Intelligent networked traffic state monitoring method and system based on multi-source data fusion

Abstract

The application discloses a multi-source data fusion intelligent networked traffic state monitoring method and system, and particularly relates to the technical field of intelligent traffic, and comprises the following steps: in a T time period, speed data of M data sources in an nth topological road section is acquired, the speed data is preprocessed to obtain a data quality coefficient of the nth topological road section; the data quality coefficient of the nth topological road section is analyzed to determine whether the M data sources meet the requirements, if not, a data optimization instruction is generated, and if yes, a state analysis instruction is generated; by analyzing the speed data of the M data sources and calculating the data quality coefficient, the accuracy and reliability of input data can be ensured. When the data sources do not meet the requirements, the data optimization instruction is generated to avoid the influence of low-quality data on the system. Through real-time detection and optimization mechanism of data quality, the interference of data noise or loss is reduced, and the accuracy of the fusion model is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and more specifically, to a method and system for intelligent connected traffic status monitoring based on multi-source data fusion. Background Technology

[0002] Traditional traffic monitoring methods primarily rely on a single data source, such as traffic cameras, ground sensors, or GPS devices. These methods have limitations in terms of data collection accuracy, real-time performance, and coverage, and cannot comprehensively reflect traffic conditions.

[0003] In the prior art, for example, Chinese patent application CN103838772A discloses a multi-source traffic data fusion method. This method includes: acquiring traffic information data from three data sources—mobile phones, floating vehicles, and traffic video images; performing pixel-level fusion on the traffic information data to remove unqualified information; performing feature-level fusion on the traffic information data from each data source to generate road segment traffic state information from the three different data sources; performing decision-level fusion on the road segment traffic state information from the three different data sources to generate consistent traffic state description information for the road segment; and outputting the traffic state description information for the road segment. While the above method acquires traffic information data from multiple data sources and performs three-level fusion to generate the final traffic state of the road segment, enabling a more accurate determination of road surface traffic conditions, research and application of the above method and prior art have revealed at least the following shortcomings:

[0004] 1. Without the use of data standardization and time alignment algorithms, it is difficult to ensure the uniformity of format and time sequence of data from different sources;

[0005] 2. Pixel-level fusion is suitable for visual data from the same source or highly correlated (such as multiple consecutive images) because it is based on the overlay and analysis of information pixel by pixel. However, for heterogeneous data (such as visible light images and thermal imaging images or data collected by different sensors), direct pixel-level fusion may lead to information distortion or redundant processing, thereby reducing the effectiveness and reliability of data fusion. When monitoring complex traffic conditions, such distortion may affect the real-time performance and accuracy of the system.

[0006] To this end, the present invention provides a method and system for intelligent connected traffic status monitoring based on multi-source data fusion. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art, this invention provides a method and system for intelligent connected traffic condition monitoring based on multi-source data fusion, in order to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] In a first aspect, the present invention provides a method for intelligent connected traffic condition monitoring by fusing multi-source data, including:

[0010] Within time period T, acquire speed data from M data sources in the nth topological segment, preprocess the speed data to obtain the data quality coefficient of the nth topological segment;

[0011] Analyze the data quality coefficient of the nth topological segment to determine whether the M data sources meet the requirements. If they do not meet the requirements, generate a data optimization instruction; if they do meet the requirements, generate a status analysis instruction.

[0012] Receive data optimization instructions, optimize the data quality coefficient of the nth topology segment to obtain the speed optimization coefficient, and convert the data optimization instructions into state analysis instructions;

[0013] Receive the state analysis instruction, obtain the road feature data of the nth topology segment, generate the road load factor within the time period T based on the road feature data, and input the generated road load factor into the pre-generated state prediction model to obtain the road load factor within the time period T+a.

[0014] Extract the road load factor of the nth topological road segment within the time period T+a, and match the nth topological road segment with the traffic status warning information based on the road load factor to obtain the corresponding status warning information.

[0015] Furthermore, methods for preprocessing speed data to obtain the data quality coefficient of the nth topological segment include:

[0016] Step a1: Synchronize the speed data from M data sources to the same time period T to obtain speed calibration data;

[0017] Step a2: Standardize the speed calibration data to obtain standard speed data;

[0018] The formula for calculating the speed standard data is:

[0019]

[0020] In the formula, Vb m V represents the speed standard data of the m-th data source; m This represents the speed of the m-th data source. This represents the average speed of M data sources;

[0021] Step a3: Merge the M speed standard data sets and calculate the data quality coefficient of the nth topological segment. The calculation formula is as follows:

[0022]

[0023] In the formula, Vx n β represents the data quality coefficient of the nth topological segment. m These are the corresponding weighting factors.

[0024] Furthermore, a preset quality coefficient threshold is established, which includes Zx1 and Zx2, where Zx1 > Zx2; the data quality coefficient is then compared with the preset quality coefficient threshold.

[0025] If Vx n >Zx1 or Vx n If <Zx2, then a data optimization instruction is generated;

[0026] If Zx1≥Vx n If Zx2 is greater than or equal to 2, then a state analysis instruction is generated.

[0027] Furthermore, methods for optimizing the data quality coefficient of the nth topological segment to obtain the speed optimization coefficient include:

[0028] Step b1: Obtain the data quality coefficient VL of the nth topological segment, and construct a time-domain graph of the coefficient with time as the horizontal axis and data quality coefficient as the vertical axis;

[0029] Step b2: Divide the coefficient time-domain graph into equal parts according to the time period to obtain a set of real-time fluctuation coefficients. The set of real-time fluctuation coefficients includes R real-time coefficient fluctuation graphs, where R is an integer greater than zero.

[0030] Step b3: Extract the r-th real-time coefficient fluctuation graph from the real-time fluctuation coefficient set, where r is a positive integer greater than zero and the initial value of r is 1;

[0031] Step b4: Extract the set of standard coefficient fluctuation graphs for the same time period, calculate the similarity between the real-time coefficient fluctuation graph and the standard coefficient fluctuation graph. If the similarity between the real-time coefficient fluctuation graph and the standard coefficient fluctuation graph is greater than or equal to the preset similarity threshold, mark the real-time coefficient fluctuation graph as a normal coefficient waveform and jump to step b5; if the similarity between the real-time coefficient fluctuation graph and the standard coefficient fluctuation graph is less than the preset similarity threshold, jump directly to step b5.

[0032] Step b5: Let r = r + 1, and jump back to step b3;

[0033] Step b6: Repeat steps b3 to b5 above until r = R, then end the loop and obtain multiple normal coefficient waveforms;

[0034] Step b7: Extract the similarity corresponding to each normal coefficient waveform, and use the data quality coefficient corresponding to the normal coefficient waveform with the highest similarity as the speed optimization coefficient.

[0035] Furthermore, the road feature data includes road segment length, total lane width, total number of actual vehicles, and maximum vehicle density;

[0036] Methods for generating the road load factor for time period T based on road characteristic data include:

[0037] The road segment length, total lane width, total number of actual vehicles, and maximum vehicle density in the road feature data are respectively labeled as Ld. n Wz n Cm n and Cs n ;

[0038] The road segment length, total lane width, actual total number of vehicles, and maximum vehicle density are formalized into formulas, and the road load factor is calculated. The calculation formula is as follows:

[0039]

[0040] In the formula, Dlm n Ld represents the road load factor of the nth topological segment. n Wz represents the length of the nth topological segment. n Cm represents the total lane width of the nth topological segment. n Let Cs represent the maximum vehicle density of the nth topological segment. n Let ln(·) represent the actual number of vehicles on the nth topological segment, and ln(·) represent the logarithmic function with base e.

[0041] Furthermore, the method for generating the state prediction model includes:

[0042] Step c1: Obtain the time series set of road load factor for the nth topological segment;

[0043] Step c2: Preset the time step B, sliding step K, and sliding window length C; convert the historical road load factor in the road load factor time series set into multiple training samples using the sliding window method; use the training samples as input to the state prediction model; use the predicted road load factor after the time step B as output; use the road load factor of each training sample as the prediction target; use the prediction accuracy as the training target to train the state prediction model; generate a state prediction model that predicts the road load factor in the time period T+a based on the historical road load factor in the road load factor time series set; wherein, the state prediction model is an RNN neural network model.

[0044] Furthermore, methods for obtaining the time series set of road load factor include:

[0045] Extract the road load factor from the historical traffic status monitoring process of the nth topological road segment and mark it as the historical road load factor. Construct a time series set of road load factors from the extracted historical road load factors. The time series set of road load factors includes i historical road load factors. The time intervals between the acquisition of the i historical road load factors are equal, and the i historical road load factors correspond to a time period.

[0046] Furthermore, methods for obtaining traffic condition warning information include:

[0047] Step e1: Obtain the road load factor of N topological road segments within the time period T+a, and obtain speed data;

[0048] Step e2: Combine the road load factor and speed data into a feature vector, and input the feature vector into the pre-built traffic early warning model to obtain traffic condition early warning information.

[0049] Furthermore, the traffic early warning model is obtained based on early warning training data, which includes early warning feature data and its corresponding traffic status early warning information; the early warning feature data includes road load factor and speed data;

[0050] The method for obtaining traffic condition early warning information from the early warning training data includes:

[0051] Step d1: Extract the road load factor of the nth topological segment within the time period T+a; simultaneously obtain the speed data of the nth topological segment within the time period T+a;

[0052] Step d2: Set H road load factor intervals and set H traffic status warning messages corresponding to the H road load factor intervals; each road load factor interval is associated with and bound to one and only one traffic status warning message.

[0053] Step d3: Compare the road load factor of the nth topological segment with the road load factor interval of each road to obtain the road load factor interval into which the road load factor of the nth topological segment falls;

[0054] Step d4: Based on the road load factor range into which the road load factor of the nth topological segment falls, classify the road load factor of the nth topological segment into the corresponding traffic status warning information; and let n = n + 1, then jump back to step d1;

[0055] Step d5: Repeat steps d1 to d4 above until n = N, at which point the loop ends, and each road load factor is sequentially assigned to H traffic condition warning messages.

[0056] Step d6: Adjust traffic status warning information based on the speed data of N topological road segments.

[0057] Furthermore, methods for adjusting traffic condition warning information based on speed data from N topological road segments include:

[0058] A preset speed threshold is set, and the speed data of each topological segment is compared with the preset speed threshold.

[0059] When the speed data is greater than or equal to the preset speed threshold, the corresponding road load factor will be reduced by one level and adjusted to the next road load factor range.

[0060] When the speed data is less than the preset speed threshold, the corresponding road load factor will be increased by one level and adjusted to the previous road load factor range.

[0061] Furthermore, the methods for generating traffic warning models include:

[0062] Each set of early warning feature data is combined into a feature vector. The elements of all feature vectors are used as input to a machine learning model. The machine learning model outputs the traffic condition early warning information predicted by each set of early warning feature data, and uses the actual traffic condition early warning information corresponding to each set of early warning feature data as the prediction target. The training objective is to minimize the sum of the prediction accuracies of all predicted traffic condition early warning information. The machine learning model is trained until the sum of prediction accuracies converges, at which point training stops. The trained machine learning model is used as a traffic early warning model, which is a deep neural network model or a deep belief network model.

[0063] Secondly, the present invention provides an intelligent connected traffic condition monitoring system based on multi-source data fusion, used to implement the aforementioned intelligent connected traffic condition monitoring method based on multi-source data fusion, comprising:

[0064] The data preprocessing module is used to acquire speed data from M data sources in the nth topological segment within a time period T, and to preprocess the speed data to obtain the data quality coefficient of the nth topological segment.

[0065] The judgment module is used to analyze the data quality coefficient of the nth topological segment and determine whether the M data sources meet the requirements. If they do not meet the requirements, a data optimization instruction is generated; if they do meet the requirements, a status analysis instruction is generated.

[0066] The optimization module is used to receive data optimization instructions, optimize the data quality coefficient of the nth topology segment to obtain the speed optimization coefficient, and convert the data optimization instructions into state analysis instructions.

[0067] The analysis module is used to receive state analysis instructions, obtain road feature data of the nth topological segment, generate the road load factor within time period T based on the road feature data, and input the generated road load factor into the pre-generated state prediction model to obtain the road load factor within time period T+a.

[0068] The matching module is used to extract the road load factor of the nth topological road segment in the time period T+a, and match the nth topological road segment with the traffic status warning information based on the road load factor to obtain the corresponding status warning information.

[0069] The technical effects and advantages of this invention are as follows:

[0070] 1. This invention analyzes speed data from M data sources and calculates a data quality coefficient, helping to ensure the accuracy and reliability of the input data. When a data source does not meet requirements, a data optimization instruction is generated to avoid the impact of low-quality data on the system. Through a real-time data quality detection and optimization mechanism, interference from data noise or missing data is reduced, improving the accuracy of the fusion model. This enhances the system's robustness and adaptability, adapting to data fluctuations in different traffic scenarios, while reducing the false alarm rate caused by data anomalies and improving the stability of traffic condition prediction.

[0071] 2. This invention enables the system to flexibly adjust its operation process according to the data situation by converting between data optimization instructions and state analysis instructions, ensuring that state analysis and prediction are based on high-quality data and improving the system's execution efficiency. Attached Figure Description

[0072] Figure 1 This is a flowchart of the intelligent connected traffic condition monitoring method based on multi-source data fusion in Example 1;

[0073] Figure 2 This is a flowchart of the method for obtaining traffic status early warning information from the early warning training data in Example 1;

[0074] Figure 3 This is a schematic diagram of the structure of the intelligent connected traffic condition monitoring system with multi-source data fusion in Example 2;

[0075] Figure 4 This is a schematic diagram of an electronic device according to Example 3. Detailed Implementation

[0076] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0077] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0078] It should be understood that although terms such as "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and a similar second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated items listed.

[0079] Example 1

[0080] Please see Figure 1 As shown in the figure, this embodiment discloses a method for intelligent connected traffic status monitoring based on multi-source data fusion, the method comprising:

[0081] Step 1: Within time period T, acquire speed data from M data sources in the nth topological segment, preprocess the speed data to obtain the data quality coefficient of the nth topological segment.

[0082] It should be noted that T and M are both integers greater than zero, n = 1, 2, ..., N, where N is the total number of topological segments.

[0083] It should be understood that the aforementioned topological road segments represent the division of the target city's transportation network into N topological road segments according to a partitioning rule. This partitioning rule includes geographical location, road attributes, and traffic flow direction, etc., and is specifically determined by those skilled in the art based on actual circumstances, without limitation. The data sources include, but are not limited to, road sensors, GPS, collected data, traffic cameras, checkpoint data, floating car data, OSM systems, and POI systems. Because the quality of speed information obtained from different data sources is inconsistent, it can easily lead to data delays. Therefore, speed data preprocessing is necessary to improve data quality.

[0084] It should be noted that the speed data refers to the average speed of the vehicle on the nth topological road segment. The time period T is a preset time span based on the data collection needs of those skilled in the art. For example, the preset time span contains Y values, where Y is a positive integer. Further, assuming the time interval is 00:00-00:30 and the preset time span is 5 minutes, the speed data packet within the preset time span contains speed values ​​between 00:00-00:05, 00:05-00:10, 00:10-00:15, ..., 00:25-00:30.

[0085] In practice, methods for preprocessing speed data to obtain the data quality coefficient of the nth topological segment include:

[0086] Step a1: Synchronize the speed data from M data sources to the same time period T to obtain speed calibration data.

[0087] It should be noted that the speed data was timestamped beforehand to ensure that the speed data from different sources were consistent in time.

[0088] Step a2: Standardize the speed calibration data to obtain standard speed data; the formula for calculating the standard speed data is:

[0089]

[0090] In the formula, Vb m V represents the speed standard data of the m-th data source; m This represents the speed of the m-th data source. This represents the average speed of M data sources.

[0091] Step a3: Merge the M speed standard data sets and calculate the data quality coefficient of the nth topological segment. The calculation formula is as follows:

[0092]

[0093] In the formula, Vxn The data quality coefficient of the nth topological segment is represented by βm, which is the corresponding weighting factor. βm is a constant greater than zero, and the weighting factors are all set by those skilled in the art based on experience.

[0094] Step 2: Analyze the data quality coefficient of the nth topological segment to determine whether the M data sources meet the requirements. If they do not meet the requirements, generate a data optimization instruction; if they do meet the requirements, generate a status analysis instruction.

[0095] A preset quality coefficient threshold is set, which includes Zx1 and Zx2, where Zx1 > Zx2; the data quality coefficient is compared with the preset quality coefficient threshold.

[0096] If Vx n >Zx1 or Vx n If <Zx2, a data optimization instruction is generated, indicating that the data quality coefficient of the nth topology path does not meet the data requirements and the speed data needs to be further optimized.

[0097] If Zx1≥Vx n If Zx2 is greater than or equal to 2, a status analysis instruction is generated, indicating that the data quality coefficient of the nth topology path meets the data requirements.

[0098] It should be noted that the preset quality coefficient threshold was determined by those skilled in the art based on actual needs and extensive experimental data. This step, through multi-source data fusion and preprocessing, obtains a unified and accurate data quality coefficient, effectively solving the problem of inconsistent data quality from multiple sources in intelligent transportation systems. This data not only reflects the traffic conditions of a road segment within a specific time period but also provides data support for subsequent traffic prediction, dynamic signal control, and congestion management.

[0099] Step 3: Receive data optimization instructions, optimize the data quality coefficient of the nth topological road segment to obtain the speed optimization coefficient, and convert the data optimization instructions into state analysis instructions to improve the accuracy of traffic state analysis of the nth topological road segment.

[0100] In implementation, methods for optimizing the data quality coefficient of the nth topological segment to obtain the speed optimization coefficient include:

[0101] Step b1: Obtain the data quality coefficient Vx of the nth topological segment. n A time-domain graph of coefficients is constructed with time as the horizontal axis and data quality coefficients as the vertical axis.

[0102] Step b2: Divide the coefficient time-domain graph into equal parts according to the time period to obtain a set of real-time fluctuation coefficients. The set of real-time fluctuation coefficients includes R real-time coefficient fluctuation graphs, where R is an integer greater than zero.

[0103] Step b3: Extract the r-th real-time coefficient fluctuation graph from the real-time fluctuation coefficient set, where r is a positive integer greater than zero and the initial value of r is 1;

[0104] Step b4: Extract the set of standard coefficient fluctuation graphs for the same time period, calculate the similarity between the real-time coefficient fluctuation graph and the standard coefficient fluctuation graph. If the similarity between the real-time coefficient fluctuation graph and the standard coefficient fluctuation graph is greater than or equal to the preset similarity threshold, mark the real-time coefficient fluctuation graph as a normal coefficient waveform and jump to step b5; if the similarity between the real-time coefficient fluctuation graph and the standard coefficient fluctuation graph is less than the preset similarity threshold, do not mark it and jump directly to step b5.

[0105] It should be noted that the system database contains a set of standard coefficient fluctuation charts for multiple time periods. Each time period is associated with a standard coefficient fluctuation chart, which reflects the coefficient fluctuation of the nth topological segment under the condition that there are no abnormalities in the speed data preprocessing of M data sources.

[0106] Step b5: Let r = r + 1, and jump back to step b3;

[0107] Step b6: Repeat steps b3 to b5 above until r = R, then end the loop and obtain multiple normal coefficient waveforms;

[0108] Step b7: Extract the similarity corresponding to each normal coefficient waveform, and use the data quality coefficient corresponding to the normal coefficient waveform with the highest similarity as the speed optimization coefficient.

[0109] It should be noted that the similarity algorithm includes cosine similarity algorithm or Euclidean distance algorithm, etc.; the real-time coefficient fluctuation map and the standard coefficient fluctuation map need to be preprocessed before comparison, and the preprocessing includes image enhancement, image denoising and image segmentation.

[0110] Step 4: Receive the state analysis instruction, obtain the road feature data of the nth topological segment, generate the road load factor within the time period T based on the road feature data, and input the generated road load factor into the pre-generated state prediction model to obtain the road load factor within the time period T+a.

[0111] The road feature data includes road segment length, total lane width, total number of actual vehicles, and maximum vehicle density.

[0112] It should be noted that the road segment length and total lane width are obtained through real-time measurement along the road using GPS devices, while the actual total number of vehicles and maximum vehicle density are obtained through automatic identification and data extraction by cameras. The methods described above for obtaining road segment length, total lane width, actual total number of vehicles, and maximum vehicle density are merely illustrative examples; any other existing technology (such as LiDAR, drone aerial photography, GIS systems, etc.) can be used to collect road feature data. The methods for collecting road feature data are flexible and diverse, and the specific selection can be adjusted and optimized according to the application scenario requirements and accuracy requirements.

[0113] In practice, methods for generating the road load factor within time period T based on road characteristic data include:

[0114] The road segment length, total lane width, total number of actual vehicles, and maximum vehicle density in the road feature data are respectively labeled as Ld. n Wz n Cm n and Cs n .

[0115] The road segment length, total lane width, actual total number of vehicles, and maximum vehicle density are formalized into formulas, and the road load factor is calculated. The calculation formula is as follows:

[0116]

[0117] In the formula, Dlm n Ld represents the road load factor of the nth topological segment. n Wz represents the length of the nth topological segment. n Cm represents the total lane width of the nth topological segment. n Let Cs represent the maximum vehicle density of the nth topological segment. n Let ln(·) represent the actual number of vehicles on the nth topological segment, and ln(·) represent the logarithmic function with base e.

[0118] It should be noted that the higher the road load factor, the more congested the road segment.

[0119] In implementation, the method for generating the state prediction model includes:

[0120] Step c1: Obtain the time series set of road load factor for the nth topological segment;

[0121] Methods for obtaining the time series set of road load factor include:

[0122] Extract the road load factor from the historical traffic status monitoring process of the nth topological road segment from the system database and mark it as the historical road load factor. Construct a time series set of road load factors from the extracted historical road load factors. The time series set of road load factors includes i historical road load factors. The time interval between the acquisition of the i historical road load factors is equal and the i historical road load factors correspond to a time period.

[0123] Step c2: Based on the actual experience of the staff, preset the time step B, sliding step K, and sliding window length C; convert the historical road load factor in the road load factor time series set into multiple training samples using the sliding window method; use the training samples as input to the state prediction model, and use the predicted road load factor after the time step B as the output; use the road load factor of each training sample as the prediction target, and use the prediction accuracy as the training target to train the state prediction model; generate a state prediction model to predict the road load factor in the time period T+a based on the historical road load factor in the road load factor time series set; wherein, the state prediction model is an RNN neural network model; by using the road load factor time series set, predict the road load factor of the nth topological road segment in the time period T+a, thereby realizing the prediction of the road load factor of the nth topological road segment in the time period T+a based on the road load factor in the time period T.

[0124] It should be noted that the time series set of road load factor is divided into sliding windows of equal size. The road load factor in each window is taken as a sample, and the road load factor in the time period T+a of that window is taken as a digital label. It should be noted that one sample corresponds to one digital label, and one sample and its corresponding digital label constitute a set of load training data. Multiple sets of load training data constitute the coefficient training set.

[0125] To further illustrate, suppose the time series set A of road load factors contains 10 sets of historical road load factors, A = {A1, A2, A3, A4, A5, ..., A...} 10}, A a For the historical road load factor in group a, multiple training samples are constructed using a sliding window. The prediction time step B is set to 1, the sliding window length H to 5, and the sliding step C to 1. Each generated training sample contains 5 consecutive historical road load factors. The next road load factor among these 5 consecutive historical road load factors is then used as the prediction target. For example:

[0126] {A1, A2, A3, A4, A5} are used as the full-load training data, and their corresponding prediction target is A6;

[0127] {A2, A3, A4, A5, A6} are used as the full-load training data, and their corresponding prediction target is A7;

[0128] Similarly, a state prediction model is used to train and predict the road load factor of the nth topological segment in the time period T+a.

[0129] By collecting road characteristic data and analyzing the resulting road load factor, and using a pre-built state prediction model, the road load factor of the nth topological road segment within the time period T+a is predicted. This process not only improves prediction accuracy and traffic flow management, but also supports managers in formulating and implementing traffic control plans in a timely manner.

[0130] Step 5: Extract the road load factor of the nth topological road segment in the time period T+a, and match the nth topological road segment with the traffic status warning information according to the road load factor to obtain the corresponding traffic status warning information.

[0131] In practice, methods for obtaining traffic condition warning information include:

[0132] Step e1: Obtain the road load factor of N topological road segments within the time period T+a, and obtain speed data;

[0133] Step e2: Combine the road load factor and speed data into a feature vector, and input the feature vector into the pre-built traffic early warning model to obtain traffic condition early warning information.

[0134] The traffic early warning model is obtained based on early warning training data, which includes early warning feature data and its corresponding traffic status early warning information; the early warning feature data includes road load factor and speed data.

[0135] Please see Figure 2 As shown, in one specific embodiment, the method for obtaining traffic status early warning information in the early warning training data includes:

[0136] Step d1: Extract the road load factor of the nth topological segment within the time period T+a; simultaneously obtain the speed data of the nth topological segment within the time period T+a;

[0137] Step d2: Set H road load factor intervals and set H traffic status warning messages corresponding to the H road load factor intervals; each road load factor interval is associated with and bound to one and only one traffic status warning message.

[0138] It should be noted that the H road load factor ranges are set in a step-by-step order from large to small, and the corresponding H traffic condition warning messages are also arranged in the same order.

[0139] Step d3: Compare the road load factor of the nth topological segment with the road load factor interval of each road to obtain the road load factor interval into which the road load factor of the nth topological segment falls;

[0140] Step d4: Based on the road load factor range into which the road load factor of the nth topological segment falls, classify the road load factor of the nth topological segment into the corresponding traffic status warning information; and let n = n + 1, then jump back to step d1;

[0141] Step d5: Repeat steps d1 to d4 above until n = N, at which point the loop ends, and each road load factor is sequentially assigned to H traffic condition warning messages.

[0142] Step d6: Adjust H traffic status warning messages based on the speed data of N topological road segments.

[0143] In a preferred embodiment, the method for adjusting H traffic condition warning messages based on speed data of N topological road segments includes:

[0144] A preset speed threshold is set, and the speed data of each topological segment is compared with the preset speed threshold.

[0145] When the speed data is greater than or equal to the preset speed threshold, the corresponding road load factor will be reduced by one level and adjusted to the next road load factor range.

[0146] When the speed data is less than the preset speed threshold, the corresponding road load factor will be increased by one level and adjusted to the previous road load factor range.

[0147] It should be noted that by dynamically comparing speed data with preset speed thresholds, the road load factor can be adjusted in real time, which helps improve the accuracy and flexibility of traffic condition warnings. Traffic condition warning information is updated promptly as vehicle speed changes, avoiding misjudgments caused by static assessments. This adjustment mechanism can also more accurately reflect traffic flow, providing timely and effective decision support for traffic management departments, optimizing traffic flow management, reducing congestion risks, and improving road traffic efficiency.

[0148] In practice, the methods for generating traffic early warning models include:

[0149] Each set of early warning feature data is combined into a feature vector. All elements of the feature vectors serve as input to a machine learning model. The machine learning model outputs the traffic condition warning information predicted by each set of early warning feature data, uses the actual traffic condition warning information corresponding to each set of early warning feature data as the prediction target, and minimizes the sum of the prediction accuracies of all predicted traffic condition warning information as the training objective. The formula for calculating the prediction accuracy is: d v =(c v -p v ) 2 Where V is the number of each set of early warning feature data, d v To improve prediction accuracy, c v For the predicted traffic condition warning information corresponding to the v-th group of warning feature data, p v The actual traffic condition warning information corresponding to the v-th group of warning feature data is used; the machine learning model is trained until the sum of the prediction accuracies converges and training stops; the trained machine learning model is used as the traffic warning model, wherein the machine learning model is a deep neural network model or a deep belief network model.

[0150] This embodiment analyzes speed data from M data sources to calculate a data quality coefficient, helping to ensure the accuracy and reliability of the input data. When a data source does not meet requirements, a data optimization instruction is generated to avoid the impact of low-quality data on the system. Through a real-time data quality detection and optimization mechanism, interference from data noise or missing data is reduced, improving the accuracy of the fusion model. This enhances the system's robustness and adaptability, adapting to data fluctuations in different traffic scenarios, while reducing the false alarm rate caused by data anomalies and improving the stability of traffic condition prediction.

[0151] This embodiment enables the system to flexibly adjust its operation process based on data conditions by converting between data optimization instructions and state analysis instructions, ensuring that state analysis and prediction are based on high-quality data and improving the system's execution efficiency.

[0152] The formulas mentioned above are all dimensionless calculations, derived from software simulation using a large amount of data to approximate the real situation. The weighting factors and preset thresholds in the formulas are set by those skilled in the art based on the actual situation or obtained through large-scale data simulation. The size of the weighting factor is a specific value obtained by quantifying each parameter to facilitate subsequent comparison. The size of the weighting factor depends on the amount of sample data and the processing coefficients initially set by those skilled in the art for each set of sample data. As long as it does not affect the proportional relationship between the parameter and the quantified value, it is acceptable.

[0153] Example 2

[0154] Please see Figure 3 As shown, this embodiment provides an intelligent connected traffic status monitoring system with multi-source data fusion. The system is applied in a cloud server and includes: a data preprocessing module, a judgment module, an optimization module, an analysis module, and a matching module. The above modules are connected by wired and / or wireless means to realize data transmission between them.

[0155] The data preprocessing module is used to acquire speed data from M data sources in the nth topological segment within a time period T, and to preprocess the speed data to obtain the data quality coefficient of the nth topological segment.

[0156] The judgment module is used to analyze the data quality coefficient of the nth topological segment and determine whether the M data sources meet the requirements. If they do not meet the requirements, a data optimization instruction is generated; if they do meet the requirements, a status analysis instruction is generated.

[0157] The optimization module is used to receive data optimization instructions, optimize the data quality coefficient of the nth topology segment to obtain the speed optimization coefficient, and convert the data optimization instructions into state analysis instructions.

[0158] The analysis module is used to receive state analysis instructions, obtain road feature data of the nth topological segment, generate the road load factor within time period T based on the road feature data, and input the generated road load factor into the pre-generated state prediction model to obtain the road load factor within time period T+a.

[0159] The matching module is used to extract the road load factor of the nth topological road segment in the time period T+a, and match the nth topological road segment with the traffic status warning information based on the road load factor to obtain the corresponding status warning information.

[0160] Example 3

[0161] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device of this embodiment includes: a processor 11 ( Figure 4 The diagram shows only one processor, memory 12, and computer program 13 stored in memory 12 and executable on processor 11. When processor 11 executes computer program 13, it implements the steps described in the above-described embodiment of the intelligent connected traffic state monitoring method for multi-source data fusion, for example... Figure 1 Steps 1 to 5 are shown. Alternatively, when the processor 11 executes the computer program 13, it implements the functions of each module / unit in the above system embodiments, for example... Figure 3 The functions of each module are shown.

[0162] For example, the computer program 13 may be divided into one or more modules / units, which are stored in the memory 12 and executed by the processor 11 to complete this application. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program 13 in the terminal device.

[0163] Those skilled in the art will understand that Figure 4 This is merely an example of an electronic device and does not constitute a limitation on the device. It may include more or fewer components than illustrated, or combine certain components, or use different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc. The processor 11 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0164] The memory 12 can be an internal storage unit of the terminal device, such as a hard drive or RAM. The memory 12 can also be an external storage device of the terminal device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or FlashCard. Furthermore, the memory 12 can include both internal and external storage units. The memory 12 is used to store the computer program and other programs and data required by the terminal device. The memory 12 can also be used to temporarily store data that has been output or will be output.

[0165] Example 4

[0166] This embodiment provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the intelligent connected traffic status monitoring method of multi-source data fusion as described in Embodiment 1.

[0167] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0168] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent connected traffic condition monitoring based on multi-source data fusion, characterized in that, include: Within time period T, acquire speed data from M data sources in the nth topological segment, preprocess the speed data to obtain the data quality coefficient of the nth topological segment; Analyze the data quality coefficient of the nth topological segment to determine whether the M data sources meet the requirements. If they do not meet the requirements, generate a data optimization instruction; if they do meet the requirements, generate a status analysis instruction. Receive data optimization instructions, optimize the data quality coefficient of the nth topology segment to obtain the speed optimization coefficient, and convert the data optimization instructions into state analysis instructions; Methods for optimizing the data quality coefficient of the nth topological segment to obtain the speed optimization coefficient include: Step b1: Obtain the data quality coefficient of the nth topological segment. A time-domain graph of coefficients is constructed with time as the horizontal axis and data quality coefficients as the vertical axis. Step b2: Divide the coefficient time-domain graph into equal parts according to the time period to obtain a set of real-time fluctuation coefficients. The set of real-time fluctuation coefficients includes R real-time coefficient fluctuation graphs, where R is an integer greater than zero. Step b3: Extract the r-th real-time coefficient fluctuation graph from the real-time fluctuation coefficient set, where r is a positive integer greater than zero and the initial value of r is 1; Step b4: Extract the set of standard coefficient fluctuation graphs for the same time period, calculate the similarity between the real-time coefficient fluctuation graph and the standard coefficient fluctuation graph. If the similarity between the real-time coefficient fluctuation graph and the standard coefficient fluctuation graph is greater than or equal to the preset similarity threshold, mark the real-time coefficient fluctuation graph as a normal coefficient waveform and jump to step b5; if the similarity between the real-time coefficient fluctuation graph and the standard coefficient fluctuation graph is less than the preset similarity threshold, jump directly to step b5. Step b5: Let r = r + 1, and jump back to step b3; Step b6: Repeat steps b3 to b5 above until r=R, then end the loop and obtain multiple normal coefficient waveforms; Step b7: Extract the similarity corresponding to each normal coefficient waveform, and use the data quality coefficient corresponding to the normal coefficient waveform with the highest similarity as the speed optimization coefficient; Receive the state analysis instruction, obtain the road feature data of the nth topology segment, generate the road load factor within the time period T based on the road feature data, and input the generated road load factor into the pre-generated state prediction model to obtain the road load factor within the time period T+a. Extract the road load factor of the nth topological road segment within the time period T+a, and match the nth topological road segment with the traffic status warning information based on the road load factor to obtain the corresponding status warning information.

2. The intelligent connected traffic condition monitoring method based on multi-source data fusion according to claim 1, characterized in that, Methods for preprocessing speed data to obtain the data quality coefficient of the nth topological segment include: Step a1: Synchronize the speed data from M data sources to the same time period T to obtain speed calibration data; Step a2: Standardize the speed calibration data to obtain standard speed data; The formula for calculating the speed standard data is: ； ； In the formula, This represents the speed standard data of the m-th data source; This represents the speed of the m-th data source. This represents the average speed of M data sources; Step a3: Merge the M speed standard data sets and calculate the data quality coefficient of the nth topological segment. The calculation formula is as follows: ； In the formula, This represents the data quality coefficient of the nth topological segment. These are the corresponding weighting factors.

3. The intelligent connected traffic condition monitoring method based on multi-source data fusion according to claim 2, characterized in that, A preset quality coefficient threshold, the quality coefficient threshold including and ,in > Compare the data quality coefficient with the preset quality coefficient threshold; like > or Then, a data optimization instruction will be generated; like Then, a state analysis instruction is generated.

4. The intelligent connected traffic condition monitoring method based on multi-source data fusion according to claim 1, characterized in that, The road feature data includes road segment length, total lane width, total number of actual vehicles, and maximum vehicle density; Methods for generating the road load factor for time period T based on road characteristic data include: The road segment length, total lane width, total number of actual vehicles, and maximum vehicle density in the road feature data are respectively labeled as follows: , , and ; The road segment length, total lane width, actual total number of vehicles, and maximum vehicle density are formalized into formulas, and the road load factor is calculated. The calculation formula is as follows: ； In the formula, This represents the road load factor of the nth topological segment. This represents the length of the nth topological segment. This represents the total lane width of the nth topological segment. This represents the maximum vehicle density of the nth topological segment. This represents the actual number of vehicles on the nth topological road segment. This represents the logarithmic function with base e.

5. The intelligent connected traffic condition monitoring method based on multi-source data fusion according to claim 4, characterized in that, The method for generating the state prediction model includes: Step c1: Obtain the time series set of road load factor for the nth topological segment; Step c2: Preset the time step B, sliding step K, and sliding window length C; convert the historical road load factor in the road load factor time series set into multiple training samples using the sliding window method; use the training samples as input to the state prediction model; use the predicted road load factor after the time step B as output; use the road load factor of each training sample as the prediction target; use the prediction accuracy as the training target to train the state prediction model; generate a state prediction model that predicts the road load factor in the time period T+a based on the historical road load factor in the road load factor time series set; wherein, the state prediction model is an RNN neural network model.

6. The intelligent connected traffic condition monitoring method based on multi-source data fusion according to claim 5, characterized in that, Methods for obtaining time series sets of road load factor include: Extract the road load factor from the historical traffic status monitoring process of the nth topological road segment and mark it as the historical road load factor. Construct a time series set of road load factors from the extracted historical road load factors. The time series set of road load factors includes i historical road load factors. The time intervals between the acquisition of the i historical road load factors are equal, and the i historical road load factors correspond to a time period.

7. The intelligent connected traffic condition monitoring method based on multi-source data fusion according to claim 6, characterized in that, Methods for obtaining traffic condition warning information include: Step e1: Obtain the road load factor of N topological road segments within the time period T+a, and obtain speed data; Step e2: Combine the road load factor and speed data into a feature vector, and input the feature vector into the pre-built traffic early warning model to obtain traffic condition early warning information.

8. The intelligent connected traffic condition monitoring method based on multi-source data fusion according to claim 7, characterized in that, The traffic early warning model is obtained based on early warning training data, which includes early warning feature data and its corresponding traffic status early warning information. The early warning feature data includes road load factor and speed data; The method for obtaining traffic condition early warning information from the early warning training data includes: Step d1: Extract the road load factor of the nth topological segment within the time period T+a; simultaneously obtain the speed data of the nth topological segment within the time period T+a; Step d2: Set H road load factor intervals and set H traffic status warning messages corresponding to the H road load factor intervals; each road load factor interval is associated with and bound to one and only one traffic status warning message. Step d3: Compare the road load factor of the nth topological segment with the road load factor interval of each road to obtain the road load factor interval into which the road load factor of the nth topological segment falls; Step d4: Based on the road load factor range into which the road load factor of the nth topological segment falls, classify the road load factor of the nth topological segment into the corresponding traffic status warning information; and let n=n+1, then jump back to step d1; Step d5: Repeat steps d1 to d4 above until n=N, at which point the loop ends, and each road load factor is sequentially assigned to H traffic condition warning messages. Step d6: Adjust traffic status warning information based on the speed data of N topological road segments.

9. The intelligent connected traffic condition monitoring method based on multi-source data fusion according to claim 8, characterized in that, Methods for adjusting traffic condition warning information based on speed data from N topological road segments include: A preset speed threshold is set, and the speed data of each topological segment is compared with the preset speed threshold. When the speed data is greater than or equal to the preset speed threshold, the corresponding road load factor will be reduced by one level and adjusted to the next road load factor range. When the speed data is less than the preset speed threshold, the corresponding road load factor will be increased by one level and adjusted to the previous road load factor range.

10. The intelligent connected traffic condition monitoring method based on multi-source data fusion according to claim 8, characterized in that, Methods for generating traffic early warning models include: Each set of early warning feature data is combined into a feature vector. The elements of all feature vectors are used as input to a machine learning model. The machine learning model outputs the traffic condition early warning information predicted by each set of early warning feature data, and uses the actual traffic condition early warning information corresponding to each set of early warning feature data as the prediction target. The training objective is to minimize the sum of the prediction accuracies of all predicted traffic condition early warning information. The machine learning model is trained until the sum of prediction accuracies converges, at which point training stops. The trained machine learning model is used as a traffic early warning model, which is a deep neural network model or a deep belief network model.

11. A multi-source data fusion intelligent connected traffic condition monitoring system, used to implement the multi-source data fusion intelligent connected traffic condition monitoring method according to any one of claims 1-10, characterized in that, include: The data preprocessing module is used to acquire speed data from M data sources in the nth topological segment within a time period T, and to preprocess the speed data to obtain the data quality coefficient of the nth topological segment. The judgment module is used to analyze the data quality coefficient of the nth topological segment and determine whether the M data sources meet the requirements. If they do not meet the requirements, a data optimization instruction is generated; if they do meet the requirements, a status analysis instruction is generated. The optimization module is used to receive data optimization instructions, optimize the data quality coefficient of the nth topological segment to obtain the speed optimization coefficient, and convert the data optimization instructions into state analysis instructions. The analysis module is used to receive state analysis instructions, obtain road feature data of the nth topological segment, generate the road load factor within time period T based on the road feature data, and input the generated road load factor into the pre-generated state prediction model to obtain the road load factor within time period T+a. The matching module is used to extract the road load factor of the nth topological road segment in the time period T+a, and match the nth topological road segment with the traffic status warning information based on the road load factor to obtain the corresponding status warning information.

Images