Intelligent control method and system for traffic signal lights

Abstract

The application discloses a traffic signal lamp intelligent control method and system, and relates to the technical field of intelligent transportation systems.The method comprises the following steps: deploying a sensor group at a target traffic area signal lamp to collect traffic operation and environmental data; acquiring an adaptive analysis model containing abnormality detection and congestion prediction through a traffic scheduling platform to analyze traffic conditions and obtain congestion parameters; starting an emergency response mechanism according to the parameters, then analyzing and determining signal lamp control strategy parameters, and finally realizing adaptive regulation and control of the traffic signal lamp.The technical problems of insufficient real-time response capability, inability to effectively predict traffic abnormalities and congestion in the existing traffic signal lamp control system are solved, real-time traffic data analysis and intelligent prediction are achieved, the real-time response capability of the traffic signal lamp system is enhanced, and traffic abnormalities and congestion can be effectively predicted and responded to.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent transportation systems, in particular to a traffic signal intelligent control method and system. BACKGROUND

[0002] With the acceleration of urbanization, traffic flow has increased dramatically, and traffic congestion and accidents have become increasingly prominent. Traditional traffic signal control systems usually rely on fixed signal cycles or manually set rules, lack real-time response and dynamic adjustment capabilities to traffic flow changes, and are difficult to cope with sudden events and peak traffic pressure. These systems cannot effectively predict and identify traffic abnormalities, congestion and other conditions, resulting in low traffic efficiency and delays. Therefore, how to improve the control efficiency of traffic signals and reduce the degree of traffic congestion through intelligent means has become a difficult problem to be solved in urban traffic management.

[0003] In the related art, there are technical problems of insufficient real-time response capability in traffic signal control systems, and inability to effectively predict traffic abnormalities and congestion. SUMMARY

[0004] The present application provides a traffic signal intelligent control method and system, which solves the technical problems of insufficient real-time response capability in existing traffic signal control systems, and inability to effectively predict traffic abnormalities and congestion.

[0005] The present application provides a traffic signal intelligent control method, comprising:

[0006] Deploying a sensor group on the target traffic area signal light, acquiring traffic operation data information and traffic environment data information through the sensor group; acquiring a traffic adaptive analysis model through a traffic scheduling platform, the traffic adaptive analysis model being composed of a traffic anomaly detection sub-model and a traffic congestion prediction sub-model; analyzing the traffic operation data information and the traffic environment data information based on the traffic adaptive analysis model to obtain traffic operation congestion parameters; starting a traffic emergency response mechanism according to the traffic operation congestion parameters; analyzing the traffic operation congestion parameters based on the traffic emergency response mechanism to determine traffic signal control strategy parameters; and adaptively regulating and controlling the target traffic area signal light based on the traffic signal control strategy parameters.

[0007] The present application provides a traffic signal intelligent control system, comprising:

[0008] The sensor group deployment module is used for deploying a sensor group on a target traffic area signal lamp, and traffic operation data information and traffic environment data information are collected by the sensor group; the traffic adaptive analysis model acquisition module is used for acquiring a traffic adaptive analysis model through a traffic scheduling platform, and the traffic adaptive analysis model is composed of a traffic anomaly detection sub-model and a traffic congestion prediction sub-model; the traffic condition analysis module is used for performing traffic condition analysis on the traffic operation data information and traffic environment data information based on the traffic adaptive analysis model, and acquiring traffic operation congestion parameters; the traffic emergency response mechanism starting module is used for starting a traffic emergency response mechanism according to the traffic operation congestion parameters; the signal lamp control strategy analysis module is used for performing signal lamp control strategy analysis on the traffic operation congestion parameters based on the traffic emergency response mechanism, and determining traffic signal lamp control strategy parameters; and the adaptive regulation and control module is used for performing adaptive regulation and control on the target traffic area signal lamp based on the traffic signal lamp control strategy parameters.

[0009] The traffic signal lamp intelligent control method and system provided in the application first deploy a sensor group on a target traffic area signal lamp to collect traffic operation and environment data. An adaptive analysis model containing anomaly detection and congestion prediction is acquired through a traffic scheduling platform to analyze traffic conditions and obtain congestion parameters. An emergency response mechanism is started according to these parameters, and signal lamp control strategy parameters are analyzed and determined, so as to finally realize adaptive regulation and control of the traffic signal lamp. Through real-time traffic data analysis and intelligent prediction, the real-time response capability of the traffic signal lamp system is enhanced, and the technical effects of effectively predicting and responding to traffic anomalies and congestion are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0010] 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. In the present application, a flowchart is used to illustrate the operations performed by the system according to the embodiments of the application. It should be understood that the foregoing or the following operations are not necessarily performed in sequence. On the contrary, various steps can be processed in reverse order or simultaneously according to needs. Meanwhile, other operations can be added to these processes, or one or more steps of operations can be removed from these processes.

[0011] Figure 1 A flowchart of a traffic signal lamp intelligent control method provided by the embodiments of the application is shown in the figure.

[0012] Figure 2 A structural diagram of a traffic signal lamp intelligent control system provided by the embodiments of the application is shown in the figure.

[0013] Explanation of reference signs: sensor group deployment module 10, traffic adaptive analysis model acquisition module 20, traffic condition analysis module 30, traffic emergency response mechanism starting module 40, signal light control strategy analysis module 50, adaptive regulation module 60. DETAILED DESCRIPTION

[0014] The above description is only a summary of the technical scheme of the present application. In order to make the technical means of the present application more clear, the present application can be implemented according to the content of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described.

[0015] In order to make the purposes, technical schemes and advantages of the present application more clear, the following will combine the drawings to further describe the present application in detail, the described embodiments should not be regarded as the limitation of the present application, all other embodiments obtained by the person skilled in the art without creative labor belong to the scope of protection of the present application.

[0016] In the following description, "some embodiments" are related to a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subset of all possible embodiments, and can be combined with each other without conflict, the term "first\second" involved only distinguishes similar objects, and does not represent the specific order of the object. The terms "include" and "have" and any variations, 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.

[0017] The embodiments of the present application provide a traffic signal intelligent control method, as shown in the figure, the method comprises: Figure 1 The method comprises the steps of:

[0018] Step S100, deploying a sensor group on a target traffic area signal light, and collecting traffic operation data information and traffic environment data information through the sensor group. Specifically, first, sensor group deployment planning is performed, and according to the characteristics of the target traffic area, the types of geomagnetic, video, meteorological, acoustic and other sensors are selected, and the sensors are reasonably arranged above or around each lane of the signal light. Professional personnel install and debug to ensure that the position, angle, line are correct and the parameter calibration is accurate. Then, traffic operation data is collected, the geomagnetic sensor senses the vehicle magnetic field change to record the in-out time, calculates the vehicle flow of each lane, the video camera tracks the vehicle feature point displacement speed according to the image processing algorithm combined with the frame rate and resolution, and monitors the head and tail of the queuing vehicles to calculate the queuing length. The data is transmitted to the traffic control center at a certain interval. In addition, traffic environment data is also collected, and the meteorological sensor measures temperature, humidity, illumination, wind speed and direction, etc., and the acoustic sensor indirectly reflects the traffic flow and driving conditions, and all environmental data are also transmitted to the control center in real time for comprehensive analysis of traffic conditions and environmental conditions, and to provide information support for management decision-making.

[0019] Step S200, obtaining a traffic adaptive analysis model through a traffic scheduling platform, the traffic adaptive analysis model being composed of a traffic anomaly detection sub-model and a traffic congestion prediction sub-model. Specifically, the traffic scheduling platform first integrates massive historical traffic data from multiple sources such as a traffic management department database and a road sensor network, and pre-processes the data such as cleaning and denoising to lay a foundation for model acquisition. Then, the traffic anomaly detection sub-model is obtained, and the normal and abnormal traffic condition sample sets are divided from the pre-processed data by using a combination of rules and machine learning, and are screened and marked by using indicators such as vehicle flow, and are trained by using a feedforward neural network, and are iteratively adjusted and optimized by using a test set to evaluate, and the model structure is adjusted until the model meets the requirements. Then, the traffic congestion prediction sub-model is obtained, and the time series data of indicators such as vehicle flow are smoothed, and time windows are divided and marked, and are trained by using a recurrent neural network and its variants, and are adjusted by minimizing the loss function, and the error indicators are evaluated by using a validation set, and the model hyperparameters are improved accordingly, and the model is determined after meeting the requirements. Finally, the two sub-models are combined, the input and output integration mode is determined to make them cooperate, the anomaly detection result helps the congestion prediction, and the congestion prediction result also optimizes the anomaly detection, thereby constructing a traffic adaptive analysis model to provide support for traffic signal light control and management decision-making.

[0020] In a possible implementation, the traffic adaptive analysis model is obtained by the traffic scheduling platform, the traffic adaptive analysis model is composed of a traffic anomaly detection sub-model and a traffic congestion prediction sub-model, and step S200 further includes step S210 of obtaining a historical traffic operation data set by the traffic scheduling platform. Specifically, the traffic scheduling platform has strong data collection and integration capabilities, it establishes real-time communication connection with traffic sensors (such as geomagnetic sensors, cameras, microwave radars and the like) distributed in various road nodes of the city, continuously collects traffic operation data of different road sections and intersections in a long time span (covering various time periods such as different seasons, weekdays and holidays and the like for several years), including detailed records of parameters such as traffic flow, vehicle speed, vehicle occupancy rate and vehicle headway. At the same time, the platform also accesses the database of the traffic management department, obtains related data such as historical traffic accident information, road construction records, traffic control measure implementation conditions, and the like, carries out preprocessing operations such as cleaning, denoising and format unification on these data of different sources and different formats, removes invalid or erroneous data points, fills in missing values, so that the data has completeness and accuracy, and finally integrates to form a comprehensive, systematic and structured historical traffic operation data set, providing a rich and reliable data basis for subsequent model training.

[0021] In step S220, a feedforward neural network is used for sample identification training on the historical traffic operation data set to generate a traffic anomaly detection sub-model. Specifically, from the historical traffic operation data set, rules are set according to professional knowledge and experience in the field of traffic, combined with data analysis techniques, to extract feature vectors that can reflect normal and abnormal states of traffic, for example, sudden and significant decrease in traffic volume, sharp decrease in vehicle speed and rapid increase in vehicle occupancy, etc. are taken as the characteristic performance of abnormal traffic, and the corresponding data set is marked as an abnormal sample; while the data set with stable traffic volume, vehicle speed and vehicle occupancy within the normal range is marked as a normal sample, thereby constructing a normal traffic condition sample set and an abnormal traffic condition sample set. A feedforward neural network architecture is constructed, which includes an input layer, one or more hidden layers and an output layer. The input layer receives the traffic data feature vectors after preprocessing and feature extraction, the hidden layer performs complex nonlinear transformation and feature learning on the input data through the weighted connection between neurons, and the output layer outputs the classification result indicating whether the traffic condition is abnormal (usually 0 for normal and 1 for abnormal). During the training process, a large number of labeled sample data are input into the feedforward neural network in batches, based on the back propagation algorithm, the weights and bias parameters of the network are continuously adjusted by minimizing the cross-entropy loss function between the predicted results and the true labels, so that the network gradually learns the pattern that can accurately distinguish normal and abnormal traffic conditions. At the same time, techniques such as Dropout regularization, Early Stopping, etc. are used to prevent overfitting and ensure the generalization ability of the model. After multiple iterations of training and optimization, when the accuracy, recall rate, F1 value and other evaluation indicators of the model on the independent validation data set reach a satisfactory performance level, the traffic anomaly detection sub-model is finally determined and generated, which can detect and warn abnormal conditions in traffic operation in real time and accurately, providing strong support for traffic management departments to take timely response measures.

[0022] Step S230, the historical traffic operation data set is trained by using a recurrent neural network to obtain a traffic congestion prediction sub-model. Specifically, for the historical traffic operation data set, first, the time series data of key traffic parameters such as traffic flow and speed are subjected to stationarity test and preprocessing operation, such as using difference method, seasonal decomposition method and the like to convert non-stationary sequence into stationary sequence, so as to better perform modeling analysis. Then, according to the formation and development law of traffic congestion, the time series data is divided into different time windows (such as 5 minutes, 10 minutes, etc.), and the data in each time window is marked, for example, the time window with continuously rising traffic flow, continuously falling speed and gradually increasing vehicle occupancy is marked as a congestion trend sample, and the time window with relatively stable traffic flow and speed is marked as a smooth trend sample, thereby constructing a traffic trend condition sample set, wherein each sample not only contains traffic data features in the current time window, but also contains historical traffic data features in the previous time windows, so as to reflect the time dependence and trend of traffic congestion. A recurrent neural network (RNN) and its variants (such as long short-term memory network (LSTM) or gated recurrent unit (GRU)) are used to construct a prediction model architecture. Due to its unique structural design, the recurrent neural network can process data with time series characteristics, and through the introduction of recurrent connections between neurons, it can remember and utilize historical information, thereby predicting future traffic congestion conditions. The constructed traffic trend condition samples are sequentially input into the recurrent neural network, and the network learns the change pattern and trend of traffic congestion according to the input historical traffic data features, predicts the traffic congestion degree (such as light congestion, medium congestion, heavy congestion, etc.) or the probability of congestion occurrence in a future period of time (such as the next 15 minutes, 30 minutes, etc.), and outputs the prediction result. In the training process, the weights and bias parameters of the network are adjusted by minimizing the mean square error loss function between the prediction result and the true label to optimize the model performance. In order to prevent overfitting and improve the stability of the model, adaptive learning rate adjustment strategies (such as Adam optimization algorithm), regularization techniques (such as L1 and L2 regularization) and the like are used, and the model hyperparameters (such as the number of hidden layers, the number of neurons, the time step, etc.) are adjusted by evaluating the mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE) and the like on an independent validation data set. After repeated training and validation, when the error indicators of the model on the validation set meet the accuracy requirements of practical applications, the traffic congestion prediction sub-model is finally determined and obtained, which can predict future traffic congestion conditions in advance according to the current traffic operation conditions and historical trends, provide an important basis for the traffic management department to develop reasonable traffic relief and signal control strategies, help to alleviate traffic congestion and improve traffic operation efficiency.

[0023] Step S240, the traffic anomaly detection sub-model and the traffic congestion prediction sub-model are combined to obtain the traffic adaptive analysis model. Specifically, when combining the two sub-models, first, the input-output relationship and the cooperative working mechanism between them are determined. The output result of the traffic anomaly detection sub-model (whether the traffic is abnormal) is an important input feature of the traffic congestion prediction sub-model. When detecting traffic abnormal situations (such as traffic accidents, temporary road control, etc.), the traffic congestion prediction sub-model can adjust its prediction strategy and parameters in a timely manner according to the type and location of the abnormality, and more accurately predict the development trend and influence range of traffic congestion under abnormal conditions. For example, if a traffic accident occurs on a certain road section, the traffic congestion prediction sub-model will focus on the traffic data changes of the road section and its surrounding area, combine the congestion evolution law under similar accident scenarios in the historical data, and make more accurate prediction of the future congestion situation. The output result of the traffic congestion prediction sub-model (the degree and trend of traffic congestion) is also fed back to the traffic anomaly detection sub-model as input information, helping it to further optimize the threshold and rules of anomaly detection, and improve the accuracy and timeliness of anomaly detection. For example, if it is predicted that there will be serious congestion in a certain area, the traffic anomaly detection sub-model will accordingly increase the sensitivity to abnormal changes in the traffic data of the area, and discover potential abnormal factors that may cause more serious congestion in advance, such as slight abnormal fluctuations in vehicle speed, sudden concentration of traffic flow, etc., so as to issue an early warning of the abnormality. By establishing such a close information interaction and cooperative working mechanism, the traffic anomaly detection sub-model and the traffic congestion prediction sub-model are organically integrated to form a complete traffic adaptive analysis model, which can comprehensively consider various complex factors and dynamic changes in traffic operation, and make real-time, comprehensive and accurate analysis and prediction of traffic conditions, providing efficient and reliable support for intelligent control of traffic signals and traffic management decisions, effectively improving the operation efficiency and intelligent level of urban traffic system, better coping with various traffic problems such as traffic congestion, accidents, etc., and ensuring the smoothness and safety of urban traffic.

[0024] In a possible implementation, the historical traffic operation data set is subjected to sample identification training by using a feedforward neural network to generate a traffic anomaly detection sub-model, and step S220 further includes step S221 of identifying samples in the historical traffic operation data set to obtain a normal traffic condition sample set and an abnormal traffic condition sample set. Specifically, the historical traffic operation data set is analyzed in depth, and scientific sample identification rules are formulated based on professional knowledge and long-term experience in traffic engineering. For normal traffic condition samples, the traffic volume should be stably and reasonably distributed within the road design capacity, the vehicle speed should be within the normal driving speed range of the road section, the vehicle queue length should be short or no queue, the driving trajectory should be smooth and compliant, and the like. For example, during a non-peak period on a city trunk road, the traffic volume is 3000-4000 vehicles per hour, the average vehicle speed is 50-60 km / h, the lanes are smooth and there is no frequent starting and stopping or accumulation, and the like. The data time period meeting these conditions is marked as a normal sample, and the normal sample set is formed. The abnormal traffic condition samples are identified by detecting significant changes and abnormal features in the traffic data, such as near-zero traffic volume in a certain lane due to a traffic accident, sharp increase in traffic volume due to the dispersal of a large event, vehicle speed reduced to less than half of the normal speed due to road construction, and the like. In combination with event records, the corresponding time period data is marked as an abnormal sample, and the abnormal sample set is constructed. The accurate construction of the two sample sets provides key basic data for subsequent model training, and ensures that the model can master the essential difference between normal and abnormal traffic states.

[0025] In step S222, the normal traffic condition sample set and the abnormal traffic condition sample set are subjected to detection training by using a feedforward neural network to obtain an initial traffic anomaly detection model. Specifically, a feedforward neural network architecture is constructed, including an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is determined according to the selected traffic feature dimensions, such as traffic volume, vehicle speed, vehicle occupancy, and vehicle queue length. For example, if the four features are selected, four nodes are set. The number of layers (firstly, 2-3 layers are tried) and the number of nodes (for example, 10-20) in each layer of the hidden layer are determined by trial optimization, the layers are connected by a weight matrix, and the neurons use a ReLU function to introduce a non-linear factor to enhance the expression. The output layer has one node, which is activated by a Sigmoid function, and outputs a value of 0-1 representing the traffic anomaly probability. The normal and abnormal traffic condition sample sets are divided into a training set and a validation set at a ratio of 80:20. The sample data in the training set is preprocessed by normalization and the like, and then input into the network in batches. During iterative training, the network obtains a prediction result by forward propagation of the input sample features. Compared with the true label of the sample (0 for normal and 1 for abnormal), the loss value is calculated by a binary cross-entropy loss function according to the difference. The network weights and biases are adjusted according to the gradient information by back propagation. After 100-500 iterations (depending on the data size and convergence), the training is stopped when the performance indicators on the validation set meet the standards, and the initial traffic anomaly detection model is obtained.

[0026] Step S223, loss calculation is performed on the initial traffic anomaly detection model to obtain traffic loss data. Specifically, during the training process, the initial traffic anomaly detection model is continuously trained and optimized using the training subset, and at the end of each training cycle, the model is verified using the validation subset, and the loss value of the model on the validation subset is calculated as part of the traffic loss data. The loss calculation uses the same loss function as in the training process (such as the binary cross-entropy loss function), and for each sample in the validation subset, the predicted output of the model and the true label are substituted into the loss function formula to calculate the loss value of the sample. Then, the average loss value of all samples in the validation subset is calculated to obtain the average loss value of the model on the validation subset in the current training cycle. The change of these average loss values with the training cycle is recorded to form a sequence of traffic loss data. These loss data reflect the deviation between the predicted results of the model on the validation set and the true labels. By analyzing the trend of the loss data, the training state and performance of the model can be understood. For example, if the loss data shows a continuous downward trend, it means that the model is continuously optimizing and improving. If the loss data tends to be stable or starts to rise at a certain stage, it may indicate that the model has overfitting or has fallen into a local optimal solution, and appropriate measures need to be taken to adjust.

[0027] Step S224, the initial traffic anomaly detection model is optimized based on the traffic loss data to generate the traffic anomaly detection sub-model. Specifically, the initial traffic anomaly detection model is optimized based on the obtained traffic loss data. When the loss data shows overfitting characteristics, i.e., the validation set loss value first decreases and then increases, and the training set accuracy is high while the validation set accuracy is low, regularization techniques such as L1 regularization (adding a weight vector L1 norm term to the loss function to make some weights tend to zero to reduce model complexity) or L2 regularization (adding a weight vector L2 norm term to make the weight distribution uniform to prevent overfitting) are used, and the diversity of training data is appropriately increased, such as random transformation of original data, expansion of abnormal sample size, etc. to improve the generalization ability of the model. If the loss data shows that the model converges slowly or falls into a local optimum, such as a very slow loss value and multiple rounds of training without reaching the expected value, the optimization algorithm parameters can be adjusted, such as using the Adam optimization algorithm (dynamically adjusting the learning rate based on the first and second moment estimates of the model parameter gradient to promote convergence), or fine-tuning the network structure, such as increasing or decreasing the number of hidden layers, adjusting the number of hidden layer nodes, etc. Then, re-training and verification are performed, the change of the loss data is observed, and the optimization strategy is continuously adjusted until the accuracy of the model on the validation set is above 90%, the recall rate is above 80%, and the F1 value is above 0.85 (the specific value depends on the actual requirements). At this time, the model is the optimized traffic anomaly detection sub-model, which can accurately and reliably detect traffic anomalies, effectively assist traffic management decision-making, and ensure the safety and smoothness of the traffic system.

[0028] In a possible implementation, the historical traffic operation data set is subjected to sample identification training by using a recurrent neural network to obtain a traffic congestion prediction sub-model, and step S230 further includes step S231, time series processing and sample identification are performed on the historical traffic operation data set to obtain a traffic trend condition sample set. Specifically, stationarity test is performed on the historical traffic operation data set, and a method such as Augmented_Dickey-Fuller (ADF) test is used to determine whether the time series data of key traffic parameters such as traffic volume, vehicle speed, and vehicle occupancy is stationary. If the data is not stationary, a suitable transformation method is used to make it stationary, for example, difference method, difference calculation is performed on the data of adjacent time points to eliminate the trend and seasonal factors in the data, so that the mean and variance of the sequence no longer change with time. For data with obvious seasonal characteristics, seasonal difference or seasonal decomposition techniques can also be used to decompose the data into trend, seasonal and residual terms, and each term is analyzed and processed to better capture the internal laws and characteristics of the data. According to the formation mechanism of traffic congestion and the laws in the historical data, the processed time series data is divided into different time windows, for example, 10 minutes as a window unit. Then, sample identification is performed by analyzing the change trend and characteristics of the traffic data in each time window. If the traffic volume shows a continuous upward trend, the vehicle speed gradually decreases, and the vehicle occupancy rate continuously increases in a time window, and combined with the traffic event information (such as whether there is a large-scale activity, road construction, etc.) in the surrounding area during this period, it is determined that the traffic condition in this window has a trend of developing towards congestion, and the data in this window is marked as a congestion trend sample; otherwise, if the traffic volume is relatively stable, the vehicle speed remains in the normal range, and the vehicle occupancy rate changes little, it is marked as a smooth trend sample. In this way, the entire historical data set is traversed and marked, and finally a traffic trend condition sample set is formed, each sample of which contains not only the traffic data characteristics in the current time window, but also the historical traffic data characteristics of the previous time windows, so as to fully reflect the time dependence and trend of traffic congestion, and provide a representative and predictive data basis for subsequent model training.

[0029] Step S232, using a recurrent neural network to perform prediction training on the traffic trend condition sample set to obtain an initial traffic congestion prediction model. Specifically, a recurrent neural network (RNN) architecture is built, including an input layer, a hidden layer, and an output layer. The hidden layer has a memory function of a recurrent neuron to retain transmitted information over a time series. The number of nodes in the input layer is determined according to the selected traffic data feature dimension and the number of historical time windows, for example, if three features of traffic flow, speed, and vehicle occupancy rate are selected and the previous three time windows are considered, then the number of nodes in the input layer is nine. The number of layers (initially set to 1-2 layers) and the number of nodes in each layer (about 10-20) of the hidden layer can be experimentally optimized. The layers are connected by a weight matrix, and the neuron activation function can be selected as a tanh or ReLU function to enhance expression and learning ability. The number of nodes in the output layer depends on the prediction target. If the congestion degree is divided into three categories, then three nodes are set to output probability values using a softmax activation function. If the congestion is predicted, then one node is set to output a congestion probability using a Sigmoid activation function. Then, the traffic trend condition sample set is divided into a training set and a validation set at a ratio of 70%:30%. The sample data in the training set is preprocessed by normalization and then sequentially input into the RNN in sequence. At each time step, the network updates the current hidden layer output based on the current input features and the output state of the previous hidden layer using a recurrent neuron update mechanism. After the sequence input is completed, the traffic congestion prediction result is obtained. The prediction result is compared with the true label, and the loss value is calculated using a cross-entropy (multi-classification) or mean square error (continuous or binary classification) loss function according to the difference. The gradient information is used to adjust the network weight and bias parameters using a backpropagation algorithm. After 200-100 iterations (depending on the data size, network complexity, and convergence), the training is stopped when the performance indicators of the validation set meet the standards, and the initial traffic congestion prediction model is obtained.

[0030] Step S233, loss calculation optimization is performed based on the initial traffic congestion prediction model to obtain the traffic congestion prediction sub-model. Specifically, during the training of the initial traffic congestion prediction model, the training subset is continuously used for training, and the validation subset is used for validation at the end of each training period. The loss value on the validation subset is calculated, the same loss function (such as cross-entropy or mean square error function) as during training is used, the predicted output of each sample and the true label are substituted into the formula to calculate the loss value, and then the average loss value of each period is obtained. Record the values to form a loss data sequence as the period changes. Analyzing its trend can understand the model training state and performance. If it continues to decline, it needs to be optimized and improved. If it is stable or rising, there may be overfitting or falling into a local optimal solution, which needs to be adjusted. According to the loss calculation result, targeted optimization strategies are adopted for the model. If there is overfitting, such as the validation set loss value first decreases and then increases, and the training set accuracy is high while the validation set accuracy is low. L1 or L2 regularization can be used to reduce complexity and balance weights by adding the corresponding weight vector norm item. At the same time, increase the diversity of training data, such as randomly transforming the original data, expanding the sample size of different scenarios, and enhancing the generalization ability. If the convergence is slow or falls into a local optimum, such as the loss value decreases slowly and does not reach a satisfactory level, the optimization algorithm parameters can be adjusted, such as using the Adam algorithm to dynamically adjust the learning rate, or fine-tuning the network structure, such as increasing or decreasing the number of hidden layers, adjusting the number of nodes, changing the type of loop unit, etc. Then retrain and validate, observe the change of loss data and continuously adjust until the model accuracy on the validation set is above 85%, the root mean square error is less than 0.2, and the mean absolute error is less than 0.15 (depending on actual requirements). The optimized traffic congestion prediction sub-model is obtained to provide decision support for traffic management departments, relieve congestion, and improve the intelligent level of traffic operation and management.

[0031] Step S300, based on the traffic adaptive analysis model, traffic condition analysis is performed on the traffic operation data information and traffic environment data information to obtain traffic operation congestion parameters. Specifically, the collected traffic operation data (such as traffic flow, vehicle speed, vehicle occupancy rate, etc.) and traffic environment data (such as weather, lighting, road construction conditions, etc.) are preprocessed, error data is checked and corrected, and operation data is normalized. At the same time, key features are extracted from the operation data, and environmental data is converted into a form that the model can process. Then, the processed data is input into the traffic adaptive analysis model, the anomaly detection sub-model judges the traffic anomaly situation, determines the location, time and influence range of the abnormal event, and the congestion prediction sub-model combines the operation and environment data and historical trends to predict the congestion degree and development trend of each road section and intersection within a certain time in the future. Finally, traffic operation congestion parameters are obtained based on the model output, including identifying the congestion road section and intersection, quantifying the congestion degree using average speed reduction ratio, vehicle queue length, traffic flow saturation, etc., and estimating the congestion duration to provide key data support for subsequent traffic management measures to govern congestion and regulate traffic flow.

[0032] Step S400, according to the traffic operation congestion parameter, start traffic emergency response mechanism. Specifically, after obtaining the traffic operation congestion parameter, first evaluate the grading, according to the average speed reduction ratio, vehicle queue length, vehicle flow saturation and other indicators, according to the preset standard, the congestion condition is divided into different levels. Then start the emergency response measures and work together, through the traffic broadcast, intelligent APP and electronic display screen to release the congestion information, provide the detour suggestion for the driver; according to the congestion situation, deploy police to the key congestion point, the police manually command the optimization of traffic order at the intersection; implement the public transport priority strategy, set the special road or priority signal for it, at the same time, guarantee the green channel for emergency vehicles; the traffic management department also cooperates with the road maintenance, meteorological and other departments, such as construction congestion, adjust the plan, prepare in advance in bad weather, so as to relieve the congestion, guarantee the smooth traffic and emergency disposal ability.

[0033] Step S500, based on the traffic emergency response mechanism, the traffic operation congestion parameter is analyzed, the signal light control strategy parameter is determined. Specifically, the traffic emergency response mechanism integrates the traffic operation congestion parameter and the road network information, presents the congestion distribution by GIS technology, evaluates the congestion cause combined with the traffic environment data, provides the scene basis for the signal light strategy development. Then build the signal light control strategy solution space, clear the value range and combination mode of each signal light parameter, such as the reasonable range of green light time, phase sequence, cycle time, etc. Then design the fitness function with the vehicle average delay time, parking times, road traffic capacity utilization rate, queue length change rate and other factors as the evaluation factors, form the comprehensive index by giving the weight to each factor to measure the strategy advantage and disadvantage. Finally, use the genetic algorithm and other optimization algorithms to optimize in the solution space, through the selection, crossover, mutation and other operations iteration, find the optimal parameter combination, that is, the determined traffic signal light control strategy parameter, including the time length and phase sequence of each intersection signal light, etc., also will adjust the parameter according to the real-time traffic change, guarantee the smooth traffic.

[0034] In a possible implementation, the traffic emergency response mechanism analyzes the traffic operation congestion parameters based on a signal light control strategy, determines traffic signal light control strategy parameters, and step S500 further includes step S510 of constructing a signal light control strategy solution space based on the traffic emergency response mechanism analyzing the traffic operation congestion parameters based on a signal light control strategy. Specifically, after the traffic emergency response mechanism is started, the traffic operation congestion parameters are first analyzed in depth, and these parameters cover key information such as traffic flow, vehicle speed, vehicle queue length, congestion road segment position and range, and the like. In combination with the topology of the traffic network, including the number of intersections, lane distribution, road connection relationship and the like, the control strategy of the signal light is comprehensively analyzed. For each signal light intersection that needs to be regulated and controlled, the adjustable control parameter range thereof is determined, for example, the value range of the green light duration can be preliminarily set according to historical traffic data and road traffic capacity, and generally speaking, there can be a basic green light duration range in a non-congestion period, and in a congestion situation, this range can be appropriately extended or shortened; the red light duration is also adjusted accordingly to ensure that the traffic flow in each direction can be reasonably allocated time; the phase setting of the signal light is also considered, for example, according to the proportion of traffic flow in different directions in different time periods, it is determined whether to use a simple two-phase (such as east-west straight and left turn are released at the same time, and the same is true for south-north) or a more complex three-phase (an additional left turn phase is added) or even more phase control mode. Through comprehensive consideration and range determination of these parameters, a solution space containing multiple possible signal light control strategy combinations is constructed, providing a basis for subsequent optimization decisions.

[0035] Step S520, the signal lamp control target is evaluated and analyzed, and the signal lamp control fitness function is constructed. Specifically, the target of signal lamp control is determined, mainly focusing on relieving traffic congestion, improving road traffic efficiency, reducing vehicle waiting time and parking times, etc. In order to quantitatively evaluate the realization degree of these targets, the signal lamp control fitness function is constructed. For the target of relieving traffic congestion, the vehicle queue length reduction rate of the congestion road section can be included as an important index in the fitness function, and the corresponding score weight is given by comparing the change of the queue length before and after adopting different signal lamp control strategies; for improving road traffic efficiency, the total number of vehicles passing through the intersection per unit time can be used as a measure index, that is, the total number of vehicles passing through the intersection in a certain time interval is counted, and the score is given according to the closeness to the road design traffic capacity; for reducing vehicle waiting time and parking times, the average waiting time and average parking times of vehicles at the intersection are calculated for evaluation, the shorter the waiting time and the fewer the parking times, the higher the fitness function score. At the same time, considering the importance difference between different targets, the weight distribution of each index in the fitness function can be determined through expert experience or data analysis, for example, if the congestion problem is more prominent in the current traffic condition, the weight of the vehicle queue length reduction rate can be appropriately increased, so as to construct a fitness function which can comprehensively and accurately reflect the advantages and disadvantages of the signal lamp control strategy, so as to effectively evaluate and compare various strategies in the signal lamp control strategy solution space in the subsequent.

[0036] Step S530, control optimization analysis of the signal light control strategy solution space is carried out by using the signal light control fitness function, and the traffic signal light control strategy parameters are determined. Specifically, optimization algorithms such as genetic algorithm, simulated annealing algorithm, particle swarm optimization algorithm, etc. are used to search and iteratively optimize in the constructed signal light control strategy solution space. Taking the genetic algorithm as an example, a group of initial signal light control strategy individuals are first randomly generated, each individual representing a specific combination of signal light control parameters (such as green light duration, red light duration, phase setting, etc. Specific numerical values), and these individuals are substituted into the signal light control fitness function to calculate their respective fitness values. Then, according to the size of the fitness value, the selection operation is adopted to select the individuals with higher fitness as the parents, and the crossover operation is carried out to simulate the gene recombination process in biological genetics, generating new offspring individuals, and at the same time, a certain probability of mutation operation is carried out on part of the individuals to introduce new gene characteristics, avoiding the algorithm falling into local optimal solution. After multiple rounds of selection, crossover and mutation operations, the individuals in the population are constantly updated, making them gradually evolve towards a higher fitness direction. In the iteration process, the change of the fitness function value and the convergence trend of the population are continuously monitored, and when the preset stopping condition (such as reaching the maximum number of iterations or the change of the fitness function value being less than a certain threshold) is met, the iteration is stopped, and at this time, the signal light control strategy parameter combination corresponding to the individual with the highest fitness in the population is determined as the traffic signal light control strategy parameters. These parameters will be transmitted to the actual traffic signal light control system to realize precise regulation and control of the signal light to cope with the current traffic congestion, improve the traffic efficiency and smoothness, and with the dynamic changes of the traffic conditions, the above process can be repeated to adjust and optimize the signal light control strategy parameters in real time, ensuring the effectiveness and adaptability of traffic management.

[0037] In a possible implementation, the signal light control strategy solution space is controlled and optimized by using the signal light control fitness function, and the traffic signal light control strategy parameters are determined, and step S530 further includes step S531 of selecting a plurality of signal light control strategy parameters in the signal light control strategy solution space. Specifically, the signal light control strategy solution space is determined according to the analysis of the traffic running congestion parameters in the previous stage and the traffic network information, and covers a set of possible control parameter value ranges of the green light duration range, the red light duration range, the signal light phase combination mode, the cycle duration range and the like of each signal light intersection. In this solution space, a plurality of signal light control strategy parameter combinations are selected by using random sampling or a specific sampling strategy as an initial candidate scheme set. For example, for a small traffic network with four intersections and three phase setting possibilities for each intersection, a green light duration in the range of 20-60 seconds, a corresponding red light duration, a cycle duration in the range of 80-120 seconds, 10-20 different parameter combinations are selected by random generation, each parameter combination explicitly indicates the green light start time, the green light duration, the red light duration and the phase switching sequence of each intersection in a signal cycle, and forms the initial plurality of signal light control strategy parameters. These parameter combinations will enter the subsequent evaluation and optimization process to provide diverse starting points for finding the optimal signal light control strategy.

[0038] Step S532, the signal light control fitness function is used to evaluate the fitness of the plurality of signal light control strategy parameters, and a plurality of control strategy parameter fitnesses are obtained. Specifically, the plurality of signal light control strategy parameters selected in the previous step are respectively substituted into the pre-constructed signal light control fitness function for evaluation. The fitness function comprehensively considers a plurality of key indicators such as the traffic congestion relief degree, the road traffic efficiency improvement situation, the vehicle average waiting time reduction situation and the like. For example, for the traffic congestion relief degree, the change of the vehicle queue length of the congestion section before and after a certain set of signal light control strategy parameters is adopted is quantitatively evaluated, if the queue length is significantly reduced, a higher score is given in this indicator; for the road traffic efficiency, the ratio of the total number of vehicles passing through each intersection in a certain period of time to the theoretical traffic capacity of the road is calculated, the higher the ratio, the higher the score; the vehicle average waiting time is calculated through the data recorded by the vehicle detection equipment of each intersection, the shorter the waiting time, the higher the score. According to the weight distribution of these indicators in the fitness function (the weight can be determined by regression analysis of historical data or expert experience), the fitness value corresponding to each signal light control strategy parameter combination is calculated, and a plurality of control strategy parameter fitnesses are obtained. These fitness values directly reflect the advantages and disadvantages of each set of signal light control strategy parameters under the current traffic condition, and provide an explicit comparison basis for subsequent iterative optimization.

[0039] Step S533, based on the plurality of control strategy parameters fitness, iterative optimization is carried out in the signal lamp control strategy solution space, and the traffic signal lamp control strategy parameters are determined. Specifically, according to the obtained plurality of control strategy parameters fitness, an optimization algorithm (such as selection, crossover, mutation operation in genetic algorithm, or cooling search strategy in simulated annealing algorithm, and particle speed and position updating mechanism in particle swarm optimization algorithm, etc.) is used to carry out iterative optimization in the signal lamp control strategy solution space. Taking genetic algorithm as an example, first of all, selection operation is carried out according to the fitness value, and signal lamp control strategy parameter combination with higher fitness is selected as parent individual, and parent individual has higher "excellent gene" proportion, and is more likely to produce excellent offspring individual. Then, the parent individual is subjected to crossover operation, the gene exchange process in biological genetics is simulated, part of the control parameters of different parent individuals are combined to generate new offspring individuals, and the diversity and search range of the solution space are increased. At the same time, a certain probability is used to carry out mutation operation on part of the individuals, and the values of some control parameters are randomly changed, so as to prevent the algorithm from falling into local optimal solution and explore other potential high-quality areas in the solution space. After each iteration, the new generated individuals (i.e. new signal lamp control strategy parameter combination) are re-evaluated by using the signal lamp control fitness function, and the selection, crossover and mutation operations are continued according to the evaluation results, and the cycle is repeated. With the increase of the number of iterations, the individuals in the solution space continuously evolve towards the direction with higher fitness, and when the preset stopping condition (such as reaching the maximum number of iterations, the change of fitness function value is less than a certain threshold or the optimal solution does not improve after continuous multiple iterations, etc.) is met, the signal lamp control strategy parameter combination corresponding to the individual with the highest fitness in the solution space at this time is the final determined traffic signal lamp control strategy parameter. These parameters will be applied to the actual traffic signal lamp control system, realize the precise regulation and control of traffic signal lamp, effectively alleviate traffic congestion, improve traffic operation efficiency, and with the real-time change of traffic condition, the optimization process can be started again to dynamically adjust and optimize the signal lamp control strategy parameters, so as to maintain the good running state of the traffic system.

[0040] Step S600, based on the traffic signal control strategy parameter of the target traffic area signal lamp adaptive regulation and control. Specifically, first of all, the determined traffic signal control strategy parameter is analyzed, and is connected with the target traffic area signal lamp control system, guarantees that it can receive and adjust according to the parameter, and upgrades the software and hardware if necessary. Then continuously monitor the target area traffic conditions, use geomagnetic sensor, camera and other collection of traffic flow, speed, queue length and other operating data and weather, construction, special events and other environmental data, transmit to the management center control system, by which the traffic condition category is judged. Then according to the monitoring feedback data, the adaptive adjustment mechanism is started, when the traffic flow increases, the queue length and the congestion intensify, the signal lamp parameters are automatically adjusted, such as increasing or decreasing the green light time of each direction, flexibly changing the phase sequence, and the cycle time, phase time and yellow light time are also adjusted under special circumstances (such as bad weather, sudden accident), guiding the vehicle to bypass, realizing the dynamic matching and optimization of signal lamp and traffic condition, and improving the traffic operation efficiency and emergency response ability.

[0041] In a possible implementation, based on the traffic signal control strategy parameter of the target traffic area signal lamp adaptive regulation and control, step S600 further comprises step S610, based on the traffic signal control strategy parameter of the target traffic area signal lamp control monitoring, obtaining signal lamp control feedback parameter. Specifically, when the traffic signal control system regulates the signal lamp of the target traffic area according to the established traffic signal control strategy parameter, a comprehensive monitoring mechanism is started at the same time. High-precision vehicle detection equipment such as geomagnetic sensor, video camera is set at each signal lamp intersection, real-time collection of traffic flow, speed, vehicle queue length and other key traffic data through the intersection, at the same time, the phase switching time, green light time, red light time and yellow light time and other running state information of the signal lamp are recorded. Use communication network to transmit these data back to the monitoring platform of traffic management center quickly, through data processing and analysis technology, integrate and refine the collected mass data, extract the key indicators reflecting the signal lamp control effect as the signal lamp control feedback parameter. For example, calculate the average delay time of vehicle, that is, the total time of all vehicles waiting for signal lamp at intersection divided by the total number of vehicles; calculate the saturation of each intersection, which is measured by the ratio of actual traffic flow to road design capacity; analyze the change trend of vehicle queue length, observe the increase and decrease of queue length before and after the implementation of signal lamp control strategy, etc. These signal lamp control feedback parameters will provide objective and accurate data support for subsequent strategy optimization, help traffic management personnel to deeply understand the implementation effect and existing problems of current signal lamp control strategy, so as to make timely adjustment and improvement.

[0042] Step S620, based on the signal light control feedback parameter, the traffic signal light control strategy parameter is adjusted, and the traffic signal light optimization control strategy parameter is obtained. Specifically, according to the signal light control feedback parameter obtained in the previous step, the intelligent optimization algorithm is used to optimize and adjust the existing traffic signal light control strategy parameter. First, an optimization objective function based on the signal light control feedback parameter is established, for example, the minimum average vehicle delay time, the balanced and near optimal value of the intersection saturation, the stable or reduced vehicle queue length, etc. The function is included, and different weights are given to each target according to the actual traffic management demand. Then, optimization tools such as genetic algorithm, simulated annealing algorithm, particle swarm optimization algorithm, etc. are used to search and iteratively optimize within the preset value range of the traffic signal light control strategy parameter. Taking genetic algorithm as an example, the current traffic signal light control strategy parameter is encoded as an individual of the initial population, and the fitness value of each individual is calculated according to the optimization objective function. Through genetic operations such as selection, crossover and mutation, new individuals are constantly generated, and the optimal solution is gradually approached. In the iteration process, the weight coefficients in the optimization objective function are constantly adjusted to balance the relationship between different optimization objectives, for example, in the period when traffic congestion is more serious, the weight of the vehicle queue length reduction target is appropriately increased to preferentially alleviate the congestion; while in the period when the traffic flow is relatively stable, more attention is paid to the reduction of the average vehicle delay time and the balance of the intersection saturation. After multiple iterations of optimization, when the preset stopping condition (such as reaching the maximum number of iterations, the change of the optimization objective function value being less than a certain threshold, etc.) is met, the traffic signal light control strategy parameter corresponding to the optimal individual obtained is the traffic signal light optimization control strategy parameter. Apply these optimized parameters to the signal light control system of the target traffic area to realize the adaptive optimization and regulation of the signal light, so that the control of the signal light can be dynamically adjusted according to the changes of the real-time traffic conditions, continuously improve the traffic operation efficiency, reduce the vehicle waiting time and congestion degree, improve the service level and operation stability of the entire traffic network, and repeat the optimization and adjustment process as the traffic conditions continue to evolve, to ensure that the signal light control strategy always remains in the best state.

[0043] The embodiment of the present application adopts the method of deploying sensor groups at the target traffic area signal lights to collect traffic operation and environmental data. Through the traffic scheduling platform, an adaptive analysis model containing anomaly detection and congestion prediction is obtained to analyze the traffic conditions and obtain congestion parameters. According to these parameters, an emergency response mechanism is started, and then the signal light control strategy parameters are analyzed and determined, finally realizing the adaptive regulation of the traffic signal light, achieving the technical effects of enhancing the real-time response capability of the traffic signal light system and effectively predicting and responding to traffic anomalies and congestion through real-time traffic data analysis and intelligent prediction.

[0044] In the foregoing, reference is made to Figure 1 The traffic signal lamp intelligent control method according to the embodiment of the application is described in detail. Next, the traffic signal lamp intelligent control system according to the embodiment of the application will be described with reference to Figure 2 The traffic signal lamp intelligent control system according to the embodiment of the application is described in detail. Next, the traffic signal lamp intelligent control system according to the embodiment of the application will be described with reference to

[0045] The traffic signal lamp intelligent control system according to the embodiment of the application is used to solve the technical problems of insufficient real-time response capability, inability to effectively predict traffic abnormalities and congestion in the existing traffic signal lamp control system. Through real-time traffic data analysis and intelligent prediction, the technical effects of enhancing the real-time response capability of the traffic signal lamp system and effectively predicting and responding to traffic abnormalities and congestion are achieved. The traffic signal lamp intelligent control system comprises a sensor group deployment module 10, a traffic adaptive analysis model acquisition module 20, a traffic condition analysis module 30, a traffic emergency response mechanism starting module 40, a signal lamp control strategy analysis module 50, and an adaptive regulation module 60.

[0046] The sensor group deployment module 10 is used to deploy a sensor group on a target traffic area signal lamp, and to collect traffic operation data information and traffic environment data information through the sensor group.

[0047] The traffic adaptive analysis model acquisition module 20 is used to acquire a traffic adaptive analysis model through a traffic scheduling platform, wherein the traffic adaptive analysis model comprises a traffic anomaly detection sub-model and a traffic congestion prediction sub-model.

[0048] The traffic condition analysis module 30 is used to analyze the traffic operation data information and the traffic environment data information based on the traffic adaptive analysis model, and to acquire traffic operation congestion parameters.

[0049] The traffic emergency response mechanism starting module 40 is used to start a traffic emergency response mechanism according to the traffic operation congestion parameters.

[0050] The signal lamp control strategy analysis module 50 is used to analyze the traffic operation congestion parameters based on the traffic emergency response mechanism, and to determine traffic signal lamp control strategy parameters.

[0051] The adaptive regulation module 60 is used to adaptively regulate the target traffic area signal lamp based on the traffic signal lamp control strategy parameters.

[0052] In the following, the specific configuration of the traffic adaptive analysis model obtaining module 20 will be described in detail. As described above, the traffic adaptive analysis model is obtained by the traffic scheduling platform, and the traffic adaptive analysis model is composed of a traffic anomaly detection sub-model and a traffic congestion prediction sub-model. The traffic adaptive analysis model obtaining module 20 further comprises: a historical traffic operation condition data set obtaining unit, configured to obtain a historical traffic operation condition data set through the traffic scheduling platform; a sample identification training unit, configured to perform sample identification training on the historical traffic operation condition data set by using a feedforward neural network, and generate a traffic anomaly detection sub-model; a traffic congestion prediction sub-model obtaining unit, configured to perform sample identification training on the historical traffic operation condition data set by using a recurrent neural network, and obtain a traffic congestion prediction sub-model; and a traffic adaptive analysis model obtaining unit, configured to combine the traffic anomaly detection sub-model and the traffic congestion prediction sub-model, and obtain the traffic adaptive analysis model.

[0053] In the above, the sample identification training on the historical traffic operation condition data set by using the feedforward neural network to generate the traffic anomaly detection sub-model further comprises: a traffic condition sample set obtaining sub-unit, configured to perform sample identification on the historical traffic operation condition data set, and obtain a normal traffic condition sample set and an abnormal traffic condition sample set; a detection training sub-unit, configured to perform detection training on the normal traffic condition sample set and the abnormal traffic condition sample set by using the feedforward neural network, and obtain an initial traffic anomaly detection model; a loss calculation sub-unit, configured to perform loss calculation on the initial traffic anomaly detection model, and obtain traffic loss data; and a model optimization sub-unit, configured to optimize the initial traffic anomaly detection model based on the traffic loss data, and generate the traffic anomaly detection sub-model.

[0054] The historical traffic operation condition dataset is trained by using a recurrent neural network for sample identification, and a traffic congestion prediction sub-model is obtained. The traffic congestion prediction sub-model acquisition unit further comprises: a traffic trend condition sample set acquisition sub-unit, configured to perform time series processing and sample identification on the historical traffic operation condition dataset to obtain a traffic trend condition sample set; an initial traffic congestion prediction model acquisition sub-unit, configured to perform prediction training on the traffic trend condition sample set by using a recurrent neural network to obtain an initial traffic congestion prediction model; and a loss calculation optimization sub-unit, configured to perform loss calculation optimization based on the initial traffic congestion prediction model to obtain the traffic congestion prediction sub-model.

[0055] The specific configuration of the signal light control strategy analysis module 50 will be described in detail below. As described above, the traffic signal light control strategy parameters are determined based on the signal light control strategy analysis of the traffic operation congestion parameters by the traffic emergency response mechanism. The signal light control strategy analysis module 50 further comprises: a signal light control strategy solution space construction unit, configured to construct a signal light control strategy solution space based on the signal light control strategy analysis of the traffic operation congestion parameters by the traffic emergency response mechanism; a signal light control fitness function construction unit, configured to evaluate and analyze the signal light control target to construct a signal light control fitness function; and a control optimization analysis unit, configured to perform control optimization analysis on the signal light control strategy solution space by using the signal light control fitness function to determine the traffic signal light control strategy parameters.

[0056] The traffic signal light control strategy parameters are determined by performing control optimization analysis on the signal light control strategy solution space by using the signal light control fitness function. The control optimization analysis unit further comprises: a strategy parameter sub-unit, configured to select a plurality of signal light control strategy parameters in the signal light control strategy solution space; a fitness evaluation sub-unit, configured to perform fitness evaluation on the plurality of signal light control strategy parameters by using the signal light control fitness function to obtain a plurality of control strategy parameter fitnesses; and an iterative optimization sub-unit, configured to perform iterative optimization in the signal light control strategy solution space based on the plurality of control strategy parameter fitnesses to determine the traffic signal light control strategy parameters.

[0057] Below, the specific configuration of the adaptive regulation module 60 will be described in detail. As described above, based on the traffic signal lamp control strategy parameter for adaptive regulation of the target traffic area signal lamp, the adaptive regulation module 60 further comprises: a control monitoring unit for control monitoring of the target traffic area signal lamp based on the traffic signal lamp control strategy parameter, obtaining a signal lamp control feedback parameter; an adaptive optimization regulation unit for optimization adjustment of the traffic signal lamp control strategy parameter based on the signal lamp control feedback parameter, obtaining a traffic signal lamp optimization control strategy parameter for adaptive optimization regulation of the target traffic area signal lamp.

[0058] The traffic signal lamp intelligent control system provided by the embodiments of the present application can execute the traffic signal lamp intelligent control method provided by any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0059] 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 name of each functional unit is only for the convenience of mutual differentiation, and does not limit the protection scope of the present application.

[0060] 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 modification, equivalent substitution and improvement made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps described in the present application can be executed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

Claims

1. A method for intelligent control of traffic lights, characterized in that, The method includes: A sensor array is deployed on the traffic lights in the target traffic area to collect traffic operation data and traffic environment data. The traffic adaptive analysis model is obtained through the traffic dispatching platform. The traffic adaptive analysis model consists of a traffic anomaly detection sub-model and a traffic congestion prediction sub-model. Based on the traffic adaptive analysis model, traffic condition analysis is performed on the traffic operation data and traffic environment data to obtain traffic congestion parameters. Based on the aforementioned traffic congestion parameters, the traffic emergency response mechanism is activated; Based on the traffic emergency response mechanism, the traffic congestion parameters are analyzed to determine the traffic signal control strategy parameters. The traffic lights in the target traffic area are adaptively adjusted based on the traffic light control strategy parameters. The process of obtaining the traffic adaptive analysis model through the traffic dispatching platform includes: Historical traffic operation data datasets are obtained through the traffic dispatch platform; A feedforward neural network is used to train the historical traffic operation dataset with sample labels to generate a traffic anomaly detection sub-model. A traffic congestion prediction sub-model is obtained by training the historical traffic operation dataset using a recurrent neural network with sample labels. The traffic anomaly detection sub-model and the traffic congestion prediction sub-model are combined to obtain the traffic adaptive analysis model; The generated traffic anomaly detection sub-model includes: The historical traffic operation dataset is labeled to obtain a normal traffic condition sample set and an abnormal traffic condition sample set. A feedforward neural network is used to detect and train the normal traffic condition sample set and the abnormal traffic condition sample set to obtain an initial traffic anomaly detection model. The initial traffic anomaly detection model is used to calculate the loss and obtain traffic loss data. The initial traffic anomaly detection model is optimized based on the traffic loss data to generate the traffic anomaly detection sub-model; The obtained traffic congestion prediction sub-model includes: The historical traffic operation dataset is subjected to time series processing and sample labeling to obtain a traffic trend status sample set; A recurrent neural network is used to train the traffic trend data set to obtain an initial traffic congestion prediction model. Based on the initial traffic congestion prediction model, loss calculation and optimization are performed to obtain the traffic congestion prediction sub-model.

2. The intelligent traffic light control method as described in claim 1, characterized in that, The determination of traffic signal control strategy parameters includes: Based on the traffic emergency response mechanism, the traffic congestion parameters are analyzed to construct a traffic light control strategy solution space. The objectives of traffic light control are evaluated and analyzed, and a fitness function for traffic light control is constructed. The traffic light control fitness function is used to perform control optimization analysis on the solution space of the traffic light control strategy to determine the parameters of the traffic light control strategy.

3. The intelligent traffic light control method as described in claim 2, characterized in that, Determining the traffic light control strategy parameters includes: Multiple traffic light control strategy parameters are selected within the solution space of the traffic light control strategy. The fitness function of the traffic light control is used to evaluate the fitness of the multiple traffic light control strategy parameters to obtain the fitness of the multiple control strategy parameters; Based on the fitness of the multiple control strategy parameters, the traffic light control strategy parameters are determined by iterative optimization within the solution space of the traffic light control strategy.

4. The intelligent traffic light control method as described in claim 3, characterized in that, The adaptive control of traffic lights in the target traffic area based on the traffic light control strategy parameters includes: Based on the traffic signal control strategy parameters, the traffic lights in the target traffic area are controlled and monitored to obtain traffic light control feedback parameters. Based on the traffic light control feedback parameters, the traffic light control strategy parameters are optimized and adjusted to obtain optimized traffic light control strategy parameters for adaptive optimization and control of the traffic lights in the target traffic area.

5. A traffic signal intelligent control system, characterized in that, The system is used to implement the intelligent traffic light control method according to any one of claims 1-4, and the system comprises: A sensor group deployment module is used to deploy sensor groups on traffic lights in a target traffic area, and to collect traffic operation data and traffic environment data through the sensor groups. A traffic adaptive analysis model acquisition module is used to acquire a traffic adaptive analysis model through a traffic dispatching platform. The traffic adaptive analysis model consists of a traffic anomaly detection sub-model and a traffic congestion prediction sub-model. The traffic condition analysis module is used to perform traffic condition analysis on the traffic operation data and traffic environment data based on the traffic adaptive analysis model, and to obtain traffic congestion parameters. A traffic emergency response mechanism activation module, which is used to activate the traffic emergency response mechanism based on the traffic congestion parameters; The traffic light control strategy analysis module is used to perform traffic light control strategy analysis on the traffic congestion parameters based on the traffic emergency response mechanism, and to determine the traffic light control strategy parameters. An adaptive control module is used to adaptively control the traffic lights in the target traffic area based on the traffic light control strategy parameters.

Images