High-speed railway intelligent track inspection operation robot running monitoring method and system

Abstract

The present application relates to the technical field of railway track inspection, and particularly relates to a high-speed railway intelligent track inspection operation robot operation monitoring method and system, the present application combines track turnout state data and seismic influence data to analyze abnormal conditions of track passage detection conditions, and according to the analysis result, analyzes the threat degree of the execution environment of the robot post-earthquake track inspection operation task; combining the robot's own operation data and track turnout state data, analyzes the abnormal conditions of the robot post-earthquake track inspection operation task's own operation control; based on the execution environment threat degree of the robot post-earthquake track inspection operation task, the abnormal conditions of the robot post-earthquake track inspection operation task's own operation control, the track inspection operation risk of the robot post-earthquake track inspection operation task is evaluated; according to the evaluation result, the robot emergency braking early warning is realized, so as to guarantee the operation safety of the inspection robot in the earthquake destructive environment, and the intelligent level of the high-speed railway post-earthquake track inspection operation is significantly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of railway track inspection, and in particular to a high-speed railway intelligent track inspection operation robot running monitoring method and system. BACKGROUND

[0002] As the artery of modern transportation, the track system of high-speed railway, especially the complex structure of turnout area, is the key infrastructure to ensure the safe, smooth and efficient operation of trains. In the long-term service, the track of high-speed railway will not only be worn and deformed due to the cyclic load of trains and the influence of natural environment, but also its structural integrity will face extreme threats in high-risk areas located in earthquake belts. Strong earthquakes and their frequent aftershocks can cause sudden deterioration of track geometry, jamming or failure of core locking components of turnouts, and may even cause landslide intrusion into the limit. If the post-disaster safety state assessment and passing condition determination cannot be carried out quickly and accurately on the line, especially in the critical turnout area, it may lead to damage of detection equipment and delay of maintenance. To cope with the challenges of track inspection, using robots to replace traditional manual inspection has become an explicit trend. In recent years, the development of intelligent inspection robots and track intelligent diagnosis systems has successfully realized the capture of millimeter-level changes in track geometry and the automatic identification of partial faults, significantly improving the detection efficiency and objectivity.

[0003] However, the existing technology still has significant deficiencies when applied to track inspection in the turnout area of post-earthquake high-risk scenarios. First, the existing track inspection robot operation monitoring method is usually designed based on the conventional environment, without deep correlation and comprehensive analysis of the track turnout geometric state data collected by the robot and the real-time seismic influence data. For example, when there is a track lateral displacement in the turnout area, if it is not considered whether the displacement is caused by long-term creep or a strong aftershock impact just occurred, the instantaneous stability and subsequent risk of the structure cannot be accurately evaluated, leading to misjudgment of the abnormal situation of passing conditions. Second, the existing technology also lacks real-time and quantitative risk assessment of the running stability of the robot itself on the distorted track after the earthquake. Local large deformation, obstacles and other factors that may exist on the post-earthquake track can seriously affect the passability of the robot, which may cause the robot to slip, overturn or collide. Traditional methods often only focus on the track state itself and ignore the operational risks faced by the robot performing tasks, which leads to inaccurate judgment of the overall risk of post-earthquake track inspection operation and limits its reliable application in truly high-risk and complex emergency scenarios. SUMMARY

[0004] In order to overcome the defects and deficiencies existing in the prior art, the application provides a high-speed railway intelligent track inspection operation robot operation monitoring method and system, which synchronously collects track, earthquake and robot body data in post-earthquake operation, constructs and applies a gradient boosting tree and other machine learning models, respectively quantitatively analyzes track passage condition abnormality, execution environment threat degree and robot body operation abnormality of the current operation position, and then evaluates the final track inspection operation risk, thereby realizing emergency braking warning, and significantly improving the intelligent level and overall safety redundancy of high-speed railway post-earthquake track inspection operation.

[0005] In order to achieve the above purpose, the application adopts the following technical solutions:

[0006] In the first aspect, the application embodiment provides a high-speed railway intelligent track inspection operation robot operation monitoring method, which comprises the following steps:

[0007] S1, in the process of robot executing post-earthquake track inspection operation task, synchronously collecting the robot's body operation data, track turnout state data and earthquake influence data;

[0008] S2, combining the track turnout state data and the earthquake influence data, analyzing the track passage detection condition abnormality, and according to the track passage detection condition abnormality analysis result, analyzing the execution environment threat degree of the robot post-earthquake track inspection operation task;

[0009] S3, combining the robot's body operation data and the track turnout state data, analyzing the body operation control abnormality of the robot post-earthquake track inspection operation task;

[0010] S4, based on the execution environment threat degree analysis result and the body operation control abnormality analysis result of the robot post-earthquake track inspection operation task, evaluating the track inspection operation risk of the robot post-earthquake track inspection operation task;

[0011] S5, based on the track inspection operation risk evaluation result of the robot post-earthquake track inspection operation task, performing robot emergency braking warning.

[0012] According to the above scheme, in step S2, the track passage detection condition abnormality is analyzed by combining the track turnout state data and the earthquake influence data, which comprises the following specific steps:

[0013] S21, extract the track state parameters of the robot operation position of the multiple historical post-earthquake track inspection operation task records from the track turnout state data, wherein the track state parameters include: track lateral displacement value, track vertical settlement value and turnout close clearance value; extract the peak ground acceleration value and the aftershock occurrence frequency of the robot operation position in the corresponding historical track inspection period of the multiple historical post-earthquake track inspection operation task records from the earthquake influence data; at the same time, extract the passing condition abnormal flag of the robot operation position in the corresponding historical track inspection period recorded after the multiple historical post-earthquake track inspection operation task is completed; based on the track lateral displacement value, the track vertical settlement value and the turnout close clearance value of the robot operation position of the multiple historical post-earthquake track inspection operation task records, and the peak ground acceleration value, the aftershock occurrence frequency and the passing condition abnormal flag of the robot operation position in the corresponding historical track inspection period, a passing condition abnormality analysis data set is constructed;

[0014] S22, statistics of the number of records of each track state parameter exceeding the corresponding preset safety threshold in different peak ground acceleration value intervals accounts for the percentage of the total number of records in the current peak ground acceleration value interval, and the corresponding percentage is defined as the reference abnormal probability of the corresponding track state parameter under different seismic levels;

[0015] S23, a passing condition abnormality analysis model is constructed and trained based on a logistic regression model; the peak ground acceleration value, the aftershock occurrence frequency of each historical post-earthquake track inspection operation task record in the passing condition abnormality analysis data set, and the product of the track lateral displacement value and the corresponding reference abnormal probability, the product of the track vertical settlement value and the corresponding reference abnormal probability, and the product of the turnout close clearance value and the corresponding reference abnormal probability are used as input features of the model, and the passing condition abnormal flag is used as the training target of the model for model training, to obtain a trained passing condition abnormality analysis model;

[0016] S24, the peak ground acceleration value, the aftershock occurrence frequency, the track lateral displacement value, the track vertical settlement value and the turnout close clearance value of the robot current operation position are obtained in real time; the reference abnormal probability of each track state parameter is queried according to the peak ground acceleration value interval of the peak ground acceleration value of the robot current operation position; the peak ground acceleration value, the aftershock occurrence frequency of the robot current operation position, and the product of the corresponding track lateral displacement value and the corresponding reference abnormal probability, the product of the corresponding track vertical settlement value and the corresponding reference abnormal probability, and the product of the corresponding turnout close clearance value and the corresponding reference abnormal probability are input into the trained passing condition abnormality analysis model for analysis, and the probability value output by the passing condition abnormality analysis model is used as the track passing condition abnormality of the robot current operation position in the robot post-earthquake track inspection operation task.

[0017] According to the above scheme, in step S2, the execution environment threat degree of the robot post-earthquake track inspection operation task is analyzed according to the track passage detection condition abnormal situation analysis result in step S2; including the following specific steps:

[0018] S25, extracting the line obstacle projection area of the section where the robot operation position of the multiple historical post-earthquake track inspection operation task record is located and the turnout locking device state code from the turnout state data, and associating and extracting the track passage detection condition abnormal situation of the corresponding robot operation position; at the same time, the environmental threat mark of the robot operation position of the multiple historical post-earthquake track inspection operation task record is extracted; the line obstacle projection area of the section where the robot operation position of the multiple historical post-earthquake track inspection operation task record is located and the turnout locking device state code, the track passage detection condition abnormal situation of the corresponding robot operation position, and the environmental threat mark are used as an environmental threat degree analysis data set;

[0019] S26, constructing and training an environmental threat degree analysis model based on a gradient boosting decision tree model; the track passage detection condition abnormal situation of the robot operation position of the multiple historical post-earthquake track inspection operation task record in the environmental threat degree analysis data set, the corresponding line obstacle projection area, and the turnout locking device state code are used as input features, and the environmental threat mark is used as an output target for model training, to obtain a trained environmental threat degree analysis model;

[0020] S27, real-time acquiring the track passage detection condition abnormal situation of the robot current operation position and the line obstacle projection area and the turnout locking device state code of the section where the robot current operation position is located, inputting the trained environmental threat degree analysis model for analysis, and using the probability value output by the environmental threat degree analysis model as the execution environment threat degree of the robot current operation position in the robot post-earthquake track inspection operation task.

[0021] According to the above scheme, in step S3, the body operation control abnormal situation of the robot post-earthquake track inspection operation task is analyzed in combination with the robot body operation data and the track turnout state data; including the following specific steps:

[0022] S31, extract the three-dimensional coordinates of the robot operation position and the running direction from the body running data of the robot, and extract the nearest track turnout position passed by the robot operation position according to the three-dimensional coordinates of the robot operation position and the running direction; based on the robot operation position and the nearest track turnout position passed, real-time collection of the robot transverse displacement deviation value sequence, the robot pitch attitude angle sequence, the robot roll attitude angle sequence, the driving motor current fluctuation value sequence in the robot running process on the track section between the robot operation position and the nearest track turnout position passed; at the same time, the track curvature and track slope of the track section between the robot operation position and the nearest track turnout position passed are obtained from the track turnout state data;

[0023] S32, based on the robot transverse displacement deviation value sequence, the robot pitch attitude angle sequence, the robot roll attitude angle sequence, and the driving motor current fluctuation value sequence, the robot transverse displacement variation degree, the robot pitch attitude angle variation degree, the robot roll attitude angle variation degree, and the robot driving motor current fluctuation variation degree on the track section between the robot operation position and the nearest track turnout position passed are calculated respectively;

[0024] S33, the track curvature and track slope on the track section between the robot and the nearest track turnout position passed corresponding to any operation position are obtained in multiple historical post-earthquake track inspection operation tasks, and the robot transverse displacement variation degree, the robot pitch attitude angle variation degree, the robot roll attitude angle variation degree, and the robot driving motor current fluctuation variation degree are obtained on the corresponding track section; at the same time, the body operation abnormal flag of the robot passing through the corresponding track section in the corresponding historical post-earthquake track inspection operation task is obtained;

[0025] S34, the track curvature, track slope, robot transverse displacement variation degree, robot pitch attitude angle variation degree, robot roll attitude angle variation degree, robot driving motor current fluctuation variation degree and robot body operation abnormal flag passing through the corresponding track section in the historical post-earthquake track inspection operation task are taken as an abnormal prediction data set, and the abnormal prediction data set is randomly divided into an abnormal prediction training set and an abnormal prediction verification set.

[0026] According to the above scheme, the body operation control abnormality of the robot post-earthquake track inspection operation task is analyzed in step S3 by combining the body running data of the robot and the track turnout state data; further comprising the following specific steps:

[0027] S35, construct a gradient boosting tree model, based on the abnormal prediction training set, the track curvature, track slope, robot lateral displacement variation degree, robot pitch attitude angle variation degree, robot roll attitude angle variation degree, and robot drive motor current fluctuation variation degree obtained by statistics on the corresponding track segment in the historical post-earthquake track inspection operation task are taken as the input features of the gradient boosting tree model, and the robot body operation abnormal flag through the corresponding track segment is taken as the output target of the model, the gradient boosting tree model is trained, and an initial abnormal prediction model is obtained; the initial abnormal prediction model is verified through the abnormal prediction verification set, and the corresponding initial abnormal prediction model whose prediction accuracy of the model on the abnormal prediction verification set reaches the preset model accuracy threshold is taken as the body operation abnormal analysis model;

[0028] S36, real-time acquisition of track curvature, track slope, and robot lateral displacement variation degree, robot pitch attitude angle variation degree, robot roll attitude angle variation degree, and robot drive motor current fluctuation variation degree on the track segment between the current operation position of the robot and the corresponding nearest track turnout position that has passed, input into the body operation abnormal analysis model, and output to obtain a prediction probability value as the body operation control abnormality of the current operation position of the robot in the post-earthquake track inspection operation task of the robot.

[0029] According to the above scheme, the track inspection operation risk of the post-earthquake track inspection operation task of the robot is evaluated based on the execution environment threat degree analysis result and the body operation control abnormality analysis result of the robot in step S4; including the following specific steps:

[0030] S41, extracting the execution environment threat degree and the body operation control abnormality of the current operation position of the robot in the post-earthquake track inspection operation task of the robot;

[0031] S42, weighted sum of the execution environment threat degree and the body operation control abnormality of the current operation position of the robot in the post-earthquake track inspection operation task of the robot, to obtain the track inspection operation risk of the current operation position of the robot in the post-earthquake track inspection operation task of the robot.

[0032] According to the above scheme, the robot emergency braking warning is performed based on the track inspection operation risk evaluation result of the post-earthquake track inspection operation task of the robot in step S5; specifically including:

[0033] S51, obtaining the track inspection operation risk evaluation result of the current operation position of the robot in the post-earthquake track inspection operation task of the robot;

[0034] S52, a preset track operation risk threshold, when the track inspection operation risk assessment result of the current operation position of the robot in the post-earthquake track inspection operation task of the robot is greater than the track operation risk threshold, the robot emergency braking warning is performed; when the track inspection operation risk assessment result of the current operation position of the robot in the post-earthquake track inspection operation task of the robot is less than or equal to the track operation risk threshold, the high-speed railway track is continuously inspected.

[0035] In a second aspect, the embodiments of the present application also provide a high-speed railway intelligent track inspection operation robot operation monitoring system, comprising:

[0036] A data acquisition module is configured to synchronously acquire the body operation data of the robot, the track turnout state data and the earthquake influence data during the execution of the post-earthquake track inspection operation task by the robot.

[0037] An execution environment monitoring module is configured to analyze the track passing detection condition abnormality in combination with the track turnout state data and the earthquake influence data, and analyze the execution environment threat degree of the post-earthquake track inspection operation task of the robot according to the track passing detection condition abnormality analysis result.

[0038] A control abnormality detection module is configured to analyze the body operation control abnormality of the post-earthquake track inspection operation task of the robot in combination with the body operation data of the robot and the track turnout state data.

[0039] A track inspection operation monitoring module is configured to evaluate the track inspection operation risk of the post-earthquake track inspection operation task of the robot based on the execution environment threat degree analysis result of the post-earthquake track inspection operation task of the robot and the body operation control abnormality analysis result.

[0040] A braking warning module is configured to perform the robot emergency braking warning based on the track inspection operation risk evaluation result of the post-earthquake track inspection operation task of the robot.

[0041] Compared with the prior art, the present application has the following advantages and beneficial effects:

[0042] 1. The present application realizes the probabilistic and quantitative evaluation of the track passing detection condition abnormality by associating and modeling the post-earthquake track turnout state data and the earthquake influence data, and improves the scientificity and accuracy of the safety evaluation in the initial emergency inspection;

[0043] 2. The present application constructs the body operation control abnormality analysis model by combining the robot body operation data and the track state data, can predict the slipping and instability risks of the robot itself in real time, and ensures the operation safety and equipment integrity of the inspection robot in the earthquake destructive environment;

[0044] 3、The application combines the execution environment threat degree and the ontology operation abnormal situation to perform comprehensive risk assessment on the track inspection operation task, and significantly improves the intelligent level and overall safety redundancy of the post-earthquake track inspection operation of the high-speed railway. BRIEF DESCRIPTION OF DRAWINGS

[0045] Other features, objects and advantages of the application will become more apparent from the following detailed description of non-limiting embodiments, made with reference to the accompanying drawings:

[0046] Figure 1 The figure is a schematic diagram of the overall process of the high-speed railway intelligent track inspection operation robot operation monitoring method of the application;

[0047] Figure 2 The figure is a work flow chart of step S2 in the high-speed railway intelligent track inspection operation robot operation monitoring method of the application;

[0048] Figure 3 The figure is a structural schematic diagram of the high-speed railway intelligent track inspection operation robot operation monitoring system of the application;

[0049] Figure 4 The figure is a work flow chart of step S3 in the high-speed railway intelligent track inspection operation robot operation monitoring method of the application. DETAILED DESCRIPTION

[0050] The technical solutions of the application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the application and the specific features in the embodiments are detailed descriptions of the technical solutions of the application, rather than limitations of the technical solutions of the application. In the case of no conflict, the technical features in the embodiments of the application and the embodiments can be combined with each other.

[0051] Embodiment 1

[0052] As shown in the figure, the embodiment provides a high-speed railway intelligent track inspection operation robot operation monitoring method, which specifically includes the following steps: Figure 1

[0053] S1, during the execution of the post-earthquake track inspection operation task by the robot, the ontology operation data of the robot, the track turnout state data and the earthquake influence data are synchronously collected;

[0054] S2, the track passing detection condition abnormal situation is analyzed in combination with the track turnout state data and the earthquake influence data, and the execution environment threat degree of the post-earthquake track inspection operation task of the robot is analyzed according to the track passing detection condition abnormal situation analysis result;

[0055] S3, the ontology operation control abnormal situation of the post-earthquake track inspection operation task of the robot is analyzed in combination with the ontology operation data of the robot and the track turnout state data.​

[0056] S4, based on the execution environment threat degree analysis result of the robot post-earthquake track inspection operation task and the ontology operation control abnormal situation analysis result, track inspection operation risk of the robot post-earthquake track inspection operation task is evaluated;

[0057] S5, based on the track inspection operation risk evaluation result of the robot post-earthquake track inspection operation task, robot emergency braking warning is carried out.

[0058] In this embodiment, as shown in Figure 2 The track passing detection condition abnormal situation is analyzed in step S2 in combination with the track turnout state data and the earthquake influence data; including the following specific steps:

[0059] S21, extract the track state parameters of the robot operation position of the multiple historical post-earthquake track inspection operation task records from the track turnout state data, wherein the track state parameters include: track lateral displacement value, track vertical settlement value and turnout close clearance value; extract the peak ground acceleration value and aftershock occurrence frequency of the robot operation position in the corresponding historical track inspection period from the multiple historical post-earthquake track inspection operation task records from the earthquake influence data; at the same time, extract the passing condition abnormal flag of the robot operation position in the corresponding historical track inspection period recorded after the multiple historical post-earthquake track inspection operation task is completed; wherein the passing condition abnormal flag is determined according to whether the track lateral displacement value, the track vertical settlement value and the turnout close clearance value of the corresponding historical track inspection period exceed the corresponding preset safety threshold value, if any track state parameter exceeds, the flag is 1, otherwise it is 0; based on the track lateral displacement value, the track vertical settlement value and the turnout close clearance value of the robot operation position of the multiple historical post-earthquake track inspection operation task records, and the peak ground acceleration value, the aftershock occurrence frequency and the passing condition abnormal flag of the robot operation position in the corresponding historical track inspection period, a passing condition abnormality analysis data set is constructed; it should be noted that the core of the embodiment is to construct an analysis data set that can accurately reflect the abnormality of the passing condition of the post-earthquake track turnout area. Unlike ordinary railway track inspection, which only needs to focus on the long-term slow changes of track geometric dimensions, the safety state of the post-earthquake turnout area of high-speed railway is a dynamic and high-risk variable affected by the duration of ground motion. In order to quantify this risk, the embodiment integrates multiple source heterogeneous data; specifically, from the massive track turnout state data of historical track inspection tasks, the track lateral displacement value, the track vertical settlement value and the turnout close clearance value recorded at each positioning point of each task are extracted. These parameters are direct physical representations of track structure integrity and turnout functionality, and their measurement relies on high-precision sensors installed on robots or fixed monitoring points, such as total station, static level and laser ranging sensor, which can obtain millimeter-level change data in real time and objectively. At the same time, from the associated earthquake influence data, the peak ground acceleration value and aftershock occurrence frequency of the corresponding time and same geographical position are extracted. The peak ground acceleration value data is derived from the high-sensitivity seismic accelerometer network laid along the railway, and the aftershock frequency is obtained by accessing the real-time data stream of the national or regional earthquake monitoring station.In the present embodiment, the most critical step is to generate the passage condition anomaly flag, which is not a subjective judgment, but is based on strict engineering safety specifications: after each historical task is completed, automatically backtrack whether the above three track state parameters of each positioning point in this task period exceed the preset safety threshold defined by the railway maintenance regulations; in the present embodiment, the preset safety threshold is set differently according to the different design standards of high-speed railway and ordinary railway, and the higher sensitivity of turnout area compared to ordinary track section; for example, for high-speed railway turnout, the threshold of close clearance is much smaller than ordinary track; the present embodiment further spatiotemporally aligns and correlates the track state parameters of all positioning points, the corresponding seismic motion parameters and the automatically determined anomaly flags in all historical tasks, to form the passage condition anomaly analysis data set required for machine learning, to correlate the accidental, high-intensity seismic disturbance with the micro response of the track structure, and to lay a solid, traceable data foundation for subsequent risk probabilistic prediction.

[0060] S22, statistics in different peak ground acceleration value interval, each track state parameter exceeds the corresponding preset safety threshold record number accounts for the total record number in the current peak ground acceleration value interval percentage, the corresponding percentage is defined as the corresponding track state parameter in different ground motion level benchmark abnormal probability; the embodiment is used for quantifying the prior probability of each track state parameter abnormality under different earthquake intensity, that is, the benchmark abnormal probability. In the track inspection of ordinary railway, parameter overrun is often regarded as an independent event for analysis. However, in the post-earthquake scene, ground motion intensity is the core driving factor of track state deterioration, and its influence is nonlinear. In order to objectively reveal this correlation, the data set constructed in S21 is finely analyzed; in the specific operation of the embodiment, first, the peak ground acceleration value range in the whole data set is divided into several continuous intervals (for example, 0-0.05g, 0.05g-0.1g, 0.1g-0.2g, etc.); then, for each interval, statistics of all data records falling in this interval are made; for each record, check whether the track lateral displacement value, track vertical settlement value and switch tightness gap value exceed the respective preset safety threshold. Then, calculate the percentage of the number of records in which a certain parameter (such as track lateral displacement) appears overrun in a certain peak ground acceleration value interval, accounts for the total number of records in the peak ground acceleration value interval as the benchmark abnormal probability of the parameter under this specific peak ground acceleration value level; in the embodiment, the mathematical meaning of the benchmark abnormal probability is the empirical possibility of track abnormality of a certain type when encountering a certain intensity ground motion in historical data; for example, the embodiment can find that when the peak ground acceleration value is in the 0.1g-0.2g interval, the benchmark probability of switch tightness gap overrun is as high as 40%, while the probability of lateral displacement overrun is only 15%. Based on the above, the role of the embodiment is as follows:

[0061] First, the embodiment realizes the change of risk assessment from binary judgment of whether to exceed the threshold to continuous probability thinking of how much probability to exceed the threshold, which is more in line with the uncertainty nature of post-earthquake risk;

[0062] Second, the embodiment gives each track state parameter a dynamic weight, which changes with the intensity of ground motion in real time, so that the model can understand that under strong earthquake, tightness gap failure is a more alarming risk than normal settlement, and under weak earthquake, it is the opposite.

[0063] S23, based on the logistic regression model is constructed and trained to pass the condition abnormal analysis model; the peak ground acceleration value of each historical post-earthquake track inspection operation task record in the abnormal analysis data set, the aftershock occurrence frequency, and the product of the track lateral displacement value and the corresponding reference abnormal probability, the product of the track vertical settlement value and the corresponding reference abnormal probability, and the product of the turnout close clearance value and the corresponding reference abnormal probability are used as the input features of the model, and the abnormal flag of the passing condition is used as the training target of the model for model training to obtain the trained passing condition abnormal analysis model; the core of the embodiment is to construct and train a passing condition abnormal analysis model that can comprehensively consider the earthquake environment and track state and output an abnormal probability; considering the complex nonlinear relationship between the input features and the output target, and the need to output an easily interpretable probability value, the embodiment selects a logistic regression model as the basis and enhances the input features; the specific process of model construction is as follows:

[0064] From the data set constructed in S21, the peak ground acceleration value (continuous value) and the aftershock occurrence frequency (continuous value) of each data record are used as basic environmental features;

[0065] The measured track lateral displacement value, track vertical settlement value, and turnout close clearance value in each record are multiplied by the reference abnormal probability of each parameter corresponding to the interval of the peak ground acceleration value of the record calculated in step S22; the significance of the multiplication operation used in the embodiment is that a risk-weighted value is used in the embodiment; for example, a smaller displacement measured under a strong earthquake (high peak ground acceleration value) may have a weighted feature value larger than a larger displacement measured under a weak earthquake (low peak ground acceleration value) because the reference probability of displacement deterioration under a strong earthquake is higher; this is equivalent to having the model learn the risk weight of the parameter under different earthquake environments;

[0066] The three weighted features, together with the two basic environmental features, form a five-dimensional input feature vector; the training target of the model is the abnormal flag of the passing condition (0 or 1) defined in S21;

[0067] Use 70% of the sample data set as the training set and input the logistic regression model for training; in the embodiment, the logistic regression model maps the linear combination of the input features to a continuous value between 0 and 1 through the logit function, which represents the predicted probability of the sample being judged as abnormal (the passing condition abnormal flag is 1); the model learns the probability mapping relationship between the input features and the passing condition abnormal flag through algorithms such as maximum likelihood estimation, and finally obtains a set of optimal model coefficients, so that the predicted probability output by the model is as close as possible to the true binary flag distribution;

[0068] After the training is completed, the remaining 30% of the validation set is used to evaluate the model performance, calculate the prediction accuracy, and determine the optimal probability classification threshold through the ROC curve. When the prediction accuracy of the model on the validation set reaches the preset standard (the preset standard in this embodiment is 90% by default), the model training is completed;

[0069] S24, the peak ground acceleration value, the aftershock occurrence frequency, the track lateral displacement value, the track vertical settlement value, and the turnout close clearance value of the current working position of the robot are obtained in real time; according to the peak ground acceleration value of the current working position of the robot, the corresponding reference abnormal probability of each track state parameter is queried; the peak ground acceleration value, the aftershock occurrence frequency, and the corresponding track lateral displacement value of the current working position of the robot are input into the trained passing condition abnormal analysis model for analysis, and the probability value output by the passing condition abnormal analysis model is used as the track passing detection condition abnormal situation of the current working position of the robot in the robot post-earthquake track inspection operation task.

[0070] In this embodiment, the execution environment threat degree of the robot post-earthquake track inspection operation task is analyzed according to the track passing detection condition abnormal situation analysis result in step S2; including the following specific steps:

[0071] S25, extract the line obstacle projection area and turnout locking device state code of the section where the robot operation position of the multiple historical post-earthquake track inspection operation task record is located from the track turnout state data, and associate and extract the track passing detection condition abnormality of the corresponding robot operation position; at the same time, extract the environmental threat mark of the robot operation position of the multiple historical post-earthquake track inspection operation task record, wherein the environmental threat mark is determined according to whether the robot of the corresponding historical post-earthquake track inspection operation task record is interrupted or canceled at the corresponding operation position due to environmental problems; if yes, the environmental threat mark is 1, otherwise the environmental threat mark is 0; the line obstacle projection area and turnout locking device state code of the section where the robot operation position of the multiple historical post-earthquake track inspection operation task record is located, the track passing detection condition abnormality of the corresponding robot operation position, and the environmental threat mark are taken as the environmental threat degree analysis data set; it should be noted that for the post-earthquake high-speed railway turnout area, the passing condition abnormality (track itself problem) is only a part of the environmental threat, and the passing condition abnormality may be combined with other environmental factors such as slope rockfall, foreign matter invading the limit, and locking state failure of the turnout key equipment to form a three-dimensional and composite threat environment; this is significantly different from the ordinary track segment inspection which usually only needs to focus on track geometry and a small amount of obstacles; specifically, from the track turnout state data of the historical track inspection task record, the line obstacle projection area and turnout locking device state code associated with each operation position of each task are extracted; the obstacle projection area is calculated by projecting the point cloud data scanned by the laser radar or binocular vision sensor carried by the robot to the track plane after coordinate conversion, which is an objective continuous value, and the calculation method of the obstacle projection area is the prior art, which will not be described here; the turnout locking device state code is derived from the sensor monitoring (such as current, voltage, and rod displacement) of the switch machine, close inspection device and other devices, and the discrete code (for example: 00-normal, 01-warning, 10-failure) output after state diagnosis algorithm; further, the embodiment accurately associates these records with the track passing detection condition abnormality calculated for the same historical task and the same operation position in step S24; and labels the environmental threat mark for each record, which is determined on the basis of objective operation logs: whether the robot is interrupted in the preset inspection process or cancels the subsequent task at the operation position due to environmental problems (such as encountering large obstacles and being unable to overcome obstacles, turnout locking failure leading to path being locked, etc.). If yes, the mark is "1" (environmental threat exists), otherwise it is "0". Based on the above, the embodiment realizes the fusion of multi-level environmental information.It will reflect the abnormal probability of the track basic state, the obstacle area reflecting the external risk, the locking state code reflecting the equipment function state, and the final resulting operation interruption consequences are associated together, so that the subsequent model can learn that not all track abnormalities will cause operation interruption (for example, slight settlement can be overcome by robots), but when track abnormalities are superimposed with large obstacles or equipment failures, the threat level will increase sharply.

[0072] S26, based on the gradient boosting decision tree model, an environment threat degree analysis model is constructed and trained; the track passing detection condition abnormality of the robot operation position in the environment threat degree analysis data set, the corresponding line obstacle projection area, and the turnout locking device state code of the multiple historical post-earthquake track inspection operation task records are taken as input features, and the environment threat sign is taken as an output target for model training to obtain a trained environment threat degree analysis model; in this embodiment, a gradient boosting decision tree model is selected, which can automatically capture high-order interaction and nonlinear relationship between features by integrating multiple weak decision trees, and is not sensitive to feature dimension, and is very suitable for processing mixed data in this step; the specific training process of the environment threat degree analysis model can be set as follows in this embodiment:

[0073] A1, an environment threat degree analysis set is obtained, which is composed of three-dimensional data points composed of feature values of three feature dimensions of abnormal conditions of passing conditions, line obstacle projection areas and turnout locking device state codes corresponding to each operation position recorded in multiple historical post-earthquake track inspection operation tasks, and environment threat signs (1 represents threat leading to operation interruption, and 0 represents normal) corresponding to each data point determined according to historical task logs; the environment threat degree analysis set is randomly divided into an environment threat degree training set and an environment threat degree verification set;

[0074] A2, three core construction parameters of the gradient boosting decision tree model are set in this embodiment; wherein the total number of decision trees is 100 by default, the maximum growth depth of a single decision tree is limited to 5 by default, and the learning rate (shrink step) of the model is 0.1 by default;

[0075] A3, for each iteration of the model to construct a new decision tree, the following operations are performed:

[0076] The predicted value of each sample in the environment threat degree analysis training set of the current integrated model (a basic constant value for the initial round) is calculated, and the pseudo-residual (i.e. the difference between the true label and the current predicted probability) is calculated based on the logarithmic loss function; the pseudo-residual is taken as a new target to be fitted in the current round;

[0077] From the analysis of the training set of all samples, a subset is randomly selected for constructing the current tree; in the process of tree construction, for any node that needs to be split, the current node data is the sample subset allocated to the node by the three-dimensional feature vector and the corresponding pseudo residual target value; from the three feature dimensions, by traversing all possible split points, the feature and split value that can make the pseudo residual sum of squares of the two child nodes after splitting decrease the most are selected, so that the current node data is divided into two subsets according to the split value, and is allocated to two child nodes respectively; the above process of selecting the best feature and split value is recursively performed on the generated child nodes to continue the division until any of the following conditions is met: the current node depth reaches 5, the number of data points in the current node is less than the preset minimum value, or the feature value split gain of all samples in the current node data is less than a certain threshold, then stop and form a leaf node.

[0078] A4、Repeat step A3 until 100 decision trees are generated by additive model integration as a gradient boosting decision tree model, as a trained environmental threat level analysis model; for receiving a new three-dimensional data point input, by letting the data point traverse each decision tree, moving from the root node to a leaf node according to the judgment rule of each tree, and performing weighted summation of all tree corresponding leaf node weight values, finally converting to a probability value between 0 and 1 through Sigmoid function as the execution environment threat level of the current job position.

[0079] S27, real-time acquisition of track access detection condition abnormal situation of robot current job position and detection of line obstacle projection area of robot current job position segment and turnout locking device state code, input trained environmental threat level analysis model for analysis, the probability value output by the environmental threat level analysis model is used as the execution environment threat level of the robot current job position in the robot post-earthquake track inspection operation task.

[0080] In this embodiment, as shown in Figure 4 , in step S3, the body operation control abnormal situation of the robot post-earthquake track inspection operation task is analyzed in combination with the body running data and track turnout state data of the robot; including the following specific steps:

[0081] S31, extract the three-dimensional coordinates of the robot operation position and the running direction from the body running data of the robot, and extract the nearest track turnout position passed by the robot operation position according to the three-dimensional coordinates of the robot operation position and the running direction; based on the robot operation position and the nearest track turnout position passed, real-time collection of the robot transverse displacement deviation value sequence, the robot pitch attitude angle sequence, the robot roll attitude angle sequence, the drive motor current fluctuation value sequence in the robot running process on the track section between the robot operation position and the nearest track turnout position passed; at the same time, the track curvature and track slope of the track section between the robot operation position and the nearest track turnout position passed are obtained from the track turnout state data;

[0082] S32, based on the robot transverse displacement deviation value sequence, the robot pitch attitude angle sequence, the robot roll attitude angle sequence, and the drive motor current fluctuation value sequence, respectively calculate the robot transverse displacement variation degree, the robot pitch attitude angle variation degree, the robot roll attitude angle variation degree, and the robot drive motor current fluctuation variation degree on the track section between the robot operation position and the nearest track turnout position passed; the core of the embodiment is to perform feature engineering on the original time sequence data, and extract statistical features capable of representing the operation stability of the robot on a specific track section. On the irregular track after the earthquake, the running state fluctuation (variation) of the robot can better reveal the potential risk than the average value; for example, on an ordinary smooth track, the transverse displacement deviation may fluctuate slightly around a very small mean value; but in the damaged turnout area, even if the average displacement is not large, there may be a dramatic and intermittent large jitter, which is a precursor of derailment; therefore, the method discards simple mean value analysis and introduces variation degree as the core feature; specifically including:

[0083] calculate the mean and standard deviation of the robot transverse displacement deviation value sequence, and take the ratio of the standard deviation and the mean of the robot transverse displacement deviation value sequence as the robot transverse displacement variation degree on the track section between the robot operation position and the nearest track turnout position passed;

[0084] calculate the mean and standard deviation of the robot pitch attitude angle sequence, and take the ratio of the standard deviation and the mean of the robot pitch attitude angle sequence as the robot pitch attitude angle variation degree on the track section between the robot operation position and the nearest track turnout position passed;

[0085] calculate the mean and standard deviation of the robot roll attitude angle sequence, and take the ratio of the standard deviation and the mean of the robot roll attitude angle sequence as the robot roll attitude angle variation degree on the track section between the robot operation position and the nearest track turnout position passed;

[0086] The mean and standard deviation of the sequence of the driving motor current fluctuation value are calculated, and the ratio of the standard deviation and the mean of the sequence of the driving motor current fluctuation value is taken as the driving motor current fluctuation variation degree of the robot on the track section between the robot operation position and the nearest track turnout position that has been passed.

[0087] S33, in the multiple historical post-earthquake track inspection operation tasks, the track curvature and track slope of the robot on the track section between any operation position and the nearest track turnout position that has been passed by the robot are obtained, and the robot lateral displacement variation degree, the robot pitch attitude angle variation degree, the robot roll attitude angle variation degree, and the robot driving motor current fluctuation variation degree on the corresponding track section are obtained, and the body operation abnormal flag of the robot passing through the corresponding track section in the corresponding historical post-earthquake track inspection operation task is obtained; wherein, the body operation abnormal flag is whether driving slip, attitude out of control or positioning abnormal failure occurs when the robot passes through the corresponding track section, if yes, the body operation abnormal flag is 1, otherwise the body operation abnormal flag is 0; specifically, the determination of the body operation abnormal flag depends on objective system logs and fault codes recorded in the task process; in this embodiment, it is inquired from the historical data whether the state monitoring system of the robot generates an explicit fault alarm immediately after the robot passes through the specific track section in the historical post-earthquake track inspection operation task, such as "driving wheel speed difference out of limit" (driving slip), "attitude angle out of safety threshold" (attitude out of control) or "positioning information jump / loss" (positioning abnormality); these alarms are generated by the robot bottom control system according to the real-time logical judgment of sensor data, which is the digital record of objective facts; as long as any of the above abnormal failures occurs, the body operation abnormal flag of the historical record is recorded as "1" (abnormal), otherwise as "0" (normal).

[0088] S34, the track curvature, track slope, robot lateral displacement variation degree, robot pitch attitude angle variation degree, robot roll attitude angle variation degree, robot driving motor current fluctuation variation degree and body operation abnormal flag of the robot passing through the corresponding track section obtained on the corresponding track section in the historical post-earthquake track inspection operation task are taken as an abnormal prediction data set, and the abnormal prediction data set is randomly divided into an abnormal prediction training set and an abnormal prediction verification set;

[0089] In this embodiment, the body operation control abnormality of the robot post-earthquake track inspection operation task is analyzed in step S3 in combination with the robot body running data and the track turnout state data; the following specific steps are further included:

[0090] S35, a gradient boosting tree model is constructed, based on the abnormality prediction training set, the track curvature, track slope, robot lateral displacement variation degree, robot pitch attitude angle variation degree, robot roll attitude angle variation degree, and robot drive motor current fluctuation variation degree obtained by counting on the corresponding track segment in the historical post-earthquake track inspection operation task are taken as the input features of the gradient boosting tree model, the robot body operation abnormality flag through the corresponding track segment is taken as the output target of the model, the gradient boosting tree model is trained, and an initial abnormality prediction model is obtained; the initial abnormality prediction model is verified through the abnormality prediction verification set, and the corresponding initial abnormality prediction model whose accuracy rate of predicting the abnormality prediction verification set reaches the preset model accuracy threshold is taken as the body operation abnormality analysis model; the specific training process of the body operation abnormality analysis model can be set as follows in this embodiment:

[0091] B1, an abnormality prediction training set is obtained, which is composed of six-dimensional data points composed of six characteristic values of track curvature, track slope, robot lateral displacement variation degree, robot pitch attitude angle variation degree, robot roll attitude angle variation degree, and drive motor current fluctuation variation degree obtained by counting on the corresponding track segment in the historical post-earthquake track inspection task, and the body operation abnormality flag (1 represents abnormality, 0 represents normality) corresponding to each data point determined according to whether the objective fault log such as drive slip, attitude out of control or positioning abnormality is recorded after passing through the segment;

[0092] B2, three core construction parameters of the gradient boosting tree model are set in this embodiment; wherein the total number of decision trees (i.e. the number of iteration rounds) is 100 by default, the maximum growth depth limit of a single decision tree is 6 by default, and the learning rate of the model is 0.1 by default;

[0093] B3, for each iteration of the model to construct a new decision tree, the following operations are performed:

[0094] The predicted value of each sample in the abnormality prediction training set of the current integrated model is calculated (the predicted value of all samples is initialized to the logarithmic ratio of the proportion of negative class samples in the data set in the initial round), and the pseudo-residual (i.e. the difference between the true label and the current model prediction probability) is calculated based on the logarithmic loss function; the pseudo-residual is taken as the new target to be fitted in the current round;

[0095] Using all or a random part of the training sample, in the process of tree construction, for any node that needs to be split, from the six feature dimensions, by traversing all possible split points, the feature and specific split value that can make the pseudo residual sum of squares of the samples in the two child nodes after splitting decrease the most are selected, so that the current node data is divided into two subsets according to the split value, and is assigned to two child nodes respectively; The process of selecting the best feature and split value described above is recursively performed on the generated child nodes to continue to divide until any of the following conditions is met: the current node depth reaches 6, the number of data points in the current node is less than 10, or all samples in the current node data have completely the same pseudo residual;

[0096] B4、Repeat step B3 until a gradient boosting tree model is generated by 100 decision trees through forward stepwise addition model integration, as the completed training ontology operation anomaly analysis model; for receiving a new six-dimensional data point input, by letting the data point traverse each decision tree in the forest, according to the judgment rule of each tree from the root node to a leaf node, and all tree corresponding leaf node weight contribution value is accumulated and summed, finally through Sigmoid function conversion to a probability value between 0 and 1 as the ontology operation control abnormal situation;

[0097] S36, real-time acquisition of the track curvature, track slope on the track section between the current working position of the robot and the corresponding nearest track turnout position that has passed, and the robot lateral displacement variation degree, robot pitch attitude angle variation degree, robot roll attitude angle variation degree, and robot drive motor current fluctuation variation degree obtained by statistics on the corresponding track section, are input into the ontology operation anomaly analysis model, and a predicted probability value is output as the ontology operation control abnormal situation of the current working position of the robot in the post-earthquake rail inspection operation task of the robot.

[0098] In this embodiment, the track inspection operation risk of the post-earthquake rail inspection operation task of the robot is evaluated based on the execution environment threat degree analysis result and the ontology operation control abnormal situation analysis result of the post-earthquake rail inspection operation task of the robot in step S4; including the following specific steps:

[0099] S41, extracting the execution environment threat degree and the ontology operation control abnormal situation of the current working position of the robot in the post-earthquake rail inspection operation task of the robot;

[0100] S42, the execution environment threat degree and the ontology operation control abnormal situation of the current working position of the robot in the post-earthquake rail inspection operation task of the robot are weighted and summed to obtain the track inspection operation risk of the current working position of the robot in the post-earthquake rail inspection operation task of the robot.

[0101] In the embodiment, the robot emergency braking warning is performed based on the track inspection operation risk assessment result of the robot post-earthquake track inspection operation task in step S5; specifically including:

[0102] S51, obtaining a track inspection operation risk assessment result of a current operation position of the robot in the robot post-earthquake track inspection operation task;

[0103] S52, presetting a track operation risk threshold value, when the track inspection operation risk assessment result of the current operation position of the robot in the robot post-earthquake track inspection operation task is greater than the track operation risk threshold value, performing the robot emergency braking warning; when the track inspection operation risk assessment result of the current operation position of the robot in the robot post-earthquake track inspection operation task is less than or equal to the track operation risk threshold value, continuing to perform the track inspection of the high-speed railway track; it should be noted that in the embodiment, the weight and the threshold value are obtained by: obtaining 3000 sets of the body operation data of the robot, the track switch state data and the earthquake influence data of the robot in the process of the robot performing the post-earthquake track inspection operation task, and calculating the track inspection operation risk of the robot post-earthquake track inspection operation task. At the same time, the judgment result of whether 3000 sets of the robot post-earthquake track inspection operation task is completed is obtained. The track inspection operation risk of the 3000 sets of the robot post-earthquake track inspection operation task and the judgment result of whether the corresponding robot post-earthquake track inspection operation task is completed are input into a fitting software for fitting, and the corresponding weight and threshold value of the highest determination coefficient are output.

[0104] Embodiment 2

[0105] As shown in Figure 3 , the embodiment provides a high-speed railway intelligent track inspection operation robot operation monitoring system, which comprises:

[0106] A data acquisition module is configured to synchronously acquire the body operation data of the robot, the track switch state data and the earthquake influence data in the process of the robot performing the post-earthquake track inspection operation task;

[0107] An execution environment monitoring module is configured to analyze the track passing detection condition abnormality in combination with the track switch state data and the earthquake influence data, and analyze the execution environment threat degree of the robot post-earthquake track inspection operation task according to the track passing detection condition abnormality analysis result;

[0108] A control abnormality detection module is configured to analyze the body operation control abnormality of the robot post-earthquake track inspection operation task in combination with the body operation data of the robot and the track switch state data;

[0109] A track inspection operation monitoring module is configured to evaluate the track inspection operation risk of the robot post-earthquake track inspection operation task based on the execution environment threat degree analysis result and the body operation control abnormality analysis result of the robot post-earthquake track inspection operation task;

[0110] The brake early warning module is used for early warning of robot emergency braking based on the track inspection operation risk assessment result of the robot post-earthquake track inspection operation task.

[0111] The steps of implementing the corresponding functions of the parameters and the unit modules in the high-speed railway intelligent track inspection operation robot operation monitoring system of the application can refer to the parameters and steps in the embodiments of the high-speed railway intelligent track inspection operation robot operation monitoring method, which will not be repeated here.

[0112] Each embodiment in the application is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other. Each embodiment mainly explains the difference from other embodiments. Especially, the Internet of Things device and medium embodiment is basically similar to the method embodiment, so the description is relatively simple, and the relevant parts can refer to the part of the method embodiment.

[0113] The system and medium provided by the embodiments of the application are one-to-one corresponding with the method, so the system and medium also have similar beneficial technical effects as the method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the system and medium will not be repeated here.

[0114] Those skilled in the art should understand that the embodiments of the application can be provided as a method, a system, or a computer program product. Therefore, the application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the application can be in the form of a computer program product implemented on one or more computer usable storage media containing computer usable program code (including but not limited to disk storage, CD-ROM, optical storage, etc.).

[0115] The application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The device that implements the functions specified in one flow or multiple flows and / or blocks. Figure 1 The device that implements the functions specified in one block or multiple blocks.

[0116] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0117] In one typical configuration, the computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0118] The memory can include non-persistent memory and / or volatile memory, such as random access memory (RAM) about which the computer stores the information. The memory is an example of computer readable media.

[0119] Computer readable media includes permanent and non-permanent, moveable and non- moveable media that can be implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile discs (DVDs) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. According to the definition provided herein, computer readable media does not include transitory media, such as modulated data signals and carrier waves.

[0120] It is also important to note that the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0121] The above merely illustrates the embodiments of the present application, and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. within the spirit and principles of the present application should be included in the scope of the claims of the present application.

Claims

1. A high-speed railway intelligent track inspection operation robot running monitoring method, characterized in that, The method comprises the following steps: S1, during the execution of the post-earthquake track inspection task by the robot, synchronously collecting the body operation data of the robot, the track turnout state data, and the earthquake influence data; S2, combining the track turnout state data and the earthquake influence data, analyzing the track passage detection condition abnormality, and according to the analysis result of the track passage detection condition abnormality, analyzing the threat degree of the execution environment of the post-earthquake track inspection task of the robot; S3, combining the body operation data of the robot and the track turnout state data, analyzing the body operation control abnormality of the post-earthquake track inspection task of the robot; S4, based on the analysis result of the threat degree of the execution environment of the post-earthquake track inspection task of the robot and the analysis result of the body operation control abnormality, evaluating the track inspection operation risk of the post-earthquake track inspection task of the robot; S5, based on the evaluation result of the track inspection operation risk of the post-earthquake track inspection task of the robot, performing the robot emergency braking pre-warning. 2.The high-speed railway intelligent track inspection robot operation monitoring method according to claim 1, characterized in that, In the step S2, the track passage detection condition abnormality is analyzed by combining the track turnout state data and the earthquake influence data; The method comprises the following specific steps: S21, extracting the track state parameters of the robot operation position in the multiple historical post-earthquake track inspection task records from the track turnout state data, wherein the track state parameters comprise a track transverse displacement value, a track vertical settlement value, and a turnout close clearance value; extracting the peak ground acceleration value and the aftershock occurrence frequency of the robot operation position in the corresponding historical track inspection period from the earthquake influence data in the multiple historical post-earthquake track inspection task records; at the same time, extracting the passage condition abnormality flag of the robot operation position in the corresponding historical track inspection period recorded after the multiple historical post-earthquake track inspection tasks are completed; based on the track transverse displacement value, the track vertical settlement value, and the turnout close clearance value of the robot operation position in the multiple historical post-earthquake track inspection task records, and the peak ground acceleration value, the aftershock occurrence frequency, and the passage condition abnormality flag of the robot operation position in the corresponding historical track inspection period, a passage condition abnormality analysis data set is constructed; S22, calculating the percentage of the number of records in which each track state parameter exceeds the corresponding preset safety threshold in the total number of records in the current peak ground acceleration value interval in different peak ground acceleration value intervals, and defining the corresponding percentage as the reference abnormal probability of the corresponding track state parameter under different ground motion levels; S23, constructing and training a passage condition abnormality analysis model based on a logistic regression model; taking the peak ground acceleration value, the aftershock occurrence frequency, the product of the track transverse displacement value and the corresponding reference abnormal probability, the product of the track vertical settlement value and the corresponding reference abnormal probability, and the product of the turnout close clearance value and the corresponding reference abnormal probability of each historical post-earthquake track inspection task record in the passage condition abnormality analysis data set as the input features of the model, and taking the passage condition abnormality flag as the training target of the model to train the model, thereby obtaining the trained passage condition abnormality analysis model; S24, real-time acquisition of the peak ground acceleration value, aftershock occurrence frequency, track lateral displacement value, track vertical settlement value, and turnout close clearance value of the current working position of the robot; according to the peak ground acceleration value interval of the peak ground acceleration value of the current working position of the robot, the corresponding reference abnormal probability of each track state parameter is queried; the peak ground acceleration value, aftershock occurrence frequency, and the product of the corresponding track lateral displacement value and the corresponding reference abnormal probability, the product of the corresponding track vertical settlement value and the corresponding reference abnormal probability, and the product of the corresponding turnout close clearance value and the corresponding reference abnormal probability of the current working position of the robot are input into the trained passing condition abnormality analysis model for analysis, and the probability value output by the passing condition abnormality analysis model is taken as the track passing detection condition abnormality situation of the current working position of the robot in the post-earthquake track inspection operation task of the robot. 3.The high-speed railway intelligent track inspection robot operation monitoring method according to claim 2, characterized in that, In step S2, the execution environment threat degree of the post-earthquake track inspection operation task of the robot is analyzed according to the track passing detection condition abnormality situation analysis result; The specific steps include: S25, extracting the line obstacle projection area and turnout locking device state code of the section where the robot working position of the multiple historical post-earthquake track inspection operation task records is located, and associating the track passing detection condition abnormality situation of the corresponding robot working position; at the same time, the environmental threat mark of the robot working position of the multiple historical post-earthquake track inspection operation task records is extracted; the line obstacle projection area and turnout locking device state code of the section where the robot working position of the multiple historical post-earthquake track inspection operation task records is located, the track passing detection condition abnormality situation of the corresponding robot working position, and the environmental threat mark are taken as the environmental threat degree analysis data set; S26, constructing and training an environmental threat degree analysis model based on a gradient boosting decision tree model; the track passing detection condition abnormality situation, corresponding line obstacle projection area, and turnout locking device state code of the robot working position of the multiple historical post-earthquake track inspection operation task records in the environmental threat degree analysis data set are taken as input features, and the environmental threat mark is taken as an output target for model training, to obtain a trained environmental threat degree analysis model; S27, real-time acquisition of the track passing detection condition abnormality situation of the current working position of the robot and the line obstacle projection area and turnout locking device state code of the section where the current working position of the robot is located, input into the trained environmental threat degree analysis model for analysis, and the probability value output by the environmental threat degree analysis model is taken as the execution environment threat degree of the current working position of the robot in the post-earthquake track inspection operation task of the robot. 4.The high-speed railway intelligent track inspection robot operation monitoring method of claim 3, wherein, In step S3, the body operation control abnormality situation of the post-earthquake track inspection operation task of the robot is analyzed in combination with the body operation data of the robot and the track turnout state data. The specific steps include: S31, extract the three-dimensional coordinates of the robot operation position and the running direction from the body running data of the robot, and extract the nearest track turnout position passed by the robot operation position according to the three-dimensional coordinates of the robot operation position and the running direction; based on the robot operation position and the nearest track turnout position passed, real-time collection of the robot transverse displacement deviation value sequence, the robot pitch attitude angle sequence, the robot roll attitude angle sequence, and the drive motor current fluctuation value sequence of the robot during operation on the track section between the robot operation position and the nearest track turnout position passed; at the same time, the track curvature and track slope of the track section between the robot operation position and the nearest track turnout position passed are obtained from the track turnout state data; S32, based on the robot transverse displacement deviation value sequence, the robot pitch attitude angle sequence, the robot roll attitude angle sequence, and the drive motor current fluctuation value sequence, the robot transverse displacement variation degree, the robot pitch attitude angle variation degree, the robot roll attitude angle variation degree, and the robot drive motor current fluctuation variation degree on the track section between the robot operation position and the nearest track turnout position passed are calculated respectively; S33, obtain the track curvature, track slope, robot transverse displacement variation degree, robot pitch attitude angle variation degree, robot roll attitude angle variation degree, and robot drive motor current fluctuation variation degree on the track section between the robot operation position and the nearest track turnout position passed in each history post-earthquake track inspection operation task, and obtain the body operation abnormal flag of the robot passing through the corresponding track section in the corresponding history post-earthquake track inspection operation task; S34, the track curvature, track slope, robot transverse displacement variation degree, robot pitch attitude angle variation degree, robot roll attitude angle variation degree, and robot drive motor current fluctuation variation degree obtained on the corresponding track section in the history post-earthquake track inspection operation task are taken as an abnormal prediction data set, and the abnormal prediction data set is randomly divided into an abnormal prediction training set and an abnormal prediction verification set. 5.The high-speed railway intelligent track inspection robot operation monitoring method according to claim 4, characterized in that, The step S3 combines the body running data of the robot and the track turnout state data to analyze the body operation control abnormality of the robot post-earthquake track inspection operation task; further comprising the following specific steps: S35. Construct a gradient boosting tree model. Based on the anomaly prediction training set, use the track curvature, track slope, robot lateral displacement variation, robot pitch angle variation, robot roll angle variation, and robot drive motor current fluctuation variation obtained from historical post-earthquake track inspection tasks on the corresponding track segments as input features of the gradient boosting tree model. Use the robot's body operation anomaly flags through the corresponding track segments as the model's output target. Train the gradient boosting tree model to obtain the initial anomaly prediction model. Validate the initial anomaly prediction model using the anomaly prediction validation set. Use the initial anomaly prediction model whose accuracy on the anomaly prediction validation set reaches the preset model accuracy threshold as the body operation anomaly analysis model. S36. In real time, acquire the track curvature and track slope on the track segment between the robot's current working position and the corresponding nearest passed track switch position, as well as the robot's lateral displacement variation, robot pitch angle variation, robot roll angle variation, and robot drive motor current fluctuation variation obtained statistically on the corresponding track segment. Input these values ​​into the body operation anomaly analysis model and output the predicted probability value as the body operation control anomaly of the robot's current working position in the robot's post-earthquake track inspection task. 6.The high-speed railway intelligent track inspection robot operation monitoring method according to claim 5, characterized in that, In step S4, the risk of the robot's post-earthquake track inspection task is assessed based on the analysis results of the threat level of the execution environment of the robot's post-earthquake track inspection task and the analysis results of abnormal operation control of the robot itself. The specific steps include the following: S41. Extract the threat level of the execution environment and abnormal operation control of the robot at its current working position during the robot's post-earthquake track inspection task; S42. The weighted sum of the threat level of the execution environment and the abnormal situation of the robot's operation and control at the current working position in the post-earthquake track inspection task is used to obtain the track inspection operation risk of the robot at the current working position in the post-earthquake track inspection task. 7.The method of claim 4, wherein, In step S5, an emergency braking warning for the robot is issued based on the risk assessment results of the robot's post-earthquake track inspection task. Specifically, it includes: S51. Obtain the risk assessment results of the robot's current working position in the post-earthquake track inspection task; S52. Preset trajectory operation risk threshold. When the risk assessment result of the robot's current working position in the post-earthquake track inspection task is greater than the trajectory operation risk threshold, the robot will perform an emergency braking warning. When the risk assessment result of the robot's current working position in the post-earthquake track inspection task is less than or equal to the trajectory operation risk threshold, the high-speed railway track will continue to be inspected.

8. The high-speed railway intelligent track inspection operation robot running monitoring system is based on the high-speed railway intelligent track inspection operation robot running monitoring method in any one of claims 1-7, and is characterized in that, The system includes: The data acquisition module is used to simultaneously collect the robot's own operation data, track switch status data, and earthquake impact data during the robot's post-earthquake track inspection operation. An execution environment monitoring module is configured to analyze track passage detection condition abnormality in combination with track turnout state data and earthquake influence data, and analyze the execution environment threat degree of the robot post-earthquake track detection operation task according to the track passage detection condition abnormality analysis result. A control abnormality detection module is configured to analyze the body operation control abnormality of the robot post-earthquake track detection operation task in combination with the body operation data of the robot and the track turnout state data. A track detection operation monitoring module is configured to evaluate the track detection operation risk of the robot post-earthquake track detection operation task based on the execution environment threat degree analysis result and the body operation control abnormality analysis result of the robot post-earthquake track detection operation task. A braking early warning module is configured to perform robot emergency braking early warning based on the track detection operation risk evaluation result of the robot post-earthquake track detection operation task.

Images