Training methods, devices, equipment, and storage media for community azimuth prediction models

Abstract

This invention discloses a method, apparatus, device, and storage medium for training a cell azimuth prediction model. The method includes: processing simulated prediction data to obtain sample point data; the sample point data includes cell sample data and sample labels, wherein the cell sample data contains at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength; and the sample label is the cell azimuth angle set in the simulation. The sample point data is then input into a neural network model for training. Training ends when the trained neural network model meets preset conditions, and the neural network model meeting the preset conditions is used as the cell azimuth prediction model. Obtaining sample point data through simulated prediction data overcomes the problem of manual measurement, greatly expands the number of sample point data, and thus improves the accuracy of the cell azimuth prediction model.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and specifically to a method, apparatus, device, and storage medium for training a cell azimuth prediction model. Background Technology

[0002] Azimuth is an important parameter describing the orientation of an antenna in a mobile communication network. Azimuth can be understood as the angle obtained by rotating a plane facing due north clockwise until it coincides with the plane where the antenna is located. In actual antenna placement, azimuth angles are typically 0 degrees, 120 degrees, and 240 degrees, corresponding to cells A, B, and C, respectively.

[0003] In existing technologies, the cell azimuth is usually predicted using a cell azimuth prediction model. The training process of this cell azimuth prediction model requires manual measurement of the azimuth as training samples, which results in a small number of samples and easily leads to low accuracy of the model prediction results. Summary of the Invention

[0004] In view of the above problems, embodiments of the present invention are proposed to provide a method, apparatus, device and storage medium for training a cell azimuth prediction model that overcomes or at least partially solves the above problems.

[0005] According to one aspect of the present invention, a method for training a cell azimuth prediction model is provided, comprising:

[0006] The simulation prediction data is processed to obtain sample point data; the sample point data includes cell sample data and sample labels. The cell sample data includes at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. The sample label is the cell azimuth angle set in the simulation.

[0007] The sample point data is input into the neural network model for training. The training ends when the trained neural network model meets the preset conditions. The neural network model that meets the preset conditions is used as the azimuth prediction model of the cell.

[0008] According to another aspect of the present invention, a cell azimuth prediction method is provided, comprising:

[0009] The cell parameters of the target cell are obtained by processing the minimized drive test MDT user data. The cell parameters include at least one of the following: sample point latitude and longitude, base station latitude and longitude, and sample point signal strength.

[0010] The cell parameters are input into the cell azimuth prediction model obtained by the above-mentioned cell azimuth prediction model training method to make predictions and obtain the predicted azimuth of the target cell.

[0011] According to another aspect of the present invention, a cell azimuth prediction model training device is provided, comprising:

[0012] The sample data acquisition module is used to process the simulation prediction data to obtain sample point data. The sample point data includes cell sample data and sample tags. The cell sample data includes at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. The sample tag is the cell azimuth angle set in the simulation.

[0013] The model training module is used to input the sample point data into the neural network model for training. When the trained neural network model meets the preset conditions, the training ends, and the neural network model that meets the preset conditions is used as the azimuth prediction model of the cell.

[0014] According to another aspect of the present invention, a cell azimuth prediction device is provided, comprising:

[0015] The parameter acquisition module is used to process the minimized drive test MDT user data to obtain the cell parameters of the target cell. The cell parameters include at least one of the following: sample point latitude and longitude, base station latitude and longitude, and sample point signal strength.

[0016] The prediction module is used to input the cell parameters into the cell azimuth prediction model obtained by the above-mentioned cell azimuth prediction model training method to make predictions and obtain the predicted azimuth of the target cell.

[0017] According to another aspect of the present invention, a computing device is provided, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;

[0018] The memory is used to store at least one executable instruction, which causes the processor to perform the operation corresponding to the above-mentioned cell azimuth prediction model training method and the operation corresponding to the above-mentioned cell azimuth prediction method.

[0019] According to another aspect of the present invention, a computer storage medium is provided, wherein the storage medium stores at least one executable instruction, the executable instruction causing a processor to perform operations corresponding to the above-described cell azimuth prediction model training method, and to perform operations corresponding to the above-described cell azimuth prediction method.

[0020] According to the solution provided in the above embodiments of the present invention, obtaining sample point data through simulation prediction data can overcome the problem of manual measurement, greatly expand the number of sample point data, and thus improve the accuracy of the cell azimuth prediction model.

[0021] The above description is merely an overview of the technical solutions of the embodiments of the present invention. In order to better understand the technical means of the embodiments of the present invention and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of the present invention more obvious and understandable, specific implementation methods of the embodiments of the present invention are described below. Attached Figure Description

[0022] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0023] Figure 1 The flowchart of the cell azimuth prediction model training method provided in Embodiment 1 of the present invention is shown.

[0024] Figure 2 This paper shows a data extraction flowchart in the cell azimuth prediction model training method provided in Embodiment 1 of the present invention;

[0025] Figure 3 This diagram illustrates the data rasterization process in the cell azimuth prediction model training method provided in Embodiment 1 of the present invention.

[0026] Figure 4 A flowchart of the cell azimuth prediction method provided in Embodiment 2 of the present invention is shown;

[0027] Figure 5 A flowchart of the cell azimuth prediction method provided in Embodiment 3 of the present invention is shown;

[0028] Figure 6 This diagram illustrates the structure of the cell azimuth prediction model training device provided in an embodiment of the present invention.

[0029] Figure 7 A schematic diagram of the structure of the cell azimuth prediction device provided in an embodiment of the present invention is shown;

[0030] Figure 8 A schematic diagram of the structure of a computing device provided in an embodiment of the present invention is shown. Detailed Implementation

[0031] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0032] Example 1

[0033] Figure 1 A flowchart illustrating the cell azimuth prediction model training method provided in Embodiment 1 of the present invention is shown. Figure 1 As shown, the method includes the following steps:

[0034] Step S110: Process the simulation prediction data to obtain sample point data; the sample point data includes cell sample data and sample tags. The cell sample data includes at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. The sample tag is the cell azimuth angle set in the simulation.

[0035] The simulation prediction data includes at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. The sample point signal strength can be represented by the Reference Signal Receiver Power (RSRP). Specifically, the simulation prediction data is obtained through simulation, for example, by conducting cell simulation experiments based on electronic maps and base station operating parameter data. The base station operating parameter data includes at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength.

[0036] Step S120: Input the sample point data into the neural network model for training. When the trained neural network model meets the preset conditions, the training ends. The neural network model that meets the preset conditions is used as the cell azimuth prediction model.

[0037] The preset condition can be the loss function in the neural network model, and training ends when the loss function converges. Alternatively, during training, the predicted cell azimuth angles output by the neural network model are detected, and training ends when the accuracy of the predicted cell azimuth angles reaches a preset threshold (e.g., 99%).

[0038] Specifically, in order to reduce the granularity of the azimuth category interval, the above neural network model can choose the VGG model. The VGG model is suitable for larger sample point data, which makes the azimuth category classification more accurate and improves the accuracy of azimuth prediction.

[0039] This embodiment obtains sample point data through simulation prediction data, which can overcome the problem of manual measurement, greatly expand the number of sample point data, and thus improve the accuracy of the cell azimuth prediction model.

[0040] In one alternative embodiment, the processing of simulation test data includes at least one of data extraction, outlier removal, data rasterization, and data normalization.

[0041] In one alternative embodiment, Figure 2 The flowchart illustrates the data extraction process in the cell azimuth prediction model training method provided in Embodiment 1 of the present invention. For example... Figure 2 As shown, the method includes the following steps:

[0042] Step S210: Extract the valid fields from the simulation prediction data.

[0043] For the simulation prediction data, the effective fields required by the neural network model are extracted after field filtering: cell identifier, sample point latitude and longitude, base station latitude and longitude, sample point signal strength, etc.

[0044] Step S220: Obtain the number of sample points under each cell identifier (unique identifier) ​​and determine whether the number of sample points is greater than M.

[0045] Where M is a preset screening value M. If the number of sample points is less than M, then step S230 is executed; otherwise, step S240 is executed.

[0046] Step S230: Remove the sample point data under this cell.

[0047] Step S240: Retain the sample point data under this cell.

[0048] In one optional embodiment, the outlier removal process employs a local anomaly factor algorithm to detect and filter outlier sample data. This prevents the neural network model from extracting sample data with excessively high weights due to interfering features such as outlier distribution during training. The local anomaly factor algorithm is as follows:

[0049] When k=t, define D in the sample point dataset. k (x ( i ) ) represents x ( i ) The k-distance, where x (k=t) Indicates distance x ( i ) The kth farthest data sample point:

[0050]

[0051] Define x ( i ) To x ( j ) Reachability Distance (RD) t (x ( i ) , x ( j ) ):

[0052]

[0053] Define x ( i ) Local Reachability Density (LRD) k (x ( i ) ):

[0054]

[0055] Define x ( i ) Local Outlier Factor (LOF) k (x ( i ) ):

[0056]

[0057] Where, x ( i ) Local anomaly factor LOF k (x ( i ) If the value of ) is less than or equal to 1, then x is considered to be... k These are normal sample points; if LOF k (x ( i ) If the value of ) is greater than 1, it is considered an outlier. Outlier points are removed as exceptions.

[0058] Outlier removal can eliminate the influence of outlier samples and other interfering features on model training, improve the model's fitting speed, and thus enhance the model's training efficiency and prediction accuracy.

[0059] In one alternative embodiment, data rasterization and data normalization processes can be performed simultaneously. Figure 3 The flowchart illustrates the data rasterization process in the cell azimuth prediction model training method provided in Embodiment 1 of the present invention. For example... Figure 3 As shown, the method includes the following steps:

[0060] Step S310: Obtain initial sample point data based on simulation prediction data.

[0061] The initial sample point data can be simulation test data or data extracted from simulation prediction data and outlier removed.

[0062] Step S320: Set the grid coordinate unit length X meters and the size of the grid matrix (A, B).

[0063] Step S330: With the base station as the center and the north direction as the Y-axis, the cell signal coverage area is rasterized according to the ratio of X meters to X meters. The number of stored sample points and the average received signal strength RSRP value in each grid are calculated to generate the cell signal strength distribution matrix.

[0064] Step S340: Calculate the value range of the grid matrix RSRP corresponding to each cell, and normalize the grid matrix according to the value range.

[0065] This embodiment determines the coordinate unit length of the grid, records the number of sample points and the received average signal strength (RSRP) within each grid, and uses these data to generate the cell signal strength distribution matrix. The rasterization method in this embodiment can preserve the continuous distribution characteristics of the signal, supports more precise pixelation of the grid matrix, and makes the prediction results more accurate.

[0066] In an alternative embodiment, the normalized raster matrix can be dimensionality reduced to generate a data type that the neural network model can accept.

[0067] In one alternative embodiment, to improve the training efficiency and prediction accuracy of the model, data normalization uses a zero-mean standardization method:

[0068]

[0069] in, and These are the mean and variance of the elements in the raster matrix, respectively.

[0070] Example 2

[0071] Figure 4 A flowchart of the cell azimuth prediction method provided in Embodiment 2 of the present invention is shown. Figure 4 As shown, the method includes the following steps:

[0072] Step S410: Process the minimized drive test MDT user data to obtain the cell parameters of the target cell. The cell parameters include at least one of the following: sample point latitude and longitude, base station latitude and longitude, and sample point signal strength.

[0073] The MDT user data refers to user data containing at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. Processing of the MDT user data includes at least one of the following: data extraction, outlier removal, data rasterization, and data normalization. The specific processing steps for each method are detailed in the simulation test data processing section above and will not be repeated here.

[0074] Step S420: Input the cell parameters into the cell azimuth prediction model obtained by the cell azimuth prediction model training method of this embodiment of the invention for prediction, and obtain the predicted azimuth of the target cell.

[0075] This embodiment uses the cell azimuth prediction model obtained by the cell azimuth prediction model training method of the present invention for prediction, which can improve the accuracy of cell azimuth prediction.

[0076] In an optional embodiment, the method of this implementation further includes:

[0077] Based on the set azimuth deviation, determine whether there is a large deviation in the predicted azimuth of the target cell, and generate a report of cells with large predicted azimuth deviations.

[0078] Specifically, the output data of the cell azimuth prediction model is analyzed and processed to output the azimuth interval with the highest probability for each batch of data. Corresponding reports are generated based on the model output; for example, reports on cells with excessive deviations from the predicted azimuth are output based on a set azimuth deviation threshold.

[0079] In an optional embodiment, the method of this implementation further includes:

[0080] Based on the on-site survey data, the azimuth data of the community report that had a large deviation was corrected, and the corrected azimuth data of the community was used as sample point data to update the azimuth prediction model of the community.

[0081] Based on the community reports, on-site surveys can be conducted to confirm the accuracy of the community azimuth prediction model's output. Inaccurate azimuth angles in the output can be corrected and used as sample point data to retrain the model, further improving the model's prediction accuracy.

[0082] In this embodiment, after the model training is completed, the MDT user data is input into the model to obtain the predicted classification (azimuth interval) of the cell's azimuth. This method has high prediction accuracy and can be iteratively optimized based on subsequent feedback, making it highly maintainable and reusable.

[0083] Example 3

[0084] Figure 5 A flowchart of the cell azimuth prediction method provided in Embodiment 3 of the present invention is shown. This embodiment is divided into a data cleaning module, an outlier detection module, a data rasterization module, a data normalization module, a model training module, and an output prediction module. The present invention trains a neural network model based on simulation prediction data. The neural network model in this embodiment is the VGG model, and antenna azimuth prediction is performed based on MDT user data.

[0085] Specifically, the data cleaning module receives simulation prediction data and MDT user data, and extracts the effective fields required by the neural network model through field filtering: sample point latitude and longitude, sample point RSRP value, cell unique identifier code and base station latitude and longitude, etc. It obtains the number of sample points under each cell unique identifier code and sets the filtering value M. If the number of sample points is less than M, the sample point data under that cell is removed; otherwise, the sample point data under that cell is retained.

[0086] Since the simulation test data and MDT user data in the cell area are basically densely distributed in local areas with a small number of outlier sample points scattered around, in order to prevent the VGG model from assigning too much weight to interfering features such as outlier distribution during training, the outlier detection module adopts the Local Anomaly Factor (LOF) algorithm to detect and filter outlier sample points. The LEF algorithm is described in the above embodiment and will not be repeated here.

[0087] The data rasterization module receives sample data filtered by the outlier detection module, and sets the grid coordinate unit length X meters and the size of the grid matrix (A, B). This module rasterizes the cell coverage area with the base station as the center and due north as the Y-axis, according to the grid size of X meters * X meters. It calculates the number of sample points stored in each grid, the average received signal strength RSRP value, and generates a cell signal strength distribution map.

[0088] The data normalization module accepts a raster matrix, calculates the range of RSRP values ​​for each cell's corresponding raster matrix, and normalizes the raster matrix based on this range. Furthermore, it performs dimensionality reduction on the normalized raster matrix to generate data types acceptable to the VGG model. The data normalization uses a zero-mean standardization method, as described in the relevant embodiments above, and will not be repeated here.

[0089] Because the training set of sample points constructed from simulation test data is large and contains multi-dimensional features such as signal strength distribution and environmental terrain, a deeper VGG model, which acquires wider feature maps, is used as the classification model. The model input data is a normalized and dimensionality-reduced raster matrix of simulation prediction data. The classification granularity is set to N degrees, and the azimuth angle is divided into... The category, or data label, is the interval category to which the azimuth of the cell belongs, as set by the simulation prediction data.

[0090] The output prediction module takes the normalized raster matrix of MDT user data as input, obtains the model's output data, analyzes and processes it, and outputs the azimuth interval with the highest probability for each batch of data. Based on the model output, corresponding cell reports are generated. For example, based on a set deviation threshold, reports on cells with excessive deviations from the predicted azimuth are output. Furthermore, the azimuth data of cells with large deviations in the cell reports can be corrected based on field survey data, and the corrected azimuth data is used as sample point data to update the cell azimuth prediction model.

[0091] This embodiment predicts cell azimuth angles based on MDT user data and simulation prediction data. The data is sufficient to support accurate cell azimuth angle prediction.

[0092] Example 4

[0093] Figure 6 A schematic diagram of the structure of the cell azimuth prediction model training device provided in an embodiment of the present invention is shown. Figure 6 As shown, the device includes: a sample data acquisition module 11 and a model training module 12; wherein,

[0094] The sample data acquisition module 11 is used to process the simulation prediction data to obtain sample point data; the sample point data includes cell sample data and sample tags. The cell sample data includes at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. The sample tag is the cell azimuth angle set in the simulation.

[0095] The model training module 12 is used to input the sample point data into the neural network model for training. When the trained neural network model meets the preset conditions, the training ends and the neural network model that meets the preset conditions is used as the cell azimuth prediction model.

[0096] Furthermore, the sample data acquisition module 11 is specifically used for at least one of the following processes: data extraction, outlier removal, data rasterization, and data normalization.

[0097] Furthermore, the neural network model is a VGG model.

[0098] The cell azimuth prediction model training device described in this embodiment of the invention is used to execute the cell azimuth prediction model training method described in the above embodiment. Its working principle and technical effect are similar, and will not be repeated here.

[0099] Example 5

[0100] Figure 7 A schematic diagram of the cell azimuth prediction device provided in an embodiment of the present invention is shown. Figure 7As shown, the device includes: a parameter acquisition module 21 and a prediction module 22; wherein,

[0101] The parameter acquisition module 21 is used to process the minimized drive test MDT user data to obtain the cell parameters of the target cell. The cell parameters include at least one of the following: sample point latitude and longitude, base station latitude and longitude, and sample point signal strength.

[0102] The prediction module 22 is used to input the cell parameters into the cell azimuth prediction model obtained by the above-mentioned cell azimuth prediction model training method to make predictions and obtain the predicted azimuth of the target cell.

[0103] Furthermore, the device also includes: a cell report generation module 23; wherein,

[0104] The cell report generation module 23 is used to determine whether there is a large deviation in the predicted azimuth of the target cell based on the set azimuth deviation, and to generate a cell report with a large predicted azimuth deviation.

[0105] Furthermore, the device also includes: a model update module 24; wherein,

[0106] The model update module 24 is used to correct the azimuth data of the community report that has a large deviation based on the field survey data, and to update the community azimuth prediction model using the corrected azimuth data as sample point data.

[0107] The cell azimuth prediction device described in this embodiment of the invention is used to execute the cell azimuth prediction method described in the above embodiment. Its working principle and technical effect are similar, and will not be repeated here.

[0108] Example 6

[0109] This invention provides a non-volatile computer storage medium storing at least one executable instruction that can execute the cell azimuth prediction model training method and the cell azimuth prediction method in any of the above method embodiments.

[0110] Example 7

[0111] Figure 8 The diagram shows a structural schematic of a computing device provided in an embodiment of the present invention. The specific embodiments of the present invention do not limit the specific implementation of the computing device.

[0112] like Figure 8 As shown, the computing device may include a processor, a communications interface, memory, and a communications bus.

[0113] The processor, communication interface, and memory communicate with each other via a communication bus. The communication interface is used to communicate with other network elements, such as clients or other servers. The processor executes programs, specifically the relevant steps in the embodiments of the cell azimuth prediction model training method and cell azimuth prediction method for computing devices described above.

[0114] Specifically, the program may include program code, which includes computer operation instructions.

[0115] The processor may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The computing device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0116] Memory is used to store programs. Memory may include high-speed RAM, and may also include non-volatile memory, such as at least one disk drive.

[0117] Specifically, the program can be used to enable the processor to execute the cell azimuth prediction model training method and the cell azimuth prediction method in any of the above method embodiments. The specific implementation of each step in the program can be found in the corresponding steps and units described in the above embodiments of the cell azimuth prediction model training method and cell azimuth prediction method, and will not be repeated here. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the devices and modules described above can be referred to the corresponding process descriptions in the foregoing method embodiments, and will not be repeated here.

[0118] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, the embodiments of the present invention are not directed to any particular programming language. It should be understood that the embodiments of the present invention described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of the embodiments of the present invention.

[0119] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0120] Similarly, it should be understood that, in order to streamline the embodiments of the invention and aid in understanding one or more of the various inventive aspects, features of the embodiments of the invention are sometimes grouped together in a single embodiment, figure, or description thereof in the above description of exemplary embodiments of the invention. However, this disclosure should not be construed as reflecting an intention that the claimed embodiments of the invention require more features than are expressly recited in each claim. Rather, as reflected in the following claims, inventive aspects lie in fewer than all features of a single foregoing disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the invention.

[0121] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0122] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination.

[0123] The various component embodiments of the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components according to the embodiments of the present invention. The embodiments of the present invention can also be implemented as device or apparatus programs (e.g., computer programs and computer program products) for performing part or all of the methods described herein. Such programs implementing the embodiments of the present invention can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0124] It should be noted that the above embodiments are illustrative of the present invention and not restrictive of the invention, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. Embodiments of the present invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. A method for training a community azimuth prediction model, characterized in that, include: Data processing is performed on the simulation prediction data to obtain sample point data. The sample point data includes cell sample data and sample labels. The cell sample data includes at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. The sample label is the cell azimuth angle set in the simulation. Initial sample point data is obtained based on the simulation prediction data. The coordinate unit length X meters and the size of the grid matrix are set. The cell signal coverage area is gridded with the base station as the center and north as the Y-axis, according to X meters * X meters. The number of sample points stored in each grid and the received average signal strength RSRP value are calculated to generate a cell signal strength distribution matrix. The range of RSRP values ​​for each cell's corresponding grid matrix is ​​calculated, and the grid matrix is ​​normalized according to the range of values. The sample point data is input into the neural network model for training. The training ends when the trained neural network model meets the preset conditions. The neural network model that meets the preset conditions is used as the azimuth prediction model of the cell.

2. The method according to claim 1, characterized in that, The data processing includes at least one of the following: data extraction and outlier removal.

3. The method according to claim 1 or 2, characterized in that, The neural network model is the VGG model.

4. A method for predicting the azimuth angle of a residential area, characterized in that, include: The cell parameters of the target cell are obtained by processing the minimized drive test MDT user data. The cell parameters include at least one of the following: sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. The cell parameters are input into the cell azimuth prediction model obtained by the cell azimuth prediction model training method according to any one of claims 1-3 for prediction, so as to obtain the predicted azimuth of the target cell.

5. The method according to claim 4, characterized in that, The method further includes: Based on the set azimuth deviation, determine whether there is a large deviation in the predicted azimuth of the target cell, and generate a report of cells with large predicted azimuth deviations.

6. The method according to claim 5, characterized in that, The method further includes: Based on the on-site survey data, the azimuth data of the community report that had a large deviation was corrected, and the corrected azimuth data of the community was used as sample point data to update the azimuth prediction model of the community.

7. A training device for a community azimuth prediction model, characterized in that, include: The sample data acquisition module is used to process simulation prediction data to obtain sample point data. The sample point data includes cell sample data and sample labels. The cell sample data contains at least one of the following: cell identifier, sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. The sample label is the cell azimuth angle set in the simulation. Specifically, initial sample point data is acquired based on the simulation prediction data. The coordinate unit length X meters and the size of the grid matrix are set. The cell signal coverage area is gridded with the base station as the center and north as the Y-axis, according to X meters * X meters. The number of sample points stored in each grid and the received average signal strength RSRP value are calculated to generate a cell signal strength distribution matrix. The range of RSRP values ​​for each cell's corresponding grid matrix is ​​calculated, and the grid matrix is ​​normalized according to the range of values. The model training module is used to input the sample point data into the neural network model for training. When the trained neural network model meets the preset conditions, the training ends, and the neural network model that meets the preset conditions is used as the azimuth prediction model of the cell.

8. A device for predicting the azimuth angle of a residential area, characterized in that, include: The parameter acquisition module is used to process the minimized drive test MDT user data to obtain the cell parameters of the target cell. The cell parameters include at least one of the following: sample point latitude and longitude, base station latitude and longitude, and sample point signal strength. The prediction module is used to input the cell parameters into the cell azimuth prediction model obtained by the cell azimuth prediction model training method according to any one of claims 1-3 for prediction, so as to obtain the predicted azimuth of the target cell.

9. A computing device, comprising: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction, which causes the processor to perform the operation corresponding to the cell azimuth prediction model training method as described in any one of claims 1-3, and to perform the operation corresponding to the cell azimuth prediction method as described in any one of claims 4-6.

10. A computer storage medium storing at least one executable instruction, the executable instruction causing a processor to perform an operation corresponding to the cell azimuth prediction model training method as described in any one of claims 1-3, and to perform an operation corresponding to the cell azimuth prediction method as described in any one of claims 4-6.

Images