Continuous longitudinal slope traffic risk factor intelligent identification method based on deep learning

Abstract

The invention relates to the technical field of traffic data processing, in particular to a continuous longitudinal slope traffic risk factor intelligent identification method based on deep learning. Longitudinal slope characteristics of the continuous longitudinal slope are obtained, and braking characteristics are recognized and obtained according to monitoring data of vehicle braking in the continuous longitudinal slope; a basic traffic risk factor recognizer is obtained, incremental learning data configuration is carried out according to the braking time and the longitudinal slope characteristics, and incremental learning data is obtained; performing incremental learning on the basic traffic risk factor identifier by adopting incremental learning data according to the braking frequency distribution to obtain a plurality of incremental identifiers; vehicle features are collected, and longitudinal slope traffic risk factor information is recognized, output and obtained based on a plurality of incremental recognizers in combination with longitudinal slope features. The characteristics of different continuous longitudinal slope road sections can be effectively adapted, and the longitudinal slope traffic risk factors can be comprehensively and accurately identified.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for intelligent identification of traffic risk factors on continuous longitudinal slopes based on deep learning. Background Technology

[0002] Traditional techniques for identifying traffic risk factors on continuous longitudinal slopes often rely on manual on-site surveys combined with historical accident data statistics, or on simple machine learning models trained with a small amount of fixed sample data. However, these techniques have significant shortcomings. They ignore the dynamic behavior characteristics of vehicles, leading to insufficient data utilization. Furthermore, simple machine learning models, due to their fixed training data, are difficult to adapt to different road sections and have poor adaptability. Moreover, they are affected by subjective human factors, data lag, and the insufficient robustness of single models, resulting in low reliability of risk factor identification results and failing to meet the needs of accurate identification. Summary of the Invention

[0003] This invention addresses the technical problem of existing technologies failing to meet the requirements for accurate identification by providing a deep learning-based intelligent identification method for traffic risk factors on continuous longitudinal slopes.

[0004] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides a method for intelligent identification of traffic risk factors on a continuous longitudinal slope based on deep learning, comprising: acquiring the longitudinal slope features of the continuous longitudinal slope, and identifying braking features based on vehicle braking monitoring data within the continuous longitudinal slope, wherein the braking features include braking frequency distribution and braking time; acquiring a basic traffic risk factor identifier, configuring incremental learning data based on the braking time and longitudinal slope features to obtain incremental learning data; incrementally learning the basic traffic risk factor identifier using the incremental learning data based on the braking frequency distribution to obtain multiple incremental identifiers; collecting vehicle features, combining the longitudinal slope features, and identifying and outputting longitudinal slope traffic risk factor information based on the multiple incremental identifiers.

[0005] Optionally, the longitudinal slope characteristics of the continuous longitudinal slope are obtained, and braking characteristics are identified based on the vehicle braking monitoring data within the continuous longitudinal slope. This includes: obtaining the longitudinal slope characteristics of the continuous longitudinal slope, wherein the longitudinal slope characteristics include the longitudinal slope gradient and longitudinal slope length; obtaining vehicle braking monitoring data within a preset time range within the continuous longitudinal slope, identifying braking time sequences, and obtaining a set of historical braking time sequences; and calculating multiple average frequencies and average total durations of braking based on the set of historical braking time sequences to obtain braking frequency distribution and braking time as braking characteristics.

[0006] Optionally, obtaining a basic traffic risk element identifier includes: collecting a set of sample longitudinal slope features and a set of sample vehicle features based on safety monitoring data of multiple sample longitudinal slopes, and collecting the proportion of longitudinal slope risk events occurring under different sample longitudinal slope features and sample vehicle features, labeling them as a set of sample longitudinal slope traffic risk element information; constructing a basic traffic risk element identifier based on machine learning; using the sample longitudinal slope feature set, sample vehicle feature set, and sample longitudinal slope traffic risk element information set, supervising the training of the basic traffic risk element identifier, completing the training after the test accuracy meets the accuracy threshold, and obtaining a basic traffic risk element identifier that has completed training and testing.

[0007] Optionally, incremental learning data is configured based on the braking time and longitudinal slope characteristics to obtain incremental learning data, including: acquiring a preset incremental learning data volume; acquiring the average braking time of multiple sample longitudinal slopes under monitoring and processing; correcting the preset incremental learning data volume based on the ratio of the braking time to the average braking time to obtain the incremental learning data volume; and collecting the incremental longitudinal slope feature set, incremental vehicle feature set, and incremental longitudinal slope traffic risk element information set of the continuous longitudinal slope with the highest similarity to the longitudinal slope characteristics according to the incremental learning data volume, until the incremental learning data volume is met to obtain incremental learning data.

[0008] Optionally, based on the braking frequency distribution, the basic traffic risk element identifier is incrementally learned using the incremental learning data to obtain multiple incremental identifiers, including: constructing a braking frequency interval based on the minimum and maximum braking frequencies within the braking frequency distribution; processing and obtaining an average braking frequency interval based on monitoring data of multiple sample longitudinal slopes; obtaining a preset number of incremental learning units; correcting the preset number of incremental learning units based on the ratio of the interval lengths of the braking frequency intervals and the average braking frequency intervals to obtain the number of incremental learning units; and incrementally learning the basic traffic risk element identifier using the incremental learning data according to the number of incremental learning units to obtain multiple incremental identifiers.

[0009] Specifically, according to the incremental learning quantity, the basic traffic risk element identifier is incrementally trained using the incremental learning data to obtain multiple incremental identifiers, including: randomly dividing the incremental learning data multiple times according to the incremental learning quantity as the total number of times to obtain multiple random incremental data; and using the multiple random incremental data to incrementally train the basic traffic risk element identifier to obtain multiple incremental identifiers.

[0010] Optionally, vehicle features are collected, and combined with the longitudinal slope features, longitudinal slope traffic risk element information is obtained by identifying and outputting multiple incremental recognizers. This includes: obtaining vehicle features of currently passing vehicles through a road monitoring system, wherein the vehicle features include vehicle length; inputting the vehicle features and longitudinal slope features into multiple incremental recognizers, identifying and outputting multiple longitudinal slope traffic risk element information, and fusing them to obtain longitudinal slope traffic risk element information.

[0011] By implementing this invention, it is possible to acquire the longitudinal slope characteristics of a continuous longitudinal slope and identify braking characteristics based on vehicle braking monitoring data within the continuous longitudinal slope. These braking characteristics include braking frequency distribution and braking time. The longitudinal slope characteristics provide fundamental road environment information for subsequent traffic risk identification; different slope gradients and lengths directly affect vehicle driving status and risk levels. Braking characteristics directly reflect the driving safety of vehicles on the continuous longitudinal slope section. The braking frequency distribution reflects the intensity of vehicle braking behavior, while the braking time reflects the duration of braking behavior. The combination of these two characteristics provides crucial vehicle driving behavior data support for subsequent risk identification, making risk identification more targeted. A basic traffic risk element identifier is obtained. Based on the braking time and longitudinal slope characteristics, incremental learning data is configured to obtain incremental learning data. The basic traffic risk element identifier provides a stable initial model framework for subsequent incremental learning. It is trained on a large number of samples and has a certain general risk identification capability. By correcting the amount of incremental learning data through braking time and longitudinal slope characteristics and collecting similar data, the incremental learning data can better adapt to the actual situation of the current continuous longitudinal slope, avoiding data redundancy or insufficiency, improving the efficiency and accuracy of subsequent incremental learning, and allowing the model to adapt to the characteristics of the current road segment more quickly. Based on the braking frequency distribution, the basic traffic risk element recognizer is incrementally trained using the incremental learning data to obtain multiple incremental recognizers. The number of incremental learners is adjusted based on the braking frequency distribution to match the complexity of vehicle braking behavior on the current road segment. When braking frequency fluctuations are large, the number of learning iterations can be appropriately increased to ensure sufficient model learning. Multiple incremental recognizers are obtained through multiple random data partitioning training, avoiding the randomness that may arise from a single training data partition, enhancing the model's robustness. Multiple recognizers can learn data features from different perspectives, improving the model's ability to identify complex traffic risks. Vehicle features are collected and combined with the longitudinal slope features. Based on multiple incremental recognizers, the output of longitudinal slope traffic risk element information is obtained. The introduction of vehicle features supplements the vehicle dimension information of traffic risk identification. Vehicles of different lengths face different risks when traveling on continuous longitudinal slopes. Combining longitudinal slope features allows for a more comprehensive risk assessment. The identification results of multiple incremental recognizers are fused, integrating the advantages of each recognizer and reducing the possibility of misjudgment or omission that may occur with a single recognizer. This significantly improves the reliability and accuracy of the longitudinal slope traffic risk element identification results, providing a more precise basis for traffic management and safety decision-making.

[0012] In summary, by implementing this invention, it is possible to effectively adapt to the characteristics of different continuous longitudinal slope road sections, comprehensively and accurately identify longitudinal slope traffic risk factors, and provide reliable data support and decision-making basis for road traffic safety management. Attached Figure Description

[0013] Figure 1 A flowchart illustrating a method for intelligent identification of traffic risk factors on continuous longitudinal slopes based on deep learning, provided by this invention. Figure 2 This invention provides a method for intelligent identification of traffic risk factors on continuous longitudinal slopes based on deep learning, which uses incremental learning data to incrementally learn the basic traffic risk factor identifier. Detailed Implementation

[0014] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0016] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0017] Example 1, as Figure 1 As shown, this embodiment of the invention provides a method for intelligent identification of traffic risk factors on continuous longitudinal slopes based on deep learning, including: S100: Obtain the longitudinal slope characteristics of the continuous longitudinal slope, and identify the braking characteristics based on the vehicle braking monitoring data within the continuous longitudinal slope. The braking characteristics include braking frequency distribution and braking time. S200: Obtain the basic traffic risk element identifier, configure incremental learning data based on the braking time and longitudinal slope characteristics, and obtain incremental learning data; S300: Based on the braking frequency distribution, the incremental learning data is used to perform incremental learning on the basic traffic risk element identifier to obtain multiple incremental identifiers; S400: Collects vehicle characteristics, combines them with the longitudinal slope characteristics, and uses multiple incremental recognizers to identify and output information on longitudinal slope traffic risk factors.

[0018] In step S100 of this application embodiment, the longitudinal slope characteristics of the continuous longitudinal slope are obtained, and braking characteristics are identified based on the vehicle braking monitoring data within the continuous longitudinal slope, including: Obtain the longitudinal slope characteristics of a continuous longitudinal slope, including the longitudinal slope gradient and longitudinal slope length; Obtain vehicle braking monitoring data within a preset time range within the continuous longitudinal slope, identify braking time sequences, and obtain a set of historical braking time sequences; Based on the set of historical braking time sequences, multiple average frequencies and average total durations of braking are calculated to obtain the braking frequency distribution and braking time, which serve as braking characteristics.

[0019] In this embodiment of the application, the purpose of step S100 is to provide comprehensive and critical basic data support for the subsequent identification of traffic risk factors on continuous longitudinal slopes. Specifically, it includes two aspects: first, clarifying the core road geometric attributes of continuous longitudinal slopes to define the basic environmental boundaries for risk identification; second, extracting the key driving behavior characteristics of vehicles on the road section to intuitively reflect the real-time safety status of the road section. The combination of the two makes the subsequent risk identification more targeted and accurate.

[0020] Specifically, it is necessary to obtain the longitudinal slope characteristics of the continuous longitudinal slope, namely the longitudinal slope gradient and longitudinal slope length. The longitudinal slope gradient reflects the degree of inclination of the road segment, and the longitudinal slope length reflects the extension range of the continuous sloping road segment. These two parameters are the basis for determining risk-related factors such as vehicle driving resistance and braking load on this road segment. The longitudinal slope characteristics of the continuous longitudinal slope can be obtained by consulting road design drawings and on-site measurements. For example, assuming that a mountain highway has a continuous longitudinal slope segment A, the longitudinal slope characteristics of segment A can be: this segment contains 3 continuous slopes, the first slope is 5%, the second slope is 7%, and the third slope is 4%, the length of the first slope is 800 meters, the second slope is 600 meters, and the third slope is 1000 meters, with a total continuous longitudinal slope length of 2400 meters.

[0021] Furthermore, it is necessary to obtain vehicle braking monitoring data within a preset time range within the continuous longitudinal slope, identify braking time sequences, and obtain a set of historical braking time sequences. That is, select vehicle braking monitoring data within a preset time range within the continuous longitudinal slope, such as the past 1 month or 3 months, identify the specific time of each braking occurrence from the data, and organize it into a set of historical braking time sequences to completely record the temporal distribution of vehicle braking behavior on this road section.

[0022] Specifically, in the example above, high-definition cameras and speed-sensing loop detectors deployed along section A can be used to collect braking data from all vehicles passing through the area over the past three months. This data can then identify the specific time of each braking action, creating a historical braking time sequence. For example, at 8:12:30 on January 5th, a truck with license plate number Gansu A·XXXX1 braked on the first slope, with a braking time of 3 seconds; at 8:15:10 on January 5th, a sedan with license plate number Gansu A·XXXX2 braked on the second slope, with a braking time of 4 seconds, and so on. This historical braking time sequence is compiled to fully record the temporal distribution of vehicle braking behavior on that section of road.

[0023] Furthermore, based on the historical braking time sequence set, it is necessary to calculate multiple average frequencies and average total durations of braking to obtain braking frequency distribution and braking time as braking characteristics.

[0024] Optionally, in the above example, the average braking frequency can be calculated according to the slope segments. The first slope segment experiences an average of 28 braking events per day, the second slope segment an average of 35 braking events per day, and the third slope segment an average of 18 braking events per day, forming a "braking frequency distribution for slopes of different gradients". In another implementation, the average braking frequency can also be calculated by time period, i.e., in section A, there are an average of 32 braking events per hour during the morning peak (7:00-9:00), 15 braking events per hour during the off-peak (9:00-17:00), and 28 braking events per hour during the evening peak (17:00-19:00), forming a "braking frequency distribution for different time periods".

[0025] Then, the duration of all braking actions is statistically analyzed, and the average total duration is calculated as a braking time feature. For example, the average braking duration of a truck on this road segment is 4.2 seconds, and the average braking duration of a car is 2.1 seconds. The average total braking duration of all vehicles is 3.0 seconds, which is used as the "braking time" feature of this road segment.

[0026] In step S200 of this application embodiment, obtaining the basic traffic risk element identifier includes: Based on safety monitoring data of multiple sample longitudinal slopes, sample longitudinal slope feature sets and sample vehicle feature sets were collected, and the proportion of longitudinal slope risk events occurring under different sample longitudinal slope features and sample vehicle features was collected and labeled as sample longitudinal slope traffic risk element information set. A basic traffic risk element identifier is constructed based on machine learning. Using the sample longitudinal slope feature set, sample vehicle feature set, and sample longitudinal slope traffic risk element information set, the basic traffic risk element identifier is trained under supervision. Training is completed after the test accuracy meets the accuracy threshold, and the trained and tested basic traffic risk element identifier is obtained.

[0027] In this embodiment, the purpose of step S200 is to construct an initial model framework with general traffic risk identification capabilities, namely a basic traffic risk element identifier. This basic traffic risk element identifier can establish the correlation between "longitudinal slope features + vehicle features" and "longitudinal slope risk events" based on historical sample data, providing a stable and reliable basic model for subsequent incremental learning of specific continuous longitudinal slope road sections, avoiding the inefficiency and insufficient accuracy problems caused by training the model from scratch.

[0028] First, it is necessary to collect and label sample data for training the basic traffic risk element identifier. That is, based on the safety monitoring data of multiple sample longitudinal slopes, collect sample longitudinal slope feature sets and sample vehicle feature sets, and collect the proportion of longitudinal slope risk events under different sample longitudinal slope features and sample vehicle features, and label them as sample longitudinal slope traffic risk element information sets.

[0029] Specifically, multiple continuous longitudinal slope sections of different types need to be selected as sample longitudinal slopes, covering sections with different slopes, lengths, and traffic flows to ensure the diversity and representativeness of the samples. Then, for each sample longitudinal slope, parameters such as longitudinal slope and longitudinal slope length are obtained through road design data and on-site measurements to form a sample longitudinal slope feature set; Using road monitoring equipment such as cameras and vehicle detectors, the characteristics of vehicles passing through each sample longitudinal slope are collected. The main characteristics of the vehicles are vehicle length, and if the monitoring equipment supports this, they can be further included, such as vehicle type (classified by truck, car, bus, etc.) and vehicle load, to form a sample vehicle feature set.

[0030] For each sample longitudinal slope, statistical analysis is conducted on longitudinal slope risk events that occur within a certain period, such as one year, including rear-end collisions and brake failures. The proportion of risk events occurring under different combinations of "sample longitudinal slope characteristics + sample vehicle characteristics" is calculated. For example, under the combination of "7% slope + 2000 meters length + truck (12 meters length)," the proportion of risk events is 8%. These proportions are labeled as a set of traffic risk element information for the sample longitudinal slope.

[0031] Furthermore, a basic traffic risk factor identifier needs to be constructed based on machine learning. Depending on the task type of this identifier, a neural network algorithm can be used.

[0032] The basic traffic risk element identifier mainly consists of three parts: the input layer, the hidden layer, and the output layer.

[0033] In the input layer, nodes are set according to the feature types of the sample longitudinal slope feature set and the sample vehicle feature set. For example, the number of input layer nodes is set to 5, corresponding to 5 types of core feature parameters: sample longitudinal slope features, including longitudinal slope gradient and longitudinal slope length; sample vehicle features, including vehicle length, vehicle load and vehicle type.

[0034] Two hidden layers are set up: the first hidden layer contains 12 neurons and the second hidden layer contains 8 neurons. Through two layers of nonlinear transformation, the deep correlation between input features and output risk information is fully explored.

[0035] The number of nodes in the output layer is set to 1, which corresponds to the core content of the "sample longitudinal slope traffic risk element information set", namely the "proportion of longitudinal slope risk events" under different sample longitudinal slope characteristics and sample vehicle characteristics. The output result is a value between 0 and 1, representing the proportion of risk events, such as 0.08, which is 8%.

[0036] In the parameter settings of the basic traffic risk factor identifier, the ReLU activation function is used between the input layer and the first hidden layer to solve the gradient vanishing problem and enhance the model's ability to extract features; the ReLU activation function is used between the first and second hidden layers; and the Sigmoid activation function is used between the second hidden layer and the output layer to map the output result to the 0-1 interval, matching the numerical range of the "proportion of risk events". The Adam optimizer is selected, the learning rate is set to 0.005, and the mean squared error loss function is used; the L2 regularization coefficient is set to 0.001; and the batch size of the input samples for each training is set to 32.

[0037] For training the basic traffic risk factor identifier, the training data consists of the sample longitudinal slope feature set, the sample vehicle feature set, and the sample longitudinal slope traffic risk factor information set. The training data is divided into a training set and a test set in a 2:8 ratio. The maximum number of training rounds is set to 1200 rounds.

[0038] In 20 consecutive training rounds, the model's loss function value (mean squared error) remained stable below 0.002, with no significant decreasing or increasing trend, indicating that the model had entered a stable state. Training was then stopped, and the basic traffic risk element identifier was finally obtained after training and testing.

[0039] In step S200 of this application embodiment, incremental learning data is configured based on the braking time and longitudinal slope characteristics to obtain incremental learning data, including: Obtain the preset incremental learning data volume; The average braking time of multiple sample longitudinal slope monitoring processes is obtained. Based on the ratio of the braking time to the average braking time, the preset incremental learning data amount is corrected to obtain the incremental learning data amount. Based on the incremental learning data volume, collect the incremental longitudinal slope feature set, incremental vehicle feature set, and incremental longitudinal slope traffic risk element information set of the continuous longitudinal slope with the highest similarity to the longitudinal slope feature, until the incremental learning data volume is met, and obtain incremental learning data.

[0040] In this embodiment, the incremental learning data obtained in step S200 provides highly adaptable and reasonably sized training data for subsequent incremental learning based on the basic traffic risk factor identifier. By combining the braking time and slope characteristics of the current continuous longitudinal slope, the amount of incremental learning data is dynamically adjusted and similar data is filtered to avoid data redundancy or insufficiency, ensuring that incremental learning can accurately adapt to the characteristics of the current road segment and improving the recognition accuracy and efficiency of the subsequent incremental identifier.

[0041] First, a preset incremental learning data volume needs to be obtained. That is, an initial, general standard for incremental learning data volume is set. This preset value is determined based on the experience data of incremental learning of similar longitudinal slope road sections in the past. It provides a basic reference for subsequent data volume correction. For example, it is preset that 100 sets of data need to be collected.

[0042] Next, it is necessary to obtain the average braking time of the monitored longitudinal slopes in multiple sample longitudinal slopes. Based on the ratio of the braking time to the average braking time, the preset incremental learning data volume is corrected to obtain the incremental learning data volume. That is, from the existing multiple sample longitudinal slopes, i.e., the safety monitoring data of the sample longitudinal slopes used in constructing the basic traffic risk element identifier, the braking time of each sample longitudinal slope is extracted, and the average braking time of all sample longitudinal slopes is calculated. For example, the braking times of multiple sample longitudinal slopes are 3.2 seconds, 2.8 seconds, and 3.0 seconds, respectively, and the average braking time is 3.0 seconds.

[0043] Next, the average braking time of the monitoring and treatment within the longitudinal slope of multiple samples is obtained. Based on the ratio of the braking time to the average braking time, the ratio of the braking time of the current continuous longitudinal slope to the above average braking time is calculated, such as 4.5 ÷ 3.0 = 1.5.

[0044] Then, the preset incremental learning data volume is corrected using this ratio. The correction method can be: incremental learning data volume = ratio of braking time to average braking time * preset incremental learning data volume. If the current braking time is greater than the average braking time (ratio > 1), it means that the braking behavior of vehicles on the current road segment is more frequent and the risk is more complex, so the data volume needs to be increased, such as adjusting the data volume to 100 × 1.5 = 150 sets; if the current braking time is less than the average braking time (ratio < 1), the data volume is appropriately reduced, and finally the incremental learning data volume adapted to the current road segment is obtained.

[0045] Furthermore, according to the incremental learning data volume, it is necessary to collect the incremental longitudinal slope feature set, incremental vehicle feature set, and incremental longitudinal slope traffic risk element information set of the continuous longitudinal slope with the highest similarity to the longitudinal slope feature, until the incremental learning data volume is met, and thus obtain incremental learning data.

[0046] Based on the current continuous longitudinal slope characteristics, the continuous longitudinal slope with the smallest difference in slope gradient and length from the sample longitudinal slope database or newly monitored longitudinal slope sections is selected. For example, for the longitudinal slope in S100 with a gradient of 7% and a length of 2000 meters, similar longitudinal slopes with gradients of 6.8% to 7.2% and lengths of 1900 to 2100 meters can be selected.

[0047] For the selected similar longitudinal slopes, incremental longitudinal slope feature sets, incremental vehicle feature sets, and incremental longitudinal slope traffic risk factor information sets are collected. The incremental longitudinal slope feature set includes parameters such as slope and length of the similar longitudinal slopes; the incremental vehicle feature set includes characteristics such as the length of vehicles traveling on the similar longitudinal slopes; and the incremental longitudinal slope traffic risk factor information set includes the proportion of longitudinal slope risk events occurring under different combinations of "longitudinal slope features + vehicle features" within the similar longitudinal slopes.

[0048] Continue collecting the above three types of data until the total amount of data reaches the incremental learning data volume determined in the aforementioned steps, such as 150 sets, and finally form incremental learning data.

[0049] In step S300 of this application embodiment, based on the braking frequency distribution, the incremental learning data is used to incrementally learn the basic traffic risk element identifier to obtain multiple incremental identifiers, including: Based on the minimum and maximum braking frequencies within the braking frequency distribution, a braking frequency range is constructed; The average braking frequency range was obtained by processing monitoring data from multiple longitudinal slope samples. Get the preset incremental learning count; The preset incremental learning quantity is corrected and calculated based on the ratio of the interval length of the braking frequency interval to the average braking frequency interval to obtain the incremental learning quantity. Based on the incremental learning quantity, the basic traffic risk element identifier is incrementally learned using the incremental learning data to obtain multiple incremental identifiers.

[0050] In Embodiment 0 of this application, the purpose of step S30 is to dynamically adjust the number of incremental learning iterations by combining the braking frequency distribution of the current continuous longitudinal slope, and to optimize and train the basic traffic risk element recognizer multiple times using highly adaptable incremental learning data, ultimately obtaining multiple incremental recognizers. This allows the model to fully learn the complex characteristics of braking behavior in the current road segment, improving the robustness and accuracy of traffic risk element recognition for that road segment, and avoiding recognition bias caused by insufficient data or training iterations in a single model.

[0051] Specifically, the first step is to construct a braking frequency range based on the minimum and maximum braking frequencies within the braking frequency distribution. That is, based on the braking frequency distribution obtained in step S100, the minimum and maximum braking frequencies are extracted, and the braking frequency range for the current continuous longitudinal slope is constructed using these two values ​​as boundaries. For example, if the minimum frequency in the braking frequency distribution is 18 times / day and the maximum frequency is 35 times / day, then the braking frequency range for the current road segment is [18 times / day, 35 times / day].

[0052] Furthermore, based on the monitoring data of multiple sample longitudinal slopes, it is necessary to process and obtain the average braking frequency range. From the safety monitoring data of multiple sample longitudinal slopes used to construct the basic traffic risk element identifier, the braking frequency distribution of each sample longitudinal slope is extracted, and the braking frequency range of each sample longitudinal slope is calculated. Then, the braking frequency ranges of all sample longitudinal slopes are statistically processed: the "minimum average frequency" and "maximum average frequency" of all sample ranges are calculated, and the average braking frequency range is formed by using these two average values ​​as boundaries. For example, if the minimum braking frequency average of 100 sample longitudinal slopes is 20 times / day and the maximum braking frequency average is 30 times / day, then the average braking frequency range is [20 times / day, 30 times / day].

[0053] Furthermore, it is necessary to obtain a preset number of incremental learning iterations. An initial, universal standard for the number of incremental learning iterations should be established. This value is determined based on past experience with incremental learning on similar longitudinal slope road sections; for example, a preset number of incremental learning iterations of 5 is recommended to provide a basic reference for subsequent adjustments.

[0054] Furthermore, based on the ratio of the interval lengths of the braking frequency interval and the average braking frequency interval, the preset incremental learning quantity is corrected and calculated to obtain the incremental learning quantity.

[0055] First, calculate the length of the braking frequency interval for the current road segment and the length of the average braking frequency interval for the sample longitudinal slope. The braking frequency interval for the current road segment is the maximum braking frequency minus the minimum braking frequency, such as 35-18=17 times / day; the length of the average braking frequency interval for the sample longitudinal slope is the average maximum braking frequency minus the average minimum braking frequency, such as 30-20=10 times / day.

[0056] Then, calculate the ratio of the two interval lengths, which is the length of the current road segment braking frequency interval ÷ the length of the average braking frequency interval of the sample longitudinal slope, such as 17 ÷ 10 = 1.7.

[0057] Next, the preset incremental learning number is corrected using the ratio of the interval length. Specifically, the incremental learning number = preset incremental learning number * the above interval length ratio. If the above interval length ratio is greater than 1, it means that the braking frequency of the current road segment fluctuates more and the vehicle driving status is more complex, so the number of incremental learning needs to be increased, such as 5 × 1.7 ≈ 9 times. If the above interval length ratio is less than 1, the number of learning is appropriately reduced, and finally the incremental learning number adapted to the current road segment is obtained.

[0058] Furthermore, according to the stated incremental learning quantity, the basic traffic risk element identifier needs to be incrementally learned using the incremental learning data to obtain multiple incremental identifiers.

[0059] In step S300 of this application embodiment, the basic traffic risk element identifier is incrementally learned using the incremental learning data according to the incremental learning quantity, to obtain multiple incremental identifiers, including: Based on the total number of incremental learning iterations, the incremental learning data is randomly divided multiple times to obtain multiple random incremental data sets. Multiple random incremental data sets are used to incrementally train the basic traffic risk element identifier, resulting in multiple incremental identifiers.

[0060] like Figure 2 As shown, in step S300 of this application embodiment, multiple differentiated incremental recognizers are generated by randomly dividing and training the incremental learning data multiple times. This is to avoid the random errors caused by a single data division method, improve the robustness and generalization ability of the model to identify traffic risk factors on the current continuous longitudinal slope, and ensure that the final output traffic risk identification result is more comprehensive and reliable.

[0061] First, the total number of data partitions needs to be determined using the "incremental learning quantity" already defined in S300. For example, if the incremental learning quantity is 9 times, then 9 data partitions are required.

[0062] For the "incremental learning data" configured in S200, the incremental learning data is randomly allocated each time it is partitioned into a training subset and a validation subset. The data partitioning ratio remains consistent each time, with the training subset accounting for 80% and the validation subset accounting for 20%, but the specific data allocation must be different.

[0063] After completing the same number of random partitions as the incremental learning, a corresponding number of "random incremental data" are obtained. For example, 9 partitions correspond to 9 random incremental data. Each random incremental data contains an independent training subset and a validation subset, and the sample distribution of each data is different, which can cover more diverse feature combination scenarios.

[0064] Furthermore, the "Basic Traffic Risk Element Identifier" is used as the initial model for each incremental learning iteration. For each random incremental data set, its training subset is input into the Basic Traffic Risk Element Identifier. By adjusting parameters such as the model's learning rate and neuron weights, the model learns the correlation patterns between features and risks in the current random incremental data. At the same time, a validation subset is used to monitor the model's training effect in real time to avoid overfitting.

[0065] Ultimately, multiple incremental recognizers are generated: After training on each random incremental data set, the parameters of the base recognizer are optimized to form an "incremental recognizer" adapted to the characteristics of that data set. This process is repeated until all random incremental data sets have been trained, resulting in "multiple incremental recognizers" equal to the number of incremental learners. Each recognizer can identify traffic risk factors on continuous longitudinal slopes based on the feature patterns it has learned.

[0066] In step S400 of this application embodiment, vehicle features are collected, and combined with the longitudinal slope features, based on multiple incremental recognizers, longitudinal slope traffic risk factor information is identified and output, including: The road monitoring system obtains the vehicle characteristics of currently passing vehicles, including vehicle length. The vehicle features and longitudinal slope features are input into multiple incremental recognizers, and the recognition outputs multiple longitudinal slope traffic risk element information are obtained. The longitudinal slope traffic risk element information is obtained by fusion processing.

[0067] In this embodiment of the application, step S400 above is to collect the key features of currently passing vehicles, combine them with the inherent longitudinal slope features of the road section, and utilize the collaborative recognition capabilities of multiple incremental recognizers to achieve accurate and comprehensive identification of traffic risk factors on continuous longitudinal slopes.

[0068] Specifically, this requires relying on road monitoring systems deployed along the roadside, including high-definition cameras, vehicle detection radar, and intelligent weighing equipment, to capture real-time data on vehicles traveling on continuous longitudinal slope sections and extract core vehicle characteristics. The focus is on acquiring the length parameters of each passing vehicle, along with data such as load and vehicle type. For example, by using camera images and image recognition technology to measure the distance between the front and rear ends of vehicles, a sedan's length can be determined to be 4.5 meters, and a truck's length to be 12 meters, thus forming the vehicle characteristic data for the current vehicle.

[0069] Then, the collected current vehicle features are integrated with the continuous longitudinal slope features obtained in step S100 to form a complete feature input set. For example, a feature input set can be "vehicle length 4.5 meters + longitudinal slope gradient 7% + longitudinal slope length 2000 meters".

[0070] Next, the complete feature set described above is simultaneously fed into multiple incremental recognizers obtained in step S300. Each incremental recognizer independently analyzes the input vehicle and longitudinal slope features based on the feature patterns it has learned through training with different random incremental data, and outputs the corresponding longitudinal slope traffic risk element information, that is, the proportion of risk events predicted by the incremental recognizer. For example, incremental recognizer A outputs a traffic risk proportion of 8%, incremental recognizer B outputs 7.5%, etc.

[0071] Finally, the traffic risk factor information of the longitudinal slope output by multiple incremental recognizers is fused. A reasonable fusion strategy is adopted to eliminate the random errors of individual recognizers and obtain comprehensive and reliable traffic risk factor information of the longitudinal slope. For example, the average value of multiple recognition results, the weighted average value, or the result consistent with most recognizers is selected. For example, if five incremental recognizers output 8%, 7.5%, 8.2%, 7.8%, and 8.1% respectively, the average value is taken to obtain the final traffic risk ratio of 7.92%, which is used as the traffic risk factor information of the longitudinal slope, representing the risk level of the corresponding vehicle driving on the corresponding longitudinal slope.

[0072] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0073] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0074] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0075] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0076] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0077] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0078] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for intelligent identification of traffic risk factors on continuous longitudinal slopes based on deep learning, characterized in that, The method includes: The longitudinal slope characteristics of a continuous longitudinal slope are obtained, and the braking characteristics are identified based on the vehicle braking monitoring data within the continuous longitudinal slope. The braking characteristics include braking frequency distribution and braking time. A basic traffic risk element identifier is obtained, and incremental learning data is configured based on the braking time and longitudinal slope characteristics to obtain incremental learning data. Based on the braking frequency distribution, the basic traffic risk element identifier is incrementally learned using the incremental learning data to obtain multiple incremental identifiers. Vehicle characteristics are collected, combined with the longitudinal slope characteristics, and based on multiple incremental recognizers, longitudinal slope traffic risk element information is obtained by identification output.

2. The intelligent identification method for traffic risk factors on continuous longitudinal slopes based on deep learning according to claim 1, characterized in that, Acquire the longitudinal slope characteristics of a continuous longitudinal slope, and based on vehicle braking monitoring data within the continuous longitudinal slope, identify and obtain braking characteristics, including: Obtain the longitudinal slope characteristics of a continuous longitudinal slope, including the longitudinal slope gradient and longitudinal slope length; Obtain vehicle braking monitoring data within a preset time range within the continuous longitudinal slope, identify braking time sequences, and obtain a set of historical braking time sequences; Based on the set of historical braking time sequences, multiple average frequencies and average total durations of braking are calculated to obtain the braking frequency distribution and braking time, which serve as braking characteristics.

3. The intelligent identification method for traffic risk factors on continuous longitudinal slopes based on deep learning according to claim 1, characterized in that, Acquire basic traffic risk element identifiers, including: Based on safety monitoring data of multiple sample longitudinal slopes, sample longitudinal slope feature sets and sample vehicle feature sets were collected, and the proportion of longitudinal slope risk events occurring under different sample longitudinal slope features and sample vehicle features was collected and labeled as sample longitudinal slope traffic risk element information set. A basic traffic risk element identifier is constructed based on machine learning. Using the sample longitudinal slope feature set, sample vehicle feature set, and sample longitudinal slope traffic risk element information set, the basic traffic risk element identifier is trained under supervision. Training is completed after the test accuracy meets the accuracy threshold, and the trained and tested basic traffic risk element identifier is obtained.

4. The intelligent identification method for traffic risk factors on continuous longitudinal slopes based on deep learning according to claim 1, characterized in that, Based on the braking time and longitudinal slope characteristics, incremental learning data is configured to obtain incremental learning data, including: Obtain the preset incremental learning data volume; The average braking time of multiple sample longitudinal slope monitoring processes is obtained. Based on the ratio of the braking time to the average braking time, the preset incremental learning data amount is corrected to obtain the incremental learning data amount. Based on the incremental learning data volume, collect the incremental longitudinal slope feature set, incremental vehicle feature set, and incremental longitudinal slope traffic risk element information set of the continuous longitudinal slope with the highest similarity to the longitudinal slope feature, until the incremental learning data volume is met, and obtain incremental learning data.

5. The intelligent identification method for traffic risk factors on continuous longitudinal slopes based on deep learning according to claim 1, characterized in that, Based on the braking frequency distribution, the basic traffic risk element identifier is incrementally learned using the incremental learning data to obtain multiple incremental identifiers, including: Based on the minimum and maximum braking frequencies within the braking frequency distribution, a braking frequency range is constructed; The average braking frequency range was obtained by processing monitoring data from multiple longitudinal slope samples. Get the preset incremental learning count; The preset incremental learning quantity is corrected and calculated based on the ratio of the interval length of the braking frequency interval to the average braking frequency interval to obtain the incremental learning quantity. Based on the stated incremental learning quantity, the basic traffic risk element identifier is incrementally learned using the incremental learning data to obtain multiple incremental identifiers.

6. The intelligent identification method for traffic risk factors on continuous longitudinal slopes based on deep learning according to claim 5, characterized in that, Based on the stated incremental learning quantity, the basic traffic risk element identifier is incrementally learned using the incremental learning data to obtain multiple incremental identifiers, including: Based on the total number of incremental learning iterations, the incremental learning data is randomly divided multiple times to obtain multiple random incremental data sets. Multiple random incremental data sets are used to incrementally train the basic traffic risk element identifier, resulting in multiple incremental identifiers.

7. The intelligent identification method for traffic risk factors on continuous longitudinal slopes based on deep learning according to claim 1, characterized in that, Vehicle characteristics are collected, combined with the longitudinal slope characteristics, and based on multiple incremental recognizers, longitudinal slope traffic risk factor information is identified and output, including: The road monitoring system obtains the vehicle characteristics of currently passing vehicles, including vehicle length. The vehicle features and longitudinal slope features are input into multiple incremental recognizers, and the recognition outputs multiple longitudinal slope traffic risk element information are obtained. The longitudinal slope traffic risk element information is obtained by fusion processing.

Images