Highway traffic inspection method and system based on big data processing
By using big data processing technology to integrate multi-source data for dynamic road condition classification and adaptive inspection route planning, the problems of high manpower consumption, data silos, and non-optimal route planning in existing highway inspections have been solved, achieving efficient and accurate highway inspections.
Patent Information
- Application Number
- CN202511325667.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-17
- Publication Date
- 2025-10-24
- Estimated Expiration
- 2045-09-17
AI Technical Summary
Existing highway inspection technologies suffer from problems such as high manpower consumption, significant susceptibility to weather and traffic conditions, severe data silos, fixed inspection frequencies that cannot be dynamically adjusted, and insufficient route planning, resulting in inadequate inspection accuracy and timeliness.
By employing a big data processing approach, multi-dimensional dynamic road condition classification, adaptive inspection frequency optimization, and optimal inspection route planning are used to integrate multi-source heterogeneous data, dynamically evaluate road conditions, and adaptively adjust the inspection rhythm and route. The improved VRPTW-PW algorithm is then used to generate the shortest time and highest priority driving route.
It has improved the accuracy and timeliness of highway inspections, eliminated information silos, dynamically adjusted inspection frequency and routes, increased inspection coverage and the rate of hazard detection, optimized resource allocation, and improved road safety and traffic efficiency.
Smart Images

Figure CN120832994A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of highway inspection, more particularly, the present application relates to a highway traffic inspection method and system based on big data processing. BACKGROUND
[0002] For a long time, highway inspection mainly relies on manual inspection or fixed period mechanical inspection. The staff carries simple instruments on the road, visually inspects cracks, pits, and sign damage by experience, and then manually enters the results into the system. This method not only consumes manpower, but also is greatly affected by weather and traffic conditions, and missed detection and wrong detection occur from time to time. More importantly, once the inspection plan is arranged, it is difficult to adjust, and it is not possible to flexibly change according to real-time road conditions or emergencies, resulting in that high-risk road sections cannot be paid attention to in time, and low-risk areas frequently occupy resources.
[0003] With the improvement of traffic informatization level, a large number of sensing devices have been laid on the road: Beidou floating car, geomagnetic vehicle detector, ETC gantry, bridge health monitor, weather station and various accident reporting platforms. Although the data types seem to be rich, they are scattered in traffic police, operation, maintenance, meteorological and other departments, and the formats, frequencies and coordinate systems are different. There is a lack of unified data cleaning, space-time alignment and standardization process, resulting in the phenomenon of "data island", it is difficult to form a complete portrait of the road state, and it is more difficult to make accurate decisions accordingly.
[0004] At present, there are automatic recognition based on video or simple early warning systems based on a single data source on the market, but there are generally the following deficiencies: first, only single-dimensional indicators such as traffic flow or structural safety are concerned, and there is a lack of a grading method that comprehensively considers real-time risk, infrastructure health and historical importance; second, the inspection frequency is fixed, and it is not possible to dynamically adjust according to the risk change, the interval between the last inspection or the emergency; third, the path planning stops at the "nearest principle" or simple scheduling, and does not consider the time efficiency constraints of vehicle endurance, real-time road conditions and high-priority road sections. Therefore, there is an urgent need for a complete technical solution that can integrate multi-source big data, dynamically evaluate road conditions, adaptively adjust the inspection rhythm and optimize the path, in order to truly improve the accuracy and timeliness of highway inspection. SUMMARY
[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a highway traffic inspection method and system based on big data processing.
[0006] In order to achieve the above-mentioned purpose, the present application provides the following technical scheme: The highway traffic inspection method based on big data processing comprises the following steps: S1, multi-dimensional dynamic road condition grading: cleaning and aligning each road real-time data, calculating congestion, accident, road surface and structure indicators, and comprehensively obtaining a 1-4 level risk grade of each kilometer road section; S2, adaptive patrol frequency tuning: give a basic patrol period according to the risk level, dynamically weight by risk change, time since last patrol and emergency, analyze the patrol priority of each road segment at this moment; S3, optimal patrol path planning: convert priority to time window constraint, use improved VRPTW-PW algorithm to generate the shortest time and highest priority driving route for all patrol vehicles under real-time traffic and vehicle endurance constraints.
[0007] Specifically, the specific process of S1 is as follows: Multi-source heterogeneous data access and preprocessing: Real-time access of traffic flow data, event and abnormal data, infrastructure state data and environmental data; Clean, time and space alignment and standardization of the data; Multi-dimensional feature extraction and calculation: Calculate congestion risk index and accident risk index of real-time risk dimension; Calculate pavement condition index PCI, structure safety index and facility integrity rate of infrastructure health dimension; Extract road function grade weight and key node weight of historical importance dimension; Dynamic classification: Based on real-time risk dimension index, infrastructure health dimension index and historical importance dimension index, through classification rule engine or lightweight classification model, dynamically assign comprehensive road condition level RCL to each evaluation unit.
[0008] Specifically, in S1: The traffic flow data includes Beidou data of floating car, geomagnetic, microwave vehicle detector data and ETC gantry data; The event and abnormal data includes traffic police accident, event platform data, traffic control information and social media or navigation App user reported data; The infrastructure state data includes historical and real-time patrol report, bridge or tunnel structure health monitoring sensor data and road property facility damage record; The environmental data includes meteorological bureau API data and road environment sensor data; The comprehensive road condition level includes: Level 1, trigger if any of the following conditions is met: High accident risk, i.e. historical accident heat map peak plus real-time emergency brake frequency greater than threshold, or structure safety alarm, i.e. bridge, tunnel monitoring data exceeds limit; Infrastructure state: PCI≤40 or visible damage area of road surface≥10%; Level2, triggered when any of the following conditions are met and Level1 is not reached: Real-time running risk: persistent congestion, i.e., vehicle speed < 40% of design value and lasts more than 30 minutes, or high accident risk, i.e., accident probability model output > 0.7; Infrastructure status: 40 < PCI ≤ 60 or facility damage density ≥ 3 per km; Level3, triggered when any of the following conditions are met and higher levels are not reached: Potential risk: periodic congestion, i.e., peak hour vehicle speed < 60% of design value, or 60 < PCI ≤ 80; Environmental warning: meteorological yellow warning affecting visibility or road adhesion coefficient; Level4, default level, requiring simultaneous satisfaction of: Traffic running: vehicle speed ≥ 80% of design value and no persistent congestion; Infrastructure: PCI > 80 and no new damage reports; Environmental conditions: no meteorological warning affecting driving; The comprehensive road condition level is dynamically updated at a preset period, and re-evaluation is triggered immediately when a major event or critical sensor data mutation occurs; Wherein, the major event includes traffic accident or disaster weather warning, and the critical sensor data includes bridge or tunnel structure health monitoring data and road surface environmental sensor data.
[0009] Specifically, the specific process of S2 is as follows: Establish a basic frequency mapping: Pre-set different road condition levels RCL corresponding to the basic target patrol frequency BIF: Level1: target frequency = 1 time per 2 hours, BIF = 0.5 times / hour; Level2: target frequency = 1 time per 4 hours, BIF = 0.25 times / hour; Level3: target frequency = 1 time per 8 hours, BIF = 0.125 times / hour; Level4: target frequency = 1 time per 24 hours, BIF ≈ 0.042 times / hour; Introduce a dynamic tuning factor: Risk change trend factor Tf: calculated based on the rate of change of real-time risk dimension index; Time interval factor Tl from last patrol time: calculate the interval from the last successful patrol to the current time t; Event impact factor Ef: assigned according to the type and severity of the event, no event = 0, event > 0; A dynamic priority weight DPW is calculated for each road segment at time ; : ; wherein: is the dynamic priority weight of road segment at time ; is the base target inspection frequency of road segment ; is the normalized value of risk change trend factor of road segment at time ; is the time since last inspection of road segment at time ; is the corresponding base period ; is the event impact factor of road segment at time ; , , are adjustable weight coefficients, controlling the importance of different factors; Output priority weight: output the , value of all road segments, the higher the value, the higher the inspection demand intensity.
[0010] Specifically, the specific process of the S3 is as follows: Problem modeling: Define resources: a set of available inspection vehicles, each vehicle containing attributes of current location, state, endurance and working time; Define tasks: a set of road segments to be inspected, each road segment is associated with a dynamic priority weight value and an estimated inspection time ; Set goals: Maximize total coverage priority: , =1 indicates that the road segment is planned to be covered, =0 indicates not covered; Minimize total travel time: generate a travel time matrix between road segments based on real-time traffic ; Meeting timeliness constraints: to achieve preset standards High priority road segments with values set time windows; Constraints: Vehicle endurance and working time limits; Path continuity; Each road segment is covered by at most one vehicle; Solution algorithm: Improved vehicle routing problem with time windows and priority weights (VRPTW-PW) algorithm is adopted: Time window setting: according to road segment priority weights Set the expected service time window for the road segment: For The highest road segment, requires coverage within the next Minutes; The second highest priority road segment requires coverage within the next Minutes, and so on; The objective function selects any of the following forms as needed: Form 1: Where Is a 0-1 variable, =1 indicates that the road segment Is planned to be covered, =0 indicates that it is not covered; Is a priority weight coefficient used to balance coverage priority and travel cost; travel time is the total time spent by all inspection vehicles in the task; Form 2: Where Is the sum of the priority weights of the road segments planned to be covered; Is the travel time budget, i.e., the maximum total travel time allowed.
[0011] Specifically, the solution algorithm in S3 further includes: Pheromone initialization and update rules are integrated Release more pheromones for Road segments greater than the threshold value; Heuristic information includes And the remaining time window size; Large-scale neighborhood search: Remove Or timeout window road segments below the threshold value in the destruction phase first; Insert Road segments greater than the threshold value and meeting the time window in the repair phase first; Real-time traffic integration: Calculate travel time At this time, real-time call map API road condition layer or traffic flow data-based travel time prediction; Output and feedback: Output each vehicle detailed route sequence: vehicle K: [start point-> segment A (priority weight ( , inspection time consumption )-> segment B (priority weight ( , inspection time consumption )->...-> end point / next start point]; Route instructions are issued to the vehicle terminal; Receive the actual position, task state and new collected data of the vehicle feedback, trigger the dynamic grading and frequency optimization module to update.
[0012] The highway traffic inspection system based on big data processing comprises: A multi-source data fusion module is used for real-time access and cleaning of Beidou, vehicle inspector, weather, accident, structure sensing multi-source heterogeneous data, and provides input for subsequent analysis; A dynamic road condition grading module is used for automatically dividing each kilometer road segment into four risk levels according to three dimensions of real-time risk, infrastructure health and historical importance, and locking the area needing attention; A patrol frequency optimization module is used for real-time calculation of the dynamic priority weight of each road segment based on the grading result and superposition of three factors of risk change, time from last patrol and sudden event, so that the adaptive patrol rhythm of high risk high frequency and low risk low frequency is realized; An intelligent path planning module is used for converting the priority weight into a VRPTW-PW model with a time window, calling an improved ant colony or large-scale neighborhood search algorithm, and generating an optimal inspection route under the constraints of vehicle endurance and real-time road condition; A closed-loop execution and feedback module is used for issuing route instructions to the vehicle terminal, real-time returning vehicle position, task completion and new collected data, and reversely refreshing the first four modules to form a rolling optimization closed loop of perception, decision, execution and re-perception.
[0013] Technical effects and advantages of the present application: The highway inspection is upgraded from the traditional mode of timing, fixed line and fixed person to the intelligent closed loop of changing with risk and moving with data: through real-time convergence and cleaning of multi-source heterogeneous data such as traffic flow, accident, structure and weather, 1-4 level risk grading of 1 kilometer road segment is automatically given every 5-15 minutes, and information island is completely eliminated; Based on the grading result, the patrol frequency is changed from fixed period to dynamic rhythm of high risk twice an hour and low risk once a day, and three factors of risk change, time from last patrol and sudden event are real-time weighted, so that resources are preferentially invested in the most needed area; Path planning introduces the VRPTW-PW algorithm with a time window to generate the optimal route with the shortest travel and highest coverage under the constraints of vehicle endurance, real-time road conditions and priority. It is instantly sent and transmitted back through the on-board terminal in real time to achieve rolling optimization of perception-decision-execution-reperception.
[0014] Ultimately, the inspection coverage rate and hidden danger discovery rate were significantly improved, the missed inspection rate and empty driving rate decreased simultaneously, maintenance funds and manpower investment were accurately allocated, road safety and traffic efficiency were improved simultaneously, providing a replicable and popularizable engineering paradigm for the sustainable operation of smart highways. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 Flow chart of the method of the present invention; Figure 2 It is a system flow chart of the present invention. DETAILED DESCRIPTION
[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.
[0017] like Figure 1 As shown, the highway traffic inspection method based on big data processing includes the following steps: Step 1: Multi-dimensional dynamic road condition classification: Clean and align real-time data from all roads, calculate indicators such as congestion, accidents, road surface, and structure, and comprehensively derive a risk level of 1-4 for each kilometer of road section; Multi-source heterogeneous data access and preprocessing: Traffic flow data: real-time access to BeiDou floating vehicle data (average speed, travel time, trajectory point density), geomagnetic / microwave vehicle inspection devices (flow rate, occupancy rate, vehicle speed), and ETC gantries (cross-sectional flow rate, vehicle type composition); event and anomaly data: access to traffic police accident / incident platforms (accident location, type, severity), traffic control information, and reports from social media / navigation app users (congestion, obstacles); infrastructure status data: access to historical and real-time inspection reports (pavement cracks, potholes, flatness index), bridge / tunnel structural health monitoring sensor data (vibration, strain), and damage records of road property facilities (guardrails, markings); environmental data: access to the Meteorological Bureau API (real-time and forecast rainfall, snow, fog, ice, strong winds), and road environment sensors (pavement temperature, humidity, visibility); Preprocessing: data cleaning (remove outliers, fill missing values), spatio-temporal alignment (unify timestamps, map geographic coordinates to specific road segments or grids), standardization / normalization.
[0018] Multi-dimensional feature extraction and calculation: Real-time risk dimension: congestion risk index: based on real-time speed, flow / capacity ratio, speed variation coefficient; accident risk index: based on historical accident heat map, real-time sudden braking / sudden lane change frequency (from Beidou), abnormal parking detection (from video or event data), visibility / slippery index; infrastructure health dimension: pavement condition index (PCI): based on historical damage data, recent inspection results, heavy load traffic (from ETC / weighting) to estimate the deterioration rate; structure safety index: based on bridge / tunnel monitoring data threshold alarm state, historical evaluation report; facility integrity rate: based on the frequency and density of road property facility damage reports; Historical importance dimension: Road function level: expressway, national highway, provincial highway, county road, etc. (preset weight); key node: interchange, long tunnel, large bridge, road section around traffic hub (preset weight).
[0019] Dynamic classification model: build classification rule engine or train lightweight classification model; Input: each dimension index (real-time risk, infrastructure health, historical importance) calculated in step two; Output: dynamically assign a comprehensive road condition level (RCL) to each evaluation unit (such as 1 km road section). For example: Level1 (extremely high risk / deterioration), trigger if any of the following conditions are met: Real-time safety risk: there is an extremely high accident risk (historical accident heat map peak + real-time sudden braking frequency > threshold) or structure safety alarm (bridge / tunnel monitoring data exceeds limit); infrastructure state: PCI ≤ 40 (severe damage) or visible damage area of road surface ≥ 10%.
[0020] Level2 (high risk / concern), trigger if any of the following conditions are met and not Level1: Real-time operation risk: persistent congestion (vehicle speed < design value 40% and lasts > 30 minutes) or high accident risk (accident probability model output > 0.7); infrastructure state: 40 < PCI ≤ 60 (moderate damage) or facility damage density ≥ 3 per km.
[0021] Level3 (medium risk / regular), trigger if any of the following conditions are met and not higher levels: Potential risk: Periodic congestion (peak hour speed < 60% of design value) or slight deterioration of PCI (60 < PCI ≤ 80); Environmental warning: Meteorological yellow warning (moderate rain / fog / snow) affecting visibility or road adhesion coefficient.
[0022] Level 4 (low risk / good), default level, requires simultaneous satisfaction of: Traffic operation: speed ≥ 80% of design value and no persistent congestion; Infrastructure: PCI > 80 (good) and no new damage reports; Environmental conditions: no weather warning affecting driving.
[0023] Dynamic: The evaluation period can be configured (e.g., 5-15 minutes), and immediate re-evaluation is triggered when significant events (accidents, disaster weather warnings) or critical sensor data mutations occur.
[0024] Step two, adaptive patrol frequency optimization: Given the basic patrol period according to the risk level, then dynamically weighted by risk change, time since last patrol, and sudden events to analyze the patrol priority of each road segment at this moment; Establish a basic frequency-level mapping: Preset different road condition levels (RCL) corresponding to basic target patrol frequencies (BIF): Level 1: Target frequency = 1 time per 2 hours (BIF = 0.5 times / hour); Level 2: Target frequency = 1 time per 4 hours (BIF = 0.25 times / hour); Level 3: Target frequency = 1 time per 8 hours (BIF = 0.125 times / hour); Level 4: Target frequency = 1 time per 24 hours (BIF ≈ 0.042 times / hour); Introduce dynamic optimization factors: Risk change trend factor (Tf): Calculate the rate of change (e.g., slope) of real-time risk dimension index in the recent period (e.g., 1 hour); if the risk rises sharply, increase the frequency weight, and if it tends to be stable or decreases, appropriately reduce it; Time since last patrol factor (Tl): Calculate the time interval since the last successful patrol of the road segment. The longer the interval, the closer it is to or exceeds the BIF required period, and the higher the current priority weight; Event impact factor (Ef): If a new event (accident, regulation) occurs on this road segment or adjacent road segments, significantly increase the frequency requirement (may temporarily exceed BIF) according to the event type and severity; Calculate the dynamic priority weight (DPW) of each road segment by weighted calculation: ; Where: For road segment At time Dynamic priority weight; For road segment Base target inspection frequency; For road segment At time Risk change trend factor (normalized value); For road segment At time Time since last inspection; For Corresponding base period (e.g. 2 hours for Level 1); For road segment At time Event impact factor (0 means no event, >0 means event, the larger the value, the greater the impact); , , Adjustable weight coefficient (a ) to control the importance of different factors; Output: All road segments , The higher the value, the more the road segment needs to be prioritized and covered more frequently in the next round of inspection planning; Directly reflects the dynamically adjusted inspection demand intensity.
[0025] Step three, optimal inspection path planning: convert priority to time window constraints, and use the improved VRPTW-PW algorithm to generate the shortest time and highest priority driving route for all inspection vehicles under real-time traffic and vehicle endurance restrictions; Problem modeling: Resources: a group of available inspection vehicles (with sensors), each vehicle has a current location, state (idle / busy), endurance / working time; Tasks: all road segments to be inspected. Each road segment has a value (representing its priority weight) and an estimated inspection time ( ) (estimated according to road segment length, historical average inspection speed); Objective: maximize total coverage priority: , =1 means the road segment is planned to be covered, =0 means not covered; Minimize total travel time: consider the travel time matrix between road segments ( ) caused by vehicle location and real-time traffic (from map API or traffic flow data); Satisfy timeliness constraints: high-priority (high DPW) sections should be covered as soon as possible (soft / hard time windows can be set); Constraints: Vehicle operating hours and range limits; each road section can be inspected by a vehicle at most once (or multiple times as needed, but with intervals); path continuity.
[0026] Solution algorithm: Adopt the improved vehicle routing problem with time windows and priority weights (VRPTW-PW) algorithm; Prioritize road segments This is converted into "urgency requirements" for high-priority road sections, which can be reflected through time window constraints or objective function weights; specifically: Time window setting: based on the road section priority weight , set the expected service time window for the road segment: right The highest section, demanding the future Coverage within minutes; Second highest priority road section, requiring the future Coverage within minutes (meeting ); Similarly, the time window width can be flexibly set to reflect coverage flexibility; Objective function: Two forms are available, choose as needed: Form 1: Optimize travel time while taking priority coverage into consideration: ; in: is a 0-1 variable, =1 indicates road section Covered by the plan, =0 means no coverage; is the priority weight coefficient, which is used to balance coverage priority and driving cost; driving time is the sum of the driving time of all inspection vehicles in the task; Form 2: With coverage priority as the optimization core, the travel time budget is constrained: ; in: is the sum of the priority weights of the planned covered road sections; The travel time budget is the maximum allowed total travel time. That is, priority is given to high Road segment coverage, while limiting the total driving time to not exceed the budget .
[0027] Solver selection; Small-scale scenarios (precise algorithm): use the column generation algorithm (ColumnGeneration); Large-scale / real-time scenario (heuristic / meta-heuristic): prefer improved Ant Colony Optimization (ACO) or Large Neighborhood Search (LNS), details as follows: ACO improvement points: pheromone initialization and update rules are integrated , to high road segments, the path releases more pheromone; heuristic information contains and time window constraints.
[0028] LNS process: Destruction phase: prefer to remove low or already timed-out window road segments; repair phase: prefer to insert high and time window satisfied road segments.
[0029] Real-time traffic integration: calculate the vehicle's travel time from the current location to the start of the candidate road segment : real-time call map API (such as traffic layer ETA function); or based on the accessed real-time traffic flow data, estimate the current / predicted travel time.
[0030] Output and execution: the algorithm output is to generate detailed route sequences for each inspection vehicle: vehicle K: [start -> road segment A (priority weight , inspection time ) -> road segment B (priority weight , inspection time ) ->...-> end / next start], including the estimated arrival / start / end time of each task; Route instructions are issued to vehicle terminals through the Internet of Vehicles / mobile network; Closed-loop feedback: during the execution of tasks, the vehicle transmits real-time position, task completion status, and newly collected traffic / infrastructure data. These data: are used to update the real-time risk dimension and infrastructure health dimension data in step one; are used to update (reset to 0) and possible in step two; trigger or participate in the next round of dynamic grading, frequency tuning, and path planning (rolling optimization).
[0031] Highway traffic inspection system based on big data processing, including multi-source data fusion module, dynamic traffic grading module, inspection frequency tuning module, intelligent path planning module, and closed-loop execution and feedback module; The multi-source data fusion module accesses and cleans Beidou, vehicle detector, weather, accident, structure sensor, and other multi-source heterogeneous data in real time, providing unified, clean, and spatiotemporally aligned input for subsequent analysis; The dynamic road condition grading module automatically divides each kilometer road segment into four risk levels according to three dimensions of real-time risk, infrastructure health and historical importance, and quickly locks the area that needs to be focused on; The inspection frequency optimization module calculates the dynamic priority weight of each road segment based on the grading results and three factors of risk change, time since last inspection and sudden events, to achieve adaptive inspection rhythm of high risk and high frequency, and low risk and low frequency. The intelligent path planning module converts the priority weight into a VRPTW-PW model with a time window, calls an improved ant colony or large-scale neighborhood search algorithm, and generates an optimal inspection route under the constraints of vehicle endurance and real-time road conditions. The closed-loop execution and feedback module sends route instructions to the vehicle terminal, and transmits vehicle position, task completion and newly collected data in real time, and reversely refreshes the first four modules to form a rolling optimization closed loop of perception, decision, execution and re-perception.
[0032] The above formulas are all dimensionless numerical calculations, and specific dimensionless methods such as standardization can be used, which will not be described here. The formula is obtained by simulating the most recent real situation with a large amount of data, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.
[0033] The above embodiments can be realized by software, hardware, firmware or any combination thereof. When realized by software, the above embodiments can be realized in the form of a computer program product in whole or in part. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer readable storage medium or transferred from one computer readable storage medium to another, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired or wireless (such as infrared, wireless, microwave, etc.) mode. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, an ATA hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid state ATA hard disk.
[0034] It should be understood that the size of the sequence number of each process described above does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0035] Those skilled in the art can appreciate that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.
[0036] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and in actual implementation, there can be another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.
[0037] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, which can be located in one place or distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.
[0038] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0039] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile ATA hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.
[0040] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.
Claims
1. A highway traffic inspection method based on big data processing, characterized in that, Comprising the following steps: S1, multi-dimensional dynamic road condition classification: clean and align real-time data of each road, calculate congestion, accident, road surface, structure indicators, and comprehensively obtain 1-4 level risk grades of each kilometer road section; S2, adaptive patrol frequency optimization: give a basic patrol cycle according to the risk level, dynamically weight three factors of risk change, time from last patrol and sudden events, and analyze the patrol priority of each road section at this moment; S3, optimal patrol path planning: convert the priority into a time window constraint, and use the improved VRPTW-PW algorithm to generate the shortest time and highest priority driving route for all patrol vehicles under real-time road conditions and vehicle endurance restrictions.
2. The method for highway traffic inspection based on big data processing according to claim 1, characterized in that, The specific process of S1 is as follows: Multi-source heterogeneous data access and preprocessing: Access real-time traffic flow data, event and abnormal data, infrastructure status data and environmental data; Clean, space-time align and standardize the data; Multi-dimensional feature extraction and calculation: Calculate real-time risk dimension congestion risk index and accident risk index; Calculate infrastructure health dimension road surface condition index PCI, structure safety index and facility integrity rate; Extract road function level weight and key node weight of historical importance dimension; Dynamic classification: Based on real-time risk dimension index, infrastructure health dimension index and historical importance dimension index, through a classification rule engine or a lightweight classification model, dynamically assign a comprehensive road condition level RCL to each evaluation unit.
3. The method as claimed in claim 2, wherein, In S1: The traffic flow data includes Beidou data of floating cars, geomagnetic, microwave vehicle detector data and ETC gantry data; The event and abnormal data include traffic police accidents, event platform data, traffic control information and social media or navigation App user reported data; The infrastructure status data includes historical and real-time patrol reports, bridge or tunnel structure health monitoring sensor data and road property facility damage records; The environmental data includes meteorological bureau API data and road environment sensor data; The comprehensive road condition level includes: Level 1, triggered when any of the following conditions is met: Very high accident risk, i.e. historical accident heat map peak plus real-time emergency brake frequency greater than threshold, or structure safety warning, i.e. bridge or tunnel monitoring data exceeds the limit; Infrastructure status: PCI≤40 or visible damage area of road surface≥10%; Level 2, triggered when any of the following conditions is met and not reached Level 1: Real-time operation risk: sustained congestion, i.e. vehicle speed less than 40% of design value and sustained for more than 30 minutes, or high accident risk, i.e. accident probability model output>0.7; Infrastructure status: 40<PCI≤60 or facility damage density≥3 per kilometer; Level 3, triggered when any of the following conditions is met and not reached a higher level: Potential risk: periodic congestion, i.e. peak vehicle speed less than 60% of design value, or 60<PCI≤80; Environmental warning: meteorological yellow warning and influence on visibility or road adhesion coefficient; Level 4, default level, requires simultaneous satisfaction of: Traffic operation: vehicle speed≥80% of design value and no sustained congestion; Infrastructure: PCI>80 and no new damage report; Environmental conditions: no weather warning affecting driving; The comprehensive road condition level is dynamically updated at a preset period, and reevaluation is triggered immediately when a major event or key sensor data mutation occurs; The major event includes a traffic accident or a disaster weather warning, and the key sensor data includes bridge or tunnel structure health monitoring data and road surface environment sensor data.
4. The method as claimed in claim 1, wherein, The specific process of S2 is as follows: Establish a basic frequency mapping: Pre-set the basic target inspection frequency BIF corresponding to different road condition levels RCL: Level 1: target frequency = 1 time every 2 hours, BIF = 0.5 times / hour; Level 2: target frequency = 1 time every 4 hours, BIF = 0.25 times / hour; Level 3: target frequency = 1 time every 8 hours, BIF = 0.125 times / hour; Level 4: target frequency = 1 time every 24 hours, BIF ≈ 0.042 times / hour; Introduce a dynamic tuning factor: Risk change trend factor Tf: calculated based on the change rate of real-time risk dimension index; Time interval factor Tl from the last inspection: calculated from the last successful inspection to the current time t; Event impact factor Ef: assigned according to the type and severity of the incident, no event = 0, event > 0; computing a dynamic priority weight DPW by weighted calculation of each road segment at a time of the dynamic priority weight : ; Wherein: for a road segment at a time dynamic priority weight; a basis target inspection frequency for a road segment of the road segment; for a road segment at a time normalized values of risk change trend factors; for the road segment at the time time since last inspection; for corresponding base period; for a road segment at a time an event influence factor; , , adjustable weight coefficients, , control the importance of different factors; Output priority weight: output all road segments' priority weight , The higher the value, the higher the inspection demand strength.
5. The method for highway traffic inspection based on big data processing according to claim 1, characterized in that, The specific process of S3 is as follows: Problem modeling: Define resources: available inspection vehicle set, each vehicle containing current location, state, endurance and working time attributes; Define task: set of road segments to be inspected, each road segment Associated dynamic priority weight Value and estimated inspection time consumption ; Set the target: Maximize total coverage priority: , = 1 if the road segment is planned to be covered, = 0 if not covered; Minimizing total travel time: generating inter-segment travel time matrix based on real-time traffic conditions ; Meeting timeliness constraints: to achieve preset standards High priority section of values set time window; Set the constraints: Vehicle endurance and working time limit; Path continuity; Each road segment is covered by at most one vehicle; Solution algorithm: Improved vehicle routing problem with time window and priority weight VRPTW-PW algorithm is adopted: Time window setting: based on link priority weight Set expected service time window for link: To The highest road segment, requiring future Covered within minutes; Next highest priority leg, required to be covered in the future minutes, and so on. The objective function is selected in the following form: Form 1: wherein is a 0-1 variable, = 1 indicates that the road segment is planned to be covered, = 0 indicates that it is not covered; is a priority weight coefficient for balancing the priority of coverage and the driving cost; Travel time is the total time spent by all inspection vehicles in the task; Form 2: wherein is the sum of the priority weights of the planned covered segments; is the travel time budget, i.e. the maximum total travel time allowed upper limit.
6. The method for highway traffic inspection based on big data processing according to claim 5, characterized in that, The solution algorithm in S3 also includes: Improved ant colony optimization: Pheromone initialization and update rules incorporated Pheromone release for segments greater than threshold Pheromone release for segments greater than threshold Heuristic information includes and a remaining time window size; Large-scale neighborhood search: The destruction phase preferentially removes below threshold or timeout window segments; Repair phase preferential insertion Segments greater than threshold and meeting time window Real-time road condition integration: Calculate travel time At this time, real-time call map API traffic layer or traffic flow data based on the prediction of travel time; Output and feedback: Output detailed route sequence for each vehicle: Vehicle K: [Start -> Link A (priority weight ( , inspection time spent ) -> Link B (priority weight ( , inspection time spent ) ->... -> End / Next Start]; Route instructions are issued to the vehicle terminal; Receive vehicle feedback of actual position, task status and new collected data, trigger dynamic grading and frequency tuning module update.
7. The system for application in the big data processing based highway traffic inspection method according to any one of claims 1-6, characterized in that, Including the following modules: Multi-source data fusion module for real-time access and cleaning of Beidou, vehicle detector, weather, accident, structure sensor multi-source heterogeneous data to provide input for subsequent analysis; Dynamic road condition grading module for automatically dividing each kilometer road segment into four risk levels according to real-time risk, infrastructure health and historical importance, and locking the area that needs to be paid attention to; Inspection frequency tuning module, based on the grading results and superimposed risk change, time interval from last inspection and three major factors of sudden event, real-time calculation of dynamic priority weight of each road segment, realizing adaptive inspection rhythm of high risk high frequency and low risk low frequency; Intelligent path planning module for converting priority weight to VRPTW-PW model with time window, calling improved ant colony or large-scale neighborhood search algorithm to generate optimal inspection route under vehicle endurance and real-time road condition constraints; A closed-loop execution and feedback module is configured to issue route instructions to the vehicle terminal, and to transmit the vehicle position, task completion and newly collected data in real time, so as to refresh the first four modules reversely, and form a rolling optimization closed loop of perception, decision, execution and re-perception.
Citation Information
Patent Citations
Urban road intelligent inspection method and system based on automatic driving vehicle
CN119151510A
Machine vision-assisted intelligent road inspection method and system
CN119479308A
Unmanned aerial vehicle patrol path optimization method and system based on mobile unmanned aerial vehicle nest
CN120335497A
Time window selection for vehicle routing problem
US20180129985A1
Route generation for vehicle routing problem and variants thereof
US20240183670A1
Cited By
Power system inspection method based on machine learning
CN121213058A
Method and system for planning inspection path of ultra-high voltage transformer substation
CN121503842A
