A terrain-adaptive track truck dispatching method

Abstract

The application discloses a kind of track freight car scheduling methods based on terrain adaptation, specifically related to the field of intelligent scheduling, for solving the problem of dispatching conflict caused by sensing failure due to terrain shielding in existing track freight transport.The method is to obtain the radio echo data and sensing failure state of the freight car in the historical running process, identify the terrain shielding section in the track line, and analyze the signal attenuation gradient before entering the shielding section, establish an identification model to determine whether to enter the sensing failure state soon;Further, the inertia propulsion analysis is carried out on the track in the shielding section, the track extension template is constructed, and the dynamic prediction envelope during the sensing failure period is generated;In the current scheduling period, based on the prediction envelope, find the potential spatial intersection conflict relationship, if conflict is found, generate a dispatching conflict signal, and implement dispatching correction for the conflict vehicle, so as to realize the precise adaptive scheduling control of terrain shielding section.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent scheduling, more particularly, to a rail freight car scheduling method based on terrain adaptation. BACKGROUND

[0002] In complex terrain scenarios such as mountainous transportation channels, tunnel staggered areas, and densely structured yard areas, the dynamic scheduling of rail freight cars highly depends on the accurate grasp of the position information of the freight cars. The traditional sensing system usually uses wireless signal feedback means to realize the occupation judgment and scheduling feedback. However, in areas with serious terrain shielding, such as mountain shielding, slope turning, tunnel group entrance, or large metal structure concentration section, this sensing means generally has the problem of sensing blind area or data feedback lag. For example, the signal propagation path of the trackside sensor is limited, the wireless signal is delayed or drifted due to terrain reflection, and even the track circuit may have signal noise interference in long slope wet slippery environment. This kind of terrain-induced sensing interruption phenomenon is not accidental, but has a clear regional characteristic and regular distribution. If the scheduling system does not form an effective identification and prediction mechanism, it is easy to make a mistake in scheduling during the sensing window period, resulting in too short following distance of the freight car, path intersection conflict, or marshalling station train control instruction error, which seriously affects the running continuity and safety redundancy of the rail system. Therefore, how to modify the rail scheduling strategy in real time in the sensing temporary loss or lag state combined with the distribution law of terrain shielding has become a key technical problem in the intelligent scheduling of rail transportation in complex terrain areas. SUMMARY

[0003] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a rail freight car scheduling method based on terrain adaptation to solve the problems raised in the background art.

[0004] To achieve the above object, the present application provides the following technical scheme:

[0005] A rail freight car scheduling method based on terrain adaptation, comprising the following steps:

[0006] S1, obtaining radio echo data and sensing failure state of the freight car in the historical running record, and identifying the terrain shielding section in the rail line;

[0007] S2, analyzing the radio echo signal attenuation gradient before entering the terrain shielding section, establishing a terrain shielding section identification model, and identifying whether the freight car is about to enter the sensing failure state;

[0008] S3, extracting the running parameters before the freight car enters the sensing failure state, performing inertia propulsion analysis on the running track in the terrain shielding section, and constructing a track extension template for each terrain shielding section;

[0009] S4, predicting a trajectory of the truck during a perception failure period based on the trajectory extension template and generating a dynamic prediction envelope;

[0010] S5, finding a potential spatial intersection conflict relationship with other trucks in the current scheduling period based on the dynamic prediction envelope, and generating a scheduling conflict signal;

[0011] S6, when the scheduling conflict signal appears, scheduling correction is performed on the truck in the intersection conflict relationship.

[0012] In a preferred embodiment, in S1, the radio echo data and perception failure state of the truck in the historical operation record are obtained, and the topographic shielding section in the track line is specifically identified, which comprises:

[0013] Based on the geographic coordinate data of the track line, the track section is divided and the corresponding section index is established;

[0014] The number of radio perception interruptions recorded in the historical operation log of the truck is obtained, and the perception failure occurrence frequency of each section is counted;

[0015] According to the distribution of the perception failure occurrence frequency in each track section, the cumulative density value of the perception failure is counted, and the track section with a density value higher than a set threshold is marked as a topographic shielding section.

[0016] In a preferred embodiment, in S2, the radio echo signal attenuation gradient before entering the topographic shielding section is analyzed, a topographic shielding section identification model is established, and whether the truck is about to enter a perception failure state is specifically identified, which comprises:

[0017] In a preset historical monitoring period, the radio echo time sequence signal strength recorded in the running process of all trucks is obtained, and a historical echo signal change curve is constructed;

[0018] Based on the historical echo signal change curve, the radio echo signal is analyzed for attenuation gradient, and a topographic shielding section identification model is established;

[0019] The real-time radio echo change curve is obtained and input into the topographic shielding section identification model, the topographic shielding section index that the truck is about to enter is determined according to the model output result, and the truck is about to enter the perception failure state is marked.

[0020] In a preferred embodiment, the radio echo signal is analyzed for attenuation gradient based on the historical echo signal change curve, and a topographic shielding section identification model is established, which specifically comprises:

[0021] Before entering the topographic shielding section, the signal strength of the historical echo signal change curve is differentially fitted in a preset analysis window, and the signal attenuation gradient feature is extracted;

[0022] The attenuation gradient characteristics in the plurality of echo signal change curves are mapped and associated with the terrain sheltering section index, a multi-class SVM classification algorithm is used to train a sheltering identification model, and the model automatically identifies the terrain sheltering section where the truck is located based on the attenuation gradient characteristics.

[0023] In a preferred embodiment, in S3, the running parameters of the truck about to enter the perception failure state are extracted, the inertia propulsion analysis of the running track in the terrain sheltering section is performed, and the track extension template of each terrain sheltering section is constructed, which specifically includes:

[0024] The truck about to enter the perception failure state is identified and the running parameters of the truck are extracted, the running parameters being the entry speed before entering the terrain sheltering section and the actual load;

[0025] When the truck enters the perception failure state, the on-board accelerometer is activated to record the acceleration sequence of the running track in the terrain sheltering section;

[0026] According to the actual load and the entry speed of the truck, the acceleration sequence in each terrain sheltering section is clustered;

[0027] Based on the clustering results, an inertia propulsion vector is established for each terrain sheltering section, and a track extension template for the terrain sheltering section is established.

[0028] In a preferred embodiment, the inertia propulsion vector is established for each terrain sheltering section based on the clustering results, and the track extension template for the terrain sheltering section is established, which specifically includes:

[0029] The acceleration sequence obtained by clustering in each terrain sheltering section is subjected to time normalization processing, and the length of the time axis is unified to obtain a standardized acceleration sequence template;

[0030] The standardized acceleration sequence is integrated once to calculate the corresponding speed change sequence;

[0031] The speed change sequence is integrated twice to obtain the sliding displacement change sequence per unit time, forming a track propulsion path;

[0032] The track propulsion path is converted into a propulsion vector group, the entry speed and the actual load are marked as indexes, and a track extension template for the terrain sheltering section is constructed.

[0033] In a preferred embodiment, in S4, based on the track extension template, the track of the truck during the perception failure is predicted and a dynamic prediction envelope is generated, which specifically includes:

[0034] For all trucks identified as about to enter the perception failure state, the terrain sheltering section index and the running parameters of the truck to be entered are obtained;

[0035] Based on the terrain shelter section index and the operating parameters, a corresponding trajectory extension template is called to perform multi-step linear superposition prediction, and a predicted trajectory of the truck during the perception failure is output;

[0036] A preset tolerance range is set, and a dynamic prediction envelope of all trucks about to enter the perception failure state is generated based on the predicted trajectory.

[0037] In a preferred embodiment, in S5, finding potential spatial intersection conflict relationships with other trucks based on the dynamic prediction envelope in the current scheduling period, and generating a scheduling conflict signal specifically includes:

[0038] The current scheduling period is converted into a unified time axis, and the dynamic prediction envelope is grouped according to the terrain shelter section index;

[0039] Comparing any two dynamic prediction envelopes in each group, the envelope interval overlap is judged at each time point on the current scheduling period time axis;

[0040] If a truck pair with envelope overlap relationship is identified, the overlap interval of the corresponding terrain shelter section is taken as the scheduling intersection conflict point of the current scheduling period, and a scheduling conflict signal is generated.

[0041] In a preferred embodiment, in S6, when the scheduling conflict signal occurs, the scheduling correction of the truck with intersection conflict relationship specifically includes:

[0042] When the scheduling conflict signal is monitored, the scheduling intersection conflict point of each terrain shelter section identified is obtained, and the truck pair with intersection conflict relationship corresponding to the intersection conflict point is extracted;

[0043] Selecting the truck in the truck pair that is not in the process of entering the terrain shelter section as the scheduling correction object, and based on the trajectory extension template, the position influence of the initial speed change into the section on the dynamic prediction envelope of the truck is simulated in sequence according to the preset speed adjustment step;

[0044] In the simulation process, the minimum initial speed adjustment amount into the section that does not produce overlap with all dynamic prediction envelopes of other trucks in the current scheduling period is selected, and the adjustment amount is taken as the scheduling correction amount of the scheduling correction object.

[0045] The technical effects and advantages of the track truck scheduling method based on terrain adaptation are as follows:

[0046] The application realizes continuous prediction and scheduling control of the running position of the track freight car in the radio echo perception interruption state, effectively solves the problem of the sensing blind area caused by shielding in the traditional sensing method in the complex terrain scene. Through analyzing the echo signal attenuation gradient, the sensing failure risk is accurately identified, and the inertial propulsion modeling is carried out combined with the entry speed and load characteristics of the freight car, so that the predicted trajectory has personalized adaptability and engineering executability. At the same time, the construction of the dynamic prediction envelope combined with the multi-car space intersection analysis mechanism can generate a scheduling conflict signal in advance, so that the scheduling system still has the ability to avoid path conflicts during the sensing window, and improves the scheduling continuity and system safety redundancy. The overall scheme does not depend on additional hardware deployment, can integrate existing radio sensing devices and operation data, has good expansibility and engineering adaptability, and is especially suitable for intelligent scheduling in sensing failure environments such as mountainous railways, tunnel-intensive areas and yard transfer areas. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 A schematic diagram of a track freight car scheduling method based on terrain adaptation. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0049] Embodiment 1

[0050] Figure 1 A track freight car scheduling method based on terrain adaptation is given, which includes the following steps:

[0051] S1, obtaining radio echo data and sensing failure state of the freight car in the historical operation record, identifying the terrain shielding section in the track line;

[0052] S2, analyzing the radio echo signal attenuation gradient before entering the terrain shielding section, establishing a terrain shielding section identification model, and identifying whether the freight car is about to enter a sensing failure state;

[0053] S3, extracting the running parameters of the freight car before entering the sensing failure state, performing inertial propulsion analysis on the running trajectory in the terrain shielding section, and constructing a trajectory extension template for each terrain shielding section;

[0054] S4, predicting the trajectory of the freight car during the sensing failure period based on the trajectory extension template and generating a dynamic prediction envelope;

[0055] S5, finding potential spatial intersection conflict relationship with other trucks in the current scheduling period based on dynamic prediction envelope, and generating scheduling conflict signal;

[0056] S6, when the scheduling conflict signal appears, scheduling correction is performed on the trucks in the intersection conflict relationship.

[0057] In S1, the radio echo data and perception failure state of the truck in the historical operation record are obtained, and the terrain shielding section in the track line is identified.

[0058] The precise coordinate data of the track line in the spatial geographic layer is obtained, and the data source is the existing track surveying and mapping results or the surveying and mapping data of the track collected through high-precision GNSS measurement. The data labels the longitude and latitude point of each track center line in a linear manner, and the corresponding elevation information is attached. Based on the coordinate data, the line segment segmentation operation is performed, and the whole track line is divided into several track sections according to the equal length principle or the characteristic change principle. If the track has continuous curve change, significant slope fluctuation or structure type (such as tunnel, bridge, yard) change point, non-equal length section division is performed with these turning points as boundaries, so that each section has relatively uniform structure characteristics. Each section will be assigned a unique section index number, and the numbering is arranged in ascending order according to the running direction of the track.

[0059] The radio echo communication state record is extracted from the historical operation log recorded in the running process of the truck in the trial operation stage. The operation log includes the time stamp, current position (provided by the vehicle-mounted camera recording in the trial operation stage, unmanned aerial vehicle aerial photography or manual recording), radio echo receiving state flag and other fields in each running process. By traversing the log file, all records of radio perception interruption are filtered out, and these points are located based on the time stamp and position data. Then, according to the established track section index, each interruption record is mapped to a specific track section, and the number of perception interruption events in each section is counted. In order to eliminate the interference of running frequency difference on the statistical result, standardization processing is introduced, that is, the number of perception interruption is divided by the total number of running times of the section in the historical log, and the perception failure frequency of the section is obtained. This frequency data will be arranged into a section frequency table, each record including the section number and its corresponding perception failure frequency value, which will be used as an indicator for subsequent shielding judgment.

[0060] The frequency of perception failure of each track section is calculated by sliding accumulation to obtain the perception failure density value of each section in the adjacent range. This operation is achieved by setting a sliding window (for example, containing 3 adjacent sections before and after) and summing or averaging the frequency values in the window. The accumulated density value reflects the concentration of perception interruption in a certain section of the track line and its adjacent area, and can reflect the distribution characteristics of the terrain shielding area. According to the maximum density value of the non-shielding section in the historical operation data as the reference lower limit, the density value judgment threshold is set, for example, the density threshold is set to 0.45. For the track section with accumulated density value higher than the threshold, it is considered that the terrain factor has a significant impact, and it is difficult to guarantee the stability of radio perception, which is marked as a terrain shielding section.

[0061] In S2, the radio echo signal attenuation gradient before entering the terrain shielding section is analyzed, a terrain shielding section identification model is established, and whether the truck is about to enter the perception failure state is identified.

[0062] The historical operation data of the truck archived in the trial operation stage is set to have a monitoring period, which is consistent with the dispatching cycle length to which the historical operation data belongs. In this time period, the radio echo time sequence signal data bound with the position information in the running record of all trucks equipped with radio positioning is extracted. Each data record contains track coordinates, time stamp and corresponding signal strength value, and the sampling frequency is set to 1 per second to ensure that the details of signal fluctuation can be captured. The data is sorted according to the track line and running time, and the signal strength change of each running process of each truck is sorted to form a complete historical echo signal change curve. In order to avoid signal errors caused by short-time shielding, equipment interference and other incidental abnormalities, median filtering and outlier rejection processing are performed on the signal curve to ensure that the echo change trend is physically reasonable.

[0063] On the basis of the historical echo signal change curve, the typical signal attenuation characteristics before entering the shielding section are extracted, and the section positioning analysis is performed on the curve. According to the established shielding section index information, the pre-shielding section in each truck running process is located, that is, an analysis window is set in the section before the truck enters the shielding area. The length of the analysis window is set according to the time when the echo signal change curve starts to attenuate, and the default setting is 10 seconds, which is used as the differential analysis area. First-order differential calculation is performed on the signal strength sequence in the window, the signal change rate sequence is extracted, and linear fitting is performed based on the least square method to obtain the descending slope of the curve, that is, the signal attenuation gradient. In order to eliminate the signal reference offset caused by the difference between equipment models, all attenuation gradient values are normalized according to the reference strength of each vehicle equipment, and the unified measurement standard is used to generate a unique attenuation gradient feature vector for each track.

[0064] After obtaining the complete set of attenuation gradient feature vectors, a label mapping relationship between the feature vectors and the actual shielding section index is established. Specifically, the position information provided by the on-board camera recording, unmanned aerial vehicle aerial photography or manual recording in the trial operation stage is time-aligned with the attenuation gradient features. The mapping process is manually verified and labeled according to the actual entry shielding section number corresponding to each operation process, to ensure the accuracy of the training samples. A multi-class support vector machine (SVM) is selected as the classification model algorithm framework, and a radial basis function (RBF) kernel is used to deal with the nonlinear characteristics of signal changes. To prevent overfitting, the sample set is divided into a training set and a validation set in a ratio of 8:2. During the training process, the cross-validation method is used to optimize the penalty parameter and the kernel function parameter combination, and the model generalization ability is improved.

[0065] After the model is established, the radio echo signal strength data of the running truck is collected in real time, and the time sequence change is continuously recorded. The current track coordinates of the truck and its real-time signal strength sequence are automatically extracted to form a real-time signal change curve. The curve is processed by sliding difference and linear fitting to extract the attenuation gradient feature vector in the current running environment, which is input into the aforementioned SVM shielding section recognition model to obtain the output shielding section prediction number. At the same time, the current truck state is marked as about to enter the perception failure state.

[0066] In S3, the running parameters of the truck before it enters the perception failure state are extracted, and the running trajectory in the terrain shielding section is analyzed by inertia to construct a trajectory extension template for each terrain shielding section.

[0067] Based on the output results of the terrain shielding section recognition model, the truck about to enter the perception failure state is recognized and extracted in real time. After the recognition and extraction are completed, the current running parameters of the truck are extracted synchronously, wherein the entry speed is obtained from the monitoring in the previous normal communication state, and the instantaneous speed of the truck at the end point of the analysis window (i.e. the time point before the boundary of the terrain shielding section) is recorded. The speed value is expressed in meters per second and is normalized according to the track section length. The actual load parameter is derived from the loading weighing record at the job departure station. Before starting the job, the loading mass of each truck should be accurately recorded by the track weighing device, and the record unit is ton. The information is attached to the running log before the truck departs. The load and speed are paired to form a running parameter pair, which describes the inertia state of the truck when it enters the terrain shielding section.

[0068] When the truck is officially passing the boundary point of the sheltered section (the radio echo is officially interrupted) and is marked as entering the perception failure state, the acceleration collection function is automatically activated. The accelerometer should have three-axis sensing capability and capture the front acceleration as the main channel, with a sampling frequency not less than 10 Hz, ensuring continuous perception of motion state changes. Since external positioning support cannot be obtained during wireless perception interruption, the acceleration sequence recorded by the vehicle-mounted accelerometer is the only available dynamic trajectory information source. During the collection process, the accelerometer records the instantaneous acceleration value at each time point and stores it in the vehicle-mounted buffer in timestamp order. The end time is determined by the time when the radio echo is restored. During the entire sheltered section, no real-time upload processing is performed on the acceleration data, but after the collection is completed, the entire section acceleration sequence is packaged with the running parameters and uploaded to the central dispatch analysis after the communication is restored.

[0069] After completing the acceleration sequence collection of multiple trucks in the same section, all acceleration sequence samples in the same terrain sheltered section are structurally classified. To avoid the influence of load and speed differences on sequence comparison, normalization processing is first performed. The load normalization method is based on the set maximum rated load of the truck (for example, set to 30 tons), and the speed normalization is based on the set track standard speed threshold (for example, set to 20 m / s). The normalized running parameters and acceleration sequence are input into the clustering analysis process. The clustering method selects a density-based clustering algorithm, and the clustering dimensions include the normalized entry speed, normalized load value, and statistical features of the acceleration sequence (such as mean, standard deviation, maximum slope, etc.), while introducing dynamic time warping distance as a sequence similarity measurement index to improve the accuracy of sequence alignment and matching. After clustering is completed, each cluster represents a type of inertia propulsion mode, records the running parameter interval and acceleration form template corresponding to each cluster, and is bound with the terrain sheltered section index to form a reference template group for trajectory extension. In the case of not collecting complete acceleration data later, the most similar acceleration template is quickly matched according to the entry parameters to realize the estimation and deduction of the motion path during the sheltered period.

[0070] After clustering the acceleration sequences, the movement path of each acceleration template sequence in the terrain sheltered section is further derived based on numerical integration to structure the acceleration signal sequence. First, select the time interval (e.g. 0.1 seconds) in units of the accelerometer sampling frequency, and integrate each normalized acceleration sequence to form a velocity change sequence. The integral is numerically approximated using the trapezoidal method to ensure the accuracy of the velocity derivation at the acceleration curve intervals, and the relative speed value sequence is obtained by accumulation at each time. The speed sequence reflects the speed change trend of the truck in the terrain sheltered section, and is an intermediate quantity connecting acceleration and spatial displacement. The speed sequence obtained by first-order integration needs to be offset corrected according to the initial speed of the section, i.e. the first speed value of each sequence is not set to 0, but the initial speed extracted before is used as the initial value, so that the speed sequence is consistent with the actual physical state. After completing the correction of the speed sequence, continue to perform the second-order integration operation, i.e. accumulate the integration of each speed value to obtain the sliding displacement in each unit time interval. Similarly, the trapezoidal numerical integration is used, and the integration results are arranged in time order to form the sliding displacement sequence in the sheltered section. Each element represents the estimated moving distance of the truck in the corresponding time interval. This sliding displacement sequence constitutes the trajectory advancing path, which is converted into a spatial advancing path according to the time step. The path starting point can be set as the spatial coordinates of the starting point of the sheltered section. Given the unit time interval and the sliding displacement length, the trajectory points of each time step are recursively constructed in a fixed advancing direction (from the previous trajectory section towards the derivation). The final trajectory advancing path is a sequence of displacement points with fixed time steps.

[0071] After constructing the trajectory advancing path, the displacement path sequence is converted into a vector representation to realize template storage and subsequent dynamic prediction operations. Each trajectory advancing path is composed of a set of consecutive spatial coordinate points, and the difference between adjacent coordinate points is calculated to obtain the displacement vector between the two points, i.e. a set of advancing vectors is formed. Each advancing vector contains direction information and carries the advancing amplitude to describe the actual sliding trend of the time period. The vector is represented in three dimensions (X, Y, T), where T is the time step number corresponding to the advancing vector, used to align with the current scheduling period. An index is bound to each set of advancing vectors to facilitate retrieval and matching based on actual operating parameters. The index consists of two parts: the entry speed gear and the load interval number. The entry speed gear is divided into several grades (e.g. 5 m / s per grade) according to the set standard, and the load interval is set according to the tonnage standard range (e.g. every 5 tons is an interval). For example, a sample with a speed of 13.4 m / s and a load of 17.8 tons will be labeled as speed gear 3 and load gear 4. Using this grading mechanism can significantly reduce the number of templates and improve matching efficiency.

[0072] After the generation of the propulsion vectors and the binding of the indexes, all generated propulsion vector groups are stored according to the terrain shelter segment indexes of the corresponding track segments, and a corresponding track extension template library is constructed. Each template library entry includes: track segment index, speed-load combination index, and propulsion vector sequence. All template storage structures should support fast retrieval according to the combination index. The template construction process is a one-time offline generation process, and template updating tasks can be uniformly executed after each running period ends. For new load structures and new trajectory propulsion generated under new scheduling speed strategies, the same process is performed to complete the supplement and update the propulsion vector distribution in the existing template according to the track maintenance changes.

[0073] In the S4, the trajectory of the truck during the perception failure is predicted based on the track extension template, and a dynamic prediction envelope is generated.

[0074] For all trucks identified as about to enter the perception failure state, the shelter segment index, entry speed, and actual load parameters of each truck are obtained in turn. This parameter group is used as the index field of the track extension template library, and through the double-key index structure of the speed and load grades, the trajectory propulsion template of the vehicle under the specified shelter segment is quickly matched. Starting from the starting point of the template, the displacement of each propulsion vector is added to the coordinates of the previous position point in the order of the time steps of the propulsion vectors, and the prediction of the current position is completed. The whole process is a linear propulsion calculation, which does not rely on the trajectory trend formed by real-time sensor data to complete the trajectory extrapolation, thereby performing trajectory prediction.

[0075] After obtaining all the predicted trajectories, enter the dynamic envelope construction phase to reflect the track space area that each truck may actually occupy under the shielding state. Since the predicted trajectory is a single central path, it needs to be further expanded into a dynamic space region with a boundary profile based on a pre-set tolerance range. The pre-set tolerance range can be set according to the running stability of the truck, safety requirements and trajectory error history distribution, and generally takes 0.5 meters to 1.0 meters as the basis for horizontal and vertical deviation limits. The specific value can be determined according to the measured data (such as the deviation between the predicted time of driving out of the terrain shielding section and the actual time), and the default is to generate an envelope with a horizontal offset of ±0.6 meters and a vertical advance and lag offset of ±0.4 meters. In the specific operation process, a two-dimensional ellipse or polygon buffer zone with the center point as the trajectory point coordinate and the space radius as the tolerance range is constructed for each predicted trajectory point, and all buffer zones are sequentially connected to form the envelope boundary. Considering the consistency of the track running direction, the envelope construction method can be unified as an expansion strategy for the main propulsion direction + horizontal offset to form a dynamic forward expansion area. If an elliptical envelope structure is used, the center line propulsion path can be taken as the long axis direction, and the horizontal tolerance is the short axis to generate a trajectory corridor shaped envelope. After completing the construction of the individual dynamic envelope of each vehicle, it is bound with the vehicle number and the corresponding scheduling period time interval to form a data entity that can participate in scheduling conflict judgment. All envelopes are stored in the form of a space region set and indexed by time steps for subsequent scheduling and control steps.

[0076] In S5, potential spatial intersection conflict relationships with other trucks are found based on the dynamic prediction envelope in the current scheduling period, and a scheduling conflict signal is generated.

[0077] After identifying all the trucks that will enter the perception failure state and completing the construction of their dynamic prediction envelopes, the current scheduling period is taken as a unified analysis time axis, and all dynamic prediction envelopes are grouped based on the index of the terrain shielding section to which they belong. First, the terrain shielding section index field attached to each prediction envelope is extracted, and all envelopes are divided into several groups according to the index value, with each group corresponding to a shielding section. This grouping operation ensures that envelope intersection judgment is only performed within the same shielding section, excluding interfering paths that do not overlap physically across sections. The construction of the unified time axis takes the start and end times of the current scheduling period as boundaries and divides it into fixed time steps to form a time sequence. For example, if the period length is set to 15 seconds and the step length is 0.1 seconds, 150 analysis time points are formed. All prediction envelopes are aligned in coordinates with reference to this unified time axis. The original timestamp sequence in the envelope needs to be mapped to the corresponding time points on this unified axis. If there is a situation of incomplete alignment, linear interpolation is used to fill in the blank positions to ensure that there are corresponding envelope boundary coordinate sets at each time point.

[0078] After the above data alignment and grouping operations are completed, a pairwise comparison and judgment is performed on any two dynamic prediction envelopes within each group. The judgment process is performed separately for each time point, and the boundary ranges of the two envelopes at the time point are compared in sequence to determine whether there is a spatial overlap relationship. The spatial overlap determination criterion is based on two-dimensional track plane coordinates, and the minimum circumscribed rectangle envelope or standard polygon contour is used to calculate the intersection area of the two regions. If the intersection area is greater than a set minimum value (such as 0.01 square meters), it is determined that there is a spatial overlap. In the process of traversing the time axis, once an envelope pair with an overlapping relationship is identified at a time point, the time point and the corresponding two vehicle numbers are immediately recorded as a dispatch intersection conflict information. At the same time, the conflict information is bound with the shielding section index to form a complete dispatch conflict identifier, which includes parameters such as truck pair number, conflict time point, shielding section index, and overlapping envelope range. For all envelope pairs in each group, the above traversal judgment process needs to be performed on the complete time axis, and finally a dispatch intersection conflict list in the shielding section is formed, and a dispatch conflict signal is sent.

[0079] In S6, when a dispatch conflict signal occurs, the trucks with intersection conflict relationship are corrected.

[0080] After receiving the dispatch conflict signal in the dispatch cycle, first, all dispatch intersection conflict lists are parsed to extract the terrain shielding section index and the corresponding dispatch intersection conflict point information involved. Each conflict record explicitly includes the shielding section where the conflict occurs, the conflict time point, and the truck pair number in the overlapping envelope relationship. By traversing the current dispatch conflict list, the truck pairs with intersection conflict relationship are extracted one by one, and the mapping of the shielding section index related to them is established to form a data input set for dispatch correction processing. For each pair of trucks with intersection conflict relationship, it is determined which truck has dispatch correction space. The correction object preferentially selects vehicles that are not yet in the corresponding shielding section or have not yet entered the corresponding shielding section. Such vehicles are still in the track perception effective area, and their running parameters can be effectively adjusted without affecting the subsequent perception strategy, thus having the feasibility of implementing dispatch correction. If both trucks in the conflict pair meet this condition, the vehicle that enters the section later is preferentially selected as the dispatch correction object to reduce the risk of subsequent dispatch chain changes.

[0081] After selecting the dispatch correction object, the current entry segment initial speed and actual load parameters are extracted, combined with the terrain shelter segment index it will enter, and the corresponding trajectory extension template is called. The template is a standard propulsion path set generated by the shelter segment under specific operating conditions, and its structure is a speed-trajectory displacement mapping group. On this template, one-way simulation adjustment of the entry segment speed parameter is performed according to a fixed speed adjustment step, for example, the adjustment step is set to 0.5 m / s, and a series of simulated entry segment speed values are generated by sequentially decreasing from the current speed. For each simulated entry segment speed value, the trajectory extension template is called to regenerate the corresponding dynamic prediction envelope, and the prediction envelope is inserted into the current dispatch period unified time axis to judge the overlap relationship with all other running or predicted cargo trucks. The judgment process is consistent with the above-mentioned intersection conflict identification step, that is, whether the prediction envelope overlaps with any envelope at all time points is determined one by one.

[0082] Once the prediction envelope generated by a certain entry segment speed adjustment value does not overlap with any envelope of other cargo trucks in the dispatch period, the speed adjustment value is marked as a feasible correction speed. Among all feasible correction speeds, the one with the smallest adjustment amount is selected as the final correction amount to ensure the minimum degree of dispatch intervention and avoid unnecessary interference with the overall operating rhythm. Finally, the dispatch correction amount is applied to the dispatch parameters of the dispatch correction object to update its dispatch start time and entry segment speed setting and generate a dispatch correction instruction, realizing real-time avoidance of potential dispatch conflicts. The entire process is based on the trajectory extension template, combined with the minimum speed adjustment principle, to complete the precision constraint and dynamic dispatch correction with an automated dispatch strategy.

[0083] The above formulas are dimensionless values calculated, and the formulas are obtained by software simulation of a large amount of data to obtain a formula closest to the actual situation. The preset parameters and threshold values in the formula are set by a person skilled in the art according to the actual situation.

[0084] The above-described embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, the above-described embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center through a wired (for example, infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0085] Those of ordinary skill in the art can realize that the modules and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0086] Those of ordinary skill in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, device, and module can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0087] In several embodiments provided in the present application, it should be understood that the disclosed system, device, and method can be implemented in other ways. For example, the above-described device embodiments are only schematic, for example, the division of the modules is only a logical function division, and actual implementation can have another division manner, for example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed ones can be indirect coupling or communication connection through some interfaces, devices, or modules, which can be electrical, mechanical, or other forms.

[0088] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment.

[0089] In addition, the functional modules in each embodiment of the present application can be integrated into one processing module, or each module can exist physically alone, or two or more modules can be integrated into one module.

[0090] The functions, if realized in the form of software function modules and sold or used as independent products, can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program code storage media.

[0091] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0092] Finally: the above is only the preferred embodiment of the present application, and is not used to limit the present application, any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application, should be included in the protection scope of the present application.

Claims

1. A terrain-adaptive rail freight car scheduling method, characterized in that, Includes the following steps: S1. Obtain radio echo data and sense failure status of freight cars in historical operation records, and identify terrain-obscured sections in the track line. S2. Analyze the attenuation gradient of radio echo signals before entering the terrain-shaded section, establish a terrain-shaded section identification model, and identify whether the truck is about to enter a state of sensing failure. S3. Extract the operating parameters of the truck before it enters the state of perception failure, perform inertial propulsion analysis on the operating trajectory in the terrain-shaded section, and construct the trajectory extension template for each terrain-shaded section. S4. Predict the trajectory of the truck during the perception failure period based on the trajectory extension template and generate a dynamic prediction envelope; S5. Within the current scheduling cycle, find potential spatial intersection and conflict relationships with other freight cars based on dynamic prediction envelopes, and generate scheduling conflict signals. S6. When a scheduling conflict signal occurs, the scheduling of the trucks with conflicting relationships shall be corrected.

2. The method for scheduling rail freight cars based on terrain adaptation according to claim 1, characterized in that, In step S1, acquiring radio echo data and sensing failure status of the freight car from its historical operation records, and identifying terrain-obscured sections in the track line specifically includes: Based on the geographical coordinate data of the rail lines, the rail sections are divided and corresponding section indexes are established; Obtain the number of radio sensing interruptions recorded in the historical operation log of the cargo truck, and count the frequency of sensing failures in each section; Based on the distribution of sensing failure frequency across different track segments, the cumulative density value of sensing failure is calculated, and track segments with density values ​​exceeding a set threshold are marked as terrain-masked segments.

3. The method for scheduling rail freight cars based on terrain adaptation according to claim 1, characterized in that, In step S2, the attenuation gradient of the radio echo signal before entering the terrain-obscured section is analyzed, and a terrain-obscured section identification model is established to identify whether the truck is about to enter a state of sensing failure. Specifically, this includes: Within a preset historical monitoring period, the intensity of radio echo timing signals recorded during the operation of all trucks is acquired, and historical echo signal change curves are constructed. Based on the historical echo signal variation curve, attenuation gradient analysis of radio echo signals is performed to establish a terrain-shaded section identification model. The real-time radio echo change curve is obtained and input into the terrain occlusion section identification model. Based on the model output, the index of the terrain occlusion section that the truck is about to enter is determined, and the truck is marked as about to enter a perception failure state.

4. The method for scheduling rail freight cars based on terrain adaptation according to claim 3, characterized in that, The specific steps for establishing a terrain-shading section identification model by performing attenuation gradient analysis of radio echo signals based on historical echo signal variation curves include: Before entering the terrain-obscured section, within the set analysis window, the signal intensity decrease rate of the historical echo signal change curve is differentially fitted to extract the signal attenuation gradient characteristics. The attenuation gradient features in multiple echo signal variation curves are mapped and associated with the terrain shading section index. A multi-class SVM classification algorithm is used to train the shading recognition model. The model automatically identifies the terrain shading section where the truck is located based on the attenuation gradient features.

5. A method for scheduling rail freight cars based on terrain adaptation according to claim 1, characterized in that, In step S3, the operating parameters of the truck before it enters the perception failure state are extracted, and the inertial propulsion analysis is performed on the operating trajectory within the terrain-obscured section. The specific steps for constructing the trajectory extension template for each terrain-obscured section include: Identify the truck that is about to enter a state of sensor failure and extract the truck's operating parameters, which are the entry speed and actual load before entering the terrain-obscured section. When the truck enters a state of sensor failure, the on-board accelerometer is activated to record the acceleration sequence of its running trajectory in the terrain-obscured section. Based on the actual load and entry speed of the trucks, the acceleration sequences in each terrain-shaded section are clustered. Based on the clustering results, an inertial propulsion vector is established for each terrain-shaded segment, and a trajectory extension template for that terrain-shaded segment is created.

6. A method for scheduling rail freight cars based on terrain adaptation according to claim 5, characterized in that, The process of establishing an inertial propulsion vector for each terrain-shading segment based on clustering results, and establishing a trajectory extension template for that terrain-shading segment, specifically includes: Time normalization is performed on the acceleration sequences obtained by clustering under each terrain-shaded segment to obtain a standardized acceleration sequence template with a uniform time axis length. Perform a single integration on the standardized acceleration sequence to calculate the corresponding velocity change sequence; By performing a second integral on the velocity change sequence, the sliding displacement change sequence per unit time is obtained, forming the trajectory advancement path; The trajectory advancement path is converted into a set of advancement vectors, and the entry speed and actual load are used as index markers to construct a trajectory extension template for terrain-covered sections.

7. A method for scheduling rail freight cars based on terrain adaptation according to claim 1, characterized in that, In step S4, predicting the trajectory of the truck during the perception failure period based on the trajectory extension template and generating a dynamic prediction envelope specifically includes: For all trucks identified as about to enter a state of perception failure, obtain the index of the terrain-obscured section they are about to enter and their operating parameters; Based on the terrain-shading section index and operating parameters, the corresponding trajectory extension template is called to perform multi-step linear overlay prediction and output the predicted trajectory of the truck during the perception failure period. With a preset tolerance range, dynamic prediction envelopes are generated for all trucks that are about to enter a state of perception failure by predicting their trajectories.

8. A method for scheduling rail freight cars based on terrain adaptation according to claim 1, characterized in that, In step S5, the process of finding potential spatial intersection and conflict relationships with other freight cars based on dynamic prediction envelopes within the current scheduling cycle and generating scheduling conflict signals specifically includes: The current scheduling cycle is converted into a unified time axis, and the dynamic prediction envelope is grouped according to the terrain shading segment index. Compare any two dynamically predicted envelopes within each group, and perform envelope interval overlap judgment at each time point on the current scheduling cycle time axis; If a pair of freight cars with overlapping envelopes is identified, the overlapping section of the corresponding terrain-shaded area is taken as the scheduling intersection conflict point of the current scheduling cycle, and a scheduling conflict signal is generated.

9. A method for scheduling rail freight cars based on terrain adaptation according to claim 1, characterized in that, In step S6, when a scheduling conflict signal occurs, the scheduling correction for trucks with intersecting conflict relationships specifically includes: When a scheduling conflict signal is detected, the scheduling intersection conflict point of each identified terrain-shaded section is obtained, and the cargo truck pairs with intersection conflict relationship corresponding to the intersection conflict point are extracted. The freight cars that are in the middle of the freight car pair that have not yet departed or are about to enter the terrain-obscured section are selected as the scheduling correction objects. The step size is adjusted according to the preset speed, and the influence of the initial speed change on the position of the dynamic prediction envelope of the freight car is simulated sequentially based on the trajectory extension template. During the simulation, the minimum initial speed adjustment of the entry segment that does not overlap with the dynamic prediction envelope of all other freight cars within the current scheduling cycle is selected, and this adjustment is used as the scheduling correction amount for the scheduling correction object.

Images