A traffic light data acquisition control system and method for intelligent networked vehicles

Abstract

The application discloses a traffic light data acquisition control system and method for intelligent networked vehicles, and belongs to the technical field of intelligent networked vehicles, and comprises the following steps: a traffic police cloud platform acquires traffic light data from a traffic light signal machine in real time and stores the traffic light data; a cloud control platform acquires a traffic light timing scheme and a small amount of low-frequency real-time traffic light data from the traffic police cloud platform, verifies the timing scheme through the small amount of low-frequency real-time traffic light data, and when the timing scheme is verified to be matched, sends the timing scheme to a vehicle; when the timing scheme is verified to be not matched, requests a large amount of high-frequency real-time traffic light data from the traffic police cloud platform, calibrates the timing scheme, and obtains a calibrated timing scheme; and the calibrated timing scheme is sent to the vehicle, and the vehicle acquires the calibrated timing scheme. The application improves data accuracy, reduces decision-making risks, enhances scheme adaptability, and covers complex scenes.

Description

Technical Field

[0001] This invention relates to the field of intelligent connected vehicle technology, specifically to a traffic light data acquisition and control system and method for intelligent connected vehicles. Background Technology

[0002] Traffic light data is crucial for intelligent connected vehicles, allowing them to know the status and countdown of traffic lights in advance, accurately plan their speed to achieve green wave passage, reduce stops and starts, and improve traffic efficiency. Simultaneously, it can provide early warnings of red lights, preventing red-light running and ensuring intersection safety, making it one of the core data sources for vehicle-to-infrastructure (V2I) communication. Currently, intelligent connected vehicles typically acquire traffic light data in three ways: 1. Through onboard cameras, radar, and other sensing devices, vehicles autonomously identify and analyze the status and countdown of traffic lights at intersections in real time. 2. Relying on vehicle-to-cloud communication (4G / 5G) to receive traffic light data pushed by the traffic police cloud platform in real time. 3. Gaode AI traffic light calculation method: using large AI models (such as the Visual Spatiotemporal Sequence Model VSTM), it comprehensively calculates and simulates massive amounts of historical traffic data, real-time traffic trajectories, and intersection signal patterns to deduce the status and countdown of traffic lights.

[0003] It should be noted that Gaode's AI traffic light calculations are highly dependent on the density and regularity of floating car trajectory data. When traffic flow is sparse (data is sparsity) or when traffic lights are subject to irregular temporary manual intervention (such as on-site traffic police takeover or sudden emergency control), the prediction accuracy will drop significantly or even fail to provide effective predictions due to the lack of effective trajectory samples or the disruption of historical patterns.

[0004] While autonomous vehicle recognition methods have low dependency, they place high demands on the performance of onboard chips and are subject to a certain probability of recognition errors. Receiving traffic light data in real-time from the traffic police cloud platform provides relatively accurate data, but the large volume of data necessitates real-time reception and processing, posing challenges to the performance of both the vehicle-side and vehicle-cloud links. Therefore, this paper proposes a traffic light data acquisition and control system for intelligent connected vehicles. Summary of the Invention

[0005] The technical problem to be solved by this invention is: to address the problems of high vehicle-cloud link load and poor real-time performance and low accuracy of traffic light data at adaptive signal control (dynamic timing) intersections in the existing technology, a traffic light data acquisition and control system for intelligent connected vehicles is provided.

[0006] The present invention solves the above-mentioned technical problems through the following technical solutions, the present invention comprising:

[0007] The traffic signal data acquisition module is used to acquire and store traffic light data from the traffic light signal controller in real time through the traffic police cloud platform.

[0008] The verification and calibration module is used to obtain the traffic light timing scheme and a small amount of low-frequency traffic light real-time data from the traffic police cloud platform through the cloud control platform. It uses the small amount of low-frequency traffic light real-time data to verify the timing scheme. When the timing scheme is verified and matches, the timing scheme is sent to the vehicle. When the timing scheme is not verified and matches, it requests a large amount of high-frequency real-time traffic light data from the traffic police cloud platform to calibrate the timing scheme and obtain the calibrated timing scheme.

[0009] The timing scheme acquisition module is used to send the calibrated timing scheme to the vehicle through the cloud control platform, and the vehicle obtains the calibrated timing scheme.

[0010] Furthermore, in the signal data acquisition module, a small amount of low-frequency real-time traffic light data is low-frequency sampling data with a sampling frequency of f1. Each sample data includes a unique intersection identifier, sampling timestamp, current execution phase, real-time status of the light group, data source identifier, and digital signature. The sampling frequency f1 is set from 30 seconds / time to 300 seconds / time.

[0011] Furthermore, in the verification and calibration module, the specific process for verifying the timing scheme is as follows:

[0012] Step S211: Match the traffic light timing scheme of the same intersection with a small amount of low-frequency real-time traffic light data according to the unique intersection identifier, and filter out the sampled data that exceeds the effective period of the traffic light timing scheme;

[0013] Step S212: Calculate the theoretical predicted state of the traffic light group at the sampling time based on the traffic light timing scheme;

[0014] Step S213: Compare the predicted state of the theoretical light group with the actual light group state in a small amount of real-time low-frequency traffic light data to obtain the sampling point matching result;

[0015] Step S214: Determine whether the traffic light timing scheme is effective based on the matching results of continuous sampling points.

[0016] Furthermore, in step S212, calculating the theoretical predicted state of the lamp group at the sampling time includes calculating the cycle offset, and the formula for calculating the cycle offset is as follows:

[0017] cycle_offset=elapsed_time mod cycle;

[0018] Where elapsed_time is the difference between the sampling point time and the start time of the traffic light timing scheme, cycle is the signal cycle duration of the traffic light timing scheme, and mod represents the modulo operation.

[0019] Furthermore, in step S214, the specific determination process is as follows:

[0020] The matching results of three consecutive sampling points are statistically analyzed. If the number of matching sampling points is ≥2, the traffic light timing scheme is determined to be valid; if the number of matching sampling points is ≤1, the traffic light timing scheme is determined to be invalid.

[0021] Furthermore, in the verification and calibration module, when requesting real-time high-frequency traffic light data from the traffic police cloud platform, the request parameters include: the unique identifier of the target intersection, the request identifier, the sampling frequency f2, and the sampling duration. The sampling data field includes the sampling timestamp, the current execution phase, the current color of each light group, and the remaining time of the current color of each light group. The sampling frequency f2 can be set from 1 second / time to 10 seconds / time, and the sampling duration is specified. The setting range is 60 seconds to 600 seconds.

[0022] Furthermore, in the verification and calibration module, the specific process for calibrating the timing scheme is as follows:

[0023] Step S221: Analyze the real-time data of high-frequency traffic lights and extract the timestamps of the state switching of each light group;

[0024] Step S222: Calculate the actual timing parameters based on the timestamp of the state transition;

[0025] Step S223: Correct the original traffic light timing scheme based on the actual timing parameters and generate the calibrated timing scheme.

[0026] Furthermore, in the verification and calibration module, after generating the calibrated timing scheme, the following verification steps are also included:

[0027] Recalibrate using three sampling points within 10 minutes after calibration. If the matching rate is ≥90%, the calibration is confirmed to be effective. If there is no matching, repeat the calibration steps and extend the sampling time of the high-frequency traffic light real-time data to 10 minutes.

[0028] Furthermore, in the verification and calibration module, the timing scheme includes a unique intersection identifier, scheme ID, signal cycle duration, effective period, phase list, and scheme issuance timestamp; the phase list includes phase ID, traffic direction, and light group status parameters in execution order; the calibrated timing scheme adds adjustment markers and calibration basis fields, and the calibrated timing scheme takes effect immediately after high-frequency sampling ends.

[0029] This invention also provides a method for acquiring and controlling traffic light data for intelligent connected vehicles, based on the above-described system implementation, comprising the following steps:

[0030] Step S1: The traffic police cloud platform obtains and stores traffic light data from the traffic light controller in real time;

[0031] Step S2: The cloud control platform obtains the traffic light timing scheme and a small amount of low-frequency real-time traffic light data from the traffic police cloud platform. It uses the small amount of low-frequency real-time traffic light data to verify the timing scheme. When the timing scheme is verified to match, it sends the timing scheme to the vehicle. When the timing scheme is not verified to match, it requests a large amount of high-frequency real-time traffic light data from the traffic police cloud platform to calibrate the timing scheme and obtain the calibrated timing scheme.

[0032] Step S3: The cloud control platform sends the calibrated timing plan to the vehicle, and the vehicle obtains the calibrated timing plan.

[0033] The present invention has the following advantages over the prior art:

[0034] 1. Improve data accuracy and reduce decision-making risks.

[0035] The "timing scheme verification + dynamic calibration" mechanism solves the problem of data disconnect between traffic police cloud platform and actual intersection status in traditional solutions:

[0036] When the timing scheme is adjusted due to unforeseen circumstances (such as temporary traffic control), the cloud control platform can quickly identify the deviation through sampled data (such as the actual green light duration being shortened by 5 seconds) and generate a calibrated scheme to ensure that the traffic light data received by the vehicle is consistent with the actual state of the intersection (error ≤ 2 seconds).

[0037] To avoid vehicles making decision-making errors due to relying on outdated timing schemes (such as driving at the "green wave speed" planned in the original scheme but encountering a traffic light that turns red early), and to reduce safety risks such as running red lights and sudden braking, tests have shown that it can reduce the accident rate at intersections by more than 30%.

[0038] 2. Enhance the adaptability of the solution to cover complex scenarios.

[0039] Traditional solutions rely on real-time data from the traffic police cloud platform, which is prone to failure when traffic lights are not connected to the network, data transmission is interrupted, or timing is temporarily adjusted. The improved solution of this invention adopts a dual-track system of "timing scheme + real-time sampling". When the real-time data of the traffic police cloud platform is interrupted (such as network failure), the cloud control platform can predict the traffic light status based on the calibrated timing scheme (error ≤ 5 seconds), ensuring that vehicles can still make normal decisions in weak network environments (such as planning the speed of vehicles passing through intersections).

[0040] It supports handling special scenarios (such as temporarily extending green lights during morning and evening rush hours, and manually adjusting timings during heavy rain), and quickly adapts to changes through high-frequency sampling to ensure that the traffic light data acquired by vehicles is always "available and reliable".

[0041] 3. Optimize traffic efficiency and support the deep application of vehicle-road cooperative systems.

[0042] The calibrated timing scheme provides intelligent connected vehicles with more accurate decision-making basis, directly improving traffic efficiency:

[0043] Vehicles can calculate the optimal speed (e.g., 50 km / h) based on precise timing schemes (e.g., 15 seconds remaining on the green light + 300 meters from the intersection), achieving a "green wave" traffic efficiency improvement of more than 20% and reducing intersection waiting time.

[0044] It provides basic data support for advanced vehicle-road cooperative functions (such as collaborative obstacle avoidance at intersections and priority passage for emergency vehicles). For example, when an ambulance passes by, the cloud control platform can quickly calibrate the timing plan and push it to surrounding vehicles to guide them to avoid it in advance.

[0045] 4. Reduce dependence on a single data source and improve system robustness.

[0046] Traditional solutions rely entirely on real-time data from the traffic police cloud platform. If the platform interface malfunctions or access is restricted, vehicles will lose access to traffic light data. The improved method of this invention achieves this through:

[0047] Using the timing scheme as the "benchmark framework" and real-time sampling data as the "calibration basis", a closed loop of "scheme prediction + actual correction" is formed, reducing the strong dependence on real-time data from the traffic police cloud platform (data interruption within 30 seconds can be tolerated).

[0048] When automakers or regulatory authorities need to access traffic light data, the cloud control platform can provide reliable data with calibration evidence (such as data labeled "verified based on 300 sampled data points"), improving the reliability of cross-entity data sharing.

[0049] 5. The deterministic logic of "transferring benchmarks + correcting deviations," rather than pure AI probabilistic prediction, results in high accuracy and reliability.

[0050] This solution uses the official timing scheme of the traffic police cloud platform as a benchmark for independent calculation. It is essentially a deterministic logic of "transmitting the benchmark + correcting deviations" rather than a purely AI probabilistic prediction. Under normal road conditions, its countdown error is determined only by transmission delay, and its accuracy and reliability are far superior to third-party AI models that rely on historical trajectory calculations (such as the Gaode AI traffic light calculation model). Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating the traffic light data acquisition and control method for intelligent connected vehicles in an embodiment of the present invention.

[0052] Figure 2 This is a schematic diagram of the timing scheme verification and calibration process in an embodiment of the present invention. Detailed Implementation

[0053] The embodiments of the present invention are described in detail below. These embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments.

[0054] This embodiment provides a technical solution: a traffic light data acquisition and control system for intelligent connected vehicles, comprising the following modules:

[0055] The traffic signal data acquisition module is used to acquire traffic light data from the traffic signal controller in real time using the traffic police cloud platform;

[0056] The verification and calibration module is used to obtain the traffic light timing scheme and a small amount of real-time traffic light data from the traffic police cloud platform using the cloud control platform. The timing scheme is verified using a small amount of real-time traffic light data. When the timing scheme is verified and matches, the timing scheme is sent to the vehicle. When the timing scheme is not verified and matches, a large amount of real-time traffic light data is requested from the traffic police cloud platform to calibrate the timing scheme and obtain the calibrated timing scheme.

[0057] The timing scheme acquisition module is used to push the adjusted traffic light timing scheme to the intelligent connected vehicle using the cloud control platform, so that the intelligent connected vehicle can obtain the accurate timing scheme.

[0058] like Figure 1 As shown, this embodiment also provides a method for acquiring and controlling traffic light data for intelligent connected vehicles. Based on the above system implementation, it includes the following steps:

[0059] Step 1: The process of the traffic police cloud platform acquiring data from the traffic light controllers remains unchanged. It still acquires traffic light data from the controllers in real time. At this stage, the traffic police cloud platform continues to serve as a data storage platform.

[0060] Step 2: The cloud control platform obtains the traffic light timing scheme and a small amount of real-time traffic light data from the traffic police cloud platform. It verifies the timing scheme using the small amount of real-time traffic light data. When the timing scheme matches, it sends the timing scheme to the vehicle. When the timing scheme does not match, it requests a large amount of real-time traffic light data from the traffic police cloud platform to calibrate the timing scheme and obtain the calibrated timing scheme.

[0061] It should be noted that this invention adopts a hierarchical sampling strategy of "first acquiring a small amount of low-frequency data, and then acquiring a large amount of high-frequency data as needed." The logical necessity of this strategy lies in minimizing the concurrent load and network bandwidth consumption of the traffic police cloud platform while ensuring data accuracy. Without this hierarchical mechanism, if the system continuously polls all intersection data at a high frequency (e.g., once per second), it will generate a massive amount of invalid data streams; while relying solely on low-frequency data will fail to capture the instantaneous changes in traffic signals. Therefore, using low-frequency data as a "routine monitoring sentinel," and triggering a high-frequency "deep scan" only when an anomaly is detected, is the optimal solution for achieving efficient vehicle-road cooperation. Using this approach significantly reduces system resource consumption: in scenarios where most intersections are operating normally (i.e., timing scheme verification and matching), the cloud control platform only needs to maintain a very low frequency (e.g., once per minute) of data interaction. Compared to high-frequency transmission (e.g., once per second) throughout the day, this solution can reduce data transmission volume and platform computing pressure by approximately 98%, avoiding excessive consumption of the traffic police cloud platform's computing power. It can accurately capture and quickly respond to abnormal states: when low-frequency sampling detects a mismatch between the "theoretical state" and the "actual state" (e.g., two out of three consecutive sampling points are mismatched), the system will immediately trigger a high-frequency data request (e.g., once per second, lasting 300 seconds). This "on-demand" mechanism can accurately capture phase changes caused by temporary signal takeover, manual intervention, or equipment failure, and can quickly calculate the corrected timing parameters through short-term high-frequency sampling, ensuring that the data obtained by the vehicle is always synchronized with the actual state of the intersection. It can also optimize communication link efficiency: this strategy effectively distinguishes between "steady-state" and "transient" data requirements. It maintains low bandwidth usage in steady-state conditions and guarantees data accuracy in transient (abnormal) conditions, solving the technical contradiction of "having to operate at high power consumption all day long to capture occasional anomalies" in traditional solutions.

[0062] In this embodiment, the input data format is defined as follows:

[0063] Traffic light timing scheme (from the traffic police cloud platform) (In this embodiment, the traffic light timing scheme is a traffic signal control strategy formulated for the intersection marked INT001_X012), detailed as follows:

[0064] Scheme Identifier: The scheme ID is PLAN_20241001_0800, indicating that the scheme was formulated on October 1, 2024, and is effective from 08:00 to 18:00 daily.

[0065] Signal period: The total duration of each complete signal cycle (i.e., one cycle) is 120 seconds.

[0066] Phase settings: The scheme includes two main passage phases, which are executed in sequence:

[0067] East-west phase (P1):

[0068] The green light lasts for 30 seconds, allowing vehicles traveling in both east and west directions to proceed.

[0069] The light then turns yellow for 3 seconds, indicating that a change is imminent;

[0070] The light remains red for 87 seconds (meaning that while north-south traffic is permitted, east-west traffic remains red).

[0071] North-South Phase (P2):

[0072] The green light lasts for 25 seconds, allowing vehicles traveling in both north and south directions to proceed.

[0073] The yellow light transitions for 3 seconds;

[0074] The light remains red for 92 seconds (meaning that while east-west traffic is permitted, north-south traffic remains red).

[0075] Light group control: Each direction is controlled by an independent light group (L1 corresponds to east-west, L2 corresponds to north-south), and the status and duration of each light color are clearly defined to ensure mutual exclusion between phases and safe switching.

[0076] Release time: The plan was released to the signal control system at timestamp 1696108800000 (corresponding to 2023-10-01 00:00:00 UTC).

[0077] The following is an example of a data organization format:

[0078] {

[0079] "intersection_id": "INT001_X012", / / Unique identifier for the intersection (format: area code_intersection number)

[0080] "plan_id": "PLAN_20241001_0800", / / Timing scheme ID (date_effective time)

[0081] "cycle": 120, / / Cycle duration (seconds)

[0082] "effective_time": { / / The effective period of the plan

[0083] "start": "08:00:00",

[0084] "end": "18:00:00"

[0085] },

[0086] "phases": [ / / List of phases (arranged in execution order)

[0087] {

[0088] "phase_id": "P1", / / Phase ID

[0089] "direction": "east-west", / / Allows passage in the east-west direction.

[0090] "lights": [ / / Light group status (red / yellow / green)]

[0091] {"light_id": "L1" , "color": "green" , "duration": 30}, / / Green light for 30 seconds

[0092] {"light_id": "L1" , "color": "yellow" , "duration": 3}, / / Yellow light for 3 seconds

[0093] {"light_id": "L1" , "color": "red" , "duration": 87} / / Red light for 87 seconds (remaining time within the cycle) ]

[0095] },

[0096] {

[0097] "phase_id": "P2",

[0098] "direction": "north-south", / / North-South direction

[0099] "lights": [

[0100] {"light_id": "L2" , "color": "green" , "duration": 25},

[0101] {"light_id": "L2" , "color": "yellow" , "duration": 3},

[0102] {"light_id": "L2" , "color": "red" , "duration": 92} ]

[0104] }

[0105] ],

[0106] "timestamp": 1696108800000 / / Scheme issuance timestamp (milliseconds)

[0107] }

[0108] The sampled real-time traffic light data (from the traffic police cloud platform, sampled every 1 minute) is a snapshot of the intersection signal status collected from the traffic police cloud platform every minute, reflecting the real-time status of the traffic lights at the specified intersection in actual operation. Specific details are as follows:

[0109] Associated intersections: The unique identifier of the intersection corresponding to the data is INT001_X012, which is consistent with the aforementioned timing scheme.

[0110] Sampling frequency and timing: The system samples at a fixed frequency of 1 minute / time, and each record contains a timestamp accurate to milliseconds (e.g., 1696109100000 corresponds to 2023-10-01 08:05:00 UTC).

[0111] Data content structure: Each sampling record includes:

[0112] The current execution phase (such as "P1" or "P2") indicates which traffic phase the intersection is currently in.

[0113] Real-time status of each light group: List the current light color (red / yellow / green) and remaining duration (seconds) for each light group (L1, L2), such as 15 seconds left for green light and 18 seconds left for red light, to show the signal countdown information.

[0114] Data integrity assurance: Includes data source identification and digital signature ("signature":"xxx") from traffic_police_cloud (traffic police cloud platform) to ensure data authenticity, traceability and tamper-proof.

[0115] The following is an example of a data organization format:

[0116] {

[0117] "intersection_id": "INT001_X012", / / Associated intersection

[0118] "sample_data": [

[0119] {

[0120] "timestamp": 1696109100000 , / / Sampling time (milliseconds, corresponding to 08:05:00)

[0121] "phase": "P1", / / Current execution phase

[0122] "lights": [

[0123] {"light_id": "L1" , "color": "green" , "remaining_time": 15}, / / 15 seconds remaining on the green light

[0124] {"light_id": "L2" , "color": "red" , "remaining_time": 18} ]

[0126] },

[0127] {

[0128] "timestamp": 1696109400000 , / / Sampled at 08:10:00

[0129] "phase": "P2",

[0130] "lights": [

[0131] {"light_id": "L1" , "color": "red" , "remaining_time": 20},

[0132] {"light_id": "L2" , "color": "green" , "remaining_time": 10} ]

[0134] }

[0135] ],

[0136] "data_source": "traffic_police_cloud", / / Data source

[0137] "signature": "xxx" / / Digital signature for traffic police cloud platform (tamper-proof)

[0138] }

[0139] It should be noted that, more preferably, the sampling frequency of the "small amount of low-frequency real-time traffic light data" is not a fixed value, but a parameter that can be dynamically adjusted according to urban traffic characteristics, signal cycle length, and network environment. Typically, the sampling frequency is set between 30 seconds / time and 300 seconds / time (i.e., 5 minutes / time). In special scenarios, this range can be adaptively adjusted according to the needs of the scenario.

[0140] Lower limit setting (30 seconds): This is mainly applicable to intersections with extremely short signal cycles (such as less than 60 seconds), or during peak hours when traffic is congested and signal controllers are frequently adjusted, to ensure that rapidly changing phases can be captured.

[0141] Upper limit setting (300 seconds): This is mainly applicable to suburban intersections during off-peak hours or when signal timing is extremely stable and the cycle is long (e.g., more than 180 seconds), in order to minimize the data transmission load.

[0142] The reason for setting it to "1 minute / time" in this embodiment is:

[0143] This embodiment selects 1 minute / sample as the default sampling frequency, which is the result of balancing the following three technical considerations and measured data:

[0144] The "least common multiple" covering the vast majority of signal periods:

[0145] Statistics show that the traffic light cycle on urban main roads and secondary roads is typically between 60 and 150 seconds. Setting the sampling frequency to 60 seconds (1 minute) ensures that at any given time, at least one key phase switching point within a complete cycle (such as the moment when the green light turns red or the red light turns green) can be captured. This effectively verifies the phase sequence and approximate duration of the timing scheme and avoids data blind spots caused by insufficient sampling.

[0146] The "golden balance" between communication load and data validity:

[0147] If the sampling frequency is too high (e.g., 10 seconds / time), although the data is more dense, it will lead to an exponential increase in data transmission volume, putting unnecessary bandwidth pressure on the traffic police cloud platform and the vehicle-to-cloud link. If the sampling frequency is too low (e.g., 5 minutes / time), when there are minor fluctuations in the traffic signal (e.g., a 10-second green light delay), the system may need a longer time to detect the deviation. A frequency of 1 minute / time ensures effective capture of anomalies (e.g., phase inversion, abrupt period changes) while keeping the data transmission volume at an extremely low level (only 1 / 60th of high-frequency sampling), achieving the best balance between system efficiency and monitoring accuracy.

[0148] Matching the vehicle's computational and refresh requirements:

[0149] When intelligent connected vehicles perform speed guidance (GLOSA) or red light violation warning (RSI), they typically do not require millisecond-level real-time countdowns (that is a task for the high-frequency calibration phase). A 1-minute data refresh rate is sufficient to support the vehicle in making macro-level driving behavior decisions (such as "all three intersections ahead have green lights, it is recommended to maintain the current speed"), which is in line with the data interaction characteristics of vehicle-to-infrastructure (V2I) systems in non-emergency situations.

[0150] In this embodiment, the timing scheme verification process in step two is as follows:

[0151] Objective: To verify whether there are any deviations in the execution of the timing scheme (such as the actual green light duration not matching the scheme) by sampling real-time data (a small amount of real-time traffic light data, which is low-frequency sampling data).

[0152] Step 1: Data Association and Alignment

[0153] Match the "timing scheme" and "sampling data" for the same intersection by intersection_id;

[0154] Filter out sampling data that are outside the effective period of the timing scheme (e.g., if the scheme is effective from 8:00 to 18:00, remove the sampling point at 7:50).

[0155] Step 2: Predict the lamp group status at the sampling time based on the timing scheme.

[0156] Using the sampling point as a reference, calculate the "theoretical phase and state" of that moment in the timing scheme:

[0157] Example: The timing scheme has a period of 120 seconds. The time of a certain sampling point is 1696109100000 (08:05:00), and the scheme takes effect from 1696108800000 (08:00:00). Then:

[0158] elapsed_time = 1696109100000 - 1696108800000 = 300 seconds (5 minutes)

[0159] cycle_offset = 300%120 = 60 seconds (offset within the current cycle)

[0160] Iterate through the phases (current execution phase) of the timing scheme and calculate the time interval for each phase:

[0161] Example: Total duration of phase P1 = 30 (green) + 3 (yellow) + 87 (red) = 120 seconds, where:

[0162] Green light period: 0-30 seconds;

[0163] Yellow light duration: 30-33 seconds;

[0164] Red light period: 33-120 seconds;

[0165] Based on cycle_offset=60 seconds, determine the theoretical state (i.e., the "predicted state" in step 3 below): During the red light period of phase P1 (33-120 seconds), the L1 light group should be red.

[0166] Step 3: Compare the predicted state with the actual state of the sampled data.

[0167] Compare the "predicted status" field with the "lights" (real-time status of the light group) field of the sampled data field:

[0168] If the light groups have the same color and the remaining time deviation is ≤1 second (allowing for signal error), then the sampling point is "matched".

[0169] If the colors of the light groups are inconsistent, or the remaining time deviation is greater than 1 second, then the sampling point is "mismatched".

[0170] Step 4: Determine if the timing plan is effective

[0171] Statistical analysis of matching results for three consecutive sampling points:

[0172] If there are ≥2 matches, the timing scheme is deemed "valid" and no adjustment is needed.

[0173] If there are ≤1 matches, the timing scheme is deemed "invalid", triggering the subsequent timing scheme adjustment process.

[0174] In this embodiment, the relevant content of the timing scheme adjustment in step two is as follows:

[0175] Added input data format (Request more real-time data)

[0176] When the timing scheme does not match, the cloud control platform requests "high-frequency sampling data" from the traffic police cloud platform, in the following format:

[0177] Request parameters: Configuration information used to initiate a real-time traffic light status sampling request to the traffic signal data service system.

[0178] Target intersection: The unique identifier of the intersection for which data needs to be collected is INT001_X012, which is consistent with the aforementioned timing scheme and real-time data.

[0179] Request Identifier: This data collection task is identified by the unique request ID REQ_20241001_0815, which facilitates tracking and association with logs.

[0180] Sampling frequency: set to 1 second / time. This request aims to obtain high-frequency, fine-grained signal state change data.

[0181] It should be noted that the sampling frequency setting range of the present invention is from 1 second / time to 10 seconds / time. In special scenarios, this range can be adaptively adjusted according to the needs of the scenario.

[0182] Sampling duration: Continuous sampling for 300 seconds (5 minutes), covering approximately 2 to 3 complete signal cycles (calculated based on the original signal cycle of 120 seconds), which is sufficient to observe the dynamic process of multiple phase cycles. It should be noted that the sampling duration setting range of this invention is 60 seconds to 600 seconds. In special scenarios, this range can be adaptively adjusted according to the needs of the scenario.

[0183] Required fields: Clearly define the key fields that the returned data must contain, including:

[0184] timestamp (sampling timestamp);

[0185] phase (current execution phase);

[0186] lights.color (current color of each light group);

[0187] lights.remaining_time (remaining time for the current light color of each light group).

[0188] The following is an example of a data organization format:

[0189] {

[0190] "intersection_id": "INT001_X012" ,

[0191] "request_id": "REQ_20241001_0815", / / Request ID

[0192] "sample_frequency": 1, / / Sampling frequency (seconds / sample, the original scheme was 300 seconds / sample)

[0193] "duration": 300, / / Sampling duration (5 minutes, covering 2-3 cycles)

[0194] "required_fields": ["timestamp", "phase", "lights.color", "lights.remaining_time"]

[0195] }

[0196] The collection method, process, and content of high-frequency sampling data (refined traffic light status records returned by the traffic police cloud platform based on the aforementioned request identifier (request ID: REQ_20241001_0815)) are summarized as follows:

[0197] Data acquisition method: The signal status of the designated intersection (INT001_X012) is continuously sampled at an extremely high frequency of 1 second / time, which is much higher than the conventional monitoring frequency (such as 5 minutes / time), so as to achieve near real-time capture of the traffic light change process.

[0198] Sampling process: Within the requested 300 seconds (5 minutes) duration, the system continuously generates and records a total of 300 time series data points. Each data point precisely corresponds to a whole second, fully covering approximately 2–3 signal cycles, and can clearly reflect dynamic details such as phase switching and light countdown.

[0199] Data content: Each record contains:

[0200] timestamp: A timestamp accurate to milliseconds (e.g., 1696109700000 corresponds to 08:15:00UTC).

[0201] phase: The currently executing phase (e.g., "P1" indicates the east-west travel phase);

[0202] lights: Current status information of the light group, including the light color (e.g., "red") and the remaining duration of that color (in seconds). (Note: Only one light group is shown in the example. In actual applications, there may be multiple light groups, such as L1 / L2. This is a simplified representation.)

[0203] The following is an example of a data organization format:

[0204] {

[0205] "intersection_id": "INT001_X012" ,

[0206] "request_id": "REQ_20241001_0815" ,

[0207] "sample_data": [ / / 1 record per second, 300 records in total]

[0208] {"timestamp": 1696109700000 , "phase": "P1" , "lights": [{"color": "red", "remaining_time": 5}]},

[0209] {"timestamp": 1696109701000 , "phase": "P1" , "lights": [{"color": "red", "remaining_time": 4}]},

[0210] / / ... 298 entries omitted ]

[0212] }

[0213] The specific adjustment steps are as follows:

[0214] Objective: To calculate the actual timing pattern based on high-frequency sampling data and correct the original timing scheme.

[0215] Step 1: Analyze the high-frequency sampling data and extract key time points.

[0216] Identify the timestamp of each light group's state transition (e.g., the time when L1 changes from red to green):

[0217] Example: Extracting the state switching records of L1 lamp group from high-frequency sampling data:

[0218] {

[0219] "L1": [

[0220] {"from_color": "red", "to_color": "green", "timestamp":1696109705000}, / / Red light turns green light

[0221] {"from_color": "green", "to_color": "yellow", "timestamp":1696109735000}, / / Green light turns yellow (lasts 30 seconds)

[0222] {"from_color": "yellow", "to_color": "red", "timestamp":1696109738000} / / Yellow light turns red (lasts 3 seconds) ]

[0224] }

[0225] Step 2: Calculate the actual timing parameters.

[0226] Calculate the duration of a single cycle: the time difference between two adjacent transitions to the same state (e.g., the interval between two red light to green light transitions in L1):

[0227] Actual cycle = 1696109705000 (first green activation) - 1696109585000 (last green activation) = 120 seconds (consistent with the original plan)

[0228] Calculate the duration of each phase:

[0229] Green light duration = Yellow light start time - Green light start time = 1696109735000 - 1696109705000 = 30 seconds (consistent with the original plan)

[0230] Yellow light duration = Red light start time - Yellow light start time = 1696109738000 - 1696109735000 = 3 seconds (consistent with the original plan)

[0231] Red light duration = next cycle green light start time - current red light start time = 1696109825000 - 1696109738000 = 87 seconds (consistent with the original plan)

[0232] If there is a deviation (e.g., the actual green light duration is 25 seconds), then record the deviation value (-5 seconds).

[0233] Step 3: Generate the adjusted timing plan

[0234] The original plan was revised based on the actual calculated timing parameters, and an "adjusted" flag and the reason for the adjustment were added, as shown in the example below:

[0235] {

[0236] "intersection_id": "INT001_X012" ,

[0237] "plan_id": "PLAN_20241001_0800_ADJ", / / The newly added ADJ marker indicates the adjusted plan.

[0238] "cycle": 120 ,

[0239] "adjusted": true,

[0240] "adjust_reason": "Sampling data shows that the actual green light duration for P1 is 25 seconds (original plan: 30 seconds)".

[0241] "phases": [

[0242] {

[0243] "phase_id": "P1",

[0244] "direction": "east-west",

[0245] "lights": [

[0246] {"light_id": "L1", "color": "green", "duration": 25}, / / Corrected to actual value

[0247] {"light_id": "L1" , "color": "yellow" , "duration": 3},

[0248] {"light_id": "L1" , "color": "red" , "duration": 92} / / Red light duration adjusted accordingly (120-25-3=92) ]

[0250] }

[0251] / / ... Other phases are adjusted similarly.

[0252] ],

[0253] "effective_time": {

[0254] "start": "08:16:00", / / Adjustments will take effect immediately.

[0255] "end": "18:00:00"

[0256] },

[0257] "calibration_data": "Calculated based on 300 sampled data points from REQ_20241001_0815" / / Calibration basis

[0258] }

[0259] The adjusted timing scheme is a new signal control strategy generated after calibration and correction based on the original scheme and high-frequency sampling measurement data. The adjustment process is further explained in the following text:

[0260] Data comparison and analysis:

[0261] The phase parameters set in the original timing scheme (PLAN_20241001_0800) are compared second by second with 300 high-frequency sampled data (1 second / time, lasting 5 minutes) from request REQ_20241001_0815. Key indicators such as the actual green light duration and phase switching time of each phase are analyzed in detail.

[0262] Identification bias:

[0263] Statistical sampling data revealed that the green light duration for the east-west phase (P1) in actual operation was 25 seconds, instead of the originally planned 30 seconds, representing a deviation of 5 seconds. This deviation may be due to factors such as equipment aging, control logic deviation, or external intervention.

[0264] Parameter correction:

[0265] The original plan was revised based on the measured values:

[0266] The green light duration for phase P1 will be adjusted from 30 seconds to 25 seconds.

[0267] The duration of the yellow light remains unchanged (3 seconds);

[0268] The red light duration has been adjusted accordingly to 92 seconds (ensuring the total cycle length remains 120 seconds: 120−25−3=92).

[0269] Other phases (such as P2) are also calibrated synchronously based on the actual performance in the sampled data, following the same logic.

[0270] Updated scheme identifier:

[0271] Generate a new scheme ID PLAN_20241001_0800_ADJ, where “_ADJ” explicitly identifies the adjusted version;

[0272] The newly added field "adjusted":true indicates that this solution has undergone data-driven calibration;

[0273] Add the "adjust_reason" field to clearly explain the basis for the adjustment: "Sampling data shows that the actual green light duration for P1 is 25 seconds (the original plan was 30 seconds)."

[0274] Effective time setting:

[0275] The adjusted scheme will take effect immediately from 08:16:00 (i.e., 1 minute after the high-frequency sampling ends) to quickly respond to measured deviations and improve the accuracy of signal control.

[0276] Calibration traceability:

[0277] The "calibration_data" field specifies the data source on which this adjustment is based: "Calculated based on 300 sample data from REQ_20241001_0815", ensuring that the adjustment process is traceable and verifiable.

[0278] In summary, the adjusted timing scheme, through a closed-loop process of "actual measurement - comparison - correction - release," achieves consistent calibration from theoretical timing to actual operation, improves the accuracy and reliability of traffic signal control, and lays the foundation for dynamic optimization and intelligent regulation.

[0279] In this embodiment, the adjusted timing scheme can be verified using the new sampling data (3 points within 10 minutes after adjustment) through the above-described verification logic:

[0280] If the matching rate is ≥90%, the adjustment is confirmed to be effective;

[0281] If the mismatch persists, repeat the timing scheme adjustment steps (increase the sampling duration to 10 minutes) until the scheme stabilizes.

[0282] Step 3: The cloud control platform pushes the adjusted traffic light timing scheme to the intelligent connected vehicle, and the intelligent connected vehicle obtains the accurate timing scheme.

[0283] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A traffic light data acquisition and control system for intelligent connected vehicles, characterized in that, include: The traffic signal data acquisition module is used to acquire and store traffic light data from the traffic light signal controller in real time through the traffic police cloud platform. The verification and calibration module is used to obtain the traffic light timing scheme and a small amount of low-frequency traffic light real-time data from the traffic police cloud platform through the cloud control platform. It uses the small amount of low-frequency traffic light real-time data to verify the timing scheme. When the timing scheme is verified and matches, the timing scheme is sent to the vehicle. When the timing scheme is not verified and matches, it requests a large amount of high-frequency real-time traffic light data from the traffic police cloud platform to calibrate the timing scheme and obtain the calibrated timing scheme. The timing scheme acquisition module is used to send the calibrated timing scheme to the vehicle through the cloud control platform, and the vehicle obtains the calibrated timing scheme.

2. The traffic light data acquisition and control system for intelligent connected vehicles according to claim 1, characterized in that, In the signal data acquisition module, a small amount of low-frequency traffic light real-time data is low-frequency sampling data with a sampling frequency of f1. Each sample data includes a unique intersection identifier, sampling timestamp, current execution phase, real-time status of the light group, data source identifier, and digital signature. The sampling frequency f1 is set from 30 seconds / time to 300 seconds / time.

3. The traffic light data acquisition and control system for intelligent connected vehicles according to claim 1, characterized in that, The specific process for verifying the timing scheme in the verification and calibration module is as follows: Step S211: Match the traffic light timing scheme of the same intersection with a small amount of low-frequency real-time traffic light data according to the unique intersection identifier, and filter out the sampled data that exceeds the effective period of the traffic light timing scheme; Step S212: Calculate the theoretical predicted state of the traffic light group at the sampling time based on the traffic light timing scheme; Step S213: Compare the predicted state of the theoretical light group with the actual light group state in a small amount of real-time low-frequency traffic light data to obtain the sampling point matching result; Step S214: Determine whether the traffic light timing scheme is effective based on the matching results of continuous sampling points.

4. A traffic light data acquisition and control system for intelligent connected vehicles according to claim 3, characterized in that, In step S212, calculating the theoretical predicted state of the lamp group at the sampling time includes calculating the cycle offset. The formula for calculating the cycle offset is as follows: cycle_offset=elapsed_time mod cycle; Where elapsed_time is the difference between the sampling point time and the start time of the traffic light timing scheme, cycle is the signal cycle duration of the traffic light timing scheme, and mod represents the modulo operation.

5. A traffic light data acquisition and control system for intelligent connected vehicles according to claim 3, characterized in that, In step S214, the specific determination process is as follows: The matching results of three consecutive sampling points are statistically analyzed. If the number of matching sampling points is ≥2, the traffic light timing scheme is determined to be valid; if the number of matching sampling points is ≤1, the traffic light timing scheme is determined to be invalid.

6. A traffic light data acquisition and control system for intelligent connected vehicles according to claim 3, characterized in that, In the verification and calibration module, when requesting real-time high-frequency traffic light data from the traffic police cloud platform, the request parameters include: unique identifier of the target intersection, request identifier, sampling frequency f2, and sampling duration. The sampling data field includes the sampling timestamp, the current execution phase, the current color of each light group, and the remaining time of the current color of each light group. The sampling frequency f2 can be set from 1 second / time to 10 seconds / time, and the sampling duration is specified. The setting range is 60 seconds to 600 seconds.

7. A traffic light data acquisition and control system for intelligent connected vehicles according to claim 6, characterized in that, The specific process for calibrating the timing scheme in the verification and calibration module is as follows: Step S221: Analyze the real-time data of high-frequency traffic lights and extract the timestamps of the state switching of each light group; Step S222: Calculate the actual timing parameters based on the timestamp of the state transition; Step S223: Correct the original traffic light timing scheme based on the actual timing parameters and generate the calibrated timing scheme.

8. A traffic light data acquisition and control system for intelligent connected vehicles according to claim 7, characterized in that, In the verification and calibration module, after generating the calibrated timing scheme, the following verification steps are also included: Recalibrate using three sampling points within 10 minutes after calibration. If the matching rate is ≥90%, the calibration is confirmed to be effective. If there is no matching, repeat the calibration steps and extend the sampling time of the high-frequency traffic light real-time data to 10 minutes.

9. A traffic light data acquisition and control system for intelligent connected vehicles according to claim 8, characterized in that, In the verification and calibration module, the timing scheme includes a unique intersection identifier, scheme ID, signal cycle duration, effective period, phase list, and scheme issuance timestamp; the phase list includes phase ID, traffic direction, and light group status parameters in execution order; the calibrated timing scheme adds adjustment mark and calibration basis fields, and the calibrated timing scheme takes effect immediately after high-frequency sampling ends.

10. A method for acquiring and controlling traffic light data for intelligent connected vehicles, characterized in that, Based on the system implementation as described in any one of claims 1 to 9, the system includes the following steps: Step S1: The traffic police cloud platform obtains and stores traffic light data from the traffic light controller in real time; Step S2: The cloud control platform obtains the traffic light timing scheme and a small amount of low-frequency real-time traffic light data from the traffic police cloud platform. It uses the small amount of low-frequency real-time traffic light data to verify the timing scheme. When the timing scheme is verified to match, it sends the timing scheme to the vehicle. When the timing scheme is not verified to match, it requests a large amount of high-frequency real-time traffic light data from the traffic police cloud platform to calibrate the timing scheme and obtain the calibrated timing scheme. Step S3: The cloud control platform sends the calibrated timing plan to the vehicle, and the vehicle obtains the calibrated timing plan.

Images