A multi-dimensional digital traffic flow data collection method, device and system

Abstract

The present invention relates to the technical field of vehicle traffic control systems, and in particular to a multi-dimensional digital traffic flow data collection method, device and system, including: obtaining the complexity of the road conditions at each moment according to the correlation between the vehicle density and average speed at each moment and a number of adjacent moments and the changes in the vehicle density and average speed; obtaining the characteristic distance between each two moments according to the correlation between the characteristics of the road conditions and the difference in the characteristics of the same road conditions at each two moments; obtaining the inverse distance weighted interpolation result and the linear interpolation result of each missing moment according to the characteristic distance between each missing moment and the other several moments and the vehicle density at the other several moments, and obtaining the vehicle density interpolation result of each missing moment. The present invention performs missing value filling according to the complexity of the road conditions, combined with the inverse distance weighted interpolation and the linear interpolation, to achieve more realistic and more accurate missing value filling.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle traffic control systems, and in particular to a multi-dimensional digital traffic flow data collection method, device and system. Background Art

[0002] Multidimensional digital traffic flow data includes vehicle type data, vehicle density data, vehicle behavior data and traffic signal data. By analyzing multidimensional digital traffic flow data, we can understand travel patterns and behaviors in different time periods and regions, and help urban planners and traffic management departments formulate effective traffic optimization strategies; the traffic management and control system based on multidimensional digital traffic flow data is an important tool for improving urban management and the quality of life of residents. The collection of multidimensional digital traffic flow data is crucial to the efficiency of the traffic management and control system. Only by ensuring the integrity and accuracy of the multidimensional digital traffic flow data can the traffic management and control system work smoothly and efficiently.

[0003] When missing data occurs in multidimensional digital traffic flow data, linear interpolation results of the time when missing data occurs are usually obtained through the multidimensional digital traffic flow data of the adjacent time moments when missing data occurs, so as to fill the missing values ​​at the time when missing data occurs; however, when the road traffic conditions are complicated, due to the possible large changes in road conditions and dense traffic of vehicles and people, the linear interpolation results may be quite different from the actual conditions, which may lead to inaccurate missing value filling results. Summary of the invention

[0004] The present invention provides a multi-dimensional digital traffic flow data collection method, device and system to solve the problem that when the existing road traffic conditions are complicated, the linear interpolation result is far from the actual situation, resulting in inaccurate missing value filling results.

[0005] The present invention provides a multi-dimensional digital traffic flow data collection method, device and system using the following technical solutions:

[0006] An embodiment of the present invention provides a method for collecting multi-dimensional digital traffic flow data, the method comprising the following steps:

[0007] Obtain vehicle density data and average vehicle speed data at each moment, and obtain all road condition influencing features at each moment; record the moment when vehicle density is null as a missing moment;

[0008] According to the correlation between the vehicle density and average speed changes at each moment and its adjacent moments, the traffic flow trend at each moment is obtained; according to the traffic flow trend at each moment and the average value of the vehicle density and average speed at several adjacent moments, the complexity of the road conditions at each moment is obtained;

[0009] According to the correlation between each two road condition influencing features, the correlation coefficient of each two road condition influencing features is obtained; according to the correlation coefficient of each two road condition influencing features and the difference of the same road condition influencing features at each two moments, the characteristic distance at each two moments is obtained;

[0010] According to the characteristic distance between each missing moment and several other moments and the vehicle density at several other moments, the inverse distance weighted interpolation result and the linear interpolation result of each missing moment are obtained; according to the inverse distance weighted interpolation result and the linear interpolation result of each missing moment, the vehicle density interpolation result of each missing moment is obtained.

[0011] Furthermore, obtaining the traffic flow trend at each moment includes:

[0012] Traffic flow related trends at the i-th moment The calculation method is:

[0013]

[0014] Where: represents the Kendall correlation coefficient between the vehicle density and the average vehicle speed at all comparison moments in the comparison period at the i-th moment, represents the vehicle density at the next moment adjacent to the i-th moment, represents the vehicle density of the i-th moment adjacent to the previous moment, represents the average vehicle speed at the next moment adjacent to the i-th moment, It represents the average vehicle speed between the i-th moment and the previous moment.

[0015] Furthermore, the specific method for obtaining all comparison moments within the comparison period is:

[0016] The comparison length is preset, and each moment is taken as the center, and the moments of each adjacent comparison length are taken forward and backward respectively, which are recorded as the comparison moment of each moment, and the set of all the comparison moments of each moment is recorded as the comparison period of each moment.

[0017] Furthermore, obtaining the complexity of the road conditions at each moment includes:

[0018] The complexity of the road conditions at the i-th moment The calculation method is:

[0019]

[0020] Where: represents the traffic flow trend at the i-th moment, represents the traffic flow trend at the i-1th moment, represents the average value of vehicle density at all comparison moments in the comparison period at the i-th moment, represents the average of the average vehicle speeds at all comparison moments in the comparison period at the i-th moment; Represents the Sigmoid function.

[0021] Furthermore, obtaining the correlation coefficient between each two road condition influencing features includes:

[0022] The result of multiplying each feature in the road condition influencing features at the i-th moment by the road condition complexity at the i-th moment is recorded as each feature parameter at the i-th moment; the sequence composed of the same feature parameters in the road condition influencing features at all moments is recorded as the feature sequence of each feature in the road condition influencing features, and the Spearman correlation coefficient between the feature sequences of every two features in the road condition influencing features is obtained, which is recorded as the correlation coefficient of every two features in the road condition influencing features.

[0023] Further, obtaining the characteristic distance between every two moments includes:

[0024] The characteristic distance between the i-th moment and the j-th moment The calculation method is:

[0025]

[0026] Where: It represents the average value of the correlation coefficient between the kth feature and all other features in the road condition influencing features. represents the kth characteristic parameter at the i-th moment, represents the kth characteristic parameter at the jth moment, Indicates the number of types of features in the road condition impact features.

[0027] Furthermore, the specific method for obtaining the inverse distance weighted interpolation result and the linear interpolation result of each missing moment is:

[0028] Preset the number of controls, obtain the time with the minimum number of controls at each missing moment, as the control time for each missing moment, and the inverse distance weight of the rth control time of the tth missing moment The calculation method is:

[0029]

[0030] Where: represents the characteristic distance between the rth control moment and the tth missing moment of the tth missing moment, Indicates the number of controls, represents the missing moment The characteristic distance between the control moment and the tth missing moment;

[0031] Taking the inverse distance weights of all control moments of the t-th missing moment as the weight, the weighted mean of the vehicle density of all control moments of the t-th missing moment is recorded as the inverse distance weighted interpolation result of the t-th missing moment;

[0032] The average value of the vehicle density at the previous moment and the next moment of the t-th missing moment is recorded as the linear interpolation result of the t missing moments.

[0033] Furthermore, obtaining the vehicle density interpolation result at each missing moment includes:

[0034] Vehicle density interpolation result at the tth missing time The calculation method is:

[0035]

[0036] Where: represents the complexity of the road conditions at the tth missing moment, represents the inverse distance weighted interpolation result of the tth missing moment, Represents the linear interpolation result of the tth missing moment.

[0037] A multi-dimensional digital traffic flow data acquisition device, the device comprises: a data acquisition module, a road condition complexity analysis module and a missing value filling module; wherein the data acquisition module is used to obtain vehicle density data and average vehicle speed data at each moment and all road condition influencing features; the road condition complexity analysis module is used to obtain the road condition complexity at each moment according to the correlation between the road condition changes at each moment and several adjacent moments, and the average value of the vehicle density data and average vehicle speed data at several adjacent moments; the missing value filling module is used to obtain the characteristic distance between every two moments according to the correlation between every two road condition influencing features and the difference between every road condition influencing features at every two moments; according to the characteristic distance between each missing moment and several other moments and the vehicle density at several other moments, the vehicle density interpolation result at each missing moment is obtained to complete the missing value filling.

[0038] A multi-dimensional digital traffic flow data collection system comprises a memory, a processor and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of the multi-dimensional digital traffic flow data collection method are implemented.

[0039] The beneficial effects of the technical solution of the present invention are as follows: according to the correlation between the road condition changes at each moment and several adjacent moments, as well as the average of the vehicle density data and the average speed data at several adjacent moments, the complexity of the road condition at each moment is obtained, and the road congestion condition is reflected by the complexity of the road condition. The greater the complexity of the road condition, the less accurate the result when the linear interpolation is used to fill the missing values; according to the correlation between every two road condition influencing features and the difference between each road condition influencing feature at every two moments, the characteristic distance between every two moments is obtained, and different weights are given to each road condition influencing feature in the calculation of the characteristic distance according to the correlation between each road condition influencing feature and other road condition influencing features, so as to better The differences in the road condition influencing characteristics between every two moments are carefully reflected to obtain more accurate inverse distance interpolation weights in subsequent analysis; the inverse distance weighted interpolation results and linear interpolation results of each missing moment are obtained according to the characteristic distance between each missing moment and several other moments and the vehicle density of several other moments, and then the vehicle density interpolation results of each missing moment are obtained; the inverse distance weighted interpolation results and linear interpolation results of each missing moment are weighted averaged according to the complexity of the road conditions. When the complexity of the road conditions is large, the weight of the inverse distance weighted interpolation results is greater, and when the complexity of the road conditions is small, the weight of the linear interpolation results is greater, so as to obtain interpolation results that are more in line with the actual situation and achieve more accurate missing value filling. 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 1 The present invention is a flowchart of the steps of a multi-dimensional digital traffic flow data collection method. 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 following is a detailed description of a multi-dimensional digital traffic flow data collection method, device and system proposed by the present invention, its specific implementation method, structure, features and effects, 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 multi-dimensional digital traffic flow data collection method, device and system provided by the present invention in conjunction with the accompanying drawings.

[0045] First, see Figure 1 , which shows a flowchart of a multi-dimensional digital traffic flow data collection method provided by an embodiment of the present invention, the method comprising the following steps:

[0046] Step S001, obtain vehicle density data and average vehicle speed data at each moment, and obtain all road condition influencing features at each moment; record the moment when the vehicle density is null as a missing moment.

[0047] It should be noted that the collection interval and collection period are preset. The purpose of this embodiment is to fill the missing values ​​in the multidimensional digital traffic flow data according to the road condition influencing characteristics. Therefore, the vehicle density and average vehicle speed at several times and the road condition influencing characteristics at several times are first obtained; wherein the road condition influencing characteristics described in this embodiment include holidays, weather and time periods. This embodiment fills the missing values ​​in the multidimensional digital traffic flow data using vehicle density data as an example. Other embodiments may fill the missing values ​​in other multidimensional digital traffic flow data, and this embodiment does not make specific limitations.

[0048] Specifically, any intersection is taken as the collection intersection, and the traffic image of the collection intersection is obtained once every collection interval through the surveillance camera at the collection intersection, and all the road condition images in the collection period are obtained; the vehicles in all the road condition images are identified by the pre-trained YOLO model, and one road condition image corresponds to one moment; the number of vehicles in each road condition image is used as the vehicle density of each road condition image at the corresponding moment; the forward distance between each road condition image and the next adjacent road condition image is divided by the collection interval and recorded as the speed of the same vehicle, and the average speed of all vehicles in each road condition image is recorded as the average vehicle speed at the corresponding moment of each road condition image; the moment when the vehicle density is a null value is recorded as a missing moment.

[0049] Further, the road condition impact features corresponding to each road condition image when it is collected are obtained; for the holiday features in the road condition impact features, the holidays are periodic with a year as a cycle. When the road condition image is collected during a statutory holiday, the holiday features in the road condition impact features of the road condition image at the corresponding moment are recorded as 1; when the road condition image is collected on the previous day or the next day during a statutory holiday, the holiday features in the road condition impact features of the road condition image at the corresponding moment are recorded as 0.5; when the road condition image is not collected during a statutory holiday and is not the previous day or the next day during a statutory holiday, the holiday features in the road condition impact features of the road condition image at the corresponding moment are recorded as 0; for the day features in the road condition impact features, Weather characteristics: weather conditions have no periodicity. The meteorological department will have a comfort score for daily weather based on temperature, precipitation, wind speed, air quality, etc. The normalized result of the meteorological department's comfort score corresponding to each road condition image is obtained when the road condition image is collected, and the normalized result of the comfort score is recorded as the weather feature in the road condition influencing feature at the corresponding moment of each road condition image; for the time period feature in the road condition influencing feature, the time period is periodic with a day as a cycle. A day from 0:00 to 24:00 is evenly divided into 48 time periods, and the 48 time periods are numbered respectively. The result of normalizing the number of the time period at the corresponding moment of each road condition image is used as the time period feature in the road condition influencing feature at the corresponding moment of each road condition image.

[0050] It should be noted that the normalization methods used in this embodiment are all normalized using a linear normalization function. Other embodiments may choose other normalization methods, which are not specifically limited in this embodiment. The holiday characteristics, weather characteristics, and time period characteristics of each road condition image at the corresponding moment constitute the road condition influencing characteristics at each moment. The preset collection interval in this embodiment is 1 second, and the collection period is 30 days. Other embodiments may be set to other values, which are not specifically limited in this embodiment.

[0051] Step S002: Obtain the traffic flow trend at each moment; obtain the complexity of the road conditions at each moment.

[0052] It should be noted that for missing moments when the road conditions are simple, linear interpolation can obtain more accurate results. However, for missing moments when the road conditions are complex, the data obtained by linear interpolation may be quite different from the actual situation due to the large changes in road conditions and the dense traffic of vehicles and people. Therefore, the complexity of the road conditions at each moment is analyzed first according to the sudden changes in road conditions at each moment and the traffic congestion at each moment.

[0053] The more inconsistent the road condition changes at each moment are with the overall change trend in a nearby period of time, the more likely the road condition at that moment will change suddenly, and the more complex the road condition at that moment will be. In a period of time adjacent to each moment, if the overall road condition is characterized by a smaller vehicle density, it means that the road is more likely to be unobstructed at that moment; if the overall road condition is characterized by a larger vehicle density and a higher vehicle speed, it means that the road is more likely to be in a stable state of high pressure at that moment; if the overall road condition is characterized by a larger vehicle density and a lower vehicle speed, it means that the road is more likely to be congested at that moment.

[0054] Specifically, the comparison length is preset, and each moment is taken as the center, and the moments of the adjacent comparison lengths are taken forward and backward respectively, and recorded as the comparison moment of each moment, and the set of all the comparison moments of each moment is recorded as the comparison period of each moment. It should be noted that this embodiment is described with the comparison length as 10, and other embodiments can be set to other values, which are not specifically limited in this embodiment; when one side of a moment is less than the comparison length moments, the maximum number of moments that can be taken is taken as the comparison moment;

[0055] Traffic flow related trends at the i-th moment The calculation method is:

[0056]

[0057] Where: represents the Kendall correlation coefficient between the vehicle density and the average vehicle speed at all comparison moments in the comparison period at the i-th moment, represents the vehicle density at the next moment adjacent to the i-th moment, represents the vehicle density of the i-th moment adjacent to the previous moment, represents the average vehicle speed at the next moment adjacent to the i-th moment, It represents the average vehicle speed between the i-th moment and the previous moment.

[0058] It should be noted that the Kendall correlation coefficient reflects the similarity between the sequences of two sets of data. represents the Kendall correlation coefficient between the vehicle density and the average vehicle speed at all comparison moments at the i-th moment, then It reflects the correlation trend of the overall vehicle density change and the overall average vehicle speed change of all comparison moments at the i-th moment. The closer the value is to 1, the more consistent the rising or falling trend of the vehicle density and the average vehicle speed in the comparison period at the i-th moment is. The closer the value is to -1, the more inconsistent the rising or falling trend of the vehicle density and the average vehicle speed in the comparison period at the i-th moment is. The Kendall correlation coefficient is a well-known technology and will not be described in detail in this embodiment. It represents the ratio of the change in vehicle density to the change in average vehicle speed between the next moment adjacent to the ith moment and the previous moment adjacent to the ith moment, reflecting the correlation trend between the change in vehicle density and the change in average vehicle speed between the ith moment and the previous moment adjacent to the ith moment. A positive value indicates that the change in vehicle density and the change in average vehicle speed between the next moment adjacent to the ith moment and the previous moment adjacent to the ith moment are positively correlated, and a negative value indicates that the change in vehicle density and the change in average vehicle speed between the next moment adjacent to the ith moment and the previous moment adjacent to the ith moment are negatively correlated; The value of is positive and the larger the absolute value is, the more the traffic flow changes at the adjacent moments before and after the i-th moment are consistent with the overall change trend of the comparison period of the i-th moment. The larger the traffic flow-related trend of the i-th moment is, the more stable the road condition at the i-th moment is. The value of is negative and the larger the absolute value is, the more inconsistent the traffic flow changes at the adjacent moments before and after the i-th moment are with the overall change trend of the comparison period of the i-th moment. The smaller the traffic flow-related trend of the i-th moment is, the more sudden the road condition at the i-th moment is. In particular, when there are missing moments in the comparison moments of the i-th moment, the missing moments are not included in the calculation. When there are missing moments in the adjacent previous moment or the adjacent next moment of the i-th moment, it is postponed to the second moment adjacent upward or the second moment adjacent downward, and so on until a moment that is not a missing moment is selected.

[0059] Furthermore, the complexity of the road conditions at the i-th moment The calculation method is:

[0060]

[0061] Where: represents the traffic flow trend at the i-th moment, represents the traffic flow trend at the i-1th moment, represents the average value of vehicle density at all comparison moments in the comparison period at the i-th moment, represents the average of the average vehicle speeds at all comparison moments in the comparison period at the i-th moment; represents a Sigmoid function, which is used for normalization processing in this embodiment.

[0062] It should be noted that The larger the value of is, the more likely the road condition is to change suddenly, and the more complex the road condition is at the i-th moment; A positive value indicates that the traffic flow correlation trend from the i-1th moment to the i-th moment increases, and a negative value indicates that the traffic flow correlation trend from the i-1th moment to the i-th moment decreases. The larger the value, the more the traffic flow correlation trend increases, and the road condition at the i-th moment is more consistent with the road condition change trend during the comparison period of the i-th moment. The larger the value of is, the larger the average value of vehicle density is during the comparison period at the i-th moment. The smaller the value of is, the smaller the average value of the average vehicle speed in the comparison period at the i-th moment is. The smaller the value of is, the more likely it is that the comparison period at the i-th moment is in a low-density, unobstructed state or a high-density, high-speed, stable state. The larger the value of , the higher the vehicle density and lower the speed during the comparison period at the i-th moment, the more likely road congestion is during the comparison period at the i-th moment, and the more complex the road conditions at the i-th moment.

[0063] Similarly, the complexity of the road conditions at each moment is obtained.

[0064] Step S003: Obtain the correlation coefficient between every two road condition impact features; and obtain the characteristic distance between every two moments.

[0065] It should be noted that when the road conditions are complex, the results obtained by linear interpolation are inaccurate, but the traffic conditions have certain regularity. For example, during the rush hours from 7:30 to 8:30 in the morning and from 6 to 7 in the afternoon on weekdays, the vehicle density will increase significantly, and the vehicle density will further increase in rainy and snowy weather, and the vehicle density will increase overall during holidays; therefore, if the missing moment and another moment are in a similar time period, similar holiday conditions, and similar weather conditions, then the missing moment and the other moment may have similar vehicle densities, so the feature distance can be constructed through the difference in features between the missing moment and the other moments, and the vehicle density at the missing moment is interpolated by inverse distance weighting through the feature distance to fill in the vehicle density at the missing moment;

[0066] There is correlation between different features. Some features will have a great impact on the change of vehicle density only when combined with other features, while some features do not need to be combined with other features to have a great impact on the change of vehicle density. Therefore, when constructing feature distance, the correlation between features must also be considered. The stronger the correlation between a feature and all other features, the smaller its weight when constructing feature distance, and the weaker the correlation between a feature and all other features, the greater its weight when constructing feature distance.

[0067] Specifically, taking any moment as an example, the result of multiplying each feature in the road condition influencing feature at the moment by the road condition complexity at the moment is recorded as each feature parameter at the moment; a sequence consisting of the same feature parameters in the road condition influencing feature at all moments is recorded as a feature sequence of each feature in the road condition influencing feature, and the Spearman correlation coefficient between the feature sequences of every two features in the road condition influencing feature is obtained, which is recorded as a correlation coefficient of every two features in the road condition influencing feature; it should be noted that the Spearman correlation coefficient is a well-known technology and will not be described in detail in this embodiment;

[0068] The characteristic distance between the i-th moment and the j-th moment The calculation method is:

[0069]

[0070] Where: It represents the average value of the correlation coefficient between the kth feature and all other features in the road condition influencing features. represents the kth characteristic parameter at the i-th moment, represents the kth characteristic parameter at the jth moment, Indicates the number of types of features in the road condition impact feature. In this embodiment, K=3.

[0071] It should be noted that The larger the value of , the stronger the correlation between the k-th feature and other features. The stronger the correlation between other features, the smaller the effect of the feature on the result. Therefore, when calculating the feature distance between the i-th moment and the j-th moment, the smaller the weight of the k-th feature. It represents the difference between the kth feature parameter at the ith moment and the jth moment. The larger the difference is, the greater the difference in the road condition impact characteristics between the ith moment and the jth moment is, and the larger the feature distance between the ith moment and the jth moment is.

[0072] Similarly, the characteristic distance between every two moments is obtained.

[0073] Step S004, obtaining the inverse distance weighted interpolation result and the linear interpolation result of each missing moment; obtaining the vehicle density interpolation result of each missing moment.

[0074] It should be noted that, at the missing moments with a high degree of road condition complexity, the vehicle density that causes road congestion is regular when the road condition influencing characteristics are similar, while at the missing moments with a low degree of road condition complexity, the regularity between the vehicle density and the road condition influencing characteristics is relatively weaker; therefore, for the missing moments with a high degree of road condition complexity, the results obtained by linear interpolation may have certain deviations, and the inverse distance weighted interpolation obtained by the feature distance is more accurate, while for the missing moments with a low degree of road condition complexity, the results obtained by linear interpolation are more accurate than those obtained by inverse distance weighted interpolation; therefore, different weights are assigned according to the complexity of the road condition to obtain the vehicle density interpolation results of the missing moments.

[0075] Specifically, the number of controls is preset, and the moment with the smallest number of controls at each missing moment is obtained as the control moment of each missing moment; the inverse distance weight of the rth control moment of the tth missing moment is The calculation method is:

[0076]

[0077] Where: represents the characteristic distance between the rth control moment and the tth missing moment of the tth missing moment, Indicates the number of controls, represents the missing moment The characteristic distance between the control moment and the tth missing moment;

[0078] Taking the inverse distance weights of all control moments of the t-th missing moment as weights, the weighted average of the vehicle densities of all control moments of the t-th missing moment is recorded as the inverse distance weight interpolation result of the t-th missing moment; similarly, the inverse distance weight interpolation results of all missing moments are obtained; it should be noted that the calculation method of the inverse distance weight is a well-known technology, which is not described in detail in this embodiment; the preset number of controls in this embodiment is described by taking 10 as an example, and other embodiments may be set to other values, which are not specifically limited in this embodiment;

[0079] The average value of the vehicle density at the previous moment and the next moment of the t-th missing moment is recorded as the linear interpolation result of the t missing moments; the linear interpolation results of all missing moments are obtained in the same way; in particular, when the previous moment or the next moment of the t-th missing moment is a missing moment, it is postponed to the second moment adjacent upwards or the second moment adjacent downwards, and so on until a moment that is not a missing moment is selected.

[0080] Furthermore, the vehicle density interpolation result at the tth missing time is The calculation method is:

[0081]

[0082] Where: represents the complexity of the road conditions at the tth missing moment, represents the inverse distance weighted interpolation result of the tth missing moment, Represents the linear interpolation result of the tth missing moment.

[0083] It should be noted that when the road condition is less complex, the linear interpolation result has a greater weight, and the vehicle density interpolation result at the tth missing time is given by Partially determined, when the road condition is more complex, the inverse distance weighted interpolation result has a greater weight, and the vehicle density interpolation result at the tth missing time is mainly determined by Partial decision.

[0084] Similarly, the vehicle density interpolation results of all missing moments are obtained to complete the missing value filling; the vehicle density data with missing value filling is transmitted to the traffic management and control system to help the traffic management and control system to carry out traffic planning and scheduling, and realize multi-dimensional digital traffic flow data collection.

[0085] In a second aspect, the present invention also provides a multi-dimensional digital traffic flow data acquisition device, which includes: a data acquisition module, a road condition complexity analysis module and a missing value filling module; wherein the data acquisition module is used to obtain vehicle density data and average vehicle speed data at each moment and all road condition influencing features; the road condition complexity analysis module is used to obtain the road condition complexity at each moment based on the correlation between the road condition changes at each moment and several adjacent moments, and the average value of the vehicle density data and average vehicle speed data at several adjacent moments of each moment; the missing value filling module is used to obtain the characteristic distance between every two moments based on the correlation between every two road condition influencing features and the difference between each road condition influencing feature at every two moments; and obtain the vehicle density interpolation result of each missing moment based on the characteristic distance between each missing moment and several other moments and the vehicle density at several other moments to complete the missing value filling.

[0086] In a third aspect, the present invention also provides a multi-dimensional digital traffic flow data acquisition system, which includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the above method steps S001 to S004 are implemented.

[0087] 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 multi-dimensional digital traffic flow data collection method, characterized in that: The method comprises the following steps: Obtain vehicle density data and average vehicle speed data at each moment, and obtain all road condition influencing features at each moment; record the moment when vehicle density is null as a missing moment; According to the correlation between the vehicle density and average speed changes at each moment and its adjacent moments, the traffic flow trend at each moment is obtained; according to the traffic flow trend at each moment and the average value of the vehicle density and average speed at several adjacent moments, the complexity of the road conditions at each moment is obtained; According to the correlation between each two road condition influencing features, the correlation coefficient of each two road condition influencing features is obtained; according to the correlation coefficient of each two road condition influencing features and the difference of the same road condition influencing features at each two moments, the characteristic distance at each two moments is obtained; According to the characteristic distance between each missing moment and several other moments and the vehicle density at several other moments, the inverse distance weighted interpolation result and the linear interpolation result of each missing moment are obtained; according to the inverse distance weighted interpolation result and the linear interpolation result of each missing moment, the vehicle density interpolation result of each missing moment is obtained; The step of obtaining the correlation coefficient between each two road condition influencing features includes: The result of multiplying each feature in the road condition influencing feature at the i-th moment by the road condition complexity at the i-th moment is recorded as each feature parameter at the i-th moment; the sequence composed of the same feature parameters in the road condition influencing feature at all moments is recorded as the feature sequence of each feature in the road condition influencing feature, and the Spearman correlation coefficient between the feature sequences of every two features in the road condition influencing feature is obtained, which is recorded as the correlation coefficient of every two features in the road condition influencing feature; The step of obtaining the characteristic distance between every two moments includes: The characteristic distance between the i-th moment and the j-th moment The calculation method is: Where: It represents the average value of the correlation coefficient between the kth feature and all other features in the road condition influencing feature. represents the kth characteristic parameter at the i-th moment, represents the kth characteristic parameter at the jth moment, Indicates the number of types of features in the road condition impact features.

2. According to claim 1, a multi-dimensional digital traffic flow data collection method is characterized in that: The obtaining of the traffic flow-related trend at each moment includes: Traffic flow related trends at the i-th moment The calculation method is: Where: represents the Kendall correlation coefficient between the vehicle density and the average vehicle speed at all comparison moments in the comparison period at the i-th moment, represents the vehicle density at the next moment adjacent to the i-th moment, represents the vehicle density of the i-th moment adjacent to the previous moment, represents the average vehicle speed at the next moment adjacent to the i-th moment, It represents the average vehicle speed between the i-th moment and the previous moment.

3. According to claim 2, a multi-dimensional digital traffic flow data collection method is characterized in that: The specific method for obtaining all comparison moments within the comparison period is: The comparison length is preset, and each moment is taken as the center, and the moments of each adjacent comparison length are taken forward and backward respectively, which are recorded as the comparison moment of each moment, and the set of all the comparison moments of each moment is recorded as the comparison period of each moment.

4. According to claim 3, a multi-dimensional digital traffic flow data collection method is characterized in that: The obtaining of the complexity of the road conditions at each moment includes: The complexity of the road conditions at the i-th moment The calculation method is: Where: represents the traffic flow trend at the i-th moment, represents the traffic flow trend at the i-1th moment, represents the average value of vehicle density at all comparison moments in the comparison period at the i-th moment, represents the average of the average vehicle speeds at all comparison moments in the comparison period at the i-th moment; Represents the Sigmoid function.

5. The multi-dimensional digital traffic flow data collection method according to claim 1, characterized in that: The specific method for obtaining the inverse distance weighted interpolation result and the linear interpolation result of each missing moment is: Preset the number of controls, obtain the time with the minimum number of controls at each missing moment, as the control time for each missing moment, and the inverse distance weight of the rth control time of the tth missing moment The calculation method is: Where: represents the characteristic distance between the rth control moment and the tth missing moment of the tth missing moment, Indicates the number of controls, represents the missing moment The characteristic distance between the control moment and the tth missing moment; Taking the inverse distance weights of all control moments of the t-th missing moment as the weight, the weighted mean of the vehicle density of all control moments of the t-th missing moment is recorded as the inverse distance weighted interpolation result of the t-th missing moment; The average value of the vehicle density at the previous moment and the next moment of the t-th missing moment is recorded as the linear interpolation result of the t missing moments.

6. A multi-dimensional digital traffic flow data collection method according to claim 1, characterized in that: The step of obtaining the vehicle density interpolation result at each missing moment includes: Vehicle density interpolation result at the tth missing time The calculation method is: Where: represents the complexity of the road conditions at the tth missing moment, represents the inverse distance weighted interpolation result of the tth missing moment, Represents the linear interpolation result of the tth missing moment.

7. A multi-dimensional digital traffic flow data collection device, characterized in that: The device comprises: a data acquisition module, a road condition complexity analysis module and a missing value filling module; The data acquisition module is used to obtain vehicle density data and average vehicle speed data at each moment, and obtain all road condition influencing features at each moment; the moment when the vehicle density is null is recorded as a missing moment; The road condition complexity analysis module is used to obtain the traffic flow trend at each moment according to the correlation between the vehicle density and average speed changes at each moment and several adjacent moments; and to obtain the road condition complexity at each moment according to the traffic flow trend at each moment and the average value of the vehicle density and average speed at several adjacent moments. The missing value filling module is used to obtain the correlation coefficient of each two road condition influencing features according to the correlation between each two road condition influencing features; according to the correlation coefficient of each two road condition influencing features and the difference of the same road condition influencing features at each two moments, obtain the characteristic distance at each two moments; According to the characteristic distance between each missing moment and several other moments and the vehicle density at several other moments, the inverse distance weighted interpolation result and the linear interpolation result of each missing moment are obtained; according to the inverse distance weighted interpolation result and the linear interpolation result of each missing moment, the vehicle density interpolation result of each missing moment is obtained; The step of obtaining the correlation coefficient between each two road condition influencing features includes: The result of multiplying each feature in the road condition influencing feature at the i-th moment by the road condition complexity at the i-th moment is recorded as each feature parameter at the i-th moment; the sequence composed of the same feature parameters in the road condition influencing feature at all moments is recorded as the feature sequence of each feature in the road condition influencing feature, and the Spearman correlation coefficient between the feature sequences of every two features in the road condition influencing feature is obtained, which is recorded as the correlation coefficient of every two features in the road condition influencing feature; The step of obtaining the characteristic distance between every two moments includes: The characteristic distance between the i-th moment and the j-th moment The calculation method is: Where: It represents the average value of the correlation coefficient between the kth feature and all other features in the road condition influencing feature. represents the kth characteristic parameter at the i-th moment, represents the kth characteristic parameter at the jth moment, Indicates the number of types of features in the road condition impact features.

8. A multi-dimensional digital traffic flow data acquisition system, characterized in that: The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of a multi-dimensional digital traffic flow data collection method as described in any one of claims 1-6 are implemented.

Images