Intelligent traffic signal lamp dynamic regulation system and method based on artificial intelligence

Abstract

The application discloses an artificial intelligence-based intelligent road traffic signal lamp dynamic regulation system and method, relates to the field of signal lamp regulation, and comprises the following steps: constructing a traffic state characteristic vector based on traffic flow data; analyzing the traffic state characteristic vector in real time through a deep learning model, predicting a traffic flow change trend in a future time window, generating a prediction result, generating a dynamic signal lamp regulation strategy, and issuing the dynamic signal lamp regulation strategy to a signal lamp control unit; adjusting the signal lamp state in real time through the signal lamp control unit, generating a regulation result; and storing the regulation result and the traffic flow data in a traffic management database for data mining optimization, generating a signal lamp control optimization suggestion, and dynamically regulating the intelligent traffic signal lamp. The technical problem that traditional signal lamps are difficult to flexibly adjust according to real-time traffic flow is solved, the technical effect that the signal lamp timing is dynamically adjusted to realize intelligent timing and reduce congestion and the incidence of traffic accidents is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of signal light regulation, in particular to an intelligent road traffic signal light dynamic regulation system and method based on artificial intelligence. BACKGROUND

[0002] The control of traffic signal lights mainly relies on preset fixed timing schemes, which are often based on historical traffic data or experience to formulate, and are difficult to accurately reflect the changes of real-time traffic conditions. With the acceleration of urbanization and the continuous increase of traffic flow, the limitations of this static control method are increasingly prominent, leading to more serious problems such as traffic congestion, low traffic efficiency, and environmental pollution, resulting in the technical problem that traditional signal lights cannot be flexibly adjusted according to real-time traffic flow. SUMMARY

[0003] The present application provides an intelligent road traffic signal light dynamic regulation system and method based on artificial intelligence, which solves the technical problem that traditional signal lights cannot be flexibly adjusted according to real-time traffic flow, and achieves the technical effect of dynamically adjusting signal light timing to realize intelligent timing, reduce congestion and traffic accident rate. The present application provides an intelligent road traffic signal light dynamic regulation system based on artificial intelligence, which is applied to an intelligent road traffic signal light dynamic regulation method based on artificial intelligence, comprising: a vector construction module for constructing a traffic state feature vector based on traffic flow data; a real-time analysis module for real-time analysis of the traffic state feature vector through a deep learning model to predict the traffic flow trend in a future time window and generate a prediction result; a regulation analysis module for generating a dynamic signal light regulation strategy according to the prediction result; a real-time adjustment module for issuing the dynamic signal light regulation strategy to a signal light control unit to adjust the signal light state in real time through the signal light control unit and generate a regulation result; and a mining optimization module for storing the regulation result and the traffic flow data to a traffic management database for data mining optimization to generate signal light control optimization suggestions for dynamic regulation of intelligent traffic signal lights.

[0004] The present application also provides an intelligent road traffic signal light dynamic regulation method based on artificial intelligence, comprising: constructing a traffic state feature vector based on traffic flow data; real-time analysis of the traffic state feature vector through a deep learning model to predict the traffic flow trend in a future time window and generate a prediction result; generating a dynamic signal light regulation strategy according to the prediction result; issuing the dynamic signal light regulation strategy to a signal light control unit to adjust the signal light state in real time through the signal light control unit and generate a regulation result; and storing the regulation result and the traffic flow data to a traffic management database for data mining optimization to generate signal light control optimization suggestions for dynamic regulation of intelligent traffic signal lights.

[0005] The one or more technical solutions provided in the application have at least the following technical effects or advantages:

[0006] The artificial intelligence-based intelligent road traffic signal lamp dynamic regulation system and method provided by the application relate to the technical field of signal lamp regulation, solve the technical problem that traditional signal lamps are difficult to flexibly adjust according to real-time traffic flow, and achieve the technical effects of dynamically adjusting signal lamp timing to realize intelligent timing and reduce congestion and traffic accident rates. BRIEF DESCRIPTION OF DRAWINGS

[0007] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings of the embodiments of the application will be briefly introduced below, and the flowcharts are used to illustrate the operations performed by the system according to the embodiments of the application in the application. It should be understood that the foregoing or the following operations are not necessarily executed in sequence. On the contrary, according to the needs, various steps can be processed in reverse order or at the same time. At the same time, other operations can be added to these processes, or one or more steps of operation can be removed from these processes.

[0008] Figure 1 The structural schematic diagram of the artificial intelligence-based intelligent road traffic signal lamp dynamic regulation system provided by the embodiments of the application is shown in the figure.

[0009] Figure 2 The flowchart of the artificial intelligence-based intelligent road traffic signal lamp dynamic regulation method provided by the embodiments of the application is shown in the figure. DETAILED DESCRIPTION

[0010] The above description is only a summary of the technical solutions of the application, in order to more clearly understand the technical means of the application, the content of the specification can be implemented, and in order to make the above and other purposes, features and advantages of the application more obvious and easy to understand, the following specific embodiments of the application are described.

[0011] In order to make the purpose, technical solutions and advantages of the application more clear, the application will be further described in detail below in combination with the drawings, and the described embodiments should not be regarded as limiting the application, all other embodiments obtained by the person skilled in the art without making creative labor are within the scope of protection of the application.

[0012] In the following description, "some embodiments" are referred to, which describe a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict, and the term "first\second" referred to is only to distinguish similar objects, and does not represent a specific order for the objects. The terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0013] The embodiments of the present application provide an intelligent road traffic signal lamp dynamic regulation method based on artificial intelligence, which is applied to an intelligent road traffic signal lamp dynamic regulation system based on artificial intelligence, as shown in Figure 1 The system comprises:

[0014] The intelligent road traffic signal lamp dynamic regulation system based on artificial intelligence according to the embodiments of the present application is used to solve the technical problem that the traditional signal lamp is difficult to flexibly adjust according to the real-time traffic flow, and achieves the technical effect of dynamically adjusting the signal lamp timing to realize intelligent timing and reduce congestion and traffic accident rate. The intelligent road traffic signal lamp dynamic regulation system based on artificial intelligence comprises a vector construction module 10 configured to construct a traffic state feature vector based on traffic flow data. In the following, the specific configuration of the vector construction module 10 will be described in detail. As described above, the vector construction module 10 can further comprise: a first calculation unit configured to extract historical traffic flow data of a target intersection, calculate average flow, speed standard deviation and queue length change rate in different time periods; a weight construction unit configured to identify abnormal event features to construct event influence weights; a first prediction unit configured to perform congestion propagation prediction based on historical congestion patterns to obtain congestion prediction results; and a data fusion unit configured to fuse the average flow, the speed standard deviation, the queue length change rate, the event influence weights and the congestion prediction results to obtain a traffic state feature vector.

[0015] First, historical traffic flow data, including vehicle flow, speed distribution, and queue length, etc. are extracted from the traffic monitoring system, inductive loops, or video detection devices at the target intersection. For different time periods (e.g., morning peak, flat peak, evening peak), the data is segmented and aggregated by time window (e.g., 15 minutes): average flow is obtained by calculating the average number of vehicles passing through the intersection per unit time in each period; speed standard deviation is calculated based on the dispersion of all vehicle instantaneous speeds in each period, reflecting the volatility of speed; the queue length change rate is obtained by normalizing the growth rate of the queue length (e.g., from the initial value to the peak value) in each time window. Abnormal event features can be identified by event detection algorithms (e.g., traffic accident identification based on computer vision or weather sensor data): for traffic accidents, the event location (whether in the core area of the intersection), duration (e.g., more than 30 minutes), and the number of affected lanes (e.g., occupying more than 2 lanes) are extracted; for severe weather (heavy rain, heavy fog), the visibility (e.g., less than 100 meters) and precipitation intensity (e.g., more than 20 mm per hour) are quantified. Based on the event type and parameters, the AHP (Analytic Hierarchy Process) is used to construct the event impact weight, for example, the traffic accident weight is 0.7 (if it affects more than 50% of the lanes), and the rain weight is 0.5 (if the visibility is less than 100 meters). The weight values are corrected by expert scoring method. Further, a congestion propagation model is constructed based on historical congestion data: the GNN (Graph Neural Network) is used to simulate the congestion diffusion path of upstream and downstream road segments of the intersection, the input is the historical congestion start time, duration, and diffusion range (e.g., the time for congestion to spread from the intersection to the adjacent road segment), and the output is the area and duration probability of congestion that may be covered in the next 30 minutes (e.g., the probability of spreading to the downstream 500-meter area is 80%).

[0016] Finally, the above indicators are fused into a traffic state feature vector: the average flow, speed standard deviation, and queue length change rate are standardized by Z-score; the event impact weight and congestion prediction results (e.g., diffusion range probability) are normalized by Min-Max. A feature vector containing 5 dimensions (e.g., [flow mean = 1.2, speed standard deviation = 0.8, queue change rate = 0.6, event weight = 0.7, congestion probability = 0.85]) is generated by feature concatenation and input into a deep learning model for traffic state classification and control strategy generation.

[0017] The real-time analysis module 20 is configured to analyze the traffic state feature vector in real time through a deep learning model, predict the traffic flow trend in a future time window, and generate a prediction result. The specific configuration of the real-time analysis module 20 will be described in detail below. As described above, the real-time analysis module 20 can further include a data training unit configured to collect traffic flow data of a target intersection for N consecutive days, build a training data set, and the training data set includes the traffic state feature vector and the corresponding actual traffic flow change, and N is an integer greater than 2; a feature construction unit configured to extract the spatio-temporal features of the traffic flow data using a convolutional neural network to generate high-dimensional features; and a data modeling unit configured to model the high-dimensional features through a long short-term memory network to obtain the deep learning model.

[0018] The traffic flow data of the target intersection for N consecutive days (N>2) is collected. These data should contain multiple dimensions of traffic state feature vectors, such as traffic volume, vehicle speed, lane occupancy, etc., and the actual traffic flow changes corresponding to these feature vectors. These raw data constitute the training data set, which provides a basis for the model to learn the internal rules of traffic flow changes.

[0019] Next, the convolutional neural network (CNN) is used to process the training data set. CNN can efficiently extract the spatio-temporal features of the data, capture the local patterns and trends of the traffic flow data through the convolution kernel sliding in the time and space dimensions, and generate high-dimensional feature representations. The high-dimensional features not only contain the spatial distribution information of the traffic flow, but also contain the dynamic changes in time. After obtaining the high-dimensional features, the high-dimensional features are input into the long short-term memory network (LSTM) for time series modeling. The LSTM network captures the long-term dependencies in the time series data, effectively remembers and forgets information according to the gating mechanism, and accurately predicts the future traffic flow changes. By training the LSTM network, the model learns the mapping relationship from the high-dimensional features to the actual traffic flow changes, and thus obtains a deep learning model that can accurately predict the traffic flow. Not only does it improve the prediction accuracy of the model, but it also enhances its adaptability to complex traffic flow changes.

[0020] Further, the deep learning model is applied to the analysis of real-time traffic state feature vectors. The feature vectors are collected in real time by sensors deployed at the target intersection and continuously input into the model. The model first uses the convolutional neural network (CNN) to quickly process the input feature vectors and extract high-dimensional features containing spatio-temporal information. Subsequently, these high-dimensional features are sent to the long short-term memory network (LSTM) for time series analysis.

[0021] The LSTM network captures the long-term dependencies hidden in the feature vectors, enabling accurate prediction of traffic flow trends in future time windows. This prediction process is online and real-time, allowing for rapid response to changes in traffic conditions. Ultimately, the model generates prediction results that include expected traffic flow values and trends over a future period, providing valuable decision support for traffic management departments.

[0022] Through real-time analysis processes, the predictive capabilities of deep learning models are utilized to achieve accurate traffic flow predictions, contributing to the smooth operation and efficient management of urban traffic.

[0023] The control analysis module 30 is configured to generate a dynamic signal light control strategy based on the prediction results. The specific configuration of the control analysis module 30 will be described in detail below. As mentioned above, the control analysis module 30 can further include: a second calculation unit for extracting traffic flow change prediction trends based on the prediction results, calculating the optimal green light duration of each phase, and obtaining multiple phase optimal green light durations; a data detection unit for identifying high-priority phases, determining phase switching priorities, and starting an emergency response mode when detecting abnormal events, which forcibly extends the green light duration of related phases and shortens the duration of other phases through the emergency response mode; a data integration unit for integrating the multiple phase optimal green light durations, phase switching priorities, and emergency response mode into the dynamic signal light control strategy.

[0024] First, the prediction trend of traffic flow changes is extracted, which represents the dynamic changes of vehicle flow in each direction over a future period. The optimal green light duration of each phase (i.e., the time period for different direction vehicle flow) is calculated based on traffic flow theory models. This calculation process aims to maximize the efficiency of the intersection, reduce vehicle waiting time, and alleviate traffic congestion.

[0025] After obtaining the optimal green light duration of multiple phases, high-priority phases are further identified, which usually correspond to directions with high traffic flow or traffic pressure. Based on priority information, the priority order of phase switching is determined to ensure that the direction with the largest traffic flow can obtain more passing opportunities within a limited time.

[0026] In addition, the emergency response mode is used to deal with traffic abnormal events such as traffic accidents and vehicle malfunctions. When such abnormal events are detected, the emergency response mode is immediately started, which forcibly extends the green light duration of related phases and shortens the duration of other phases to ensure that emergency vehicles (such as ambulances and fire trucks) can quickly pass through the intersection, while also providing necessary passing convenience for affected vehicles.

[0027] Finally, the optimal green light duration of multiple phases, phase switching priority, and emergency response mode are integrated into a complete dynamic signal light control strategy. The dynamic signal light control strategy can flexibly adjust the signal light control according to the real-time traffic conditions, effectively improving the traffic capacity and safety of the intersection.

[0028] Next, the specific configuration of the regulation analysis module 30 will be described in detail. As described above, the regulation analysis module 30 can further include: a first event generation unit for analyzing road construction plan periods according to target roads to generate temporary road closure events; a second event generation unit for conducting accident impact analysis on target roads to generate traffic accident events; a third event generation unit for conducting weather visibility analysis on target roads to generate adverse weather events; and a fourth event generation unit for adding the temporary road closure events, the traffic accident events, and the adverse weather events to the abnormal events.

[0029] In order to comprehensively cover abnormal events that may affect traffic flow, first, multi-dimensional analysis is performed on the target road. Temporary road closure events are planned and generated according to road construction plan periods. The temporary road closure events can include specific times of construction start and end, specific road sections closed, and predicted traffic impact range, providing an important basis for subsequent traffic management.

[0030] Secondly, accident impact analysis is conducted on the target road. By analyzing historical accident data and road characteristics, possible traffic accident events are predicted and generated. The traffic accident events record in detail the possible locations, times, and potential impact degrees of accidents, which helps to develop timely response measures.

[0031] In addition, in-depth analysis is also conducted on the weather visibility of the target road. By combining meteorological data and road environment, adverse weather events are identified and generated. The adverse weather events can include fog, heavy rain, snowstorm and other weather conditions that may affect traffic safety, as well as their potential impact on road traffic capacity.

[0032] Finally, these temporary road closure events, traffic accident events, and adverse weather events are added to the abnormal event library. The abnormal event library provides comprehensive abnormal event information support for the development of dynamic signal light control strategies, ensuring that the strategies can make timely and effective adjustments to various sudden situations.

[0033] The real-time adjustment module 40 is configured to issue the dynamic signal light control strategy to the signal light control unit, and to generate a regulation result by adjusting the signal light state in real time through the signal light control unit.

[0034] After the dynamic signal light regulation strategy is formulated, it is timely issued to the signal light control unit. This step can be realized through a dedicated communication channel, ensuring the rapid transmission and accurate reception of the regulation strategy. The signal light control unit, as a core component of the traffic management system, is responsible for adjusting the state of the signal light in real time according to the received regulation strategy.

[0035] The issuance of the regulation strategy triggers the automatic response mechanism of the signal light control unit. The control unit first analyzes the key information in the regulation strategy, including the optimal green light duration of each phase, phase switching priority, and emergency response mode, etc. Then, it accurately adjusts the timing scheme of the signal light according to these information to ensure efficient and orderly management of traffic flow at the intersection.

[0036] During the implementation of the regulation strategy, the signal light control unit also continuously monitors the traffic conditions at the intersection and generates real-time regulation results. The regulation results are used to represent key indicators such as traffic flow changes, vehicle waiting time, and intersection passing efficiency after the strategy is executed. By comparing the data before and after the regulation, the effectiveness of the dynamic signal light regulation strategy can be evaluated, and necessary adjustments and optimizations can be made in a timely manner.

[0037] Finally, the dynamic signal light regulation strategy is transformed into actual traffic management effect through real-time adjustment of the signal light control unit. Not only does it improve the passing capacity of the intersection, but also enhances the flexibility and adaptability of the traffic system, providing a strong guarantee for the smooth operation of urban traffic.

[0038] The mining optimization module 50 is used to store the regulation results and the traffic flow data to the traffic management database for data mining optimization, and generate signal light control optimization suggestions for dynamic regulation of intelligent traffic signal lights.

[0039] In the following, the specific configuration of the mining optimization module 50 will be described in detail. As described above, the mining optimization module 50 can further include: a data recognition unit for statistically analyzing the regulation results and the traffic flow data of the target road for M consecutive months, identifying high-frequency congestion periods and inefficient phases, and M is an integer greater than 1; a scheme adjustment unit for adjusting the default timing scheme of the signal light based on the high-frequency congestion period to generate a first timing scheme; a switching optimization unit for optimizing phase switching based on the inefficient phase to determine phase switching data; a data mining unit for storing the first timing scheme and the phase switching data to the traffic management database for data mining optimization to generate the signal light control optimization suggestions.

[0040] To continuously optimize the signal control strategy, the regulation results and traffic flow data of the target road for M consecutive months (M > 1) can be statistically analyzed. This long-term data accumulation can provide rich traffic information, which helps to identify high-frequency congestion periods and inefficient phases.

[0041] First, statistical analysis is used to screen out periods of frequent congestion from a large amount of data. These periods usually correspond to key moments of high traffic flow and low traffic efficiency. Based on these high-frequency congestion periods, the default timing scheme of the signal is adjusted to generate a first timing scheme that is more adaptive to actual traffic demand.

[0042] Second, the identified inefficient phases are further optimized. By analyzing the traffic flow changes during phase switching, the optimal phase switching time and sequence are determined, generating phase switching data. These data provide an important basis for fine-tuning signal control.

[0043] Finally, the first timing scheme and phase switching data are stored in the traffic management database, and data mining techniques are used for further analysis and optimization. By mining the relevance and regularity of the data, signal control optimization suggestions are generated. These suggestions aim to further improve the traffic efficiency at intersections and reduce traffic congestion, providing more scientific and effective decision support for urban traffic management. Not only does it embody the concept of data-driven management, but it also demonstrates the great potential of intelligent transportation systems in urban traffic optimization.

[0044] Next, the specific configuration of the mining optimization module 50 will be described in detail. As mentioned above, the mining optimization module 50 can further include: an index formulation unit for implementing the first timing scheme and the phase switching data at the target intersection and continuously monitoring traffic flow data to formulate a vehicle delay index; a data judgment unit for calculating a vehicle delay improvement rate based on the vehicle delay index and determining whether the vehicle delay improvement rate is lower than a preset improvement rate; a data updating unit for updating the first timing scheme and the phase switching data when the vehicle delay improvement rate is lower than the preset improvement rate until the preset improvement rate is met, generating an i-th timing scheme and a j-th phase switching data, i and j being positive integers not equal to 0; a data storage unit for storing the i-th timing scheme and the j-th phase switching data in the traffic management database.

[0045] After implementing the first timing scheme and phase switching data at the target intersection, traffic flow data is continuously monitored, and a vehicle delay index is formulated to evaluate the regulation effect. By calculating the average delay time of vehicles passing through the intersection, the impact of the regulation strategy on traffic flow is intuitively reflected.

[0046] Then, the vehicle delay improvement rate is calculated according to the vehicle delay index, and is compared with the preset improvement rate. If the improvement rate is lower than the preset value, it indicates that the current regulation strategy still has optimization space. The first timing plan and phase switching data are iteratively updated by adjusting the green light duration, phase switching sequence and other parameters, so as to achieve higher traffic efficiency.

[0047] This updating process continues until the vehicle delay improvement rate reaches or exceeds the preset improvement rate. At this time, the i-th timing plan and j-th phase switching data (i, j are positive integers not equal to 0) that meet the requirements are obtained, and these optimized schemes are stored in the traffic management database. Not only the traffic capacity of the intersection is improved, but also strong support is provided for subsequent optimization work.

[0048] In the foregoing, with reference to Figure 1 The intelligent road traffic signal lamp dynamic regulation system based on artificial intelligence according to the embodiments of the present application is described in detail. Next, the intelligent road traffic signal lamp dynamic regulation method based on artificial intelligence according to the embodiments of the present application will be described with reference to Figure 2

[0049] Step A100, constructing a traffic state feature vector based on traffic flow data; in one possible implementation, step A100 further includes step A110, extracting historical traffic flow data of a target intersection, calculating average flow, speed standard deviation and queue length change rate in different time periods; performing step A120, identifying abnormal event features to construct event influence weight; performing step A130, predicting congestion propagation based on historical congestion patterns to obtain congestion prediction results; performing step A140, fusing the average flow, the speed standard deviation, the queue length change rate, the event influence weight and the congestion prediction results to obtain a traffic state feature vector.

[0050] Performing step A200, real-time analyzing the traffic state feature vector through a deep learning model to predict traffic flow change trend in a future time window and generate a prediction result; in one possible implementation, step A200 further includes step A210, collecting traffic flow data of a target intersection for consecutive N days to construct a training data set, the training data set including the traffic state feature vector and corresponding actual traffic flow change, N being an integer greater than 2; performing step A220, using a convolutional neural network to extract spatio-temporal features of the traffic flow data to generate high-dimensional features; performing step A230, modeling the high-dimensional features through a long short-term memory network to obtain the deep learning model.

[0051] ​performing step A300, generating a dynamic signal lamp control strategy according to the prediction result; in a possible implementation, step A300 further includes step A310, extracting a traffic flow change prediction trend according to the prediction result, calculating an optimal green light duration of each phase, and obtaining a plurality of phase optimal green light durations; performing step A320, identifying a high-priority phase, determining a phase switching priority, starting an emergency response mode when an abnormal event is detected, and forcibly extending the green light duration of the related phase and shortening the duration of other phases through the emergency response mode; performing step A330, integrating the plurality of phase optimal green light durations, the phase switching priority, and the emergency response mode into the dynamic signal lamp control strategy.

[0052] In a possible implementation, step A320 further includes step A signal lamp control, generating a temporary road closure event according to a target road analysis of a road construction plan period; performing step A322, generating a traffic accident event according to a target road accident impact analysis; performing step A323, generating a severe weather event according to a target road weather visibility analysis; performing step A324, adding the temporary road closure event, the traffic accident event, and the severe weather event to the abnormal event.

[0053] performing step A400, issuing the dynamic signal lamp control strategy to a signal lamp control unit, adjusting the signal lamp state in real time through the signal lamp control unit, and generating a control result; performing step A500, storing the control result and the traffic flow data to a traffic management database for data mining optimization, and generating a signal lamp control optimization suggestion for dynamic control of the intelligent traffic signal lamp.

[0054] In a possible implementation, step A500 further includes step A510, counting the control result and the traffic flow data of a target road for consecutive M months, identifying a high-frequency congestion period and an inefficient phase, and M is an integer greater than 1; performing step A520, adjusting a default timing scheme of the signal lamp based on the high-frequency congestion period, and generating a first timing scheme; performing step A530, performing phase switching optimization based on the inefficient phase, and determining phase switching data; performing step A540, storing the first timing scheme and the phase switching data to the traffic management database for data mining optimization, and generating the signal lamp control optimization suggestion.

[0055] In a possible implementation, the step A540 further includes a step A541 of implementing the first timing plan and the phase switching data at the target intersection and continuously monitoring traffic flow data to develop a vehicle delay index; a step A542 of calculating a vehicle delay improvement rate according to the vehicle delay index and determining whether the vehicle delay improvement rate is lower than a preset improvement rate; a step A543 of updating the first timing plan and the phase switching data until the preset improvement rate is met when the vehicle delay improvement rate is lower than the preset improvement rate to generate an i-th timing plan and a j-th phase switching data, i and j being positive integers not equal to 0; and a step A544 of storing the i-th timing plan and the j-th phase switching data to the traffic management database.

[0056] The embodiments of the present application solve the technical problem that the conventional signal lamp is difficult to flexibly adjust according to real-time traffic flow, and achieve the technical effects of dynamically adjusting signal lamp timing to realize intelligent timing and reduce congestion and traffic accident rates.

[0057] The intelligent road traffic signal lamp dynamic regulation system based on artificial intelligence provided by the embodiments of the present application can execute the intelligent road traffic signal lamp dynamic regulation method based on artificial intelligence provided by any embodiment of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0058] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server, and each unit and module included is only divided according to the functional logic, but is not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy mutual differentiation, and do not limit the protection scope of the present application.

[0059] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. An artificial intelligence-based intelligent road traffic signal lamp dynamic regulation system, characterized in that, The system comprises: a vector construction module for constructing a traffic state feature vector based on traffic flow data; a real-time analysis module for predicting traffic flow trends in a future time window by real-time analysis of the traffic state feature vector through a deep learning model, and generating a prediction result; a regulation analysis module for generating a dynamic signal light regulation strategy according to the prediction result; a real-time adjustment module for issuing the dynamic signal light regulation strategy to a signal light control unit, and adjusting the signal light state through the signal light control unit in real time to generate a regulation result; a mining optimization module for storing the regulation result and the traffic flow data to a traffic management database for data mining optimization, and generating a signal light control optimization suggestion for dynamic regulation of intelligent traffic signal lights; the mining optimization module comprises: a data identification unit for identifying high-frequency congestion periods and inefficient phases by statistically analyzing the regulation result and the traffic flow data of a target road for M consecutive months, M being an integer greater than 1; a scheme adjustment unit for adjusting the default timing scheme of the signal light based on the high-frequency congestion periods, and generating a first timing scheme; a switching optimization unit for performing phase switching optimization based on the inefficient phases, and determining phase switching data; a data mining unit for storing the first timing scheme and the phase switching data to the traffic management database for data mining optimization, and generating the signal light control optimization suggestion; an index formulation unit for implementing the first timing scheme and the phase switching data at a target intersection, and continuously monitoring traffic flow data to formulate a vehicle delay index; a data determination unit for calculating a vehicle delay improvement rate according to the vehicle delay index, and determining whether the vehicle delay improvement rate is lower than a preset improvement rate; a data updating unit for updating the first timing scheme and the phase switching data until the preset improvement rate is met when the vehicle delay improvement rate is lower than the preset improvement rate, and generating an i-th timing scheme and a j-th phase switching data, i and j being positive integers not equal to 0; a data storage unit for storing the i-th timing scheme and the j-th phase switching data to the traffic management database.

2. The artificial intelligence-based intelligent traffic signal lamp dynamic regulation system of claim 1, wherein, The vector construction module comprises: a first calculation unit for extracting historical traffic flow data of a target intersection, and calculating average flow, speed standard deviation and queue length change rate in different time periods; a weight construction unit for identifying abnormal event features to construct event influence weights; a first prediction unit for predicting congestion propagation based on historical congestion patterns to obtain a congestion prediction result; a data fusion unit for fusing the average flow, the speed standard deviation, the queue length change rate, the event influence weights and the congestion prediction result to obtain a traffic state feature vector. 3.The artificial intelligence-based intelligent traffic signal lamp dynamic regulation system of claim 1, wherein, The real-time analysis module comprises: a data training unit for collecting traffic flow data of a target intersection for N consecutive days, and constructing a training data set, the training data set comprising the traffic state feature vector and corresponding actual traffic flow changes, N being an integer greater than 2; The feature construction unit is configured to extract spatiotemporal features of the traffic flow data using a convolutional neural network to generate high-dimensional features. The data modeling unit is configured to model the high-dimensional features as a time series using a long short-term memory network to obtain the deep learning model.

4. The artificial intelligence-based intelligent traffic signal lamp dynamic regulation system of claim 1, wherein, The control analysis module includes: The second calculation unit is configured to extract traffic flow change prediction trends from the prediction results, calculate optimal green light durations for each phase, and obtain optimal green light durations for multiple phases. The data detection unit is configured to identify high-priority phases, determine phase switching priorities, and start an emergency response mode when an abnormal event is detected, which forcibly extends the green light duration of the relevant phase and shortens the duration of other phases. The data integration unit is configured to integrate the optimal green light durations for the multiple phases, the phase switching priorities, and the emergency response mode into the dynamic traffic signal control strategy. 5.The artificial intelligence-based intelligent traffic signal lamp dynamic regulation system of claim 4, wherein, The control analysis module includes: The first event generation unit is configured to analyze road construction plan periods based on target roads to generate temporary road closure events. The second event generation unit is configured to analyze the impact of accidents based on target roads to generate traffic accident events. The third event generation unit is configured to analyze weather visibility based on target roads to generate adverse weather events. The fourth event generation unit is configured to add the temporary road closure events, the traffic accident events, and the adverse weather events to the abnormal events.

6. The method for dynamic regulation of intelligent traffic signal based on artificial intelligence, characterized in that, The method is used to implement the artificial intelligence-based intelligent road traffic signal dynamic control system according to any one of claims 1-5, and the method includes: Based on traffic flow data, a traffic state feature vector is constructed. A deep learning model is used to analyze the traffic state feature vector in real time, predict traffic flow change trends in a future time window, and generate prediction results. Based on the prediction results, a dynamic traffic signal control strategy is generated. The dynamic traffic signal control strategy is sent to a signal control unit, which adjusts the signal light state in real time to generate a control result. The control result and the traffic flow data are stored in a traffic management database for data mining optimization to generate signal control optimization suggestions for dynamic control of intelligent traffic signals.

Images