the pantograph remains unchanged to obtain the original pantograph image at the current moment.

[0057] Step 2, using a pantograph area precise positioning method that eliminates dirty interference, to achieve pantograph positioning in a clean pantograph image, and obtain pantograph two-dimensional rectangular coordinate frame information;

[0058] The step 2 specifically includes the following steps:

[0059] Step 2.1, construct a dirty pantograph dataset, which includes two categories: dirty and clean;

[0060] Step 2.2, a real-time target detection algorithm is used to train, verify and test the dirty pantograph dataset to obtain the optimal weight and realize the binary classification of dirty pantograph and clean pantograph images;

[0061] Step 2.3, saving the two-dimensional rectangular coordinate frame information of the pantograph in the clean pantograph image to form an array rect1;

[0062] Step 3, using a pantograph anomaly automatic labeling method based on a visual large model to automatically generate a pantograph anomaly data set;

[0063] The step 3 specifically comprises the following steps:

[0064] Step 3.1, prepare one image of each type of pantograph abnormality, and annotate it with corresponding labels to form a one-to-one corresponding pantograph abnormality data set, which includes the original image and the annotation file;

[0065] Step 3.2, input the pantograph abnormal data group of each abnormal category into the visual big model for data generation, and obtain a complete pantograph abnormal data set;

[0066] Step 4: Build a real-time pantograph anomaly detection model to accurately detect three types of anomalies: large pantograph sparks, structural anomalies, and foreign objects, and obtain the abnormal two-dimensional rectangular coordinate frame information;

[0067] The step 4 specifically comprises the following steps:

[0068] Step 4.1, a real-time target detection algorithm is used to train, verify and test the pantograph anomaly data set to obtain the optimal weight and achieve preliminary detection results of three pantograph anomaly categories: large sparks, structural anomalies, and foreign objects;

[0069] Step 4.2, if there is an abnormal category, the abnormal two-dimensional rectangular coordinate frame information of each abnormal category in the preliminary detection result of the pantograph abnormality is saved in sequence to form an array rect2 respectively;

[0070] Step 5, design an efficient verification method for pantograph anomaly detection, and use the relative position relationship of the coordinate frame to determine the reliability of the anomaly detection result;

[0071] The step 5 specifically comprises the following steps:

[0072] Step 5.1, extracting four coordinate points point1, point2, point3 and point4 of the pantograph two-dimensional rectangular coordinate frame rect1;

[0073]

[0074] Step 5.2, extract the four coordinate points point1, point2, point3 and point4 of the abnormal two-dimensional rectangular coordinate frame rect2 of each abnormal category in turn, and stack them in order to facilitate subsequent coordinate determination;

[0075]

[0076] Step 5.3, respectively determine whether any point of the abnormal two-dimensional rectangular coordinate frame of each abnormal category is within the pantograph two-dimensional rectangular coordinate frame. If it exists, that is, inside = 1, it is considered that the detection result of the abnormal category belongs to the detected pantograph, and its reliability is strong; if it does not exist, that is, inside = 0, it is considered that the detection result of the abnormal category may not belong to the detected pantograph, which may be a misjudgment, and its reliability is poor, and manual screening is required;

[0077]

[0078] Step 6: For abnormal detection results with high reliability, the abnormal category label and abnormal coordinate frame are drawn on the original pantograph image to output the abnormal detection result image.

[0079] In this embodiment, a pantograph positioning method that eliminates interference from dirt is used to solve the problem of low accuracy in pantograph abnormality detection caused by interference from complex environments. Compared with the existing pantograph abnormality detection method based on image processing, the present invention uses a large visual model to automatically generate a data set, which no longer requires a large amount of manpower and material resources, and uses a deep learning neural network method in the process of pantograph abnormality detection. There is no need to manually adjust parameters, and full automation and intelligence are achieved. At the same time, the reliability of the detection results is additionally judged, which makes up for the credibility defects brought by the AI ​​algorithm and improves the detection accuracy. In addition, this method has no limitations on the type of pantograph abnormalities and has strong universality. This method realizes accurate, real-time and reliable detection of various types of pantograph abnormalities.

[0080] Example 4

[0081] like Figure 1 As shown, this embodiment provides a pantograph abnormality detection method, comprising the following steps:

[0082] Step 1: A camera is installed at a fixed position on the roof of the train to monitor the pantograph in real time and obtain the original pantograph image at the current moment;

[0083] Step 2, using a pantograph area precise positioning method that eliminates dirty interference, to achieve pantograph positioning in a clean pantograph image, and obtain pantograph two-dimensional rectangular coordinate frame information;

[0084] The step 2 specifically includes the following steps:

[0085] Step 2.1, use LabelImg labeling software to label dirty and clean pantograph data and construct a dirty pantograph dataset;

[0086] Step 2.2, use the YOLOv8 target detection algorithm to train, verify and test the dirty pantograph dataset to obtain the optimal weight and realize the binary classification of dirty pantograph and clean pantograph images;

[0087] Step 2.3, saving the two-dimensional rectangular coordinate frame information of the pantograph in the clean pantograph image to form an array rect1;

[0088] Step 3, using a pantograph anomaly automatic labeling method based on a visual large model to automatically generate a pantograph anomaly data set;

[0089] The step 3 specifically comprises the following steps:

[0090] Step 3.1, prepare one image of each type of pantograph abnormality, use LabelImg to annotate the corresponding labels, and form a one-to-one corresponding pantograph abnormality data set, which includes the original image and the annotated file;

[0091] Step 3.2, input the pantograph abnormal data group of each abnormal category into the automatic annotation visual large model for data generation, and obtain a complete pantograph abnormal data set;

[0092] Step 4: Build a real-time pantograph anomaly detection model to accurately detect three types of anomalies: large pantograph sparks, structural anomalies, and foreign objects, and obtain the abnormal two-dimensional rectangular coordinate frame information;

[0093] The step 4 specifically comprises the following steps:

[0094] Step 4.1, use the YOLOv8 target detection algorithm to train, verify and test the pantograph anomaly data set to obtain the optimal weight and achieve preliminary detection results of the three pantograph anomaly categories of large sparks, structural anomalies and foreign objects;

[0095] Step 4.2, if there is an abnormal category, the abnormal two-dimensional rectangular coordinate frame information of each abnormal category in the pantograph image is saved in sequence to form an array rect2 respectively;

[0096] Step 5, design an efficient verification method for pantograph anomaly detection, and use the relative position relationship of the coordinate frame to determine the reliability of the anomaly detection result;

[0097] The step 5 specifically comprises the following steps:

[0098] Step 5.1, extracting four coordinate points point1, point2, point3 and point4 of the pantograph two-dimensional rectangular coordinate frame rect1;

[0099]

[0100] Step 5.2, extract the four coordinate points point1, point2, point3 and point4 of the abnormal two-dimensional rectangular coordinate frame rect2 of each abnormal category in turn, and stack them in order to facilitate subsequent coordinate determination;

[0101]

[0102] Step 5.3, respectively determine whether any point of the abnormal two-dimensional rectangular coordinate frame of each abnormal category is within the pantograph two-dimensional rectangular coordinate frame. If it exists, that is, inside = 1, it is considered that the detection result of the abnormal category belongs to the detected pantograph, and its reliability is strong; if it does not exist, that is, inside = 0, it is considered that the detection result of the abnormal category may not belong to the detected pantograph, which may be a misjudgment, and its reliability is poor, and manual screening is required;

[0103]

[0104] Step 6: For abnormal detection results with high reliability, the abnormal category label and abnormal coordinate frame are drawn on the original pantograph image to output the abnormal detection result image.

[0105] In summary, in this embodiment, the detection accuracy in the pantograph anomaly detection data set reached 97%, and the detection speed was 65FPS, achieving accurate and real-time pantograph anomaly detection. The innovation of the proposed pantograph anomaly detection method is mainly reflected in the combination of four intelligent methods, namely, eliminating dirt interference, automatic data set annotation, real-time anomaly detection, and efficient verification of detection results, to achieve accurate, real-time, and highly reliable pantograph anomaly detection, and its detection efficiency, detection accuracy, detection speed, and detection reliability are all effectively guaranteed.

[0106] Example 5

[0107] This embodiment 5 provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the pantograph abnormality detection method as described above is implemented. The method includes:

[0108] Obtain pantograph images;

[0109] The acquired pantograph image is processed using a pre-trained anomaly detection model to obtain the detection results and the abnormal two-dimensional rectangular coordinate frame information, thereby realizing accurate detection of three types of anomalies: large pantograph sparks, structural anomalies, and foreign objects.

[0110] Example 6

[0111] This embodiment 6 provides a computer device, including a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions executable by the processor, the processor calls the program instructions to execute the pantograph abnormality detection method as described above, the method comprising:

[0112] Obtain pantograph images;

[0113] The acquired pantograph image is processed using a pre-trained anomaly detection model to obtain the detection results and the abnormal two-dimensional rectangular coordinate frame information, thereby realizing accurate detection of three types of anomalies: large pantograph sparks, structural anomalies, and foreign objects.

[0114] Example 7

[0115] This embodiment 7 provides an electronic device, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes instructions for implementing the pantograph abnormality detection method as described above, the method including:

[0116] Obtain pantograph images;

[0117] The acquired pantograph image is processed using a pre-trained anomaly detection model to obtain the detection results and the abnormal two-dimensional rectangular coordinate frame information, thereby realizing accurate detection of three types of anomalies: large pantograph sparks, structural anomalies, and foreign objects.

[0118] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0119] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0120] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0121] These computer program instructions can also be loaded onto a computer or other programmable data processing device, and a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0122] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative work on the basis of the technical solution disclosed in the present invention should be included in the scope of protection of the present invention.

Claims

1. A pantograph abnormality detection method, characterized in that: include: Obtain pantograph images; The acquired pantograph image is processed using a pre-trained anomaly detection model to obtain the detection results, and the abnormal two-dimensional rectangular coordinate frame information with high reliability is obtained to achieve accurate detection of three types of anomalies in the pantograph: large sparks, structural anomalies, and foreign objects.

2. The pantograph abnormality detection method according to claim 1, characterized in that: Training the anomaly detection model includes: installing a camera at a fixed position on the roof of the train to monitor the pantograph in real time to obtain the original pantograph image at the current moment; using a method for accurately locating the pantograph area that eliminates interference from dirt to achieve pantograph positioning in a clean pantograph image, and obtaining pantograph two-dimensional rectangular coordinate frame information; using a method for automatically labeling pantograph anomalies based on a large visual model to automatically generate a pantograph anomaly data set; using the pantograph anomaly data set to train the constructed pantograph anomaly detection model to obtain two-dimensional rectangular coordinate frame information of three types of anomalies, namely, large pantograph sparks, structural anomalies, and foreign objects; and using an efficient verification method for pantograph detection to achieve accurate detection of pantograph anomalies.

3. The pantograph abnormality detection method according to claim 2, characterized in that: A target detection algorithm is used to train, verify and test the dirty pantograph dataset to obtain the optimal weight, realize binary classification of dirty pantograph and clean pantograph images, and obtain the two-dimensional rectangular coordinate frame of the clean pantograph.

4. The pantograph abnormality detection method according to claim 2, characterized in that: An automatic pantograph anomaly labeling method based on a visual big model is used to automatically generate a pantograph anomaly data set, including: preparing one pantograph anomaly image of each type and labeling them accordingly to form a one-to-one pantograph anomaly data set, which includes the original image and the labeling file; the pantograph anomaly data set of each anomaly category is input into the visual big model for data generation to obtain a complete pantograph anomaly data set.

5. The pantograph abnormality detection method according to claim 2, characterized in that: A real-time target detection algorithm is used to train, verify and test the pantograph anomaly data set to obtain the optimal weights and achieve preliminary detection results for three pantograph anomaly categories: large sparks, structural anomalies and foreign objects. If there is an abnormal category, the abnormal two-dimensional rectangular coordinate frame information of each abnormal category in the preliminary detection result of the pantograph abnormality is saved in sequence to form arrays respectively.

6. The pantograph abnormality detection method according to claim 2, characterized in that: Design an efficient verification method for pantograph anomaly detection, and use the relative position relationship of the coordinate frame to determine the reliability of the anomaly detection results; including: Extract the four coordinate points point1, point2, point3 and point4 of the pantograph two-dimensional rectangular coordinate frame rect1; Sequentially extract the four coordinate points point1, point2, point3 and point4 of the abnormal two-dimensional rectangular coordinate box rect2 of each abnormal category and stack them in order; Determine whether any point of the abnormal two-dimensional rectangular coordinate box of each abnormal category is within the two-dimensional rectangular coordinate box of the pantograph. If it exists, that is, inside = 1, it is considered that the detection result of the abnormal category belongs to the detected pantograph, and its reliability is strong; if it does not exist, that is, inside = 0, it is considered that the detection result of the abnormal category may not belong to the detected pantograph, which may be a misjudgment, and its reliability is poor, and manual screening is required.

7. The pantograph abnormality detection method according to claim 2, characterized in that: For the anomaly detection results with strong reliability, the anomaly category label and the anomaly coordinate frame are drawn on the original pantograph image to output the anomaly detection result image.

8. A pantograph abnormality detection system, characterized in that: include: An acquisition module, used for acquiring a pantograph image; The processing module is used to process the acquired pantograph image using a pre-trained anomaly detection model to obtain the detection result, obtain the abnormal two-dimensional rectangular coordinate frame information with high reliability, and realize the accurate detection of three types of anomalies of the pantograph: large sparks, structural anomalies, and foreign objects.

9. A computer device, characterized in that: It includes a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the pantograph abnormality detection method as described in any one of claims 1-7.

10. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes instructions for implementing the pantograph abnormality detection method as described in any one of claims 1-7.

Images