Method for assessing and simulating resilience of regional transportation system based on inclined orbit satellite

Abstract

The application discloses a regional traffic system resilience evaluation and simulation deduction method based on an inclined orbit satellite, relates to the technical field of data processing, and collects fine-grained space-time data of traffic operation by using the multi-orbit covering characteristics of the inclined orbit satellite to form a traffic state feature set; in the process of space-time sequence preprocessing of the traffic state feature set, an event influence feature set is generated, and combined with structure information extracted from a traffic network topology set, an event propagation initial state set is constructed, thereby avoiding the problem that previous methods only rely on average flow or average speed indexes and ignore the initial disturbance difference of events; further, by diffusing and deducing the event propagation initial state set under the network node and path space adjacency relationship, an event diffusion stage feature set is generated, layered representation of event propagation dynamics is realized, and the resilience level of the traffic system under different event conditions can be clearly shown.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a regional traffic system resilience evaluation and simulation deduction method based on eccentric orbit satellites. BACKGROUND

[0002] In modern regional traffic system research, how to quickly grasp the accident influence range and propagation process after the occurrence of a sudden event has become a key direction of traffic resilience analysis and emergency management. With the development of satellite remote sensing and traffic simulation technology, the observation ability based on eccentric orbit satellites is gradually introduced into the field of traffic system evaluation. Among them, "eccentric orbit satellite driven traffic sudden event space-time diffusion graph construction" is a new method proposed in this background. This direction emphasizes making full use of the characteristics of multi-time and space resolution and coverage difference of eccentric orbit satellites to describe and visualize the propagation law of traffic sudden events in time and space dimensions, providing more dynamic and comprehensive support for traffic system resilience evaluation.

[0003] At present, in the analysis and management of traffic sudden events, the existing methods mostly rely on single-time ground sensor data or static observation results of a single satellite. Although this kind of means can reflect the congestion around the accident point, it can only provide local information of a certain time slice, lacking continuity and globality. As a result, only the static congestion caused by the accident can be seen, and it is impossible to clearly grasp how the traffic influence gradually spreads over time, conducts to upstream and downstream, and further affects the adjacent road network. This deficiency makes the emergency management department often lag behind the actual situation when diverting and scheduling or warning, which easily leads to the rapid evolution of local problems into a larger range of systematic failure. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a regional traffic system resilience evaluation and simulation deduction method based on eccentric orbit satellites, which solves the problems mentioned in the background art.

[0005] To achieve the above purpose, the present application realizes the following technical scheme: a regional traffic system resilience evaluation and simulation deduction method based on eccentric orbit satellites, comprising the following steps:

[0006] S1, collecting traffic running data acquired by eccentric orbit satellites under multi-orbit coverage to form a traffic state feature set;

[0007] S2, performing time-space sequence preprocessing on the traffic state feature set, generating an event influence feature set in combination with the coverage time difference of the eccentric orbit satellite, and matching with the extracted traffic network topology set to construct an event propagation initial state set;

[0008] S3, generating an event diffusion stage feature set according to the event propagation initial state set and the spatial adjacency relationship between the nodes and paths in the generated traffic network topology set;

[0009] S4, calculating a resilience index set after processing based on the event diffusion stage feature set and performing evaluation and use.

[0010] Preferably, the S1 includes S11;

[0011] S11, based on the multi-orbit coverage of the inclined orbit satellite, periodically observing the target area to obtain remote sensing original image Img;

[0012] According to the multi-orbit coverage plan of the inclined orbit satellite, record the time stamp and sampling interval of each observation to form a continuous observation time sequence as a time sequence index Tim;

[0013] Using the basic geographic information system data of regional traffic, combining the road boundary recognition result in the remote sensing original image Img, establishing the spatial coordinate mapping relationship according to the road number and node number, forming the road space index Loc;

[0014] Through optical multi-frame difference processing of the remote sensing original image Img, the displacement trajectory of the vehicle in the continuous frame is extracted, and the traffic speed sequence Spe is calculated according to the trajectory interval and the observation time difference;

[0015] At the same time, the remote sensing original image Img is analyzed by microwave remote sensing echo intensity, and the echo change rate of the road section per unit time is counted to generate the traffic flow sequence Flo;

[0016] The remote sensing original image Img is segmented according to the road vector boundary information to obtain the road vector segmentation result Seg, and the proportion of road pixels covered by vehicles in the segmentation result is calculated to obtain the road occupancy rate sequence Occ;

[0017] Integrate the traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy rate sequence Occ to obtain the initial traffic operation data set.

[0018] Preferably, the S1 further includes S12;

[0019] S12, performing consistency detection and noise correction on the initial traffic operation data set:

[0020] Among them, the consistency detection takes the time sequence index Tim as the basis to perform stationarity test on the traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy rate sequence Occ, and adopts ADF test and autocorrelation function analysis to eliminate abnormal paragraphs that do not conform to the time sequence rule;

[0021] The noise correction applies a wavelet threshold denoising method to the traffic speed sequence Spe, traffic flow sequence Flo and road occupancy sequence Occ sequence data after consistency detection within the spatial range defined by the road space index Loc to correct high-frequency random noise;

[0022] The consistency detection and noise correction of the traffic speed sequence Spe, traffic flow sequence Flo and road occupancy sequence Occ are uniformly integrated according to the time sequence index Tim and the road space index Loc to form a traffic state feature set Dat.

[0023] Preferably, S2 includes S21;

[0024] S21, based on the traffic state feature set Dat, combines the time sequence index Tim and the road space index Loc to identify abnormal fluctuations in traffic operation:

[0025] Wherein, the identification of abnormal fluctuations is completed through steps S211, S212 and S213; the definition of the observation period Cyc, the continuous observation period threshold m, the single observation period and the window length w is based on the time sequence index Tim;

[0026] The observation period Cyc is the smallest time sampling interval in the time sequence index Tim;

[0027] The continuous observation period threshold m, based on the observation period Cyc, represents the minimum time length of continuous occurrence when the speed drops suddenly;

[0028] The single observation period represents an observation period Cyc, which is used to determine that the change needs to occur within one time step in the flow mutation feature Ffl;

[0029] The window length w represents an integer multiple of the observation period Cyc, and is used to count the occupancy rate anomaly through a detection window composed of a plurality of observation periods.

[0030] Preferably, S211 defines the observation period Cyc as the smallest sampling interval determined by the time sequence index Tim, and differentiates the traffic speed sequence Spe according to the time sequence index Tim to obtain the speed change rate sequence Δv(t, Loc) of the time point t in the road space index Loc;

[0031] Wherein, the specific calculation formula of the speed change rate sequence Δv(t, Loc) of the time point t in the road space index Loc is as follows:

[0032] Δv(t, Loc) = (Spe(t, Loc) - Spe(t-1, Loc)) ÷ Spe(t-1, Loc);

[0033] In the formula, Spe(t, Loc) and Spe(t-1, Loc) represent the traffic speed sequence Spe at time point t and time t-1 at road space index Loc respectively;

[0034] When the speed change rate sequence Δv(t, Loc) at time point t at road space index Loc is less than a preset -α within a continuous observation period threshold m, where α represents a preset speed drop threshold, it is determined that the traffic speed sequence Spe at time point t at road space index Loc is abnormal, and the traffic speed sequence Spe is marked as a speed drop feature Fsp;

[0035] S212, the traffic flow sequence Flo is differentially processed according to the time sequence index Tim, to obtain the flow change amount Δf(t, Loc) at time point t at road space index Loc;

[0036] Wherein, the flow change amount Δf(t, Loc) at time point t at road space index Loc is specifically calculated as follows:

[0037] Δf(t, Loc) = Flo(t, Loc) - Flo(t-1, Loc);

[0038] In the formula, Flo(t, Loc) and Flo(t-1, Loc) represent the traffic flow sequence Flo at time point t and time t-1 at road space index Loc respectively;

[0039] When the absolute value of the flow change amount Δf(t, Loc) at time point t at road space index Loc is greater than a preset flow mutation threshold γ within a single observation period Cyc, and the change occurs within a single observation period, it is determined that the traffic flow sequence Flo at time point t at road space index Loc is abnormal, and the traffic flow sequence Flo is marked as a flow mutation feature Ffl;

[0040] S213, in the road occupancy rate sequence Occ, the mean μ and the standard deviation σ of the road occupancy rate sequence Occ(t, Loc) at time point t at road space index Loc are calculated, to establish a reference for abnormality determination;

[0041] A detection window with an observation period Cyc as a basic unit is reconstructed, and the window length is defined as w = q × Cyc; Wherein q is an integer;

[0042] In each detection window, the values of the road occupancy rate sequence Occ(t, Loc) of the time point t at the road space index Loc are checked one by one, and when the values in the detection window are all greater than μ+k×σ, it is determined that the road occupancy rate sequence Occ(t, Loc) of the time point t at the road space index Loc has persistent abnormality in the time interval [t, t+w], and the road occupancy rate sequence Occ is marked as the occupancy rate abnormality feature Foc;

[0043] wherein k is an abnormality determination coefficient;

[0044] After integrating the speed sudden drop feature Fsp, the flow mutation feature Ffl and the occupancy rate abnormality feature Foc, the event influence feature set Evt is obtained.

[0045] Preferably, the S2 further comprises S22;

[0046] S22, based on the event influence feature set Evt, combining the three types of features with the topological structure of the road system to map, determine the initial distribution of the event disturbance in the traffic network, and form the event propagation initial state set Ini based on the extracted traffic network topology set Net;

[0047] wherein the traffic network topology set Net is based on the geographic information system data of the regional traffic, extracts the spatial connection relationship of the road nodes and the road edges, and establishes a one-to-one correspondence index in combination with the road space index Loc to form the traffic network topology set Net;

[0048] wherein the road nodes include intersections and entrances and exits; and the road edges include road segments and lane segments;

[0049] The event propagation initial state set Ini is specifically mapped by steps S221 and S222;

[0050] S221, with the road space index Loc as a reference, the three types of features in the event influence feature set Evt are mapped to the corresponding nodes or edges in the traffic network topology set Net one by one;

[0051] S222, after completing the mapping of step S221, based on the adjacency relationship in the traffic network topology set Net, the edges and neighbor nodes directly connected to the affected nodes are identified, and the edges and neighbor nodes are included in the initial affected range to form the event propagation initial state set Ini.

[0052] Preferably, the S3 comprises S31;

[0053] S31, based on the event propagation initial state set Ini, using the adjacency relationship of the traffic network topology set Net, an event propagation space-time dynamics model Dyn is established;

[0054] The event propagation spatiotemporal dynamics model Dyn is established through steps S311 and S312;

[0055] S311, in the traffic network topology set Net, all nodes marked as disturbed by the event propagation initial state set Ini are defined as node i;

[0056] Through the adjacency relationship of the traffic network topology set Net, the nodes directly connected to node i are determined, denoted as node j;

[0057] When a road edge connecting node i and node j exhibits any one of the speed sudden drop feature Fsp, the flow mutation feature Ffl or the occupancy abnormal feature Foc within the observation period Cyc, it is determined that the road edge exhibits abnormality, and the event disturbance has the transmission condition of propagating from node i to node j;

[0058] S312, after satisfying the disturbance transmission condition, the probability of disturbance propagation from node i to node j is calculated, and the probability value P(t, i→j) of event disturbance propagation from node i to node j at time t is obtained;

[0059] The specific calculation formula is as follows:

[0060]

[0061] In the formula, e represents a sigmoid function, △Spe, △Flo and △Occ represent the relative difference of the traffic speed sequence Spe, the relative difference of the traffic flow sequence Flo and the relative difference of the road occupancy rate sequence Occ respectively; a, b and c represent the weight coefficients of the relative difference △Spe of the traffic speed sequence Spe, the relative difference △Flo of the traffic flow sequence Flo and the relative difference △Occ of the road occupancy rate sequence Occ respectively, and a+b+c=1, the specific numerical value is set by the user; θ represents a threshold parameter;

[0062] Wherein, the relative difference △Spe of the traffic speed sequence Spe is obtained by the calculation formula ΔSpe=|Spe(t,i)-Spe(t,j)|÷Spe(t,i); in the formula, Spe(t,i) and Spe(t,j) represent the traffic speed sequence Spe at node i and node j at time point t respectively;

[0063] The relative difference △Flo of the traffic flow sequence Flo is obtained by the calculation formula ΔFlo=|Flo(t,i)-Flo(t,j)|÷Flo(t,i); in the formula, Flo(t,i) and Flo(t,j) represent the traffic flow sequence Flo at node i and node j at time point t respectively;

[0064] The relative difference ΔOcc of the road occupancy sequence Occ is obtained by the formula ΔOcc = |Occ(t, i) - Occ(t, j)| ÷ Occ(t, i); wherein, Occ(t, i) - Occ(t, j) respectively represent the road occupancy sequence Occ at the time point t at the node i and the node j.

[0065] Preferably, the S3 further comprises S32.

[0066] The S32 calculates the propagation speed Vp and the diffusion radius Rp of the event diffusion process based on the output of the event propagation spatiotemporal dynamics model Dyn using the probability value P(t, i→j), and performs integration processing to generate an event diffusion stage feature set Pro.

[0067] The propagation speed Vp is obtained by the following formula:

[0068]

[0069] In the formula, VP(t) represents the propagation speed at the time point t; E represents a disturbed edge set, which is determined by the event propagation spatiotemporal dynamics model Dyn at the time t, and is a set of all road edges where the disturbance propagation occurs; Cyc represents an observation period.

[0070]

[0071] In the formula, Rp(t) represents the diffusion radius at the time point t; V represents a potential disturbed node set, which is obtained by statistically outputting all disturbed nodes of the event propagation spatiotemporal dynamics model Dyn at the time point t; Nc represents a center node, which is a reference node selected in the event propagation initial state set Ini and serves as a reference node of the diffusion range; d(Nc, j) represents the shortest path distance from the center node Nc to the node j, wherein d represents a shortest path function; and P(t, Nc→j) represents the probability value of the disturbance propagating from the center node Nc to the node j.

[0072] Preferably, the S4 comprises S41.

[0073] The S41 performs normalization processing on the propagation speed Vp and the diffusion radius Rp using a min-max normalization method based on the event diffusion stage feature set Pro, eliminates the dimensional difference between different data, and inputs the normalized propagation speed Vp and diffusion radius Rp into an evaluation function to calculate the resilience level of the traffic system and output a resilience index set Res.

[0074] The evaluation function has the following specific form:

[0075]

[0076] In the formula, r1 and r2 respectively represent the weight coefficients of the normalized propagation speed Vp and the diffusion radius Rp, the specific values of which are set by the user, and r1+r2=1; max(Vp) and max(Rp) respectively represent the maximum values of the propagation speed Vp and the diffusion radius Rp in the entire Tim range.

[0077] Preferably, the S4 further comprises S42;

[0078] S41, the toughness index set Res is converted into a toughness evaluation value in percentage form, which is quantitatively expressed in the interval of 0-100%;

[0079] When the toughness evaluation value is high, it indicates that the regional traffic system resilience is strong in the resilience level under the emergency;

[0080] When the toughness evaluation value is low, it indicates that the regional traffic system resilience is weak in the resilience level under the emergency.

[0081] The present application provides a regional traffic system resilience evaluation and simulation deduction method based on inclined orbit satellites, which has the following beneficial effects:

[0082] (1) The multi-orbit coverage characteristics of inclined orbit satellites are used to collect fine-grained spatio-temporal data of traffic operation, form a traffic state feature set, and ensure high coverage rate and multi-dimensional integrity of the data source. Secondly, in the process of spatio-temporal sequence preprocessing of the traffic state feature set, the coverage time difference between inclined orbit satellite observations is fully considered, so as to generate an event influence feature set, and combined with the structure information extracted from the traffic network topology set, an initial state set of event propagation is constructed, avoiding the problem that the previous method only relies on average flow or average speed indicators and ignores the initial disturbance difference of the event. Further, through the diffusion deduction of the initial state set of event propagation under the adjacency relationship of network nodes and path space, an event diffusion stage feature set is generated, realizing the hierarchical representation of event propagation dynamics, rather than only estimating a single overall indicator. Finally, the toughness index set Res is calculated by taking the event diffusion stage feature set as input, and is directly quantified in percentage form for evaluation and use, which can clearly show the resilience level of the traffic system under different event conditions.

[0083] (2) Based on the traffic state feature set Dat, combined with the time index Tim and the road space index Loc, the observation period Cyc, the threshold value m of the number of continuous observation periods, the single observation period and the window length w can be strictly defined, so as to identify the event influence feature set Evt including the speed drop feature Fsp, the flow mutation feature Ffl and the occupancy abnormal feature Foc, and ensure that the determination conditions have traceability in time and space dimensions. Subsequently, the event influence feature set Evt is mapped to the traffic network topology set Net, and the initial state set Ini of event propagation is formed by adjacency relationship extension, so that the disturbance range of the event is directly related to the network structure, and the starting position and diffusion boundary of the event are accurately marked.

[0084] (3) Based on the adjacency relationship between the initial state set Ini of event propagation and the traffic network topology set Net, the established event propagation spatio-temporal dynamics model Dyn can not only determine whether the disturbance has the condition of propagating from node i to node j, but also convert the uncertainty of such propagation into a calculable quantitative index through the probability value P(t, i→j), avoiding the shortcomings of the existing methods that the propagation process can only rely on hard determination. On this basis, the propagation speed Vp and the diffusion radius Rp are further introduced to generate the event propagation stage feature set Pro, which can simultaneously depict the dynamic evolution process of the disturbance from the time efficiency and the spatial range. BRIEF DESCRIPTION OF DRAWINGS

[0085] Figure 1 The figure is a schematic diagram of the steps of the regional traffic system resilience evaluation and simulation deduction method based on the inclined orbit satellite of the present application.

[0086] Figure 2 The figure is a schematic diagram of the data processing P(t, i→j) data processing probability value data processing curve changing with time. DETAILED DESCRIPTION

[0087] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0088] Embodiment 1

[0089] The present application provides a regional traffic system resilience evaluation and simulation deduction method based on an inclined orbit satellite, please refer to Figure 1 , which comprises the following steps:

[0090] S1, collect the traffic operation data acquired by the inclined orbit satellite under multi-orbit coverage to form a traffic state feature set;

[0091] S2, spatiotemporal sequence preprocessing is performed on the traffic state feature set, event influence feature set is generated combined with the coverage time difference of the inclined orbit satellite, and is matched with the extracted traffic network topology set to construct an initial state set of event propagation;

[0092] S3, according to the initial state set of event propagation, combined with the spatial adjacency relationship of each node and path in the generated traffic network topology set, an event diffusion stage feature set is generated;

[0093] S4, after processing based on the event diffusion stage feature set, a resilience index set is calculated and generated, and is evaluated for use.

[0094] In this embodiment, the multi-orbit coverage characteristics of the inclined orbit satellite are used to collect fine-grained spatiotemporal data of traffic operation, form a traffic state feature set, and ensure high coverage rate and multi-dimensional integrity of the data source. Secondly, in the process of spatiotemporal sequence preprocessing of the traffic state feature set, the coverage time difference between the observations of the inclined orbit satellite is fully considered to generate an event influence feature set, and the structural information extracted from the traffic network topology set is combined to construct an initial state set of event propagation, avoiding the problem that the previous method only relies on average flow or average speed indicators and ignores the differences in initial disturbance of events. Further, by propagating and deducing the initial state set of event propagation under the spatial adjacency relationship of network nodes and paths, an event diffusion stage feature set is generated, which realizes the hierarchical representation of event propagation dynamics, rather than only estimating a single overall indicator. Finally, the resilience index set Res is calculated by taking the event diffusion stage feature set as input, and is directly quantified as a percentage for evaluation and use, which can clearly show the resilience level of the traffic system under different event conditions. For example, when a sudden traffic accident occurs on a core arterial road in the city, the traditional method may only give the numerical value of the overall average delay increase, while through this method, the resilience percentage performance of the accident on a specific time window and a specific road hierarchy (such as trunk road, secondary road, auxiliary road) can be accurately quantified in the simulation and deduction results, so that the traffic manager can intuitively identify the difference between the risk of rapid spread of the accident on the trunk road and the buffering effect of the secondary road, thereby providing a scientific basis for shunting and dispatching and emergency resource allocation.

[0095] Embodiment 2

[0096] Specifically, the S1 includes S11;

[0097] S11, based on the multi-orbit coverage of the inclined orbit satellite, periodically observing the target area to obtain remote sensing raw images Img;

[0098] According to the multi-orbit coverage plan of the inclined orbit satellite, the time stamp and sampling interval of each observation are recorded to form a continuous observation time sequence as a time sequence index Tim;

[0099] The road space index Loc is formed by using the basic geographic information system data of regional traffic, combining the road boundary recognition result in the remote sensing original image Img, and establishing a spatial coordinate mapping relationship according to the road number and node number;

[0100] The traffic speed sequence Spe is obtained by performing optical multi-frame difference processing on the remote sensing original image Img, extracting the displacement trajectory of the vehicle in the continuous frames, and calculating the traffic speed sequence according to the trajectory interval and the observation time difference;

[0101] The traffic flow sequence Flo is obtained by performing microwave remote sensing echo intensity analysis on the remote sensing original image Img, and counting the echo change rate of the road section in unit time;

[0102] The road vector segmentation result Seg is obtained by segmenting the remote sensing original image Img according to the road vector boundary information, and the road occupancy rate sequence Occ is obtained by calculating the proportion of the road pixels covered by the vehicle in the segmentation result;

[0103] The initial traffic operation data set is obtained by integrating the traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy rate sequence Occ;

[0104] It should be noted that:

[0105] The road space index Loc is a spatial marking method established based on road geographic information and image segmentation result, and each road and road node has a unique index number; it is used to ensure that the speed, flow and occupancy rate correspond to the same road or node in space; if there is no road space index Loc, different traffic parameters cannot be mapped to a unified road network, and the subsequent diffusion path deduction will be distorted;

[0106] The remote sensing original image Img is a continuous time sequence image of the target region obtained by a cross-track satellite sensor under multi-orbit coverage, which is the only basic source data for traffic state parameter extraction;

[0107] The road vector segmentation result Seg is obtained by spatial segmentation based on the remote sensing original image Img and road basic geographic information, and is used to identify the road range, so as to count the vehicle pixel ratio in the road area; if there is no road vector segmentation result Seg, the vehicle pixel cannot be corresponded to the road boundary, and the calculated road occupancy rate sequence Occ will be distorted or even invalid;

[0108] The traffic speed sequence Spe is a vehicle average speed sequence obtained by optical image frame difference calculation, which is used to reflect the road passing efficiency and is a core index for measuring the dynamicity of traffic state; if the traffic speed sequence Spe is missing, the direct impact of the sudden event on the road operation efficiency cannot be judged;

[0109] Traffic flow sequence Flo is a vehicle passing quantity sequence per unit time obtained by microwave remote sensing echo statistics, and is used to reflect road capacity and traffic carrying condition; if the traffic flow sequence Flo is absent, the carrying limit of the traffic network at the time of the event cannot be quantified;

[0110] Road occupancy sequence Occ is a road utilization intensity sequence calculated by the road pixel occupancy in the road vector segmentation result Seg, and is used to reflect the saturation degree of road resources in actual use; if the road occupancy sequence Occ is absent, it cannot be revealed whether the traffic congestion is caused by local high occupancy, and it cannot support the diffusion path deduction.

[0111] The S1 further includes S12;

[0112] S12, consistency detection and noise correction are performed on the initial traffic operation data set:

[0113] The consistency detection takes the time sequence index Tim as the reference to perform stationarity test on the traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy sequence Occ, and adopts ADF test and autocorrelation function analysis to eliminate abnormal paragraphs that do not conform to the time sequence rule, so as to ensure that the sequence has modeling property in the time dimension;

[0114] The noise correction applies wavelet threshold denoising method to correct high-frequency random noise of the traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy sequence Occ sequence data in the spatial range defined by the road space index Loc after the consistency detection, and then combines the chi-square test to verify the residual distribution, so as to ensure that the sequence after noise processing conforms to the statistical characteristics;

[0115] The traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy sequence Occ after the consistency detection and the noise correction are uniformly integrated according to the time sequence index Tim and the road space index Loc, to form a traffic state feature set Dat;

[0116] It should be noted that:

[0117] The traffic state feature set Dat is uniformly integrated by the traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy sequence Occ after the consistency detection and the noise correction under the double constraints of the time sequence index Tim and the road space index Loc; as the direct input of the subsequent event feature extraction, the consistency in the time and space dimensions is ensured.

[0118] In this embodiment, through the processing flow of S1, the data quality and spatial mapping accuracy can be significantly improved at the front end of the traffic system resilience assessment. Due to the dual constraints of the time index Tim and the road space index Loc, this method not only ensures the one-to-one correspondence of the traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy rate sequence Occ in the time and space dimensions, but also eliminates non-stationary segments and high-frequency noise through ADF test and wavelet threshold denoising, and finally forms the traffic state feature set Dat which is significantly better than existing methods in integrity and consistency. This advantage is particularly prominent in practical applications, for example, when an urban expressway experiences abnormal congestion during the morning peak, traditional traffic monitoring often leads to data misjudgment due to noise interference of a single sensor, while this method can accurately distinguish whether the capacity saturation is caused by the sudden increase of actual traffic flow Flo or the false signal caused by occasional interference in the remote sensing image, thereby avoiding misjudgment and false scheduling. This not only improves data reliability, but also provides a stable and reliable input basis for subsequent event propagation modeling and resilience index calculation.

[0119] Embodiment 3

[0120] Specifically, the S2 comprises S21;

[0121] S21, based on the traffic state feature set Dat, combined with the time index Tim and the road space index Loc, identifies abnormal fluctuations in traffic operation:

[0122] Wherein, the identification of abnormal fluctuations is completed through steps S211, S212 and S213; the definition of the observation period Cyc, the continuous observation period threshold m, the single observation period and the window length w is based on the time index Tim;

[0123] The observation period Cyc is the smallest time sampling interval in the time index Tim;

[0124] The continuous observation period threshold m is based on the observation period Cyc, representing the minimum time length of continuous occurrence when the speed drops suddenly;

[0125] The single observation period represents an observation period Cyc, which is used to determine that the change needs to occur within one time step in the flow mutation feature Ffl;

[0126] The window length w represents an integer multiple of the observation period Cyc, and is used to count the occupancy rate anomaly through a detection window composed of a plurality of continuous observation periods.

[0127] The S2 comprises S21;

[0128] S21, identifying abnormal fluctuation in traffic operation based on the traffic state feature set Dat, combining the time sequence index Tim and the road space index Loc;

[0129] Wherein, the identification of abnormal fluctuation is completed through steps S211, S212 and S213; the definition of observation period Cyc, the continuous observation period threshold m, the single observation period and the window length w is based on the time sequence index Tim;

[0130] The observation period Cyc is the smallest time sampling interval in the time sequence index Tim;

[0131] The continuous observation period threshold m represents the minimum time length of continuous occurrence of speed drop based on the observation period Cyc;

[0132] The single observation period represents an observation period Cyc, which is used to determine that the change needs to occur within one time step in the flow mutation feature Ffl;

[0133] The window length w represents an integer multiple of the observation period Cyc, and a detection window composed of a plurality of observation periods is used to count the occupancy rate anomaly.

[0134] S211, defining the observation period Cyc as the smallest sampling interval determined by the time sequence index Tim, and performing difference between adjacent time points on the traffic speed sequence Spe according to the time sequence index Tim to obtain the speed change rate sequence Δv(t, Loc) of the time point t at the road space index Loc;

[0135] Wherein, the specific calculation formula of the speed change rate sequence Δv(t, Loc) of the time point t at the road space index Loc is as follows:

[0136] Δv(t, Loc)=(Spe(t, Loc)-Spe(t-1, Loc))÷Spe(t-1, Loc);

[0137] In the formula, (Spe(t, Loc) and Spe(t-1, Loc) represent the traffic speed sequence Spe of the time point t and the time t-1 at the road space index Loc, respectively;

[0138] When the speed change rate sequence Δv(t, Loc) of the time point t at the road space index Loc is less than the preset -α within the continuous observation period threshold m, wherein α represents the preset speed drop threshold, it is determined that the traffic speed sequence Spe has an abnormality at the time point t at the road space index Loc, and the traffic speed sequence Spe is marked as the speed drop feature Fsp;

[0139] S212, difference processing is performed on the traffic flow sequence Flo according to the time sequence index Tim, to obtain a flow change amount Δf(t, Loc) of the time point t in the road space index Loc;

[0140] The flow change amount Δf(t, Loc) of the time point t in the road space index Loc is specifically calculated according to the following formula:

[0141] Δf(t, Loc) = Flo(t, Loc) - Flo(t-1, Loc);

[0142] In the formula, Flo(t, Loc) and Flo(t-1, Loc) respectively represent the traffic flow sequence Flo of the time point t and the time t-1 in the road space index Loc;

[0143] When the absolute value of the flow change amount Δf(t, Loc) of the time point t in the road space index Loc is greater than a preset flow mutation threshold γ within a single observation period Cyc, and the change occurs within a single observation period, it is determined that the traffic flow sequence Flo has an anomaly at the time point t in the road space index Loc, and the traffic flow sequence Flo is marked as a flow mutation feature Ffl;

[0144] S213, in the road occupancy rate sequence Occ, the mean μ and the standard deviation σ of the road occupancy rate sequence Occ(t, Loc) of the time point t in the road space index Loc are calculated, to establish a reference for anomaly determination;

[0145] A detection window is further constructed with the observation period Cyc as a basic unit, and the window length is defined as w=qxCyc; wherein q is an integer;

[0146] In each detection window, the values of the road occupancy rate sequence Occ(t, Loc) of the time point t in the road space index Loc are checked one by one, and when all the values in the detection window are greater than μ+k×σ, it is determined that the road occupancy rate sequence Occ(t, Loc) of the time point t in the road space index Loc has a persistent anomaly in the time interval [t, t+w], and the road occupancy rate sequence Occ is marked as an occupancy rate anomaly feature Foc;

[0147] In the formula, k is an anomaly determination coefficient;

[0148] After integrating the speed drop feature Fsp, the flow mutation feature Ffl, and the occupancy rate anomaly feature Foc, an event influence feature set Evt is obtained.

[0149] The S2 further includes S22;

[0150] S22, based on the event influence feature set Evt, the three types of features are combined with the topological structure of the road system to map, determine the initial distribution of the event disturbance in the traffic network, and form the initial state set Ini of the event propagation based on the extracted traffic network topology set Net;

[0151] The traffic network topology set Net is based on the geographic information system data of the regional traffic, extracts the spatial connection relationship of the road nodes and the road edges, and establishes a one-to-one index combined with the road space index Loc to form the traffic network topology set Net. The traffic network topology set Net includes a node set, an edge set, and an adjacency relationship between nodes, and is a network framework for event propagation deduction;

[0152] The road nodes include intersections and entrances and exits; the road edges include road segments and lane segments;

[0153] The initial state set Ini of the event propagation is specifically mapped by steps S221 and S222;

[0154] S221, with the road space index Loc as a reference, the three types of features in the event influence feature set Evt are mapped to the corresponding nodes or edges in the traffic network topology set Net one by one, for example, when the traffic speed sequence Spe is marked as a speed drop feature Fsp at time point t under the road space index Loc, the corresponding road node or road edge of the road space index Loc in the traffic network topology set Net is marked;

[0155] S222, after the mapping of step S221 is completed, based on the adjacency relationship in the traffic network topology set Net, the edges and neighbor nodes directly connected to the affected nodes are identified, and the edges and neighbor nodes are included in the initial affected range to form the initial state set Ini of the event propagation;

[0156] It should be noted that:

[0157] The traffic network topology set Net is a set composed of road nodes and road edges, each element corresponds to the road space index Loc, and is attached with adjacency relationship information; it represents that the road structure is extracted through the regional geographic information system data, and the index is established combined with Loc; it provides a structural framework for spatial mapping and propagation path of event influence features;

[0158] The initial state set Ini of the event propagation is obtained by mapping the event influence feature set Evt to the traffic network topology set Net combined with the adjacency relationship to expand the range; as the starting state of diffusion deduction, it defines the initial position and range of the event disturbance in the network; without the initial state set Ini of the event propagation, the diffusion model cannot establish the starting point of the propagation, and the event evolution process cannot be expanded.

[0159] In this embodiment, through the processing flow of S2, the method realizes accurate mapping from data features to network topology, making up for the deficiencies of unclear event disturbance positioning and ambiguous propagation starting point in the prior art. Specifically, based on the traffic state feature set Dat, combined with the time sequence index Tim and the road space index Loc, the observation period Cyc, the threshold value m of the number of consecutive observation periods, the single observation period and the window length w can be strictly defined, so as to identify the event influence feature set Evt including the speed drop feature Fsp, the flow mutation feature Ffl and the occupancy abnormal feature Foc, and ensure that its determination conditions have traceability in time and space dimensions. Subsequently, the event influence feature set Evt is mapped to the traffic network topology set Net, and the initial state set Ini of event propagation is formed by adjacency relationship expansion, so that the disturbance range of the event is directly related to the network structure, and the starting position and diffusion boundary of the event are accurately marked. In real traffic scenarios, the advantages of this mapping mechanism are very prominent. For example, when a multi-car rear-end accident occurs on an urban viaduct during the evening peak, the traditional method can only identify the congestion state of the road, but cannot accurately infer its immediate impact on adjacent ramps or downstream nodes; and the method can not only locate the road edge where the accident occurs, but also automatically identify the ramps and intersections directly adjacent to it, forming the starting disturbance range that can be used as input for diffusion deduction, thereby significantly improving the accuracy and practicality of subsequent event propagation simulation.

[0160] Embodiment 4

[0161] Please refer to Figure 1 and Figure 2 Specifically, the S3 comprises S31;

[0162] S31, on the basis of the initial state set Ini of event propagation, the adjacency relationship of the traffic network topology set Net is used to establish the spatio-temporal dynamics model Dyn of event propagation;

[0163] The establishment of the spatio-temporal dynamics model Dyn of event propagation is completed through steps S311 and S312;

[0164] S311, in the traffic network topology set Net, all nodes disturbed by the initial state set Ini of event propagation are defined as node i;

[0165] Through the adjacency relationship of the traffic network topology set Net, the nodes directly connected to the node i are determined and recorded as node j;

[0166] When a road edge connecting node i and node j has any one of the features Fsp, Ffl or Foc in the observation period Cyc, it is determined that the road edge exhibits abnormality, and the event disturbance has the transmission condition to propagate from node i to node j;

[0167] In S312, after the disturbance transmission condition is met, the probability of the disturbance propagating from node i to node j is calculated, and the probability value P(t, i→j) of the event disturbance propagating from node i to node j at time t is obtained, reflecting the propagation possibility of the propagation road edge abnormality.

[0168] The specific calculation formula is as follows:

[0169]

[0170] In the formula, e represents a sigmoid function, △Spe, △Flo and △Occ represent the relative differences of the traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy rate sequence Occ respectively; a, b and c represent the weight coefficients of the relative differences △Spe, △Flo and △Occ of the traffic speed sequence Spe, the traffic flow sequence Flo and the road occupancy rate sequence Occ respectively, and a+b+c=1, and the specific values are set by the user; θ represents a threshold parameter for controlling the cumulative difference intensity that the disturbance needs to reach to significantly propagate;

[0171] The relative difference △Spe of the traffic speed sequence Spe is obtained by the calculation formula ΔSpe=|Spe(t,i)-Spe(t,j)|÷Spe(t,i); in the formula, Spe(t,i) and Spe(t,j) represent the traffic speed sequence Spe at node i and node j at time point t respectively.

[0172] The relative difference △Flo of the traffic flow sequence Flo is obtained by the calculation formula ΔFlo=|Flo(t,i)-Flo(t,j)|÷Flo(t,i); in the formula, Flo(t,i) and Flo(t,j) represent the traffic flow sequence Flo at node i and node j at time point t respectively.

[0173] The relative difference △Occ of the road occupancy rate sequence Occ is obtained by the calculation formula ΔOcc=|Occ(t,i)-Occ(t,j)|÷Occ(t,i); in the formula, Occ(t,i) and Occ(t,j) represent the road occupancy rate sequence Occ at node i and node j at time point t respectively.

[0174] It should be noted that:

[0175] Node i has been marked as a disturbed node in the initial state set Ini of event propagation; is the mapping result of Evt→ Ini; as the propagation source point of disturbance;

[0176] Node j is a node that has a direct adjacency relationship with node i in the traffic network topology set Net; is the adjacency relationship of Net; as a potential target of disturbance propagation;

[0177] Road edge (i, j) connects the topological channel of node i and node j; is the edge set of Net; as the carrying path of three types of feature propagation;

[0178] Probability value P(t, i→j) is the probability value of event disturbance propagating from node i to node j through road edge (i, j) at time t; is the quantification of the uncertainty and strength of propagation; if there is no probability value P(t, i→j), the propagation process can only be a hard decision, lacking the characterization of the dynamic characteristics of the traffic system;

[0179] When the probability value P(t, i→j) of node i propagating to node j is much smaller than the threshold parameter θ, P(t, i→j) approaches 0, indicating that the propagation almost does not occur;

[0180] When the probability value P(t, i→j) of node i propagating to node j approaches the threshold parameter θ, P(t, i→j) rapidly rises, and the propagation possibility significantly increases;

[0181] When the probability value P(t, i→j) of node i propagating to node j is much greater than the threshold parameter θ, P(t, i→j) approaches 1, indicating that the propagation almost certainly occurs.

[0182] The S3 further includes S32;

[0183] S32, based on the output of the event propagation spatiotemporal dynamics model Dyn, uses the probability value P(t, i→j) to calculate the propagation speed Vp and the diffusion radius Rp of the event diffusion process, and performs integration processing to generate an event diffusion stage feature set Pro;

[0184] The propagation speed Vp is obtained by the following calculation formula:

[0185]

[0186] In the formula, VP(t) represents the propagation speed at time point t; E represents the disturbed edge set, which is determined by the event propagation spatiotemporal dynamics model Dyn at time t, and is the set of all disturbed propagation road edges, which is based on the disturbed node i and its adjacent node j; Cyc represents the observation period; the physical meaning of the formula is that the propagation speed Vp(t) describes the diffusion efficiency of the disturbance through the traffic network in unit time;

[0187] When the propagation speed Vp(t) is large, it means that a large number of roadside areas may participate in the propagation of disturbances in a very short time, and the system is more likely to experience cascading failures.

[0188] When the propagation speed Vp(t) is small or close to zero, it indicates that the disturbance is in local propagation or has been effectively suppressed, and the system has better immediate resistance capability.

[0189]

[0190] In the formula, Rp(t) represents the diffusion radius at time t, reflecting the spatial scale of the disturbance propagating outward from the central node; V represents the set of potentially disturbed nodes, which is obtained by statistically analyzing all disturbed nodes output by the event propagation spatiotemporal dynamics model Dyn at time t; Nc represents the central node, which is selected as the reference node for the diffusion range by the reference node in the initial state set Ini of the event propagation, usually the first disturbed node or the geometric / topological center of the disturbed region; d(Nc,j) represents the shortest path distance from the central node Nc to node j, where d represents the shortest path function, including processing using Dijkstra or BFS; P(t,Nc→j) represents the probability value of the disturbance propagating from the central node Nc to node j; the physical meaning of this formula is that the diffusion radius Rp(t) describes the spatial scale of the disturbance's influence range.

[0191] When the diffusion radius Rp(t) continues to increase, it indicates that the scope of the event's influence is expanding, more road nodes are affected, and the spatial stability of the system is significantly reduced.

[0192] When the diffusion radius Rp(t) remains stable or shrinks, it indicates that the propagation is limited or gradually declines, and the system's resilience begins to emerge.

[0193] In this embodiment, through the processing flow of S3, the method realizes the spatiotemporal dynamic quantitative modeling of disturbance propagation in traditional traffic incident analysis for the first time. Based on the adjacency relationship between the initial state set Ini of incident propagation and the topological set Net of the traffic network, the established spatiotemporal dynamic model Dyn of incident propagation can not only determine whether the disturbance has the condition to propagate from node i to node j, but also convert the uncertainty of such propagation into a calculable quantitative index through the probability value P(t, i→j), avoiding the deficiency of the existing method that “the propagation process can only rely on hard determination”. On this basis, the propagation speed Vp and the diffusion radius Rp are further introduced to generate the characteristic set Pro of the event diffusion stage, which can simultaneously describe the dynamic evolution process of the disturbance from the angles of time efficiency and spatial range. In real applications, the value of this capability is very outstanding. For example, when a sudden flow aggregation occurs at the entrance of a subway in the city, causing the traffic flow of the surrounding roads to increase sharply, the traditional method can only reflect the instantaneous congestion of the node, but cannot predict the diffusion speed and range of the congestion to the adjacent roads. The method can accurately determine whether the incident will form a cascading propagation in a short time through the propagation speed Vp, and predict the range of the affected roads by using the diffusion radius Rp, so as to provide the decision basis for the city traffic managers to deploy the detour scheme in advance and add emergency channels. This capability of dynamically quantifying the propagation efficiency and spatial range of influence enables the resilience evaluation to be upgraded from static detection to dynamic control of the whole process of the sudden event.

[0194] Embodiment 5

[0195] Specifically, the S4 comprises S41;

[0196] S41, based on the characteristic set Pro of the event diffusion stage, the propagation speed Vp and the diffusion radius Rp are normalized by using the min-max normalization method to eliminate the dimensional difference between different data, and the normalized propagation speed Vp and diffusion radius Rp are combined to input the evaluation function to calculate the resilience level of the traffic system, and output the resilience index set Res;

[0197] The specific form of the evaluation function is as follows:

[0198]

[0199] In the formula, r1 and r2 respectively represent the weight coefficients of the normalized propagation speed Vp and the diffusion radius Rp, the specific values of which are set by the user, and r1+r2=1; max(Vp) and max(Rp) respectively represent the maximum values of the propagation speed Vp and the diffusion radius Rp in the entire Tim range; the physical meaning of the formula is that the greater the propagation speed Vp or the greater the diffusion radius Rp, the faster the event diffusion and the wider the influence range, and the lower the system resilience; therefore, the inverse form is used to represent that the resilience is inversely proportional to the resilience; the maximum values max(Vp) and max(Rp) are taken to reflect the most unfavorable diffusion state in the entire Tim range, so as to ensure that the evaluation result covers the extreme disturbance situation.

[0200] The S4 further includes S42;

[0201] S42, the resilience index set Res is converted into a percentage form of the resilience evaluation value, which is quantitatively expressed in the 0-100% interval;

[0202] When the resilience evaluation value is high, it indicates that the resilience level of the regional transportation system resilience under the emergency is strong;

[0203] When the resilience evaluation value is low, it indicates that the resilience level of the regional transportation system resilience under the emergency is weak.

[0204] In this embodiment, through the processing flow of S4, the method effectively solves the problem that the evaluation results are not comparable due to the non-uniform dimensions of different indexes in the resilience evaluation stage. Specifically, based on the event diffusion stage feature set Pro, the min-max normalization method is used to normalize the propagation speed Vp and the diffusion radius Rp, which not only eliminates the dimensional difference between the two, but also forms the resilience index set Res by inputting the evaluation function, ensuring the mathematical interpretability and fairness of the evaluation result. Further, the resilience index set Res is converted into a percentage form of the resilience evaluation value, which is quantitatively expressed in the intuitive 0-100% interval to reflect the resilience level of the regional transportation system, so that the complex propagation dynamics results can be directly understood and used by the management department. In real-world scenarios, this quantitative effect is particularly important, for example, when a certain urban area encounters sudden heavy rain causing multiple road waterlogging, the traditional evaluation method often only gives a qualitative "resilience is weak" conclusion, which is difficult to support cross-department emergency response; while the resilience evaluation value of the present method can clearly indicate that the resilience is only 45%, which intuitively reflects that the system is in a critical fragile state, thereby providing quantitative decision-making basis for the transportation management department to quickly dispatch drainage vehicles and evacuate traffic flow. This conversion from complex calculation to intuitive percentage quantization is a unique advantage that cannot be achieved by existing technologies.

[0205] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A method for resilience assessment and simulation of regional transportation systems based on heterogeneous satellite orbits, characterized by: Includes the following steps: S1. Collect traffic operation data obtained by satellites with different orbits under multi-orbit coverage to form a set of traffic status features; S2. Perform spatiotemporal sequence preprocessing on the traffic state feature set, combine the coverage time difference of satellites with different orbits to generate an event impact feature set, and match it with the extracted traffic network topology set to construct the initial state set for event propagation. S3. Based on the initial state set of event propagation and combined with the spatial adjacency relationships between each node and path in the generated traffic network topology set, generate a feature set for the event propagation stage. S4. After processing the feature set based on the event diffusion stage, calculate and generate a set of resilience indicators, and then use them for evaluation.

2. The method for assessing and simulating the resilience of regional transportation systems based on heterogeneous satellites according to claim 1, characterized in that: S1 includes S11; S11. Based on multi-orbit coverage by satellites with different orbits, periodic observations are conducted on the target area to obtain raw remote sensing images (Img). Based on the multi-orbit coverage plan of the satellites with different orbits, the timestamps and sampling intervals of each observation are recorded to form a continuous observation time series, which serves as the time series index Tim. Using basic geographic information system data of regional transportation, combined with road boundary identification results in remote sensing raw image Img, a spatial coordinate mapping relationship is established according to road number and node number to form road spatial index Loc; By performing optical multi-frame differential processing on the original remote sensing image Img, the displacement trajectory of vehicles in consecutive frames is extracted, and the traffic speed sequence Spe is calculated based on the trajectory spacing and the observation time difference. Simultaneously, microwave remote sensing echo intensity analysis was performed on the original remote sensing image Img, and the echo change rate on the road cross section per unit time was statistically analyzed to generate a traffic flow sequence Flo. The original remote sensing image Img is segmented according to the road vector boundary information to obtain the road vector segmentation result Seg. The proportion of road pixels covered by vehicles in the segmentation result is calculated to obtain the road occupancy rate sequence Occ. Integrate the traffic speed sequence Spe, the traffic flow sequence Flo, and the road occupancy rate sequence Occ to obtain an initial traffic operation data set.

3. The method for assessing and simulating the resilience of regional transportation systems based on heterogeneous satellites according to claim 2, characterized in that: S1 further includes S12; S12. Perform consistency checks and noise corrections on the initial traffic operation dataset: Among them, the consistency test uses the time series index Tim as the benchmark to conduct stationarity tests on the traffic speed series Spe, traffic flow series Flo, and road occupancy rate series Occ. The ADF test and autocorrelation function analysis are used to remove abnormal segments that do not conform to the time series pattern. Within the spatial range defined by the Road Spatial Index (Loc), wavelet thresholding is applied to correct high-frequency random noise in the traffic speed sequence (Spe), traffic flow sequence (Flo), and road occupancy rate sequence (Occ) after consistency detection. The traffic speed sequence Spe, traffic flow sequence Flo, and road occupancy rate sequence Occ, after consistency detection and noise correction, are integrated according to the time index Tim and the road spatial index Loc to form the traffic state feature set Dat.

4. The method for assessing and simulating the resilience of regional transportation systems based on heterogeneous satellites according to claim 3, characterized in that: S2 includes S21; S21. Based on the traffic state feature set Dat, combined with the time-series index Tim and the road spatial index Loc, identify abnormal fluctuations in traffic operation: The identification of abnormal fluctuations is completed through steps S211, S212 and S213; the observation period Cyc, the threshold m for the number of consecutive observation periods, the single observation period and the window length w are defined based on the time series index Tim; The observation period Cyc is the smallest time sampling interval in the time series index Tim; The threshold m for the number of consecutive observation periods is based on the observation period Cyc, representing the minimum time length during which a sudden drop in velocity occurs consecutively. A single observation period represents one observation period Cyc, which is used in the flow mutation feature Ffl to determine that the change must occur within a time step; The window length w represents an integer multiple of the observation period Cyc. The detection window, composed of multiple consecutive observation periods, is used to statistically analyze occupancy anomalies.

5. The method for assessing and simulating the resilience of regional transportation systems based on heterogeneous satellites according to claim 4, characterized in that: S211. Define the observation period Cyc as the minimum sampling interval determined by the time index Tim. Perform adjacent time difference on the traffic speed sequence Spe according to the time index Tim to obtain the speed change rate sequence Δv(t, Loc) at time point t in the road spatial index Loc. The specific calculation formula for the velocity change rate sequence Δv(t, Loc) at time point t in the road spatial index Loc is as follows: Δv(t,Loc)=(Spe(t,Loc)-Spe(t-1,Loc))÷Spe(t-1,Loc); In the formula, (Spe(t, Loc) and Spe(t-1, Loc) represent the traffic speed sequences Spe at time point t and time t-1 at the road spatial index Loc, respectively; When the speed change rate sequence Δv(t, Loc) at time point t in the road spatial index Loc is less than the preset -α within the consecutive observation period threshold m, where α represents the preset speed drop threshold, it is determined that the traffic speed sequence Spe has an anomaly at time point t in the road spatial index Loc, and the traffic speed sequence Spe is marked as a speed drop feature Fsp. S212. Perform differential processing on the traffic flow sequence Flo according to the time index Tim to obtain the traffic flow change Δf(t, Loc) at the road spatial index Loc at time point t. The specific calculation formula for the traffic flow change Δf(t, Loc) at time point t in the road spatial index Loc is as follows: Δf(t,Loc)=Flo(t,Loc)-Flo(t-1,Loc); In the formula, Flo(t, Loc) and Flo(t-1, Loc) represent the traffic flow sequences Flo at time point t and time t-1 at the road spatial index Loc, respectively; When the absolute value of the traffic flow change Δf(t, Loc) at time point t in the road spatial index Loc is greater than the preset traffic flow mutation threshold γ within a single observation period Cyc, and the change occurs within a single observation period, it is determined that the traffic flow sequence Flo has an anomaly at time point t in the road spatial index Loc, and the traffic flow sequence Flo is marked as the traffic flow mutation feature Ffl. S213. In the road occupancy rate sequence Occ, calculate the mean μ and standard deviation σ of the road occupancy rate sequence Occ(t,Loc) at the road spatial index Loc at time point t, which are used to establish the benchmark for anomaly judgment. Then, construct a detection window with the observation period Cyc as the basic unit, and define the window length as w = q × Cyc; where q is an integer; Within each detection window, the values ​​of the road occupancy rate sequence Occ(t, Loc) at time point t in the road spatial index Loc are checked one by one. When all values ​​in the detection window are greater than μ+k×σ, it is determined that the road occupancy rate sequence Occ(t, Loc) at time point t in the road spatial index Loc has a persistent anomaly in the time interval [t, t+w], and the road occupancy rate sequence Occ is marked as the occupancy rate anomaly feature Foc. Where k is the anomaly detection coefficient; After integrating the speed drop feature Fsp, traffic mutation feature Ffl, and occupancy anomaly feature Foc, the event impact feature set Evt is obtained.

6. The method for resilience assessment and simulation of regional transportation systems based on heterogeneous satellites according to claim 5, characterized in that: S2 further includes S22; S22. Based on the event impact feature set Evt, the three types of features are combined with the topology of the road system for mapping to determine the initial distribution of event disturbances in the traffic network, and the initial state set Ini of event propagation is formed based on the extracted traffic network topology set Net. Among them, the traffic network topology set Net is based on the geographic information system data of regional traffic, extracts the spatial connection relationship between road nodes and road edges, and establishes a one-to-one corresponding index in combination with the road spatial index Loc to form the traffic network topology set Net; Among them, road nodes include intersections and entrances / exits; roadside includes road sections and lane sections; The initial state set Ini for event propagation is specifically mapped through steps S221 and S222. S221. Using the road spatial index Loc as a reference, map the three types of features in the event impact feature set Evt to the corresponding nodes or edges in the traffic network topology set Net. S222. After completing the mapping in step S221, based on the adjacency relationship in the traffic network topology set Net, identify the edges and neighboring nodes directly connected to the affected nodes, and include the edges and neighboring nodes in the initial affected range to form the initial state set Ini for event propagation.

7. The method for assessing and simulating the resilience of regional transportation systems based on heterogeneous satellites according to claim 6, characterized in that: S3 includes S31; S31. Based on the initial state set Ini of event propagation, establish the spatiotemporal dynamic model Dyn of event propagation using the adjacency relationship of the traffic network topology set Net. The establishment of the event propagation spatiotemporal dynamics model Dyn is completed through steps S311 and S312; S311. In the traffic network topology set Net, all nodes marked as disturbed by the initial state set Ini of event propagation are defined as node i. By using the adjacency relationships of the traffic network topology set Net, the node directly connected to node i is determined and denoted as node j; If a roadside connecting node i and node j exhibits any one of the following characteristics within the observation period Cyc: a sudden drop in speed (Fsp), a sudden change in traffic flow (Ffl), or an abnormal occupancy rate (Foc), then the roadside is determined to be abnormal, and the event disturbance has the conditions to propagate from node i to node j. S312. After satisfying the disturbance propagation condition, calculate the probability that the disturbance propagates from node i to node j, and obtain the probability value P(t, i→j) of the event disturbance propagating from node i to node j at time t. The specific calculation formula is as follows: In the formula, e represents the sigmoid function; △Spe, △Flo, and △Occ represent the relative differences of the traffic speed sequence Spe, the traffic flow sequence Flo, and the road occupancy sequence Occ, respectively; a, b, and c represent the weighting coefficients of the relative differences △Spe, △Flo, and △Occ of the traffic speed sequence Spe, the traffic flow sequence Flo, and the road occupancy sequence Occ, respectively, and a+b+c=1, with specific values ​​set by the user; θ represents the threshold parameter. The relative difference ΔSpe of the traffic speed sequence Spe is obtained by the formula ΔSpe=|Spe(t,i)-Spe(t,j)|÷Spe(t,i); where Spe(t,i) and Spe(t,j) represent the traffic speed sequences Spe at node i and node j at time point t, respectively. The relative difference ΔFlo of the traffic flow sequence Flo is obtained by the formula ΔFlo=|Flo(t,i)-Flo(t,j)|÷Flo(t,i); where Flo(t,i) and Flo(t,j) represent the traffic flow sequences Flo at node i and node j at time point t, respectively. The relative difference ΔOcc of the road occupancy rate sequence Occ is obtained by the formula ΔOcc=|Occ(t,i)-Occ(t,j)|÷Occ(t,i); where Occ(t,i)-Occ(t,j) represent the road occupancy rate sequences Occ at node i and node j at time point t, respectively.

8. The method for resilience assessment and simulation of regional transportation systems based on heterogeneous satellites according to claim 7, characterized in that: S3 further includes S32; S32. Based on the output of the event propagation spatiotemporal dynamics model Dyn, use the probability value P(t, i→j) to calculate the propagation speed Vp and diffusion radius Rp of the event diffusion process, and integrate them to generate the event diffusion stage feature set Pro. The propagation speed Vp is obtained by the following formula: In the formula, VP(t) represents the propagation speed at time t; E represents the set of disturbed edges, which is the set of all road edges where disturbances propagate, determined by the event propagation spatiotemporal dynamics model Dyn at time t; Cyc represents the observation period. In the formula, Rp(t) represents the diffusion radius at time t; V represents the set of potential disturbed nodes, which is obtained by statistically analyzing all disturbed nodes output by the event propagation spatiotemporal dynamics model Dyn at time t; Nc represents the central node, which is selected as the reference node for the diffusion range by the reference node in the initial state set Ini of the event propagation; d(Nc,j) represents the shortest path distance from the central node Nc to node j, where d represents the shortest path function; P(t,Nc→j) represents the probability value of the disturbance propagating from the central node Nc to node j.

9. The method for assessing and simulating the resilience of regional transportation systems based on heterogeneous satellites according to claim 8, characterized in that: S4 includes S41; S41. Based on the feature set Pro of the event diffusion stage, the propagation speed Vp and diffusion radius Rp are normalized using the min-max normalization method to eliminate the difference in dimensions between different data. The normalized propagation speed Vp and diffusion radius Rp are then combined with the input evaluation function to calculate the resilience level of the traffic system and output the resilience index set Res. The evaluation function takes the following specific form: In the formula, r1 and r2 represent the weighting coefficients of the normalized propagation velocity Vp and diffusion radius Rp, respectively, and the specific values ​​are set by the user, and r1+r2=1; max(Vp) and max(Rp) represent the maximum values ​​of the propagation velocity Vp and diffusion radius Rp in the entire Tim range, respectively.

10. The method for assessing and simulating the resilience of regional transportation systems based on heterogeneous satellites according to claim 9, characterized in that: S4 also includes S42; S42. Convert the set of resilience indicators Res into percentage-based resilience assessment values, expressed quantitatively in the range of 0–100%. When the resilience assessment value increases, it indicates that the regional transportation system has a strong resilience level under emergencies. When the resilience assessment value decreases, it indicates that the regional transportation system is less resilient to emergencies.

Images