A three-dimensional traffic flow prediction and dispatching system

Abstract

The present application proposes a three-dimensional traffic vehicle flow prediction and dispatching system, which relates to the technical field of traffic dispatching. Lane flow correlation analysis is performed based on historical vehicle flow data of multiple basic lanes and target lanes to obtain flow correlation coefficients, multiple flow fluctuations of multiple basic vehicle flows are analyzed, and vehicle flow prediction of target lanes is performed to obtain multiple predicted vehicle flows. Multiple predicted vehicle flows are weightedly calculated according to multiple flow correlation coefficients and multiple flow fluctuations to obtain predicted target vehicle flows, and vehicle flow dispatching is performed. The technical problems of insufficient dynamic correlation analysis of vehicle flows and limited prediction accuracy in existing traffic vehicle flow prediction and dispatching, which make it difficult for vehicle flow dispatching to adapt to complex three-dimensional traffic environments and slow dispatching response, are solved, and the technical effect of improving the accuracy and timeliness of vehicle flow prediction is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of traffic dispatching, and in particular to a three-dimensional traffic vehicle flow prediction and dispatching system. Background Art

[0002] With the rapid development of urban transportation, the three-dimensional transportation system has become a key solution to relieve urban traffic pressure and improve traffic efficiency. Three-dimensional transportation includes multi-level road structures, such as viaducts, underground roads and aerial passages. Especially in cities, the three-dimensional transportation network with elevated roads and underground roads is becoming increasingly complex, and the problem of traffic flow prediction and scheduling is becoming increasingly prominent. In order to effectively manage the traffic flow in three-dimensional transportation, many studies have proposed prediction models based on traditional traffic flow data, which mainly rely on historical data and current real-time data for prediction and scheduling.

[0003] However, most of the current traffic flow prediction technologies rely on methods based on statistical analysis and time series analysis. Although these methods can predict traffic flow to a certain extent, their disadvantage is that they cannot effectively handle nonlinear relationships in multi-dimensional and complex three-dimensional traffic environments. In addition, traditional methods usually only rely on data from a single lane or adjacent lanes to predict traffic flow, but there is significant traffic correlation and volatility between lanes in three-dimensional traffic, especially in multi-lane and intersection areas, where the traffic of multiple lanes has a great influence on each other. The correlation and volatility analysis of lane traffic in existing technologies has failed to deeply explore the dynamic relationship between different lanes, resulting in low prediction accuracy.

[0004] In summary, the existing traffic flow forecasting and scheduling often have technical problems such as insufficient dynamic correlation analysis of traffic flow and limited prediction accuracy, which makes it difficult for traffic flow scheduling to adapt to complex three-dimensional traffic environments and slow scheduling response. Summary of the invention

[0005] The present application provides a three-dimensional traffic vehicle flow prediction and dispatching system, which is used to solve the technical problems existing in the existing traffic vehicle flow prediction and dispatching, such as insufficient dynamic correlation analysis of vehicle flow, limited prediction accuracy, which makes it difficult for vehicle flow dispatching to adapt to the complex three-dimensional traffic environment and slow dispatching response.

[0006] The present application provides a three-dimensional traffic flow prediction and dispatching system, the system comprising:

[0007] A vehicle flow collection module is used to collect multiple basic vehicle flows of other multiple basic lanes in three-dimensional traffic; a correlation analysis module is used to perform lane flow correlation analysis based on the historical vehicle flow data of the multiple basic lanes and the target lane, and obtain multiple flow correlation coefficients; a fluctuation analysis module is used to analyze the multiple flow fluctuations of the multiple basic vehicle flows based on the historical vehicle flow data of the multiple basic lanes; a vehicle flow scheduling module is used to predict the vehicle flow of the target lane based on the multiple basic vehicle flows, obtain multiple predicted vehicle flows, perform weighted calculation on the multiple predicted vehicle flows according to the multiple flow correlation coefficients and the multiple flow fluctuations, obtain the predicted target vehicle flow of the target lane, and perform vehicle flow scheduling.

[0008] The present application proposes a three-dimensional traffic flow prediction and dispatching system, which collects multiple basic traffic flows of multiple other basic lanes in three-dimensional traffic; performs lane flow correlation analysis based on the historical traffic flow data of the multiple basic lanes and the target lane to obtain multiple flow correlation coefficients; analyzes multiple flow fluctuations of the multiple basic traffic flows based on the historical traffic flow data of the multiple basic lanes; predicts the flow of the target lane based on the multiple basic traffic flows to obtain multiple predicted traffic flows, performs weighted calculation on the multiple predicted traffic flows according to the multiple flow correlation coefficients and the multiple flow fluctuations to obtain the predicted target traffic flow of the target lane, and performs traffic flow dispatching, thereby solving the technical problems of insufficient dynamic correlation analysis of traffic flow and limited prediction accuracy in the existing traffic traffic flow prediction and dispatching, which makes it difficult for traffic flow dispatching to adapt to the complex three-dimensional traffic environment and slow dispatching response, and achieves the technical effect of improving the accuracy and timeliness of traffic flow forecasting. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of the structure of a three-dimensional traffic vehicle flow prediction and scheduling system provided in an embodiment of the present application.

[0010] Explanation of the reference numerals: vehicle flow collection module 11 , correlation analysis module 12 , fluctuation analysis module 13 , vehicle flow scheduling module 14 . DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0012] The present application embodiment provides a three-dimensional traffic flow prediction and scheduling system, such as Figure 1 As shown, the system comprises:

[0013] The vehicle flow collection module 11 is used to collect multiple basic vehicle flows of other multiple basic lanes in three-dimensional traffic.

[0014] Furthermore, in three-dimensional traffic, multiple basic vehicle flows of other multiple basic lanes are collected, and the vehicle flow collection module 11 is also used to: in three-dimensional traffic, determine the target lane for vehicle flow prediction and scheduling, and record other lanes as multiple basic lanes; collect multiple basic vehicle flows of the multiple basic lanes.

[0015] Specifically, it is necessary to first determine the target lane for traffic flow prediction and scheduling. The target lane refers to a specific lane in the three-dimensional transportation network for which traffic flow prediction and optimized scheduling are currently required. The target lane can be a city main road, an elevated road or an intersection lane, etc., which is usually a key area where traffic flow optimization is required. After determining the target lane, the system collects the traffic flow data of multiple basic lanes adjacent to or related to the target lane in real time through various sensors, cameras, radars and other equipment deployed in the three-dimensional transportation network. Among them, the basic lane refers to other lanes that have traffic flow correlation or influence on the target lane. The selection of the basic lane is based on its traffic flow correlation with the target lane. It can be different lanes of the same road, or it can be other road lanes parallel to or intersecting with the target lane. For example, if the target lane is a lane of an elevated road, the basic lane can be other lanes of the elevated road, or a ground road parallel to the elevated road. By selecting these basic lanes with traffic flow correlation, the comprehensiveness and accuracy of data collection can be ensured, providing basic data support for subsequent traffic flow correlation analysis and prediction.

[0016] The collected traffic volume refers to the vehicle quantity data collected by sensors and monitoring equipment within a preset time window. Multiple basic traffic volumes refer to the traffic volume data of multiple basic lanes. Traffic volume data usually includes key parameters such as the number of vehicles passing through the lanes, vehicle speed, lane occupancy rate, etc. These data help to understand the traffic status of different lanes and provide a reference for subsequent traffic volume prediction and scheduling. In actual operation, in order to ensure the accuracy and timeliness of the data, the system can set multiple data collection cycles. For example, the collection of traffic volume data can be updated in real time based on a time window of seconds or minutes. After each data collection cycle, the system will automatically summarize and record all traffic volume data within the time period to form a historical data sequence. By summarizing and storing these data, it can provide key support for subsequent traffic correlation analysis and volatility analysis.

[0017] The key role of this step is to provide the necessary data foundation for subsequent traffic analysis and prediction models. By collecting traffic flow data from multiple basic lanes, the system can fully understand the traffic status of the target lane and its surrounding environment, laying a solid data foundation for subsequent traffic correlation analysis, fluctuation calculation and traffic flow prediction. This process not only ensures the accuracy of the prediction, but also ensures that the system can adaptively adjust in a dynamically changing traffic environment and respond to changes in traffic flow in a timely manner.

[0018] The correlation analysis module 12 is used to perform lane flow correlation analysis based on the historical traffic flow data of the multiple basic lanes and the target lanes to obtain multiple flow correlation coefficients.

[0019] Optionally, historical traffic flow data refers to the traffic data of each lane collected during a certain period of time in the past. Historical traffic flow data includes traffic flow data sequences of each lane at multiple time nodes. These historical data cover different time periods, such as hours, days or weeks, to ensure that the analysis has a sufficient sample size and can reflect the changing trends and fluctuation characteristics of traffic flow. For each lane, historical traffic flow data includes key indicators such as the number of vehicles, speed, lane occupancy rate, etc. in the lane during a specific time period. These data can be collected through sensors or cameras installed on the lane and organized into data sequences in chronological order. Through these historical data, the traffic flow change patterns of each lane can be analyzed to provide a data basis for subsequent correlation analysis.

[0020] Next, the system will use the collected historical traffic data to perform lane traffic correlation analysis. The purpose of traffic correlation analysis is to identify the degree of correlation between the target lane and multiple basic lanes, that is, to determine the degree of mutual influence between them in traffic flow by analyzing the traffic flow change patterns between the lanes. Through traffic correlation analysis, multiple traffic correlation coefficients can be obtained, which reflect the strength of the correlation between traffic flows in each lane. Among them, the traffic correlation coefficient is a numerical indicator used to measure the degree of correlation between lane flows, which is usually calculated based on statistical methods. For example, commonly used calculation methods include Pearson correlation coefficient, gray correlation degree, etc. The Pearson correlation coefficient can measure the linear relationship between two variables, while the gray correlation degree can measure whether the changing trends of traffic flows between different lanes are similar.

[0021] In actual operation, in order to conduct traffic correlation analysis, the system first organizes the historical traffic data of multiple basic lanes and target lanes to ensure that the data format of each lane is consistent and the time nodes match. If there are missing values ​​or abnormal data, the system will perform data interpolation or remove abnormal data. Then, the appropriate correlation analysis method is selected according to the data characteristics and analysis requirements. For example, if the historical data is relatively complete and presents a linear relationship, the system can use the Pearson correlation coefficient for analysis; if the data is uncertain or incomplete, the system can choose the gray correlation analysis method. Through these methods, the system can calculate the traffic correlation coefficient between each pair of lanes. By performing correlation analysis on the historical traffic data of multiple basic lanes and target lanes, the system will generate multiple traffic correlation coefficients, each of which reflects the traffic similarity between the corresponding lanes. For example, if the traffic correlation coefficient between a basic lane and a target lane is close to 1, it means that the traffic flow change trends of the two are very similar; if it is close to 0, it means that there is almost no linear relationship between the two. By calculating the traffic correlation coefficient, the system can identify which basic lanes and target lanes have a strong correlation in traffic changes, thereby providing more accurate reference data for subsequent traffic flow forecasts, better understanding the relationship between different lanes, and thus improving the accuracy of overall traffic flow forecasts.

[0022] The fluctuation analysis module 13 is used to analyze multiple flow fluctuations of the multiple basic vehicle flows according to the historical vehicle flow data of the multiple basic lanes.

[0023] For example, by deeply analyzing the historical traffic flow data of multiple basic lanes, the traffic flow fluctuation of each lane is calculated and analyzed. Traffic flow fluctuation is a key indicator used to measure the change range of traffic flow within a certain time range. It reflects the stability or volatility of traffic flow and is crucial for traffic flow prediction and scheduling. The analysis of traffic flow fluctuation can reveal the fluctuation trend of lane flow and provide valuable reference information for subsequent traffic flow prediction.

[0024] Specifically, the traffic flow data sequence of each basic lane is extracted from the historical traffic flow data, which includes the number of vehicles or flow values ​​at multiple time nodes. The system will ensure that these data are arranged in chronological order to avoid time confusion affecting subsequent analysis. In order to ensure the accuracy of the data, it is necessary to process missing values ​​or abnormal data. Common methods include data interpolation or removal of outliers. Subsequently, the system selects a data window according to the time interval that needs to be analyzed. For example, you can choose daily average data, hourly data, or more fine-grained minute-by-minute data. By analyzing historical data, the system can derive the traffic flow changes of each lane at different time nodes.

[0025] Traffic flow fluctuation refers to the degree of fluctuation of lane traffic flow within a certain time range. It measures the fluctuation amplitude of traffic flow, that is, the drastic degree of traffic flow change. In order to calculate the traffic flow fluctuation, the historical traffic flow series of each basic lane is first averaged, that is, the average value of the lane traffic flow is calculated within the selected time window and recorded as the average traffic flow. For example, if the traffic flow data of a basic lane for one day is considered, the average traffic flow within one day is calculated. Furthermore, the degree of traffic flow fluctuation is reflected by calculating the deviation between the traffic flow at each time node and the average traffic flow of the lane. The deviation percentage can be calculated by the following formula: By calculating the deviation percentage at each time node, the system can obtain the traffic flow fluctuation data of the lane. Finally, the traffic flow fluctuation of each basic lane will be obtained through statistical analysis of the deviation percentage data. For example, the system can calculate the mean value, variance and other statistics of the fluctuation of each lane to comprehensively evaluate the volatility of traffic flow.

[0026] At this stage, the system will calculate the flow fluctuation of each basic lane separately and obtain the flow fluctuation data of multiple basic lanes. For example, if five basic lanes are involved, the system will obtain the flow fluctuation of these five lanes. These fluctuation data can be used to evaluate the stability of traffic flow changes in each lane. A lane with a larger fluctuation means that the traffic flow of the lane fluctuates more violently and may change greatly in different time periods; while a lane with a smaller fluctuation indicates that the traffic flow of the lane is relatively stable and changes less. Flow fluctuation analysis plays an important role in traffic flow prediction. Lanes with large fluctuations often show more uncertainty, and the difficulty of prediction increases accordingly. Therefore, in the subsequent traffic flow prediction, the system will assign weights to different lanes in the prediction model according to the size of the fluctuation. For example, a lane with a larger fluctuation may require more consideration and weighting to cope with the uncertainty in its prediction.

[0027] For example, suppose that in a three-dimensional traffic network, the target lane is the main lane of an elevated bridge, while the basic lanes are other lanes on the elevated bridge and the ground roads adjacent to the elevated bridge. By collecting data from the past week, the system extracts the historical traffic volume of each basic lane and calculates the traffic volume fluctuation of these lanes. For example, for one of the basic lanes, the system calculates that its average traffic volume is 1,200 vehicles / hour, and the deviation percentage of the hourly traffic volume is concentrated within 15% most of the time, but there is a 30% fluctuation during peak hours. This means that the traffic volume fluctuation of this lane is large, and the system will pay more attention to its uncertainty in subsequent traffic volume forecasts. By analyzing the traffic volume fluctuations of multiple basic lanes, the system can accurately evaluate the traffic volume changes of each lane and provide more valuable information for subsequent traffic volume forecasts. The calculation of traffic volume fluctuations helps to reveal the stability differences of traffic volumes in different lanes, thereby optimizing the weight distribution of the traffic volume forecast model and improving the accuracy and real-time performance of the entire traffic volume forecast system.

[0028] The vehicle flow scheduling module 14 is used to predict the vehicle flow of the target lane according to the multiple basic vehicle flows, obtain multiple predicted vehicle flows, perform weighted calculation on the multiple predicted vehicle flows according to the multiple flow correlation coefficients and the multiple flow fluctuations, obtain the predicted target vehicle flow of the target lane, and perform vehicle flow scheduling.

[0029] Specifically, based on the historical traffic flow data of multiple basic lanes, the prediction model is used to predict the future traffic flow of the target lane, usually predicting the traffic flow in a certain period of time in the future, such as half an hour later. First, the system constructs a traffic flow prediction model based on the historical traffic flow data of multiple basic lanes collected in the previous steps and the results of lane flow correlation analysis to predict the traffic flow of the target lane. The prediction model will be based on the traffic flow data of each basic lane, combined with the flow correlation coefficient and volatility data between the target lane and the basic lane.

[0030] In a specific embodiment, in order to predict the traffic flow, the system first needs to build a traffic flow prediction channel for the target lane, which includes multiple prediction branches, each branch corresponding to a basic lane. The system takes the historical traffic flow data of multiple basic lanes as input, and the time window of the input data should be consistent with the time of the prediction target. For example, if the system wants to predict the traffic flow of the target lane in the next half hour, the historical data of the basic lane should cover the traffic flow of the corresponding time period in the past (such as the past half hour or 1 hour). The traffic flow prediction of each basic lane is based on a machine learning algorithm or a statistical model. For example, support vector machine (SVM), regression analysis, neural network and other methods can be used to train a corresponding traffic flow prediction model for the historical traffic flow data of each basic lane, wherein each basic lane is predicted through its exclusive prediction branch, thereby outputting the corresponding predicted traffic flow value. Assuming that the predicted time period is half an hour later, the system will predict the traffic flow of each basic lane in this period separately, and obtain multiple predicted traffic flow values. Subsequently, after obtaining the predicted traffic flow of multiple basic lanes, the system performs weighted calculation on these predicted traffic flows according to the multiple traffic correlation coefficients and multiple traffic fluctuations calculated previously. First, the flow correlation coefficient reflects the degree of flow correlation between the basic lane and the target lane. The flow fluctuation reflects the volatility of the traffic flow of each basic lane. In order to improve the accuracy of the prediction results, the system adjusts multiple predicted traffic flows according to two weighting principles: flow correlation coefficient weighting and flow fluctuation weighting. Among them, the flow correlation coefficient weighting means that the predicted traffic flow of each basic lane will be weighted according to the flow correlation coefficient between it and the target lane. The higher the correlation coefficient, the greater the impact of the predicted traffic flow on the target lane, so it should be given a higher weight; the flow fluctuation weighting means that for lanes with larger fluctuations, the predicted value of its traffic flow may have higher uncertainty, so the system will adjust the prediction weight according to the size of the flow fluctuation. Lanes with larger fluctuations will receive lower weights to reduce the impact of uncertainty on the overall prediction results.

[0031] After weighted calculation, the system will obtain the final predicted traffic volume of the target lane. This prediction result reflects the traffic volume level of the target lane in the future and is usually used for traffic volume scheduling decisions. According to the predicted traffic volume, the traffic control system can optimize the scheduling strategies such as signal light timing, lane allocation, and traffic flow guidance of the target lane, so as to achieve the purpose of optimizing traffic flow, reducing congestion, and improving road traffic efficiency. For example, if the system predicts that the traffic volume of the target lane will increase sharply in half an hour, the signal light cycle can be adjusted in advance, or the traffic flow guidance can be adjusted dynamically to divert traffic, thereby avoiding traffic jams or accidents. On the contrary, if the traffic volume of the target lane is predicted to be low, the system can appropriately extend the signal light cycle to improve the road traffic efficiency.

[0032] By making predictions based on the traffic flow data of multiple basic lanes and performing weighted calculations based on the traffic correlation coefficient and volatility, the traffic flow of the target lane can be accurately predicted and a scientific basis can be provided for traffic scheduling. This process effectively improves the accuracy of traffic flow prediction in three-dimensional transportation, reduces traffic congestion, and improves the overall road traffic efficiency.

[0033] Furthermore, based on the historical traffic flow data of the multiple basic lanes and the target lanes, a lane traffic flow correlation analysis is performed to obtain multiple traffic flow correlation coefficients. The correlation analysis module 12 is also used to: collect the historical traffic flow data of the multiple basic lanes and the target lanes at multiple time nodes in the past to obtain multiple historical basic traffic flow sequences and historical target traffic flow sequences; based on the multiple historical basic traffic flow sequences and historical target traffic flow sequences, a lane traffic flow correlation analysis of the multiple basic lanes and the target lanes is performed to obtain multiple traffic flow correlation coefficients.

[0034] Furthermore, the system first needs to collect and construct multiple historical traffic flow data sequences from the historical traffic flow data of multiple basic lanes and target lanes, especially selecting those traffic flow data corresponding to specific time nodes. Assuming that the goal is to predict the traffic flow of the target lane half an hour later, the system will collect the traffic flow data of multiple basic lanes and target lanes at the time point based on multiple past time nodes, especially at a fixed time period every day (such as 9 o'clock). Specifically, select the target lane and determine multiple basic lanes with traffic correlation. Then, collect the traffic flow data of these basic lanes and target lanes at multiple past time nodes respectively. Assuming that the goal is to predict the traffic flow of the target lane in the next half hour (for example, the 9 o'clock period), the system will go back to the historical data to collect the traffic flow data of each basic lane and target lane at the 9 o'clock period of each day in the past multiple days. These data include the number of vehicles, speed, occupancy rate and other information of each lane at the 9 o'clock time node. The historical traffic flow data of each lane will be organized in chronological order to form a series of traffic flow sequences at time nodes. For example, the system will obtain a historical basic traffic flow sequence, which includes the traffic flow data of the basic lane at multiple past time nodes (such as the 9 o'clock period on multiple days); at the same time, the system will also construct a historical target traffic flow sequence, which includes the traffic flow data of the target lane at these same time nodes.

[0035] After obtaining multiple historical traffic flow sequences, the system will perform lane flow correlation analysis based on these historical data, aiming to quantify the flow correlation between each basic lane and the target lane. The goal of flow correlation analysis is to identify the similarity of flow change patterns between the target lane and the basic lane, and then evaluate the degree of influence of the basic lane on the traffic flow of the target lane. Specifically, the system will use statistical or mathematical methods, such as Pearson correlation coefficient, grey correlation analysis, etc., to calculate the correlation coefficient between each basic lane and the target lane. The Pearson correlation coefficient is used to measure the linear relationship between two variables. The value ranges from -1 to 1, indicating the strength of the correlation between the two. A value close to 1 indicates a strong positive correlation between the two, while a value close to -1 indicates a negative correlation. Grey correlation is suitable for processing incomplete data or more complex nonlinear relationships, and can evaluate the similarity of flow data sequences between different lanes. The system obtains multiple flow correlation coefficients by calculating the flow correlation coefficient between each basic lane and the target lane. These coefficients will be used as the basis for weighted calculation in subsequent traffic flow prediction. The higher the correlation coefficient of the basic lane, the greater the impact of its traffic flow prediction on the target lane, so it needs to be given a higher weight in the prediction model. The traffic correlation coefficient provides a reliable basis for the traffic flow prediction model. The basic lanes with high correlation will be given higher prediction weights, thereby improving the accuracy and credibility of the traffic flow prediction of the target lane, and further improving the accuracy of the traffic flow prediction of the target lane.

[0036] Furthermore, based on the multiple historical basic vehicle flow sequences and the historical target vehicle flow sequences, a lane flow correlation analysis of the multiple basic lanes and the target lanes is performed, and the correlation analysis module 12 is also used to: use the multiple historical basic vehicle flow sequences as multiple influencing sequences and the historical target vehicle flow sequence as the main sequence; based on the grey correlation analysis, process and obtain the correlation coefficients of the multiple influencing sequences and the main sequence as multiple flow correlation coefficients.

[0037] Optionally, the historical traffic flow data of multiple basic lanes are used as the influence sequence, and the historical traffic flow data of the target lane is used as the main sequence. In this way, the system can analyze the traffic flow change relationship between each basic lane and the target lane, thereby providing the necessary correlation support for the prediction model. Specifically, the influence sequence refers to the historical traffic flow data sequence of multiple basic lanes. These basic lanes may have different degrees of traffic correlation with the target lane, and are therefore used as influencing factors to affect the traffic flow prediction of the target lane. For example, if the traffic flow of the target lane changes greatly in a certain period of time in the past, the traffic flow of the basic lane may also change similarly, thereby affecting the target lane. The main sequence refers to the historical traffic flow data sequence of the target lane. The historical traffic flow data sequence of the target lane will be used as the main sequence for the system to compare and analyze with the influence sequences of multiple basic lanes. The main sequence is the key data used for prediction, and its traffic flow changes will be affected by multiple basic lanes.

[0038] Grey correlation analysis is an effective nonlinear data analysis method, which is widely used in the correlation analysis of complex systems. In this embodiment, the system adopts the grey correlation analysis method to process the relationship between multiple influence sequences and the main sequence. The core idea of ​​grey correlation is to calculate the similarity or correlation by comparing the change trends between data sequences, so as to obtain the degree of influence of each influence sequence on the main sequence. Specifically, to ensure the accuracy of the analysis, the system first standardizes the influence sequence and the main sequence. The purpose of standardization is to uniformly convert data of different dimensions or scales into dimensionless data, thereby eliminating the effects caused by different units or magnitudes. A common standardization method is range standardization, that is, scaling the data to the range of [0,1]. In the standardized influence sequence and main sequence, the system will calculate the grey correlation coefficient of each time node. The correlation coefficient measures the similarity between two data sequences at a certain moment. Specifically, the calculation formula of the grey correlation coefficient is: Among them, x i (k) and x j (k) are the standardized data values ​​of the influencing sequence and the main sequence at time node k, respectively. ρ is the discrimination coefficient, which is usually taken as 0.5. Through this calculation, the system can obtain the grey correlation coefficient between each basic lane and the target lane, reflecting the similarity of their traffic flow change trends. After calculating the correlation coefficient at each time node, the system will average the correlation coefficient of each basic lane to obtain the overall correlation between each basic lane and the target lane. These average correlation coefficients are the flow correlation coefficients, which represent the flow correlation strength between each basic lane and the target lane.

[0039] Through grey correlation analysis, the system obtains multiple flow correlation coefficients. Each flow correlation coefficient reflects the similarity between the traffic flow change of a certain basic lane and the traffic flow change of the target lane. The higher the correlation coefficient, the more similar the traffic flow change trend of the basic lane is to that of the target lane, indicating that the basic lane has a greater impact on the traffic flow of the target lane. These flow correlation coefficients will be used as a weighted basis in the subsequent traffic flow prediction and scheduling process to help the system determine which basic lanes are most important for the traffic flow prediction of the target lane. Basic lanes with high flow correlation coefficients will obtain higher prediction weights, thereby having a greater impact on the traffic flow prediction of the target lane. Moreover, obtaining multiple flow correlation coefficients can quantify the flow correlation between each basic lane and the target lane. These coefficients provide an accurate weighted basis for subsequent traffic flow prediction, ensuring that the traffic flow prediction of the target lane is more accurate and reliable.

[0040] Furthermore, based on the historical traffic flow data of the multiple basic lanes, multiple flow fluctuations of the multiple basic traffic flows are analyzed, and the fluctuation analysis module 13 is also used to: calculate and obtain multiple average basic traffic flows based on the multiple historical basic traffic flow sequences; calculate the deviation percentages between the multiple basic traffic flows and the multiple average basic traffic flows as multiple flow fluctuations.

[0041] Specifically, the historical traffic flow data collected for each basic lane is processed to construct a historical traffic flow sequence for each lane. These sequences include traffic flow data at each time point over a period of time. The system will use these historical traffic flow sequences to calculate the average traffic flow of each basic lane. Specifically, the traffic flow data of each lane is accumulated and summed, and then divided by the total number of data points to obtain the average traffic flow value of each lane. This average value reflects the typical traffic flow level of the lane during the observation period.

[0042] Next, the system calculates the deviation between the traffic volume of each basic lane at each time point and its average traffic volume. This deviation is determined by comparing the difference between the actual traffic volume at each time point and the average traffic volume. These deviations are then converted into deviation percentages, that is, the difference between the traffic volume at each time point and the average traffic volume is divided by the average traffic volume and multiplied by 100%, resulting in a percentage value that represents the degree of change in traffic volume at a specific time point relative to the average level. This deviation percentage helps quantify the volatility of traffic volume. Traffic volume volatility is the statistical analysis result based on these deviation percentages, which can be obtained by calculating the standard deviation or variance of the deviation percentage. This volatility indicator measures the fluctuation of traffic volume at different time points and reflects the stability of traffic volume. The higher the volatility, the more unstable the traffic volume of the lane and the more violent the fluctuation; the lower the volatility, the more stable the traffic volume. Through the calculated traffic volume volatility, the system can better understand the stability of traffic volume in each basic lane. This information is very critical when predicting and scheduling traffic volume. For example, for lanes with higher volatility, more uncertainties may need to be considered in the traffic flow prediction model, or more flexible measures may need to be taken in the scheduling strategy to cope with possible fluctuations. On the contrary, lanes with lower volatility can use more stable prediction and scheduling models.

[0043] By analyzing and calculating the historical traffic flow data of multiple basic lanes to obtain the average traffic flow and its fluctuation of each lane, the system can effectively evaluate the traffic flow stability of each lane. This evaluation provides key reference information for traffic flow prediction and scheduling, and helps to improve the efficiency and accuracy of traffic management.

[0044] Furthermore, based on the multiple basic vehicle flows, the vehicle flow of the target lane is predicted to obtain multiple predicted vehicle flows. The vehicle flow scheduling module 14 is also used to: pre-train a target flow prediction channel for predicting the vehicle flow of the target lane, wherein the target flow prediction channel includes multiple target flow prediction branches, and the multiple target flow prediction branches correspond one-to-one to the multiple basic lanes; the multiple basic vehicle flows are respectively input into the multiple target flow prediction branches, and the prediction output obtains multiple predicted vehicle flows.

[0045] Exemplarily, the system uses a comprehensive prediction model to predict the future traffic flow of the target lane, which covers multiple prediction branches directly related to multiple basic lanes. Each prediction branch is trained on the historical traffic flow data of a basic lane to improve the accuracy and relevance of the prediction. The design of this branch structure enables the prediction model to process information from multiple data sources at the same time, thereby more accurately predicting the traffic flow of the target lane.

[0046] Specifically, the system first builds a target traffic prediction channel, which contains multiple independent prediction branches, each corresponding to a basic lane. This structural design allows each branch to specifically process data related to its corresponding lane, so that the prediction model can fully utilize the traffic flow characteristics and historical data of each lane for learning. The main function of the target traffic prediction channel is to integrate the influences from different basic lanes and make a comprehensive prediction of the future traffic flow of the target lane.

[0047] Before actual prediction, each prediction branch needs to be pre-trained. The purpose of pre-training is to enable the model of each branch to learn and understand the traffic flow change patterns and trends of its corresponding basic lane. Through machine learning techniques such as neural networks or regression analysis, the pre-training process will adjust the model parameters based on historical traffic flow data so that it can accurately predict future traffic flow. These historical data usually include information such as the size of traffic flow, time series, seasonal changes, etc.

[0048] After the pre-training is completed, the system inputs the latest or selected historical traffic flow data of the basic lane into the corresponding prediction branch. Each branch uses its trained model to independently predict the traffic flow of its corresponding lane at a certain point in the future. These prediction results will represent the traffic flow prediction of the target lane from the perspective of each basic lane. The system integrates the output results of each prediction branch (i.e., the predicted traffic flow), usually using weighted average or other statistical methods to combine these prediction values ​​to obtain the final target lane traffic flow prediction result. In this process, the weight of each prediction branch may be adjusted according to the traffic correlation coefficient and traffic fluctuation between each basic lane and the target lane to ensure the accuracy and reliability of the prediction results.

[0049] By constructing a target traffic flow prediction channel containing multiple prediction branches, the system can effectively use the traffic flow data of multiple basic lanes to predict the future traffic flow of the target lane. This method improves the accuracy and reliability of the prediction, provides strong support for traffic management, and makes traffic scheduling more effective and efficient.

[0050] Furthermore, the pre-training is used to predict the target traffic flow prediction channel of the target lane, and the traffic flow scheduling module 14 is also used to: collect multiple historical basic traffic flow sets of the multiple basic lanes, and collect the traffic flow of the target lane after a preset time length of each historical basic traffic flow, and mark it as a sample predicted traffic flow set; respectively use the multiple historical basic traffic flow sets as input data, use the sample predicted traffic flow set as supervised training, and train multiple target traffic flow prediction branches based on machine learning to obtain a target traffic flow prediction channel.

[0051] Furthermore, historical traffic flow data of multiple basic lanes in the past preset time length (for example, every half hour) are collected, and these data constitute a historical basic traffic flow set, including traffic flow information of each basic lane at different time points. At the same time, the system will also collect traffic flow data of the target lanes corresponding to these time points half an hour later to form a sample predicted traffic flow set. This data collection method allows the system to analyze the time correlation between the basic traffic flow and the traffic flow of the target lane, and provide the necessary input and label data for subsequent machine learning training.

[0052] Using the obtained historical basic traffic flow set as input data and the sample predicted traffic flow set as the target (label) of supervised training, the system will train multiple target traffic flow prediction branches based on machine learning technology. Each prediction branch is responsible for processing the basic lane data related to it and predicting the impact of the data on the traffic flow of the target lane. In this process, a variety of machine learning algorithms, such as support vector machines, random forests, or deep learning networks, can be used to learn and simulate the law of traffic flow changes based on historical data. Each target traffic flow prediction branch works independently to provide a prediction of the traffic flow of the target lane under the influence of its corresponding basic lane. After training, the system will integrate the outputs of these branches to form a comprehensive prediction result. This is usually achieved through weighted averaging or more complex integrated learning techniques to ensure that the prediction results obtained from multiple data sources are both accurate and reliable.

[0053] After the initial prediction model training and verification, the system may need to adjust and optimize the model based on actual performance, which includes adjusting model parameters, optimizing data input format, or redefining prediction targets. The optimized model will be deployed in the actual traffic management system as a complete target traffic flow prediction channel to predict the traffic flow of the target lane in real time and support traffic scheduling and management decisions.

[0054] By establishing a target flow prediction channel including multiple pre-trained branches, the system can effectively use the historical traffic data of multiple base lanes to predict the traffic flow of the target lane at a specific time in the future. This method provides a powerful tool for traffic management by integrating multi-source data and applying advanced machine learning technology, which helps to achieve more accurate and efficient traffic flow management and scheduling.

[0055] Furthermore, according to the multiple flow correlation coefficients and the multiple flow fluctuations, the multiple predicted traffic flows are weightedly calculated to obtain the predicted target traffic flow of the target lane, and traffic flow scheduling is performed. The traffic flow scheduling module 14 is also used to: subtract the multiple flow fluctuations from 1 to obtain multiple flow accuracy coefficients; respectively use the multiple flow accuracy coefficients and multiply them by the multiple flow correlation coefficients to obtain multiple corrected flow correlation coefficients; calculate the ratio of each corrected flow correlation coefficient to the sum of the multiple corrected flow correlation coefficients to obtain multiple prediction weights; use the multiple prediction weights to weightedly calculate the multiple predicted traffic flows to obtain the predicted target traffic flow of the target lane, and perform traffic flow scheduling.

[0056] Specifically, the flow fluctuation of each basic lane is evaluated. Lanes with large flow fluctuations indicate that their traffic flow data fluctuates violently, and the corresponding prediction accuracy is low. In order to adjust the impact of this instability on the prediction results, the system uses 1 minus the flow fluctuation method to calculate the flow accuracy coefficient of each basic lane. This coefficient reflects the stability of the traffic flow data of each lane and the credibility of the prediction. For example, if the flow fluctuation of a lane is 0.4, its flow accuracy coefficient is 0.6, indicating that the traffic flow data of this lane is relatively stable and the prediction credibility is high.

[0057] Next, the system combines the calculated flow accuracy coefficient with the flow correlation coefficient of each lane and performs multiplication to obtain the corrected flow correlation coefficient. This step aims to comprehensively consider the traffic flow fluctuation of the base lane and its flow correlation with the target lane to adjust the weight of each lane in the prediction model. The higher the corrected flow correlation coefficient, the greater the impact of the lane data on the predicted target lane traffic flow. The ratio of each corrected flow correlation coefficient to the sum of all corrected flow correlation coefficients is then calculated to determine the prediction weight of each base lane. This ratio reflects the relative importance of each base lane in the overall prediction model. A higher prediction weight for a lane means that the historical traffic flow data of the lane plays a greater role in predicting the future traffic flow of the target lane.

[0058] Finally, the system will use the prediction weights calculated above to perform weighted calculations on the predicted traffic flow of each basic lane. In this way, the system combines the prediction results of all basic lanes to obtain the comprehensive predicted traffic flow of the target lane. This prediction result will be used to guide the traffic dispatching strategy of the target lane, such as signal control, lane use optimization, etc., in order to achieve effective management and dispatching of traffic.

[0059] For example, assuming that the corrected traffic correlation coefficients of basic lanes A, B, and C are 0.2, 0.5, and 0.3 respectively, and if the predicted traffic volumes of these lanes are 800, 1000, and 900 vehicles / hour respectively, the predicted traffic volume of the target lane is calculated as follows: predicted traffic volume = (800×0.2)+(1000×0.5)+(900×0.3)=160+500+270=930 vehicles / hour. This weighted calculation method not only considers the impact of the basic lane on the target lane, but also the stability and reliability of its traffic volume data, making the prediction result more accurate and reliable.

[0060] By using the flow accuracy coefficient and the modified flow correlation coefficient to weightedly calculate the predicted traffic flow in the basic lane, the accuracy of the traffic flow prediction in the target lane can be effectively improved. This refined prediction and scheduling method helps traffic management departments to more effectively control and optimize traffic flow, reduce traffic congestion, and improve road use efficiency.

[0061] Through the technical solutions of the above-mentioned embodiments, a three-dimensional traffic vehicle flow prediction and dispatching system provided by the present application solves the technical problems existing in the existing traffic vehicle flow prediction and dispatching, such as insufficient dynamic correlation analysis of vehicle flow and limited prediction accuracy, which makes it difficult for vehicle flow dispatching to adapt to the complex three-dimensional traffic environment and slow dispatching response, thereby achieving the technical effect of improving the accuracy and timeliness of vehicle flow prediction.

[0062] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art 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. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A three-dimensional traffic flow prediction and dispatching system, characterized in that: The system comprises: The vehicle flow collection module is used to collect multiple basic vehicle flows of other basic lanes in three-dimensional traffic; A correlation analysis module, configured to perform lane flow correlation analysis based on historical traffic flow data of the plurality of basic lanes and the target lane to obtain a plurality of flow correlation coefficients; A fluctuation analysis module, configured to analyze a plurality of flow fluctuations of the plurality of basic traffic flows according to the historical traffic flow data of the plurality of basic lanes; The vehicle flow scheduling module is used to predict the vehicle flow of the target lane according to the multiple basic vehicle flows, obtain multiple predicted vehicle flows, perform weighted calculation on the multiple predicted vehicle flows according to the multiple flow correlation coefficients and the multiple flow fluctuations, obtain the predicted target vehicle flow of the target lane, and perform vehicle flow scheduling.

2. The three-dimensional traffic vehicle flow prediction and dispatching system according to claim 1 is characterized in that: The vehicle flow collection module is also used for: In three-dimensional traffic, the target lane for traffic flow prediction and scheduling is determined, and other lanes are recorded as multiple basic lanes; A plurality of basic vehicle flows of the plurality of basic lanes are collected.

3. The three-dimensional traffic vehicle flow prediction and dispatching system according to claim 1 is characterized in that: The correlation analysis module is also used for: Collecting historical traffic flow data of the plurality of basic lanes and the target lanes at a plurality of time nodes in the past to obtain a plurality of historical basic traffic flow sequences and a plurality of historical target traffic flow sequences; According to the multiple historical basic vehicle flow sequences and the historical target vehicle flow sequences, a lane flow correlation analysis of the multiple basic lanes and the target lanes is performed to obtain multiple flow correlation coefficients.

4. The three-dimensional traffic vehicle flow prediction and dispatching system according to claim 3 is characterized in that: The correlation analysis module is also used for: The multiple historical basic traffic flow sequences are used as multiple influencing sequences, and the historical target traffic flow sequence is used as a main sequence; Based on the grey correlation analysis, correlation coefficients between the multiple influencing sequences and the main sequence are obtained as multiple flow correlation coefficients.

5. The three-dimensional traffic vehicle flow prediction and dispatching system according to claim 3 is characterized in that: The volatility analysis module is also used for: Calculating and obtaining a plurality of average basic vehicle flows according to the plurality of historical basic vehicle flow sequences; Deviation percentages between the multiple basic vehicle flows and the multiple average basic vehicle flows are calculated as multiple flow fluctuations.

6. The three-dimensional traffic vehicle flow prediction and dispatching system according to claim 1 is characterized in that: The vehicle flow scheduling module is also used for: Pre-training a target flow prediction channel for predicting the flow of vehicles in a target lane, wherein the target flow prediction channel includes a plurality of target flow prediction branches, and the plurality of target flow prediction branches correspond one to one to the plurality of basic lanes; The multiple basic vehicle flows are respectively input into the multiple target flow prediction branches, and the prediction outputs are used to obtain multiple predicted vehicle flows.

7. The three-dimensional traffic vehicle flow prediction and dispatching system according to claim 6 is characterized in that: The vehicle flow scheduling module is also used for: Collecting multiple historical basic traffic flow sets of the multiple basic lanes, and collecting the traffic flow of the target lane after a preset time length of each historical basic traffic flow, and marking them as a sample predicted traffic flow set; The multiple historical basic vehicle flow sets are respectively used as input data, the sample predicted vehicle flow set is used as supervised training, and multiple target flow prediction branches are trained based on machine learning to obtain a target flow prediction channel.

8. The three-dimensional traffic vehicle flow prediction and dispatching system according to claim 1 is characterized in that: The vehicle flow scheduling module is also used for: Subtracting the plurality of flow fluctuations from 1 to obtain a plurality of flow accuracy coefficients; The plurality of flow accuracy coefficients are respectively adopted and multiplied by the plurality of flow correlation coefficients to obtain a plurality of corrected flow correlation coefficients; Calculate the ratio of each corrected flow correlation coefficient to the sum of multiple corrected flow correlation coefficients to obtain multiple prediction weights; The multiple prediction weights are used to perform weighted calculation on the multiple predicted traffic flows, to obtain the predicted target traffic flow of the target lane, and to perform traffic flow scheduling.

Images