Traffic scene identification method, device, equipment and storage medium

Abstract

The embodiments of the present application disclose a method, apparatus, device and storage medium for identifying traffic scenarios. The method comprises: constructing multiple data features based on user data of a cell to be identified within a set time period; the data features include cell-level features and sector-level features; selecting a set number of data features corresponding to a candidate traffic scenario from the multiple data features; inputting the set number of data features into a binary classification model for scene identification corresponding to the candidate traffic scenario to obtain the target traffic scenario of the cell to be identified. The method for identifying traffic scenarios provided by the embodiments of the present application selects a set number of features from the constructed data features and inputs them into a binary classification model for scene identification corresponding to the candidate traffic scenario to obtain the target traffic scenario of the cell to be identified, which can quickly and accurately determine the traffic scenario of the cell.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a method, apparatus, device, and storage medium for identifying a traffic scenario. Background Art

[0002] With the rapid development of communication technology, base station strategies are becoming increasingly diverse and complex. However, applying different strategies to different base stations would waste significant manpower and financial resources. Applying the same strategy to all base stations would ignore the specific characteristics of their locations and fail to achieve the desired effect. For example, base stations in subways often have a large number of users during the day, but no users late at night when the subway closes. Therefore, a unified strategy can be adopted for these base stations in subways, such as entering energy-saving mode late at night and resuming it during the day. There are many other base stations in similar subway locations, such as children's palaces and elementary schools.

[0003] Existing methods for identifying traffic scenarios are based on the geographic location of base stations during base station planning, which determines traffic model characteristics. Examples include schools, hospitals, subways, and high-speed rail stations. However, in reality, different traffic patterns can occur in the same location. For example, in a school, classrooms experience traffic during the day but not late at night, while dormitories experience traffic at night and not during the day. These are all school base stations, but the traffic scenarios differ. Furthermore, traffic scenarios may be incomplete or incorrectly recorded during planning, leading to subsequent analysis errors. Furthermore, a cell's traffic scenario changes over time and cannot remain constant. Summary of the Invention

[0004] The embodiments of the present application provide a method, apparatus, device, and storage medium for identifying a traffic scenario, which can quickly and accurately determine the traffic scenario of a cell.

[0005] To achieve the above objectives, the present invention discloses a method for identifying a traffic scenario, including:

[0006] Constructing multiple data features based on user data of the cell to be identified within a set time period; the data features include cell-level features and sector-level features;

[0007] Selecting a set number of data features corresponding to the candidate traffic scenario from the plurality of data features;

[0008] The set number of data features are input into a scene recognition binary classification model corresponding to the candidate traffic scene to obtain a target traffic scene of the cell to be identified.

[0009] To achieve the above objectives, the present invention discloses a device for identifying a traffic scenario, including:

[0010] A data feature construction module is used to construct multiple data features based on user data of the cell to be identified within a set time period; the data features include cell-level features and sector-level features;

[0011] a data feature selection module, configured to select a set number of data features corresponding to a candidate traffic scenario from the plurality of data features;

[0012] The target traffic scene determination module is used to input the set number of data features into the scene recognition binary classification model corresponding to the candidate traffic scene to obtain the target traffic scene of the cell to be identified.

[0013] In order to achieve the above-mentioned purpose, an embodiment of the present application discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for identifying the traffic scenario as described in the embodiment of the present application is implemented.

[0014] In order to achieve the above-mentioned objectives, an embodiment of the present application discloses a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the method for identifying a traffic scenario as described in the embodiment of the present application is implemented.

[0015] The embodiments of the present application disclose a method, apparatus, device and storage medium for identifying traffic scenarios, which constructs multiple data features based on user data of a cell to be identified within a set time period; the data features include cell-level features and sector-level features; a set number of data features corresponding to candidate traffic scenarios are selected from the multiple data features; the set number of data features are input into a binary classification model for scene identification corresponding to the candidate traffic scenario to obtain a target traffic scenario for the cell to be identified. The method for identifying traffic scenarios provided by the embodiments of the present application selects a set number of features from the constructed data features and inputs them into a binary classification model for scene identification corresponding to the candidate traffic scenario to obtain a target traffic scenario for the cell to be identified, which can quickly and accurately determine the traffic scenario of the cell. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of a method for identifying a traffic scenario disclosed in an embodiment of the present application;

[0017] Figure 2 This is a user data diagram of a subway-like scenario disclosed in an embodiment of the present application;

[0018] Figure 3 This is a user data diagram of a weekday peak scenario disclosed in an embodiment of the present application;

[0019] Figure 4 This is a flowchart of the training method of the scene recognition binary classification model disclosed in the embodiment of the present application;

[0020] Figure 5 This is a schematic diagram of the structure of a device for identifying a service scenario disclosed in an embodiment of the present application;

[0021] Figure 6 It is a structural diagram of a computer device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0022] To make the purpose, technical solutions and advantages of this application more clear, the embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of this application can be combined with each other in any way.

[0023] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0024] In the following description, suffixes such as "module," "component," or "unit" used to represent elements are used only to facilitate the description of the present invention and have no specific meaning. Therefore, "module," "component," or "unit" may be used interchangeably.

[0025] In one embodiment, Figure 1 This is a flow chart of a method for identifying a traffic scene provided in an embodiment of the present application. This method can be applied to the case of identifying a traffic scene in a cell. This method can be executed by a device for identifying a traffic scene. Figure 1 As shown, the method includes S110-S130.

[0026] S110 , constructing a plurality of data features according to user data of the cell to be identified within a set time period.

[0027] The data features include cell-level features and sector-level features. The set time period may be one week, one month, or six months. The user data may be the number of users establishing a Radio Resource Control (RRC) connection. The user data within the set time period may be user data divided at a set granularity within the set time period, for example, at an hourly granularity.

[0028] In an embodiment of the present application, a method for constructing multiple data features based on user data of a cell to be identified within a set time period may be: obtaining user data of the cell to be identified and the sector where the cell to be identified is located, divided at a set granularity within a set time period; wherein the sector includes the cell to be identified and the cell with the same coverage as the cell to be identified; constructing cell-level features based on user data corresponding to the cell to be identified; and constructing sector-level features based on user data corresponding to the sector where the cell to be identified is located.

[0029] The sectors include the cell to be identified and the cells with the same coverage as the cell to be identified. The cells with the same coverage meet the following conditions: the longitude and latitude of the cell to be identified differ by a first set value (e.g., 100 meters); the azimuth angle of the cell to be identified differs by a second set value (e.g., 30 degrees); the cell is an outdoor cell; and the frequency is different from that of the cell to be identified.

[0030] In this embodiment, the method for constructing data features can be understood as statistically analyzing user data to obtain multiple data features. Cell-level features include basic information, low traffic, high traffic, correlation, difference, mutation, and distribution. Sector-level features include basic sector information, low sector traffic, and high sector traffic. For example, Table 1 shows the constructed cell-level features; Table 2 shows the constructed sector-level features.

[0031] Table 1

[0032]

[0033] Among them, TA is the timing advance (TA) distribution; RSRP is the reference signal received power (RSRP) of the cell; NI is noise and interference (NI); PL is path loss (PL); MCS is the modulation and coding scheme (MCS); CQI is the downlink channel quality indicator (CQI).

[0034] Table 2

[0035]

[0036]

[0037] S120: Select a set number of data features corresponding to the candidate traffic scenario from the plurality of data features.

[0038] There may be multiple candidate traffic scenarios, which may be manually set, such as subway-like scenarios, weekday peak scenarios, and emergency scenarios. Figure 2 This is a user data diagram of a subway-like scenario in an embodiment of the present application. Figure 3 This is a user data diagram of a weekday peak scenario in the embodiment of the present application. Figure 2-Figure 3The figure shows user data for one month. In this example, different traffic scenarios may require different set numbers of selected data features. Assuming that the candidate traffic scenarios include subway-like scenarios, weekday peak scenarios, and emergency scenarios, you need to select a first set number of data features corresponding to subway-like scenarios, a second set number of data features corresponding to weekday peak scenarios, and a third set number of data features corresponding to emergency scenarios.

[0039] Step 130: Input a set number of data features into a scene recognition binary classification model corresponding to the candidate traffic scene to obtain a target traffic scene of the cell to be identified.

[0040] In this embodiment, different traffic scenarios correspond to different scene recognition binary classification models. Assuming that the candidate traffic scenarios include subway-like scenarios, weekday peak scenarios, and emergency scenarios, a first set number of data features are input into the scene recognition binary classification model corresponding to the subway-like scenario to determine whether the cell to be identified belongs to the subway-like scenario; a second set number of data features are input into the scene recognition binary classification model corresponding to the weekday peak scenario to determine whether the cell to be identified belongs to the weekday peak scenario; and a third set number of data features are input into the scene recognition binary classification model corresponding to the emergency scenario to determine whether the cell to be identified belongs to the emergency scenario.

[0041] The technical solution of this embodiment is to construct multiple data features based on user data of the cell to be identified within a set time period; the data features include cell-level features and sector-level features; a set number of data features corresponding to candidate traffic scenarios are selected from the multiple data features; the set number of data features are input into a scene recognition binary classification model corresponding to the candidate traffic scenario to obtain the target traffic scenario of the cell to be identified. The method for identifying traffic scenarios provided by the embodiment of the present application selects a set number of features from the constructed data features and inputs them into a scene recognition binary classification model corresponding to the candidate traffic scenario to obtain the target traffic scenario of the cell to be identified, which can quickly and accurately determine the traffic scenario of the cell.

[0042] Optional, Figure 4 This is a flow chart of a method for training a scene recognition binary classification model in an embodiment of the present application. The training method for a scene recognition binary classification model as shown in the figure includes the following steps:

[0043] S410: Construct multiple data features based on user data of a sample cell within a set time period.

[0044] The sample cell may be a cell in a specific location (such as a subway, shopping mall, office building, park, school, or stadium). The set time period may be one week, one month, or six months. The user data may be the number of users establishing a Radio Resource Control (RRC) connection. The user data within the set time period may be user data divided at a set granularity within the set time period, for example, at an hourly granularity.

[0045] Specifically, the process of constructing multiple data features based on the user data of the sample cell within a set time period can be: obtaining the user data of the sample cell and the sector where the sample cell is located, divided at a set granularity within a set time period; constructing cell-level features based on the user data corresponding to the sample cell; and constructing sector-level features based on the user data corresponding to the sector where the sample cell is located.

[0046] In this embodiment, the constructed cell-level features refer to Table 1, and the constructed sector-level features refer to Table 2, which will not be repeated here.

[0047] S420: Select a set number of data features corresponding to the candidate traffic scenarios from the plurality of data features.

[0048] In this embodiment, different candidate traffic scenarios may have different set numbers of selected data features. Specifically, a method for selecting the set number of data features corresponding to each candidate traffic scenario from multiple data features may be: dividing the sample cell into candidate traffic scenario classes and non-candidate traffic scenario classes; determining a classification index for each data feature based on the candidate traffic scenario classes and the non-candidate traffic scenario classes; and selecting the set number of data features from the multiple data features based on the classification index.

[0049] Specifically, the method of determining the classification index of each data feature based on the candidate traffic scenario class and the non-candidate traffic scenario class can be: for each data feature, determine the feature center of the data feature within the candidate traffic scenario class; calculate the average intra-class distance between the data feature and the feature center within the candidate traffic scenario class; calculate the average inter-class distance between the data feature and the feature center within the non-candidate traffic scenario class; perform weighted summation of the average intra-class distance and the average inter-class distance to obtain the classification index of the data feature.

[0050] The feature center of each feature in the candidate traffic scenario class is calculated according to the following formula: Among them, S j represents the feature center of the jth feature, x ij represents the jth feature of the i-th sample cell of the candidate traffic scenario class, and n represents that there are n sample cells in the candidate traffic scenario class.

[0051] The average intra-class distance between data features and feature centers within a candidate traffic scenario class is calculated using the following formula: The average inter-class distance between the data features and feature centers in the non-candidate traffic scenario class is calculated using the following formula: Among them, y ij represents the jth feature of the i-th sample cell of the non-candidate traffic scenario class, m represents that there are m sample cells in the non-candidate traffic scenario class, and p represents the distance index, which is a positive integer and can be set to 2.

[0052] The calculation formula for the weighted sum of the average intra-class distance and the average inter-class distance can be: j =c1*D jout -c2*D jin , where the interval of c1 and c2 is [0,1], c1 can be 0.3, and c2 can be 0.7. In the embodiment of the present application, the larger the classification index, the more favorable the data feature is for classification.

[0053] In this embodiment of the present application, after obtaining the classification index of each data feature, the data features are sorted in descending order of classification index. Then, a set number of data features ranked at the top of the list are extracted as data features for candidate traffic scenarios. The set number can be any value between 10 and 20. The number selected can vary for different candidate traffic scenarios.

[0054] S430: Determine sample weights of sample cells belonging to the candidate traffic scenario.

[0055] In the embodiment of the present application, the number of samples in some traffic scenarios is often very small, and the samples are very unbalanced. In order to avoid the impact of the imbalance between samples on the model, it is necessary to weight the samples.

[0056] The sample weights of the sample cells belonging to the candidate traffic scenario may be calculated according to the following formula: Among them, P n is the number ratio of samples of a certain type of candidate traffic scenario, v n It is the importance of custom categories, set by the user, with a value range of [0,1] and a default value of 1. n is the weight of a candidate traffic scenario category.

[0057] S440: Training a scenario recognition binary classification model corresponding to the candidate traffic scenario based on the sample weight and a set number of data features.

[0058] In this embodiment, one candidate traffic scenario corresponds to one scenario recognition binary classification model.

[0059] Specifically, the method of training the scene recognition binary classification model corresponding to the candidate traffic scene based on the sample weight and a set number of data features can be: configuring enumeration parameters for the set neural network to obtain multiple initial binary classification models; dividing the sample cells into training sets and test sets according to the set ratio; training the multiple initial binary classification models based on the training set to obtain multiple intermediate binary classification models; testing the multiple intermediate binary classification models based on the test set to obtain test results; and determining the scene recognition binary classification model based on the test results.

[0060] The neural network can be constructed using algorithms such as random forest, gradient boosting decision tree (GBDT), or xgboost. Configuring enumerated parameters can be understood as adjusting the parameters in the neural network using a grid method. The set ratio can be 3:1. The intermediate binary classification model with the highest accuracy in the test results is selected as the final scene recognition binary classification model.

[0061] In an embodiment of the present application, multiple initial binary classification models are trained separately based on the training set, and a method for obtaining multiple intermediate binary classification models can be: for each initial binary classification model, the training set is divided into N parts of data; the N parts of data are traversed, and one part of the traversed data is used as a verification set, and the remaining N-1 parts of data are used as sub-training sets; the initial binary classification model is trained based on the sub-training set, and the trained initial binary classification model is verified based on the verification set to obtain a verification result; until the traversal of N parts of data is completed, N trained initial binary classification models and N verification results are obtained; according to the verification results, the intermediate binary classification model corresponding to the initial binary classification model is determined from the N trained initial binary classification models.

[0062] Among them, N can be set to 5. Specifically, the one with the highest accuracy in the verification results is determined as the intermediate two-classification model corresponding to the initial two-classification model.

[0063] The technical solution of the embodiment of the present application constructs multiple data features based on user data of a sample cell within a set time period; selects a set number of data features corresponding to candidate traffic scenarios from the multiple data features; determines the sample weights of the sample cells belonging to the candidate traffic scenarios; and trains a binary classification model for scene recognition corresponding to the candidate traffic scenarios based on the sample weights and the set number of data features. The embodiment of the present application provides a method for identifying traffic scenarios, which trains a binary classification model for scene recognition corresponding to the candidate traffic scenarios based on the sample weights and the set number of data features, and determines the target traffic scenario of the cell to be identified based on the trained binary classification model for scene recognition, thereby quickly and accurately determining the traffic scenario of the cell.

[0064] For example, the following embodiment trains a scene recognition binary classification model corresponding to a subway-like scene:

[0065] Table 3 shows the category codes of the sample communities:

[0066] Table 3

[0067] Category Code Category Name 0 Subway-like scene 1 Non-subway-like scenarios

[0068] The classification index of each data feature of the subway-like scene and the non-subway-like scene is determined in the manner disclosed in the above embodiment. The data features with the highest classification index ranking by a set number are selected as the data features corresponding to the subway-like scene.

[0069] The sample weights of sample cells belonging to the subway-like scenario can be calculated according to the following formula: Among them, P n is the number ratio of samples of a certain type of candidate traffic scenario, v n It is the importance of custom categories, set by the user, with a value range of [0,1] and a default value of 1. n is the weight of a candidate traffic scenario category. Table 4 shows the values ​​of v.

[0070] Table 4

[0071] Category Code Category Name v 0 Subway-like scene 1 1 Non-subway-like scenarios 0.8

[0072] In this embodiment, the neural network is constructed using the xgboost algorithm. The parameters to be adjusted include the number of base classifiers n_estimators, the maximum depth of the base classifiers max_depth, the minimum threshold gamma for splitting leaf nodes, the learning rate learning_rate, and the column sampling ratio colsample_bytree.

[0073] Set the enumeration variables for each parameter: n_estimators = [50, 100, 150, 200]; max_depth = [6, 8, 10, 15]; gamma = [0, 0.1, 0.2]; learning_rate = [0.1, 0.3, 0.5]; colsample_bytree = [0.5, 0.8, 1]. This will give us 4*4*3*3*3=432 initial binary classification models.

[0074] For each initial binary classification model, the sample plot was divided into a training set:test set ratio of 0.75:0.25. The training set was further divided into five parts, with four parts used for training and one for validation. This cycle was repeated five times to obtain five models. The model with the highest accuracy was selected as the training model for this round. This model was then tested on the test set to obtain the classification results for the test set.

[0075] According to the classification process of xgboost, the confusion matrix is ​​used to represent the quality of the classification results. The confusion matrix is ​​shown in Table 5:

[0076] Table 5

[0077]

[0078] The evaluation process of the model test results is calculated according to the following formula:

[0079]

[0080] F1 is the final evaluation score of the model. The higher the F1, the more accurate the model. The model with the highest F1 is selected as the final scene recognition binary classification model.

[0081] For example, the following embodiment trains a scene recognition binary classification model corresponding to the weekday peak scene:

[0082] Table 6 shows the category codes of the sample communities:

[0083] Table 6

[0084]

[0085] The classification index of each data feature of the weekday peak scene and the non-weekday peak scene is determined in the manner disclosed in the above embodiment. The data features with the highest classification index ranking of a set number are selected as the data features corresponding to the weekday peak scene.

[0086] The sample weight of the sample cells belonging to the weekday peak scenario can be calculated according to the following formula: Among them, P n is the number ratio of samples of a certain type of candidate traffic scenario, v n It is the importance of custom categories, set by the user, with a value range of [0,1] and a default value of 1. n is the weight of a candidate traffic scenario category. Table 7 shows the values ​​of v.

[0087] Table 7

[0088] Category Code Category Name v 0 Weekday peak scene 1 1 Non-working day peak scene 0.8

[0089] In this example, the neural network is constructed using the GBDT algorithm. The parameters to be adjusted include: the number of base classifiers n_estimators, the maximum depth of the base classifiers max_depth, the maximum number of features selected by the base classifiers max_features, and the learning rate learning_rate. Set the enumeration variables for each parameter: n_estimators = [50, 100, 150, 200]; max_depth = [6, 8, 10, 12]; max_features = [sqrt, 0.7, 0.9, 1]; learning_rate = [0.1, 0.2, 0.4, 0.8]. This yields 4*4*4*4=256 initial binary classification models.

[0090] For each initial binary classification model, the sample plot was divided into a training set:test set ratio of 0.75:0.25. The training set was further divided into five parts, with four parts used for training and one for validation. This cycle was repeated five times to obtain five models. The model with the highest accuracy was selected as the training model for this round. This model was then tested on the test set to obtain the classification results for the test set.

[0091] According to the classification process of GBDT, the confusion matrix is ​​used to represent the quality of the classification results. The confusion matrix is ​​shown in Table 8:

[0092] Table 8

[0093]

[0094] The evaluation process of the model test results is calculated according to the following formula:

[0095]

[0096] F1 is the final evaluation score of the model. The higher the F1, the more accurate the model. The model with the highest F1 is selected as the final scene recognition binary classification model.

[0097] Figure 5 This is a schematic diagram of the structure of a device for identifying a service scenario disclosed in an embodiment of the present application. Figure 5 As shown, the device includes:

[0098] The data feature construction module 210 is configured to construct a plurality of data features based on user data of the cell to be identified within a set period of time; the data features include cell-level features and sector-level features;

[0099] A data feature selection module 220 is configured to select a set number of data features corresponding to a candidate traffic scenario from the plurality of data features;

[0100] The target traffic scenario determination module 230 is configured to input the set number of data features into a scenario recognition binary classification model corresponding to the candidate traffic scenario to obtain the target traffic scenario of the cell to be identified.

[0101] Optionally, the data feature construction module 210 is further configured to:

[0102] Obtain user data of a sample cell and a sector where the sample cell is located, divided at a set granularity within a set time period; wherein the sector includes the sample cell and a cell with the same coverage as the sample cell;

[0103] Construct cell-level features based on user data corresponding to the sample cells;

[0104] Sector-level features are constructed based on the user data corresponding to the sector where the sample cell is located.

[0105] Optionally, it also includes a model training module for:

[0106] Construct multiple data features based on user data of the sample cell within a set time period;

[0107] Selecting a set number of data features corresponding to the candidate traffic scenarios respectively from the multiple data features;

[0108] Determining sample weights of sample cells belonging to the candidate traffic scenario;

[0109] A scene recognition binary classification model corresponding to the candidate traffic scene is trained based on the sample weight and the set number of data features.

[0110] Optionally, the model training module is also used to:

[0111] Divide the sample cells into candidate traffic scenario classes and non-candidate traffic scenario classes;

[0112] Determine the classification index of each data feature according to the candidate traffic scenario class and the non-candidate traffic scenario class;

[0113] A set number of data features are selected from the plurality of data features according to a classification index.

[0114] Optionally, the model training module is also used to:

[0115] For each data feature, determine the feature center of the data feature within the candidate traffic scenario class;

[0116] Calculate the average intra-class distance between data features and feature centers within the candidate traffic scenario class;

[0117] Calculate the average inter-class distance between the data features and feature centers within the non-candidate traffic scenario class;

[0118] The weighted sum of the average intra-class distance and the average inter-class distance is performed to obtain the classification index of the data feature.

[0119] Optionally, the model training module is also used to:

[0120] Configure enumerated parameters for the neural network to obtain multiple initial binary classification models;

[0121] Divide the sample cells into training sets and test sets according to the set ratio;

[0122] Based on the training set, multiple initial binary classification models are trained separately to obtain multiple intermediate binary classification models;

[0123] Test multiple intermediate binary classification models based on the test set to obtain test results;

[0124] Determine the scene recognition binary classification model based on the test results.

[0125] Optionally, the model training module is also used to:

[0126] For each initial two-class model, the training set is divided into N parts of data;

[0127] Traverse N copies of data, use one copy of the data as the validation set, and the remaining N-1 copies of data as the sub-training set;

[0128] The initial binary classification model is trained based on the sub-training set, and the trained initial binary classification model is verified based on the verification set to obtain the verification results; until N pieces of data are traversed, N trained initial binary classification models and N verification results are obtained;

[0129] According to the verification results, an intermediate binary classification model corresponding to the initial binary classification model is determined from the N trained initial binary classification models.

[0130] In one embodiment, Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present application. Figure 6 As shown, the device provided by this application includes: a processor 310 and a memory 320. The number of processors 310 in the device can be one or more. Figure 6 In the example, a processor 310 is used. The number of memories 320 in the device can be one or more. Figure 6 In the example, a memory 320 is used. The processor 310 and the memory 320 of the device can be connected via a bus or other means. Figure 6 In the embodiment, the device is a computer device.

[0131] The memory 320, as a computer-readable storage medium, can be configured to store software programs, computer executable programs, and modules, such as program instructions / modules corresponding to the device of any embodiment of the present application (for example, the encoding module and the first sending module in the data transmission device). The memory 320 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the device, etc. In addition, the memory 320 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 320 may further include a memory remotely located relative to the processor 310, and these remote memories may be connected to the device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0132] The above-mentioned device can be configured to execute the identification method applied to the traffic scenario provided in any of the above-mentioned embodiments, and has corresponding functions and effects.

[0133] The program stored in the corresponding memory 320 may be a program instruction / module corresponding to the interrupt handling method provided in the embodiments of the present application. The processor 310 executes the software program, instructions, and modules stored in the memory 320 to execute one or more functional applications and data processing of the computer device, that is, to implement the association query method applied to data in the above method embodiments. It is understood that when the above device is a receiving end, it can execute the interrupt handling method provided in any embodiment of the present application and have the corresponding functions and effects.

[0134] An embodiment of the present application also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a method for identifying traffic scenarios, the method comprising: constructing multiple data features based on user data of a cell to be identified within a set time period; the data features include cell-level features and sector-level features; selecting a set number of data features corresponding to a candidate traffic scenario from the multiple data features; inputting the set number of data features into a scene recognition binary classification model corresponding to the candidate traffic scenario to obtain a target traffic scenario for the cell to be identified.

[0135] It will be appreciated by those skilled in the art that the term user equipment encompasses any suitable type of wireless user equipment, such as a mobile phone, a portable data processing device, a portable web browser or a car-mounted mobile station.

[0136] In general, various embodiments of the present application may be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. For example, some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device, although the present application is not limited thereto.

[0137] Embodiments of the present application may be implemented by executing computer program instructions by a data processor of a mobile device, for example, in a processor entity, or by hardware, or by a combination of software and hardware. The computer program instructions may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages.

[0138] The block diagram of any logical flow in the drawings of the present application may represent program steps, or may represent interconnected logical circuits, modules and functions, or may represent a combination of program steps and logical circuits, modules and functions. A computer program may be stored on a memory. The memory may be of any type suitable for the local technical environment and may be implemented using any suitable data storage technology, such as, but not limited to, a read-only memory (ROM), a random access memory (RAM), an optical storage device and system (a digital versatile disc (DVD) or a compact disk (CD)). Computer-readable media may include non-transient storage media. A data processor may be of any type suitable for the local technical environment, such as, but not limited to, a general-purpose computer, a special-purpose computer, a microprocessor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), and a processor based on a multi-core processor architecture.

[0139] The above description is merely an exemplary embodiment of the present application and is not intended to limit the scope of protection of the present application.

[0140] Embodiments of the present application may be implemented by executing computer program instructions by a data processor of a mobile device, for example, in a processor entity, or by hardware, or by a combination of software and hardware. The computer program instructions may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages.

[0141] The above description of exemplary embodiments of the present application has been provided by way of exemplary and non-limiting examples. However, various modifications and adaptations to the above embodiments will be apparent to those skilled in the art, when considered in conjunction with the accompanying drawings and claims, without departing from the scope of the present invention. Therefore, the proper scope of the present invention will be determined by reference to the claims.

Claims

1. A method for identifying a traffic scenario, characterized in that: include: Constructing multiple data features based on user data of the cell to be identified within a set time period; The data features include cell-level features and sector-level features; Selecting a set number of data features corresponding to the candidate traffic scenario from the plurality of data features; Inputting the set number of data features into a scene recognition binary classification model corresponding to the candidate traffic scene to obtain a target traffic scene of the cell to be identified; Construct multiple data features based on user data of the cell to be identified within a set period of time, including: Acquire user data of a cell to be identified and a sector where the cell to be identified is located, divided at a set granularity within a set time period; wherein the sector includes the cell to be identified and a cell with the same coverage as the cell to be identified; Constructing a cell-level feature according to user data corresponding to the cell to be identified; A sector-level feature is constructed according to user data corresponding to the sector where the to-be-identified cell is located.

2. The method according to claim 1, characterized in that The training method of the scene recognition binary classification model is: Construct multiple data features based on user data of the sample cell within a set time period; Selecting a set number of data features corresponding to the candidate traffic scenarios respectively from the multiple data features; Determining sample weights of sample cells belonging to the candidate traffic scenario; A scene recognition binary classification model corresponding to the candidate traffic scene is trained based on the sample weight and the set number of data features.

3. The method according to claim 2, characterized in that Selecting a set number of data features corresponding to the candidate traffic scenarios from the plurality of data features includes: Classifying the sample cells into candidate traffic scenario classes and non-candidate traffic scenario classes; Determining a classification index for each data feature according to the candidate traffic scenario classes and the non-candidate traffic scenario classes; A set number of data features are selected from the plurality of data features according to the classification index.

4. The method according to claim 3, characterized in that Determining a classification index for each data feature according to the candidate traffic scenario class and the non-candidate traffic scenario class includes: For each data feature, determining a feature center of the data feature within the candidate traffic scenario class; Calculating the average intra-class distance between the data features and the feature centers within the candidate traffic scenario class; Calculating the average inter-class distance between the data features and the feature centers in the non-candidate traffic scene class; A weighted sum is performed on the average intra-class distance and the average inter-class distance to obtain a classification index of the data feature.

5. The method according to claim 2, characterized in that Training a scene recognition binary classification model corresponding to the candidate traffic scene based on the sample weight and the set number of data features includes: Configure multiple parameters of the neural network to obtain multiple initial binary classification models; Dividing the sample cells into a training set and a test set according to a set ratio; Training the multiple initial binary classification models respectively based on the training set to obtain multiple intermediate binary classification models; Testing the multiple intermediate binary classification models respectively based on the test set to obtain test results; A scene recognition binary classification model is determined according to the test results.

6. The method according to claim 5, characterized in that The multiple initial binary classification models are trained separately based on the training set to obtain multiple intermediate binary classification models, including: For each initial two-class model, the training set is divided into N parts of data; Traverse the N data sets, use the first data set as the validation set, and the remaining N-1 data sets as the sub-training sets; The initial binary classification model is trained based on the sub-training set, and the trained initial binary classification model is verified based on the verification set to obtain a verification result; until the N pieces of data are traversed, N trained initial binary classification models and N verification results are obtained; According to the verification result, an intermediate binary classification model corresponding to the initial binary classification model is determined from the N trained initial binary classification models.

7. A device for identifying a traffic scene, characterized in that: include: A data feature construction module is used to construct multiple data features based on user data of the cell to be identified within a set time period; the data features include cell-level features and sector-level features; a data feature selection module, configured to select a set number of data features corresponding to a candidate traffic scenario from the plurality of data features; a target traffic scenario determination module, configured to input the set number of data features into a scenario recognition binary classification model corresponding to the candidate traffic scenario to obtain a target traffic scenario for the cell to be identified; The data feature construction module includes: obtaining user data of the cell to be identified and the sector where the cell to be identified is located, divided at a set granularity within a set time period; wherein the sector includes the cell to be identified and the cell with the same coverage as the cell to be identified; constructing cell-level features based on the user data corresponding to the cell to be identified; and constructing sector-level features based on the user data corresponding to the sector where the cell to be identified is located.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for identifying a traffic scenario as described in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for identifying a traffic scenario as described in any one of claims 1 to 6 is implemented.

Images