Traffic data analysis system and method based on big data

Abstract

The present invention discloses a traffic data analysis system and method based on big data, which belongs to the technical field of traffic data analysis, and includes an urban road route division unit, an urban road route calibration unit, a vehicle vibration detection module, a data preliminary screening module, a vehicle vibration classification model, a data processing unit, a data merging module, and a lane situation analysis unit; wherein the data processing unit is used for data correction processing; and the lane situation analysis unit calculates the situation value T of each lane. When recommending the best travel route, the present invention comprehensively considers the road congestion and the road conditions of the road to ensure that the driver selects the optimal route with relatively good road conditions and relatively good congestion. The present invention can perform situation analysis based on traffic data, and timely warn of road situations, so that the warning efficiency is improved and the accuracy is high.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic data analysis, and in particular to a traffic data analysis system and method based on big data. Background Art

[0002] Traffic data is large in volume and diverse in types, and has the typical characteristics of big data.

[0003] The traffic data analysis system is used to monitor and analyze traffic data (such as highway data) and obtain road condition information. On the one hand, it can guide travelers to choose appropriate travel routes, and on the other hand, it can provide references for road optimization, traffic control, congestion relief, etc. to traffic planning departments.

[0004] For example, Zhao Bo conducted research on the design of an urban road traffic operation status monitoring and guidance system. Based on real-time monitoring of traffic operation conditions, the traffic guidance screen provides traffic participants with real-time road congestion conditions, notifies and broadcasts sudden traffic incidents, and guides the majority of traffic participants to choose the best travel routes, thereby providing intuitive and accurate travel services to the majority of traffic participants and improving the efficiency of traffic guidance services and urban road control of the public security traffic management department.

[0005] The defects of the prior art can be divided into the following two aspects:

[0006] (1) The above optimal travel route only considers the road congestion, but does not consider the road condition (damage, subsidence), etc., which may result in the driver missing the optimal route when he has multiple options.

[0007] (2) At the same time, the system’s analysis and processing of traffic data is not strong enough, and it is unable to conduct situation analysis based on traffic data and issue timely warnings of road conditions.

[0008] Based on this, the present invention designs a traffic data analysis system and method based on big data to solve the above problems. Summary of the invention

[0009] In view of the above-mentioned shortcomings of the prior art, the present invention provides a traffic data analysis system and method based on big data.

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0011] A traffic data analysis system based on big data, comprising:

[0012] The urban road route division unit is used to configure n lane routes; each lane route is equally divided into p lane units, and the lane unit is named A 1～i -B 1～t -C 1～p;

[0013] The urban road route calibration unit is used to calibrate the road endpoints, turning points, traffic markings, and lane unit coordinates in the city map with the GPS coordinates;

[0014] A vehicle vibration detection module is placed or installed on the vehicle and is used to detect vibration data of the vehicle during driving in each lane unit;

[0015] The data preliminary screening module is used to compare the vibration data with the set time and set value. If it is not within the set time or is less than the set value, it is considered to be a normal state and no processing is performed; if it is within the set time and higher than the set value, the vibration data is input into the vehicle vibration classification model;

[0016] Vehicle vibration classification model: establish a vehicle vibration classification model based on a three-dimensional convolutional neural network, input vibration data into the vehicle vibration classification model, output lane units and their road defect types: null value, crack, subsidence or damage; the output data is sent to the data processing unit;

[0017] The data processing unit is used to collect data of each vehicle passing through the lane within a set time, and correct the data to obtain corrected data; the specific steps are:

[0018] Input the data of each vehicle in the same lane unit;

[0019] Determine whether the road defect types of each data are all null values;

[0020] If yes, it means that the lane unit is normal and no processing is done; this step can prioritize the lane unit data without any defects;

[0021] If not, then count the data of each vehicle in the lane unit received within the last hour and calculate the null value rate β of the data:

[0022] β=U 空 / U 总

[0023] Among them, U 空 Indicates the number of null values ​​for road defect types in the last hour of data, U 总 Indicates the total amount of data;

[0024] Determine whether 1-β is ≤5%;

[0025] If yes, it means that the defect of the lane unit is eliminated or there is a misjudgment; the road defect type of the lane unit is corrected to a null value;

[0026] If not, calculate the null value rate α of the total data received:

[0027] α=V空 / U 总

[0028] Among them, V 空 Indicates the number of null values ​​of road defect types in the total data, U 总 Indicates the total amount of data;

[0029] Determine whether 1-α is ≥ 20%;

[0030] If not, it is considered that the defect of the lane unit is not prominent, and the road defect type of the lane unit is corrected to a null value;

[0031] If so, it is considered that the defect of the lane unit is prominent, and it is determined whether the defect type of the non-empty road is of the same type;

[0032] If they are of the same type, the road defect type of the lane unit is modified to the non-empty road defect type;

[0033] If they are not all of the same type, calculate the proportion of each non-empty road defect type:

[0034] e=K 非空 / U 总

[0035] Among them, K 非空 represents the number of non-empty road defect types, U 总 Indicates the total amount of data;

[0036] Sort the proportions e of each non-empty road defect type from largest to smallest, and select the non-empty road defect type with the largest e to correct the road defect type of the lane unit;

[0037] Then continue to input the data of each vehicle in the next lane unit for correction processing;

[0038] After the repair is completed, the data of all non-empty road defect types are output as valid data;

[0039] A data merging module is used to merge valid data with the same road defect type within the range of g lane units to obtain the length j of the merged data;

[0040] The lane situation analysis unit is used to obtain the merged valid data and calculate the situation value T of each lane according to the following formula:

[0041] T=a 3 *m3+b 2 *m2+c*m1;

[0042] Among them, m1 represents the proportion of cracks in the valid data, m2 represents the proportion of subsidence, and m3 represents the proportion of damage; a represents the number of valid data with j≥3, b represents the number of valid data with j=2, and c represents the number of valid data with j=1;

[0043] The smaller the lane situation value T is, the better the road condition is;

[0044] When recommending the best travel route, the route recommendation value is calculated in the following way:

[0045] Route recommendation value = ∑ traffic congestion index*w+∑ lane situation value T*(1-w);

[0046] Among them, ∑ traffic congestion index represents the sum of traffic congestion indexes of lanes passed by each route, and ∑ lane situation value represents the sum of lane situation values ​​of lanes passed by each route;

[0047] Calculate the route recommendation values ​​respectively, and obtain the route with the highest route recommendation value, which is the best travel route.

[0048] Furthermore, the urban road route division unit is used to number the roads in the city map A 1～i , i represents the number of roads, and then the lanes are numbered according to the number of lanes t. 1～i -B 1～t ; According to the city map, road length S 1～i , a total of n lane routes are configured; each lane route is equally divided into p lane units, that is, the lane unit length is S 1～i / p; Lane unit named A 1～i -B 1～t -C 1～p .

[0049] Furthermore, the vibration data includes the time when the vibration occurs and the vibration value.

[0050] Furthermore, in the model training phase of the three-dimensional convolutional neural network classification model, vibration data of vehicles on normal roads, vibration data of vehicles on cracked roads, vibration data of vehicles on subsidence roads, and vibration data of vehicles on damaged roads are collected for model training.

[0051] Furthermore, d takes a value of 2-3.

[0052] Furthermore, the specific method of data merging is:

[0053] Enter the corrected lane unit valid data A i -B t -C p and types of road defects;

[0054] Obtain whether there is the same road defect type within the range of g lane units in front of the lane where the lane unit is located;

[0055] If not, then directly output A i -B t -C p and types of road defects;

[0056] If so, then output lane unit range A i -B t -C p-j～p and its road defect type, and data length j.

[0057] Furthermore, j is an integer and is less than g.

[0058] Furthermore, when T>2.2%p, w=0.85~0.88; when T≤2.2%p, w=0.71~0.73.

[0059] Furthermore, it also includes a display unit for displaying lane information of T>2.2%p, including the lane situation value T, cracks, subsidence, and damaged lane units / lane unit range information of the lane.

[0060] In order to better achieve the purpose of the present invention, the present invention also provides a traffic data analysis method based on big data, comprising the following steps:

[0061] 1. Configure n lane routes and divide each lane route into p lane units. The lane units are named A 1～i -B 1～t -C 1～p ;

[0062] 2. Calibrate the road endpoints, turning points, traffic markings, and lane unit coordinates in the city map with the GPS coordinates;

[0063] 3. Detect the vibration data of the vehicle during driving in each lane unit; the vibration data includes the time when the vibration occurs and the vibration value;

[0064] 4. Compare the vibration data with the set time and set value. If it is not within the set time or is less than the set value, it is considered to be a normal state and no processing is done. If it is within the set time and higher than the set value, the vibration data is input into the vehicle vibration classification model.

[0065] 5. Establish a vehicle vibration classification model based on a three-dimensional convolutional neural network, input vibration data into the vehicle vibration classification model, output lane units and their road defect types: null value, crack, subsidence or damage; the output data is sent to the data processing unit;

[0066] 6. Collect data of each vehicle passing through the lane within a set time, and correct the data to obtain corrected data; the specific steps are:

[0067] 6.1. Input the data of each vehicle in the same lane unit;

[0068] 6.2. Determine whether the road defect types of each data are all null values;

[0069] If yes, it means that the lane unit is normal and no processing is done; this step can prioritize the lane unit data without any defects;

[0070] If not, then count the data of each vehicle in the lane unit received within the last hour and calculate the null value rate β of the data:

[0071] β=U 空 / U 总

[0072] Among them, U 空 Indicates the number of null values ​​for road defect types in the last hour of data, U 总 Indicates the total amount of data;

[0073] Determine whether 1-β is ≤5%;

[0074] If yes, it means that the defect of the lane unit is eliminated or there is a misjudgment; the road defect type of the lane unit is corrected to a null value;

[0075] If not, calculate the null value rate α of the total data received:

[0076] α=V 空 / U 总

[0077] Among them, V 空 Indicates the number of null values ​​of road defect types in the total data, U 总 Indicates the total amount of data;

[0078] Determine whether 1-α is ≥ 20%;

[0079] If not, it is considered that the defect of the lane unit is not prominent, and the road defect type of the lane unit is corrected to a null value;

[0080] If so, it is considered that the defect of the lane unit is prominent, and it is determined whether the defect type of the non-empty road is of the same type;

[0081] If they are of the same type, the road defect type of the lane unit is modified to the non-empty road defect type;

[0082] If they are not all of the same type, calculate the proportion of each non-empty road defect type:

[0083] e=K 非空 / U 总

[0084] Among them, K 非空 represents the number of non-empty road defect types, U 总 Indicates the total amount of data;

[0085] Sort the proportions e of each non-empty road defect type from largest to smallest, and select the non-empty road defect type with the largest e to correct the road defect type of the lane unit;

[0086] 6.3. Then continue to input the data of each vehicle in the next lane unit, and then use step 6.2 to perform correction processing;

[0087] 6.4. After the repair is completed, the data of all non-empty road defect types are output as valid data;

[0088] 7. Merge the valid data with the same road defect type within the g lane units to obtain the length j of the merged data, where g takes a value of 2 to 3;

[0089] 8. The lane situation analysis unit obtains the merged valid data and calculates the situation value T of each lane according to the following formula:

[0090] T=a 3 *m3+b 2 *m2+c*m1;

[0091] Among them, m1 represents the proportion of cracks in the valid data, m2 represents the proportion of subsidence, and m3 represents the proportion of damage; a represents the number of valid data with j≥3, b represents the number of valid data with j=2, and c represents the number of valid data with j=1;

[0092] Sending lane information with a lane situation value T>2.2%p to a display unit;

[0093] The smaller the lane situation value T is, the better the road condition is;

[0094] IX. When recommending the best travel route, the route recommendation value is calculated in the following manner:

[0095] Route recommendation value = ∑ traffic congestion index*w+∑ lane situation value T*(1-w)

[0096] Among them, ∑ traffic congestion index represents the sum of traffic congestion indexes of lanes passed by each route, and ∑ lane situation value represents the sum of lane situation values ​​of lanes passed by each route;

[0097] When T>2.2%p, w=0.85~0.88;

[0098] When T≤2.2%p, w=0.71~0.73;

[0099] Calculate the route recommendation values ​​respectively, and obtain the route with the highest route recommendation value, which is the best travel route.

[0100] 10. The display unit displays lane information with T>2.2%p, including the lane situation value T, cracks, subsidence, and damaged lane units / lane unit range information of the lane.

[0101] Compared with the prior art, the present invention has the following beneficial effects:

[0102] When recommending the best travel route, the present invention comprehensively considers the road congestion and road conditions to ensure that the driver selects the best route with relatively good road conditions and relatively less congestion.

[0103] The present invention can perform situation analysis based on traffic data and timely warn of road situations, which is conducive to timely understanding the status of each lane and issuing warnings when major defects occur on the road as a whole, thereby improving warning efficiency and increasing accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] 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 prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0105] Figure 1 This is a schematic diagram of the urban road route division.

[0106] Figure 2 This is a schematic diagram of a traffic data analysis system based on big data of the present invention.

[0107] Figure 3 The data processing flow chart is shown in Figure 2.

[0108] Figure 4 This is a schematic diagram showing situation warning. DETAILED DESCRIPTION

[0109] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0110] Embodiment 1

[0111] In some embodiments, please refer to the attached instructions. Figure 1 , a traffic data analysis system based on big data, comprising:

[0112] Urban road route division unit, used to number the roads in the city map A 1～i , i represents the number of roads, and then the lanes are numbered according to the number of lanes t. 1～i -B 1～t ; According to the city map, road length S 1～i , a total of n lane routes are configured; each lane route is equally divided into p lane units, that is, the lane unit length is S 1～i / p (e.g. 50cm); lane unit is named A 1～i -B 1～t -C 1～p ;

[0113] For example, A1-B2-C 1300 represents the 1300th lane unit of lane B2 of road A1;

[0114] The urban road route calibration unit is used to calibrate the road endpoints, turning points, traffic markings, and lane unit coordinates in the city map with the GPS coordinates;

[0115] The vehicle vibration detection module is placed or installed on the vehicle and is used to detect the vibration data of the vehicle during the driving process of each lane unit; the vibration data includes the time when the vibration occurs and the vibration value;

[0116] The data preliminary screening module is used to compare the vibration data with the set time and set value. If it is not within the set time or is less than the set value, it is considered to be in a normal state and no processing is performed; if it is within the set time and higher than the set value, the vibration data is input into the vehicle vibration classification model; vibrations that are not within the detection time range (for example, only detect within 24 hours today) and are small and do not affect the safe driving of the vehicle can be ignored;

[0117] Vehicle vibration classification model, establish a vehicle vibration classification model based on a three-dimensional convolutional neural network (CNN). In the model training phase of the three-dimensional convolutional neural network classification model, collect vibration data of normal vehicles on the road, vibration data of vehicles with cracks on the road (2cm<crack size≤5cm), vibration data of vehicles with subsidence on the road (no damage, only downward subsidence size>5cm, depth greater than 1cm), and vibration data of vehicles with damaged roads (damaged with pits, damage size>5cm), and perform model training. The vibration data is input into the vehicle vibration classification model, and the lane unit and its road defect type are output: null value (representing normal), crack, subsidence, damage; for example, A1-B2-C1300 cracks, the output data is sent to the data processing unit;

[0118] The data processing unit is used to collect data of each vehicle passing through the lane within a set time, and correct the data to obtain corrected data; the specific steps are:

[0119] Input the data of each vehicle in the same lane unit;

[0120] Example 1: Same lane unit: Car 1: A1-B1-C1 empty value, Car 2: A1-B1-C1 sinking, Car 3: A1-B1-C1 empty value;

[0121] Determine whether the road defect types of each data are all null values;

[0122] If yes, it means that the lane unit is normal and no processing is done; this step can prioritize the lane unit data without any defects;

[0123] If not, then count the data of each vehicle in the lane unit received within the last hour and calculate the null value rate β of the data:

[0124] β=U 空 / U 总

[0125] Among them, U 空 Indicates the number of null values ​​for road defect types in the last hour of data, U 总 Indicates the total amount of data;

[0126] Determine whether 1-β is ≤5%;

[0127] If yes, it means that the defect of the lane unit is eliminated or there is a misjudgment; the road defect type of the lane unit is corrected to a null value, for example, the correction data: A1-B1-C1 null value (correction) (Example 1);

[0128] If not, calculate the null value rate α of the total data received:

[0129] α=V 空 / U 总

[0130] Among them, V 空 Indicates the number of null values ​​of road defect types in the total data, U 总 Indicates the total amount of data;

[0131] Determine whether 1-α is ≥ 20%;

[0132] If not, it is considered that the defect of the lane unit is not prominent, and the road defect type of the lane unit is corrected to a null value;

[0133] If so, it is considered that the defect of the lane unit is prominent, and it is determined whether the defect type of the non-empty road is of the same type;

[0134] If they are of the same type, the road defect type of the lane unit is corrected to the non-empty road defect type, for example, the correction data: A1-B1-C2 subsidence (correction);

[0135] If they are not all of the same type, calculate the proportion of each non-empty road defect type:

[0136] e=K 非空 / U 总

[0137] Among them, K 非空 represents the number of non-empty road defect types, U 总 Indicates the total amount of data;

[0138] Sort the proportions e of each non-empty road defect type from largest to smallest, and select the non-empty road defect type with the largest e to correct the road defect type of the lane unit;

[0139] Then continue to input the data of each vehicle in the next lane unit for correction processing;

[0140] After the repair is completed, the data of all non-empty road defect types are output as valid data;

[0141] For example, first correct the A1-B1-C1 lane unit of A1-B1, and then correct the A1-B1-C1 lane unit of A1-B1. 2······ The lane unit is corrected, and then the lane units of A1-B2 are corrected in sequence, until all lane units of A1 are corrected, and then all lane units of A2 are corrected;

[0142] Through the above steps, road defects within the lane unit / range can be discovered in time, and the road can be corrected regularly;

[0143] The data merging module is used to merge the valid data with the same road defect type within the range of g lane units to obtain the length j of the merged data, where g takes a value of 2 to 3;

[0144] The merged data length j refers to the number of lane units contained in the merged data;

[0145] For example, A1-B2-C 1300 Subsidence, A1-B2-C 1301 Subsidence, A1-B2-C 1302 Subsidence; after merging, it becomes A1-B2-C 1300～1302 Subsidence, j is 1302-1300+1＝3 ;

[0146] The specific method is:

[0147] Enter the corrected lane unit valid data A i -B t -C p and its road defect type; e.g. A1-B2-C 1302 crack;

[0148] Obtain whether there is the same road defect type within the range of g lane units in front of the lane where the lane unit is located;

[0149] If not, then directly output A i -B t -C p and its road defect type; e.g. A1-B2-C 1302 Cracks; indicating that there is no similar road defect near the lane unit;

[0150] If so, then output lane unit range A i -B t -C p-j～p Its road defect type, and data length j (j is an integer and less than g); for example A1-B2-C 1300～1302 crack, j = 3;

[0151] It is beneficial to reduce data and merge events, and reduce the amount of calculation in the next step;

[0152] The lane situation analysis unit is used to obtain the merged valid data and calculate the situation value T of each lane according to the following formula:

[0153] T=a 3 *m3+b 2 *m2+c*m1;

[0154] Among them, m1 represents the proportion of cracks in the valid data, m2 represents the proportion of subsidence, and m3 represents the proportion of damage; a represents the number of valid data with j≥3, b represents the number of valid data with j=2, and c represents the number of valid data with j=1;

[0155] The lane information with a lane situation value T>2.2%p is sent to the display unit. T>2.2%p indicates that the road condition may affect traffic safety and serves to alert the road management department.

[0156] The smaller the lane situation value T is, the better the road condition is;

[0157] When recommending the best travel route, the route recommendation value is calculated in the following way:

[0158] Route recommendation value = ∑ traffic congestion index*w+∑ lane situation value T*(1-w);

[0159] Among them, ∑ traffic congestion index represents the sum of traffic congestion indexes of lanes passed by each route, and ∑ lane situation value represents the sum of lane situation values ​​of lanes passed by each route;

[0160] When T>2.2%p, w=0.85~0.88;

[0161] When T≤2.2%p, w=0.71~0.73;

[0162] Select the top d (value 2-3) lane situation values ​​T to calculate the route recommendation values ​​respectively, and obtain the route with the highest route recommendation value, which is the best travel route.

[0163] When recommending the best travel route, the present invention comprehensively considers the road congestion and road conditions to ensure that the driver selects the best route with relatively good road conditions and relatively less congestion.

[0164] The display unit is used to display lane information with T>2.2%p, including the lane situation value T, cracks, subsidence, and damaged lane units / lane unit range information of the lane, which is conducive to timely grasping the status of each lane and issuing an early warning when major defects appear on the road as a whole. The early warning efficiency is improved and the accuracy is high. The present invention can perform situation analysis based on traffic data and issue a timely early warning of road situation.

[0165] Embodiment 2

[0166] This embodiment discloses a traffic data analysis method based on big data, comprising the following steps:

[0167] 1. Configure n lane routes and divide each lane route into p lane units. The lane units are named A 1～i -B 1～t -C 1～p ;

[0168] For example, A1-B2-C 1300 represents the 1300th lane unit of lane B2 of road A1;

[0169] 2. Calibrate the road endpoints, turning points, traffic markings, and lane unit coordinates in the city map with the GPS coordinates;

[0170] 3. Detect the vibration data of the vehicle during driving in each lane unit; the vibration data includes the time when the vibration occurs and the vibration value;

[0171] Fourth, compare the vibration data with the set time and set value. If it is not within the set time or is less than the set value, it is considered to be in a normal state and no processing is performed; if it is within the set time and higher than the set value, the vibration data is input into the vehicle vibration classification model; vibrations that are not within the detection time range (for example, only detect within 24 hours today) and are small and do not affect the safe driving of the vehicle can be ignored;

[0172] 5. Establish a vehicle vibration classification model based on a three-dimensional convolutional neural network (CNN). In the model training phase of the three-dimensional convolutional neural network classification model, collect vibration data of vehicles on normal roads, vehicles on cracked roads, vehicles on subsidence roads, and vehicles on damaged roads for model training. The vibration data is input into the vehicle vibration classification model, and the lane unit and its road defect type are output: null value (representing normal), crack, subsidence, and damage; for example, A1-B2-C 1300 cracks, the output data is sent to the data processing unit;

[0173] 6. Collect data of each vehicle passing through the lane within a set time, and correct the data to obtain corrected data; the specific steps are:

[0174] 6.1. Input the data of each vehicle in the same lane unit;

[0175] Example 1: Same lane unit: Car 1: A1-B1-C1 empty value, Car 2: A1-B1-C1 sinking, Car 3: A1-B1-C1 empty value;

[0176] 6.2. Determine whether the road defect types of each data are all null values;

[0177] If yes, it means that the lane unit is normal and no processing is done; this step can prioritize the lane unit data without any defects;

[0178] If not, then count the data of each vehicle in the lane unit received within the last hour and calculate the null value rate β of the data:

[0179] β=U 空 / U 总

[0180] Among them, U 空 Indicates the number of null values ​​for road defect types in the last hour of data, U 总 Indicates the total amount of data;

[0181] Determine whether 1-β is ≤5%;

[0182] If yes, it means that the defect of the lane unit is eliminated or there is a misjudgment; the road defect type of the lane unit is corrected to a null value, for example, the correction data: A1-B1-C1 null value (correction) (Example 1);

[0183] If not, calculate the null value rate α of the total data received:

[0184] α=V 空 / U 总

[0185] Among them, V 空 Indicates the number of null values ​​of road defect types in the total data, U 总 Indicates the total amount of data;

[0186] Determine whether 1-α is ≥ 20%;

[0187] If not, it is considered that the defect of the lane unit is not prominent, and the road defect type of the lane unit is corrected to a null value;

[0188] If so, it is considered that the defect of the lane unit is prominent, and it is determined whether the defect type of the non-empty road is of the same type;

[0189] If they are of the same type, the road defect type of the lane unit is corrected to the non-empty road defect type, for example, the correction data: A1-B1-C2 subsidence (correction);

[0190] If they are not all of the same type, calculate the proportion of each non-empty road defect type:

[0191] e=K 非空 / U 总

[0192] Among them, K 非空 represents the number of non-empty road defect types, U 总 Indicates the total amount of data;

[0193] Sort the proportions e of each non-empty road defect type from largest to smallest, and select the non-empty road defect type with the largest e to correct the road defect type of the lane unit;

[0194] 6.3. Then continue to input the data of each vehicle in the next lane unit, and then use step 6.2 to perform correction processing;

[0195] 6.4. After the repair is completed, the data of all non-empty road defect types are output as valid data;

[0196] For example, first correct the A1-B1-C1 lane unit of A1-B1, and then correct the A1-B1-C1 lane unit of A1-B1. 2······ The lane unit is corrected, and then the lane units of A1-B2 are corrected in sequence, until all lane units of A1 are corrected, and then all lane units of A2 are corrected;

[0197] Through the above steps, road defects within the lane unit / range can be discovered in time, and the road can be corrected regularly;

[0198] 7. Merge the valid data with the same road defect type within the g lane units to obtain the length j of the merged data, where g takes a value of 2 to 3;

[0199] For example, A1-B2-C 1300 Subsidence, A1-B2-C 1301 Subsidence, A1-B2-C 1302 Subsidence; after merging, it becomes A1-B2-C 1300～1302 sinking;

[0200] The specific method is:

[0201] 7.1. Input the corrected lane unit valid data A i -B t -C p and its road defect type; e.g. A1-B2-C 1302 crack;

[0202] 7.2. Obtain whether there is the same road defect type within the range of g lane units in front of the lane where the lane unit is located;

[0203] If not, then directly output A i -B t -C p and its road defect type; e.g. A1-B2-C 1302 crack;

[0204] If so, then output lane unit range A i -B t -C p-j～p Its road defect type, and data length j (j is an integer and less than g); for example A1-B2-C 1300～1302 crack, j = 2;

[0205] It is beneficial to reduce data and merge events, and reduce the amount of calculation in the next step;

[0206] 8. The lane situation analysis unit obtains the merged valid data and calculates the situation value T of each lane according to the following formula:

[0207] T=a 3 *m3+b 2 *m2+c*m1;

[0208] Among them, m1 represents the proportion of cracks in the valid data, m2 represents the proportion of subsidence, and m3 represents the proportion of damage; a represents the number of valid data with j≥3, b represents the number of valid data with j=2, and c represents the number of valid data with j=1;

[0209] The lane information with a lane situation value T>2.2%p is sent to the display unit. T>2.2%p indicates that the road condition may affect traffic safety and serves to alert the road management department.

[0210] The smaller the lane situation value T is, the better the road condition is;

[0211] IX. When recommending the best travel route, the route recommendation value is calculated in the following manner:

[0212] Route recommendation value = ∑ traffic congestion index*w+∑ lane situation value T*(1-w)

[0213] Among them, ∑ traffic congestion index represents the sum of traffic congestion indexes of lanes passed by each route, and ∑ lane situation value represents the sum of lane situation values ​​of lanes passed by each route;

[0214] When T>2.2%p, w=0.85~0.88;

[0215] When T≤2.2%p, w=0.71~0.73;

[0216] Calculate the route recommendation values ​​respectively, and obtain the route with the highest route recommendation value, which is the best travel route.

[0217] When recommending the best travel route, the present invention comprehensively considers the road congestion and road conditions to ensure that the driver selects the best route with relatively good road conditions and relatively less congestion.

[0218] 10. The display unit displays lane information with T>2.2%p, including the lane situation value T, cracks, subsidence, and damaged lane units / lane unit range information of the lane, which is conducive to timely grasping the status of each lane and issuing an early warning when major defects appear on the road as a whole. The early warning efficiency is improved and the accuracy is high. The present invention can perform situation analysis based on traffic data and issue an early warning of road situation in a timely manner.

[0219] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A traffic data analysis system based on big data, characterized in that: include: The urban road route division unit is used to configure n lane routes and divide each lane route into p lane units. The lane unit is named A 1～i -B 1～t -C 1～p ; i represents the number of roads, t represents the number of lanes; The urban road route calibration unit is used to calibrate the road endpoints, turning points, traffic markings, and lane unit coordinates in the city map with the GPS coordinates; The vehicle vibration detection module is used to detect the vibration data of the vehicle during the driving process of each lane unit; A data preliminary screening module is used to preliminarily screen the vibration data and input it into the vehicle vibration classification model; A vehicle vibration classification model that outputs lane units and their road defect types; The data processing unit is used to collect data of each vehicle passing through the lane within a set time, and correct the data to obtain corrected data; the specific steps are: Input the data of each vehicle in the same lane unit; Determine whether the road defect types of each data are all null values; If so, no action will be taken; If not, then count the data of each vehicle in the lane unit received within the last hour and calculate the null value rate β of the data: β=U 空 / IN 总 Among them, U 空 Indicates the number of null values ​​for road defect types in the last hour of data, U 总 Indicates the total amount of data; Determine whether 1-β is ≤5%; If yes, it means that the defect of the lane unit is eliminated or there is a misjudgment; the road defect type of the lane unit is corrected to a null value; If not, calculate the null value rate α of the total data received: α=V 空 / U 总 Among them, V 空 Indicates the number of null values ​​of road defect types in the total data, U 总 Indicates the total amount of data; Determine whether 1-α is ≥ 20%; If not, it is considered that the defect of the lane unit is not prominent, and the road defect type of the lane unit is corrected to a null value; If yes, it is considered that the defect of the lane unit is prominent, the road defect type of the lane unit is a non-empty road defect type, and it is determined whether the non-empty road defect type is of the same type; If they are of the same type, the road defect type of the lane unit is modified to the non-empty road defect type; If they are not all of the same type, calculate the proportion of each non-empty road defect type: e=K 非空 / YOU 总 Among them, K 非空 represents the number of non-empty road defect types, U 总 Indicates the total amount of data; Sort the proportions e of each non-empty road defect type from largest to smallest, and select the non-empty road defect type with the largest e to correct the road defect type of the lane unit; Then, the data of each vehicle in the next lane unit is input for correction processing; after the correction is completed, the data of all non-empty road defect types are output as valid data; A data merging module is used to merge valid data with the same road defect type within the range of g lane units to obtain the length j of the merged data; The lane situation analysis unit is used to obtain the merged valid data and calculate the situation value T of each lane according to the following formula: T=a 3 *m3+b 2 *m2+c*m1; Among them, m1 represents the proportion of cracks in the valid data, m2 represents the proportion of subsidence, and m3 represents the proportion of damage; a represents the number of valid data with j≥3, b represents the number of valid data with j=2, and c represents the number of valid data with j=1; When recommending the best travel route, the route recommendation value is calculated in the following way: Route recommendation value = ∑ traffic congestion index*w+∑ lane situation value T*(1-w); Among them, ∑ traffic congestion index represents the sum of traffic congestion indexes of lanes passed by each route, and ∑ lane situation value represents the sum of lane situation values ​​of lanes passed by each route; Calculate the route recommendation values ​​respectively, and obtain the route with the highest route recommendation value, which is the best travel route.

2. The traffic data analysis system based on big data according to claim 1 is characterized in that: Urban road route division unit, used to number the roads in the city map A 1～i , i represents the number of roads, and then the lanes are numbered according to the number of lanes t. 1～i -B 1～t ; According to the city map, road length S 1～i , a total of n lane routes are configured; each lane route is equally divided into p lane units, that is, the lane unit length is S 1～i / p; Lane unit named A 1～i -B 1～t -C 1～p .

3. The traffic data analysis system based on big data according to claim 2 is characterized in that: The vibration data includes the time when the vibration occurs and the vibration value.

4. The traffic data analysis system based on big data according to claim 3 is characterized in that: During the model training phase of the three-dimensional convolutional neural network classification model, vibration data of vehicles on normal roads, vibration data of vehicles on cracked roads, vibration data of vehicles on subsidence roads, and vibration data of vehicles on damaged roads are collected for model training.

5. The traffic data analysis system based on big data according to claim 4 is characterized in that: The value of g is 2 to 3, and the value of d is 2-3.

6. The traffic data analysis system based on big data according to claim 5 is characterized in that: The specific method of data merging is as follows: Enter the corrected lane unit valid data A i -B t -C p and types of road defects; Obtain whether there is the same road defect type within the range of g lane units in front of the lane where the lane unit is located; If not, then directly output A i -B t -C p and types of road defects; If so, then output lane unit range A i -B t -C p-j～p and its road defect type, and data length j.

7. The traffic data analysis system based on big data according to claim 6 is characterized in that: j is an integer and is less than g.

8. The traffic data analysis system based on big data according to claim 7 is characterized in that: When T>2.2%p, w=0.85~0.88; when T≤2.2%p, w=0.71~0.

73.

9. The traffic data analysis system based on big data according to claim 8 is characterized in that: It also includes a display unit for displaying lane information with T>2.2%p, including the lane situation value T, cracks, subsidence, and damaged lane units / lane unit range information of the lane.

10. A traffic data analysis method based on big data, characterized in that: Using the traffic data analysis system based on big data as claimed in claim 9, comprising the following steps:

1. Configure n lane routes and divide each lane route into p lane units. The lane units are named A 1～i -B 1～t -C 1～p ; i represents the number of roads, t represents the number of lanes; 2. Calibrate the road endpoints, turning points, traffic markings, and lane unit coordinates in the city map with the GPS coordinates; 3. Detect the vibration data of the vehicle during driving in each lane unit; the vibration data includes the time when the vibration occurs and the vibration value; 4. Compare the vibration data with the set time and set value. If it is not within the set time or is less than the set value, it is considered to be a normal state and no processing is done. If it is within the set time and higher than the set value, the vibration data is input into the vehicle vibration classification model.

5. Establish a vehicle vibration classification model based on a three-dimensional convolutional neural network, input vibration data into the vehicle vibration classification model, output lane units and their road defect types: null value, crack, subsidence or damage; the output data is sent to the data processing unit; 6. Collect data of each vehicle passing through the lane within a set time, and correct the data to obtain corrected data; the specific steps are: 6.

1. Input the data of each vehicle in the same lane unit; 6.

2. Determine whether the road defect types of each data are all null values; If yes, it means that the lane unit is normal and no processing is done; this step can prioritize the lane unit data without any defects; If not, then count the data of each vehicle in the lane unit received within the last hour and calculate the null value rate β of the data: β=U 空 / IN 总 in, U 空 Indicates the number of null values ​​for road defect types in the last hour of data, U 总 Indicates the total amount of data; Determine whether 1-β is ≤5%; If yes, it means that the defect of the lane unit has been eliminated or there is a misjudgment; Correct the road defect type of the lane unit to be null; If not, calculate the null value rate α of the total data received: α=V 空 / U 总 Among them, V 空 Indicates the number of null values ​​of road defect types in the total data, U 总 Indicates the total amount of data; Determine whether 1-α is ≥ 20%; If not, it is considered that the defect of the lane unit is not prominent, and the road defect type of the lane unit is corrected to a null value; If yes, it is considered that the defect of the lane unit is prominent, the road defect type of the lane unit is a non-empty road defect type, and it is determined whether the non-empty road defect type is of the same type; If they are of the same type, the road defect type of the lane unit is modified to the non-empty road defect type; If they are not all of the same type, calculate the proportion of each non-empty road defect type: e=K 非空 / YOU 总 Among them, K 非空 represents the number of non-empty road defect types, U 总 Indicates the total amount of data; Sort the proportions e of each non-empty road defect type from largest to smallest, and select the non-empty road defect type with the largest e to correct the road defect type of the lane unit; 6.

3. Then continue to input the data of each vehicle in the next lane unit, and then use step 6.2 to perform correction processing; 6.

4. After the repair is completed, the data of all non-empty road defect types are output as valid data; 7. Merge the valid data with the same road defect type within the g lane units to obtain the length j of the merged data, where g takes a value of 2 to 3; 8. The lane situation analysis unit obtains the merged valid data and calculates the situation value T of each lane according to the following formula: T=a 3 *m3+b 2 *m2+c*m1; Among them, m1 represents the proportion of cracks in the valid data, m2 represents the proportion of subsidence, and m3 represents the proportion of damage; a represents the number of valid data with j≥3, b represents the number of valid data with j=2, and c represents the number of valid data with j=1; Sending lane information with a lane situation value T>2.2%p to a display unit; The smaller the lane situation value T is, the better the road condition is; IX. When recommending the best travel route, the route recommendation value is calculated in the following manner: Route recommendation value = ∑ traffic congestion index*w+∑ lane situation value T*(1-w) Among them, ∑ traffic congestion index represents the sum of traffic congestion indexes of lanes passed by each route, and ∑ lane situation value represents the sum of lane situation values ​​of lanes passed by each route; When T>2.2%p, w=0.85~0.88; When T≤2.2%p, w=0.71~0.73; Calculate the route recommendation values ​​respectively, and obtain the route with the highest route recommendation value, which is the best travel route; 10. The display unit displays lane information with T>2.2%p, including the lane situation value T, cracks, subsidence, and damaged lane units / lane unit range information of the lane.

Images