Automatic control method for tooth entering operation of driving gear of rack rail train

Abstract

The invention relates to an automatic control method for tooth entering operation of a driving gear of a rack rail train. The method comprises the following steps: firstly, acquiring historical driving gear tooth-entering vibration signals under different working conditions, and preprocessing the historical driving gear tooth-entering vibration signals; then, a tooth contact state prediction model is built based on a time convolutional network and a time attention mechanism, and training is carried out; then, collecting a real-time driving gear entering vibration signal, and inputting the signal into the trained entering gear contact state prediction model to obtain an entering gear contact state index; and finally, when the input tooth contact state index exceeds a preset threshold value, outputting an input and output tooth start signal instruction and executing electric power cut-off operation, and when the input tooth contact state index does not exceed the preset threshold value, outputting an input and output tooth completion signal instruction and switching on electric power output. By means of the method, the power control strategy can be optimized, the driving gear tooth entering process is more stable, impact is reduced, tooth entering smoothness is improved, the operation stability of a rack rail train is improved, and the service life of a system is prolonged.

Description

Technical Field

[0001] The present application relates to the technical field of rail transit, and particularly to an automatic control method for the tooth engagement operation of a gear-rack train drive gear. Background Art

[0002] A gear-rack railway is a rail transit system designed specifically for steep sections. Its core feature is to add drive gears to the train axles and configure rack tracks in the track structure to provide additional traction and braking force, enabling the train to operate stably on lines with relatively large gradients. During the operation of a gear-rack train, the meshing quality between the drive gear and the rack track directly affects the running smoothness of the train and the service life of the gear-rack system. However, during the tooth engagement process of the train, the drive gear and the tooth engagement device may experience the phenomenon of tooth jamming, that is, abnormal impacts occur at the meshing part between the gear tooth tip and the rack, resulting in local overload. Such impacts not only reduce the riding comfort of passengers but may also exacerbate the wear of the gear and the rack, and even cause fatigue fracture of the tooth root, seriously affecting the safety and stability of the gear-rack system.

[0003] In the prior art, the tooth engagement control of gear-rack railways mainly relies on mechanical design optimization or empirical adjustment, but lacks real-time monitoring and intelligent control means for the tooth engagement process. Since the tooth engagement process involves complex non-linear dynamic characteristics, traditional rigid control methods are difficult to accurately adjust the gear meshing state, resulting in difficulty in ensuring the smoothness of tooth engagement. Especially under different loads, speeds, and changes in track conditions, there is a large uncertainty in the tooth engagement state of the drive gear, and existing methods are difficult to effectively adapt to.

[0004] Therefore, in the related art, there is an urgent need for a method that can flexibly and accurately control the tooth engagement process of the drive gear of a gear-rack train. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide an automatic control method for the tooth engagement operation of a gear-rack train drive gear that can flexibly and accurately control the tooth engagement process.

[0006] In a first aspect, the present application provides an automatic control method for the tooth engagement operation of a gear-rack train drive gear. The method includes:

[0007] Collect historical drive gear tooth engagement vibration signals under different working conditions and perform preprocessing;

[0008] Build a tooth engagement contact state prediction model based on a time convolutional network and a time attention mechanism and train it;

[0009] Collect real-time drive gear tooth engagement vibration signals, input them into the trained tooth engagement contact state prediction model, and obtain tooth engagement contact state indicators;

[0010] When the tooth engagement contact state index exceeds a preset threshold, an instruction for the start of tooth engagement is output, and a power cut-off operation is performed. When the tooth engagement contact state index does not exceed the preset threshold, an instruction for the completion of tooth engagement is output, and power output is switched on.

[0011] Optionally, in an embodiment of the present application, the preprocessing includes:

[0012] Normalize the historical vibration signal of the driving gear during tooth engagement using a normalization method, and perform data annotation according to the corresponding tooth engagement state.

[0013] Optionally, in an embodiment of the present application, the tooth engagement contact state prediction model includes:

[0014] Extract features from the input vibration signal using multiple convolutional layers, where the convolutional layers include causal convolution and dilated convolution;

[0015] Input the extracted features into a temporal attention mechanism to identify the key time points of tooth engagement, and weight the output of the convolutional layer with the obtained attention weights;

[0016] Input the weighted temporal features into a fully connected layer to calculate the tooth engagement contact state index.<>

[0017] Optionally, in an embodiment of the present application, building and training the tooth engagement contact state prediction model based on a temporal convolutional network and a temporal attention mechanism includes:

[0018] Minimize the error between the predicted value and the true label using a mean squared error loss function.

[0019] Optionally, in an embodiment of the present application, building and training the tooth engagement contact state prediction model based on a temporal convolutional network and a temporal attention mechanism further includes:

[0020] Evaluate the fitting ability of the model using dynamic time warping error, and select the model with the minimum error for online detection.

[0021] In a second aspect, the present application also provides an automatic control device for the tooth engagement operation of a toothed-rail train driving gear. The device includes:

[0022] A data acquisition and preprocessing module, configured to acquire historical vibration signals of the driving gear during tooth engagement under different working conditions, and perform preprocessing;

[0023] A tooth engagement contact state prediction model building and training module, configured to build and train a tooth engagement contact state prediction model based on a temporal convolutional network and a temporal attention mechanism;

[0024] The incoming tooth contact state prediction module is used to collect the real-time incoming tooth vibration signal of the driving gear, input it into the trained incoming tooth contact state prediction model, and obtain the incoming tooth contact state index;

[0025] The incoming tooth operation control module is used to output an incoming tooth start signal instruction and perform a power cut-off operation when the incoming tooth contact state index exceeds a preset threshold, and output an incoming tooth completion signal instruction and turn on the power output when the incoming tooth contact state index does not exceed the preset threshold.

[0026] In a third aspect, the present application also provides a computer device. The computer device includes a memory and a processor. The memory stores a computer program, and the processor executes the steps of the methods described in the above various embodiments.

[0027] In a fourth aspect, the present application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, it implements the steps of the methods described in the above various embodiments.

[0028] For the above automatic control method for the incoming tooth operation of the driving gear of a rack and pinion train, first, collect the historical incoming tooth vibration signals of the driving gear under different working conditions and perform preprocessing; then, build an incoming tooth contact state prediction model based on a temporal convolutional network and a temporal attention mechanism and train it; then, collect the real-time incoming tooth vibration signal of the driving gear, input it into the trained incoming tooth contact state prediction model, and obtain the incoming tooth contact state index; finally, when the incoming tooth contact state index exceeds a preset threshold, output an incoming tooth start signal instruction and perform a power cut-off operation, and when the incoming tooth contact state index does not exceed the preset threshold, output an incoming tooth completion signal instruction and turn on the power output. That is to say, by installing an acceleration sensor on the rack and pinion truck of the rack and pinion train to collect the vibration signals of the driving gear and the incoming tooth device in real time, and combining with a temporal convolutional network (TCN) for feature extraction, the key temporal features during the incoming tooth process are effectively identified. The temporal attention mechanism is introduced to focus on the critical moments before and after the incoming tooth on the basis of TCN feature extraction, improve the accuracy of incoming tooth state determination, construct an incoming tooth contact state index, and realize the precise monitoring of the incoming tooth state. Through this method, the power control strategy can be optimized, the incoming tooth process of the driving gear can be made smoother, thereby reducing the impact, improving the incoming tooth smoothness, and enhancing the running stability and system service life of the rack and pinion train. Description of the Drawings

[0029] Figure 1 It is a schematic flow chart of an automatic control method for the incoming tooth operation of a driving gear of a rack and pinion train in an embodiment;

[0030] Figure 2 It is a schematic structural diagram of an incoming tooth contact state prediction model in an embodiment;

[0031] Figure 3 Schematic diagram of the loss function curve for model training in an embodiment;

[0032] Figure 4 Schematic diagram of the process of offline model training and online detection in an embodiment;

[0033] Figure 5 Schematic diagram of the process of tooth entry state recognition and control response in an embodiment;

[0034] Figure 6 Structural block diagram of an automatic control device for the tooth entry operation of the drive gear of a rack and pinion train in an embodiment;

[0035] Figure 7 Internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0036] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0037] In one embodiment, as Figure 1 shown, an automatic control method for the tooth entry operation of the drive gear of a rack and pinion train is provided, including the following steps:

[0038] S101: Collect historical drive gear tooth entry vibration signals under different working conditions and perform preprocessing.

[0039] In the embodiment of the present application, first, acceleration sensors are installed on each rack and pinion bogie to collect vibration signals between the drive gear and the tooth entry device in real time. These vibration signals contain important dynamic characteristics during the tooth entry process and are preprocessed after signal collection. The preprocessed signals will be used for subsequent analysis of tooth entry contact state indicators and model training. Specifically, speed sensors are installed on each rack and pinion bogie to collect vibration signals between the drive gear and the tooth entry device in real time. Assuming that the vibration signal collected at time t is represented as x t , then the signal sequence recorded during the entire tooth entry process is:

[0040] x = {x1, x2, ···, x T}

[0041] where T is the total time step of the tooth entry process. To ensure the integrity and stability of the signal, the sampling frequency f s needs to satisfy the Nyquist sampling theorem, that is, f s > 2f max , where fmax is the highest characteristic frequency in the signal.

[0042] Specifically, in an embodiment of the present application, the preprocessing includes:

[0043] Normalize the historical driving gear tooth entry vibration signal and perform data annotation according to the corresponding tooth entry state.

[0044] In an embodiment of the present application, the collected original signals may have different amplitude ranges. To eliminate the differences in signal scale, a normalization method is used for standardization processing to stabilize the signal amplitude. The calculation formula is as follows:

[0045]

[0046] where μ and σ are the mean and standard deviation of this batch of data. The normalized signal is used for subsequent model training to improve the convergence and generalization ability of the model.

[0047] Collect the tooth entry vibration signals under different working conditions (such as load changes, track gradients, running speeds), and the data is annotated by experts according to the tooth entry state. Construct a data set where y i represents the tooth entry state label, which is defined as follows:

[0048]

[0049] S103: Build a tooth entry contact state prediction model based on a temporal convolutional network and a temporal attention mechanism and train it.

[0050] In the embodiment of the present application, a temporal convolutional network (TCN) is used as the basic structure to build a deep learning model, and a temporal attention mechanism is introduced to enhance the attention to key time steps. The model uses the tooth entry moment as the supervision signal, and by optimizing the mean square error (MSE) loss function, it can output the "tooth entry contact state index" to characterize the contact state between the driving gear and the tooth entry device. The temporal attention mechanism dynamically assigns weights to improve the model's ability to identify key tooth entry features. During training, the model learns the tooth entry feature patterns in the vibration signal and evaluates the generalization ability of the model through cross-validation. After training is completed, save the offline trained model for online detection.

[0051] Specifically, in an embodiment of the present application, the tooth entry contact state prediction model includes:

[0052] S201: Extract features from the input vibration signal using multiple convolutional layers, and the convolutional layers include causal convolution and dilated convolution.

[0053] S203: Input the extracted features into the temporal attention mechanism to identify the key time points of tooth entry, and weight the obtained attention weights with the output of the convolutional layer.

[0054] S205: Input the weighted temporal features into the fully connected layer to calculate the tooth entry contact state index.

[0055] In one embodiment of the present application, as Figure 2 shown, it is a schematic structural diagram of the tooth entry contact state prediction model. In the vibration signal during the tooth entry process, there are complex dynamic changes in the signal patterns at different time points. To effectively extract the temporal features in the tooth entry signal, a temporal convolutional network (TCN) is constructed to train the input vibration signal. The TCN model includes multiple convolutional layers, and each layer extracts features from the vibration signal through causal convolution and dilated convolution. Causal convolution ensures that the model only depends on past information and does not use future information, thus meeting the constraints of time series prediction. Dilated convolution expands the convolutional receptive field in the time dimension to capture long-term temporal features. The signal output by the l-th convolutional layer is calculated as:

[0056]

[0057] where ω lk is the weight of the l-th convolutional kernel, h l-1 (t) is the output of the (l - 1)-th layer, d is the dilation factor, k is the convolution size, b l is the bias term, and f(·) is the ReLU activation function.

[0058] During the tooth entry process, the vibration signal contains complex dynamic changes, but the information at all time steps is not equally important for determining the tooth entry state. For example, the vibration characteristics before and after tooth entry may be more critical, while the information in the steady state phase is relatively less important. Therefore, after the convolutional layer extracts features, directly inputting the flattened features into the fully connected layer may cause the model to treat all time steps equally and ignore the influence of key time points. To solve this problem, a temporal attention mechanism is introduced to identify the key time points during the tooth entry process, improve the accuracy of the tooth entry contact state index, thereby optimizing the dynamic control strategy, reducing shocks, and improving the smoothness of tooth entry. The calculation formula of the temporal attention mechanism is as follows:

[0059] h l = Flatten(h l )

[0060] e t = W a h l + b a

[0061]

[0062] Among them, α t is the attention weight at the t-th moment, and W a and b a are the weight and bias. Then, the attention weight is weighted with the output h l of the convolutional layer to obtain the weighted feature

[0063]

[0064] After that, the time series feature weighted by the attention mechanism is input into the fully connected layer to calculate the tooth contact state index

[0065]

[0066] Among them, W o and b o are the weight and bias of the fully connected layer, and σ(·) is the Sigmoid function, which maps the output to between (0, 1) as the tooth contact state index.

[0067] In an embodiment of the present application, building and training the tooth contact state prediction model based on the temporal convolutional network and the temporal attention mechanism includes:

[0068] Using the mean squared error loss function to minimize the error between the predicted value and the true label.

[0069] In an embodiment of the present application, the training objective of the TCN is to minimize the error between the predicted value and the true label, and the mean squared error loss function is adopted:

[0070]

[0071] Among them, N is the batch size, y i is the true label, is the model predicted value.

[0072] As Figure 3 shown, during the model training process, the value of the loss function (L mse ) gradually decreases, indicating that the model is continuously optimized and converges to the optimal solution. By optimizing the loss function, the prediction accuracy and stability of the model for the tooth state can be effectively improved.

[0073] In an embodiment of the present application, building and training the tooth contact state prediction model based on the temporal convolutional network and the temporal attention mechanism further includes:

[0074] Using the dynamic time warping error to evaluate the fitting ability of the model, and selecting the model with the minimum error for online detection.

[0075] In one embodiment of the present application, a rolling prediction method is adopted to cross-validate the fitting ability of the model for the tooth entry state index. The fitting error of the model is calculated through Dynamic Time Warping (DTW), which is defined as follows:

[0076]

[0077]

[0078]

[0079] Select the model with the minimum DTW error for online detection, which is defined as follows:

[0080]

[0081] where is the set of all trained models, and M * is the model with the minimum DTW error. After selecting the model, fix the parameters and deploy them to the online detection system to calculate the tooth entry contact state index in real time and execute the power control strategy of the driving gear.

[0082] In one embodiment of the present application, as Figure 4 shown, this method combines a Temporal Convolutional Network (TCN) with a temporal attention mechanism for model training. In the offline training stage, first extract the temporal features in the vibration signal from historical data to capture the dynamic change patterns during the tooth entry process. The Temporal Convolutional Network can efficiently capture the temporal information in the vibration signal, while the temporal attention mechanism highlights the feature information of key time points by assigning weights to different time steps. Use the Mean Squared Error (MSE) as the loss function to optimize the parameters of the model, and evaluate the generalization ability of the model through cross-validation. At the same time, use the Dynamic Time Warping (DTW) algorithm to evaluate the fitting effect of the model to ensure that the selected model has good accuracy and stability. After training is completed, save the optimal model for online detection. During the train operation, the real-time collected vibration signal will be input into the trained model, and the model will judge the tooth entry process by calculating the "tooth entry contact state index".

[0083] S105: Collect the real-time tooth entry vibration signal of the driving gear, input it into the trained tooth entry contact state prediction model, and obtain the tooth entry contact state index.

[0084] In the embodiment of the present application, during the actual operation of the gear-rail train, the real-time tooth entry vibration signal x of the driving gear is obtained through a sliding window strategy, and this data is input into the trained tooth entry contact state prediction model for processing. The TCN extracts the features of the input data through multi-layer causal convolution and dilated convolution to obtain the temporal feature h l(t), the contribution of features at different time steps to the tooth entry detection task is calculated using the temporal attention mechanism to obtain the weight α t , and then the features are weighted The extracted weighted features are calculated through a fully connected layer to obtain the tooth entry contact state index at the current moment which is used to quantify the contact state between the current driving gear and the tooth entry device

[0085] S107: When the tooth entry contact state index exceeds a preset threshold, an instruction for the start of tooth entry signal is output to perform a power cut-off operation; when the tooth entry contact state index does not exceed the preset threshold, an instruction for the completion of tooth entry signal is output to turn on the power output

[0086] In the embodiment of the present application, when the model calculates the tooth entry contact state index , it is determined according to a preset threshold τ. When , it is determined that the driving gear enters the tooth entry state and an instruction for the start of tooth entry signal is output; when returns to the normal value, it indicates that the tooth entry process is over and an instruction for the completion of tooth entry signal is output. Specifically, as Figure 5 shown, when it is detected that the "tooth entry contact state index" exceeds the preset threshold and the start of tooth entry signal is confirmed, the control system issues an instruction through the traction control unit to control the power output of the driving motor. This instruction will perform a power cut-off operation through the motor control module, immediately interrupting the electric driving force of the driving motor, so that the driving gear meshes with the tooth entry device only relying on the adhesion between the bogie and the track. When it is detected that the "tooth entry contact state index" returns to the normal range and remains stable, the control system identifies the completion of tooth entry signal and issues an instruction to resume driving through the traction control unit. This instruction is transmitted to the motor control module to turn on the power output again, and the driving motor resumes normal power supply, and the driving gear switches back to the traction mode. For each tooth-rail bogie on the train, the whole process of tooth entry state recognition, control response, and power recovery is executed in sequence. The tooth entry state of each bogie is monitored in real time, and the power control is adjusted according to the "tooth entry contact state index" to ensure that each bogie completes the tooth entry process smoothly and stably

[0087] In the above automatic control method for the tooth-rail train driving gear engaging operation, first, historical driving gear engaging vibration signals under different working conditions are collected and preprocessed; then, a prediction model for the engaging contact state is built based on the temporal convolutional network (TCN) and the temporal attention mechanism and trained; then, real-time driving gear engaging vibration signals are collected and input into the trained prediction model for the engaging contact state to obtain the engaging contact state index; finally, when the engaging contact state index exceeds a preset threshold, an engaging start signal command is output to perform the power cut-off operation, and when the engaging contact state index does not exceed the preset threshold, an engaging completion signal command is output to turn on the power output. That is to say, by installing an acceleration sensor on the tooth-rail bogie of the tooth-rail train to collect the vibration signals of the driving gear and the engaging device in real time, and combining with the temporal convolutional network (TCN) for feature extraction to effectively identify the key temporal features in the engaging process, and introducing the temporal attention mechanism to focus on the critical moments before and after engaging on the basis of TCN feature extraction, the accuracy of engaging state determination is improved, an engaging contact state index is constructed, and the accurate monitoring of the engaging state is realized. Through this method, the dynamic control strategy can be optimized to make the driving gear engaging process smoother, thereby reducing impact, improving the engaging smoothness, and enhancing the running stability and system service life of the tooth-rail train.

[0088] It should be understood that although the steps in the flowcharts involved in the above embodiments are shown in sequence according to the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order limit for the execution of these steps, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages do not necessarily have to be executed at the same time, but can be executed at different times, and the execution order of these steps or stages does not necessarily have to be sequential, but can be executed alternately or in turn with at least some of the steps or stages in other steps or other steps.

[0089] Based on the same inventive concept, an embodiment of the present application also provides a tooth-rail train driving gear engaging operation automatic control device for implementing the above-mentioned automatic control method for the tooth-rail train driving gear engaging operation. The solution provided by this device for solving the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the tooth-rail train driving gear engaging operation automatic control device provided below can refer to the limitations for the automatic control method for the tooth-rail train driving gear engaging operation in the above text, and will not be repeated here.

[0090] In one embodiment, as Figure 6As shown, an automatic control device 600 for the tooth engagement operation of a rack and pinion train is provided, including: a data acquisition and preprocessing module 601, a tooth engagement contact state prediction model construction and training module 603, a tooth engagement contact state prediction module 605, and a tooth engagement operation control module 607, where:

[0091] The data acquisition and preprocessing module 601 is used to acquire the historical tooth engagement vibration signals of the driving gear under different working conditions and perform preprocessing;

[0092] The tooth engagement contact state prediction model construction and training module 603 is used to construct and train a tooth engagement contact state prediction model based on a temporal convolutional network and a temporal attention mechanism;

[0093] The tooth engagement contact state prediction module 605 is used to acquire the real-time tooth engagement vibration signal of the driving gear, input it into the trained tooth engagement contact state prediction model, and obtain the tooth engagement contact state index;

[0094] The tooth engagement operation control module 607 is used to output a tooth engagement start signal instruction and perform a power cut-off operation when the tooth engagement contact state index exceeds a preset threshold, and output a tooth engagement completion signal instruction and turn on the power output when the tooth engagement contact state index does not exceed the preset threshold.

[0095] In an embodiment of the present application, the data acquisition and preprocessing module is further used for:

[0096] Normalize and standardize the historical tooth engagement vibration signals of the driving gear using a normalization method, and perform data annotation according to the corresponding tooth engagement states.

[0097] In an embodiment of the present application, the tooth engagement contact state prediction model includes:

[0098] Extract features from the input vibration signal using multiple convolutional layers, where the convolutional layers include causal convolution and dilated convolution;

[0099] Input the extracted features into the temporal attention mechanism to identify the key time points of tooth engagement, and weight the obtained attention weights with the output of the convolutional layer;

[0100] Input the weighted temporal features into a fully connected layer to calculate the tooth engagement contact state index.

[0101] In an embodiment of the present application, the tooth engagement contact state prediction model construction and training module is further used for:

[0102] Minimize the error between the predicted value and the true label using a mean square error loss function.

[0103] In an embodiment of the present application, the tooth engagement contact state prediction model construction and training module is further used for:

[0104] Adopt the fitting ability of the dynamic time warping error evaluation model, and select the model with the minimum error for online detection.

[0105] Each module in the above-mentioned automatic control device for the tooth-rail train driving gear engaging operation can be implemented in whole or in part by software, hardware, and their combination. Each of the above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so as to facilitate the processor to call and execute the operations corresponding to each of the above modules.

[0106] In one embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 7 shown. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it realizes an automatic control method for the tooth-rail train driving gear engaging operation. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the shell of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0107] Those skilled in the art can understand that Figure 7 the structure shown in

[0108] is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0109] In one embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, it realizes the steps in the above method embodiments.

[0110] In one embodiment, a computer program product is provided, including a computer program which, when executed by a processor, implements the steps in the above method embodiments.

[0111] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data that have been authorized by the user or fully authorized by all parties.

[0112] Those of ordinary skill in the art can understand that all or part of the processes of implementing the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0113] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0114] The above-described embodiments merely represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. An automatic control method for the tooth engagement operation of a rack and pinion train, characterized in that, The method includes: Collecting historical driving gear tooth - entry vibration signals under different working conditions and performing pre - processing; Building and training a tooth - entry contact state prediction model based on a temporal convolutional network and a temporal attention mechanism; Collecting real - time driving gear tooth - entry vibration signals, inputting them into the trained tooth - entry contact state prediction model, and obtaining tooth - entry contact state indicators; When the tooth - entry contact state indicator exceeds a preset threshold, outputting a tooth - entry start signal instruction and performing a power - off operation; when the tooth - entry contact state indicator does not exceed the preset threshold, outputting a tooth - entry completion signal instruction and turning on the power output.

2. The automatic control method for the tooth-engaging operation of the driving gear of a rack railway vehicle according to claim 1, wherein The pre - processing includes: Normalizing the historical driving gear tooth - entry vibration signals using a normalization method and performing data annotation according to the corresponding tooth - entry states.

3. The automatic control method for the tooth-engaging operation of the drive gear of a rack and pinion train according to claim 1, characterized in that, The tooth - entry contact state prediction model includes: Extracting features from the input vibration signals using multiple convolutional layers, where the convolutional layers include causal convolution and dilated convolution; Inputting the extracted features into a temporal attention mechanism to identify key tooth - entry time points, and weighting the obtained attention weights with the output of the convolutional layer; Inputting the weighted temporal features into a fully - connected layer to calculate the tooth - entry contact state indicator.

4. The automatic control method for the tooth-engaging operation of the drive gear of a rack railway vehicle according to claim 1, wherein, Building and training the tooth - entry contact state prediction model based on the temporal convolutional network and the temporal attention mechanism includes: Minimizing the error between the predicted value and the true label using a mean - squared - error loss function.

5. The automatic control method for the tooth-engaging operation of the drive gear of a rack railway train according to claim 4, wherein, Building and training the tooth - entry contact state prediction model based on the temporal convolutional network and the temporal attention mechanism further includes: Evaluating the fitting ability of the model using a dynamic time warping error and selecting the model with the minimum error for online detection.

6. An automatic control device for the tooth engagement operation of a rack railway drive gear, characterized in that, The device includes: A data collection and pre - processing module for collecting historical driving gear tooth - entry vibration signals under different working conditions and performing pre - processing; A tooth - entry contact state prediction model building and training module for building and training a tooth - entry contact state prediction model based on a temporal convolutional network and a temporal attention mechanism; A tooth - entry contact state prediction module for collecting real - time driving gear tooth - entry vibration signals, inputting them into the trained tooth - entry contact state prediction model, and obtaining tooth - entry contact state indicators; A tooth - entry operation control module for outputting a tooth - entry start signal instruction and performing a power - off operation when the tooth - entry contact state indicator exceeds a preset threshold, and outputting a tooth - entry completion signal instruction and turning on the power output when the tooth - entry contact state indicator does not exceed the preset threshold.

7. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 5.

Images