Train idling and slipping intelligent prediction method and device, storage medium

Abstract

The application provides a train idling and slipping intelligent prediction method, device and storage medium. The method determines line information of a track where a train is located, and obtains environment perception information of the train. According to the environment perception information, a running label, environment characteristics, working condition characteristics and track surface attachment type are determined. Based on the line information, the running label, the environment characteristics, the working condition characteristics, the track surface attachment type and a pre-trained idling and slipping detection model, a prediction result is obtained. The method provided by the application obtains the prediction result based on the line information, the running label, the environment characteristics, the working condition characteristics, the track surface attachment type and the pre-trained idling and slipping detection model. Since the line information, the running label, the environment characteristics, the working condition characteristics and the track surface attachment type can effectively identify characteristic parameters of adhesion reduction, the method provided by the application can predict the probability of adhesion reduction, and intelligent prediction of idling and slipping is realized.

Description

Technical Field

[0001] This application relates to the field of rail transit technology, and in particular to a method, device, and storage medium for intelligent prediction of train slippage. Background Technology

[0002] A healthy and proper frictional relationship between the wheel and rail is crucial for train starting, braking, and wheel-rail dynamics. Under normal circumstances, the adhesion coefficient between the wheel and rail is sufficient to provide the necessary traction and braking force. However, in actual operation, the accumulation of fallen leaves, oil, and rain / snow on the rails forms a third medium layer, significantly reducing the adhesion coefficient. This results in insufficient traction and braking force, weak train acceleration, excessive braking distance, and wheel slippage—the so-called low wheel-rail adhesion problem. A certain degree of low adhesion can affect normal train operation and, in severe cases, may lead to safety accidents. Furthermore, low wheel-rail adhesion causes wear and scratches on the wheels and rails, increasing system maintenance costs and reducing train lifespan.

[0003] To prevent wheel slippage and wheel idling, common detection methods include speed difference judgment, creep rate judgment, and acceleration / deceleration and change judgment. However, these methods often rely on wheel-rail behavior after wheel slippage has occurred, failing to predict slippage in a timely and accurate manner. Therefore, there is an urgent need to propose an intelligent slippage prediction method. Summary of the Invention

[0004] To address one of the aforementioned technical deficiencies, this application provides a method, device, and storage medium for intelligent prediction of train slippage.

[0005] The first aspect of this application provides a method for intelligent prediction of train slippage, the method comprising:

[0006] Determine the track information of the train and obtain the train's environmental perception information;

[0007] Based on environmental perception information, determine the operation tag, environmental characteristics, operating condition characteristics, and type of rail surface attachments;

[0008] Based on track information, operation tags, environmental characteristics, operating condition characteristics, types of rail surface attachments, and a pre-trained idling and slippage detection model, prediction results are obtained.

[0009] Optionally, determine the track information of the train, including:

[0010] Laser beams are periodically emitted using a lidar sensor, and the emission time is recorded.

[0011] The laser radar sensor receives the reflected beam of the laser beam and records the reception time.

[0012] Calculate the train's position information at each time point based on the launch and reception times;

[0013] The location information at each time point is matched with the electronic map of the line to determine the track on which the train is located.

[0014] Based on the track where the train is located, determine the track information corresponding to the train's current position;

[0015] Among them, the track information is related to wheel-rail adhesion.

[0016] Optionally, the route information may include one or more of the following: gradient, inclination angle, and turning radius.

[0017] Optionally, the environmental perception information includes: axle speed, vehicle speed, temperature, humidity, track surface image, passenger-carrying conditions, and basic operating conditions.

[0018] Acquire environmental perception information of the train, including:

[0019] The axle speed of the train is obtained through an axle speed sensor;

[0020] Train speed is obtained using millimeter-wave radar;

[0021] The temperature and humidity of the train's environment are obtained through environmental sensors;

[0022] Images of the train's track surface are obtained using a high-speed camera;

[0023] The onboard signaling system obtains the train's passenger-carrying and basic operating conditions.

[0024] Optionally, based on environmental perception information, the operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment types are determined, including:

[0025] The running label is determined based on the difference between vehicle speed and axle speed;

[0026] Temperature and humidity are defined as environmental characteristics;

[0027] The passenger-carrying operating condition and the basic operating condition are defined as the operating condition characteristics;

[0028] Image recognition technology is used to identify the type of rail surface attachments in the rail surface images.

[0029] Optionally, the running label is slippage, normal, or idling;

[0030] The operating label is determined based on the difference between vehicle speed and axle speed, including:

[0031] Calculate the difference between vehicle speed and axle speed;

[0032] If the difference is not less than the maximum threshold, the running label is slippage;

[0033] If the difference is not greater than the minimum threshold, the running label is considered normal.

[0034] If the difference is less than the maximum threshold and greater than the minimum threshold, the running label is "idle".

[0035] Optionally, the operating characteristics include passenger-carrying operating conditions and basic operating conditions;

[0036] Based on track information, operational tags, environmental characteristics, operating condition characteristics, types of rail surface attachments, and a pre-trained idling and slippage detection model, prediction results are obtained, including:

[0037] Encode the passenger-carrying operation with distance differences;

[0038] The basic operating conditions and rail surface attachment types are one-hot encoded;

[0039] The encoded data, line information, running labels, and environmental features are input into a pre-trained idling and slippage detection model to obtain prediction results.

[0040] The idling and slippage detection model is an XGBoost model. During training, the idling and slippage detection model uses early stopping to prevent overfitting and cross-validation to adjust the hyperparameters.

[0041] Optionally, the prediction results may include information on the sections where idling and / or slippage occurred;

[0042] After obtaining the prediction results, it also includes:

[0043] It shares section information with the following vehicle so that the following vehicle can adjust its vehicle control strategy.

[0044] A second aspect of this application provides an electronic device, comprising:

[0045] Memory;

[0046] Processor; and

[0047] Computer programs;

[0048] The computer program is stored in the memory and configured to be executed by the processor to implement the method described in the first aspect above.

[0049] A third aspect of this application provides a computer-readable storage medium having a computer program stored thereon; the computer program is executed by a processor to implement the method described in the first aspect above.

[0050] This application provides a method, device, and storage medium for intelligent prediction of train slippage. The method determines the track information of the train and acquires the train's environmental perception information. Based on the environmental perception information, it determines the operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment type. Based on the track information, operating tag, environmental characteristics, operating condition characteristics, rail surface attachment type, and a pre-trained slippage detection model, it obtains the prediction result. The method provided in this application, based on track information, operating tag, environmental characteristics, operating condition characteristics, rail surface attachment type, and a pre-trained slippage detection model, obtains the prediction result. Since track information, operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment type can effectively identify characteristic parameters that cause adhesion descent, the method provided in this application can predict the probability of adhesion descent, thus achieving intelligent prediction of slippage. Attached Figure Description

[0051] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0052] Figure 1 A flowchart illustrating a train slippage intelligent prediction method provided in this application embodiment;

[0053] Figure 2 This is a schematic diagram of environmental perception information acquisition provided in an embodiment of this application;

[0054] Figure 3 This is a schematic diagram illustrating another method for acquiring environmental perception information, provided in an embodiment of this application.

[0055] Figure 4 A schematic diagram illustrating the relationship between speed difference and the probability of train slippage provided in this application embodiment;

[0056] Figure 5 This is a schematic diagram of the track surface attachment type identification model provided in the embodiments of this application;

[0057] Figure 6 A schematic diagram illustrating one-hot encoding of the basic working conditions and rail surface attachment types provided in the embodiments of this application;

[0058] Figure 7 This is a schematic diagram of the implementation module of the intelligent prediction method for train slippage provided in the embodiments of this application. Detailed Implementation

[0059] To make the technical solutions and advantages of the embodiments of this application clearer, the exemplary embodiments of this application will be described in further detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0060] In the process of developing this application, the inventors discovered that commonly used detection methods to prevent wheel slippage and idling include speed difference judgment method, creep rate judgment method, and acceleration / deceleration and change judgment method. However, these methods often make judgments based on wheel-rail performance after wheel slippage and idling have occurred, and cannot predict the occurrence of slippage and idling in a timely and accurate manner.

[0061] To address the aforementioned problems, this application provides a method, device, and storage medium for intelligent prediction of train slippage. The method determines the track information of the train and acquires the train's environmental perception information. Based on the environmental perception information, it determines the operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment type. Based on the track information, operating tag, environmental characteristics, operating condition characteristics, rail surface attachment type, and a pre-trained slippage detection model, it obtains a prediction result. The method provided in this application, based on track information, operating tag, environmental characteristics, operating condition characteristics, rail surface attachment type, and a pre-trained slippage detection model, obtains a prediction result. Since track information, operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment type can effectively identify characteristic parameters that cause adhesion descent, the method provided in this application can predict the probability of adhesion descent, thus achieving intelligent prediction of slippage.

[0062] See Figure 1 This embodiment provides a method for intelligent prediction of train slippage, and the execution process of the method is as follows:

[0063] 101. Determine the track information of the train and obtain the train's environmental perception information.

[0064] Step 101 includes two processes: one is to determine the track information of the train, and the other is to acquire the train's environmental perception information. The two processes can be executed sequentially or in parallel. If executed sequentially, this embodiment does not restrict the order of the two processes.

[0065] • The process of determining the track information of the train

[0066] 1.1 The position information of the train at each time is calculated using a lidar sensor.

[0067] Step 1.1 can be implemented based on a single lidar sensor or multiple lidar sensors.

[0068] For example, if it is a lidar sensor, then step 1.1 is implemented as follows:

[0069] 1. A laser beam is periodically emitted using a lidar sensor, and the emission time is recorded. After each laser beam is emitted, the reflected beam is received by the lidar sensor, and the reception time is recorded.

[0070] After the laser beam is emitted, the emission time is recorded. The laser beam travels at the speed of light and is reflected back when it intersects with an object. When the laser beam is reflected back by the object, the lidar sensor records the reception time of the reflected beam. Environmental information can be obtained in real time through the reflected beam.

[0071] 2. Calculate the train's location information at each time point based on the launch and reception times.

[0072] For example, for a laser beam, its propagation time can be obtained from its emission and reception times. The distance traveled by the laser beam during its round trip can then be calculated, yielding its position information, which is the position information of the train at the time the laser beam was emitted. By periodically transmitting laser beams, the position information of the train at each laser beam emission time can be obtained, thus providing the train's position information at each specific moment.

[0073] If there are multiple radar sensors, then step 1.1 is implemented as follows:

[0074] 1. Laser beams are periodically emitted using multiple lidar sensors, and the emission time is recorded. Each time a laser beam is emitted, the reflected beam is received by the corresponding lidar sensor, and the reception time is recorded.

[0075] After the laser beam is emitted, the emission time is recorded. The laser beam travels at the speed of light and is reflected back when it intersects with an object. When the laser beam is reflected back by the object, the lidar sensor records the reception time of the reflected beam. Environmental information can be obtained in real time through the reflected beam.

[0076] 2. Calculate the train's location information at each time point based on the launch and reception times.

[0077] For example, for a laser beam, its propagation time can be obtained from its emission and reception times. This allows calculation of the distance traveled during the round trip, thus determining the train's position information at the moment the laser beam was emitted. By periodically transmitting laser beams, the train's position information at each emission moment can be obtained, leading to its position information at each specific time. Multiple lidar sensors can operate simultaneously, emitting laser beams at the same time, resulting in multiple positions at the same moment. By combining the different periods and data from different lidar sensors, all position information can be compiled into a three-dimensional point cloud. The coordinates of each point represent the train's position in the vehicle coordinate system at the moment the laser beam was emitted. This point cloud data, representing the position information at each moment, includes detailed information about the vehicle's surrounding environment.

[0078] 1.2 Based on the location information at each time point and the electronic map of the line, determine the line information of the track where the train is located.

[0079] After obtaining the location information at each time, the line information of the train's track can be determined based on the location information at each time and the electronic map of the line.

[0080] For example, the location information at each moment is matched with the electronic map of the route to determine the track the train is on. Based on the track the train is on, the corresponding line information for the train's current location is determined.

[0081] The track information is related to wheel-rail adhesion. For example, track information includes, but is not limited to, one or more of the following: gradient, inclination angle, and turning radius.

[0082] Electronic maps include the accurate location and geometry of stations, tracks, tunnels, and other underground structures.

[0083] Taking point cloud data as an example, the real-time point cloud data is matched with the electronic map in step 1.2. By comparing the point cloud data with the electronic map, the position of the vehicle relative to the train track (i.e., the track line on which the train is located) is determined. Based on the track line on which the train is located, the gradient, inclination angle, turning radius, and other track information related to wheel-rail adhesion, such as the current position of the train, are determined.

[0084] • Regarding the process of acquiring environmental perception information for trains

[0085] Environmental perception information can be obtained through onboard sensors and trackside sensors.

[0086] The environmental perception information includes at least: axle speed, vehicle speed, temperature, humidity, track surface image, passenger-carrying conditions, and basic operating conditions.

[0087] Specifically, such as Figure 2 As shown, the process of acquiring environmental perception information of the train is as follows:

[0088] 2.1 Obtain the train's axle speed using an axle speed sensor.

[0089] For example, an onboard axle speed sensor records axle speed information during train operation. This axle speed information recorded by the axle speed sensor is then used as the train's axle speed.

[0090] 2.2 Train speed is obtained using millimeter-wave radar.

[0091] For example, vehicle-mounted millimeter-wave radar records the speed information of a train during its journey, and this speed information recorded by the millimeter-wave radar is used as the train's speed.

[0092] 2.3 The temperature and humidity of the train's environment are obtained through environmental sensors.

[0093] For example, trackside environmental sensors record temperature and humidity information of the track surface environment. The temperature and humidity information recorded by the environmental sensors is obtained and used as the temperature and humidity of the environment in which the train is located.

[0094] 2.4 Acquire images of the train's track surface using a high-speed camera.

[0095] For example, a high-speed camera on a vehicle captures images of the track surface conditions, and these images are used as the track surface images of the train.

[0096] 2.5 Obtain the passenger-carrying and basic operating conditions of the train through the onboard signaling system.

[0097] For example, the onboard signaling system records the passenger-carrying operating conditions of the train (such as AW0, AW1, AW2, AW3) and the basic operating conditions of the train (such as traction, inertia, braking). The passenger-carrying operating conditions and basic operating conditions recorded by the onboard signaling system are obtained and used as the passenger-carrying operating conditions and basic operating conditions of the train.

[0098] It should be noted that the order of steps 2.1 to 2.5 above is only an example. In actual application, the execution order can be determined according to the situation. This embodiment does not limit the execution order of steps 2.1 to 2.5.

[0099] In addition to the seven environmental perception information items mentioned above, the onboard LiDAR sensor obtains the train's position during the process of determining the track information. This position can also be used as environmental perception information. Therefore, the environmental perception information consists of eight items: axle speed, vehicle speed, temperature, humidity, track surface image, passenger operating conditions, basic operating conditions, and train position. Figure 3 As shown.

[0100] 102. Based on environmental perception information, determine the operating label, environmental characteristics, operating condition characteristics, and type of rail surface attachments.

[0101] Step 102 is the process of identifying key features in the environmental perception information. The implementation details of this process are as follows:

[0102] 102-1, Determine the running label based on the difference between vehicle speed and axle speed.

[0103] The running labels are slipping, idling, or normal.

[0104] Due to adhesion, the wheel-axle speed and the train's actual speed are never equal during train operation. During traction, the wheel-axle speed is greater than the actual train speed, and conversely, during braking, the wheel-axle speed is less than the actual train speed. The difference between wheel-axle speed and train speed increases with increasing creep rate. Therefore, comparing whether the difference between wheel-axle speed and train speed exceeds a certain set threshold can be used to determine whether the train is experiencing wheel spin or slippage.

[0105] Reference Figure 4 The relationship between the speed difference and the probability of train slippage is shown by calculating the train speed (i.e., vehicle speed, such as v). r (t) and wheel and axle speed (i.e., axle speed, i.e., v) w The difference between (t) (i.e., Δv(t) = v) r (t)-v w The system uses the speed difference (Δv(t)) to detect whether the train is skidding. If the calculated speed difference exceeds the maximum threshold Δv2′, the train is considered to have skidded; if the calculated speed difference is less than the minimum threshold Δv1′, the train is considered to have idled.

[0106] Therefore, the implementation process of step 102-1 is as follows: calculate the difference between the vehicle speed and the axle speed. If the difference is not less than the maximum threshold, the running label is "slippage". If the difference is not greater than the minimum threshold, the running label is "normal". If the difference is less than the maximum threshold but greater than the minimum threshold, the running label is "idling".

[0107] That is, run the label

[0108] 102-3, temperature and humidity are defined as environmental characteristics.

[0109] 102-4, passenger-carrying operating condition and basic operating condition are defined as operating condition characteristics.

[0110] 102-5, using image recognition technology to identify the type of rail surface attachments in the rail surface image.

[0111] The purpose of steps 102-5 is to extract features from the track surface image using image recognition technology and identify the type of the third medium layer attached to the track surface, such as: dry state, water film, oil stains, fallen leaves, etc.

[0112] Step 102-5 can be implemented using a pre-trained track surface attachment type recognition model, which is a CNN (Convolutional Neural Network) model. In step 102-5, the track surface image is input into the pre-trained track surface attachment type recognition model. This model uses image recognition technology to identify the track surface image and outputs a classification prediction. The Softmax function is used to convert the output classification prediction result into a probability distribution, and then the category with the highest probability is selected as the final determined track surface attachment type.

[0113] The training process for this track surface attachment type recognition model is as follows:

[0114] The training process for this model is as follows:

[0115] 1. Data Preparation

[0116] 1) Collect labeled image datasets, including dry track surfaces, track surfaces containing water, oil, and fallen leaves;

[0117] 2) Use annotation software to mark the rail area in the rail surface condition image;

[0118] 3) Assign the correct labels to the third medium layer deposits on the rail surface;

[0119] 4) Divide the dataset into training set and test set.

[0120] 2. Data Preprocessing

[0121] 1) Adjust the size and dimensions of the image to fit the network's input requirements;

[0122] 2) Normalize the image by scaling the pixel values ​​to between 0 and 1;

[0123] 3. Establish a model for identifying the types of rail surface attachments.

[0124] like Figure 5 As shown, create a CNN model including convolutional layers, pooling layers, fully connected layers, and activation functions; the number of neurons in the output layer should match the number of categories (e.g., 4 categories: dry, water, oil, and leaves); choose the cross-entropy loss function.

[0125] This CNN model is a track surface attachment type recognition model.

[0126] 4. Training of the track surface attachment type recognition model

[0127] 1) The training set is used to train the track surface attachment type recognition model, and the model weights are updated through the backpropagation algorithm;

[0128] 2) Adjust the weights using an optimizer (such as Adam, SGD, etc.) to minimize the loss function;

[0129] 3) Monitor the performance of the validation set through early stopping to avoid overfitting.

[0130] 5. Evaluation of the track surface attachment type identification model:

[0131] Use the test set to evaluate the model's performance and calculate metrics such as accuracy, precision, recall, and F1 score.

[0132] If the evaluation results meet the preset requirements, the training of the track surface attachment type recognition model is complete; otherwise, the above training process is repeated.

[0133] It should be noted that the order of steps 102-1 to 102-5 above is only an example. In actual application, the execution order can be determined according to the situation. This embodiment does not limit the execution order of steps 102-1 to 102-5.

[0134] 103. Based on track information, operation tags, environmental characteristics, operating condition characteristics, types of rail surface attachments, and a pre-trained idling and slippage detection model, prediction results are obtained.

[0135] The operating conditions here include passenger-carrying conditions and basic operating conditions.

[0136] In addition to being based on track information, operation tags, environmental characteristics, operating conditions, and types of rail surface attachments, step 103 can also be based on information such as axle speed and vehicle speed to increase the accuracy of the final prediction result.

[0137] The implementation process of step 103 is as follows:

[0138] 103-1 encodes the passenger-carrying operation with distance difference.

[0139] For example, the codes are 0, 1, 2, 3.

[0140] 103-2, One-hot encoding is performed on the basic working conditions and the type of rail surface attachments.

[0141] like Figure 6As shown, the dry state of rail surface attachments is coded as 001, the water state as 0010, the oil state as 0100, and the leaves state as 1000. The traction state of the basic working condition is coded as 001, the coasting state as 010, and the braking state as 100.

[0142] In addition, after performing step 103-2, missing value processing will be performed before proceeding to step 103-3.

[0143] 103-3. The encoded data, line information, running labels, and environmental features are input into a pre-trained idling and slippage detection model to obtain the prediction results.

[0144] If the running label in step 102 is used as label information, and then aligned with the line information, environmental characteristics, working condition characteristics, and rail surface attachment type, it is input together with the encoded data into the pre-trained idling and slippage detection model to obtain the prediction result.

[0145] Alternatively, the running label in step 102 can be used as label information, and then aligned with the track information, environmental characteristics, operating condition characteristics, type of rail surface attachments, axle speed, and vehicle speed. This information, along with the encoded data, can be input into a pre-trained idling and slippage detection model to obtain the prediction result.

[0146] The prediction results should include at least information on the sections where idling and / or slippage occurred.

[0147] The idling and slippage detection model is an XGBoost model. During training, the model uses early stopping to prevent overfitting and cross-validation to adjust hyperparameters.

[0148] Step 103 is implemented based on a pre-trained idling and slippage detection model, which is an XGBoost model. The training process of this idling and slippage detection model is as follows:

[0149] The XGBoost (Extreme Gradient Boosting) algorithm was chosen to train the intelligent prediction model for train slippage. XGBoost is a powerful gradient boosting algorithm commonly used for classification and regression tasks. The algorithm flow is as follows:

[0150] 1. Dataset partitioning

[0151] The training sample data is obtained. This data is also processed in step 101 to obtain line information and environmental perception information. It is also processed in step 102 to determine the operation tag, environmental characteristics, working condition characteristics and rail surface attachment type based on the environmental perception information.

[0152] At this point, the running label will be used as the label information, and then aligned with the track information, environmental characteristics, working condition characteristics, type of rail surface attachments, axle speed, vehicle speed, etc., and divided into training set and test set in a 7:3 ratio.

[0153] 2. Data Preprocessing

[0154] 1) Encode the passenger-carrying conditions (such as AW0, AW1, AW2, AW3) into codes with distance differences (such as 0, 1, 2, 3).

[0155] 2) Use one-hot encoding to process basic operating conditions and rail surface attachment types, such as... Figure 6 As shown.

[0156] 3) Handling missing values

[0157] 3. Establish a slippage detection model

[0158] The idling slippage detection model is an XGBoost model, therefore the establishment process is as follows:

[0159] 1) Import the XGBoost library and create an XGBoost classifier.

[0160] 2) Set the hyperparameters of XGBoost, such as learning rate, tree depth, number of trees, etc.

[0161] 3) Initialize the model; you can use the default parameters or adjust them based on experience.

[0162] 4. Training of the slippage detection model

[0163] 1) Train the XGBoost model using the training set.

[0164] During training, the model will be gradually optimized to minimize the loss function.

[0165] 2) Use early stopping to prevent overfitting.

[0166] That is, monitor performance on the validation set and stop training when performance no longer improves.

[0167] 5. Evaluation of the idling slippage detection model

[0168] Use a test set to evaluate the performance of the trained spin and slip detection model. For example, use various metrics to evaluate classification performance, such as accuracy, precision, recall, F1 score, etc.

[0169] 6. Parameter optimization

[0170] Use methods such as cross-validation to adjust the model's hyperparameters to improve performance.

[0171] In addition, after obtaining the prediction result by performing step 103, the segment information in the prediction result will be shared with the following vehicle so that the following vehicle can adjust its vehicle control strategy.

[0172] For example, based on inter-train communication technology, section information from the prediction results (such as information on sections where track slippage / idleage occurs) can be shared with the following train. Adjusting the train's control strategy based on the prediction results can effectively mitigate problems such as wheel and rail wear and abrasion caused by train slippage / idleage. Simultaneously, the train will also receive section information from the prediction results shared by the preceding train (such as information on sections where track slippage / idleage occurs). Adjusting the train's control strategy based on the preceding train's prediction results can also effectively mitigate problems such as wheel and rail wear and abrasion caused by train slippage / idleage.

[0173] The intelligent prediction method for train slippage provided in this embodiment employs machine learning technology. It uses a high-speed camera to capture images of the rail surface ahead of the train, identifies rail surface conditions, and detects the presence of a third medium layer such as water film, oil, or fallen leaves. Then, an axle speed sensor records axle speed data, and a speed difference threshold method is used to determine the degree of train slippage, labeling the captured rail surface images. By training a machine learning classification model, it can assess the risk of train slippage based on the captured images. This method can effectively identify characteristic parameters of adhesion descent, predict the probability of adhesion descent, and avoid slippage by adjusting control strategies, effectively reducing wear and abrasion on wheels and rails, and has broad application prospects.

[0174] The intelligent prediction method for train slippage provided in this embodiment can be executed by... Figure 7 The vehicle track environment perception module, train precise positioning module, and model training module shown are implemented.

[0175] The train precise positioning module is used to determine the track information of the train in step 101. Specifically, it uses lidar equipment to perform high-precision positioning of the train, continuously outputting the train's position information at various times, and combining this with an electronic map of the track to obtain the current track information of the train.

[0176] The track environment perception module is used to acquire the train's environmental perception information in step 101. Specifically, it uses axle speed sensors, millimeter-wave radar, environmental sensors, and high-speed cameras to acquire the train's axle speed, vehicle speed, the train's surrounding environment, and track surface images. This perception information serves as input for the subsequent slippage detection model. Additionally, the track environment perception module also executes step 102, obtaining the operating label (normal / slippage / freezing) based on the train's axle speed and vehicle speed; and simultaneously obtaining environmental features and operating condition features based on the type of large track surface attachments in the track surface image.

[0177] The model training module is used to train the slippage detection model. This training process is based on multi-dimensional feature fusion training, which incorporates track information, operating labels, environmental characteristics, operating condition characteristics, and types of rail surface attachments from the sample data. Once trained, this slippage detection model can be embedded into the train to ensure train operation safety and reduce operational losses. By integrating the intelligent slippage prediction model, the train can achieve real-time sharing of rail surface conditions.

[0178] like Figure 7 As shown, a train slippage and idling intelligent prediction model is trained using key features and slippage labels output by the train precise positioning module and the track environment perception module. This model is integrated into the onboard system, collecting model inputs through onboard sensors and signal systems, and outputting real-time train slippage and idling conditions through a machine learning classification model. In single-car operation scenarios or multi-car cooperative scenarios such as virtual train formations, the intelligent prediction model integrated with the preceding trains can effectively identify characteristic parameters of adhesion descent, predict the probability of adhesion descent, adjust control strategies, avoid slippage and idling, and effectively reduce wear and abrasion on wheels and rails.

[0179] The intelligent prediction method for train slippage provided in this embodiment can effectively identify the characteristic parameters of adhesion descent, predict the probability of adhesion descent, and avoid slippage by adjusting the control strategy, thereby effectively reducing wear and scratches on wheels and rails.

[0180] This embodiment provides an intelligent prediction method for train slippage. It determines the track information of the train and acquires the train's environmental perception information. Based on the environmental perception information, it determines the operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment type. Based on the track information, operating tag, environmental characteristics, operating condition characteristics, rail surface attachment type, and a pre-trained slippage detection model, it obtains the prediction result. The method provided in this embodiment obtains the prediction result based on track information, operating tag, environmental characteristics, operating condition characteristics, rail surface attachment type, and a pre-trained slippage detection model. Since track information, operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment type can effectively identify characteristic parameters that cause adhesion descent, the method provided in this embodiment can predict the probability of adhesion descent, thus achieving intelligent prediction of slippage.

[0181] Based on the same inventive concept of intelligent prediction method for train slippage, this embodiment provides an electronic device, which includes: a memory, a processor, and a computer program.

[0182] The computer program is stored in memory and configured to be executed by the processor to implement the above-mentioned intelligent prediction method for train slippage.

[0183] Specifically,

[0184] Determine the track information of the train and obtain the train's environmental perception information.

[0185] Based on environmental perception information, determine the operating tags, environmental characteristics, operating condition characteristics, and types of rail surface attachments.

[0186] Based on track information, operation tags, environmental characteristics, operating condition characteristics, types of rail surface attachments, and a pre-trained idling and slippage detection model, prediction results are obtained.

[0187] Optionally, determine the track information of the train, including:

[0188] The laser beam is periodically emitted by a lidar sensor, and the emission time is recorded.

[0189] The laser radar sensor receives the reflected beam of the laser beam and records the reception time.

[0190] Calculate the train's location information at each time point based on the launch and reception times.

[0191] The location information at each time point is matched with the electronic map of the line to determine the track on which the train is located.

[0192] Based on the track where the train is located, determine the track information corresponding to the train's current position.

[0193] Among them, the track information is related to wheel-rail adhesion.

[0194] Optionally, the route information may include one or more of the following: gradient, inclination angle, and turning radius.

[0195] Optionally, environmental perception information includes: axle speed, vehicle speed, temperature, humidity, track surface image, passenger-carrying conditions, and basic operating conditions.

[0196] Acquire environmental perception information of the train, including:

[0197] The axle speed of the train is obtained through an axle speed sensor.

[0198] Train speed is obtained using millimeter-wave radar.

[0199] The temperature and humidity of the train's environment are obtained through environmental sensors.

[0200] Images of the train's track surface are obtained using a high-speed camera.

[0201] The onboard signaling system obtains the train's passenger-carrying and basic operating conditions.

[0202] Optionally, based on environmental perception information, the operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment types are determined, including:

[0203] The running label is determined based on the difference between vehicle speed and axle speed.

[0204] Temperature and humidity are defined as environmental characteristics.

[0205] Passenger-carrying operating conditions and basic operating conditions are defined as operating condition characteristics.

[0206] Image recognition technology is used to identify the type of rail surface attachments in the rail surface images.

[0207] Optionally, the running label can be set to slip, normal, or idling.

[0208] The operating label is determined based on the difference between vehicle speed and axle speed, including:

[0209] Calculate the difference between vehicle speed and axle speed.

[0210] If the difference is not less than the maximum threshold, the running label is "slippery".

[0211] If the difference is not greater than the minimum threshold, the running label is considered normal.

[0212] If the difference is less than the maximum threshold and greater than the minimum threshold, the running label is "idle".

[0213] Optionally, the operating characteristics include passenger-carrying operating conditions and basic operating conditions.

[0214] Based on track information, operational tags, environmental characteristics, operating condition characteristics, types of rail surface attachments, and a pre-trained idling and slippage detection model, prediction results are obtained, including:

[0215] The passenger-carrying operation is encoded with distance differences.

[0216] The basic operating conditions and rail surface attachment types are one-hot encoded.

[0217] The encoded data, line information, running labels, and environmental features are input into a pre-trained idling and slippage detection model to obtain prediction results.

[0218] The idling and slippage detection model is an XGBoost model. During training, the idling and slippage detection model uses early stopping to prevent overfitting and cross-validation to adjust the hyperparameters.

[0219] Optionally, the prediction results may include information on sections where idling and / or slippage occurred.

[0220] After obtaining the prediction results, it also includes:

[0221] It shares section information with the following vehicle so that the following vehicle can adjust its vehicle control strategy.

[0222] The electronic device provided in this embodiment has a computer program executed by a processor to obtain prediction results based on track information, running tags, environmental characteristics, operating condition characteristics, rail surface attachment types, and a pre-trained idling slip detection model. Since track information, running tags, environmental characteristics, operating condition characteristics, and rail surface attachment types can effectively identify characteristic parameters that cause adhesion descent, the electronic device provided in this embodiment can predict the probability of adhesion descent and achieve intelligent prediction of idling slip.

[0223] Based on the same inventive concept as the intelligent prediction method for train slippage, this embodiment provides a computer-readable storage medium on which a computer program is stored. The computer program is executed by a processor to implement the aforementioned intelligent prediction method for train slippage.

[0224] Specifically,

[0225] Determine the track information of the train and obtain the train's environmental perception information.

[0226] Based on environmental perception information, determine the operating tags, environmental characteristics, operating condition characteristics, and types of rail surface attachments.

[0227] Based on track information, operation tags, environmental characteristics, operating condition characteristics, types of rail surface attachments, and a pre-trained idling and slippage detection model, prediction results are obtained.

[0228] Optionally, determine the track information of the train, including:

[0229] The laser beam is periodically emitted by a lidar sensor, and the emission time is recorded.

[0230] The laser radar sensor receives the reflected beam of the laser beam and records the reception time.

[0231] Calculate the train's location information at each time point based on the launch and reception times.

[0232] The location information at each time point is matched with the electronic map of the line to determine the track on which the train is located.

[0233] Based on the track where the train is located, determine the track information corresponding to the train's current position.

[0234] Among them, the track information is related to wheel-rail adhesion.

[0235] Optionally, the route information may include one or more of the following: gradient, inclination angle, and turning radius.

[0236] Optionally, environmental perception information includes: axle speed, vehicle speed, temperature, humidity, track surface image, passenger-carrying conditions, and basic operating conditions.

[0237] Acquire environmental perception information of the train, including:

[0238] The axle speed of the train is obtained through an axle speed sensor.

[0239] Train speed is obtained using millimeter-wave radar.

[0240] The temperature and humidity of the train's environment are obtained through environmental sensors.

[0241] Images of the train's track surface are obtained using a high-speed camera.

[0242] The onboard signaling system obtains the train's passenger-carrying and basic operating conditions.

[0243] Optionally, based on environmental perception information, the operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment types are determined, including:

[0244] The running label is determined based on the difference between vehicle speed and axle speed.

[0245] Temperature and humidity are defined as environmental characteristics.

[0246] Passenger-carrying operating conditions and basic operating conditions are defined as operating condition characteristics.

[0247] Image recognition technology is used to identify the type of rail surface attachments in the rail surface images.

[0248] Optionally, the running label can be set to slip, normal, or idling.

[0249] The operating label is determined based on the difference between vehicle speed and axle speed, including:

[0250] Calculate the difference between vehicle speed and axle speed.

[0251] If the difference is not less than the maximum threshold, the running label is "slippery".

[0252] If the difference is not greater than the minimum threshold, the running label is considered normal.

[0253] If the difference is less than the maximum threshold and greater than the minimum threshold, the running label is "idle".

[0254] Optionally, the operating characteristics include passenger-carrying operating conditions and basic operating conditions.

[0255] Based on track information, operational tags, environmental characteristics, operating condition characteristics, types of rail surface attachments, and a pre-trained idling and slippage detection model, prediction results are obtained, including:

[0256] The passenger-carrying operation is encoded with distance differences.

[0257] The basic operating conditions and rail surface attachment types are one-hot encoded.

[0258] The encoded data, line information, running labels, and environmental features are input into a pre-trained idling and slippage detection model to obtain prediction results.

[0259] The idling and slippage detection model is an XGBoost model. During training, the idling and slippage detection model uses early stopping to prevent overfitting and cross-validation to adjust the hyperparameters.

[0260] Optionally, the prediction results may include information on sections where idling and / or slippage occurred.

[0261] After obtaining the prediction results, it also includes:

[0262] It shares section information with the following vehicle so that the following vehicle can adjust its vehicle control strategy.

[0263] The computer-readable storage medium provided in this embodiment has a computer program thereon that is executed by a processor to obtain prediction results based on track information, running tags, environmental characteristics, operating condition characteristics, rail surface attachment types, and a pre-trained idling slip detection model. Since track information, running tags, environmental characteristics, operating condition characteristics, and rail surface attachment types can effectively identify characteristic parameters that cause adhesion descent, the computer-readable storage medium provided in this embodiment can predict the probability of adhesion descent, thus achieving intelligent prediction of idling slip.

[0264] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0265] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0266] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0267] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0268] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0269] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for intelligent prediction of train slippage, characterized in that, The method includes: Determine the track information of the train and obtain the environmental perception information of the train; Based on the environmental perception information, the operating tag, environmental characteristics, operating condition characteristics, and type of rail surface attachments are determined; Based on the line information, the operation tag, the environmental characteristics, the operating condition characteristics, the type of rail surface attachments, and the pre-trained idling and slippage detection model, the prediction result is obtained; The environmental perception information includes: axle speed, vehicle speed, temperature, humidity, track surface image, passenger-carrying conditions, and basic operating conditions. The acquisition of the train's environmental perception information includes: The axle speed of the train is obtained through an axle speed sensor; The train's speed is obtained using millimeter-wave radar; The temperature and humidity of the environment in which the train is located are obtained through environmental sensors; The train's track surface image was acquired using a high-speed camera; The passenger-carrying and basic operating conditions of the train are obtained through the onboard signaling system; The step of determining the operating tag, environmental characteristics, operating condition characteristics, and rail surface attachment type based on the environmental perception information includes: The running label is determined based on the difference between the vehicle speed and the axle speed; The temperature and humidity are defined as environmental characteristics; The passenger-carrying operating condition and the basic operating condition are defined as operating condition characteristics; The type of rail surface attachment is determined by identifying the rail surface image using image recognition technology.

2. The method according to claim 1, characterized in that, The determination of the track information for the train includes: Laser beams are periodically emitted using a lidar sensor, and the emission time is recorded. The reflected beam of the laser beam is received by a lidar sensor, and the reception time is recorded. Calculate the train's position information at each time point based on the transmission time and the reception time; The location information at each time point is matched with the electronic map of the line to determine the track on which the train is located. Based on the track where the train is located, determine the track information corresponding to the train's current position; The track information is related to wheel-rail adhesion.

3. The method according to claim 2, characterized in that, The route information includes one or more of the following: gradient, inclination angle, and turning radius.

4. The method according to claim 1, characterized in that, The running label indicates slippage, normal operation, or idling. The step of determining the running label based on the difference between the vehicle speed and the axle speed includes: Calculate the difference between the vehicle speed and the axle speed; If the difference is not less than the maximum threshold, then the running label is slippage; If the difference is not greater than the minimum threshold, the running label is considered normal. If the difference is less than the maximum threshold and greater than the minimum threshold, then the running label is "idle".

5. The method according to claim 1, characterized in that, The operating conditions include passenger-carrying operating conditions and basic operating conditions; The prediction results obtained based on the track information, the operation tag, the environmental characteristics, the operating condition characteristics, the type of rail surface attachments, and the pre-trained idling and slippage detection model include: Encode the passenger-carrying operation with distance differences; The basic operating conditions and rail surface attachment types are one-hot encoded; The encoded data, the line information, the running label, and the environmental features are input into a pre-trained idling slippage detection model to obtain the prediction result. The idling slip detection model is an XGBoost model, and during training, the idling slip detection model uses early stopping to prevent overfitting and cross-validation to adjust hyperparameters.

6. The method according to claim 1, characterized in that, The prediction results include information on the sections where idling and / or slippage occurred; After obtaining the prediction result, the process also includes: The vehicle shares the section information with the following vehicle so that the following vehicle can adjust its vehicle control strategy.

7. An electronic device, characterized in that, include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and configured to be executed by the processor to implement the method as described in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, It stores a computer program thereon; the computer program is executed by a processor to implement the method as described in any one of claims 1-6.

Images