Illegal operation vehicle identification method and system based on big data

Through a vehicle evaluation model based on big data, the identification feature indicators in historical driving data are extracted, the total vehicle evaluation value is calculated, and illegally operated vehicles are judged, which solves the problem of inefficient identification in the existing technology and realizes efficient detection of illegal operation vehicles. and investigate and deal with it.

CN118964945APending Publication Date: 2024-11-15ZHEJIANG SUPCON INFORMATION TECH CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411149368.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-21
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing technology cannot accurately identify illegally operated vehicles, resulting in inefficient investigation and punishment, and requires manual check-up and inspection, which wastes manpower and material resources.

Method used

By obtaining the historical driving data of the sample vehicles, extracting identification feature indicators to establish a vehicle evaluation model, calculating the total vehicle evaluation value of the vehicle to be identified, determining whether it is an illegal vehicle, and then conducting targeted inspections to improve the identification accuracy and investigation efficiency.

Benefits of technology

It improves the identification accuracy and investigation efficiency of illegally operated vehicles, avoids the waste of manual card setting, and realizes automated inspection of illegally operated vehicles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118964945A_ABST
    Figure CN118964945A_ABST
Patent Text Reader

Abstract

The invention discloses an illegal operation vehicle identification method and system based on big data, and the method comprises the steps: obtaining a historical driving data set of a plurality of sample vehicles, extracting a plurality of identification feature indexes of the sample vehicles, and building a vehicle evaluation model of the illegal operation vehicles; the vehicle evaluation model is a corresponding relation between a plurality of identification characteristic indexes of the vehicle and a vehicle evaluation total value; after driving data of a to-be-recognized vehicle is obtained and recognition feature indexes are extracted, a vehicle evaluation total value is calculated by using the vehicle evaluation model; and when the total vehicle evaluation value exceeds a judgment threshold value, determining that the to-be-identified vehicle is an illegal operation vehicle. According to the invention, whether the vehicle is an illegal operation vehicle is evaluated and judged according to the vehicle evaluation model, and then the target which is judged to be the illegal operation vehicle is inspected in a targeted manner, so that the identification accuracy of the illegal operation vehicle can be improved, the investigation efficiency can be improved, and the waste of manpower and material resources caused by one-by-one inspection by manual card setting is avoided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a method and system for identifying illegally operated vehicles based on big data. Background Art

[0002] The development and application of illegal vehicle identification technology is an important topic, especially in the field of urban traffic management. This technology is mainly aimed at vehicles that have not obtained legal operating licenses but are engaged in profitable passenger or freight activities. These illegal operations not only disrupt the normal market order and affect the rights and interests of legal operators, but may also bring safety hazards, because illegal vehicles often lack the necessary safety inspections and insurance guarantees, which seriously threaten the personal safety of passengers and traffic order. Therefore, the identification and crackdown on illegal vehicles has become particularly important. In the prior art, the identification of illegal vehicles is mainly carried out by manually setting up checkpoints at fixed checkpoints. However, due to the mutual notification of drivers of illegal vehicles, the efficiency of the investigation and handling of illegal vehicles is low, and it is impossible to accurately identify illegal vehicles for targeted inspections.

[0003] The "method and system for inspecting illegal operating vehicles" disclosed in the Chinese patent literature has a publication number of CN102881153A and a publication date of 2013-01-16, which includes: the control center sets the inspection task and the inspection period, inspection area and identification information corresponding to the inspection task; the control center sends the inspection task information to each legally operating vehicle located in the inspection area during the inspection period; each legally operating vehicle located in the inspection area during the inspection period displays the identification information on a preset display screen according to the received inspection task information. By using the method and system for inspecting illegal operating vehicles in this technology, illegal operating vehicles can be inspected effectively and conveniently. However, this technology also requires manual setting of inspections to inspect the vehicle, and requires the installation of a display screen on each legally operating vehicle for inspection personnel to inspect, but for illegal operating vehicles, they themselves do not have a display screen installed to display information. Therefore, this technology still cannot accurately identify illegal operating vehicles for the current illegal operating vehicle situation, and cannot carry out targeted vehicle inspections, which is inefficient. Summary of the invention

[0004] The present invention aims to overcome the problem that the prior art cannot accurately identify illegally operated vehicles for targeted inspection and the efficiency of investigating and handling illegally operated vehicles is low, and provides a method and system for identifying illegally operated vehicles based on big data.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: A method for identifying illegally operated vehicles based on big data, comprising: Obtain historical driving data sets of several sample vehicles, extract several identification feature indicators of the sample vehicles, and establish a vehicle assessment model for illegally operated vehicles; The vehicle evaluation model is a correspondence between several identification characteristic indicators of a vehicle and the total evaluation value of the vehicle; After obtaining the driving data of the vehicle to be identified and extracting the identification feature indicators, the vehicle evaluation model is used to calculate the total vehicle evaluation value; when the total vehicle evaluation value exceeds the judgment threshold, the vehicle to be identified is an illegally operated vehicle.

[0006] In the present invention, the driving data of the sample vehicle is extracted from the historical vehicle operation data, and the identification feature index that can represent the operating vehicle is extracted based on the driving data, so as to establish a vehicle evaluation model for the illegally operated vehicle. After the identification feature index is also extracted for the vehicle to be identified, it is evaluated according to the vehicle evaluation model to determine whether it is an illegally operated vehicle, and then the target determined as the illegally operated vehicle is subjected to targeted inspection, so as to improve the identification accuracy of the illegally operated vehicle, and at the same time improve the efficiency of investigation and handling, and avoid the waste of manpower and material resources of manually setting up checkpoints and checking one by one; after the identification feature index is specifically selected, the present invention obtains the specific corresponding relationship between the identification feature index and the vehicle evaluation total value based on the historical driving data training, and the determination threshold for determining whether it is an illegally operated vehicle; for the vehicle to be identified whose vehicle evaluation total value exceeds the determination threshold, because it is very similar to the identification feature index of the operating vehicle, it can be considered to have a great suspicion of illegal operation, and then through manual targeted inspection, the efficiency of investigation and handling of the illegally operated vehicle can be greatly improved, and at the same time, the vehicle evaluation model can be optimized through continuous use, so as to further improve the accuracy of identification.

[0007] Preferably, in the vehicle evaluation model, a corresponding weight coefficient is assigned to each identification feature index; The identification feature indicators are weighted and summed based on their corresponding weight coefficients to obtain the total vehicle evaluation value; A vehicle evaluation total value set of the sample vehicle is calculated based on the vehicle evaluation model and the identification characteristic indicators of the sample vehicle, and a determination threshold is determined based on the data boundary in the vehicle evaluation total value set.

[0008] Preferably, the identification feature index includes a hot area index: Obtain a set of times that sample vehicles pass through the popular area; and calculate the average number of times that the sample vehicles pass through the popular area; Taking the average number of times as the middle value, the number of times is divided into 2n intervals at equal intervals according to the maximum value of the number difference in the number set; The percentage of sample vehicles falling in the corresponding frequency interval to the total sample vehicles is used as the hot area indicator.

[0009] Preferably, the method for determining the hot area includes: Obtain the total number of times that several sample vehicles pass through a certain area, and traverse each area to obtain the area frequency set; The length of the hot interval is calculated by multiplying the difference between the maximum and minimum times in the regional frequency concentration by the set percentage, and the hot lower limit number is obtained by subtracting the hot interval length from the maximum number; An area where the total number of times that several sample vehicles pass through is between the lower limit and the maximum limit is a hot area.

[0010] Preferably, the identification feature index includes a stop number index: Obtain a stop set containing the number of stops of each sample vehicle; and calculate the average number of stops for all sample vehicles; take the average number of stops as the middle value, and divide it into 2n stop number intervals at equal intervals according to the maximum value of the number difference in the stop set; take the percentage of sample vehicles falling in the corresponding stop number interval to the total number of sample vehicles as the stop number indicator.

[0011] Preferably, the identification characteristic index includes an average speed index: Obtain a speed set containing the average speed of each sample vehicle; and average the speeds of all sample vehicles to obtain an average speed value; take the average speed value as the middle value, and divide the speed set into 2n speed intervals at equal intervals according to the maximum value of the speed difference; The percentage of sample vehicles falling in the corresponding speed range to the total sample vehicles is taken as the average speed indicator.

[0012] Preferably, the identification feature index includes a driving distance index: Obtain a distance set containing the total distance traveled by each sample vehicle; and average the distances of all sample vehicles to obtain an average distance value; take the average distance value as the middle value, and divide the distance set into 2n distance intervals at equal intervals according to the maximum value of the distance difference; The percentage of sample vehicles falling in the corresponding distance interval to the total sample vehicles is taken as the average speed indicator.

[0013] Preferably, when the sample vehicle is an operating vehicle, the identification feature index of the sample vehicle is extracted to obtain a sample index set; the value of the weighted sum of the identification feature index based on its corresponding weight coefficient is set to be one, and the sample index set is used for fitting training to obtain the weight coefficient value corresponding to each identification feature index; Based on the weight coefficient value, the vehicle evaluation value is calculated for the sample indicator set, and the minimum value of the vehicle evaluation value obtained is used as the judgment threshold.

[0014] Preferably, when the sample vehicle is a non-operating vehicle, the identification characteristic index of the sample vehicle is extracted to obtain a sample index set; Set the weighted sum of the identification feature indicators based on their corresponding weight coefficients to zero, and use the sample indicator set for fitting training to obtain the weight coefficient values ​​corresponding to each identification feature indicator; Based on the weight coefficient value, the vehicle evaluation value of the sample indicator set is calculated, and the maximum value of the vehicle evaluation value obtained is used as the judgment threshold.

[0015] An illegal vehicle identification system based on big data, including: An index extraction module extracts corresponding identification feature indexes from the vehicle's driving data; The evaluation model module stores the vehicle evaluation model and updates the model according to the historical driving data of the updated sample vehicles; the vehicle retrieval module extracts the driving data of the sample vehicles or the vehicles to be identified from the driving data of all vehicles.

[0016] The present invention has the following beneficial effects: the driving data of sample vehicles is extracted from historical vehicle operation data, and identification feature indicators of the vehicles are extracted based on these driving data, so as to establish a vehicle evaluation model for illegally operated vehicles. After the identification feature indicators of the vehicles to be identified are also extracted, they are evaluated according to the vehicle evaluation model to determine whether they are illegally operated vehicles, and then targeted inspections are carried out on the targets determined to be illegally operated vehicles, thereby improving the identification accuracy of illegally operated vehicles and at the same time improving the efficiency of investigation and punishment, avoiding the waste of manpower and material resources of manually setting up checkpoints and checking one by one. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the method for identifying illegally operated vehicles based on big data in the present invention. DETAILED DESCRIPTION

[0018] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0019] like Figure 1 As shown, a method for identifying illegal operating vehicles based on big data includes: Obtain historical driving data sets of several sample vehicles, extract several identification feature indicators of the sample vehicles, and establish a vehicle assessment model for illegally operated vehicles; The vehicle evaluation model is the correspondence between several identification characteristic indicators of the vehicle and the total evaluation value of the vehicle; After obtaining the driving data of the vehicle to be identified and extracting the identification feature indicators, the vehicle evaluation model is used to calculate the total vehicle evaluation value; when the total vehicle evaluation value exceeds the judgment threshold, the vehicle to be identified is an illegally operated vehicle.

[0020] It should be noted that the present invention extracts the driving data of sample vehicles from the historical vehicle operation data, and extracts identification feature indicators that can represent operating vehicles based on these driving data, thereby establishing a vehicle evaluation model for illegally operated vehicles. After the identification feature indicators are also extracted for the vehicle to be identified, it is evaluated according to the vehicle evaluation model to determine whether it is an illegally operated vehicle, and then a targeted inspection is carried out on the target determined to be an illegally operated vehicle, thereby improving the identification accuracy of illegally operated vehicles, and at the same time improving the efficiency of investigation and punishment, and avoiding the waste of manpower and material resources of manually setting up checkpoints and checking one by one.

[0021] After the present invention makes specific selection of identification feature indicators, the specific corresponding relationship between the identification feature indicators and the total vehicle evaluation value is obtained based on historical driving data training, as well as a determination threshold for determining whether it is an illegally operated vehicle; for vehicles to be identified whose total vehicle evaluation value exceeds the determination threshold, because the identification feature indicators of operating vehicles are very similar, they can be considered to have a great suspicion of illegal operation, and then through manual targeted inspection, the efficiency of investigating and handling illegally operated vehicles can be greatly improved, and at the same time, the vehicle evaluation model can be optimized through continuous use, thereby further improving the accuracy of identification.

[0022] It is worth noting that the operating vehicles in the present invention generally refer to taxis, online ride-hailing cars and other passenger-carrying cars. The data of sample vehicles comes from the vehicle driving data captured by urban traffic radar and cameras, including license plate number, area, speed, longitude and latitude, etc. The corresponding identification feature indicators are extracted through these vehicle driving data to determine whether private cars or other small vehicles are non-operating vehicles. Large vehicles such as buses and coaches are not considered.

[0023] As a specific embodiment, in the vehicle evaluation model, a corresponding weight coefficient is assigned to each identification feature index; The identification feature indicators are weighted and summed based on their corresponding weight coefficients to obtain the total vehicle evaluation value; A vehicle evaluation total value set of the sample vehicle is calculated based on the vehicle evaluation model and the identification characteristic indicators of the sample vehicle, and a determination threshold is determined based on the data boundary in the vehicle evaluation total value set.

[0024] It should be noted that in the vehicle evaluation model, a trained neural network model or other machine learning model can be used to output the corresponding recognition result according to the input recognition feature index; however, the use of these machine learning models requires a large amount of data calculation process in each actual recognition process of the object to be identified, which has high requirements for hardware configuration and long calculation time. For traffic law enforcement officers patrolling on the road, the small equipment they carry with them cannot meet the configuration requirements of the calculation. The technical features in this application only need to perform weighted calculations on the recognition feature indicators to obtain the total vehicle evaluation value, and then directly determine whether it is an illegally operated vehicle. The amount of calculation is small and the response speed is fast, which is more suitable for traffic law enforcement officers to patrol on the road.

[0025] Based on the illegal operation vehicle identification method, a big data-based illegal operation vehicle identification system is also provided, including: An index extraction module extracts corresponding identification feature indexes from the vehicle's driving data; The evaluation model module stores the vehicle evaluation model and updates the model according to the historical driving data of the updated sample vehicles; the vehicle retrieval module extracts the driving data of the sample vehicles or the vehicles to be identified from the driving data of all vehicles.

[0026] Specifically, when traffic law enforcement officers are patrolling on the road, they carry a portable device containing the system of the present invention; an evaluation model module is stored in the portable device, that is, the weight coefficients corresponding to various identification feature indicators and the judgment thresholds of illegally operated vehicles are stored; as for the indicator extraction module and the vehicle retrieval module, they can be set at the control center of the traffic control system, and the traffic law enforcement officers can input suspicious vehicle information such as license plates into the portable device to inquire whether it is an operating vehicle. If it is not an operating vehicle, the driving data or identification feature indicators of the vehicle to be identified can be retrieved from the control center through the portable device, and then the vehicle evaluation is performed to determine whether it is an illegally operated vehicle. When it is determined to be an illegally operated vehicle, the vehicle can be inspected immediately to improve the efficiency of investigation and punishment.

[0027] Example 1: In this example, a number of operating vehicle samples (taxis, online ride-hailing vehicles) are obtained from the traffic management bureau or relevant departments. Car = [car1, car2, ..., carN], where N represents a total of N operating vehicle samples. The historical driving data set car of the operating vehicle samples within a period of time (such as 30 days in a month) is extracted from the driving data of all vehicles. data =(car1 data =[(t1, car1 t1-data ),...,(tm,car1 tm-data )],...,carN data =[(t1, carN t1-data),...,(tm,carN tm-data )]), t1 to tm represent consecutive moments in a period of time, carN t1-data It represents the driving data of the Nth operating vehicle sample at time t1. After extracting several identification feature indicators of operating vehicles from the historical driving data set, a vehicle evaluation model for illegal operating vehicles is established.

[0028] The vehicle evaluation model is the correspondence between several identification characteristic indicators of operating vehicles and the total vehicle evaluation value; that is, the high-area index ratio , parking ratio , average speed index speed ratio and driving distance ratio The calculation relationship between these four indicators and the total vehicle evaluation result is: result = area weight *high-area ratio +parking weight *parking ratio +distance weight *distance ratio +speed weight *speed ratio At the same time, the weight coefficients corresponding to the identification feature indicators and the corresponding judgment thresholds are obtained, thereby obtaining a complete vehicle evaluation model.

[0029] After obtaining the driving data of the vehicle to be identified and extracting the identification feature indicators, the vehicle evaluation model is used to calculate the total vehicle evaluation value; when the total vehicle evaluation value exceeds the judgment threshold, the vehicle to be identified is an illegally operated vehicle.

[0030] As an optional embodiment, the identification feature index includes a hot area index high-area ratio : Obtain a set of times that sample vehicles pass through the popular area; and calculate the average number of times that the sample vehicles pass through the popular area; Taking the average number of times as the middle value, the number of times is divided into 2n intervals at equal intervals according to the maximum value of the number difference in the number set; The percentage of sample vehicles falling in the corresponding frequency interval to the total sample vehicles is used as the hot area indicator.

[0031] It should be noted that the reason for using popular areas as one of the identification feature indicators is that as an operating vehicle, its purpose is to carry passengers to the destinations they want to reach. Most of these destinations are concentrated in various shopping malls, scenic spots or other areas with concentrated traffic. Although ordinary non-operating private cars will also appear in these concentrated areas, the frequency and frequency of operating vehicles appearing in these concentrated areas is far greater than that of ordinary family cars. Therefore, it can be used as one of the characteristics to identify operating vehicles.

[0032] Furthermore, the method for determining the hot area includes: Obtain the total number of times that several sample vehicles pass through a certain area, and traverse each area to obtain the area frequency set; The length of the hot interval is calculated by multiplying the difference between the maximum and minimum times in the regional frequency concentration by the set percentage, and the hot lower limit number is obtained by subtracting the hot interval length from the maximum number; An area where the total number of times that several sample vehicles pass through is between the lower limit and the maximum limit is a hot area.

[0033] Specifically, regional grid management is performed for a certain traffic range, that is, the traffic range where illegal operating vehicles need to be checked is divided into several areas, and for each area, the total number of times the operating vehicle sample passes through the area during the data extraction period (one month and thirty days) is counted, so as to obtain the regional frequency set area count =[(area1,count1),...,(area p ,count p )], that is, the number of operating vehicle samples passing through the area within a month p The total number of times is count p When dividing popular areas, the more times the operating vehicle sample passes through, the greater the probability that it belongs to the popular area.

[0034] For the value of the number of times the operating vehicle samples pass through a specific popular area, you can select the area frequency set area count The maximum number of times the area is passed count-max =Max(count1,…,count p ) and the minimum number of times area is passed count-min =Max(count1,…,count p ). With the maximum number of times area count-max =Max(count1,…,count p ) and the minimum number of times area is passed count-min =Max(count1,…,count p) divided by three to get the length of the popular interval D = (area count-max -area count-min ) / 3; then the total number of times the operating vehicle samples pass through a certain area is [area count-max -D,area count-max ] can be divided into hot areas. You can choose, or you can choose the corresponding percentage to calculate the length of the hot interval according to actual needs, so as to get the corresponding hot area range.

[0035] After obtaining the specific hot areas, the average number of times the operating vehicle samples pass through the hot areas in a month is calculated. data Get the number of times the operating vehicle sample passes through the popular area [high-area1 count ,…,high-areaN count ], where high-areaN count Indicates the number of times the Nth operating vehicle sample passes through the popular area in a month.

[0036] The average number of times you pass through a popular area is high-area count-avg =(high-area1 count +…+high-areaN count ) / N. Average number of high-area count-avg is the middle value, and is divided into 2n frequency intervals at equal intervals according to the maximum value of the frequency difference in the frequency set, that is, there are n equally spaced frequency intervals before and after the middle value. In this embodiment, n=5 can be selected, and a total of 10 frequency intervals are divided.

[0037] According to the number of times the operating vehicle sample passes through the popular area [high-area1 count ,…,high-areaN count ], assign each frequency value in the frequency set to the corresponding frequency interval, obtain the number of operating vehicle samples corresponding to each frequency interval, and then divide it by the total number of operating vehicle samples to obtain the percentage of the number of sample vehicles in the corresponding frequency interval to the total number of sample vehicles. That is, each frequency interval corresponds to a percentage value as a hot area indicator.

[0038] As an optional embodiment, the identification characteristic index includes a parking number index parking ratio : Obtain a stop set containing the number of stops of each sample vehicle; and calculate the average number of stops for all sample vehicles; take the average number of stops as the middle value, and divide it into 2n stop number intervals at equal intervals according to the maximum value of the number difference in the stop set; take the percentage of sample vehicles falling in the corresponding stop number interval to the total number of sample vehicles as the stop number indicator.

[0039] It should be noted that the reason why the number of stops is used as an identification characteristic indicator is that as an operating vehicle, after it transports passengers to different destinations, there will be changes in the number of passengers getting on and off. Therefore, the operating vehicle will frequently stop at the roadside waiting for passengers to get on or off. That is, the number of stops of an operating vehicle will far exceed that of a normal private car. It can be used as an identification characteristic indicator to distinguish between operating vehicles and normal private cars.

[0040] Specifically, before obtaining the parking set, it is necessary to calculate the number of stops for each operating vehicle sample in a month. The driving data of a single vehicle is time-linear, that is, the vehicle driving data points arranged in chronological order can be obtained through video, checkpoints or radar data, and each data point includes specific time, location, longitude and latitude and other parameters; the difference in longitude and latitude between two adjacent data points can be used to calculate the distance traveled, and the time difference between two adjacent data points can be obtained. The speed of the operating vehicle sample between two adjacent data points is calculated by the quotient of distance and time. If the speed is less than the preset speed threshold (such as 10KM / H), it is judged that the vehicle has stopped once. Through this process, the parking set parking of the number of stops of the operating vehicle sample in a month is calculated. count =[parking1 count ,...,parkingN count ], where parkingN count Represents the number of stops of the Nth operating vehicle sample in a month.

[0041] According to the parking set parking count Average all elements in it to get the average number of parking count-avg =(parking1 count +...+parkingN count ) / N; average number of stops parking count-avg The middle value is divided into 2n stop number intervals at equal intervals according to the maximum value of the stop number difference in the stop set, that is, there are n equally spaced stop number intervals before and after the middle value. In this embodiment, n=5 can be selected, which is divided into 10 stop number intervals in total.

[0042] Parking according to the parking set of the operating vehicle sample count =[parking1 count,...,parkingN count ], assign each stop count value in the stop count set to the corresponding stop count interval, obtain the number of operating vehicle samples corresponding to each stop count interval, and then divide it by the total number of operating vehicle samples to obtain the percentage of sample vehicles in the corresponding stop count interval to the total number of sample vehicles. That is, each stop count interval corresponds to a percentage value as the stop count indicator.

[0043] As an optional embodiment, the identification characteristic index includes an average speed index speed ratio : Obtain a speed set containing the average speed of each sample vehicle; and average the speeds of all sample vehicles to obtain an average speed value; take the average speed value as the middle value, and divide the speed set into 2n speed intervals at equal intervals according to the maximum value of the speed difference; The percentage of sample vehicles falling in the corresponding speed range to the total sample vehicles is taken as the average speed indicator.

[0044] It should be noted that normal private cars stop when a red light appears or after arriving at the destination, while for operating vehicles, in addition to stopping at red lights, they will stop at the place where they wait for passengers to get on and the place where passengers get off. In addition, the number of destinations for private cars in a day or a period of time is very small, and the number of stops is far less than that of operating vehicles. For operating vehicles, the number of destinations in a day or a period of time is far more than that of private cars. Because of the time spent on stopping, there will be a big difference in calculating the average speed compared with normal private cars, which can be used as an identification feature indicator to distinguish operating vehicles.

[0045] Specifically, before obtaining the speed set, it is necessary to calculate the average speed of each operating vehicle sample within a month. The driving data of a single vehicle is time-linear, that is, the vehicle driving data points arranged in time order can be obtained through video, checkpoints or radar data, and each data point includes specific time, location, longitude and latitude and other parameters; the difference in longitude and latitude between two adjacent data points can be used to calculate the distance traveled, and the time difference between two adjacent data points can be obtained. The speed of the operating vehicle sample between two adjacent data points is calculated by the quotient of distance and time, and the speed of a single vehicle between each pair of adjacent data points is averaged to obtain the average speed of a single vehicle. Through this process, the speed set speed of the average speed of the operating vehicle sample within a month is calculated. avg =[speed1, ..., speed N ], where speed N Represents the average speed of the Nth operating vehicle sample in a month.

[0046] According to the speed collection speed avgAverage all the elements in it to get the average speed of all operating vehicle samples all-speed avg =(speed1+...+speed N ) / N; average speed all-speed avg The intermediate value is divided into 2n speed intervals at equal intervals according to the maximum value of the speed difference in the speed set, that is, there are n equally spaced speed intervals before and after the intermediate value. In this embodiment, n=5 can be selected, which is divided into 10 speed intervals in total.

[0047] According to the speed set of the operating vehicle sample speed avg =[speed1, ..., speed N ], assign each average speed value in the speed set to the corresponding speed interval, obtain the number of operating vehicle samples corresponding to each speed interval, and then divide it by the total number of operating vehicle samples to obtain the percentage of sample vehicles in the corresponding speed interval to the total number of sample vehicles. That is, each speed interval corresponds to a percentage value as the average speed indicator.

[0048] As an optional embodiment, the identification characteristic index includes a driving distance index. ratio : Obtain a distance set containing the total distance traveled by each sample vehicle; and average the distances of all sample vehicles to obtain an average distance value; take the average distance value as the middle value, and divide the distance set into 2n distance intervals at equal intervals according to the maximum value of the distance difference; The percentage of sample vehicles falling in the corresponding distance interval to the total sample vehicles is taken as the average speed indicator.

[0049] It should be noted that the driving distance index is used as the identification characteristic index because compared to normal private cars, operating vehicles will constantly travel within the traffic range due to the need to carry passengers, which makes the driving distance of operating vehicles far exceed that of normal private cars. Therefore, the difference in driving distance can be used to distinguish operating vehicles from normal private cars.

[0050] Specifically, before obtaining the distance set, it is necessary to calculate the distance traveled by each operating vehicle sample within a month. The driving data of a single vehicle is time-linear, that is, the vehicle driving data points arranged in time order can be obtained through video, checkpoints or radar data, and each data point includes specific time, location, longitude and latitude and other parameters; the difference in longitude and latitude between two adjacent data points can be used to calculate the distance traveled, and the distance traveled by a single vehicle between each pair of adjacent data points is summed to obtain the distance traveled by a single vehicle. Through this process, the distance set of the driving distance of the operating vehicle sample within a month is calculated. avg =[distance1,...,distanceN ], where distance N Represents the driving distance of the Nth operating vehicle sample in one month.

[0051] According to the distance collection avg The average of all elements in it is obtained to get the average distance of all operating vehicle samples. avg =(distance1+...+distance N ) / N; average distance all-distance avg is the middle value, and is divided into 2n distance intervals at equal intervals according to the maximum value of the distance difference in the distance set, that is, there are n equally spaced distance intervals before and after the middle value. In this embodiment, n=5 can be selected, and a total of 10 distance intervals are divided.

[0052] According to the distance set of the operating vehicle sample avg =[distance1,...,distance N ], assign each driving distance value in the distance set to the corresponding distance interval, obtain the number of operating vehicle samples corresponding to each distance interval, and then divide it by the total number of operating vehicle samples to obtain the percentage of the number of sample vehicles in the corresponding distance interval to the total number of sample vehicles. That is, each distance interval corresponds to a percentage value as the driving distance indicator.

[0053] The sample vehicles in this embodiment are operating vehicles, and the identification feature index of each sample vehicle is extracted to obtain a sample index set; that is, each operating vehicle sample corresponds to its hot area index high-area ratio , parking ratio , average speed index speed ratio and driving distance ratio .

[0054] Set the weighted sum of the identification feature index based on its corresponding weight coefficient to 1, and use the sample index set for fitting training to obtain the weight coefficient value corresponding to each identification feature index. The weight coefficients of the four indicators are hot coefficient area weight , parking coefficient weight , speed coefficient speed weight And the distance coefficient weight ; The calculation relationship between these four indicators and the total vehicle evaluation value result is: result = area weight *high-area ratio +parkingweight *parking ratio +distance weight *distance ratio +speed weight * speed ratio . When the sample vehicle is an operating vehicle, the relationship formula is established with result = 1, and a method such as linear regression is used to train the sample indicator set to obtain the values ​​of the four weight coefficients. That is, the four weight coefficients are calculated so that the sum of the squares of the difference between the total vehicle evaluation value of all operating vehicle samples and 1 is minimized, thereby obtaining the four weight coefficients in the vehicle identification model. The specific method of obtaining the weight coefficients based on the sample indicator set and the calculation relationship training belongs to the prior art, so it is not described in detail.

[0055] Based on the weight coefficient values, the vehicle evaluation value is calculated for the sample indicator set, and the minimum value of the vehicle evaluation value is used as the judgment threshold. After obtaining the values ​​of the four weight coefficients, the sample indicator set is used to calculate the final vehicle evaluation value of each operating vehicle sample. Since most vehicle evaluation values ​​are not necessarily 1, there is a range of fluctuations around 1, so the minimum value of all vehicle evaluation values ​​can be taken as the judgment threshold. When the vehicle evaluation value is greater than the judgment threshold, it can be said that it is very similar to the characteristics of an operating vehicle.

[0056] When identifying illegal operating vehicles based on this, first select the target vehicle, then query the traffic control bureau or relevant departments to see if the vehicle is an operating vehicle based on the license plate information of the target vehicle. If it is an operating vehicle, no further identification is required. If it is not an operating vehicle, the target vehicle is used as the vehicle to be identified, and the hot area index high-area is extracted based on the driving data of the vehicle to be identified. ratio , parking ratio , average speed index speed ratio and driving distance ratio .

[0057] Specifically, the number of times the vehicle to be identified passes through the hot area in a month is calculated from the driving data of the vehicle to be identified, and the percentage value corresponding to the number interval to which the number of times it passes through the hot area belongs is the hot area index of the vehicle to be identified. ratio ; Calculate the number of parking times of the vehicle to be identified within a month, and use the percentage value corresponding to the parking time interval to which the parking times belong as the parking time index of the vehicle to be identified ratio ; Calculate the average speed of the vehicle to be identified within a month, and use the percentage value corresponding to the speed range to which the average speed belongs as the average speed index speed of the vehicle to be identifiedratio Calculate the distance traveled by the vehicle to be identified within a month, and use the percentage value corresponding to the distance interval to which the travel distance belongs as the distance indicator of the vehicle to be identified ratio .

[0058] The four indicators of the vehicle to be identified are input into the vehicle identification model, and the corresponding weight coefficients are used for weighted summation to obtain the total vehicle evaluation value result of the vehicle to be identified; when its total vehicle evaluation value is greater than the judgment threshold, it indicates that the vehicle is an illegally operated vehicle.

[0059] Embodiment 2: In Embodiment 1, the vehicle evaluation model is set up with operating vehicles as sample vehicles. By comparing the identification feature index of the vehicle to be identified with the operating vehicle sample, it is determined whether the driving characteristics of the vehicle to be identified are similar to those of the operating vehicle, thereby determining whether it is an illegally operated vehicle. In addition to setting up with operating vehicles as samples, the vehicle evaluation model can also be set up with non-operating vehicles as samples. By comparing the identification feature index of the vehicle to be identified with the non-operating vehicle sample, if the driving characteristics of the vehicle to be identified are similar to those of the non-operating vehicle, it means that it is not an illegally operated vehicle, otherwise it is suspected of illegal operation.

[0060] In this embodiment, a number of non-operating vehicles (such as private cars and official cars) are obtained from the traffic management bureau or relevant departments. Non-operating vehicle samples car = [car1, car2, ..., carN], N represents a total of N non-operating vehicle samples. The historical driving data set car of non-operating vehicle samples within a period of time (such as 30 days in a month) is extracted from the driving data of all vehicles. data =(car1 data =[(t1, car1 t1-data ),...,(tm,car1 tm-data )],...,carN data =[(t1, carN t1-data ),...,(tm,carN tm-data )]), t1 to tm represent consecutive moments in a period of time, carN t1-data It represents the driving data of the Nth non-operating vehicle sample at time t1. After extracting several identification feature indicators of non-operating vehicles from the historical driving data set, a vehicle evaluation model for illegal operating vehicles is established.

[0061] The vehicle evaluation model is the correspondence between several identification characteristic indicators of non-operating vehicles and the total vehicle evaluation value; that is, the high-area index ratio , parking ratio , average speed index speed ratioand driving distance ratio The calculation relationship between these four indicators and the total vehicle evaluation result is: result = area weight *high-area ratio +parking weight *parking ratio +distance weight *distance ratio +speed weight *speed ratio At the same time, the weight coefficients corresponding to the identification feature indicators and the corresponding judgment thresholds are obtained, thereby obtaining a complete vehicle evaluation model.

[0062] After obtaining the driving data of the vehicle to be identified and extracting the identification feature indicators, the vehicle evaluation model is used to calculate the total vehicle evaluation value; when the total vehicle evaluation value exceeds the judgment threshold, the vehicle to be identified is an illegally operated vehicle.

[0063] The selection and setting of the vehicle identification characteristic index in this embodiment is the same as that in the first embodiment, so it will not be described in detail. The sample vehicles in this embodiment are non-operating vehicles, and the identification characteristic index of each sample vehicle is extracted to obtain a sample index set; that is, each non-operating vehicle sample corresponds to its hot area index high-area ratio , parking ratio , average speed index speed ratio and driving distance ratio .

[0064] Set the weighted sum of the identification feature index based on its corresponding weight coefficient to zero, and use the sample index set for fitting training to obtain the weight coefficient value corresponding to each identification feature index. The weight coefficients of the four indicators are hot coefficient area weight , parking coefficient weight , speed coefficient speed weight And the distance coefficient weight ; The calculation relationship between these four indicators and the total vehicle evaluation value result is: result = area weight *high-area ratio +parking weight *parking ratio +distance weight *distance ratio +speed weight * speed ratio . When the sample vehicle is a non-operating vehicle, the relationship formula is established with result = 0, and a method such as linear regression is used to train the sample indicator set to obtain the values ​​of the four weight coefficients. That is, the four weight coefficients are calculated so that the sum of the squares of the difference between the total vehicle evaluation value of all non-operating vehicle samples and 0 is minimized, thereby obtaining the four weight coefficients in the vehicle identification model. The specific method of obtaining the weight coefficients based on the sample indicator set and the calculation relationship training belongs to the prior art, so it is not described in detail.

[0065] Based on the weight coefficient values, the vehicle evaluation value is calculated for the sample indicator set, and the maximum value of the vehicle evaluation value is used as the judgment threshold. After obtaining the values ​​of the four weight coefficients, the sample indicator set is used to calculate the final vehicle evaluation value of each non-operational vehicle sample. Since most of the vehicle evaluation values ​​are not necessarily 0, there is a range of fluctuations around 0, so the maximum value of all vehicle evaluation values ​​can be taken as the judgment threshold. When the vehicle evaluation value is greater than the judgment threshold, it can be explained that it does not belong to the driving characteristics of non-operational vehicles, and therefore there is suspicion of illegal operation.

[0066] When identifying illegal operating vehicles based on this, first select the target vehicle, then query the traffic control bureau or relevant departments to see if the vehicle is an operating vehicle based on the license plate information of the target vehicle. If it is an operating vehicle, no further identification is required. If it is not an operating vehicle, the target vehicle is used as the vehicle to be identified, and the hot area index high-area is extracted based on the driving data of the vehicle to be identified. ratio , parking ratio , average speed index speed ratio and driving distance ratio .

[0067] Specifically, the number of times the vehicle to be identified passes through the hot area in a month is calculated from the driving data of the vehicle to be identified, and the percentage value corresponding to the number interval to which the number of times it passes through the hot area belongs is the hot area index of the vehicle to be identified. ratio ; Calculate the number of parking times of the vehicle to be identified within a month, and use the percentage value corresponding to the parking time interval to which the parking times belong as the parking time index of the vehicle to be identified ratio ; Calculate the average speed of the vehicle to be identified within a month, and use the percentage value corresponding to the speed range to which the average speed belongs as the average speed index speed of the vehicle to be identified ratio Calculate the distance traveled by the vehicle to be identified within a month, and use the percentage value corresponding to the distance interval to which the travel distance belongs as the distance indicator of the vehicle to be identified ratio .

[0068] The four indicators of the vehicle to be identified are input into the vehicle identification model, and the corresponding weight coefficients are used for weighted summation to obtain the total vehicle evaluation value result of the vehicle to be identified; when the total vehicle evaluation value is greater than the judgment threshold, it indicates that the vehicle is an illegally operated vehicle.

[0069] The above embodiments are further elaborations and illustrations of the present invention for ease of understanding, and are not limitations of the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for identifying illegal operating vehicles based on big data, characterized in that: include: Obtain historical driving data sets of several sample vehicles, extract several identification feature indicators of the sample vehicles, and establish a vehicle assessment model for illegally operated vehicles; The vehicle evaluation model is a correspondence between several identification characteristic indicators of a vehicle and the total evaluation value of the vehicle; After obtaining the driving data of the vehicle to be identified and extracting the identification feature index, the vehicle evaluation model is used to calculate the total vehicle evaluation value; When the total vehicle assessment value exceeds the judgment threshold, the vehicle to be identified is an illegally operated vehicle.

2. According to the method for identifying illegal vehicles based on big data in claim 1, it is characterized in that: In the vehicle evaluation model, a corresponding weight coefficient is assigned to each identification feature index; The identification feature indicators are weighted and summed based on their corresponding weight coefficients to obtain the total vehicle evaluation value; A vehicle evaluation total value set of the sample vehicle is calculated based on the vehicle evaluation model and the identification characteristic indicators of the sample vehicle, and a determination threshold is determined based on the data boundary in the vehicle evaluation total value set.

3. The method for identifying illegal vehicles based on big data according to claim 1 or 2, characterized in that: The identification feature indicators include hot area indicators: Obtain a set of times that sample vehicles pass through the popular area; and calculate the average number of times that the sample vehicles pass through the popular area; Taking the average number of times as the middle value, the number of times is divided into 2n intervals at equal intervals according to the maximum value of the number difference in the number set; The percentage of sample vehicles falling in the corresponding frequency interval to the total sample vehicles is used as the hot area indicator.

4. The method for identifying illegal vehicles based on big data according to claim 3 is characterized in that: The method for determining the hot area includes: Obtain the total number of times that several sample vehicles pass through a certain area, and traverse each area to obtain the area frequency set; The length of the hot interval is calculated by multiplying the difference between the maximum and minimum times in the regional frequency concentration by the set percentage, and the hot lower limit number is obtained by subtracting the hot interval length from the maximum number; An area where the total number of times that several sample vehicles pass through is between the lower limit and the maximum limit is a hot area.

5. The method for identifying illegal vehicles based on big data according to claim 1 or 2, characterized in that: The identification feature index includes the number of stops index: Get a stop set containing the number of stops of each sample vehicle; and calculate the average number of stops of all sample vehicles; Taking the average number of stops as the middle value, the stop number intervals are divided into 2n intervals at equal intervals according to the maximum value of the number difference in the stop number set; the percentage of sample vehicles falling in the corresponding stop number interval to the total sample vehicles is used as the stop number indicator.

6. The method for identifying illegal vehicles based on big data according to claim 1 or 2, characterized in that: The identification characteristic index includes an average speed index: Obtain a speed set containing the average speed of each sample vehicle; and calculate the average speed of all sample vehicles; Taking the average speed as the middle value, the speed set is divided into 2n speed intervals at equal intervals according to the maximum value of the speed difference in the speed set; The percentage of sample vehicles falling in the corresponding speed range to the total sample vehicles is taken as the average speed indicator.

7. The method for identifying illegal vehicles based on big data according to claim 1 or 2, characterized in that: The identification feature index includes a driving distance index: Obtain a distance set containing the total distance traveled by each sample vehicle; and calculate the average distance of all sample vehicles; Taking the average distance as the middle value, the distance set is divided into 2n distance intervals at equal intervals according to the maximum value of the distance difference in the distance set; The percentage of sample vehicles falling in the corresponding distance interval to the total sample vehicles is taken as the average speed indicator.

8. The method for identifying illegal vehicles based on big data according to claim 2 is characterized in that: When the sample vehicle is an operating vehicle, extracting identification feature indicators of the sample vehicle to obtain a sample indicator set; The weighted sum of the identification feature indicators based on their corresponding weight coefficients is set to 1, and the weight coefficient values ​​corresponding to each identification feature indicator are obtained by fitting training using the sample indicator set; Based on the weight coefficient value, the vehicle evaluation value is calculated for the sample indicator set, and the minimum value of the vehicle evaluation value obtained is used as the judgment threshold.

9. The method for identifying illegal vehicles based on big data according to claim 2 or 3, characterized in that: When the sample vehicle is a non-operating vehicle, extracting identification feature indicators of the sample vehicle to obtain a sample indicator set; Set the weighted sum of the identification feature indicators based on their corresponding weight coefficients to zero, and use the sample indicator set for fitting training to obtain the weight coefficient values ​​corresponding to each identification feature indicator; Based on the weight coefficient value, the vehicle evaluation value of the sample indicator set is calculated, and the maximum value of the vehicle evaluation value obtained is used as the judgment threshold.

10. A system for identifying illegal vehicles based on big data, applicable to the method for identifying illegal vehicles according to any one of claims 1 to 9, characterized in that: include: An index extraction module extracts corresponding identification feature indexes from the vehicle's driving data; The evaluation model module stores the vehicle evaluation model and updates the model according to the historical driving data of the updated sample vehicles; the vehicle retrieval module extracts the driving data of the sample vehicles or the vehicles to be identified from the driving data of all vehicles.

Citation Information

Patent Citations

  • Taxi operation data-based illegal-operating vehicle identification method

    CN105427620A

  • Illegal operating vehicle identification method based on automotive electronic mark data

    CN110136448A

  • Illegal operating vehicle identification method based on multi-source data

    CN111862606A

  • Illegal operation vehicle identification method and electronic equipment

    CN115841243A