A road-air integrated intelligent traffic management system based on multi-source data
By building a road segment similarity module in the integrated road-air intelligent traffic management system, calculating the road segment similarity and similarity, adaptive filling of multi-source data is achieved, data deviation problems caused by sensor failure are solved, and system efficiency is improved .
Patent Information
- Application Number
- CN202410549122.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-06
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-05-06
AI Technical Summary
In the integrated intelligent traffic management system of road-air and air, due to sensor failure or environmental impact, the existing interpolation method fails to effectively consider the correlation characteristics between roads, resulting in large deviations in the interpolation results and reducing system efficiency.
By constructing a section similarity module, the average vehicle speed and traffic timing sequence of the target section and the reference section are used to calculate the similarity and similarity, and the adaptive fill value is obtained, the missing data is corrected, and the complete multi-source data is obtained.
It improves the adaptive filling accuracy of missing data, enhances the processing efficiency of multi-source data, and ensures the accuracy and efficiency of the integrated road-air intelligent traffic management system.
Smart Images

Figure CN118470962B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent sensor technology, and in particular to a road-to-air integrated intelligent traffic management system based on multi-source data. Background Art
[0002] At present, while ensuring traditional mechanical performance and safety characteristics, automotive electronics pays more attention to the coordinated interaction between vehicles, roads and airspace and the integration capabilities of vehicle systems. During vehicle driving, multi-source data has become the main basis for road condition analysis. In the integrated road-air smart traffic management system, the processing of multi-source data can correlate and match and process multi-level and multi-system data from different traffic participants to form unified traffic or vehicle status information. This information can help people understand and predict traffic conditions more accurately, optimize traffic management and control strategies, and improve road safety and efficiency.
[0003] When acquiring multi-source data, due to sensor failure or the influence of the working environment, some of the collected data is missing. When processing multi-source data, the missing data needs to be filled. Nearest neighbor interpolation is a commonly used interpolation method, but when interpolating multi-source data, it does not consider the correlation characteristics between roads and only interpolates based on the time series data of the current road, resulting in large deviations in the interpolation results, which reduces the efficiency of road-to-air integrated intelligent traffic management based on multi-source data. Summary of the invention
[0004] The present invention provides a road-air integrated intelligent traffic management system based on multi-source data to solve the existing problems.
[0005] The present invention adopts the following technical solutions for a road-air integrated intelligent traffic management system based on multi-source data:
[0006] An embodiment of the present invention provides a road-air integrated intelligent traffic management system based on multi-source data, the system comprising the following modules:
[0007] Data acquisition preprocessing module: obtain vehicle speed time series data set, traffic flow data set and road weather data set; the vehicle speed time series data set includes the average vehicle speed time series sequence of several road sections, and the traffic flow data set includes the vehicle flow time series sequence of several road sections; according to the road weather data set, obtain the complete road weather data set;
[0008] Road segment similarity construction module: at any traffic intersection, any road segment in the direction of exiting the intersection is recorded as the target road segment, and all road segments in the direction of entering the intersection are recorded as reference road segments; according to the average vehicle speed time series and vehicle flow time series of the target road segment and the reference road segment, the first source sequence, the second source sequence, the first source reference sequence and the second source reference sequence are obtained; all missing data on the first source sequence are obtained; any missing data on the first source sequence is recorded as the target missing data; the time position of the target missing data in the first source sequence is recorded as the marked position; according to the first source sequence, the second source sequence, the first source reference sequence and the second source reference sequence, the similarity between the target road segment and each reference road segment at the marked position is obtained;
[0009] Target filling value correction module: obtain the target filling value according to the target missing data; obtain the target left difference and the reference left difference according to the target filling value and the marked position; obtain the adaptive filling value of the target missing data according to the similarity of the target section and each reference section in the marked position, the target left difference and the reference left difference;
[0010] Multi-source data prediction module: obtain the complete average vehicle speed time series and the complete vehicle flow time series according to the adaptive filling value of the target missing data.
[0011] Furthermore, obtaining the first source sequence, the second source sequence, the first source reference sequence and the second source reference sequence according to the average vehicle speed time series sequence and the vehicle flow time series sequence of the target road section and the reference road section includes:
[0012] The traffic flow time series and the average vehicle speed time series of the target section are obtained, which are recorded as the first source sequence and the second source sequence respectively; the traffic flow time series and the average vehicle speed time series of each reference section are obtained, which are recorded as the first source reference sequence and the second source reference sequence.
[0013] Further, obtaining the similarity between the target road segment and each reference road segment at the marked position according to the first source sequence, the second source sequence, the first source reference sequence and the second source reference sequence includes:
[0014] Acquire extreme value points on the first source sequence, the second source sequence, the first source reference sequence, and the second source reference sequence respectively, record the interval between two adjacent extreme value points in the same sequence as a subinterval, and obtain several subintervals on the first source sequence, the second source sequence, the first source reference sequence, and the second source reference sequence; record the element value of the marked position of any first source reference sequence as reference data;
[0015] According to the subinterval where the target missing data is located and the subinterval where each reference data is located, the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located is obtained;
[0016] Obtaining a first similarity and a second similarity between the target road segment and the reference road segment according to the second source sequence, the second source reference sequence, and the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located;
[0017] According to the first similarity and the second similarity between the target road section and the reference road section, the similarity between the target road section and each reference road section at the marked position is obtained.
[0018] Further, obtaining the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located according to the subinterval where the target missing data is located and the subinterval where each reference data is located includes:
[0019] The ratio of the difference between the leftmost element value and the rightmost element value of any subinterval to the length of the subinterval is recorded as the rate of change of the subinterval; the similarity between the subinterval where the target missing data is located and the subinterval where the ath reference data is located is calculated as follows:
[0020]
[0021] In the formula, θ a is the similarity between the subinterval where the target missing data is located and the subinterval where the ath reference data is located; l is the rate of change of the subinterval where the target missing data is located; l' a is the rate of change of the subinterval where the ath reference data is located; H is the difference between the valid data adjacent to the left and right sides of the target missing data; H' a is the difference between the valid data on the left and right sides of the ath reference data of the target missing data; exp() is an exponential function with a natural constant as the base.
[0022] Further, obtaining the first similarity and the second similarity between the target road segment and the reference road segment according to the second source sequence, the second source reference sequence, and the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located includes:
[0023] The similarity between the subinterval where the target missing data is located and the subinterval where any reference data is located is recorded as the first similarity between the target road section and the reference road section where the reference data is located;
[0024] For the element value of the marked position in the second source sequence, the element value of the marked position of each second source reference sequence is recorded as the reference data of the element value of the marked position in the second source sequence, and the similarity between the subinterval where the element value of the marked position in the second source sequence is located and the subinterval where any reference data of the element value of the marked position in the second source sequence is located is obtained, and recorded as the second similarity between the target road section and the reference road section where the reference data is located.
[0025] Furthermore, the similarity between the target road section and each reference road section at the marked position is obtained according to the first similarity and the second similarity between the target road section and the reference road section. The specific calculation method is:
[0026]
[0027] In the formula, β c is the similarity between the target segment and the cth reference segment at the marked position; is the variance of the first similarity and the second similarity between the target road segment and the cth reference road segment; A is the number of reference road segments; is the variance of the first similarity and the second similarity between the target road segment and the jth reference road segment; θ' c is the mean of the first similarity and the second similarity between the target road segment and the cth reference road segment; norm() is the normalization function; | | is the absolute value function.
[0028] Further, obtaining the target left difference value and the reference left difference value according to the target filling value and the mark position includes:
[0029] The difference between the target filling value and the valid data adjacent to its left is recorded as the target left difference; the difference between the element value of the marked position in any first source reference sequence and the valid data adjacent to its left is recorded as the reference left difference of the reference section corresponding to the first source reference sequence.
[0030] Further, the step of obtaining the adaptive filling value of the target missing data according to the similarity between the target road section and each reference road section at the marked position, the target left difference value and the reference left difference value comprises:
[0031] According to the similarity of the target section and each reference section in the marked position, the target left difference value and the reference left difference value, the corrected distance of the target filling value is obtained;
[0032] According to the similarity between the target section and each reference section in the marked position, the target left difference, the reference left difference and the corrected distance of the target filling value, the adaptive filling value of the target missing data is obtained.
[0033] Furthermore, the correction distance of the target filling value is obtained according to the similarity of the target section and each reference section in the marked position, the target left difference and the reference left difference. The specific calculation method is:
[0034]
[0035] Where ρ is the corrected distance of the target filling value; A is the number of reference sections; β c is the similarity between the target segment and the cth reference segment at the marked position; Q' cis the reference left difference of the cth reference section; Q is the target left difference; softmax() is the softmax normalization function; || is the absolute value function.
[0036] Furthermore, the adaptive filling value of the target missing data is obtained according to the similarity between the target section and each reference section at the marked position, the target left difference, the reference left difference and the corrected distance of the target filling value. The specific calculation method is:
[0037]
[0038] Where S' is the adaptive filling value of the target missing data; S is the target filling value; ρ is the correction distance of the target filling value; A is the number of reference sections; β c is the similarity between the target segment and the cth reference segment at the marked position; β j is the similarity between the target segment and the jth reference segment at the marked position; Q' c is the reference left difference of the cth reference section; Q is the target left difference; SGN() is the sign function.
[0039] The beneficial effects of the technical solution of the present invention are as follows: the present invention obtains the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located, preliminarily judges the similarity between the target missing data and each reference data in each source, and provides an initial basis for the correction of the target filling value of the target missing data; then the similarity between the target section and each reference section at the marked position is obtained, and the similarity between the target missing data and each reference data is made more accurate by combining multi-source data, thereby improving the accuracy of the adaptive filling value of the target missing data; according to the similarity between the target section and each reference section at the marked position, the target left difference and the reference left difference, the correction distance of the target filling value is obtained, and then the adaptive filling value of the target missing data is obtained, thereby further improving the accuracy of the adaptive filling value of the target missing data. So far, the present invention obtains complete multi-source data through accurate and reliable adaptive filling values, thereby improving the efficiency of road-air integrated intelligent traffic management based on multi-source data. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0041] Figure 1This is a module flow chart of a road-air integrated intelligent traffic management system based on multi-source data of the present invention. DETAILED DESCRIPTION
[0042] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation method, structure, features and effects of a road-air integrated intelligent traffic management system based on multi-source data proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0043] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0044] The following is a detailed description of a specific solution of a road-to-air integrated intelligent traffic management system based on multi-source data provided by the present invention in conjunction with the accompanying drawings.
[0045] See also Figure 1 , which shows a module flow chart of a road-air integrated intelligent traffic management system based on multi-source data provided by an embodiment of the present invention, the system includes the following modules:
[0046] Module 101: Data acquisition preprocessing module.
[0047] It should be noted that the purpose of this embodiment is to interpolate missing data in multi-source data based on the correlation characteristics between roads, improve the accuracy of the interpolation results, and improve the efficiency of road-air integrated intelligent traffic management of multi-source data, so it is necessary to collect multi-source data.
[0048] Specifically, the vehicle speed time series data set, traffic flow data set and road weather data set are obtained according to the traffic management data related webpage; wherein the vehicle speed time series data set contains the average vehicle speed time series sequence of several sections, and any average vehicle speed time series sequence contains the average vehicle speed of all vehicles at each time in the section corresponding to the average vehicle speed time series sequence; the traffic flow data set contains the vehicle flow time series sequence of several sections, and any vehicle flow time series sequence contains the vehicle flow at each time in the section corresponding to the vehicle flow time series sequence; the road weather data set contains the rainfall time series sequence of several sections and the visibility time series sequence of each section, and any rainfall time series sequence contains the rainfall at each time in the section corresponding to the rainfall time series sequence, and any visibility time series sequence contains the visibility at each time in the section corresponding to the visibility time series sequence. In the vehicle speed time series data set, traffic flow data set and road weather data set, the data marked as "null" are recorded as missing data, and the other data are recorded as valid data.
[0049] It should be noted that in the collected vehicle speed time series dataset, traffic flow dataset and road weather dataset, some of the collected data are missing due to sensor failure or the influence of the working environment. The missing data will be marked as "null". Since the data marked as "null" will affect the efficiency of the road-to-air integrated intelligent traffic management of multi-source data, it is necessary to interpolate the missing data marked as "null".
[0050] Furthermore, for the missing data in the rainfall time series of each road section and the visibility time series of each road section, the nearest neighbor interpolation algorithm is used to interpolate the missing data to obtain the complete rainfall time series and the complete visibility time series; the nearest neighbor interpolation algorithm is a well-known technology and the specific method is not introduced here; the road weather data set after interpolation processing is recorded as the complete road weather data set.
[0051] Module 102: Road segment similarity construction module.
[0052] Specifically, at any traffic intersection, any road section in the direction of exiting the intersection is recorded as a target section, and all road sections in the direction of entering the intersection are recorded as reference sections.
[0053] It should be noted that any road section heading into the intersection is recorded as a directly connected section, and the road sections heading into the intersection other than the reference section are recorded as bifurcated sections. During normal traffic operation, the bifurcated section will divert the traffic flow of the target section, so the traffic data fluctuations of the target section and the bifurcated section show an opposite change relationship; while the target section and the directly connected section will share the traffic flow, and the traffic data fluctuations of the target section and the directly connected section show a positive change relationship, so the similarity between the target section and the reference section at the missing position is calculated based on this.
[0054] Specifically, the traffic flow time series sequence and the average vehicle speed time series sequence of the target section are obtained, which are recorded as the first source sequence and the second source sequence respectively; the traffic flow time series sequence and the average vehicle speed time series sequence of each reference section are obtained, which are recorded as the first source reference sequence and the second source reference sequence; the extreme points on the first source sequence, the second source sequence, the first source reference sequence and the second source reference sequence are obtained respectively, and the interval between two adjacent extreme points in the same sequence is recorded as a sub-interval, so as to obtain several sub-intervals on the first source sequence, the second source sequence, the first source reference sequence and the second source reference sequence.
[0055] Furthermore, any missing data on the first source sequence is recorded as the target missing data; the time position of the target missing data in the first source sequence is recorded as the mark position; the element value of the mark position of any first source reference sequence is recorded as the reference data; the ratio of the difference between the leftmost element value and the rightmost element value of any subinterval to the length of the subinterval is recorded as the change rate of the subinterval; the similarity between the subinterval where the target missing data is located and the subinterval where the ath reference data is located is calculated as follows:
[0056]
[0057] In the formula, θ a is the similarity between the subinterval where the target missing data is located and the subinterval where the ath reference data is located; l is the rate of change of the subinterval where the target missing data is located; l' a is the rate of change of the subinterval where the ath reference data is located; H is the difference between the valid data adjacent to the left and right sides of the target missing data; H' a is the difference between the valid data on the left and right sides of the ath reference data of the target missing data; exp() is an exponential function with a natural constant as the base; 0.1 is to prevent the denominator from being 0.
[0058] It should be noted that ||l|-|l' a The smaller the value of ||, the more similar the fluctuation trend of the subinterval where the target missing data is located is to the subinterval where the ath reference data is located; ||H|-|H' aThe smaller the value of ||, the more similar the nearest neighbor volatility of the target missing data is to the nearest neighbor volatility of the a-th reference data.
[0059] Further, the similarity between the subinterval where the target missing data is located and the subinterval where any reference data is located is recorded as the first similarity between the target road section and the reference road section where the reference data is located;
[0060] For the element value at the marked position in the second source sequence, the element value at the marked position of each second source reference sequence is recorded as the reference data of the element value at the marked position in the second source sequence. According to the calculation method of the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located, the similarity between the subinterval where the element value at the marked position in the second source sequence is located and the subinterval where any reference data of the element value at the marked position in the second source sequence is located is obtained, and recorded as the second similarity between the target road section and the reference road section where the reference data is located.
[0061] It should be noted that the above steps obtain the first similarity and the second similarity. In order to improve the accuracy of data interpolation, it is necessary to combine multi-source data for analysis.
[0062] Specifically, the similarity between the target road segment and the cth reference road segment at the marked position is calculated as follows:
[0063]
[0064] In the formula, β c is the similarity between the target segment and the cth reference segment at the marked position; is the variance of the first similarity and the second similarity between the target road segment and the cth reference road segment; A is the number of reference road segments; is the variance of the first similarity and the second similarity between the target road segment and the jth reference road segment; θ' c is the mean of the first similarity and the second similarity between the target road segment and the cth reference road segment; || is the absolute value function; norm() is the linear normalization function, and the normalized object is the similarity between the target road segment and all reference road segments.
[0065] It should be noted that The larger it is, the greater the similarity between the target road segment and the cth reference road segment in different sources; The smaller it is, the higher the degree of aggregation between the target segment and the cth reference segment, and the greater the similarity between the target segment and the reference segment in the marked position.
[0066] Module 103: Target fill value correction module.
[0067] It should be noted that if the similarity between the target section and any reference section is higher, and the data fluctuation trend direction of the target section and the reference section is the same, then the reference weight of the reference section when adaptive interpolation calculation is performed on missing data will be greater; if the similarity between the target section and any reference section is higher, and the data fluctuation trend direction of the target section and the reference section is opposite, then the reverse reference weight of the reference section when adaptive interpolation calculation is performed on missing data will be greater.
[0068] Specifically, the target missing data is filled in using the nearest neighbor interpolation method to obtain the target missing data, which is recorded as the target filling value; the nearest neighbor interpolation method is a well-known technology and the specific method is not introduced here; the difference between the target filling value and the valid data adjacent to its left is recorded as the target left difference; the difference between the element value of the marked position in any first source reference sequence and the valid data adjacent to its left is recorded as the reference left difference of the reference section corresponding to the first source reference sequence.
[0069] It should be noted that if the reference left difference is greater than the target left difference, it means that the target filling value is too small; if the reference left difference is greater than the target left difference, it means that the target filling value is too large.
[0070] Specifically, the calculation method of the correction distance of the target filling value is:
[0071]
[0072] Where ρ is the corrected distance of the target filling value; A is the number of reference sections; β c is the similarity between the target segment and the cth reference segment at the marked position; Q' c is the reference left difference of the cth reference section; Q is the target left difference; softmax() is the softmax normalization function, and the normalized object is the similarity between the target section and all reference sections at the marked position; || is the absolute value function.
[0073] It should be noted that |Q' c -Q| represents the distance between the target left difference and the reference left difference of the cth reference section, softmax(β c ) is |Q' c -Q| weight value, softmax(β c ) value, the larger the value of |Q' c The more effective -Q| is.
[0074] Furthermore, the calculation method of the adaptive filling value of the target missing data is:
[0075]
[0076] Where S' is the adaptive filling value of the target missing data; S is the target filling value; ρ is the correction distance of the target filling value; A is the number of reference sections; β c is the similarity between the target segment and the cth reference segment at the marked position; β j is the similarity between the target segment and the jth reference segment at the marked position; Q' c is the reference left difference of the cth reference section; Q is the target left difference; SGN() is the sign function.
[0077] It should be noted that Indicates the correction direction of the target filling value. If its value is greater than 0, it means that the target filling value is too small and needs to be corrected to a larger value. If its value is less than 0, it means that the target filling value is too large and needs to be corrected to a smaller value. If its value is equal to 0, it means that the accuracy of the target filling value is high and there is no need to correct the target filling value.
[0078] Furthermore, according to the above method, adaptive filling values of all missing data on the first source sequence are obtained.
[0079] Furthermore, according to the method for obtaining the adaptive filling value of all the missing data on the first source sequence, the adaptive filling value of all the missing data on the second source sequence is obtained.
[0080] Module 104: Multi-source data prediction module.
[0081] Replace all missing data with the corresponding adaptive filling values to obtain a complete average vehicle speed time series sequence and a complete vehicle flow time series sequence. Input the complete road weather data set consisting of all complete rainfall time series sequences and complete visibility time series sequences, the complete vehicle speed time series data set consisting of all complete average vehicle speed time series sequences, and the complete traffic flow data set consisting of all complete vehicle flow time series sequences into the trained neural network to obtain the predicted data of the road weather data set, the vehicle speed time series data set, and the traffic flow data set, and complete the road-air integrated intelligent traffic management based on multi-source data; the neural network model in this embodiment is the RMGRU model, and other models can be used in other embodiments, which are not limited in this embodiment; the specific process is to obtain the data from each source and divide the data into a training set, a validation set, and a test set according to a ratio of 8:1:1. Then input all the data in the training set into the RMGRU model for model training to finally obtain the optimal RMGRU model, in which the loss function uses the LOSS loss function; the specific training process is well known, and this embodiment will not be described in detail.
[0082] This embodiment uses the exp(-x) model to present the inverse proportional relationship and normalization processing, x is the input of the model, and the implementer can set the inverse proportional function and normalization function according to the actual situation.
[0083] At this point, this embodiment is completed.
[0084] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A road-air integrated intelligent traffic management system based on multi-source data, characterized in that: The system includes the following modules: Data acquisition preprocessing module: obtain vehicle speed time series data set, traffic flow data set and road weather data set; the vehicle speed time series data set includes the average vehicle speed time series sequence of several road sections, and the traffic flow data set includes the vehicle flow time series sequence of several road sections; according to the road weather data set, obtain the complete road weather data set; Road segment similarity construction module: At any traffic intersection, any road segment in the direction of exiting the intersection is recorded as the target road segment, and all road segments in the direction of entering the intersection are recorded as reference roads. According to the average vehicle speed time series and vehicle flow time series of the target road section and the reference road section, a first source sequence, a second source sequence, a first source reference sequence and a second source reference sequence are obtained; all missing data on the first source sequence are obtained; any missing data on the first source sequence is recorded as target missing data; Recording the time position of the target missing data in the first source sequence as the marked position; obtaining the similarity between the target section and each reference section at the marked position according to the first source sequence, the second source sequence, the first source reference sequence and the second source reference sequence; Target filling value correction module: obtains the target filling value according to the target missing data; Obtain the target left difference and the reference left difference according to the target filling value and the marking position; Obtaining adaptive filling values for target missing data according to the similarity between the target segment and each reference segment at the marked position, the target left difference value, and the reference left difference value; Multi-source data prediction module: obtain the complete average vehicle speed time series and the complete vehicle flow time series according to the adaptive filling value of the target missing data; The obtaining the similarity between the target road segment and each reference road segment at the marked position according to the first source sequence, the second source sequence, the first source reference sequence and the second source reference sequence comprises: Acquire extreme value points on the first source sequence, the second source sequence, the first source reference sequence, and the second source reference sequence respectively, record the interval between two adjacent extreme value points in the same sequence as a subinterval, and obtain several subintervals on the first source sequence, the second source sequence, the first source reference sequence, and the second source reference sequence; record the element value of the marked position of any first source reference sequence as reference data; According to the subinterval where the target missing data is located and the subinterval where each reference data is located, the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located is obtained; Obtaining a first similarity and a second similarity between the target road segment and the reference road segment according to the second source sequence, the second source reference sequence, and the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located; According to the first similarity and the second similarity between the target road section and the reference road section, obtaining the similarity between the target road section and each reference road section at the marked position; The obtaining, according to the subinterval where the target missing data is located and the subinterval where each reference data is located, the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located comprises: The ratio of the difference between the leftmost element value and the rightmost element value of any subinterval to the length of the subinterval is recorded as the rate of change of the subinterval; the similarity between the subinterval where the target missing data is located and the subinterval where the ath reference data is located is calculated as follows: In the formula, θ a is the similarity between the subinterval where the target missing data is located and the subinterval where the ath reference data is located; l is the rate of change of the subinterval where the target missing data is located; l′ a is the rate of change of the subinterval where the ath reference data is located; H is the difference between the valid data adjacent to the left and right sides of the target missing data; H′ a is the difference between the valid data on the left and right sides of the ath reference data of the target missing data; exp() is an exponential function with a natural constant as the base; The method of obtaining the adaptive filling value of the target missing data according to the similarity between the target section and each reference section at the marked position, the target left difference and the reference left difference includes: According to the similarity of the target section and each reference section in the marked position, the target left difference value and the reference left difference value, the corrected distance of the target filling value is obtained; Obtaining adaptive filling values for target missing data based on the similarity between the target segment and each reference segment at the marked position, the target left difference, the reference left difference, and the corrected distance of the target filling value; The correction distance of the target filling value is obtained according to the similarity of the target section and each reference section in the marked position, the target left difference and the reference left difference. The specific calculation method is: Where ρ is the corrected distance of the target filling value; A is the number of reference sections; β c is the similarity between the target segment and the cth reference segment at the marked position; Q′ c is the reference left difference of the cth reference section; Q is the target left difference; softmax() is the softmax normalization function; || is the absolute value function; According to the similarity between the target section and each reference section at the marked position, the target left difference, the reference left difference and the corrected distance of the target filling value, the adaptive filling value of the target missing data is obtained. The specific calculation method is: In the formula, S ′ is the adaptive filling value of the target missing data; S is the target filling value; ρ is the correction distance of the target filling value; A is the number of reference sections; β c is the similarity between the target segment and the cth reference segment at the marked position; β j is the similarity between the target segment and the jth reference segment at the marked position; Q′ c is the reference left difference of the cth reference section; Q is the target left difference; SGN() is the sign function.
2. According to claim 1, a road-air integrated intelligent traffic management system based on multi-source data is characterized in that: The step of obtaining the first source sequence, the second source sequence, the first source reference sequence, and the second source reference sequence according to the average vehicle speed time series sequence and the vehicle flow time series sequence of the target road section and the reference road section includes: The traffic flow time series and the average vehicle speed time series of the target section are obtained, which are recorded as the first source sequence and the second source sequence respectively; the traffic flow time series and the average vehicle speed time series of each reference section are obtained, which are recorded as the first source reference sequence and the second source reference sequence.
3. According to claim 1, a road-air integrated intelligent traffic management system based on multi-source data is characterized in that: The obtaining the first similarity and the second similarity between the target road segment and the reference road segment according to the second source sequence, the second source reference sequence, and the similarity between the subinterval where the target missing data is located and the subinterval where each reference data is located comprises: The similarity between the subinterval where the target missing data is located and the subinterval where any reference data is located is recorded as the first similarity between the target road section and the reference road section where the reference data is located; For the element value of the marked position in the second source sequence, the element value of the marked position of each second source reference sequence is recorded as the reference data of the element value of the marked position in the second source sequence, and the similarity between the subinterval where the element value of the marked position in the second source sequence is located and the subinterval where any reference data of the element value of the marked position in the second source sequence is located is obtained, and recorded as the second similarity between the target road section and the reference road section where the reference data is located.
4. According to claim 1, a road-air integrated intelligent traffic management system based on multi-source data is characterized in that: The similarity between the target road segment and each reference road segment at the marked position is obtained according to the first similarity and the second similarity between the target road segment and the reference road segment. The specific calculation method is: In the formula, β c is the similarity between the target segment and the cth reference segment at the marked position; is the variance of the first similarity and the second similarity between the target road segment and the cth reference road segment; A is the number of reference road segments; is the variance of the first similarity and the second similarity between the target road segment and the jth reference road segment;θ′ c is the mean of the first similarity and the second similarity between the target road segment and the cth reference road segment; norm() is the normalization function; || is the absolute value function.
5. According to claim 1, a road-air integrated intelligent traffic management system based on multi-source data is characterized in that: The obtaining of the target left difference value and the reference left difference value according to the target filling value and the marking position comprises: The difference between the target filling value and the valid data adjacent to its left is recorded as the target left difference; the difference between the element value of the marked position in any first source reference sequence and the valid data adjacent to its left is recorded as the reference left difference of the reference section corresponding to the first source reference sequence.
