Base station traffic prediction method, device and electronic equipment

By identifying the target scenario and traffic change patterns of the base station and selecting a matching prediction model for prediction, the problem of the influence of feature differences in the base station traffic prediction is solved and the accuracy of the prediction results is improved.

CN114374983BActive Publication Date: 2024-11-12CHINA TELECOM CORP LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202111647710.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-29
Publication Date
2024-11-12
Estimated Expiration
2041-12-29

AI Technical Summary

Technical Problem

The existing base station traffic prediction model ignores the differences in base station characteristics during training, resulting in inaccuracy of traffic prediction results.

Method used

By determining the target scenario of the base station to be predicted, and selecting the traffic change mode matching the scene from the corresponding relationship between the preset scene and the traffic change mode, selecting the corresponding traffic prediction model for prediction, ensuring that the model training sample and the base station to be predicted are to be predicted. The historical data characteristics are consistent.

Benefits of technology

The accuracy of the base station traffic prediction results is improved, the characteristic differences between different base stations are weakened, and the accuracy of the traffic prediction model is enhanced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114374983B_ABST
    Figure CN114374983B_ABST
Patent Text Reader

Abstract

The embodiments of the present invention provide a base station traffic prediction method, device and electronic device, which relate to the field of wireless communication technology. The method includes: determining the target scene in which the base station to be predicted is located; determining the target traffic change pattern corresponding to the target scene from the target correspondence between the preset scene and the traffic change pattern; selecting a target model that matches the target traffic change pattern from multiple traffic prediction models obtained in advance; inputting the target scene into the target model for training, and obtaining the prediction result as the traffic prediction result of the base station to be predicted. Compared with the prior art, the solution provided by the embodiment of the present invention is applied to improve the accuracy of the traffic prediction model, thereby improving the accuracy of the traffic prediction results of the corresponding base station and the accuracy of the traffic prediction results.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to a base station traffic prediction method, device and electronic equipment. Background Art

[0002] With the continuous development of network technology, people's dependence on the network is increasing. The prediction research of network traffic can realize the pre-allocation of network resources and improve the utilization efficiency of existing communication resources.

[0003] In the related art, in terms of base station traffic prediction, it is usually implemented using a time-related traffic prediction model, and the traffic prediction model is obtained by training based on multi-period historical traffic of multiple base stations over a long period of time.

[0004] However, in the above-mentioned related technologies, when training the above-mentioned traffic prediction model, multiple base stations are analyzed together, which often ignores the characteristic differences of the base stations, affecting the accuracy of the above-mentioned traffic prediction model, and further affecting the accuracy of the traffic prediction results of the corresponding base stations. Summary of the invention

[0005] The purpose of the embodiments of the present invention is to provide a base station traffic prediction method, device and electronic device to improve the accuracy of the traffic prediction model, thereby improving the accuracy of the traffic prediction results of the corresponding base station. The specific technical solution is as follows:

[0006] In a first aspect, an embodiment of the present invention provides a base station traffic prediction method, the method comprising:

[0007] Determine the target scene where the base station to be predicted is located;

[0008] Determine a target traffic change pattern corresponding to the target scene from a target correspondence relationship between a preset scene and a traffic change pattern; wherein the target correspondence relationship is established based on a correspondence relationship between a sample scene in which each sample base station is located and a traffic change pattern that conforms to the change characteristics of historical traffic data of each sample base station;

[0009] From a plurality of traffic prediction models obtained by pre-training, a target model matching the target traffic change pattern is selected; wherein each traffic prediction model is trained based on historical traffic data of a group of sample base stations, and the change characteristics of the historical traffic data of the group of sample base stations conform to the same traffic change pattern; and each traffic prediction model is added with sample scene labels of each sample base station used to train the traffic prediction model; the traffic change pattern conformed to by the change characteristics of the historical traffic data of each sample base station used to train the target model is the target traffic change pattern;

[0010] The target scenario is input into the target model for training, and a prediction result is obtained as the traffic prediction result of the base station to be predicted.

[0011] Optionally, a training method of multiple traffic prediction models in a specific implementation includes:

[0012] Determine each sample base station and the sample scene where each sample base station is located;

[0013] Obtain historical traffic data of each sample base station, and determine the change characteristics of the historical traffic data of each sample base station;

[0014] The sample base stations whose change characteristics of historical traffic data conform to the same traffic change pattern are divided into the same group to obtain groups of sample base stations;

[0015] For each group of sample base stations, the preset initial model is trained using the historical traffic data of each base station in the group of sample base stations;

[0016] When the initial model meets the preset conditions, the training is stopped to obtain a traffic prediction model that matches the traffic change pattern corresponding to the group of sample base stations, and the sample scene labels of each sample base station in the group of sample base stations are added to the obtained traffic prediction model; wherein the traffic change pattern corresponding to each group of sample base stations is: the traffic change pattern that is consistent with the change rules of the historical traffic data of each base station in the group of sample base stations.

[0017] Optionally, in a specific implementation manner, the step of determining each sample base station includes:

[0018] Determine each candidate base station, and group each candidate base station according to location information of each candidate base station to obtain each candidate base station group;

[0019] For each candidate base station group, calculate the aggregated result of the historical traffic data of each base station in the candidate base station group within a specified time range;

[0020] Based on the calculated aggregation results, the candidate base station groups that do not meet the specified traffic conditions are eliminated to obtain sample base stations.

[0021] Optionally, in a specific implementation manner, before the step of grouping the candidate base stations according to the location information of the candidate base stations to obtain the candidate base station groups, the method further includes:

[0022] Acquire historical traffic data of each candidate base station within a specified time range, and perform data preprocessing on each acquired historical traffic data to obtain preprocessed historical traffic data of each candidate base station within the specified time range;

[0023] The step of grouping the candidate base stations according to the location information of the candidate base stations to obtain the candidate base station groups includes:

[0024] According to the position information of each pre-processed candidate base station, the pre-processed candidate base stations are grouped to obtain groups of candidate base stations;

[0025] The step of calculating, for each candidate base station group, an aggregate result of historical traffic data of each base station in the candidate base station group within a specified time range includes:

[0026] For each candidate base station group, the aggregation result of the pre-processed historical traffic data of each base station in the candidate base station group within a specified time range is calculated.

[0027] Optionally, in a specific implementation, the step of eliminating the candidate base station group that does not meet the specified traffic condition based on the calculated aggregation results includes:

[0028] For each candidate base station group, based on the calculated aggregation result, the candidate base station group is normalized to obtain a normalized result of the candidate base station group;

[0029] Eliminate the candidate base station groups whose normalized results are lower than a preset threshold.

[0030] Optionally, in a specific implementation manner, the step of normalizing each candidate base station group based on the calculated aggregation result to obtain a normalized result of the candidate base station group includes:

[0031] The standardized results of each candidate base station group are calculated using a preset target formula; wherein the target formula is:

[0032] A i =(A 0i -A min )÷(A max -A min )

[0033] Among them, A i is the normalized result of the i-th candidate base station group; A 0i A is the aggregation result calculated for the i-th candidate base station group; min is the preset minimum value; A max is the preset maximum value.

[0034] In a second aspect, an embodiment of the present invention provides a base station traffic prediction device, the device comprising:

[0035] A scene determination module, used to determine the target scene where the base station to be predicted is located;

[0036] A mode determination module, used to determine a target traffic change mode corresponding to a target scene from a target correspondence relationship between a preset scene and a traffic change mode; wherein the target correspondence relationship is established based on a correspondence relationship between a sample scene in which each sample base station is located and a traffic change mode that conforms to the change characteristics of historical traffic data of each sample base station;

[0037] A model determination module is used to select a target model that matches the target traffic change pattern from a plurality of traffic prediction models obtained by pre-training; wherein each traffic prediction model is obtained by training based on historical traffic data of a group of sample base stations, and the change characteristics of the historical traffic data of the group of sample base stations conform to the same traffic change pattern; and each traffic prediction model is added with sample scene labels of each sample base station used to train the traffic prediction model; the traffic change pattern conformed to by the change characteristics of the historical traffic data of each sample base station used to train the target model is the target traffic change pattern;

[0038] The result acquisition module is used to input the target scenario into the target model for training and obtain the prediction result as the traffic prediction result of the base station to be predicted.

[0039] Optionally, in a specific implementation, the device further includes:

[0040] A determination module, used to determine each sample base station and a sample scene where each sample base station is located;

[0041] A feature determination module obtains historical traffic data of each sample base station and determines the change characteristics of the historical traffic data of each sample base station;

[0042] The sample base station grouping module divides the sample base stations whose change characteristics of historical traffic data conform to the same traffic change pattern into the same group to obtain sample base stations in each group;

[0043] The model training module is used to train a preset initial model for each group of sample base stations using the historical traffic data of each base station in the group of sample base stations; when the initial model meets the preset conditions, the training is stopped to obtain a traffic prediction model that matches the traffic change pattern corresponding to the group of sample base stations, and the sample scene labels of each sample base station in the group of sample base stations are added to the obtained traffic prediction model; wherein the traffic change pattern corresponding to each group of sample base stations is: the traffic change pattern that is consistent with the change rules of the historical traffic data of each base station in the group of sample base stations.

[0044] Optionally, in a specific implementation, the determining module includes:

[0045] The candidate base station determination submodule is used to determine each candidate base station and group each candidate base station according to the location information of each candidate base station to obtain each candidate base station group;

[0046] The aggregation result acquisition submodule is used to calculate the aggregation result of the historical traffic data of each base station in the candidate base station group within a specified time range for each candidate base station group;

[0047] The sample base station acquisition submodule is used to eliminate the candidate base station group that does not meet the specified traffic conditions based on the calculated aggregation results to obtain sample base stations.

[0048] Optionally, in a specific implementation, the device further includes:

[0049] A data acquisition module, used for obtaining historical traffic data of each candidate base station within a specified time range before the step of grouping each candidate base station according to the location information of each candidate base station to obtain each group of candidate base stations, and performing data preprocessing on each acquired historical traffic data to obtain the preprocessed historical traffic data of each candidate base station within the specified time range;

[0050] The candidate base station determination submodule is specifically used to: group the pre-processed candidate base stations according to the pre-processed location information of each candidate base station to obtain each group of candidate base stations;

[0051] The aggregation result acquisition submodule is specifically used to: for each candidate base station group, calculate the aggregation result of the pre-processed historical traffic data of each base station in the candidate base station group within a specified time range.

[0052] Optionally, in a specific implementation, the sample base station acquisition submodule includes:

[0053] A standard processing unit, configured to standardize each candidate base station group based on the calculated aggregation result to obtain a standardized result of the candidate base station group;

[0054] The base station elimination unit is used to eliminate the candidate base station group whose normalization results are lower than a preset threshold.

[0055] Optionally, in a specific implementation, the standard processing unit is specifically used to:

[0056] The standardized results of each candidate base station group are calculated using a preset target formula; wherein the target formula is:

[0057] A i =(A 0i -A min )÷(Amax -A min )

[0058] Among them, A i is the normalized result of the i-th candidate base station group; A 0i A is the aggregation result calculated for the i-th candidate base station group; min is the preset minimum value; A max is the preset maximum value.

[0059] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0060] Memory, used to store computer programs;

[0061] The processor is used to implement the steps of any base station traffic prediction method provided in the first aspect when executing the program stored in the memory.

[0062] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of any base station traffic prediction method provided in the first aspect are implemented.

[0063] In a fifth aspect, an embodiment of the present invention provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the steps of any base station traffic prediction method provided in the first aspect above.

[0064] Beneficial effects of the embodiments of the present invention:

[0065] As can be seen from the above, when applying the solution provided by the embodiment of the present invention, in the process of traffic prediction for the base station to be predicted, it is first necessary to determine the target scene in which the base station to be predicted is located, and then determine the target traffic change pattern corresponding to the above target scene from the target correspondence relationship established based on the correspondence between the sample scenes in which each sample set station is located and the traffic change pattern that conforms to the change characteristics of the historical traffic data of each sample base station, and finally, select the target model that matches the above target traffic change pattern from the multiple traffic prediction models obtained in advance, input the above target scene into the above target model for training, and obtain the prediction result of the model as the traffic prediction result of the above base station to be predicted.

[0066] By applying the solution provided by the embodiment of the present invention, for each traffic prediction model, the change characteristics of the historical traffic data of each sample base station used to train the traffic prediction model conform to the same traffic change pattern, so that each trained traffic prediction model matches a traffic change pattern respectively.

[0067] That is to say, when using multiple sample base stations to construct a traffic prediction model, the change characteristics of the historical traffic data of different sample base stations are taken into account, and the sample base stations with the same change characteristics are divided into a group. Using the divided groups of sample base stations, different traffic prediction models are trained respectively. Therefore, each traffic prediction model is trained based on base stations with the same change characteristics of historical traffic data, rather than trained based on base stations with different change characteristics of historical traffic data.

[0068] In this way, when performing traffic prediction for the base station to be predicted, a traffic prediction model that matches the traffic change pattern of the base station to be predicted can be selected, that is, the traffic change pattern that the historical traffic data of the sample base station used to train the selected traffic prediction model conforms to is the traffic change pattern of the base station to be predicted. Therefore, since the training samples of the selected traffic prediction model have the same change characteristics as the traffic data of the base station to be predicted, when using the selected traffic prediction model to perform traffic prediction for the base station to be predicted, the accuracy of the traffic prediction result can be improved.

[0069] Based on this, the solution provided by the embodiment of the present invention is applied to group the sample base stations according to the traffic change patterns that the historical traffic data of the sample base stations conform to, and use the groups of sample base stations with the same traffic change patterns that the historical traffic data conforms to to respectively train traffic prediction models that match various traffic change patterns, which can weaken the feature differences between the base stations used to train each traffic prediction model and improve the accuracy of each traffic prediction model obtained through training. Furthermore, when the traffic prediction model that matches the traffic change pattern of the base station to be predicted is used to predict the traffic of the base station to be predicted, the accuracy of the obtained prediction results can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other embodiments can also be obtained based on these drawings.

[0071] Figure 1 A flow chart of a base station traffic prediction method provided by an embodiment of the present invention;

[0072] Figure 2(a) is a schematic diagram of the mainstream flow change pattern;

[0073] Figure 2(b) is a schematic diagram of the scenic area type flow change pattern;

[0074] Figure 2(c) is a schematic diagram of the return-to-home traffic change pattern;

[0075] Figure 2(d) is a schematic diagram of the holiday traffic change pattern;

[0076] Figure 2(e) is a schematic diagram of the office traffic change pattern;

[0077] Figure 2(f) is a schematic diagram of the campus-type traffic change pattern;

[0078] Figure 2(g) is a schematic diagram of the residential traffic flow variation pattern;

[0079] Figure 3 A flowchart of a method for training multiple traffic prediction models provided by an embodiment of the present invention;

[0080] Figure 4 A schematic diagram of the result of training a GRU base station prediction model using historical traffic data of each sample base station in a cluster corresponding to a holiday traffic change pattern;

[0081] Figure 5 for Figure 3 A flowchart of a specific implementation method of S301;

[0082] Figure 6 for Figure 3 A flowchart of another specific implementation method of S301;

[0083] Figure 7 A schematic diagram of the structure of a base station traffic prediction device provided by an embodiment of the present invention;

[0084] Figure 8 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0085] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field based on this application belong to the scope of protection of the present invention.

[0086] In terms of base station traffic prediction, it is usually implemented using a time-related traffic prediction model, which is trained based on multi-period historical traffic of multiple base stations over a long period of time. However, when training the above traffic prediction model, multiple base stations are analyzed together, which often ignores the characteristic differences of the base stations, affecting the accuracy of the above traffic prediction model, and further affecting the accuracy of the traffic prediction results of the corresponding base stations.

[0087] In order to solve the above technical problems, an embodiment of the present invention provides a base station traffic prediction method.

[0088] Among them, this method can be applicable to various application scenarios of traffic prediction for base stations, for example, traffic prediction for a newly built base station to determine the communication resources allocated to the newly built base station, and for example, traffic prediction for an existing base station to optimize the current base station communication resource allocation plan, etc. In addition, this method can be applied to various electronic devices such as servers, laptops, and desktop computers, hereinafter referred to as electronic devices. Among them, the electronic device can be an independent electronic device, such as an independent desktop computer; or it can be multiple electronic devices, such as a device cluster composed of multiple desktop computers. Based on this, the embodiment of the present invention does not limit the application scenario and execution subject of the method.

[0089] A flow prediction method provided by an embodiment of the present invention may include the following steps:

[0090] Determine the target scene where the base station to be predicted is located;

[0091] Determine a target traffic change pattern corresponding to the target scene from a target correspondence relationship between a preset scene and a traffic change pattern; wherein the target correspondence relationship is established based on a correspondence relationship between a sample scene in which each sample base station is located and a traffic change pattern that conforms to the change characteristics of historical traffic data of each sample base station;

[0092] From a plurality of traffic prediction models obtained by pre-training, a target model matching the target traffic change pattern is selected; wherein each traffic prediction model is obtained by training based on historical traffic data of a group of sample base stations, and the change characteristics of the historical traffic data of the group of sample base stations conform to the same traffic change pattern, and each traffic prediction model is added with sample scene labels of each sample base station used to train the traffic prediction model; the traffic change pattern conformed to by the change characteristics of the historical traffic data of each sample base station used to train the target model is the target traffic change pattern;

[0093] The target scenario is input into the target model for training, and a prediction result is obtained as the traffic prediction result of the base station to be predicted.

[0094] As can be seen from the above, when applying the solution provided by the embodiment of the present invention, in the process of traffic prediction for the base station to be predicted, it is first necessary to determine the target scene in which the base station to be predicted is located, and then determine the target traffic change pattern corresponding to the above target scene from the target correspondence relationship established based on the correspondence between the sample scenes in which each sample set station is located and the traffic change pattern that conforms to the change characteristics of the historical traffic data of each sample base station, and finally, select the target model that matches the above target traffic change pattern from the multiple traffic prediction models obtained in advance, input the above target scene into the above target model for training, and obtain the prediction result of the model as the traffic prediction result of the above base station to be predicted.

[0095] By applying the solution provided by the embodiment of the present invention, for each traffic prediction model, the change characteristics of the historical traffic data of each sample base station used to train the traffic prediction model conform to the same traffic change pattern, so that each trained traffic prediction model matches a traffic change pattern respectively.

[0096] That is to say, when using multiple sample base stations to construct a traffic prediction model, the change characteristics of the historical traffic data of different sample base stations are taken into account, and the sample base stations with the same change characteristics are divided into a group. Using the divided groups of sample base stations, different traffic prediction models are trained respectively. Therefore, each traffic prediction model is trained based on base stations with the same change characteristics of historical traffic data, rather than trained based on base stations with different change characteristics of historical traffic data.

[0097] In this way, when performing traffic prediction for the base station to be predicted, a traffic prediction model that matches the traffic change pattern of the base station to be predicted can be selected, that is, the traffic change pattern that the historical traffic data of the sample base station used to train the selected traffic prediction model conforms to is the traffic change pattern of the base station to be predicted. Therefore, since the training samples of the selected traffic prediction model have the same change characteristics as the traffic data of the base station to be predicted, when using the selected traffic prediction model to perform traffic prediction for the base station to be predicted, the accuracy of the traffic prediction result can be improved.

[0098] Based on this, the solution provided by the embodiment of the present invention is applied to group the sample base stations according to the traffic change patterns that the historical traffic data of the sample base stations conform to, and use the groups of sample base stations with the same traffic change patterns that the historical traffic data conforms to to respectively train traffic prediction models that match various traffic change patterns, which can weaken the feature differences between the base stations used to train each traffic prediction model and improve the accuracy of each traffic prediction model obtained through training. Furthermore, when the traffic prediction model that matches the traffic change pattern of the base station to be predicted is used to predict the traffic of the base station to be predicted, the accuracy of the obtained prediction results can be improved.

[0099] A base station traffic prediction method provided by an embodiment of the present invention is described in detail below in conjunction with the accompanying drawings.

[0100] Figure 1 A method for predicting base station traffic provided by an embodiment of the present invention is as follows: Figure 1 As shown, the method includes the following steps S101-S104.

[0101] S101: Determine a target scene in which a base station to be predicted is located.

[0102] In order to provide better network coverage for various areas such as cities and villages, it is usually necessary to establish multiple base stations at different locations in the area, so that the base stations located at different locations can be in different scenarios or in the same scenario.

[0103] For example, in cities, base stations are established in multiple locations such as commercial areas, residential areas, schools, public transportation hubs, etc.; in rural areas, base stations are established in multiple locations such as residential areas, agricultural production areas, schools, etc.

[0104] For each base station, the scene where the base station is located can be represented by the name of the block where the base station is located, the longitude and latitude of the location of the base station, the name of the base station, etc., or can be comprehensively determined by using the name of the block where the base station is located, the longitude and latitude of the location of the base station, the name of the base station, etc. This is all reasonable.

[0105] For example, if the name of a base station is Base Station No. 5, B Shopping Mall, Block A, then it can be determined that the scene where the base station is located is a shopping mall. For another example, if a base station is located in School C, then it can be determined that the scene where the base station is located is School C.

[0106] In this way, when performing traffic prediction on a base station to be predicted, the target scenario in which the base station to be predicted is located can be determined first.

[0107] For example, the name of the base station to be predicted is used as the target scene where the base station to be predicted is located; for another example, the name of the block where the base station to be predicted is located and the name of the base station to be predicted are used to comprehensively determine the target scene where the base station to be predicted is located.

[0108] Among them, the above-mentioned base station to be predicted can be a newly built base station or an existing base station, which is reasonable.

[0109] S102: Determine a target flow change pattern corresponding to a target scene from a preset target correspondence relationship between scenes and flow change patterns;

[0110] The target correspondence is established based on the correspondence between the sample scenarios in which each sample base station is located and the traffic change patterns that conform to the change characteristics of the historical traffic data of each sample base station.

[0111] For each scenario, the pattern of changes in the flow of people in each place in different scenarios over time may be different, so the pattern of changes in the communication resource usage in each place in different scenarios over time may also be different. Similarly, the pattern of changes in the flow data of each base station in different scenarios over time may also be different.

[0112] Based on this, when the time range is expanded and the historical traffic data of each base station is analyzed, it can be found that the change characteristics of the historical traffic data of each base station in different scenarios can be different, and then, the traffic change patterns that the change characteristics of the historical traffic data of each base station in different scenarios conform to can also be different. Among them, the so-called traffic change pattern can be understood as: the law that the historical traffic data changes periodically over time.

[0113] For example, for each base station in a school or office building, the historical traffic data shows a change characteristic of low traffic on holidays and high traffic on weekdays. For each base station in a shopping mall or amusement park, the historical traffic data shows a change characteristic of high traffic on holidays and low traffic on weekdays.

[0114] For example, Figure 2(a)-Figure 2(g) is a schematic diagram of various flow change modes, among which Figure 2(a)-Figure 2(g) In the figure, the horizontal axis represents time and the vertical axis represents flow data. Figure 2(a) is a schematic diagram of the mainstream flow change pattern; Figure 2(b) is a schematic diagram of the scenic area flow change pattern; Figure 2(c) is a schematic diagram of the return home flow change pattern; Figure 2(d) is a schematic diagram of the holiday flow change pattern;

[0115] Figure 2(e) is a schematic diagram of the office-type flow change pattern; Figure 2(f) is a schematic diagram of the campus-type flow change pattern; Figure 2(g) is a schematic diagram of the residential-type flow change pattern.

[0116] In this way, before performing traffic prediction on the prediction model, we can first determine each sample base station. Then, for each sample base station, determine the sample scenario in which the sample base station is located and the traffic change pattern that conforms to the changing characteristics of the historical traffic data, and establish a correspondence between the sample scenario and the traffic change pattern, so as to obtain the correspondence between the sample scenario in which each sample base station is located and the traffic change pattern that conforms to the changing characteristics of the historical traffic data of each sample base station.

[0117] Taking into account that the change characteristics of the historical traffic data of sample base stations in different sample scenarios can conform to the same traffic change pattern, it is possible to further extract the sample scene corresponding to each traffic change pattern from the above-mentioned correspondence, thereby establishing a correspondence between the traffic change pattern and the sample scene, that is, obtaining the target correspondence between the scene and the traffic change pattern.

[0118] Among them, in the target correspondence relationship between the above-mentioned scenarios and traffic change patterns, each scenario corresponds to a traffic change pattern, and each traffic change pattern can correspond to one or more scenarios.

[0119] Exemplarily, the semantic representation of the sample base station can be obtained according to the semantic characteristics of the geographical location information, name, etc. of the sample base station, so as to calculate the word library about each sample base station. For example, the base station word library can be shown in Table 1, wherein types 1-8 in Table 1 represent a traffic change mode respectively, the semantic words school, hospital, cabinet and garden represent a scene respectively, and the numbers represent the number of sample base stations in various scenes in each sample base station corresponding to each traffic change mode.

[0120] Table 1

[0121] Semantic words Type 1 Type 2 Type 3 Type 4 Type 5 Type 6 Type 7 Type 8 School 14 11 27 5 8 17 18 3 Hospital 8 4 4 5 3 3 2 3 Cabinet 6 17 2 32 15 8 23 21 garden 13 6 4 3 12 17 6 3

[0122] In the above word library, the correspondence between the scene and the traffic change pattern is established by establishing word vectors. The method of establishing word vectors is as follows:

[0123]

[0124] in, is the word vector of the γth semantic word e in the vocabulary; is the vector corresponding to the βth sample base station name containing the γth semantic word e; α is the number of sample base station names including the semantic word e in the sample base station names of each sample base station; 1≤γ≤n, n is the number of semantic words in the vocabulary.

[0125] Then, the obtained word vectors are standardized using the specified formula shown below to obtain the standardized word library shown in Table 2, where the specified formula is:

[0126]

[0127] in, is the standardized word vector of the γth semantic word e; is the vector corresponding to the βth standardized sample base station name containing the γth semantic word e; u represents the vector after standardized processing.

[0128] The above-mentioned standardized processing formula represents the probability that the sample base station whose name contains the semantic word e belongs to the traffic change pattern corresponding to each scene represented by the semantic word e, then the maximum probability e=max a βγu The corresponding traffic change pattern is the traffic change pattern that the change characteristics of the historical traffic data of the sample base station conform to.

[0129] Table 2

[0130] Semantic words Type 1 Type 2 Type 3 Type 4 Type 5 Type 6 Type 7 Type 8 School 0.1359 0.1068 0.2621 0.0485 0.0777 0.1650 0.1748 0.0291 Hospital 0.2500 0.1250 0.1250 0.1563 0.0938 0.0938 0.0625 0.0938 Cabinet 0.0484 0.1371 0.0161 0.2581 0.1210 0.0645 0.1855 0.1694 garden 0.2031 0.0938 0.0625 0.0469 0.1875 0.2656 0.0938 0.0469

[0131] After determining the target scene where the base station to be predicted is located, the traffic change pattern corresponding to the target scene can be found from the target correspondence between the above scenes and traffic change patterns, and the found traffic change pattern can be used as the target traffic change pattern corresponding to the target scene where the base station to be predicted is located.

[0132] S103: Selecting a target model that matches the target traffic change pattern from a plurality of traffic prediction models obtained by pre-training;

[0133] Among them, each traffic prediction model is trained based on the historical traffic data of a group of sample base stations. The change characteristics of the historical traffic data of this group of sample base stations conform to the same traffic change pattern, and each traffic prediction model is added with sample scene labels of each sample base station used to train the traffic prediction model; the traffic change pattern conformed to the change characteristics of the historical traffic data of each sample base station used to train the target model is the target traffic change pattern.

[0134] When training the above-mentioned multiple traffic prediction models, in order to weaken the feature differences between the base stations used to train each traffic prediction model, improve the accuracy of the traffic prediction model obtained by training, and further improve the accuracy of the traffic prediction results of the model to be predicted, the same traffic prediction model can be trained using sample base stations whose changing characteristics of historical traffic data conform to the same traffic change pattern, while different traffic prediction models can be trained using sample base stations whose changing characteristics of historical traffic data conform to different traffic change patterns.

[0135] That is to say, after determining each sample base station, the change characteristics of the historical traffic data of each sample base station can be further determined, so that each sample base station whose change characteristics of historical traffic data conform to the same traffic change pattern is divided into a group to obtain multiple groups of sample base stations. Among them, the change characteristics of the historical traffic data of each group of sample base stations conform to the same traffic change pattern, that is, the change characteristics of the historical traffic data of each sample base station in each group of sample base stations conform to the same traffic change pattern.

[0136] For each group of sample base stations obtained by division, the historical traffic data of each sample base station in the group can be used to train the model to obtain a traffic prediction model, and the traffic prediction model matches the traffic change pattern consistent with the change characteristics of the historical traffic data of the group of sample base stations.

[0137] Furthermore, since the sample scene where each sample base station is located corresponds to a traffic change pattern, when a traffic prediction model matching each traffic change pattern is obtained, in order to enable the traffic prediction model to further reflect the correspondence between the traffic change pattern and the sample scene, the sample scene labels of each sample base station used to train the traffic prediction model can be added to the traffic prediction model. In this way, for each traffic prediction model, the model can reflect the correspondence between the scene, the traffic change pattern, and the traffic data.

[0138] Thus, after obtaining the target traffic change pattern, a target model matching the target traffic change pattern can be selected from the multiple traffic prediction models obtained by pre-training. The change characteristics of the historical traffic data of each sample base station in the sample base station group used to train the target model conform to the target traffic change pattern.

[0139] S104: Input the target scenario into the target model for training, and obtain a prediction result as the traffic prediction result of the base station to be predicted.

[0140] After determining the target model, the target scenario of the base station to be predicted can be directly input into the target model, so that the target model can train the target scenario, and obtain the flow prediction result corresponding to the base station to be predicted by using the established correspondence between the target scenario, the target flow change pattern and the flow data, and output the flow prediction result. In this way, the electronic device can obtain the output result of the target model, which is the flow prediction result of the base station to be predicted.

[0141] Optionally, the base station to be predicted may be a newly built base station. After obtaining the traffic prediction result of the base station to be predicted, communication resources may be allocated to the base station to be predicted according to the traffic prediction result.

[0142] Optionally, the base station to be predicted can be an existing base station. After obtaining the traffic prediction result of the base station to be predicted, it can be determined whether the communication resources currently allocated to the base station to be predicted are reasonable based on the traffic prediction result, and then, corresponding strategies can be taken to meet the communication resource requirements in the scene where the predicted base station is located, for example, adjusting the communication resources allocated to the base station to be predicted, building a new base station in the scene where the predicted base station is located, etc.

[0143] As can be seen from the above, by applying the solution provided by the embodiment of the present invention, the sample base stations are grouped according to the traffic change patterns that the historical traffic data of the sample base stations conform to, and the groups of sample base stations with the same traffic change patterns that the historical traffic data conform to are used to train the traffic prediction models that match the various traffic change patterns, respectively. This can weaken the feature differences between the base stations used to train each traffic prediction model and improve the accuracy of the various traffic prediction models obtained through training. Furthermore, when the traffic prediction model that matches the traffic change pattern of the base station to be predicted is used to predict the traffic of the base station to be predicted, the accuracy of the prediction results obtained can be improved.

[0144] Optionally, in a specific implementation, such as Figure 3 As shown, a training method for multiple traffic prediction models provided by an embodiment of the present invention may include the following steps S301-S306:

[0145] S301: Determine each sample base station;

[0146] S302: Determine a sample scenario in which each sample base station is located;

[0147] S303: Acquire historical traffic data of each sample base station, and determine the change characteristics of the historical traffic data of each sample base station;

[0148] S304: dividing each sample base station whose change characteristics of historical traffic data conform to the same traffic change pattern into the same group to obtain each group of sample base stations;

[0149] S305: For each group of sample base stations, use the historical traffic data of each base station in the group of sample base stations to train a preset initial model;

[0150] S306: When the initial model meets the preset conditions, stop training, obtain a traffic prediction model that matches the traffic change pattern corresponding to the group of sample base stations, and add sample scene labels of each sample base station in the group of sample base stations to the obtained traffic prediction model;

[0151] The traffic change pattern corresponding to each group of sample base stations is: the traffic change pattern that the change rule of the historical traffic data of each base station in the group of sample base stations conforms to.

[0152] In this specific implementation, each sample base station and the sample scene in which each sample base station is located can be first determined; then, the historical traffic data of each sample base station is obtained, and the change characteristics of the historical traffic data of each sample base station are determined.

[0153] In this way, each sample base station whose change characteristics of historical traffic data conform to the same traffic change pattern can be divided into the same group to obtain each group of sample base stations. Among them, for each group of sample base stations, the traffic change pattern conformed to by the change rules of the historical traffic data of each base station in the group of sample base stations can be used as the traffic change pattern corresponding to the group of sample base stations.

[0154] In the process of training to obtain the above-mentioned multiple traffic prediction models, an initial model can be established first, and then, for each group of sample base stations, the historical traffic data of each sample base station in the group of sample base stations is input into the initial model for training. Thus, when the above-mentioned initial model meets the preset conditions, the training is stopped to obtain a traffic prediction model that matches the traffic change pattern corresponding to the group of sample base stations.

[0155] Among them, for each group of sample base stations, during the training process, the above-mentioned initial model can learn the change characteristics of the historical traffic data of each sample base station in the group of sample base stations, and after learning the change characteristics of a large amount of historical traffic data, gradually establish the corresponding relationship between the traffic change pattern and the traffic data corresponding to the group of sample base stations, and when the initial model meets the preset conditions, stop training to obtain a traffic prediction model that matches the traffic change pattern corresponding to the group of sample base stations.

[0156] Optionally, the above preset condition may be that the number of iterations of each historical traffic data reaches a preset number.

[0157] Optionally, the above preset condition may be that the error between the true value and the predicted value of each historical flow data is less than a preset error.

[0158] Exemplarily, each sample base station is clustered according to the change characteristics of historical traffic data by using a K-means algorithm (k-means clustering algorithm) to obtain multiple clusters after clustering. For each cluster, each sample base station belonging to the cluster is: each sample base station whose change characteristics of historical traffic data conform to the same traffic change pattern.

[0159] Furthermore, the above initial model can be a GRU (Gated Recurrent Unit networks) base station prediction model. For each cluster, the historical traffic data of each sample base station in the cluster is used as the training set and test set data of the traffic prediction model corresponding to the cluster. Thus, the historical traffic data of each sample base station in the cluster is input into the above GRU base station prediction model for training, and multiple traffic prediction models are obtained.

[0160] For example, Figure 4The following is a schematic diagram of the training results of the GRU base station prediction model using the historical traffic data of each sample base station in the cluster corresponding to the holiday traffic change pattern. Figure 4 As shown, the figure includes the trainPredict (training prediction) line and the trainY (training result) line. The traffic change patterns of the above two lines match, which indicates that the traffic prediction result of the traffic prediction model has a small error with the actual traffic data of the sample base station, that is, the traffic prediction result can be the actual traffic data.

[0161] Furthermore, after obtaining a traffic prediction model that matches each traffic change pattern, for each traffic prediction model, sample scene labels of each sample base station used to train the traffic prediction model can be added to the traffic prediction model, so that the traffic prediction model can reflect: the correspondence between the sample scenes of each sample base station used to train the traffic prediction model, the traffic change pattern matching the traffic prediction model, and the traffic data.

[0162] In this way, after determining the target scene in which the base station to be predicted is located, the target scene can be directly input into the traffic prediction model that matches the target traffic change pattern corresponding to the target scene. Thus, the traffic prediction model can be trained with the above-mentioned target scene, and the established correspondence between the target scene, the target traffic change pattern and the traffic data can be used to obtain the traffic prediction result corresponding to the base station to be predicted.

[0163] Optionally, in a specific implementation, such as Figure 5 As shown, the above step S301, determining each sample base station, may include the following steps S3011-S3014:

[0164] S3011: Determine each candidate base station;

[0165] S3012: Grouping the candidate base stations according to their location information to obtain candidate base station groups.

[0166] When determining each sample base station, each candidate base station may be determined first among each base station currently in operation, for example, all base stations in operation within a specified range may be determined as each candidate base station.

[0167] Then, the candidate base stations can be grouped according to the location information of the candidate base stations to obtain candidate base station groups.

[0168] For example, for a certain urban area, all base stations in the urban area can be determined as candidate base stations, and the longitude and latitude of each candidate base station can be used as the location information of each candidate base station. After that, the geographical area of ​​the urban area can be equally divided into squares according to the longitude and latitude to obtain the divided squares. Among them, for each square, the candidate base stations whose locations indicated by the longitude and latitude are located in the square constitute a candidate base station group. In this way, the number of squares obtained by the division is the number of candidate base station groups obtained.

[0169] S3013: For each candidate base station group, calculate the aggregation result of the historical traffic data of each base station in the candidate base station group within a specified time range;

[0170] S3014: Based on the calculated aggregation results, the candidate base station groups that do not meet the specified traffic conditions are eliminated to obtain sample base stations.

[0171] After obtaining each candidate base station group, for each candidate base station group, historical traffic data of each base station in the candidate base station group within a specified time range may be obtained.

[0172] Exemplarily, the above-mentioned specified time range may be two weeks, and the process of obtaining the historical traffic data of each candidate base station within the specified time range may include:

[0173] First, obtain the traffic data sequence of each candidate base station within two weeks, which can be:

[0174] R m =(r m1 , r m2 , r m3 , ..., r ma , ..., r mi )

[0175] Among them, R m is the traffic data sequence of the mth candidate base station within two weeks; r ma is the traffic value of the mth candidate base station within the ath hour when traffic data is generated; i is the total number of hours when the mth candidate base station generates traffic data within two weeks.

[0176] Thus, by using the traffic data sequence of each candidate base station in two weeks, the daily average traffic sequence of the candidate base station in two weeks can be obtained; wherein the daily average traffic sequence of the candidate base station is:

[0177] L mn =(l mn1 , l mn2 , l mn3 , ..., l mnb )...,lmnj ,)

[0178] Among them, L mn is the traffic sequence of the mth candidate base station on the nth day within two weeks; l mnb is the traffic value of the mth candidate base station in the bth hour when the traffic data is generated on the nth day within two weeks; j is the total number of hours when the mth candidate base station generates traffic data on the nth day within two weeks.

[0179] Based on the above two sequences, we can get the formula for calculating the daily average traffic sequence of each candidate base station within two weeks, which is:

[0180]

[0181] Among them, L m is the daily average traffic sequence of the mth base station in two weeks; n is the nth day in two weeks, and 1≤n≤14; L mn is the traffic sequence of the mth base station on the nth day.

[0182] In this way, the obtained daily average flow sequence of each candidate base station within two weeks is the obtained historical flow data of the candidate base station within two weeks.

[0183] Then, the aggregation result of each of the historical traffic data obtained above is calculated to obtain the aggregation result of the historical traffic data of each base station in the candidate base station group within a specified time range.

[0184] In this way, for each candidate base station group, after obtaining the historical traffic data of each base station in the candidate base station within a specified time range, the candidate base station group that does not meet the specified traffic conditions can be eliminated based on the various aggregation results obtained by the above calculation to obtain a sample base station.

[0185] Optionally, in a specific implementation, the above step S3014 may include the following steps 11-12:

[0186] Step 11: for each candidate base station group, based on the calculated aggregation result, the candidate base station group is normalized to obtain a normalized result of the candidate base station group;

[0187] Step 12: Eliminate candidate base station groups whose normalized results are lower than a preset threshold.

[0188] In this specific implementation, for each candidate base station group, the candidate base station group can be standardized based on the calculated aggregation result of the flow data of each base station in the candidate base station group within a specified time range to obtain the standardized result of the candidate base station group. In this way, it can be determined whether the standardized result of the candidate base station group is lower than the preset threshold. If it is lower, the candidate base station group is eliminated, that is, each base station in the candidate base station group cannot be used as a sample base station. If it is not lower, the candidate base station group is retained, that is, each base station in the candidate base station group can be used as a sample base station.

[0189] For example, when the preset threshold is 10%, for a candidate base station group, when the calculated normalized result of the candidate base station group is lower than 10%, the candidate base station group is eliminated, and each base station in the candidate base station group cannot be used as a sample base station.

[0190] Optionally, in a specific implementation, the above step 11 may include the following step 111:

[0191] Step 111: Calculate the normalization result of each candidate base station group using a preset target formula; wherein the target formula is:

[0192] A i =(A 0i -A min )÷(A max -A min )

[0193] Among them, A i is the normalized result of the i-th candidate base station group; A 0i A is the aggregation result calculated for the i-th candidate base station group; min is the preset minimum value; A max is the preset maximum value.

[0194] For example, for the first candidate base station group, A 01 is the aggregation result calculated for the first candidate base station group, and A 01 is 20, set the minimum value A min is 10, the maximum value A max is 50. If the normalized result is less than 10%, the candidate base station group is eliminated; otherwise, the candidate base station group is retained.

[0195] Then, the standardized result of the first candidate base station group can be calculated as: A1=(20-10)÷(80-10)=25%, wherein, since 25%>10%, the first candidate base station group is retained.

[0196] When determining each sample base station, since the historical traffic data of each candidate base station may contain data duplication, data errors, data missing, etc., for each candidate base station group, after obtaining the historical traffic data of each base station in the candidate base station within a specified time range, the obtained historical traffic data can be pre-processed such as data cleaning.

[0197] Correspondingly, optionally, in a specific implementation, such as Figure 6 As shown, the above step S301, determining each sample base station, may also include the following step S601:

[0198] S601: Acquire historical traffic data of each candidate base station within a specified time range, and perform data preprocessing on each acquired historical traffic data to obtain preprocessed historical traffic data of each candidate base station within the specified time range;

[0199] Accordingly, in this specific implementation, the above step S3012: grouping the candidate base stations according to the location information of the candidate base stations to obtain the candidate base station groups may include the following step S602:

[0200] S602: Grouping the pre-processed candidate base stations according to the pre-processed position information of the candidate base stations to obtain groups of candidate base stations.

[0201] Accordingly, in this specific implementation, the above step S3013, for each candidate base station group, calculates the aggregation result of the historical traffic data of each base station in the candidate base station group within the specified time range, which may include the following step S603:

[0202] S603: For each candidate base station group, calculate the aggregation result of the pre-processed historical traffic data of each base station in the candidate base station group within a specified time range.

[0203] In this specific implementation, after obtaining each historical traffic data, each historical traffic data can be preprocessed by data cleaning and other methods, for example, deleting duplicate data, correcting erroneous data, supplementing missing data, etc., so that each preprocessed historical traffic data can be obtained.

[0204] Among them, since part of the historical traffic data may be deleted in the above preprocessing process, the number of each historical traffic data obtained after preprocessing may be equal to the number of each historical traffic data obtained, or may be less than the number of each historical traffic data obtained.

[0205] Afterwards, the preprocessed candidate base stations can be grouped according to their location information to obtain groups of candidate base stations, and for each candidate base station group, the aggregation results of the historical traffic data of the preprocessed base stations in the candidate base station group within a specified time range can be calculated.

[0206] Corresponding to the base station traffic prediction method provided by the above-mentioned embodiment of the present invention, the embodiment of the present invention provides a base station traffic prediction device.

[0207] Figure 7 A schematic diagram of a flow prediction device provided by an embodiment of the present invention is shown in FIG. Figure 7 As shown, the device may include the following modules:

[0208] A scene determination module 710 is used to determine a target scene in which a base station to be predicted is located;

[0209] The mode determination module 720 is used to determine a target traffic change mode corresponding to the target scene from a target correspondence relationship between a preset scene and a traffic change mode; wherein the target correspondence relationship is established based on a correspondence relationship between a sample scene in which each sample base station is located and a traffic change mode that conforms to the change characteristics of the historical traffic data of each sample base station;

[0210] The model determination module 730 is used to select a target model that matches the target traffic change pattern from a plurality of traffic prediction models obtained by pre-training; wherein each traffic prediction model is obtained by training based on the historical traffic data of a group of sample base stations, and the change characteristics of the historical traffic data of the group of sample base stations conform to the same traffic change pattern; and each traffic prediction model is added with sample scene labels of each sample base station used to train the traffic prediction model; the traffic change pattern conformed to by the change characteristics of the historical traffic data of each sample base station used to train the target model is the target traffic change pattern;

[0211] The result acquisition module 740 is used to input the target scenario into the target model for training and obtain a prediction result as the traffic prediction result of the base station to be predicted.

[0212] As can be seen from the above, when applying the solution provided by the embodiment of the present invention, in the process of traffic prediction for the base station to be predicted, it is first necessary to determine the target scene in which the base station to be predicted is located, and then determine the target traffic change pattern corresponding to the above target scene from the target correspondence relationship established based on the correspondence between the sample scenes in which each sample set station is located and the traffic change pattern that conforms to the change characteristics of the historical traffic data of each sample base station, and finally, select the target model that matches the above target traffic change pattern from the multiple traffic prediction models obtained in advance, input the above target scene into the above target model for training, and obtain the prediction result of the model as the traffic prediction result of the above base station to be predicted.

[0213] By applying the solution provided by the embodiment of the present invention, for each traffic prediction model, the change characteristics of the historical traffic data of each sample base station used to train the traffic prediction model conform to the same traffic change pattern, so that each trained traffic prediction model matches a traffic change pattern respectively.

[0214] That is to say, when using multiple sample base stations to construct a traffic prediction model, the change characteristics of the historical traffic data of different sample base stations are taken into account, and the sample base stations with the same change characteristics are divided into a group. Using the divided groups of sample base stations, different traffic prediction models are trained respectively. Therefore, each traffic prediction model is trained based on base stations with the same change characteristics of historical traffic data, rather than trained based on base stations with different change characteristics of historical traffic data.

[0215] In this way, when performing traffic prediction for the base station to be predicted, a traffic prediction model that matches the traffic change pattern of the base station to be predicted can be selected, that is, the traffic change pattern that the historical traffic data of the sample base station used to train the selected traffic prediction model conforms to is the traffic change pattern of the base station to be predicted. Therefore, since the training samples of the selected traffic prediction model have the same change characteristics as the traffic data of the base station to be predicted, when using the selected traffic prediction model to perform traffic prediction for the base station to be predicted, the accuracy of the traffic prediction result can be improved.

[0216] Based on this, the solution provided by the embodiment of the present invention is applied to group the sample base stations according to the traffic change patterns that the historical traffic data of the sample base stations conform to, and use the groups of sample base stations with the same traffic change patterns that the historical traffic data conforms to to respectively train traffic prediction models that match various traffic change patterns, which can weaken the feature differences between the base stations used to train each traffic prediction model and improve the accuracy of each traffic prediction model obtained through training. Furthermore, when the traffic prediction model that matches the traffic change pattern of the base station to be predicted is used to predict the traffic of the base station to be predicted, the accuracy of the obtained prediction results can be improved.

[0217] Optionally, in a specific implementation, the device further includes:

[0218] A determination module, used to determine each sample base station and a sample scene where each sample base station is located;

[0219] A feature determination module obtains historical traffic data of each sample base station and determines the change characteristics of the historical traffic data of each sample base station;

[0220] The sample base station grouping module divides the sample base stations whose change characteristics of historical traffic data conform to the same traffic change pattern into the same group to obtain sample base stations in each group;

[0221] The model training module is used to train a preset initial model for each group of sample base stations using the historical traffic data of each base station in the group of sample base stations; when the initial model meets the preset conditions, the training is stopped to obtain a traffic prediction model that matches the traffic change pattern corresponding to the group of sample base stations, and the sample scene labels of each sample base station in the group of sample base stations are added to the obtained traffic prediction model; wherein the traffic change pattern corresponding to each group of sample base stations is: the traffic change pattern that is consistent with the change rules of the historical traffic data of each base station in the group of sample base stations.

[0222] Optionally, in a specific implementation, the determining module includes:

[0223] The candidate base station determination submodule is used to determine each candidate base station and group each candidate base station according to the location information of each candidate base station to obtain each candidate base station group;

[0224] The aggregation result acquisition submodule is used to calculate the aggregation result of the historical traffic data of each base station in the candidate base station group within a specified time range for each candidate base station group;

[0225] The sample base station acquisition submodule is used to eliminate the candidate base station group that does not meet the specified traffic conditions based on the calculated aggregation results to obtain sample base stations.

[0226] Optionally, in a specific implementation, the device further includes:

[0227] A data acquisition module, used for obtaining historical traffic data of each candidate base station within a specified time range before the step of grouping each candidate base station according to the location information of each candidate base station to obtain each group of candidate base stations, and performing data preprocessing on each acquired historical traffic data to obtain the preprocessed historical traffic data of each candidate base station within the specified time range;

[0228] The candidate base station determination submodule is specifically used to: group the pre-processed candidate base stations according to the pre-processed location information of each candidate base station to obtain each group of candidate base stations;

[0229] The aggregation result acquisition submodule is specifically used to: for each candidate base station group, calculate the aggregation result of the pre-processed historical traffic data of each base station in the candidate base station group within a specified time range.

[0230] Optionally, in a specific implementation, the sample base station acquisition submodule includes:

[0231] A standard processing unit, configured to standardize each candidate base station group based on the calculated aggregation result to obtain a standardized result of the candidate base station group;

[0232] The base station elimination unit is used to eliminate the candidate base station group whose normalization results are lower than a preset threshold.

[0233] Optionally, in a specific implementation, the standard processing unit is specifically used to:

[0234] The standardized results of each candidate base station group are calculated using a preset target formula; wherein the target formula is:

[0235] A i =(A 0i -A min )÷(A max -A min )

[0236] Among them, A i is the normalized result of the i-th candidate base station group; A 0i A is the aggregation result calculated for the i-th candidate base station group; min is the preset minimum value; A max is the preset maximum value.

[0237] Corresponding to the flow prediction device provided by the above embodiment of the present invention, the embodiment of the present invention also provides an electronic device, such as Figure 8 As shown, it includes a processor 801, a communication interface 802, a memory 803 and a communication bus 804, wherein the processor 801, the communication interface 802, and the memory 803 communicate with each other through the communication bus 804.

[0238] Memory 803, used for storing computer programs;

[0239] The processor 801 is used to implement the steps of any base station traffic prediction method provided by the above-mentioned embodiment of the present invention when executing the program stored in the memory 803.

[0240] The communication bus mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0241] The communication interface is used for communication between the above electronic device and other devices.

[0242] The memory may include a random access memory (RAM) or a non-volatile memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0243] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0244] In another embodiment provided by the present invention, a computer-readable storage medium is also provided, which stores a computer program. When the computer program is executed by a processor, the steps of any base station traffic prediction method provided by the above-mentioned embodiments of the present invention are implemented.

[0245] In another embodiment provided by the present invention, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute the steps of any base station traffic prediction method in the above-mentioned embodiments.

[0246] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk Solid State Disk (SSD)), etc.

[0247] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0248] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, electronic device embodiment, computer-readable storage medium embodiment, and computer program product embodiment, since they are basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0249] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A base station traffic prediction method, characterized in that: The method comprises: Determine the target scene where the base station to be predicted is located; Determine a target traffic change pattern corresponding to the target scene from a target correspondence relationship between a preset scene and a traffic change pattern; wherein the target correspondence relationship is established based on a correspondence relationship between a sample scene in which each sample base station is located and a traffic change pattern that conforms to the change characteristics of historical traffic data of each sample base station; From a plurality of traffic prediction models obtained by pre-training, a target model matching the target traffic change pattern is selected; wherein each traffic prediction model is trained based on historical traffic data of a group of sample base stations, and the change characteristics of the historical traffic data of the group of sample base stations conform to the same traffic change pattern; and each traffic prediction model is added with sample scene labels of each sample base station used to train the traffic prediction model; the traffic change pattern conformed to by the change characteristics of the historical traffic data of each sample base station used to train the target model is the target traffic change pattern; Inputting the target scenario into the target model for training, and obtaining a prediction result as a traffic prediction result of the base station to be predicted; The training method of the multiple traffic prediction models includes: determining each sample base station and the sample scene in which each sample base station is located; obtaining the historical traffic data of each sample base station and determining the change characteristics of the historical traffic data of each sample base station; dividing each sample base station whose change characteristics of the historical traffic data conform to the same traffic change pattern into the same group to obtain each group of sample base stations; for each group of sample base stations, using the historical traffic data of each base station in the group of sample base stations to train a preset initial model; when the initial model meets the preset conditions, stop training to obtain a traffic prediction model that matches the traffic change pattern corresponding to the group of sample base stations, and add the sample scene labels of each sample base station in the group of sample base stations to the obtained traffic prediction model; wherein, the traffic change pattern corresponding to each group of sample base stations is: the traffic change pattern conformed to the change rules of the historical traffic data of each base station in the group of sample base stations.

2. The method according to claim 1, characterized in that The step of determining each sample base station includes: Determine each candidate base station, and group each candidate base station according to location information of each candidate base station to obtain each candidate base station group; For each candidate base station group, calculate the aggregated result of the historical traffic data of each base station in the candidate base station group within a specified time range; Based on the calculated aggregation results, the candidate base station groups that do not meet the specified traffic conditions are eliminated to obtain sample base stations.

3. The method according to claim 2, characterized in that Before the step of grouping the candidate base stations according to the location information of the candidate base stations to obtain the candidate base station groups, the method further includes: Acquire historical traffic data of each candidate base station within a specified time range, and perform data preprocessing on each acquired historical traffic data to obtain preprocessed historical traffic data of each candidate base station within the specified time range; The step of grouping the candidate base stations according to the location information of the candidate base stations to obtain the candidate base station groups includes: grouping the pre-processed candidate base stations according to the pre-processed location information of the candidate base stations to obtain the candidate base station groups; The step of calculating, for each candidate base station group, an aggregate result of historical traffic data of each base station in the candidate base station group within a specified time range includes: For each candidate base station group, the aggregation result of the pre-processed historical traffic data of each base station in the candidate base station group within a specified time range is calculated.

4. The method according to claim 2, characterized in that: The step of eliminating the candidate base station group that does not meet the specified traffic condition based on the calculated aggregation results includes: For each candidate base station group, based on the calculated aggregation result, the candidate base station group is normalized to obtain a normalized result of the candidate base station group; Eliminate the candidate base station groups whose normalized results are lower than a preset threshold.

5. The method according to claim 4, characterized in that The step of normalizing each candidate base station group based on the calculated aggregation result to obtain the normalized result of the candidate base station group includes: The standardized results of each candidate base station group are calculated using a preset target formula; wherein the target formula is: A i =(A 0i -A min )÷(A max -A min ) Among them, A i is the normalized result of the i-th candidate base station group; A 0i A is the aggregation result calculated for the i-th candidate base station group; min is the preset minimum value; A max is the preset maximum value.

6. A base station traffic prediction device, characterized in that: The device comprises: A scene determination module, used to determine the target scene where the base station to be predicted is located; A mode determination module, used to determine a target traffic change mode corresponding to a target scene from a target correspondence relationship between a preset scene and a traffic change mode; wherein the target correspondence relationship is established based on a correspondence relationship between a sample scene in which each sample base station is located and a traffic change mode that conforms to the change characteristics of historical traffic data of each sample base station; A model determination module is used to select a target model that matches the target traffic change pattern from a plurality of traffic prediction models obtained by pre-training; wherein each traffic prediction model is obtained by training based on historical traffic data of a group of sample base stations, and the change characteristics of the historical traffic data of the group of sample base stations conform to the same traffic change pattern; and each traffic prediction model is added with sample scene labels of each sample base station used to train the traffic prediction model; the traffic change pattern conformed to by the change characteristics of the historical traffic data of each sample base station used to train the target model is the target traffic change pattern; A result acquisition module, used for inputting the target scenario into the target model for training, and acquiring a prediction result as a traffic prediction result of the base station to be predicted; The device also includes: a determination module, which is used to determine each sample base station and the sample scene in which each sample base station is located; a feature determination module, which obtains the historical traffic data of each sample base station and determines the change characteristics of the historical traffic data of each sample base station; a sample base station grouping module, which divides each sample base station whose change characteristics of historical traffic data conform to the same traffic change pattern into the same group to obtain each group of sample base stations; a model training module, which is used to train a preset initial model for each group of sample base stations using the historical traffic data of each base station in the group of sample base stations; when the initial model meets the preset conditions, the training is stopped to obtain a traffic prediction model that matches the traffic change pattern corresponding to the group of sample base stations, and the sample scene labels of each sample base station in the group of sample base stations are added to the obtained traffic prediction model; wherein, the traffic change pattern corresponding to each group of sample base stations is: the traffic change pattern conformed to the change rules of the historical traffic data of each base station in the group of sample base stations.

7. The device according to claim 6, characterized in that The determination module comprises: The candidate base station determination submodule is used to determine each candidate base station and group each candidate base station according to the location information of each candidate base station to obtain each candidate base station group; The aggregation result acquisition submodule is used to calculate the aggregation result of the historical traffic data of each base station in the candidate base station group within a specified time range for each candidate base station group; The sample base station acquisition submodule is used to eliminate the candidate base station group that does not meet the specified traffic conditions based on the calculated aggregation results to obtain sample base stations.

8. The device according to claim 7, characterized in that The device also includes: A data acquisition module, used for obtaining historical traffic data of each candidate base station within a specified time range before the step of grouping each candidate base station according to the location information of each candidate base station to obtain each group of candidate base stations, and performing data preprocessing on each acquired historical traffic data to obtain the preprocessed historical traffic data of each candidate base station within the specified time range; The candidate base station determination submodule is specifically used to: group the pre-processed candidate base stations according to the pre-processed location information of each candidate base station to obtain each group of candidate base stations; The aggregation result acquisition submodule is specifically used to: for each candidate base station group, calculate the aggregation result of the pre-processed historical traffic data of each base station in the candidate base station group within a specified time range.

9. The device according to claim 7, characterized in that The sample base station acquisition submodule includes: A standard processing unit, configured to standardize each candidate base station group based on the calculated aggregation result to obtain a standardized result of the candidate base station group; The base station elimination unit is used to eliminate the candidate base station group whose normalization results are lower than a preset threshold.

10. The device according to claim 9, characterized in that The standard processing unit is specifically used for: The standardized results of each candidate base station group are calculated using a preset target formula; wherein the target formula is: A i =(A 0i -A min )÷(A max -A min ) Among them, A i is the normalized result of the i-th candidate base station group; A 0i A is the aggregation result calculated for the i-th candidate base station group; min is the preset minimum value; A max is the preset maximum value.

11. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, for implementing the method steps described in any one of claims 1 to 5 when executing a program stored in a memory.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps described in any one of claims 1 to 5 are implemented.

Citation Information

Patent Citations

  • Resource expansion planning method and system

    CN111294812A

  • Abnormal event detection method and device

    CN111598610A

  • Mobile network base station traffic prediction method

    CN112054943A

  • Festival and holiday wireless flow prediction method, system and device based on correlation clustering hybrid algorithm model and medium

    CN112954726A

  • Network traffic prediction method, equipment, storage medium and device

    CN113746696A