CN-121999624-A - Traffic signal cooperative control method and system based on AI (advanced technology attachment) double-core architecture

CN121999624ACN 121999624 ACN121999624 ACN 121999624ACN-121999624-A

Abstract

The application relates to the technical field of intelligent traffic signal control, and discloses a traffic signal cooperative control method and system based on an AI (advanced technology interface) double-core framework, wherein the method comprises the steps of obtaining detection data of an intersection entrance lane group in a current rolling control period; the traffic carrying capacity of the import lane group is evaluated, a phase benefit prejudging task is executed, a timing correction decision task is executed, a local safety chip executes safety constraint verification, and an actual execution result is fed back to the local AI intelligent chip. Compared with the prior art, the method mainly adopts a fixed timing mode to carry out signal control adjustment, and particularly the technical problem that the synchronous comparison of the continuous release benefits and the phase switching benefits of the current phase cannot be realized under the condition that the main road and the branch road traffic flow are unbalanced in the peak period. The application combines the dual-path rolling prediction mechanism through the fusion of the video detection data and the millimeter wave detection data, thereby realizing the self-adaptive coordination of the current phase extension control and the phase switching control.

Inventors

ZHAO XINYONG
LIU CHUNRUI
ZOU WEI
ZHAO YI
ZHANG PENG
MA WEI
Song Ruidan
MA XUGUANG
WANG LIANG
LIU GUANGLEI
ZHAI YUNFENG
YANG ZITONG
SHE HONGYAN
LIN JUN
ZHOU WEN

Assignees

华路易云科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260403

Claims (10)

1. The traffic signal cooperative control method based on the AI double-core architecture is characterized by comprising the following steps: S10, acquiring detection data of an entrance lane group of an intersection in a current rolling control period, carrying out fusion analysis on a queuing state, a vehicle arrival state and a traffic state of the entrance lane group based on the detection data, and outputting a traffic state result of the entrance lane group; Step S20, estimating the traffic bearing capacity of the import lane group based on the traffic state result, and outputting a service capacity estimation result; Step S30, a dual-path rolling prediction mechanism is adopted to execute a phase benefit pre-judging task based on a service capability assessment result, and a current phase continuous release cost result and a phase switching cost result are output; step S40, executing a timing correction decision task by adopting a dynamic threshold decision mechanism based on the current phase continuous release cost result and the phase switching cost result, and outputting a timing correction instruction; And S50, based on the timing correction instruction, executing safety constraint verification by a preset local safety chip, executing the timing correction instruction after the verification is passed, generating an actual execution result, and finally feeding back the actual execution result to the preset local AI intelligent chip.
2. The traffic signal cooperative control method based on the AI dual-core architecture of claim 1, wherein in step S10, specifically comprising: Step S101, acquiring detection data of an intersection entrance lane group in a current rolling control period, wherein the detection data comprises video detection data and millimeter wave detection data, the video detection data comprises vehicle target contour information, queuing tail position information, detection area occupation information and lane traffic track information, and the millimeter wave detection data comprises target distance information, target speed information, target distance information and target arrival time sequence information; step S102, cross checking is carried out on video detection data and millimeter wave detection data by adopting a space-time registration and weighted fusion method based on Kalman filtering and Markov distance threshold matching based on an intersection entrance lane group, so as to form a fusion queuing state result, a fusion vehicle arrival state result and a fusion passing state result; and step 103, constructing a traffic state result of the entrance lane group by adopting a characteristic parameter extraction and state vector generation mode based on Hotelling principal component analysis based on the fusion queuing state result, the fusion vehicle arrival state result and the fusion traffic state result.
3. The traffic signal cooperative control method based on the AI dual-core architecture of claim 1, wherein in step S20, specifically, the method comprises: Step S201, carrying out lane group classification processing by adopting a phase release relation mapping mode based on the output traffic state result, and outputting a first entrance lane group corresponding to the current release phase and a second entrance lane group corresponding to the non-release phase; step S202, aiming at a first entrance lane group and a second entrance lane group respectively, through combining the average head interval, the lane occupation rate, the vehicle arrival intensity and the preset saturation release capacity of each entrance lane group, the passing bearing capacity of the first entrance lane group and the second entrance lane group in unit time in the current rolling control period is evaluated, and a service capacity evaluation result is formed.
4. The traffic signal cooperative control method based on the AI dual-core architecture as set forth in claim 1, wherein the local security chip is configured to perform basic signal control, phase conflict verification, minimum green light duration constraint, maximum green light duration constraint, yellow light control, full red control, and abnormal degradation control, and the local AI intelligent chip is configured to perform traffic state calculation, queuing risk assessment, timing correction calculation, and phase switching decision.
5. The traffic signal cooperative control method based on the AI dual-core architecture of claim 3, wherein in step S30, specifically comprising: Step 301, based on the service capability evaluation result, taking a first import lane group corresponding to the current releasing phase and a second import lane group corresponding to the unreleased phase as objects, and performing rolling prediction on the vehicle dispersion change of the first import lane group and the queuing growth change of the second import lane group when the current phase keeps releasing in the next rolling control period to obtain a first prediction result under the condition that the current phase continues releasing; Step S302, based on a service capability evaluation result, taking a first inlet lane group and a second inlet lane group as objects, and performing rolling prediction on queuing accumulated changes of the first inlet lane group and vehicle dissipation changes of the second inlet lane group when entering a yellow light transition stage, a full red transition stage and a candidate next phase initial release stage in sequence after the current phase is ended, so as to obtain a second prediction result under a phase switching condition; Step S303, based on a first prediction result and a second prediction result, respectively carrying out phase loss quantization processing in a cost function construction mode, and outputting a current phase continuous release cost result and a phase switching cost result, wherein the current phase continuous release cost result is used for representing the comprehensive relationship between release benefits of a first import lane group and waiting losses of a second import lane group when the current phase continuous release is carried out, and the phase switching cost result is used for representing the comprehensive relationship between phase transition losses and queuing dissipation benefits of the second import lane group when the phase switching is carried out.
6. The traffic signal cooperative control method based on the AI dual-core architecture of claim 1, wherein in step S40, specifically comprising: Step S401, reading a current phase continuous release cost result and a phase switching cost result, and carrying out difference comparison processing on the current phase continuous release cost result and the phase switching cost result to obtain a cost difference result, wherein the cost difference result is used for representing the relative relation between the current phase continuous release benefit and the phase switching benefit; Step S402, when the cost difference result exceeds a preset phase switching trigger threshold, judging that the waiting loss and queuing growth influence of the continuous release of the current phase on the inlet lane group corresponding to the unreleased phase are greater than the continuous release benefits of the current phase, generating a switching control result that the current phase is ended and switched to the candidate next phase; step S403, determining timing correction content corresponding to the current rolling control period based on the switching control result or the extended release control result generated in step S402, and finally outputting a timing correction instruction.
7. The traffic signal cooperative control method based on the AI dual-core architecture of claim 1, wherein in step S50, specifically comprising: Step S501, receiving a timing correction instruction by a preset local security chip, and performing security constraint verification on the timing correction instruction based on the current signal phase operation state, a preset minimum green light duration constraint, a preset maximum green light duration constraint, a phase conflict mutual exclusion constraint, a yellow light transition duration constraint and a full red transition duration constraint to generate a security verification result; step S502, when the safety verification result shows that the timing correction instruction meets the execution condition, executing the corresponding timing correction instruction by a preset local safety chip; When the timing correction instruction is a phase switching instruction, the current phase yellow lamp control, the full red control and the target phase green lamp starting control are sequentially executed according to a preset lamp color time sequence, and corresponding transitional execution results are recorded; And step S503, after the execution of the timing correction instruction is completed, generating an actual execution result comprising a current actual execution phase, an actual release time result, a yellow light execution result, a full red execution result and an updated phase operation state result, and feeding back the actual execution result to a preset local AI intelligent chip.
8. The traffic signal cooperative control system based on the AI double-core architecture, which is applied to the traffic signal cooperative control method based on the AI double-core architecture as claimed in any one of claims 1 to 7, is characterized in that the traffic signal cooperative control system based on the AI double-core architecture comprises: The state acquisition and fusion analysis module is used for acquiring detection data of the entrance lane group of the intersection in the current rolling control period, carrying out fusion analysis on the queuing state, the vehicle arrival state and the traffic state of the entrance lane group based on the detection data, and outputting a traffic state result of the entrance lane group; the traffic bearing capacity evaluation module is used for evaluating the traffic bearing capacity of the import lane group based on the traffic state result and outputting a service capacity evaluation result; The dual-path rolling prediction module is used for executing a phase benefit pre-judging task by adopting a dual-path rolling prediction mechanism based on the service capability evaluation result and outputting a current phase continuous release cost result and a phase switching cost result; the timing correction decision module is used for executing a timing correction decision task by adopting a dynamic threshold decision mechanism based on the current phase continuous release cost result and the phase switching cost result and outputting a timing correction instruction; and the safety verification and execution feedback module is used for executing safety constraint verification by a preset local safety chip based on the timing correction instruction, executing the timing correction instruction and generating an actual execution result after the verification is passed, and finally feeding back the actual execution result to the preset local AI intelligent chip.
9. The traffic signal cooperative control device based on the AI double-core architecture is characterized by comprising a memory, a processor and a traffic signal cooperative control program based on the AI double-core architecture, wherein the traffic signal cooperative control program based on the AI double-core architecture is stored on the memory and can run on the processor, and the traffic signal cooperative control method based on the AI double-core architecture is realized when the traffic signal cooperative control program based on the AI double-core architecture is executed by the processor.
10. A computer program product, characterized in that the computer program product comprises an AI-dual-core architecture-based traffic signal cooperative control program, which, when executed by a processor, implements a AI-dual-core architecture-based traffic signal cooperative control method as claimed in any one of claims 1 to 7.

Description

Traffic signal cooperative control method and system based on AI (advanced technology attachment) double-core architecture Technical Field The invention relates to the technical field of intelligent traffic signal control, in particular to a traffic signal cooperative control method and system based on an AI (advanced technology interface) double-core architecture. Background At present, the traffic signal control of the urban plane intersection mostly adopts a fixed timing control mode, an induction type control mode or a self-adaptive adjustment mode based on a single detection source. The fixed timing control mode is usually used for presetting signal periods, green light durations and phase sequences of all time periods according to historical traffic flow investigation results, and is relatively simple in implementation, but relatively weak in response capability to real-time traffic state changes, the induction control mode can be used for adjusting local release durations to a certain extent according to incoming vehicle detection results, but the adjustment range is usually limited to a single phase or a single direction, queuing differences among a plurality of entrance lane groups are difficult to consider, and particularly, the conditions that traffic flows of main roads continuously arrive in high density, vehicles of branches intermittently arrive, queuing differences of a plurality of entrance roads are obvious and the phase release requirements are rapidly changed often appear at an urban intersection where main roads and branches intersect in the early and late peak periods. Under the scene, if a fixed timing mode is still adopted, the problems of insufficient green light duration of a main road, continuous accumulation of queuing and even expansion of an upstream road section often occur, and if the green light duration of the current phase is simply prolonged, the long waiting time of a branch is easily caused, and even the concentrated accumulation of the vehicles of the branch is caused. There is also a problem with existing schemes that rely on only a single video detection or a single millimeter wave detection that the characterization of complex traffic conditions is inadequate. For example, under the conditions of large vehicle shielding, small bus mixing, night illumination change, rain, fog, weather or dense detection area boundary targets, accurate queuing tail position and vehicle arrival information are difficult to stably obtain by single video detection, and the single millimeter wave detection has limited capability of representing lane occupation state, target contour and fine granularity track change although distance and speed information can be provided, so that the prior art cannot fully meet the requirements of fine recognition of traffic state, real-time phase benefit comparison and timing instruction safety execution of complex intersections under the condition of peak dynamic fluctuation. Therefore, there is a need for a traffic signal cooperative control method that can still realize stable perception of traffic state of an intersection, continuous release of current phase and real-time comparison of phase switching benefits, and complete timing correction instruction local closed-loop execution on the premise of meeting signal safety constraint under the conditions that main road continuous high-density arrival, branch intermittent arrival, queuing difference of a plurality of entrance lane groups is obvious and complex fluctuation of detection environment is detected, so as to improve traffic efficiency of the intersection, signal control instantaneity and operation safety of the intersection. Disclosure of Invention Aiming at the technical defects, the invention aims to provide a traffic signal cooperative control method based on an AI (advanced technology) double-core architecture, and aims to solve the technical problem that the synchronous comparison of the continuous release benefits and the phase switching benefits of the current phase cannot be realized in the prior art by mainly adopting a fixed timing mode to carry out signal control adjustment, especially under the condition that the main road and the branch traffic flow are unbalanced in the peak period. In order to solve the technical problems, the invention adopts the following technical scheme that the invention provides a traffic signal cooperative control method based on an AI double-core architecture. The traffic signal cooperative control method based on the AI double-core architecture comprises the following steps: S10, acquiring detection data of an entrance lane group of an intersection in a current rolling control period, carrying out fusion analysis on a queuing state, a vehicle arrival state and a traffic state of the entrance lane group based on the detection data, and outputting a traffic state result of the entrance lane group; Step S20, estimating the traffic bearing capacity of the imp