Method, device and system for preventing derailment of a bogie wheel pair of a roadheader

Abstract

Embodiments of the present application provide a derailing prevention method, device and system for a bogie wheel pair of a heading machine. The method comprises: acquiring multi-source motion posture data of a derailing detection mechanical trolley and track image data of a heading direction, and performing normalization processing on the multi-source motion posture data to obtain normalized multi-source motion posture data; inputting the normalized multi-source motion posture data into a derailing prediction model trained by historical motion posture data of the bogie wheel pair to obtain a derailing risk value; subsequently, obtaining a derailing detection result of the heading machine according to the derailing risk value and the track image data, and when a driving state of the derailing detection result is a derailing precursor, generating a derailing prevention signal corresponding to a driving danger level according to the driving danger level of the derailing detection result, the derailing prevention signal being used to control the heading machine to perform a corresponding derailing prevention action. The method is used to achieve the effect of accurate prediction and effective early prevention and control of the derailing accident of the bogie wheel pair of the heading machine.

Description

Technical Field

[0001] This application relates to the field of automatic control technology, and in particular to a method, device and system for preventing derailment of the trolley wheelset of a tunneling machine. Background Technology

[0002] During tunnel excavation, the tunnel boring machine (TBM) trolley system, as the core equipment, directly affects the efficiency of construction and the safety of on-site operations. Tunnel excavation operations are characterized by complex environments and variable geological conditions, often involving fractured rock strata and soft ground. Furthermore, the TBM's working face track is subjected to the constant pressure and vibration of heavy equipment and is susceptible to groundwater seepage and erosion, making it highly prone to localized settlement, misalignment, or even breakage of the track. This can lead to derailment accidents involving the TBM's wheelsets, seriously threatening construction safety.

[0003] Existing derailment detection technologies are mostly designed for scenarios such as high-speed railways and industrial rail transportation. Their core detection logic is to collect equipment operating status data by deploying detection elements such as accelerometers and tilt sensors on the train or track, and then determine whether a derailment has occurred by combining the data with preset thresholds.

[0004] However, the core design of existing derailment detection technologies is adapted to high-speed operating scenarios, making it difficult to meet the dynamic detection requirements of the wheels of tunneling machines under millimeter-level low-speed operating conditions. This makes it impossible to effectively identify derailment precursors under low-speed conditions, resulting in a lack of foresight and timeliness in derailment detection results, and making it difficult to predict and effectively prevent derailment accidents in advance. Summary of the Invention

[0005] This application provides a method, device, and system for preventing derailment of the wheelset of a tunneling machine, so as to achieve accurate prediction and effective early prevention of derailment accidents of the tunneling machine wheelset.

[0006] In a first aspect, embodiments of this application provide a method for preventing derailment of a tunneling machine's trolley wheelset. The trolley wheelset is externally connected to a derailment detection trolley via a rigid connection. The trolley wheelset and the derailment detection trolley move synchronously along the track in the tunneling direction with the tunneling machine. The method includes:

[0007] Acquire multi-source motion attitude data and track image data in the tunneling direction of the derailment detection trolley.

[0008] The multi-source motion posture data is normalized to obtain normalized multi-source motion posture data.

[0009] Normalized multi-source motion attitude data are input into the derailment prediction model to obtain the derailment risk value; the derailment prediction model is trained using historical motion attitude data of the trolley wheelsets.

[0010] Based on the derailment risk value and track image data, the derailment detection results of the trolley wheelset are obtained; the derailment detection results include the running status and the running hazard level.

[0011] When the driving state shows signs of derailment, a derailment prevention signal corresponding to the driving hazard level is generated; the derailment prevention signal is used to control the tunneling machine to perform corresponding derailment prevention actions.

[0012] In one possible implementation, in conjunction with the first aspect, the training process of the derailment prediction model includes:

[0013] Obtain multiple sets of motion posture samples of the tunnel boring machine's trolley wheelsets; the motion posture sample sets include historical motion posture data and derailment risk label values ​​corresponding to the historical motion posture data.

[0014] By inputting historical motion posture data into the derailment prediction model, a derailment risk prediction value is obtained.

[0015] By using a preset loss function, the error between the predicted derailment risk value and the derailment risk label value is calculated.

[0016] Based on the error, the parameters of the derailment prediction model are updated; and the historical motion posture data is repeatedly input into the derailment prediction model to obtain the derailment risk prediction value, until the preset convergence condition is met, and the trained derailment prediction model is obtained.

[0017] In one possible implementation, in conjunction with the first aspect, acquiring multi-source motion attitude data of the derailment detection trolley and track image data in the tunneling direction includes:

[0018] The first sensor on the derailment detection trolley collects track image data in the direction of tunneling.

[0019] The original multi-source motion attitude dataset of the derailment detection mechanical trolley is collected by multiple second sensors on the trolley; the multiple second sensors adopt a redundant sensor layout.

[0020] The original multi-source motion posture dataset is weighted and fused to obtain the multi-source motion posture data of the derailment detection mechanical vehicle.

[0021] In one possible implementation, in conjunction with the first aspect, the method further includes:

[0022] By uploading multi-source motion attitude data, track image data, and derailment detection results to the cloud platform, maintenance suggestions for the tunneling machine can be obtained.

[0023] In one possible implementation, in conjunction with the first aspect, the derailment detection results of the train wheelset are obtained based on the derailment risk value and track image data, including:

[0024] Image recognition is performed on the track image data to obtain the track recognition results.

[0025] Based on the derailment risk value and track identification results, the derailment detection results of the trolley wheelset are obtained.

[0026] In one possible implementation, in conjunction with the first aspect, image recognition is performed on the orbital image data to obtain an orbital recognition result, including:

[0027] Multi-scale feature extraction is performed on the orbital image data to obtain the multi-scale features of the orbital image.

[0028] Feature enhancement processing is performed on the multi-scale features of the track image to obtain the track recognition result.

[0029] In one possible implementation, in conjunction with the first aspect, based on the derailment risk value and the track identification result, the derailment detection result of the trolley wheelset is obtained, including:

[0030] Based on the derailment risk value and track identification results, the running status and running hazard level of the trolley wheelset are determined by using a pre-set derailment detection rule library.

[0031] The driving status and driving hazard level are determined as the derailment detection results.

[0032] In one possible implementation, in conjunction with the first aspect, the derailment prevention signal includes at least a primary derailment signal and a secondary derailment signal; the primary derailment signal is used to control the tunneling machine to stop tunneling and activate a primary alarm; the secondary derailment signal is used to control the tunneling machine to activate a secondary alarm.

[0033] Secondly, embodiments of this application provide a derailment prevention device for the wheelset of a tunneling machine, comprising:

[0034] The acquisition module is used to acquire multi-source motion attitude data and track image data in the tunneling direction of the derailment detection trolley.

[0035] The data processing module is used to normalize multi-source motion posture data to obtain normalized multi-source motion posture data.

[0036] The derailment detection module is used to input normalized multi-source motion attitude data into the derailment prediction model to obtain the derailment risk value. The derailment prediction model is trained using historical motion attitude data of the trolley wheelset. Based on the derailment risk value and track image data, the derailment detection result of the trolley wheelset is obtained. The derailment detection result includes the running status and running hazard level.

[0037] The generation module is used to generate a derailment prevention signal corresponding to the driving hazard level when the driving state shows signs of derailment. The derailment prevention signal is used to control the tunneling machine to perform corresponding derailment prevention actions.

[0038] In one possible implementation, in conjunction with the second aspect, a derailment detection module is used to train a derailment prediction model.

[0039] In one possible implementation, in conjunction with the second aspect, the derailment detection module is specifically used for:

[0040] Obtain multiple sets of motion posture samples of the tunnel boring machine's trolley wheelsets; the motion posture sample sets include historical motion posture data and derailment risk label values ​​corresponding to the historical motion posture data.

[0041] By inputting historical motion posture data into the derailment prediction model, a derailment risk value is obtained.

[0042] The error between the derailment risk value and the derailment risk label value is calculated by using a preset loss function.

[0043] Based on the error, the parameters of the derailment prediction model are updated; and the historical motion posture data is repeatedly input into the derailment prediction model to obtain the derailment risk value, until the preset convergence condition is met, and the trained derailment prediction model is obtained.

[0044] In one possible implementation, in conjunction with the second aspect, the acquisition module is specifically used for:

[0045] The first sensor on the derailment detection trolley collects track image data in the direction of tunneling.

[0046] The original multi-source motion attitude dataset of the derailment detection mechanical trolley is collected by multiple second sensors on the trolley; the multiple second sensors adopt a redundant sensor layout.

[0047] The original multi-source motion posture dataset is weighted and fused to obtain the multi-source motion posture data of the derailment detection mechanical vehicle.

[0048] In one possible implementation, in conjunction with the second aspect, the acquisition module is used to obtain maintenance suggestions for the tunneling machine by uploading multi-source motion attitude data, track image data, and derailment detection results to a cloud platform.

[0049] In one possible implementation, in conjunction with the second aspect, the derailment detection module is specifically used for:

[0050] Normalized multi-source motion attitude data are input into the derailment prediction model to obtain the derailment risk value.

[0051] Image recognition is performed on the track image data to obtain the track recognition results.

[0052] Based on the derailment risk value and track identification results, the derailment detection results of the trolley wheelset are obtained.

[0053] In one possible implementation, in conjunction with the second aspect, the derailment detection module is specifically used for:

[0054] Normalized multi-source motion attitude data are input into the derailment prediction model to obtain the derailment risk value.

[0055] Multi-scale feature extraction is performed on the orbital image data to obtain the multi-scale features of the orbital image.

[0056] Feature enhancement processing is performed on the multi-scale features of the track image to obtain the track recognition result.

[0057] Based on the derailment risk value and track identification results, the derailment detection results of the trolley wheelset are obtained.

[0058] In one possible implementation, in conjunction with the second aspect, the derailment detection module is specifically used for:

[0059] Normalized multi-source motion attitude data are input into the derailment prediction model to obtain the derailment risk value.

[0060] Multi-scale feature extraction is performed on the orbital image data to obtain the multi-scale features of the orbital image.

[0061] Feature enhancement processing is performed on the multi-scale features of the track image to obtain the track recognition result.

[0062] Based on the derailment risk value and track identification results, the running status and running hazard level of the trolley wheelset are determined by using a pre-set derailment detection rule library.

[0063] The driving status and driving hazard level are determined as the derailment detection results.

[0064] In one possible implementation, in conjunction with the second aspect, the derailment prevention signal includes at least a primary derailment signal and a secondary derailment signal; the primary derailment signal is used to control the tunneling machine to stop tunneling and activate a primary alarm; the secondary derailment signal is used to control the tunneling machine to activate a secondary alarm.

[0065] Thirdly, embodiments of this application provide a derailment prevention system for the wheelset of a tunneling machine, including a derailment detection trolley and a tunneling machine main control device; the derailment detection trolley includes a multi-modal sensor and a transmitting module; the tunneling machine main control device includes a receiving module, a derailment prevention device, and an execution control module; the derailment detection trolley is rigidly connected to the wheelset of the tunneling machine.

[0066] Among them, the multimodal sensor is used to collect multi-source motion attitude data of the derailment detection trolley and track image data in the tunneling direction.

[0067] The transmitting module is used to transmit multi-source motion attitude data and orbital image data to the receiving module.

[0068] The receiving module is used to receive multi-source motion attitude data and track image data, and transmit them to the derailment prevention device.

[0069] The derailment prevention device is used to generate a derailment prevention signal based on multi-source motion attitude data and track image data, so that the derailment prevention device performs the first aspect and / or various possible implementations of the first aspect as described above.

[0070] The execution control module is used to receive derailment prevention signals and control the tunneling machine to perform corresponding derailment prevention actions based on the derailment prevention signals.

[0071] Fourthly, embodiments of this application provide an electronic device, including: a processor, and a memory communicatively connected to the processor.

[0072] The memory stores the instructions that the computer executes.

[0073] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0074] Fifthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0075] In a sixth aspect, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0076] This application provides a method, device, and system for preventing derailment of the wheelset of a tunneling machine. It acquires multi-source motion attitude data of the derailment detection trolley and track image data along the tunneling direction, and normalizes the multi-source motion attitude data to a unified scale, obtaining normalized multi-source motion attitude data. This eliminates the differences in dimensions and numerical scales between acceleration, tilt angle, and roll angle in the multi-source motion attitude data, adapting to the millimeter-level low-speed tunneling conditions of the tunneling machine. It effectively highlights the subtle acceleration changes in low-speed scenarios, preventing this core weak signal from being masked by other data dimensions or environmental noise. The normalized multi-source motion attitude data is then input into a derailment prediction model to obtain a derailment risk value. This derailment prediction model, trained using historical motion attitude data of the wheelset, can accurately capture subtle acceleration changes under millimeter-level low-speed conditions. Subsequently, based on the derailment risk value and track image data, the derailment detection results of the tunneling machine are obtained. When the derailment detection results indicate that the driving state is a precursor to derailment, a derailment prevention signal corresponding to the driving hazard level is generated according to the driving hazard level of the derailment detection results. This derailment prevention signal is used to control the tunneling machine to perform corresponding derailment prevention actions. Ultimately, the effect of accurate prediction and effective early prevention of derailment accidents of the tunneling machine's wheel pair is achieved. Attached Figure Description

[0077] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0078] Figure 1 A schematic diagram illustrating a scenario for a method to prevent derailment of the trolley wheelset of a tunneling machine provided in this application;

[0079] Figure 2 A flowchart illustrating a method for preventing derailment of the trolley wheelset of a tunneling machine provided in this application. Figure 1 ;

[0080] Figure 3 A flowchart illustrating a method for preventing derailment of the trolley wheelset of a tunneling machine provided in this application. Figure 2 ;

[0081] Figure 4 A flowchart illustrating a method for preventing derailment of the trolley wheelset of a tunneling machine provided in this application. Figure 3 ;

[0082] Figure 5 This is a schematic diagram of the orbital image data processing process provided in this application;

[0083] Figure 6 A specific example of a derailment prevention method for the trolley wheelset of a tunneling machine provided in this application. Figure 1 ;

[0084] Figure 7 A specific example of a derailment prevention method for the trolley wheelset of a tunneling machine provided in this application. Figure 2 ;

[0085] Figure 8 A schematic diagram of the structure of a derailment prevention device for a tunneling machine's trolley wheelset provided in this application;

[0086] Figure 9 A schematic diagram of the structure of a derailment prevention system for a tunneling machine's trolley wheelset provided in this application;

[0087] Figure 10 A schematic diagram of the structure of the electronic device provided in this application.

[0088] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0089] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0090] The application background of the embodiments of this application will be explained below:

[0091] In tunnel excavation, the tunnel boring machine (TBM) trolley system, as the core equipment, directly affects construction efficiency and on-site safety. Tunnel excavation operations are characterized by complex environments and variable geological conditions, often involving fractured rock strata and soft ground. Furthermore, the TBM's working face track is subjected to the constant pressure and vibration of heavy equipment and is susceptible to groundwater seepage and erosion, making it highly prone to localized settlement, misalignment, or even breakage. This can lead to derailment accidents involving the TBM's wheelsets, seriously threatening construction safety. Existing derailment detection technologies are mostly designed for high-speed railways and industrial rail transport. Their core detection logic involves collecting equipment operating status data by deploying accelerometers, tilt sensors, and other detection elements on the train or track, and then combining this data with preset thresholds to determine whether derailment has occurred. However, the core design of existing derailment detection technologies is adapted to high-speed operating scenarios, making it difficult to meet the dynamic detection requirements of the wheels of tunneling machines under millimeter-level low-speed operating conditions. This makes it impossible to effectively identify derailment precursors under low-speed conditions, resulting in a lack of foresight and timeliness in derailment detection results, and making it difficult to predict and effectively prevent derailment accidents in advance.

[0092] To address the aforementioned issues, the inventors investigated whether multimodal sensor fusion, derailment detection models, and image recognition technology could be used to predict and effectively prevent derailment accidents of tunnel boring machine (TBM) wheelsets under millimeter-level low-speed operating conditions. The inventors proposed a derailment prevention method for TBM wheelsets. This method acquires multi-source motion attitude data of the derailment detection trolley and track image data along the tunneling direction. The multi-source motion attitude data is then normalized and mapped to a unified scale to obtain normalized multi-source motion attitude data. This eliminates the dimensional and numerical scale differences between acceleration, tilt angle, and roll angle in the multi-source motion attitude data, adapting to the millimeter-level low-speed tunneling conditions of the TBM. It effectively highlights the subtle acceleration changes in low-speed scenarios, preventing this core weak signal from being masked by other data dimensions or environmental noise. The normalized multi-source motion attitude data is then input into a derailment prediction model to obtain a derailment risk value. This derailment prediction model, trained using historical motion attitude data of the TBM wheelsets, can accurately capture subtle acceleration changes under millimeter-level low-speed conditions. Subsequently, based on the derailment risk value and track image data, the derailment detection results of the tunneling machine are obtained. When the derailment detection results indicate that the driving state is a precursor to derailment, a derailment prevention signal corresponding to the driving hazard level is generated according to the driving hazard level of the derailment detection results. This derailment prevention signal is used to control the tunneling machine to perform corresponding derailment prevention actions. Ultimately, the effect of accurate prediction and effective early prevention of derailment accidents of the tunneling machine's wheel pair is achieved.

[0093] Taking the scenario of track safety monitoring and prevention for tunneling machine wheelsets as an example, combined with Figure 1 This illustrates the specific application scenario of the derailment prevention method for the trolley wheelset of a tunneling machine provided in this application. For example... Figure 1As shown, the specific application scenarios of this application include a derailment detection trolley, a tunneling machine wheelset, the tunneling machine's main control equipment, and a track. The tunneling machine wheelset is rigidly connected to the derailment detection trolley, and the two move synchronously on the track along the tunneling direction, ensuring that their motion postures are completely identical. The derailment detection trolley is equipped with a multi-modal sensor and a connected wireless communication module: the multi-modal sensor is used to collect the original multi-source motion posture data of the derailment detection trolley in real time, as well as track image data of the track ahead in the tunneling direction; the wireless communication module is used to communicate with the tunneling machine's main control equipment. The tunneling machine's main control equipment obtains all the data collected by the multi-modal sensor of the derailment detection trolley through the wireless communication module of the derailment detection trolley, and generates the derailment detection result of the tunneling machine wheelset based on the fusion analysis of the original multi-source motion posture data and the track image data.

[0094] For example, if the track ahead is missing in the tunneling direction, the derailment detection trolley will derail first. The driving status in the derailment detection result generated at this time is a precursor to derailment. The tunneling machine's main control equipment will generate a derailment prevention signal based on the driving danger level in the derailment detection result, and control the tunneling machine's wheels to perform corresponding derailment prevention actions on the tunneling machine to achieve early intervention in derailment accidents.

[0095] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0096] Figure 2 A flowchart illustrating a method for preventing derailment of the trolley wheelset of a tunneling machine provided in this application. Figure 1 In this embodiment, the trolley wheelset is rigidly connected to a derailment detection trolley, and the trolley wheelset and the derailment detection trolley move synchronously along the track in the tunneling direction together with the tunneling machine. Figure 2 As shown, the method includes:

[0097] S201. Acquire multi-source motion attitude data and track image data in the tunneling direction of the derailment detection trolley.

[0098] The multi-source motion attitude data includes acceleration, tilt angle, and roll angle.

[0099] In this step, track image data along the tunneling direction is acquired using the first sensor on the derailment detection trolley. Simultaneously, multiple second sensors on the trolley acquire raw multi-source motion attitude datasets. These raw multi-source motion attitude datasets are then weighted and fused to obtain the multi-source motion attitude data of the derailment detection trolley. The multiple second sensors employ a redundant sensor layout and can be multi-modal sensors.

[0100] The aforementioned derailment detection trolley is rigidly connected to the wheel pair of the trolley, effectively avoiding data acquisition deviations caused by speed differences.

[0101] S202. Normalize the multi-source motion posture data to obtain normalized multi-source motion posture data.

[0102] In this step, the acceleration, tilt angle, and roll angle are normalized to obtain normalized multi-source motion attitude data.

[0103] In one possible implementation, the following normalization formula is used to normalize the multi-source motion attitude data within a certain acquisition period:

[0104]

[0105] in, This is the actual value, which is a single data point in the multi-source motion attitude data collected by the second sensor; This represents the maximum value of all actual values ​​in the multi-source motion attitude data within the acquisition period. It is the minimum value of all actual values ​​in the multi-source motion attitude data within this acquisition period; Actual value After calculation by the formula, it is mapped to Dimensionless data for an interval.

[0106] For example, the range of lateral acceleration values ​​collected within a certain acquisition period is 0.02~0.88 m / s². 2 The lateral acceleration collected at a certain moment is ,right Normalization process is performed to obtain The calculation process is as follows:

[0107]

[0108] S203. Input the normalized multi-source motion attitude data into the derailment prediction model to obtain the derailment risk value.

[0109] The derailment prediction model is trained using a motion posture sample set, which includes historical motion posture data of the tunnel boring machine's trolley wheelsets and their corresponding derailment risk label values.

[0110] In this step, the normalized acceleration, tilt angle, and roll angle are input into the trained derailment prediction model to obtain the normalized model output value. Then, the model output value is denormalized to obtain the derailment risk value.

[0111] In one possible implementation, the formula for inverse normalization of the model output values ​​is as follows:

[0112]

[0113] in, This represents the risk value for derailment. Output values ​​for the model; To set a preset risk ceiling; The preset risk lower limit is set; the preset risk upper limit and preset risk lower limit are determined based on the derailment risk label values ​​in the motion posture sample set.

[0114] S204. Based on the derailment risk value and track image data, the derailment detection results of the trolley wheelset are obtained. The derailment detection results include the running status and the running hazard level.

[0115] In this step, image recognition is first performed on the track image data to obtain the track recognition result. Then, based on the derailment risk value and the track recognition result, the derailment detection result of the tunneling machine is obtained.

[0116] Specifically, multi-scale feature extraction is performed on the track image data to obtain multi-scale features of the track image. Then, feature enhancement processing is performed on the multi-scale features of the track image to obtain the track recognition result. Subsequently, based on the derailment risk value and the track recognition result, the driving status and driving hazard level of the trolley wheelset are determined through a preset derailment detection rule library, and the driving status and driving hazard level are determined as the derailment detection result.

[0117] S205. When the driving state shows signs of derailment, generate a derailment prevention signal corresponding to the driving hazard level.

[0118] Among them, the derailment prevention signal is used to control the tunneling machine to perform corresponding derailment prevention actions, including at least a first-level derailment signal and a second-level derailment signal. The first-level derailment signal is used to control the tunneling machine to stop tunneling and activate the first-level alarm, and the second-level derailment signal is used to control the tunneling machine to activate the second-level alarm.

[0119] In one possible implementation, the preset derailment detection rule base includes the following rules: if the track identification result is that the track ahead is abnormal or missing, the travel status of the trolley wheelset can be determined as a pre-derailment warning, and the travel hazard level of the trolley wheelset is Level 1 derailment hazard; if the derailment risk value is greater than the preset derailment threshold, the travel status of the trolley wheelset can be determined as a pre-derailment warning, and the travel hazard level of the trolley wheelset is Level 2 derailment hazard; if the track identification result is normal, and the derailment risk value is less than or equal to the preset derailment threshold, the travel status of the trolley wheelset can be determined as a pre-derailment warning, and the travel hazard level of the trolley wheelset is no derailment hazard.

[0120] Correspondingly, if the driving hazard level is Level 1 derailment hazard, a Level 1 derailment signal is generated, which can be used to control the tunneling machine to stop tunneling and activate the Level 1 alarm; if the driving hazard level is Level 2 derailment hazard, a Level 2 derailment signal is generated, which can be used to control the tunneling machine to stop tunneling and activate the Level 2 alarm.

[0121] This application provides a method for preventing derailment of the wheelset of a tunneling machine. It acquires multi-source motion attitude data of the derailment detection trolley and track image data along the tunneling direction. The multi-source motion attitude data is normalized and mapped to a uniform scale to obtain normalized multi-source motion attitude data. This eliminates the differences in dimensions and numerical scales between acceleration, tilt angle, and roll angle in the multi-source motion attitude data, adapting to the millimeter-level low-speed tunneling conditions of the tunneling machine. It effectively highlights the subtle acceleration changes in low-speed scenarios, preventing this core weak signal from being masked by other data dimensions or environmental noise. The normalized multi-source motion attitude data is then input into a derailment prediction model to obtain a derailment risk value. This derailment prediction model is trained using historical motion attitude data of the wheelset and can accurately capture subtle acceleration changes in millimeter-level low-speed conditions. Subsequently, based on the derailment risk value and track image data, the derailment detection results of the tunneling machine are obtained. When the derailment detection results indicate that the driving state is a precursor to derailment, a derailment prevention signal corresponding to the driving hazard level is generated according to the driving hazard level of the derailment detection results. This derailment prevention signal is used to control the tunneling machine to perform corresponding derailment prevention actions. Ultimately, the effect of accurate prediction and effective early prevention of derailment accidents of the tunneling machine's wheel pair is achieved.

[0122] Figure 3 A flowchart illustrating a method for preventing derailment of the trolley wheelset of a tunneling machine provided in this application. Figure 2 ,like Figure 3 As shown, in this embodiment... Figure 2 Based on the examples, the training process of the derailment prediction model is described in detail, which includes the following steps:

[0123] S301. Obtain multiple sets of motion posture sample sets; the motion posture sample sets include historical motion posture data of the tunneling machine's trolley wheelsets and the derailment risk label values ​​corresponding to the historical motion posture data.

[0124] Among them, the derailment risk label value can be manually marked on historical motion posture data, indicating the derailment risk under that historical motion posture data.

[0125] S302. Input historical motion posture data into the derailment prediction model to obtain the derailment risk prediction value.

[0126] S303. The error between the predicted derailment risk value and the derailment risk label value is calculated by using a preset loss function.

[0127] S304. Based on the error, update the parameters of the derailment prediction model; and repeatedly input historical motion posture data into the derailment prediction model to obtain the derailment risk prediction value until the preset convergence condition is met, and obtain the trained derailment prediction model.

[0128] In one possible implementation, the derailment prediction model has an input layer dimension of 3, a hidden layer number of 5 nodes, an output layer dimension of 1, and a learning rate of [missing value]. Excitation function Take the sigmoid function; its corresponding formula is as follows:

[0129]

[0130] Before training the derailment prediction model, network initialization is performed: the learning rate and error convergence threshold are set; all model parameters are randomly initialized. The default convergence condition for the derailment prediction model is that the error is less than the error convergence threshold.

[0131] Then input the training samples and perform forward propagation. Specifically, input a set of normalized historical motion posture data. Input the derailment prediction model, and calculate the hidden layer output and output layer prediction values ​​(referred to as derailment risk values ​​in some embodiments) layer by layer. Calculate the hidden layer output. The formula is as follows:

[0132]

[0133] in, For the first The input layer node to the first The weights of each hidden layer node; For input layer nodes to the first The bias of each hidden layer node; For the first The input of each input node.

[0134] Calculate the predicted risk value of derailment at the output layer. The formula is as follows:

[0135]

[0136] in, For the first The weights from each hidden layer node to the output layer node; This is the bias for the output layer nodes.

[0137] Subsequently, the mean squared error was used to calculate the error between the model's predicted values ​​and the true labels. The corresponding formula is as follows:

[0138]

[0139] in, The expected output of the derailment prediction model is referred to as the derailment risk label value in some embodiments; This is an error term, which is understandable and exists. .

[0140] Next, if the error is greater than or equal to the error convergence threshold, backpropagation is performed to update all model parameters, including weight parameters and bias parameters.

[0141] First, update the weight parameters from the hidden layer to the output layer. The corresponding formula is as follows:

[0142]

[0143] in, For the updated number The weight parameters from each hidden layer node to the output layer node; For the first time before the update The weight parameters from each hidden layer node to the output layer node.

[0144] Next, update the output layer bias parameters, using the following formula:

[0145]

[0146] in, The bias parameters for the updated output layer nodes; These are the bias parameters for the output layer nodes before the update.

[0147] Then update the weight parameters from the input layer to the hidden layer, using the following formula:

[0148]

[0149] in, For the updated number The input layer node to the first The weight parameters of each hidden layer node; For the previous version The input layer node to the first The weight parameters of each hidden layer node.

[0150] Next, update the bias parameters of the hidden layer, using the following formula:

[0151]

[0152] in, For the updated number The bias parameters of each hidden layer node; For the previous version The bias parameters of each hidden layer node.

[0153] Repeat the process of "input training samples and perform forward propagation" for all training samples until the error is less than the error convergence threshold, then stop training and obtain the trained derailment prediction model.

[0154] This application provides a method for preventing derailment of the wheelset of a tunneling machine. The method involves acquiring multiple sets of motion posture sample sets, which include historical motion posture data of the tunneling machine's wheelset and corresponding derailment risk label values. The historical motion posture data is input into a derailment prediction model to obtain a predicted derailment risk value. Then, using a preset loss function, the error between the predicted derailment risk value and the derailment risk label value is calculated. Subsequently, based on the error, the parameters of the derailment prediction model are updated, and the process of inputting historical motion posture data into the derailment prediction model to obtain the predicted derailment risk value is repeated until a preset convergence condition is met, resulting in a trained derailment prediction model. This technique uses the motion posture sample set of the tunneling machine's wheelset as training data to train the derailment prediction model, obtaining a derailment prediction model adapted to the millimeter-level low-speed operating conditions of the tunneling machine, thereby improving the accuracy of derailment prediction for the tunneling machine's wheelset.

[0155] Figure 4 A flowchart illustrating a method for preventing derailment of the trolley wheelset of a tunneling machine provided in this application. Figure 3 ,like Figure 4 As shown, this embodiment, based on any of the above embodiments, provides a detailed description of a method for preventing derailment of the trainset wheels of a tunneling machine. The method includes:

[0156] S401. The first sensor on the derailment detection trolley collects track image data in the tunneling direction.

[0157] S402. Perform multi-scale feature extraction on the orbital image data to obtain multi-scale features of the orbital image.

[0158] In this step, the orbital image data is first enhanced by preprocessing to improve its robustness under low light and dust interference. Then, multi-scale basic feature extraction is performed on the enhanced orbital image data to obtain an initial feature set at multiple scales. Next, the initial feature set at multiple scales is recalibrated to enhance the feature response of key locations such as orbital edges and wear areas through a spatial attention mechanism. Finally, the recalibrated initial feature set at multiple scales is fused to integrate detailed textures and global structural information at different scales to obtain complete multi-scale features of the orbital image.

[0159] S403. Perform feature enhancement processing on the multi-scale features of the track image to obtain the track recognition result.

[0160] In this step, the multi-scale features of the track image are first subjected to targeted feature enhancement processing to further enhance the feature discrimination of abnormal areas such as track wear and gauge deviation, and highlight the key features of track status. Then, the enhanced multi-scale features of the track image are sent to a classifier to classify and determine the track status, and the output track recognition results include the overall track status, abnormal location and abnormality degree, providing the basis for image-dimensional judgment for subsequent comprehensive derailment judgment.

[0161] In one possible implementation, the processing of orbital image data is as follows: Figure 5 As shown, the specific steps are as follows:

[0162] The first step is raw image input. The raw image data of the track in the tunneling direction collected by the derailment detection trolley is input into the track image recognition model.

[0163] The second step is data augmentation. The original image is preprocessed using operations such as brightness compensation, noise suppression, and image cropping to improve its robustness under low light and dust interference, providing a high-quality data foundation for subsequent feature extraction.

[0164] The third step involves depthwise separable convolutional layer processing. Multi-scale basic feature extraction is performed on the enhanced image. Depthwise separable convolutions are used to operate within different receptive fields, efficiently acquiring multi-scale initial features such as track texture, contours, and spatial structure.

[0165] The fourth step is spatial attention module processing. Feature recalibration is performed on the multi-scale basic features, and the feature response of key locations such as track edges, wear areas, and track breaks is enhanced through the spatial attention mechanism to optimize the expressive power of multi-scale features.

[0166] The fifth step is feature pyramid fusion processing. The recalibrated multi-scale features are fused to integrate detailed textures and global structural information at different scales, resulting in multi-scale fused features that contain complete orbital state features.

[0167] Step 6: Classifier Output. The fused multi-scale features are fed into the classifier to determine the orbital state. Based on the preset orbital state recognition rules, the classifier outputs state categories such as normal orbit, local anomaly, and no orbit.

[0168] Step 7: Output the recognition results. The final output includes the location and severity of the track anomaly, providing image-level information for subsequent comprehensive derailment determination.

[0169] S404. Collect the original multi-source motion attitude dataset of the derailment detection mechanical trolley by using multiple second sensors with redundant sensor layout on the derailment detection mechanical trolley.

[0170] S405. Perform weighted fusion processing on the original multi-source motion posture dataset to obtain the multi-source motion posture data of the derailment detection mechanical vehicle.

[0171] S406. Normalize the multi-source motion posture data to obtain normalized multi-source motion posture data.

[0172] S407. Input the normalized multi-source motion attitude data into the derailment prediction model to obtain the derailment risk value.

[0173] The derailment prediction model is trained using historical motion posture data of the trolley wheelsets.

[0174] S408. Based on the derailment risk value and track identification results, the driving status and driving hazard level of the trolley wheelset are determined through a preset derailment detection rule library.

[0175] S409. When the driving state shows signs of derailment, generate a derailment prevention signal corresponding to the driving hazard level.

[0176] In one possible implementation, preset derailment acceleration thresholds, preset derailment tilt angle thresholds, and preset derailment roll angle thresholds are determined through dynamic simulation analysis of the tunneling machine's wheel-to-wheel alignment and derailment tests of the derailment detection trolley. When the acceleration value of the derailment detection trolley exceeds the preset derailment acceleration threshold, or when the tilt angle value exceeds the preset derailment tilt angle threshold, or when the roll angle value exceeds the preset derailment roll angle threshold, the derailment detection trolley is determined to have derailed, and a derailment prevention signal is generated. The aforementioned acceleration, tilt angle, and roll angle values ​​are all key acquisition parameters included in the aforementioned multi-source motion attitude data.

[0177] S410: By uploading multi-source motion attitude data, track image data, and derailment detection results to the cloud platform, maintenance suggestions for the tunneling machine are obtained.

[0178] In this step, multi-source motion attitude data, track image data, and derailment detection results are uploaded to the cloud platform. Based on the multi-source motion attitude data, track image data, and derailment detection results, the cloud platform generates maintenance suggestions for the tunneling machine.

[0179] This application provides a method for preventing derailment of the wheelset of a tunneling machine. By using a first sensor on the derailment detection trolley, track image data in the tunneling direction is collected. The track image data is then subjected to multi-scale feature extraction and feature enhancement processing to obtain track recognition results. The combination of multi-scale feature extraction and feature enhancement processing improves the accuracy of identifying track anomalies, especially in complex working conditions, enabling early detection of potential risks and proactive prevention of derailment risks. Meanwhile, multiple second sensors with redundant sensor layouts on the derailment detection trolley collect the original multi-source motion attitude dataset of the trolley. The original multi-source motion attitude dataset is then weighted and fused to obtain the multi-source motion attitude data of the trolley. The combination of redundant sensor layout and weighted fusion processing solves the problems of sensor data bias and environmental noise interference, effectively improving data accuracy. The multi-source motion attitude data is then normalized to obtain normalized multi-source motion attitude data, eliminating the differences in dimensions and numerical scales between acceleration, tilt angle, and roll angle in the multi-source motion attitude data, effectively highlighting the subtle acceleration changes in low-speed scenarios. Finally, the normalized multi-source motion attitude data is input into the derailment prediction model to obtain the derailment risk value. Subsequently, based on the derailment risk value and track identification results, the driving status and driving hazard level of the tunnel boring machine's wheelset are determined through a pre-set derailment detection rule library. When the driving status shows signs of impending derailment, a derailment prevention signal corresponding to the driving hazard level is generated to control the tunnel boring machine to perform corresponding derailment prevention actions, thus achieving proactive identification and intervention of derailment risks. Furthermore, by uploading multi-source motion attitude data, track image data, and derailment detection results to a cloud platform, maintenance suggestions for the tunnel boring machine are obtained to reduce derailment risks through early maintenance. In summary, this embodiment ultimately achieves accurate prediction and effective early prevention of derailment accidents involving the tunnel boring machine's wheelset.

[0180] Based on any of the above embodiments, the following, in conjunction with Figure 6 and Figure 7 Taking a tunneling machine operating at a low speed of 0.4 m / s in an underground roadway as a scenario, this paper selects two typical derailment risk prevention procedures as specific examples to provide a detailed explanation of a derailment prevention method for the trolley wheels of a tunneling machine.

[0181] Figure 6 A specific example of a derailment prevention method for the trolley wheelset of a tunneling machine provided in this application. Figure 1 ,like Figure 6 As shown, taking a trackless operation scenario in which a tunneling machine is tunneling in the tunneling direction as an example, the specific implementation steps of a method for preventing derailment of the tunneling machine's trolley wheelset are as follows:

[0182] S601 uses an industrial array image sensor mounted on a derailment detection trolley to collect track image data in the tunneling direction in real time.

[0183] The industrial area array image sensor is the first sensor in the aforementioned embodiments.

[0184] S602. Perform image recognition on the track image data to obtain the track recognition result of "track missing ahead".

[0185] S603. Based on the track identification result of "missing track ahead", generate a first-level derailment signal.

[0186] In this step, by using a pre-set derailment detection rule base and based on the track identification result of "missing track ahead", the travel status of the trolley wheelset is determined to be a derailment precursor, and the travel hazard level is Level 1 derailment hazard. Based on the travel hazard level of Level 1 derailment hazard, a Level 1 derailment signal is generated.

[0187] S604. Send the first-level derailment signal to the execution control module of the tunneling machine to control the tunneling machine to stop tunneling and activate the audible and visual alarm.

[0188] It should be noted that while acquiring track image data along the tunneling direction, the original multi-source motion attitude dataset of the derailment detection trolley is also being acquired simultaneously, and derailment detection is performed based on this dataset. In this specific example, the track image data is sufficient to detect that the wheelset's travel status is a precursor to derailment and the hazard level is Level 1 derailment risk; therefore, the remaining derailment detection process is not described in detail. Figure 6 The processing steps S601-S604 shown in the embodiments do not constitute a specific limitation on a method for preventing derailment of a tunneling machine's trolley wheelset. In other embodiments of this application, a method for preventing derailment of a tunneling machine's trolley wheelset may include... Figure 6 The embodiments may include more or fewer steps; for example, a method for preventing derailment of the wheelset of a tunneling machine may include... Figure 6 Some steps in the embodiments, or, Figure 6 Some steps in the embodiments can be replaced by steps with the same function, or, Figure 6 Some steps in the embodiments can be broken down into multiple steps, etc.

[0189] Figure 7 A specific example of a derailment prevention method for the trolley wheelset of a tunneling machine provided in this application. Figure 2 ,like Figure 7 As shown, taking the abnormal motion posture of the tunneling machine's trolley wheelset as an example, the specific implementation steps of a method for preventing derailment of the tunneling machine's trolley wheelset are as follows:

[0190] S701 uses an industrial array image sensor mounted on a derailment detection trolley to collect track image data in real time along the tunneling direction.

[0191] The industrial area array image sensor is the first sensor in the aforementioned embodiments.

[0192] S702. Perform image recognition on the track image data to obtain the track recognition result of "slight track wear".

[0193] S703. Using two triaxial accelerometers, two tilt sensors, and two roll sensors with redundant sensor layout on the derailment detection trolley, the multi-source motion attitude data of the derailment detection trolley are obtained as follows: lateral acceleration 0.35 m / s², longitudinal tilt angle 1.9°, and roll angle 1.6°.

[0194] In this step, the original multi-source motion attitude dataset of the derailment detection trolley is first acquired through multiple sensors. Specifically, the main lateral acceleration is 0.36 m / s² and the redundant lateral acceleration is 0.33 m / s², the main longitudinal tilt angle is 2.0° and the redundant longitudinal tilt angle is 1.75°, and the main roll angle is 1.7° and the redundant roll angle is 1.4°. Then, based on the differentiated weights of 0.6 for the main sensor and 0.4 for the redundant sensor, the original multi-source motion attitude dataset is weighted and fused to obtain the multi-source motion attitude data of the derailment detection trolley, specifically the lateral acceleration of 0.35 m / s², the longitudinal tilt angle of 1.9°, and the roll angle of 1.6°.

[0195] S704, the normalized multi-source motion attitude data obtained are lateral acceleration 0.41, longitudinal tilt angle 0.44 and roll angle 0.40.

[0196] In this step, the min-max normalization method is used to map the multi-source motion attitude data to a unified numerical scale of 0 to 1, eliminating the differences in dimensions and numerical scales between acceleration, tilt angle, and roll angle, and obtaining normalized multi-source motion attitude data, specifically lateral acceleration 0.41, longitudinal tilt angle 0.44, and roll angle 0.40.

[0197] S705. Input the normalized multi-source motion attitude data into the derailment prediction model to obtain a derailment risk value of 90.

[0198] S706. Simultaneously input the derailment risk value and track identification result into the preset derailment detection rule library to determine the running status of the trolley wheelset as a derailment precursor and the running hazard level as a level two derailment hazard.

[0199] In this step, the derailment risk value and the track identification result are simultaneously input into the preset derailment detection rule library. Based on the preset judgment rules in the library, a comprehensive judgment is made to determine that the driving status of the tunneling machine wheel pair is a precursor to derailment and the driving hazard level is level two derailment hazard.

[0200] S707. Based on the secondary derailment hazard, a secondary derailment signal is generated.

[0201] S708: Send the secondary derailment signal to the execution control module of the tunneling machine to control the tunneling machine to trigger the yellow warning prompt on the human-machine interface.

[0202] Among them, the yellow warning on the human-machine interface prompts the user to remind the staff to manually stop the tunneling operation.

[0203] S709. Upload multi-source motion attitude data, track image data, and derailment detection results to the intelligent operation and maintenance cloud platform for mining tunneling machines.

[0204] It should be noted that, in Figure 7 The various processing steps S701-S709 shown in the embodiments do not constitute a specific limitation on a method for preventing derailment of a tunneling machine's trolley wheelset. In other embodiments of this application, a method for preventing derailment of a tunneling machine's trolley wheelset may include... Figure 7 The embodiments may include more or fewer steps; for example, a method for preventing derailment of the wheelset of a tunneling machine may include... Figure 7 Some steps in the embodiments, or, Figure 7 Some steps in the embodiments can be replaced by steps with the same function, or, Figure 7 Some steps in the embodiments can be broken down into multiple steps, etc.

[0205] Figure 8 A structural schematic diagram of a derailment prevention device for a tunneling machine's trolley wheelset provided in this application is shown below. Figure 8 As shown, the derailment prevention device 80 for the trolley wheelset of a tunneling machine provided in this embodiment includes:

[0206] The acquisition module 801 is used to acquire multi-source motion attitude data and track image data in the tunneling direction of the derailment detection trolley.

[0207] The data processing module 802 is used to normalize the multi-source motion posture data to obtain normalized multi-source motion posture data.

[0208] The derailment detection module 803 is used to input normalized multi-source motion attitude data into the derailment prediction model to obtain the derailment risk value. The derailment prediction model is trained using historical motion attitude data of the trolley wheelset. Based on the derailment risk value and track image data, the derailment detection result of the trolley wheelset is obtained. The derailment detection result includes the driving status and driving hazard level.

[0209] The generation module 804 is used to generate a derailment prevention signal corresponding to the driving hazard level when the driving state shows signs of derailment. The derailment prevention signal is used to control the tunneling machine to perform corresponding derailment prevention actions.

[0210] In one possible implementation, the derailment detection module 803 is used to train a derailment prediction model.

[0211] In one possible implementation, the derailment detection module 803 is specifically used for:

[0212] Obtain multiple sets of motion posture samples of the tunnel boring machine's trolley wheelsets; the motion posture sample sets include historical motion posture data and derailment risk label values ​​corresponding to the historical motion posture data.

[0213] By inputting historical motion posture data into the derailment prediction model, a derailment risk prediction value is obtained.

[0214] By using a preset loss function, the error between the predicted derailment risk value and the derailment risk label value is calculated.

[0215] Based on the error, the parameters of the derailment prediction model are updated; and the historical motion posture data is repeatedly input into the derailment prediction model to obtain the derailment risk prediction value, until the preset convergence condition is met, and the trained derailment prediction model is obtained.

[0216] In one possible implementation, the acquisition module 801 is specifically used for:

[0217] The first sensor on the derailment detection trolley collects track image data in the direction of tunneling.

[0218] The original multi-source motion attitude dataset of the derailment detection mechanical trolley is collected by multiple second sensors on the trolley; the multiple second sensors adopt a redundant sensor layout.

[0219] The original multi-source motion posture dataset is weighted and fused to obtain the multi-source motion posture data of the derailment detection mechanical vehicle.

[0220] In one possible implementation, the acquisition module 801 is used to obtain maintenance suggestions for the tunneling machine by uploading multi-source motion attitude data, track image data and derailment detection results to a cloud platform.

[0221] In one possible implementation, the derailment detection module 803 is specifically used for:

[0222] Normalized multi-source motion attitude data are input into the derailment prediction model to obtain the derailment risk value.

[0223] Image recognition is performed on the track image data to obtain the track recognition results.

[0224] Based on the derailment risk value and track identification results, the derailment detection results of the trolley wheelset are obtained.

[0225] In one possible implementation, the derailment detection module 803 is specifically used for:

[0226] Normalized multi-source motion attitude data are input into the derailment prediction model to obtain the derailment risk value.

[0227] Multi-scale feature extraction is performed on the orbital image data to obtain the multi-scale features of the orbital image.

[0228] Feature enhancement processing is performed on the multi-scale features of the track image to obtain the track recognition result.

[0229] Based on the derailment risk value and track identification results, the derailment detection results of the trolley wheelset are obtained.

[0230] In one possible implementation, the derailment detection module 803 is specifically used for:

[0231] Normalized multi-source motion attitude data are input into the derailment prediction model to obtain the derailment risk value.

[0232] Multi-scale feature extraction is performed on the orbital image data to obtain the multi-scale features of the orbital image.

[0233] Feature enhancement processing is performed on the multi-scale features of the track image to obtain the track recognition result.

[0234] Based on the derailment risk value and track identification results, the running status and running hazard level of the trolley wheelset are determined by using a pre-set derailment detection rule library.

[0235] The driving status and driving hazard level are determined as the derailment detection results.

[0236] In one possible implementation, the derailment prevention signal includes at least a primary derailment signal and a secondary derailment signal; the primary derailment signal is used to control the tunneling machine to stop tunneling and activate a primary alarm; the secondary derailment signal is used to control the tunneling machine to activate a secondary alarm.

[0237] This embodiment provides a derailment prevention device for the wheelset of a tunneling machine, which can perform the method provided in the above-described method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0238] Figure 9 A schematic diagram of a derailment prevention system for a tunneling machine's trolley wheelset is provided in this application, as shown below. Figure 9 As shown, the derailment prevention system 90 for the wheelset of a tunneling machine provided in this embodiment includes a derailment detection mechanical trolley 901 and a tunneling machine main control device 902; the derailment detection mechanical trolley 901 includes a multi-modal sensor 9011 and a transmitting module 9012; the tunneling machine main control device 902 includes a receiving module 9021, a derailment prevention device 9022 and an execution control module 9023; the derailment detection mechanical trolley 901 is rigidly connected to the wheelset of the tunneling machine.

[0239] In the specific implementation process, the multi-modal sensor 9011 is used to collect multi-source motion attitude data and track image data in the tunneling direction from the derailment detection trolley 901. The transmitting module 9012 transmits the multi-source motion attitude data and track image data to the receiving module 9021. The receiving module 9021 receives the multi-source motion attitude data and track image data and transmits them to the derailment prevention device 9022. The derailment prevention device 9022 generates a derailment prevention signal based on the multi-source motion attitude data and track image data, causing the device to execute the aforementioned method. The execution control module 9023 receives the derailment prevention signal and controls the tunneling machine to perform corresponding derailment prevention actions based on the signal.

[0240] The specific implementation process of the derailment prevention device 9022 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so it will not be repeated here.

[0241] Figure 10 A schematic diagram of the structure of the electronic device provided in this application. Figure 10 As shown, the electronic device 100 provided in this embodiment includes at least one processor 1001 and a memory 1002. Optionally, the device 100 further includes a communication component 1003. The processor 1001, memory 1002, and communication component 1003 are connected via a bus 1004.

[0242] In a specific implementation, at least one processor 1001 executes computer execution instructions stored in memory 1002, causing at least one processor 1001 to perform the above-described method.

[0243] The specific implementation process of processor 1001 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0244] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0245] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0246] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0247] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0248] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0249] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0250] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0251] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0252] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0253] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0254] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0255] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0256] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for preventing derailment of the wheelset of a tunneling machine, characterized in that, The trolley wheelset is rigidly connected to a derailment detection trolley. The trolley wheelset and the derailment detection trolley move synchronously along the track in the tunneling direction together with the tunneling machine. The method includes: Acquire the multi-source motion attitude data of the derailment detection trolley and the track image data in the tunneling direction; The multi-source motion posture data is normalized to obtain normalized multi-source motion posture data. The normalized multi-source motion attitude data is input into the derailment prediction model to obtain the derailment risk value; the derailment prediction model is trained using the historical motion attitude data of the trolley wheelset. Based on the derailment risk value and the track image data, the derailment detection result of the trolley wheelset is obtained; the derailment detection result includes the driving status and driving hazard level; When the driving state shows signs of derailment, a derailment prevention signal corresponding to the driving hazard level is generated according to the driving hazard level; the derailment prevention signal is used to control the tunneling machine to perform corresponding derailment prevention actions.

2. The method according to claim 1, characterized in that, The training process of the derailment prediction model includes: Obtain multiple sets of motion posture sample sets; the motion posture sample sets include historical motion posture data of the trolley wheels of the tunneling machine and derailment risk label values ​​corresponding to the historical motion posture data; The historical motion posture data is input into the derailment prediction model to obtain the derailment risk prediction value; The error between the predicted derailment risk value and the derailment risk label value is calculated by using a preset loss function. Based on the error, the parameters of the derailment prediction model are updated; and the process of inputting the historical motion posture data into the derailment prediction model to obtain the derailment risk prediction value is repeated until the preset convergence condition is met, and the trained derailment prediction model is obtained.

3. The method according to claim 1, characterized in that, The acquisition of the multi-source motion attitude data of the derailment detection trolley and the track image data in the tunneling direction includes: The first sensor on the derailment detection trolley collects track image data in the tunneling direction; The derailment detection trolley uses multiple second sensors to collect the original multi-source motion attitude dataset of the trolley; the multiple second sensors adopt a redundant sensor layout. The original multi-source motion posture dataset is subjected to weighted fusion processing to obtain the multi-source motion posture data of the derailment detection mechanical vehicle.

4. The method according to claim 1, characterized in that, The method further includes: By uploading the multi-source motion attitude data, the track image data, and the derailment detection results to the cloud platform, maintenance recommendations for the tunneling machine are obtained.

5. The method according to claim 1, characterized in that, The step of obtaining the derailment detection result of the trolley wheelset based on the derailment risk value and the track image data includes: Image recognition is performed on the track image data to obtain track recognition results; Based on the derailment risk value and the track identification result, the derailment detection result of the trolley wheelset is obtained.

6. The method according to claim 5, characterized in that, The process of performing image recognition on the orbital image data to obtain orbital recognition results includes: Multi-scale feature extraction is performed on the orbital image data to obtain multi-scale features of the orbital image; The track image is subjected to multi-scale feature enhancement processing to obtain the track recognition result.

7. The method according to claim 5, characterized in that, The step of obtaining the derailment detection result of the trolley wheelset based on the derailment risk value and the track identification result includes: Based on the derailment risk value and the track identification result, the driving status and driving hazard level of the trolley wheelset are determined by using a preset derailment detection rule library; The driving status and the driving hazard level are determined as the derailment detection result.

8. The method according to claim 1, characterized in that, The derailment prevention signal includes at least a primary derailment signal and a secondary derailment signal; the primary derailment signal is used to control the tunneling machine to stop tunneling and activate a primary alarm; the secondary derailment signal is used to control the tunneling machine to activate a secondary alarm.

9. A derailment prevention device for the wheelset of a tunneling machine, characterized in that, include: The acquisition module is used to acquire multi-source motion attitude data of the derailment detection mechanical trolley and track image data in the tunneling direction; The data processing module is used to normalize the multi-source motion posture data to obtain normalized multi-source motion posture data. The derailment detection module is used to input the normalized multi-source motion attitude data into the derailment prediction model to obtain a derailment risk value. The derailment prediction model is trained using the historical motion attitude data of the trolley wheelset. Based on the derailment risk value and the track image data, the derailment detection result of the trolley wheelset is obtained. The derailment detection result includes the driving status and driving hazard level. The generation module is used to generate a derailment prevention signal corresponding to the driving hazard level when the driving state shows signs of derailment. The derailment prevention signal is used to control the tunneling machine to perform corresponding derailment prevention actions.

10. A derailment prevention system for the wheelset of a tunneling machine, characterized in that, The system includes a derailment detection trolley and a tunneling machine main control unit; the derailment detection trolley includes a multi-modal sensor and a transmitting module; the tunneling machine main control unit includes a receiving module, a derailment prevention device, and an execution control module; the derailment detection trolley is rigidly connected to the wheelset of the tunneling machine. The multimodal sensor is used to collect multi-source motion attitude data of the derailment detection trolley and track image data in the tunneling direction; The transmitting module is used to transmit the multi-source motion attitude data and the orbital image data to the receiving module; The receiving module is used to receive the multi-source motion attitude data and the track image data, and transmit them to the derailment prevention device; The derailment prevention device is used to generate a derailment prevention signal based on the multi-source motion attitude data and the track image data, so that the derailment prevention device performs the method as described in any one of claims 1 to 8; The execution control module is used to receive the derailment prevention signal and control the tunneling machine to perform corresponding derailment prevention actions according to the derailment prevention signal.

Images