Train vehicle chassis parts abnormality detection method and device, electronic equipment and medium

Abstract

Embodiments of the present invention provide a method and device, electronic device, and medium for detecting abnormalities in train vehicle chassis parts. The method includes: capturing images of the train vehicle chassis from different perspectives using a camera; constructing a U-Net abnormality detection model using a deep learning algorithm; analyzing the images using the U-Net abnormality detection model to obtain image data of target parts in the train vehicle chassis, wherein the U-Net abnormality detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module; converting the image data into a feature vector; performing dimensionality reduction and reconstruction on the feature vector using the Autoencoder module to calculate a reconstruction error; and determining that a train vehicle chassis part abnormality event has occurred in the target part if the reconstruction error exceeds a preset threshold. Therefore, it is possible to effectively detect abnormal events in different types of train vehicle chassis parts.

Description

Technical Field

[0001] The present invention relates to the technical field of train operation safety management, and in particular to a method and device for detecting abnormalities in chassis parts of train vehicles, electronic equipment, and a medium. Background Art

[0002] Current anomaly detection for train chassis parts relies primarily on manual inspection, which is inefficient. Existing detection systems can only detect a single event and cannot detect different types of anomalies. Furthermore, due to the complex nature of train chassis inspection scenarios, existing detection methods are unable to effectively detect part anomalies. Summary of the Invention

[0003] The main purpose of the embodiments of the present invention is to provide a method and device for detecting abnormalities in chassis parts of train vehicles, electronic equipment and media, which can effectively detect abnormal events of different types of chassis parts of train vehicles.

[0004] To achieve the above-mentioned object, a first aspect of an embodiment of the present invention provides a method for detecting abnormalities in chassis parts of a train vehicle, the method comprising:

[0005] Capturing images of the train vehicle chassis from different perspectives by using a camera;

[0006] Use deep learning algorithms to build a U-Net anomaly detection model;

[0007] Analyzing the image using the U-Net anomaly detection model to obtain image data of a target part in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module;

[0008] converting the image data into a feature vector;

[0009] Performing dimensionality reduction and reconstruction on the feature vector through the Autoencoder module, and calculating the reconstruction error;

[0010] When it is determined that the reconstruction error exceeds a preset threshold, it is determined that an abnormal event of a train vehicle chassis part occurs in the target part.

[0011] In some embodiments, the training method of the U-Net anomaly detection model is as follows:

[0012] Use the Autoencoder module to calculate the reconstruction error of each normal sample;

[0013] Generate, from the reconstruction error, a larger number of abnormal samples than the normal samples using a GAN module;

[0014] The normal samples and the generated abnormal samples are mixed together to form a new data set, and a binary classification model is used for training to obtain the trained U-Net anomaly detection model.

[0015] In some embodiments, before analyzing the image using the U-Net anomaly detection model to obtain image data of the target part in the train vehicle chassis, the U-Net anomaly detection model includes the EfficientNets network model, the Autoencoder module, and the GAN module, further comprising:

[0016] The image is processed using the YOLOv7 target detection algorithm to identify the target parts in the image.

[0017] In some embodiments, the EfficientNets are used as an encoder for the U-Net anomaly detection model to improve the feature extraction and characterization capabilities of the U-Net anomaly detection model.

[0018] In some embodiments, the method further comprises:

[0019] The world coordinates of the target part are obtained through a calibration algorithm.

[0020] In some embodiments, obtaining the world coordinates of the target part by a calibration algorithm includes:

[0021] Obtaining internal parameters of the camera;

[0022] Calculating a translation vector of the camera model perpendicular to the target part;

[0023] Calculate the rotation matrix of the world coordinate system around the image coordinate system;

[0024] Obtaining a coordinate conversion relationship between image coordinates and world coordinates based on the actual position of the camera;

[0025] The world coordinates of the target part are obtained based on the coordinate transformation relationship.

[0026] In some embodiments, when it is determined that the reconstruction error exceeds a preset threshold, after determining that the target part has an abnormal train vehicle chassis part event, the method further includes:

[0027] Provide early warning and report on abnormal events of the train vehicle chassis parts, wherein the abnormal events of the train vehicle chassis parts include a broken train vehicle chassis drive shaft, loose train vehicle brakes, and damaged or lost train vehicle shock absorbers.

[0028] To achieve the above-mentioned object, a second aspect of an embodiment of the present invention provides a train vehicle chassis parts abnormality detection device, the device comprising:

[0029] A shooting module is used to capture images of the train chassis from different perspectives through a camera;

[0030] A building block for building a U-Net anomaly detection model using deep learning algorithms;

[0031] An analysis module is configured to analyze the image using the U-Net anomaly detection model to obtain image data of a target part in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module;

[0032] A conversion module, configured to convert the image data into a feature vector;

[0033] A reconstruction module, configured to reduce the dimension of the feature vector and reconstruct it through the Autoencoder module, and calculate a reconstruction error;

[0034] The judgment module is used to judge that an abnormal event of a train vehicle chassis part occurs in the target part when it is determined that the reconstruction error exceeds a preset threshold.

[0035] To achieve the above objectives, a third aspect of an embodiment of the present invention provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the method described in the first aspect when executing the computer program.

[0036] To achieve the above objectives, a fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect is implemented.

[0037] The present invention proposes a train vehicle chassis part anomaly detection method and device, electronic device, and medium. The method captures images of a train vehicle chassis from different perspectives using a camera; a U-Net anomaly detection model is constructed using a deep learning algorithm; the image is analyzed using the U-Net anomaly detection model to obtain image data of a target part in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module; the image data is converted into a feature vector; the feature vector is subjected to dimensionality reduction and reconstruction using the Autoencoder module to calculate a reconstruction error; and when it is determined that the reconstruction error exceeds a preset threshold, it is determined that a train vehicle chassis part anomaly event has occurred in the target part. Based on this, compared to existing manual detection methods, the embodiment of the present invention uses machine vision technology and a U-Net anomaly detection model to achieve anomaly detection of train vehicle chassis parts. The U-Net anomaly detection model is constructed by using a deep learning algorithm; the image is analyzed using the U-Net anomaly detection model to obtain image data of the target parts in the train vehicle chassis. The U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module. Among them, EfficientNets uses a technology called Neural Architecture Search (NAS) to automatically search for an efficient neural network structure to adapt to different image classification tasks. U-Net is a network used for image segmentation. It can divide an image into different regions to facilitate analysis of different parts of the image. Incorporating EfficientNets into U-Net and using EfficientNets as the encoder of U-Net can improve the feature extraction and representation capabilities of U-Net. To improve the robustness of the model, an Autoencoder module was added to the U-Net decoding component. A GAN was also used to further optimize the model's generative capabilities and accuracy. By combining different algorithmic modules, the inability to effectively detect abnormal train chassis components due to the complexities of the inspection scenario was addressed. Consequently, this embodiment of the present invention is capable of effectively detecting various abnormal train chassis component events. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flow chart of a method for detecting abnormalities in chassis parts of train vehicles provided by an embodiment of the present invention;

[0039] Figure 2 1 is a schematic structural diagram of a train vehicle chassis parts abnormality detection device provided by an embodiment of the present invention;

[0040] Figure 3It is a schematic diagram of the hardware structure of the electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0042] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and so on, in the specification, claims, and drawings, are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0043] In view of the fact that the detection of abnormal events of train vehicle chassis parts in the prior art mainly relies on manual detection, which is inefficient, and due to the complexity of the train vehicle chassis detection scenario, the existing detection method has the problem of being unable to effectively detect part abnormalities. The embodiments of the present invention provide a train vehicle chassis part abnormality detection method and device, electronic equipment and medium, which obtain images of the train vehicle chassis from different perspectives by camera shooting; use a deep learning algorithm to construct a U-Net anomaly detection model; analyze the image through the U-Net anomaly detection model to obtain image data of the target part in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module and a GAN module; convert the image data into a feature vector; reduce the dimension and reconstruct the feature vector through the Autoencoder module, and calculate the reconstruction error; when it is determined that the reconstruction error exceeds a preset threshold, it is determined that the target part has a train vehicle chassis part abnormal event. Based on this, compared to existing manual detection methods, the embodiment of the present invention uses machine vision technology and a U-Net anomaly detection model to achieve anomaly detection of train vehicle chassis parts. The U-Net anomaly detection model is constructed by using a deep learning algorithm; the image is analyzed using the U-Net anomaly detection model to obtain image data of the target parts in the train vehicle chassis. The U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module. Among them, EfficientNets uses a technology called Neural Architecture Search (NAS) to automatically search for an efficient neural network structure to adapt to different image classification tasks. U-Net is a network used for image segmentation. It can divide an image into different regions to facilitate analysis of different parts of the image. Incorporating EfficientNets into U-Net and using EfficientNets as the encoder of U-Net can improve the feature extraction and representation capabilities of U-Net. To improve the robustness of the model, an Autoencoder module was added to the U-Net decoding component. A GAN was also used to further optimize the model's generative capabilities and accuracy. By combining different algorithmic modules, the inability to effectively detect abnormal train chassis components due to the complexities of the inspection scenario was addressed. Consequently, this embodiment of the present invention is capable of effectively detecting various abnormal train chassis component events.

[0044] The train vehicle chassis parts abnormality detection method and device, electronic equipment and medium provided by the embodiments of the present invention are specifically described through the following embodiments. First, the train vehicle chassis parts abnormality detection method in the embodiment of the present invention is described.

[0045] Figure 1 This is an optional flow chart of a method for detecting abnormalities in a train chassis part provided by an embodiment of the present invention. Figure 1 The method may include but is not limited to steps S101 to S106.

[0046] Step S101, capturing images of a train chassis from different perspectives using a camera;

[0047] Step S102, using a deep learning algorithm to build a U-Net anomaly detection model;

[0048] Step S103: Analyze the image using a U-Net anomaly detection model to obtain image data of the target part in the train chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module;

[0049] Step S104, converting the image data into a feature vector;

[0050] Step S105, reducing the dimension and reconstructing the feature vector through the Autoencoder module, and calculating the reconstruction error;

[0051] Step S106 : When it is determined that the reconstruction error exceeds the preset threshold, it is determined that an abnormal event of the train vehicle chassis part occurs in the target part.

[0052] In some embodiments, abnormal events of train vehicle chassis parts include a broken train vehicle chassis drive shaft, loose train vehicle brakes, broken or lost train vehicle shock absorbers, etc.

[0053] In some embodiments, since the chassis of a train vehicle is composed of different components, in order to comprehensively inspect the chassis of the train vehicle, it is necessary to obtain views of the train from multiple different angles, for example, side views of the left and right sides of the train vehicle chassis, a top view of the train vehicle (if the train vehicle chassis has been disassembled), and a bottom view of the train vehicle chassis. Pictures can be taken from different positions according to the parts that need to be inspected.

[0054] In some embodiments, the present invention adopts the YOLOv7 target detection algorithm when identifying road facilities. The YOLOv7 target detection algorithm can simultaneously identify multiple different part targets.

[0055] In some embodiments, when constructing the algorithm model, the embodiments of the present invention take into account the problems such as gradient explosion of the algorithm model caused by too many layers of the deep neural network when using the deep neural network model for anomaly detection. Therefore, the present invention adopts the U-Net anomaly detection model and adds the EfficientNets network to the U-Net. U-Net is mainly used for image anomaly detection tasks. EfficientNets is an efficient deep neural network model that can improve the feature extraction and characterization capabilities of U-Net. At the same time, the Autoencoder module and the GAN module are also added, so that the entire model has better feature extraction and characterization capabilities, reconstruction capabilities, generalization capabilities, generation capabilities and accuracy, and can achieve more refined and efficient part image anomaly detection tasks.

[0056] In some embodiments, EfficientNets and U-Net are both image analysis frameworks based on convolutional neural networks. EfficientNets uses a technique called Neural Architecture Search (NAS) to automatically search for an efficient neural network structure suitable for different image classification tasks. U-Net, on the other hand, is a network used for image segmentation, which can divide an image into different regions to facilitate analysis of different parts of the image. Both frameworks have excellent feature extraction and dimensionality reduction capabilities, converting high-dimensional image data into low-dimensional feature vectors. Secondly, Autoencoder is a commonly used unsupervised learning algorithm that can map high-dimensional data to a low-dimensional space and then restore the low-dimensional data to its original dimension. In anomaly detection, Autoencoder can be used to reduce the dimensionality of image data and perform anomaly detection based on reconstruction error. The feature vector extracted by EfficientNets or U-Net is used as input to the Autoencoder, which performs encoding and decoding, and then calculates the reconstruction error. If the reconstruction error exceeds a certain threshold, the sample is considered an anomaly.

[0057] In some embodiments, the GAN module is a generative model that can generate new data samples from random noise. In anomaly detection, GAN can be used to generate more abnormal samples of parts, and these samples can be merged with the original normal samples to form a richer data set. Specifically, the reconstruction error of each normal sample can be calculated using Autoencoder, and then GAN can be used to generate more abnormal samples from the reconstruction error. This process is equivalent to expanding the data set to increase the richness of the data set. Finally, the normal samples and the generated abnormal samples are mixed together to form a new data set, and a binary classification model is used for training. During testing, the new sample is input into the model to detect whether it is an abnormal sample. Specifically, the feature vector extracted by EfficientNets or U-Net can be used as input, and it can be reduced in dimension and reconstructed by Autoencoder, and then the reconstruction error is calculated. If the reconstruction error is higher than a certain threshold, the sample can be regarded as an abnormal sample.

[0058] In some embodiments, EfficientNets is added to U-Net, and an Autoencoder module is added to the decoding part of U-Net. At the same time, a GAN module is used to further optimize the generation capability and accuracy of the model. By combining different algorithm modules, the problem of being unable to effectively detect abnormal parts due to the complexity of the train vehicle chassis detection scenario is solved. Moreover, the use of the above model can realize the detection of different types of abnormal events of vehicle chassis parts, and effectively detect problems such as deformation, loss, and loosening of parts.

[0059] In some embodiments, to pinpoint the actual location of an abnormal event occurring on a road facility and improve the accuracy of abnormal event detection, the present invention calibrates the camera and uses a calibration algorithm to determine the actual position of the target part being inspected. First, the camera's internal parameters are acquired, and then the translation vector of the camera model perpendicular to the target part being inspected is calculated. The rotation matrix of the world coordinate system around the image coordinate system is then calculated. Combined with the camera's actual position, the coordinate transformation relationship between the image coordinates and the world coordinates is derived. Based on this coordinate transformation relationship, the world coordinates of the target part are obtained. The world coordinate position is the actual position of the target in real space.

[0060] In some embodiments, during the U-Net anomaly detection model training phase, the training data is input into the Yolov7+ResNet model for data training to obtain the detection accuracy of the detection model. When training the ResNet network model, 80% of the data is used for training and 20% for testing. If the required detection accuracy is not achieved, the ResNet network model is trained using the data again until the required accuracy is achieved.

[0061] In some embodiments, images of a train vehicle chassis from different perspectives are obtained by taking pictures with a camera; a U-Net anomaly detection model is constructed using a deep learning algorithm; the images are analyzed using the U-Net anomaly detection model to obtain image data of a target part in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module; the image data is converted into a feature vector; the feature vector is reduced in dimension and reconstructed using the Autoencoder module to calculate a reconstruction error; and when it is determined that the reconstruction error exceeds a preset threshold, it is determined that a train vehicle chassis part abnormality event has occurred in the target part. Based on this, compared to existing manual detection methods, the embodiment of the present invention uses machine vision technology and a U-Net anomaly detection model to achieve anomaly detection of train vehicle chassis parts. The U-Net anomaly detection model is constructed by using a deep learning algorithm; the image is analyzed using the U-Net anomaly detection model to obtain image data of the target parts in the train vehicle chassis. The U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module. Among them, EfficientNets uses a technology called Neural Architecture Search (NAS) to automatically search for an efficient neural network structure to adapt to different image classification tasks. U-Net is a network used for image segmentation. It can divide an image into different regions to facilitate analysis of different parts of the image. Incorporating EfficientNets into U-Net and using EfficientNets as the encoder of U-Net can improve the feature extraction and representation capabilities of U-Net. To improve the robustness of the model, an Autoencoder module was added to the U-Net decoding component. A GAN was also used to further optimize the model's generative capabilities and accuracy. By combining different algorithmic modules, the inability to effectively detect abnormal train chassis components due to the complexities of the inspection scenario was addressed. Consequently, this embodiment of the present invention is capable of effectively detecting various abnormal train chassis component events.

[0062] In some embodiments, step S106 may be followed by, but not limited to, step S107:

[0063] Step S107: issuing an early warning and reporting of abnormal events of the train vehicle chassis parts, wherein the abnormal events of the train vehicle chassis parts include a broken train vehicle chassis drive shaft, loose train vehicle brakes, and damaged or lost train vehicle shock absorbers.

[0064] In some embodiments, the embodiments of the present invention use machine vision technology and U-Net anomaly detection model to detect anomalies in train vehicle chassis parts, and can report the incidents to relevant departments in real time, playing the role of real-time detection and real-time warning.

[0065] See also Figure 2 The embodiment of the present invention further provides a train vehicle chassis parts abnormality detection device, which can implement the above train vehicle chassis parts abnormality detection method, and the device includes:

[0066] A shooting module 210 is used to capture images of the train chassis from different perspectives using a camera;

[0067] A construction module 220 is used to construct a U-Net anomaly detection model using a deep learning algorithm;

[0068] An analysis module 230 is configured to analyze the image using a U-Net anomaly detection model to obtain image data of target parts in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module;

[0069] A conversion module 240, configured to convert image data into a feature vector;

[0070] The reconstruction module 250 is used to reduce the dimension and reconstruct the feature vector through the Autoencoder module, and calculate the reconstruction error;

[0071] The judgment module 260 is configured to judge that an abnormal train vehicle chassis part event occurs in the target part when it is determined that the reconstruction error exceeds a preset threshold.

[0072] Based on this, in the train vehicle chassis parts abnormality detection device of an embodiment of the present invention, the shooting module 210 obtains images of the train vehicle chassis from different perspectives through camera shooting; the construction module 220 uses a deep learning algorithm to construct a U-Net abnormality detection model; the analysis module 230 analyzes the image through the U-Net abnormality detection model to obtain image data of the target part in the train vehicle chassis, wherein the U-Net abnormality detection model includes an EfficientNets network model, an Autoencoder module and a GAN module; the conversion module 240 converts the image data into a feature vector; the reconstruction module 250 reduces the dimension and reconstructs the feature vector through the Autoencoder module to calculate the reconstruction error; the judgment module 260 determines that a train vehicle chassis part abnormality event has occurred in the target part when it is determined that the reconstruction error exceeds a preset threshold. An embodiment of the present invention captures images of a train vehicle chassis from different perspectives through camera photography; constructs a U-Net anomaly detection model using a deep learning algorithm; analyzes the images using the U-Net anomaly detection model to obtain image data of a target part in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module; converts the image data into a feature vector; performs dimensionality reduction and reconstruction on the feature vector using the Autoencoder module to calculate a reconstruction error; and when it is determined that the reconstruction error exceeds a preset threshold, determines that a train vehicle chassis part anomaly event has occurred in the target part. Based on this, compared to existing manual detection methods, the embodiment of the present invention uses machine vision technology and a U-Net anomaly detection model to achieve anomaly detection of train vehicle chassis parts. The U-Net anomaly detection model is constructed by using a deep learning algorithm; the image is analyzed using the U-Net anomaly detection model to obtain image data of the target parts in the train vehicle chassis. The U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module. Among them, EfficientNets uses a technology called Neural Architecture Search (NAS) to automatically search for an efficient neural network structure to adapt to different image classification tasks. U-Net is a network used for image segmentation. It can divide an image into different regions to facilitate analysis of different parts of the image. Incorporating EfficientNets into U-Net and using EfficientNets as the encoder of U-Net can improve the feature extraction and representation capabilities of U-Net.To improve the robustness of the model, an Autoencoder module was added to the U-Net decoding component. A GAN was also used to further optimize the model's generative capabilities and accuracy. By combining different algorithmic modules, the inability to effectively detect abnormal train chassis components due to the complexities of the inspection scenario was addressed. Consequently, this embodiment of the present invention is capable of effectively detecting various abnormal train chassis component events.

[0073] The specific implementation of the train vehicle chassis parts abnormality detection device is basically the same as the specific embodiment of the above-mentioned train vehicle chassis parts abnormality detection method, and will not be repeated here.

[0074] An embodiment of the present invention further provides an electronic device comprising a memory and a processor. The memory stores a computer program, and the processor implements the aforementioned train vehicle chassis part anomaly detection method when executing the computer program. The electronic device can be any intelligent terminal, including a tablet computer and an onboard computer.

[0075] See also Figure 3 , Figure 3 The hardware structure of an electronic device according to another embodiment is shown. The electronic device includes:

[0076] The processor 301 can be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided by the embodiments of the present invention.

[0077] The memory 302 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 302 can store an operating system and other applications. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 302, and the processor 301 is called to execute the train vehicle chassis part abnormality detection method of the embodiment of the present invention, that is, obtaining images of the train vehicle chassis from different perspectives by shooting with a camera; constructing a U-Net abnormality detection model using a deep learning algorithm; analyzing the image by the U-Net abnormality detection model to obtain image data of the target part in the train vehicle chassis, wherein the U-Net abnormality detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module; converting the image data into a feature vector; performing dimensionality reduction and reconstruction on the feature vector by the Autoencoder module to calculate the reconstruction error; and when it is determined that the reconstruction error exceeds a preset threshold, determining that a train vehicle chassis part abnormality event has occurred in the target part. Based on this, compared to existing manual detection methods, the embodiment of the present invention uses machine vision technology and a U-Net anomaly detection model to achieve anomaly detection of train vehicle chassis parts. The U-Net anomaly detection model is constructed by using a deep learning algorithm; the image is analyzed using the U-Net anomaly detection model to obtain image data of the target parts in the train vehicle chassis. The U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module. Among them, EfficientNets uses a technology called Neural Architecture Search (NAS) to automatically search for an efficient neural network structure to adapt to different image classification tasks. U-Net is a network used for image segmentation. It can divide an image into different regions to facilitate analysis of different parts of the image. Incorporating EfficientNets into U-Net and using EfficientNets as the encoder of U-Net can improve the feature extraction and representation capabilities of U-Net. To improve the robustness of the model, an Autoencoder module was added to the U-Net decoding component. A GAN was also used to further optimize the model's generative capabilities and accuracy. By combining different algorithmic modules, the inability to effectively detect abnormal train chassis components due to the complexities of the inspection scenario was addressed. Consequently, this embodiment of the present invention is capable of effectively detecting various abnormal train chassis component events.

[0078] The input / output interface 303 is used to implement information input and output.

[0079] The communication interface 304 is used to realize communication interaction between this device and other devices. Communication can be realized through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0080] The bus transmits information between various components of the device (eg, the processor 301 , the memory 302 , the input / output interface 303 , and the communication interface 304 ).

[0081] The processor 301 , the memory 302 , the input / output interface 303 and the communication interface 304 are connected to each other in communication within the device via a bus.

[0082] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, which implements the above-mentioned train vehicle chassis parts abnormality detection method when executed by a processor.

[0083] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0084] The embodiments of the present invention provide a train vehicle chassis part anomaly detection method, a train vehicle chassis part anomaly detection device, an electronic device, and a storage medium. The methods capture images of a train vehicle chassis from different perspectives using a camera; a U-Net anomaly detection model is constructed using a deep learning algorithm; the images are analyzed using the U-Net anomaly detection model to obtain image data of target parts in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module; the image data is converted into a feature vector; the feature vector is subjected to dimensionality reduction and reconstruction using the Autoencoder module to calculate a reconstruction error; and when it is determined that the reconstruction error exceeds a preset threshold, it is determined that a train vehicle chassis part anomaly event has occurred in the target part. Based on this, compared to existing manual detection methods, the embodiment of the present invention uses machine vision technology and a U-Net anomaly detection model to achieve anomaly detection of train vehicle chassis parts. The U-Net anomaly detection model is constructed by using a deep learning algorithm; the image is analyzed using the U-Net anomaly detection model to obtain image data of the target parts in the train vehicle chassis. The U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module. Among them, EfficientNets uses a technology called Neural Architecture Search (NAS) to automatically search for an efficient neural network structure to adapt to different image classification tasks. U-Net is a network used for image segmentation. It can divide an image into different regions to facilitate analysis of different parts of the image. Incorporating EfficientNets into U-Net and using EfficientNets as the encoder of U-Net can improve the feature extraction and representation capabilities of U-Net. To improve the robustness of the model, an Autoencoder module was added to the U-Net decoding component. A GAN was also used to further optimize the model's generative capabilities and accuracy. By combining different algorithmic modules, the inability to effectively detect abnormal train chassis components due to the complexities of the inspection scenario was addressed. Consequently, this embodiment of the present invention is capable of effectively detecting various abnormal train chassis component events.

[0085] Those skilled in the art will appreciate that all or some of the steps and systems in the method disclosed above can be implemented as software, firmware, hardware, and appropriate combinations thereof. Some physical components or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, and the computer-readable medium can include computer storage media (or non-transitory media) and communication media (or temporary media). As known to those skilled in the art, the term computer storage media is included in any method or technology for storing information (such as a computer-readable program, a data structure, a program module, or other data) and is volatile and non-volatile, removable, and non-removable. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory, or other memory technology, CD-ROM, digital versatile disks (DVD), or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage, or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, as is well known to those skilled in the art, communication media typically embodies computer-readable programs, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0086] The embodiments described in the embodiments of the present invention are intended to more clearly illustrate the technical solutions of the embodiments of the present invention and do not constitute a limitation on the technical solutions provided by the embodiments of the present invention. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present invention are also applicable to similar technical problems.

[0087] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present invention, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0088] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0089] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0090] The terms "first," "second," "third," "fourth," and the like (if any) in the description of the present invention and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments of the present invention described herein can be implemented in orders other than those illustrated or described herein. In addition, the terms "including" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such processes, methods, products, or apparatus.

[0091] It should be understood that in the present invention, "at least one (item)" refers to one or more, and "plurality" refers to two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can represent: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0092] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the above units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0093] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0094] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0095] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes multiple instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, and other media that can store programs.

[0096] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the invention is not limited thereby. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the invention should be within the scope of the invention.

Claims

1. A method for detecting abnormalities in chassis parts of train vehicles, characterized in that: The method comprises: Capturing images of the train vehicle chassis from different perspectives by using a camera; Use deep learning algorithms to build a U-Net anomaly detection model; Analyzing the image using the U-Net anomaly detection model to obtain image data of a target part in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module; converting the image data into a feature vector; Performing dimensionality reduction and reconstruction on the feature vector through the Autoencoder module, and calculating the reconstruction error; If it is determined that the reconstruction error exceeds a preset threshold, it is determined that an abnormal event of a train vehicle chassis part occurs in the target part; The training method of the U-Net anomaly detection model is as follows: Use the Autoencoder module to calculate the reconstruction error of each normal sample; Generate, from the reconstruction error, a larger number of abnormal samples than the normal samples using a GAN module; The normal samples and the generated abnormal samples are mixed together to form a new data set, and a binary classification model is used for training to obtain the trained U-Net anomaly detection model.

2. The method according to claim 1, characterized in that Before analyzing the image by the U-Net anomaly detection model to obtain image data of the target part in the train vehicle chassis, the U-Net anomaly detection model includes the EfficientNets network model, the Autoencoder module, and the GAN module, and further includes: The image is processed using the YOLOv7 target detection algorithm to identify the target parts in the image.

3. The method according to claim 2, characterized in that The EfficientNets serves as an encoder for the U-Net anomaly detection model to improve the feature extraction and representation capabilities of the U-Net anomaly detection model.

4. The method according to claim 1, wherein The method further comprises: The world coordinates of the target part are obtained through a calibration algorithm.

5. The method according to claim 4, characterized in that The obtaining of the world coordinates of the target part by a calibration algorithm includes: Obtaining internal parameters of the camera; Calculating a translation vector of the camera model perpendicular to the target part; Calculate the rotation matrix of the world coordinate system around the image coordinate system; Obtaining a coordinate conversion relationship between image coordinates and world coordinates based on the actual position of the camera; The world coordinates of the target part are obtained based on the coordinate transformation relationship.

6. The method according to any one of claims 1 to 5, characterized in that When it is determined that the reconstruction error exceeds a preset threshold, after determining that the target part has an abnormal train vehicle chassis part event, the method further includes: Provide early warning and report on abnormal events of the train vehicle chassis parts, wherein the abnormal events of the train vehicle chassis parts include a broken train vehicle chassis drive shaft, loose train vehicle brakes, and damaged or lost train vehicle shock absorbers.

7. A train chassis parts abnormality detection device, characterized in that: The device comprises: A shooting module is used to capture images of the train chassis from different perspectives through a camera; A building block for building a U-Net anomaly detection model using deep learning algorithms; An analysis module is configured to analyze the image using the U-Net anomaly detection model to obtain image data of a target part in the train vehicle chassis, wherein the U-Net anomaly detection model includes an EfficientNets network model, an Autoencoder module, and a GAN module; A conversion module, configured to convert the image data into a feature vector; A reconstruction module, configured to reduce the dimension of the feature vector and reconstruct it through the Autoencoder module, and calculate a reconstruction error; a judgment module, configured to judge that an abnormal train vehicle chassis part event occurs in the target part when it is determined that the reconstruction error exceeds a preset threshold; The training method of the U-Net anomaly detection model is as follows: Use the Autoencoder module to calculate the reconstruction error of each normal sample; Generate, from the reconstruction error, a larger number of abnormal samples than the normal samples using a GAN module; The normal samples and the generated abnormal samples are mixed together to form a new data set, and a binary classification model is used for training to obtain the trained U-Net anomaly detection model.

8. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the train vehicle chassis parts abnormality detection method according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the train vehicle chassis parts abnormality detection method according to any one of claims 1 to 6 is implemented.

Images