Rail transit train fault detection method, device, equipment and medium

Abstract

Embodiments of the present application provide a rail transit train fault detection method, device, equipment and medium, the method comprises: obtaining the vehicle vibration data and track surface data collected in the process of rail transit train operation;The vehicle vibration data and track surface data are preprocessed to obtain a data set;Using the first part of data in the data set, track surface fault detection is carried out, and the second part of data in the data set is used to carry out bearing fault detection and / or vehicle imbalance fault detection;Wherein, the first part of data and the second part of data are different. Through the embodiments of the present application, the joint detection of track surface fault, bearing fault and vehicle imbalance fault is realized, and the efficiency of fault detection is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicles, in particular to a fault detection method and device based on a rail transit train, equipment and a medium. BACKGROUND

[0002] With the development of urban rail transit, more stringent requirements are put forward for the safe operation of rail transit trains (such as the Yunba). In the prior art, the faults of rail transit trains and rails are detected manually, which is cumbersome and requires a large amount of manpower, and the fault detection efficiency is low. SUMMARY

[0003] In view of the above problems, a fault detection method and device based on a rail transit train, equipment and a medium are provided to overcome the above problems or at least partially solve the above problems, comprising:

[0004] A fault detection method based on a rail transit train, comprising:

[0005] Obtaining vehicle vibration data and rail surface data collected during the operation of the rail transit train;

[0006] Pretreating the vehicle vibration data and the rail surface data to obtain a data set;

[0007] Using a first part of data in the data set to detect rail surface faults, and using a second part of data in the data set to detect bearing faults and / or vehicle imbalance faults; wherein the first part of data and the second part of data are different.

[0008] Optionally, the first part of data includes a plurality of acceleration data segments and corresponding video frames, and the using the first part of data in the data set to detect rail surface faults comprises:

[0009] Determining the light and shade degree of the video frames in the data set, and using different data in the first part of data to detect rail surface faults according to the light and shade degree.

[0010] Optionally, the using different data in the first part of data to detect rail surface faults according to the light and shade degree comprises:

[0011] In the case where the light and shade degree is greater than a preset degree, using the acceleration data segments and the corresponding video frames in the first part of data to detect rail surface faults;

[0012] In the case where the light and shade degree is less than or equal to the preset degree, using the acceleration data segments in the first part of data to detect rail surface faults.

[0013] Optionally, further comprising:

[0014] Fusing track surface fault detection results of multiple rail transit trains.

[0015] Optionally, the fusing track surface fault detection results of multiple rail transit trains comprises:

[0016] Clustering defect points in the track surface fault detection results of multiple rail transit trains to obtain a clustering cluster, and determining a center point of the clustering cluster as a defect point.

[0017] Optionally, further comprising:

[0018] According to the fused track surface fault detection results, marking in a preset map.

[0019] Optionally, further comprising:

[0020] According to the fused track surface fault detection results, updating a track surface fault database.

[0021] Optionally, track surface fault detection is performed using a track surface fault detection network, and bearing fault and vehicle imbalance fault detection is performed using a bearing fault and vehicle imbalance fault detection network.

[0022] Optionally, the track surface fault detection network and / or the bearing fault and vehicle imbalance fault detection network is a convolutional neural network.

[0023] Optionally, the preprocessing of the vehicle vibration data and track surface data comprises:

[0024] Cleaning the vehicle vibration data and track surface data;

[0025] and / or, time aligning the vehicle vibration data and track surface data.

[0026] Optionally, the vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and track surface data comprises:

[0027] When there is inconsistency in data time intervals between multiple data in the acceleration data, resampling the acceleration data.

[0028] Optionally, the vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and track surface data comprises:

[0029] Converting the acceleration data from an acceleration sensor coordinate system to a vehicle coordinate system.

[0030] Optionally, the vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and track surface data comprises:

[0031] removing acceleration data caused by acceleration or deceleration of the vehicle.

[0032] Optionally, the vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and track surface data comprises:

[0033] removing acceleration data with a frequency less than a frequency threshold.

[0034] Optionally, the vehicle vibration data comprises acceleration data, and the track surface data comprises video frames, and the preprocessing of the vehicle vibration data and track surface data comprises:

[0035] segmenting the acceleration data into a plurality of acceleration data segments, and establishing a correspondence between the plurality of acceleration data segments and the video frames, position information, and timestamps.

[0036] Optionally, the preprocessing of the vehicle vibration data and track surface data comprises:

[0037] removing acceleration data segments and corresponding video frames that do not have a fault.

[0038] Optionally, the removing of acceleration data segments and corresponding video frames that do not have a fault comprises:

[0039] determining a root mean square sum of three-axis acceleration, and removing acceleration data segments and corresponding video frames with a root mean square sum less than an acceleration threshold.

[0040] A fault detection device based on rail transit trains, configured to:

[0041] obtain vehicle vibration data and track surface data collected during operation of a rail transit train;

[0042] preprocess the vehicle vibration data and track surface data to obtain a data set;

[0043] use a first portion of data in the data set to detect track surface faults, and use a second portion of data in the data set to detect bearing faults and / or vehicle imbalance faults; wherein the first portion of data and the second portion of data are different.

[0044] An electronic device comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the method as described above.

[0045] A computer-readable storage medium having stored thereon a computer program, the computer program, when executed by a processor, implementing the method as described above.

[0046] The embodiments of the present application have the following advantages:

[0047] In the embodiments of the present application, by acquiring vehicle vibration data and track surface data collected in the running process of the rail transit train, the vehicle vibration data and the track surface data are preprocessed to obtain a data set, the first part of the data in the data set is used for track surface fault detection, and the second part of the data in the data set is used for bearing fault detection and / or vehicle imbalance fault detection, the first part of the data and the second part of the data are different, the joint detection of the track surface fault, the bearing fault and the vehicle imbalance fault is realized, and manual participation is not required, and the efficiency of fault detection is improved. BRIEF DESCRIPTION OF DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the present application, the following will briefly introduce the drawings needed to be used in the description of the present application. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0049] Figure 1 is a step flow chart of a fault detection method based on a rail transit train provided by some embodiments of the present application;

[0050] Figure 2 is a schematic diagram of a track surface fault detection process provided by some embodiments of the present application;

[0051] Figure 3 is a schematic diagram of another fault detection process based on a rail transit train provided by some embodiments of the present application. DETAILED DESCRIPTION

[0052] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the following will further describe the present application in detail with reference to the drawings and specific embodiments. Obviously, the described embodiments are some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0053] Referring toFigure 1 FIG. 11 shows a flowchart of steps of a method for detecting faults of a rail transit train according to some embodiments of the present application, which can include the following steps:

[0054] In step 101, vehicle vibration data and track surface data collected during operation of the rail transit train are obtained.

[0055] In actual applications, sensors such as acceleration sensors and cameras can be deployed on the rail transit train. During operation of the rail transit train, the sensors can collect vehicle vibration data and track surface data. The client can upload the data to the server for analysis.

[0056] As an example, the vehicle vibration data can include acceleration data collected by the acceleration sensor, including acceleration data of X, Y, and Z axes. When there is a problem with the vehicle bearing or an imbalance problem with the wheel, the acceleration data will fluctuate significantly. The track surface data can include video frames collected by the camera, which can be used to reflect whether the track surface is damaged, broken, etc.

[0057] In related technologies, track surface defect detection is about track inspection trolleys, i.e., installing a camera on the inspection trolley to collect track surface information. This design is costly and cannot achieve real-time detection during vehicle operation, which cannot meet the demand of ensuring normal use of the front track during vehicle operation. Expensive and complex instruments are needed for data collection, which is costly and inefficient, and cannot meet the needs of large-scale and real-time observation. In the embodiments of the present application, a motion camera is installed on the vehicle head, and the camera can upload data to the cloud in real time. Then the video and the corresponding acceleration data are preprocessed, and the preprocessed data are input into the detection model.

[0058] In step 102, the vehicle vibration data and the track surface data are preprocessed to obtain a data set.

[0059] After obtaining the vehicle vibration data and the track surface data, the vehicle vibration data and the track surface data can be preprocessed, such as data resampling, acceleration coordinate conversion, and filtering of the acceleration data, and video frame interception, reading, and normalization of the video data.

[0060] After preprocessing, a data set can be obtained, which is used for subsequent analysis, such as a data set including preprocessed acceleration data segments, corresponding video frames, and coordinate information.

[0061] In some examples, during model training, the data set can further include corresponding track surface condition labels (such as normal or abnormal track surface conditions) and bearing condition labels (such as normal or abnormal bearing conditions), which are used as the target of model training.

[0062] In some embodiments of the present application, the preprocessing of the vehicle vibration data and track surface data comprises: cleaning the vehicle vibration data and track surface data; and / or time aligning the vehicle vibration data and track surface data.

[0063] In practical applications, the collected vehicle vibration data and track surface data can be cleaned, such as removing outliers, processing missing data, etc. The vehicle vibration data and track surface data can also be time aligned to ensure consistency in time.

[0064] In some embodiments of the present application, the preprocessing of the vehicle vibration data and track surface data comprises: resampling the acceleration data when there is inconsistency in data time intervals between multiple data in the acceleration data.

[0065] Due to equipment limitations, some data is not uniformly sampled, i.e., there is inconsistency in data time intervals between multiple data. In subsequent feature extraction, the acceleration data is subjected to fast Fourier transform. The fast Fourier transform algorithm assumes that the data is uniformly sampled. In the case of inconsistency in data time intervals, the fast Fourier transform cannot be directly performed on the signal sequence.

[0066] In order to ensure that the fast Fourier transform algorithm can accurately and effectively calculate, it is necessary to resample the data to make the time intervals between the data consistent. In some examples, spline interpolation technology can be used to resample the data to ensure that the time intervals between the data are equal, which helps to extract features in the frequency domain later and improves the processability of the data.

[0067] In some embodiments of the present application, the preprocessing of the vehicle vibration data and track surface data comprises: converting the acceleration data from an acceleration sensor coordinate system to a vehicle coordinate system.

[0068] In practical applications, Euler angles can be used for coordinate conversion to convert the acceleration data from an acceleration sensor coordinate system to a vehicle coordinate system, so that the acceleration data more accurately reflects the vehicle vibration information.

[0069] In some embodiments of the present application, the preprocessing of the vehicle vibration data and track surface data comprises: removing acceleration data that causes acceleration changes in the horizontal direction due to vehicle acceleration or deceleration.

[0070] In the case of conventional vehicle acceleration or deceleration, it can cause acceleration changes in the horizontal direction, in order to avoid this change being mistaken for track surface abnormalities, after aligning the sensor coordinate axis with the vehicle coordinate axis, this part of the data can be removed.

[0071] In some embodiments of the present application, the preprocessing of the vehicle vibration data and track surface data includes removing acceleration data with a vehicle vibration frequency less than a frequency threshold.

[0072] In order to avoid the interference of vehicle acceleration, deceleration, turning and gravity and other factors on data analysis, a filter can be used to process the acceleration data. Specifically, the acceleration data with a vehicle vibration frequency less than a frequency threshold can be removed by the filter, and the high-frequency signal component related to track surface defects is retained.

[0073] In some embodiments of the present application, the preprocessing of the vehicle vibration data and track surface data includes dividing the acceleration data into multiple acceleration data segments and establishing a correspondence between the multiple acceleration data segments and video frames, position information, and timestamps.

[0074] In practical applications, the continuous acceleration data can be divided into multiple acceleration data segments using the sliding window technique according to the vehicle speed and body length, and the video frames can be intercepted using a tool. The acceleration data segments and video frames can be one-to-one corresponding, and the position information and timestamps obtained from the Beidou system or GPS system can be matched with the data segments.

[0075] In some embodiments of the present application, the preprocessing of the vehicle vibration data and track surface data includes removing acceleration data segments and corresponding video frames that do not have faults.

[0076] In order to reduce the complexity and amount of subsequent calculations, some data is almost impossible to be faulty or have defects, and acceleration data segments and corresponding video frames that do not have faults can be removed.

[0077] In some embodiments of the present application, removing acceleration data segments and corresponding video frames that do not have faults includes determining the root mean square sum of three-axis acceleration and removing acceleration data segments and corresponding video frames with a root mean square sum less than an acceleration threshold.

[0078] In practical applications, the threshold method is used to screen the data, the root mean square sum of three-axis acceleration is calculated, and the acceleration data segments and corresponding video frames with a root mean square sum less than an acceleration threshold are removed.

[0079] In step 103, track surface fault detection is performed using a first portion of the data set, and bearing fault detection and / or vehicle imbalance fault detection is performed using a second portion of the data set; wherein the first portion of data and the second portion of data are different.

[0080] The second portion of data can include a plurality of acceleration data segments.

[0081] After obtaining the data set, for track surface fault, a first portion of data can be selected from the data set for track surface fault detection, such as the first portion of data including a plurality of acceleration data segments and corresponding video frames, for bearing fault detection and / or vehicle imbalance fault, a second portion of data can be selected from the data set for bearing fault detection and / or vehicle imbalance fault detection.

[0082] In the related art, it is not possible to simultaneously realize bearing problem, wheel imbalance and track surface joint detection in one system, while the embodiment of the present application can simultaneously realize the fault detection system of the above three functions, the system collects vehicle vibration information and track surface video information through an acceleration sensor and a camera, and performs a series of pretreatments such as resampling, coordinate conversion and filtering on the acceleration data, applies OpenCV to pretreat the video data, inputs the pretreated acceleration data set into the bearing problem and vehicle imbalance detection module, inputs the data set combined by the acceleration data and the video frames into the track surface defect detection module. Finally, the perception results of multiple vehicles for track surface defect problems are fused to improve the detection accuracy, and the detection results are fed back to the maintenance department.

[0083] In some embodiments of the present application, the first portion of data includes a plurality of acceleration data segments and corresponding video frames, and the track surface fault detection using the first portion of data in the data set includes determining the light and shade degree of the video frames in the data set, and using different data in the first portion of data for track surface fault detection according to the light and shade degree.

[0084] In some embodiments of the present application, the track surface fault detection using different data in the first portion of data according to the light and shade degree includes, in the case that the light and shade degree is greater than a preset degree, using the acceleration data segments and corresponding video frames in the first portion of data for track surface fault detection; in the case that the light and shade degree is less than or equal to the preset degree, using the acceleration data segments in the first portion of data for track surface fault detection.

[0085] In the track surface defect detection module, for the input data set, the data can be first input to the hue saturation value (HSV) color model and the openCV constructed picture classifier. Among them, the hue saturation value color model is used for feature extraction, and the openCV is used for image processing to realize the classification of video frames under different light brightness.

[0086] By analyzing the lightness and darkness of the video frame, the data can be divided into two categories, such as Figure 2 whether it is bright in the detection environment, and then different detection model 1 and detection model 2 are used for fault detection.

[0087] In the case where the lightness and darkness is greater than the preset degree, the acceleration data segment (the segment of the video frame whose lightness and darkness is greater than the preset degree) in the first part of the data and the corresponding video frame can be used for track surface fault detection. In the case where the lightness and darkness is less than or equal to the preset degree, the acceleration data segment (the segment of the video frame whose lightness and darkness is less than or equal to the preset degree) in the first part of the data, that is, without using the video frame, can be used for track surface fault detection, so as to avoid the influence of environmental factors on the track surface defect detection based on image information.

[0088] Among them, the track surface fault detection network is used for track surface fault detection, and the track surface fault detection network can be a convolutional neural network. In actual application, according to the lightness and darkness of the video frame, the data set will be transmitted to different recognition modules.

[0089] For the segment whose lightness and darkness is greater than the preset degree, it will be uploaded to the recognition module based on the fusion of bicycle acceleration and video data. This module establishes a deep learning model based on convolutional neural network (CNN) and long short-term memory network (LSTM).

[0090] Specifically, for a given video, a sequence of n frames is uniformly extracted at a pre-specified 1-second interval. To prevent the model performance from being affected by weakly correlated regions, 0 is used to fill in the boundary positions of the picture. For each frame, deep features are extracted using the Image-Net CNN, and then the extracted feature information and acceleration data are input into the LSTM network for sequence classification. The application uses an attention mechanism for feature fusion, and a multilayer perceptron (MLP) is used to calculate the similarity between the two modalities. Specifically, the image features and acceleration features are input into two different fully connected layers, respectively, and then their outputs are connected and input into a fully connected layer with an activation function. The sigmoid function is used to map the output to the [0, 1] interval as the similarity between the two modalities. Then the attention weight is used to weight and average the features. The image features and acceleration features are weighted and averaged using the attention weight calculated in the previous step to obtain the final cross-modal fusion features. Finally, the cross-modal fusion features are concatenated or added to other features (features extracted from the acceleration part) to obtain the final feature vector. The final feature vector is input into a fully connected layer with a sigmoid activation function to realize the binary classification task, i.e., to determine whether there is a defect on the track surface.

[0091] For segments with light and dark levels less than or equal to a preset level, the data is uploaded to the track surface defect recognition module based on acceleration data. This module establishes a track surface defect detection model based on a Long Short-Term Memory-Convolutional Neural Network (LSTM-FCN). The input is the acceleration data of the X, Y, and Z axes at 50 time steps, and the output is a binary classification result, i.e., whether the data is abnormal road surface. The LSTM branch is used to extract time series features, using a single-layer LSTM unit with 128 neurons and a tanh activation function. The output of the LSTM branch is sent to a Dropout layer to reduce overfitting. The FCN branch uses 2 convolutional layers and 1 fully connected layer, where each convolutional layer includes 128 filters with a size of 8 and 5, a stride of 1 and 2, and a ReLU activation function. Each convolutional layer is followed by a batch normalization and a Dropout layer. The fully connected layer includes 64 neurons with a ReLU activation function. Finally, the outputs of the LSTM and FCN branches are concatenated, and the data is mapped to a binary classification result through a fully connected layer with a sigmoid activation function.

[0092] In some embodiments of the present application, bearing fault detection and / or vehicle imbalance fault detection are performed using a bearing fault and vehicle imbalance fault detection network, which can be a convolutional neural network. In the bearing problem and vehicle imbalance detection module, a detection model based on an LSTM network can be established, which has 3 input parameters and 3 outputs. The input parameters include the pre-processed three-axis acceleration. The network has 3 output indicators, label 1 is no fault, label 2 is bearing fault, and label 3 is vehicle imbalance. The Adam optimizer is used in the LSTM unit for cost minimization, and the softmax is used as the activation function. In order to prevent overfitting, a dropout regularization with a value of 0.5 is added. The output of the LSTM layer is transmitted to the fully connected layer, which converts it into the identified type. Since the output result is divided into two categories, the size of the fully connected layer is 3.

[0093] In the related art, research on bearing problem detection using vehicle vibration analysis mostly uses traditional machine learning methods such as support vector machines and decision trees. These methods have good performance for small-scale data sets, but are not suitable for processing large-scale time series data. At the same time, traditional machine learning treats vehicle vibration information as independent time series that do not affect each other, and detects the segmented time series samples through machine learning, signal processing methods, and time series anomaly detection algorithms. This method ignores the temporal correlation between vibration signals collected from the same vehicle. In the embodiments of the present application, a bearing problem detection model based on a long short-term memory fully convolutional neural network is proposed. The LSTM-FCN model uses the structure of a convolutional neural network, which can extract features faster when processing sequence data, thereby speeding up the training speed of the model. The LSTM-FCN model can capture both long-term dependencies and local features, so it has better accuracy when processing time series data. By analyzing the characteristics of vehicle vibration, a road surface disease detection model based on LSTM-FCN is proposed, which can learn the spatiotemporal correlation of various factors affecting vehicle vibration during the detection process. Real acceleration sensor data and Beidou or GPS data can be used for model training and testing.

[0094] In some embodiments of the present application, the bearing fault detection and vehicle imbalance fault detection network can further include:

[0095] The track surface fault detection results of multiple rail transit trains are fused.

[0096] After obtaining the track surface fault detection results of a single vehicle, the track surface fault detection results of multiple rail transit trains are fused. Figure 2The multi-vehicle perception result fusion can be performed based on an OPTICS (Ordering Points To Identify the Clustering Structure, an algorithm for density clustering) algorithm to obtain a final track surface defect detection result, and thus the accuracy and reliability of track surface fault detection can be improved.

[0097] In some embodiments of the present application, the fusion of the track surface fault detection results of the multiple rail transit trains comprises: clustering defect points in the track surface fault detection results of the multiple rail transit trains to obtain clustering clusters, and determining a center point of the clustering cluster as a defect point.

[0098] In actual application, the defect points in the track surface fault detection results of the multiple rail transit trains can be clustered to realize spatial density-based clustering, each class is regarded as representing a single track surface defect, and by calculating the center point of each clustering cluster, the points of the multiple single-vehicle detection results are combined into one point, which represents the optimized position of the detected track surface defect.

[0099] In some examples, the center point can be defined by finding the point closest to other points in the cluster, such as Figure 2 The k-Means algorithm is used to find the center point.

[0100] In some embodiments of the present application, the method further comprises:

[0101] According to the fused track surface fault detection result, a preset map is marked.

[0102] In actual application, the defect points in the fused track surface fault detection result can be marked in a preset map for visualization and feedback to the maintenance department.

[0103] In some embodiments of the present application, the method further comprises:

[0104] According to the fused track surface fault detection result, a track surface fault database is updated.

[0105] In actual application, the fused track surface fault detection result can be used to update a preset track surface fault database, and the database can be updated and maintained in real time, such as deleting an anomaly from the detection result database after determining that a track surface defect has been repaired.

[0106] In the embodiment of the present application, by acquiring vehicle vibration data and track surface data collected in the running process of rail transit train, the vehicle vibration data and the track surface data are preprocessed to obtain a data set, the first part of the data in the data set is used for track surface fault detection, and the second part of the data in the data set is used for bearing fault detection and / or vehicle imbalance fault detection, the first part of the data and the second part of the data are different, the joint detection of the track surface fault, the bearing fault and the vehicle imbalance fault is realized, and the efficiency of fault detection is improved.

[0107] The following will be described in combination with Figure 3 The present application is exemplarily described:

[0108] 1. For the data acquisition and preprocessing process: by collecting acceleration data and video data, the acceleration data is resampled, coordinate conversion, filtering, and segmented data preprocessing, and the video data is preprocessed by using OpenCV tool to intercept video frames.

[0109] 2. For the process of constructing a data set: by threshold method, then by hue saturation value color model, data set 1 containing acceleration data and data set 2 containing acceleration and corresponding video frames are obtained, and then different fault detection is carried out.

[0110] 3. For the system fault detection process: in the bearing problem and wheel imbalance problem detection module, data augmentation is carried out on the acceleration data set, and then time-frequency domain feature extraction is carried out, and then fault detection is carried out based on the LSTM-FCN detection model to obtain the detection result and feedback to the maintenance department. In the track surface defect detection module, the acceleration data and the corresponding video frame data set are subjected to feature extraction, feature fusion, full connection layer processing, and detection result is obtained.

[0111] 4. For the multi-vehicle perception result fusion process: based on clustering algorithm, the track surface defect detection result is fused with the multi-vehicle perception result, and the fused track surface defect is visualized.

[0112] It should be noted that for the method embodiment, in order to simply describe, it is expressed as a series of action combinations, but those skilled in the art should know that the embodiment of the present application is not limited by the described action sequence, because according to the embodiment of the present application, certain steps can be performed in other order or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions involved are not necessarily necessary for the embodiment of the present application.

[0113] Some embodiments of the present application provide a fault detection device based on rail transit train, which can be specifically used for:

[0114] acquire vehicle vibration data and track surface data collected in the process of rail transit train operation;

[0115] preprocess the vehicle vibration data and track surface data to obtain a data set;

[0116] use a first part of data in the data set to perform track surface fault detection, and use a second part of data in the data set to perform bearing fault detection and / or vehicle imbalance fault detection; wherein the first part of data and the second part of data are different.

[0117] In some embodiments of the present application, the first part of data includes a plurality of acceleration data segments and corresponding video frames, and the use of the first part of data in the data set to perform track surface fault detection includes:

[0118] determining the light and shade degree of the video frames in the data set, and using different data in the first part of data to perform track surface fault detection according to the light and shade degree.

[0119] In some embodiments of the present application, the use of different data in the first part of data to perform track surface fault detection according to the light and shade degree includes:

[0120] in the case where the light and shade degree is greater than a preset degree, using the acceleration data segments and corresponding video frames in the first part of data to perform track surface fault detection;

[0121] in the case where the light and shade degree is less than or equal to the preset degree, using the acceleration data segments in the first part of data to perform track surface fault detection.

[0122] In some embodiments of the present application, it is also used for:

[0123] fusing track surface fault detection results of a plurality of rail transit trains.

[0124] In some embodiments of the present application, the fusion of track surface fault detection results of a plurality of rail transit trains includes:

[0125] clustering defect points in the track surface fault detection results of a plurality of rail transit trains to obtain a clustering cluster, and determining a center point of the clustering cluster as a defect point.

[0126] In some embodiments of the present application, it is also used for:

[0127] According to the fused track surface fault detection results, marking is performed in a preset map.

[0128] In some embodiments of the present application, the method further comprises:

[0129] According to the track surface fault detection result after fusion, updating the track surface fault database.

[0130] In some embodiments of the present application, the track surface fault detection network is used for track surface fault detection, and the bearing fault and vehicle imbalance fault detection network is used for bearing fault detection and / or vehicle imbalance fault detection.

[0131] In some embodiments of the present application, the track surface fault detection network and / or the bearing fault and vehicle imbalance fault detection network is a convolutional neural network.

[0132] In some embodiments of the present application, the preprocessing of the vehicle vibration data and track surface data comprises:

[0133] cleaning the vehicle vibration data and track surface data;

[0134] and / or, time aligning the vehicle vibration data and track surface data.

[0135] In some embodiments of the present application, the vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and track surface data comprises:

[0136] when there is inconsistency in data time intervals between multiple data in the acceleration data, resampling the acceleration data.

[0137] In some embodiments of the present application, the vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and track surface data comprises:

[0138] converting the acceleration data from an acceleration sensor coordinate system to a vehicle coordinate system.

[0139] In some embodiments of the present application, the vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and track surface data comprises:

[0140] removing acceleration data caused by acceleration or deceleration of the vehicle to produce acceleration changes in the horizontal direction.

[0141] In some embodiments of the present application, the vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and track surface data comprises:

[0142] removing acceleration data with a vehicle vibration frequency less than a frequency threshold.

[0143] In some embodiments of the present application, the vehicle vibration data comprises acceleration data, the track surface data comprises video frames, and the preprocessing of the vehicle vibration data and the track surface data comprises:

[0144] The acceleration data is divided into a plurality of acceleration data segments, and a corresponding relationship between the plurality of acceleration data segments and the video frames, the position information, and the time stamp is established.

[0145] In some embodiments of the present application, the preprocessing of the vehicle vibration data and the track surface data comprises:

[0146] Acceleration data segments and corresponding video frames that do not exist faults are removed.

[0147] In some embodiments of the present application, the removing of the acceleration data segments and the corresponding video frames that do not exist faults comprises:

[0148] A root mean square sum of three-axis acceleration is determined, and acceleration data segments and corresponding video frames with a root mean square sum less than an acceleration threshold value are removed.

[0149] In an embodiment of the present application, by obtaining vehicle vibration data and track surface data collected during operation of a rail transit train, the vehicle vibration data and the track surface data are preprocessed to obtain a data set, a first part of data in the data set is used for track surface fault detection, and a second part of data in the data set is used for bearing fault detection and / or vehicle imbalance fault detection, the first part of data and the second part of data are different, the joint detection of track surface faults, bearing faults, and vehicle imbalance faults is realized, and the efficiency of fault detection is improved.

[0150] Some embodiments of the present application further provide an electronic device, which can include a processor, a memory, and a computer program stored on the memory and capable of running on the processor, and the computer program is executed by the processor to implement the method as above.

[0151] Some embodiments of the present application further provide a computer readable storage medium, and the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method as above.

[0152] For the device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the related parts refer to the part of the method embodiments.

[0153] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of countries and regions, and provide corresponding operation portal for user to choose authorization or refusal.

[0154] Each of the embodiments in the present specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and the same and similar parts between the embodiments can be referred to each other.

[0155] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, device or computer program product. Therefore, the embodiments of the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present application can be in 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.

[0156] The embodiments of the present application are described with reference to flowcharts and / or block diagrams according to the method, terminal device (system) and computer program product of the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of the flows and / or blocks in the flowchart and / or block diagram can be realized by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing terminal device to produce a machine, so that the instructions executed by the computer or other programmable data processing terminal device produce a device for implementing the functions specified in the flowchart and / or block diagram. Figure 1 The functions specified in one flow or multiple flows and / or blocks Figure 1 The device for implementing the functions specified in one block or multiple blocks.

[0157] These computer program instructions can also be stored in a computer readable storage medium that can guide the computer or other programmable data processing terminal device to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction devices, which implement the functions specified in the flowchart and / or block diagram. Figure 1 The functions specified in one flow or multiple flows and / or blocks Figure 1 The device for implementing the functions specified in one block or multiple blocks.

[0158] These computer program instructions can also be loaded into a computer or other programmable data processing terminal device, so that a series of operational steps are performed on the computer or other programmable terminal device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide a process for implementing the functions specified in the flowchart Figure 1 one flow or multiple flows and / or the functions specified in the block Figure 1 one block or multiple blocks.

[0159] Although the preferred embodiments of the present application have been described, those skilled in the art who understand the basic inventive concept can make additional changes and modifications to the embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the embodiments of the present application.

[0160] Finally, it should also be noted that, in this document, relational terms such as first and second and the like can only be used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or terminal device. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or terminal device including the above element.

[0161] The above provides a detailed description of the provided rail transit train-based fault detection method, device, equipment and medium. The principles and implementation manners of the present application are described by applying specific examples. The above embodiment descriptions are only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manners and application ranges will be changed; in conclusion, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method for detecting a failure of a rail transit train, characterized in that, The method comprises: obtaining vehicle vibration data and track surface data collected during the operation of a rail transit train; preprocessing the vehicle vibration data and the track surface data to obtain a data set; the data set comprises a first part of data, and the first part of data comprises a plurality of acceleration data segments and corresponding video frames; determining the light and shade degree of the video frames in the data set; in a case where the light and shade degree is greater than a preset degree, using the acceleration data segments and the corresponding video frames in the first part of data to perform track surface fault detection; in a case where the light and shade degree is less than or equal to the preset degree, using the acceleration data segments in the first part of data to perform track surface fault detection; using a second part of data in the data set to perform bearing fault detection and / or vehicle imbalance fault detection; wherein the first part of data and the second part of data are different.

2. The method of claim 1, wherein, The method further comprises: fusing track surface fault detection results of a plurality of rail transit trains.

3. The method of claim 2, wherein, The fusing of the track surface fault detection results of the plurality of rail transit trains comprises: clustering defect points in the track surface fault detection results of the plurality of rail transit trains to obtain clustering clusters, and determining a center point of the clustering clusters as a defect point.

4. The method of claim 2, wherein, The method further comprises: labeling in a preset map according to the fused track surface fault detection results.

5. The method of claim 2, wherein, The method further comprises: updating a track surface fault database according to the fused track surface fault detection results.

6. The method of claim 1, wherein, The track surface fault detection is performed by using a track surface fault detection network, and the bearing fault detection and / or the vehicle imbalance fault detection are performed by using a bearing fault and vehicle imbalance fault detection network.

7. The method of claim 6, wherein, The track surface fault detection network and / or the bearing fault and vehicle imbalance fault detection network is a convolutional neural network.

8. The method of claim 1, wherein, The preprocessing of the vehicle vibration data and the track surface data comprises: cleaning the vehicle vibration data and the track surface data; and / or time aligning the vehicle vibration data and the track surface data.

9. The method of claim 1, wherein, The vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and the track surface data comprises: resampling the acceleration data in a case where data time intervals between a plurality of data in the acceleration data are inconsistent.

10. The method of claim 1, wherein, The vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and the track surface data comprises: converting the acceleration data from an acceleration sensor coordinate system to a vehicle coordinate system.

11. The method of claim 1, wherein, The vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and the track surface data comprises: removing acceleration data caused by acceleration or deceleration of the vehicle to generate acceleration changes in the horizontal direction.

12. The method of claim 1, wherein, The vehicle vibration data comprises acceleration data, and the preprocessing of the vehicle vibration data and the track surface data comprises: removing acceleration data with a vehicle vibration frequency less than a frequency threshold.

13. The method of claim 1, wherein, The vehicle vibration data comprises acceleration data, and the track surface data comprises video frames, and the preprocessing of the vehicle vibration data and the track surface data comprises: The acceleration data is divided into a plurality of acceleration data segments, and a corresponding relationship between the plurality of acceleration data segments and video frames, position information, and timestamps is established.

14. The method of claim 13, wherein, The preprocessing of the vehicle vibration data and the track surface data comprises: Removing acceleration data segments and corresponding video frames that do not have faults.

15. The method of claim 14, wherein, The removing of acceleration data segments and corresponding video frames that do not have faults comprises: Determining a root mean square sum of three-axis acceleration, and removing acceleration data segments and corresponding video frames whose root mean square sum is less than an acceleration threshold.

16. A rail transit train based fault detection apparatus, characterized by, For: Obtaining vehicle vibration data and track surface data collected during operation of a rail transit train; Preprocessing the vehicle vibration data and the track surface data to obtain a data set; the data set comprises a first part of data, and the first part of data comprises a plurality of acceleration data segments and corresponding video frames; Determining a light and shade degree of light of video frames in the data set; In a case where the light and shade degree of light is greater than a preset degree, using acceleration data segments and corresponding video frames in the first part of data to perform track surface fault detection; In a case where the light and shade degree of light is less than or equal to the preset degree, using acceleration data segments in the first part of data to perform track surface fault detection; Using a second part of data in the data set to perform bearing fault detection and / or vehicle imbalance fault detection; the first part of data and the second part of data are different.

17. An electronic device, comprising: A computer program is stored on the computer readable storage medium and is executable by a processor to implement the method of any one of claims 1 to 15.

18. A computer-readable storage medium, characterized in that, A computer program is stored on the computer readable storage medium and is executable by a processor to implement the method of any one of claims 1 to 15.

Images