Base station load prediction method and device and readable storage medium

Abstract

The invention provides a base station load prediction method and device and a readable storage medium, and relates to the technical field of networks, and the method comprises the steps: generating corresponding load index data according to the flow, utilization rate and user number collected by a base station every preset time period; sequentially inputting the generated load index data into a trained neural network model to enable the trained neural network model to fit the input load index data into a curve, and predicting the load type of the base station according to the curve; and performing advanced intervention by adopting an optimization means according to the predicted load type of the base station. According to the method, the device and the readable storage medium, the problems that the existing base station load prediction method needs network optimization personnel to monitor in real time, manpower is consumed, problems cannot be timely and actively found, and the base station load change trend cannot be directly obtained can be solved.

Description

Technical Field

[0001] The present invention relates to the field of network technology, and in particular to a base station load prediction method, device and readable storage medium. Background Art

[0002] With the rapid development of mobile communication technology, base stations, as the core infrastructure of wireless communication networks, have become increasingly important in load forecasting. Base station load is not only directly related to network stability and service quality, but also affects the operator's operating costs and user experience. During the operation of communication equipment networks, operators hope to predict load conditions so that they can take appropriate measures in advance to intervene and ensure user perception of mobile networks. The current network management system can count load indicators, and the load of base stations can be seen in the form of reports.

[0003] However, the existing base station load prediction methods have the following disadvantages:

[0004] (1) The existing technology consumes a lot of manpower and time costs. The existing technology requires first downloading reports from the network management side, and then the network optimization personnel screen out the high-load base stations. This process takes a long time and requires real-time monitoring by the network optimization personnel, which consumes manpower.

[0005] (2) Existing technologies cannot proactively detect problems in a timely manner and cannot predict the load of base stations. The network management side obtains the traffic that has already been generated on the base station, which means that the base station is already highly loaded and has affected user perception. Only then can network optimization personnel detect problems and take corresponding measures.

[0006] (3) It is impossible to directly obtain the load change trend of the base station. After the network optimization personnel extract the report, if they want to know the load change trend of the base station, they need to manually calculate the traffic flow in the corresponding time period. If they want to know the change trend of the trend, they need to manually draw the diagram, which is more complicated. Summary of the invention

[0007] The technical problem to be solved by the present invention is to provide a base station load prediction method, device and readable storage medium in response to the above-mentioned shortcomings of the prior art, so as to solve the problem that the existing base station load prediction method requires real-time monitoring by network optimization personnel, which consumes manpower and cannot proactively discover problems in a timely manner and cannot directly obtain the base station load change trend.

[0008] In a first aspect, the present invention provides a base station load prediction method, the method comprising:

[0009] Generate corresponding load index data according to the traffic, utilization rate and number of users collected by the base station every preset time period;

[0010] Inputting the generated load index data into the trained neural network model in sequence, so that the trained neural network model fits the input load index data into a curve, and predicts the load type of the base station according to the curve;

[0011] Take optimization measures to intervene in advance based on the predicted base station load type.

[0012] Furthermore, before generating corresponding load index data according to the flow, utilization rate and number of users collected by the base station every preset time period, the method further includes:

[0013] Collecting historical load data sets, wherein the historical load data sets include flow, utilization, and number of users in different time periods;

[0014] Obtaining a curve of a certain number of load indicators changing over time according to the historical load data set;

[0015] The neural network model is trained using the curves of the certain number of load indicators changing with time and the corresponding load type labels to obtain the trained neural network model.

[0016] Furthermore, before obtaining a certain number of curves of load indexes changing over time according to the historical load data set, the method further includes:

[0017] Abnormal data with a flow rate of 0, a user number of 0, or a utilization rate greater than 99.99% are eliminated from the historical load data set.

[0018] Furthermore, the load types include general high load, high load and super high load.

[0019] Furthermore, the generating of corresponding load index data according to the flow, utilization and number of users collected by the base station every preset time period specifically includes:

[0020] Normalize the traffic, utilization and number of users collected by the base station at every preset time period;

[0021] The corresponding load index data is generated based on the normalized traffic, utilization and number of users.

[0022] Furthermore, the load index data is generated according to the following formula:

[0023] L=D flow *W1+D prb *W2+D rrc *W3

[0024] In the formula, L represents the load index, D f1ow represents the normalized flow rate, D prbrepresents the utilization rate after normalization, D rrc represents the number of users after normalization, W1 represents the weight coefficient corresponding to the traffic, W2 represents the weight coefficient corresponding to the utilization, and W3 represents the weight coefficient corresponding to the number of users, where the value range of W1 is [0.5, 0.6].

[0025] Furthermore, the taking of optimization measures for early intervention according to the predicted base station load type specifically includes:

[0026] If the predicted base station load type is an overload, optimization measures are taken to intervene in advance;

[0027] The optimization means include at least one of the following: reducing base station power, adjusting base station parameters, and increasing base station resource capacity.

[0028] In a second aspect, the present invention provides a base station load prediction device, the device comprising:

[0029] A load index generation module, used to generate corresponding load index data according to the flow, utilization rate and number of users collected by the base station every preset time period;

[0030] A load type prediction module, connected to the load index generation module, for sequentially inputting the generated load index data into the trained neural network model, so that the trained neural network model fits the input load index data into a curve, and predicts the load type of the base station according to the curve;

[0031] The early intervention module is connected to the load type prediction module and is used to take optimization measures to perform early intervention according to the predicted base station load type.

[0032] In a third aspect, the present invention provides a base station load prediction device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to implement the base station load prediction method described in the first aspect above.

[0033] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the base station load prediction method described in the first aspect is implemented.

[0034] The present invention provides a base station load prediction method, device and readable storage medium. First, the corresponding load index data is generated according to the traffic, utilization rate and number of users collected by the base station at preset time periods; then the generated load index data is sequentially input into the trained neural network model, so that the trained neural network model fits the input load index data into a curve, and predicts the load type of the base station according to the curve; then, optimization measures are taken according to the predicted base station load type to intervene in advance. The present invention does not require too much manual intervention, and automatically extracts features from the data through a neural network model, thus avoiding tedious manual feature engineering and saving manpower and time costs.

[0035] At the same time, the present invention can predict the load trend of the base station and discover problems in a timely and proactive manner, solving the problem that the existing base station load prediction method requires real-time monitoring by network optimization personnel, which is manpower-consuming, and cannot discover problems in a timely and proactive manner and cannot directly obtain the base station load change trend. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of a base station load prediction method according to Embodiment 1 of the present invention;

[0037] Figure 2 It is a load curve diagram of different scenario types in Example 1 of the present invention;

[0038] Figure 3 This is a flowchart of another base station load prediction method according to Embodiment 1 of the present invention;

[0039] Figure 4 This is a schematic diagram of the structure of a neural network model according to Embodiment 1 of the present invention;

[0040] Figure 5 A schematic diagram of the process of training a neural network model according to Embodiment 1 of the present invention;

[0041] Figure 6 This is a schematic diagram of the structure of a base station load prediction device according to Embodiment 2 of the present invention;

[0042] Figure 7 This is a structural diagram of a base station load prediction device according to Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0043] In order to enable those skilled in the art to better understand the technical solution of the present invention, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0044] It should be understood that the specific embodiments and drawings described herein are only used to explain the present invention rather than to limit the present invention.

[0045] It can be understood that, in the absence of conflict, the various embodiments of the present invention and the various features in the embodiments can be combined with each other.

[0046] It can be understood that, for the convenience of description, the drawings of the present invention only show the parts related to the present invention, while the parts irrelevant to the present invention are not shown in the drawings.

[0047] It can be understood that each unit and module involved in the embodiments of the present invention may correspond to only one physical structure, or may be composed of multiple physical structures, or multiple units and modules may be integrated into one physical structure.

[0048] It can be understood that the terms "first", "second", etc. in the embodiments of the present invention are used to distinguish different objects, or to distinguish different processing of the same object, rather than to describe a specific order of objects.

[0049] It can be understood that, without conflict, the functions and steps marked in the flowcharts and block diagrams of the present invention may occur in an order different from that marked in the drawings.

[0050] It is understood that the flowcharts and block diagrams of the present invention illustrate the possible architectures, functions, and operations of the systems, devices, equipment, and methods according to the various embodiments of the present invention. Each box in the flowchart or block diagram may represent a unit, module, program segment, or code, which contains executable instructions for implementing the specified functions. Moreover, each box or combination of boxes in the block diagram and flowchart may be implemented by a hardware-based system that implements the specified functions, or may be implemented by a combination of hardware and computer instructions.

[0051] It can be understood that the units and modules involved in the embodiments of the present invention can be implemented by software or hardware. For example, the units and modules can be located in a processor.

[0052] Embodiment 1:

[0053] This embodiment provides a base station load prediction method, such as Figure 1 As shown, the method includes:

[0054] Step S101: Generate corresponding load index data according to the flow, utilization rate and number of users collected by the base station every preset time period.

[0055] In this embodiment, since 15 minutes is the minimum granularity of the load indicator that can be obtained by the base station network management platform, the preset time period is preferably 15 minutes, the traffic specifically refers to the downlink traffic, the utilization rate specifically refers to the downlink PRB (Physical Resource Block) utilization rate, and the number of users specifically refers to the average number of users in the RRC (Radio Resource Control) connection state.

[0056] Optionally, before generating corresponding load index data according to the flow, utilization rate and number of users collected by the base station every preset time period, the method further includes:

[0057] Collecting historical load data sets, wherein the historical load data sets include flow, utilization, and number of users in different time periods;

[0058] Obtaining a curve of a certain number of load indicators changing over time according to the historical load data set;

[0059] The neural network model is trained using the curves of the certain number of load indicators changing with time and the corresponding load type labels to obtain the trained neural network model.

[0060] In this embodiment, the neural network model can be any existing neural network model, such as a BP neural network model. A certain number of load index time-varying curves are divided into 20-minute equal lengths, and the corresponding output labels form a neural network data set. The training set, validation set, and test set are divided in a ratio of 6:2:2, and the neural network is trained to obtain the weight parameters of the neural network model. Assume that the load curves of different scenario types are as follows Figure 2 As shown, when the input curve is ①, the output is a general high-load cell, when the input curve is ②, the output is a high-load cell, and when the input curve is ③, the output is an ultra-high-load cell.

[0061] Specifically, in the process of training the neural network model, the base station load data set is first input into the neural network model in sequence to determine whether the loss function of the neural network model is less than a preset value. When the loss function is less than the preset value, the current neural network model is used as the trained neural network model. When the loss function is not less than the preset value, the weight parameters of the neural network model are adjusted using the gradient descent method until the loss function of the neural network model after adjusting the weight parameters is less than the preset value, thereby obtaining a trained neural network model.

[0062] Optionally, before obtaining a certain number of curves of load index changes over time according to the historical load data set, the method further includes:

[0063] Abnormal data with a flow rate of 0, a user number of 0, or a utilization rate greater than 99.99% are eliminated from the historical load data set.

[0064] In this embodiment, before model training, the abnormal data can be improved and supplemented. The flow and number of users should be greater than 0. If they are 0, it means that the base station is faulty and can be discarded. The maximum utilization rate is 99.99%, and the abnormal data greater than 99.99% is removed. After that, sample normalization processing can be implemented to normalize the flow, number of users and utilization rate.

[0065] Optionally, the load types include normal high load, high load and super high load.

[0066] In this embodiment, the base station load types mainly include: general high load, high load, super high load, and may also include non-high load.

[0067] Optionally, generating corresponding load index data according to the flow, utilization rate and number of users collected by the base station every preset time period specifically includes:

[0068] Normalize the traffic, utilization and number of users collected by the base station at every preset time period;

[0069] The corresponding load index data is generated based on the normalized traffic, utilization and number of users.

[0070] Specifically, the load index data is generated according to the following formula:

[0071] L=D flow *W1+D prb *W2+D rrc *W3

[0072] In the formula, L represents the load index, D flow represents the normalized flow rate, D prb represents the utilization rate after normalization, D rrc represents the number of users after normalization, W1 represents the weight coefficient corresponding to the traffic, W2 represents the weight coefficient corresponding to the utilization, and W3 represents the weight coefficient corresponding to the number of users. The value range of W1 is [0.5, 0.6], and the sum of the weight coefficients is 1.

[0073] In an optional embodiment, after the data processing is completed, different weights are assigned to the three values: traffic weight 0.5, number of users and utilization weights are 0.25 respectively. The weights are based on the high load judgment conditions of the base station: downlink traffic > 8.4GB, average number of users (number) in RRC connection state > 160 or downlink PRB utilization > 80%, that is, the downlink traffic condition must be met, and the number of users and utilization conditions can be judged as high load if one of them is met. The load index can be defined by the following formula: load index = traffic * 0.5 + utilization * 0.25 + number of users * 0.25, where the traffic, utilization, and number of users are all normalized.

[0074] It should be noted that traffic is the most direct indicator for measuring base station load. It not only reflects the actual data transmission needs, but also directly affects the use of network resources. For example, high traffic may lead to increased network latency and decreased connection quality. Therefore, when predicting base station load, the weight of traffic is set larger to more accurately reflect the current load situation. The traffic weight range can be [0.5, 0.6], preferably 0.5, and the user number weight = utilization weight = (1-traffic weight) / 2.

[0075] Step S102: input the generated load index data into the trained neural network model in sequence, so that the trained neural network model fits the input load index data into a curve and predicts the load type of the base station according to the curve.

[0076] In this embodiment, taking the preset time period of 15 minutes as an example, the base station collects traffic, utilization, and number of users every 15 minutes to generate corresponding load index data, and inputs the load index collected within this period into the trained neural network for processing every 20 minutes to obtain the load prediction result. Figure 2 As shown in , the neural network model can predict that the cell will be an ultra-high load cell.

[0077] It should be noted that some existing base station load prediction methods directly use user traffic, number of users and utilization rate to train neural network models and predict base station load. In practice, if only one parameter exceeds the limit, it is impossible to determine whether the base station is overloaded. It is necessary to comprehensively consider the three parameters to judge the base station load situation. Therefore, the present invention uses the load indicators obtained according to these three parameters for prediction, which can make the prediction results more comprehensive and accurate.

[0078] Step S103: taking optimization measures to intervene in advance according to the predicted base station load type.

[0079] In this embodiment, according to the network optimization rules, corresponding optimization measures are taken to intervene in the base stations of different predicted load types to avoid excessive load and affect user perception.

[0080] Optionally, taking optimization measures to intervene in advance according to the predicted base station load type specifically includes:

[0081] If the predicted base station load type is an overload, optimization measures are taken to intervene in advance;

[0082] The optimization means include at least one of the following: reducing base station power, adjusting base station parameters, and increasing base station resource capacity.

[0083] In this embodiment, when the fitted curve is Figure 2 When the new load curve is shown, the neural network model predicts that the cell is overloaded. At this time, the network optimization personnel can take optimization measures at time point A to intervene in advance to avoid the situation of overload in subsequent base stations such as curve ③. Among them, the optimization measures include: (1) reducing the power of the base station to reduce the coverage area and thus reduce the number of access users; (2) adjusting the base station parameters to divert traffic to surrounding base stations; (3) increasing the capacity of base station resources, etc.

[0084] It should be noted that the base station load prediction method provided by the present invention performs base station load prediction based on a neural network model. Since the neural network does not require too much manual intervention, it can self-learn and train, and automatically extract features from the data, thereby avoiding cumbersome manual feature engineering and saving labor costs and time costs. At the same time, the neural network can learn some rules from the training data and apply them to new data. It has strong generalization performance and can solve the problem that the existing technology cannot proactively discover problems in time and cannot predict the base station load. The present invention is based on the BP neural network. As long as the input and the corresponding output are defined, the features can be automatically extracted without manual parameter adjustment, and it has strong adaptability.

[0085] In a specific embodiment, Figure 3 As shown, the base station load prediction method may include the following steps:

[0086] S1. Obtain the previous base station load data sets during idle and busy hours.

[0087] Among them, the default off-peak hours are generally from 0:00 to 7:00, and the rest of the time is busy hours.

[0088] S2. Build a neural network model.

[0089] User traffic, number of users, and utilization rate in different time periods are used as inputs to the neural network.

[0090] The load type is used as the output to train the BP neural network.

[0091] The load types that affect the scenarios mainly include: general high load, high load, ultra-high load, and may also include non-high load.

[0092] Specifically, the structure of the neural network model is as follows: Figure 4 As shown, the input layer in the neural network model is used to input the corresponding data of the previous base station time dimension, such as traffic, number of users, and utilization, which are transmitted to the hidden layer through the activation function. The hidden layer in this embodiment can be composed of multiple fully connected layers, each of which can use different activation functions. A fully connected layer is connected behind the hidden layer, and the fully connected layer is connected to the output layer through an activation function to output the base station load type (general high load, high load, ultra-high load).

[0093] S3. Use the previous load data set to train the neural network model to obtain a trained neural network model.

[0094] In the process of training the neural network model, the base station load data set is first input into the neural network model in sequence to determine whether the loss function of the neural network model is less than a preset value. If the loss function is less than the preset value, the current neural network model is used as the trained neural network model. If the loss function is not less than the preset value, the weight parameters of the neural network model are adjusted using the gradient descent method until the loss function of the neural network model after the weight parameters are adjusted is less than the preset value, thereby obtaining a trained neural network model. Exemplarily, Figure 5 The figure shows the training flow chart of the neural network model. The data sets marked with the scenario load types are sequentially input into the neural network model for training until a stable and high verification accuracy is achieved and the corresponding loss value meets the preset conditions. At this time, a trained neural network model is obtained.

[0095] Specifically, before inputting data, historical data can be preprocessed at two levels. First, abnormal data should be improved and supplemented. The flow and number of users should be greater than 0. If the flow is 0, it means that the base station is faulty and can be discarded. The maximum utilization rate is 99.99%. Abnormal data greater than 99.99% can be discarded. Second, sample normalization is performed. The flow, number of users and utilization need to be normalized.

[0096] After the data is processed, different weights are assigned to the three values: traffic weight 0.5, user number and utilization weight 0.25 respectively. The weights are based on the high load judgment conditions of the base station: downlink traffic > 8.4GB, average number of users in RRC connection state (number) > 160 or downlink PRB utilization > 80%, that is, the downlink traffic condition must be met, and the user number and utilization condition can be judged as high load if one of them is met.

[0097] The load index can be defined by the following formula: load index = traffic*0.5+utilization*0.25+number of users*0.25, where the traffic, utilization, and number of users are all normalized results.

[0098] It should be noted that traffic is the most direct indicator for measuring base station load. It not only reflects the actual data transmission needs, but also directly affects the use of network resources. For example, high traffic may lead to increased network latency and decreased connection quality. Therefore, when predicting base station load, the weight of traffic is set larger to more accurately reflect the current load situation. The traffic weight range can be [0.5, 0.6], preferably 0.5, and the user number weight = utilization weight = (1-traffic weight) / 2.

[0099] A certain number of load index curves with time are divided into 20-minute curves of equal length, and the corresponding output labels are used to form a neural network data set. The training set, validation set, and test set are divided in a ratio of 6:2:2, and the neural network is trained to obtain the neural network model weight parameters. Assume that the load curves of different scenario types are as follows Figure 2 As shown in the figure, when the input curve is ①, the output is a general high-load cell, when the input curve is ②, the output is a high-load cell, and when the input curve is ③, the output is an ultra-high-load cell. After training, when new data is input, the neural network will output the corresponding result.

[0100] S4. Use the trained neural network model to predict the base station load type.

[0101] Judgment process: Fit the newly input data into a curve and predict the load type based on the trends of the new and old curves.

[0102] Specifically, the base station collects traffic, utilization, and number of users every 15 minutes (15 minutes is the minimum granularity of load indicators that the base station network management platform can obtain) to generate corresponding load indicator data, and inputs the load indicators collected within this period into the trained neural network for processing every 20 minutes to obtain the load forecast results. Figure 2 As shown in , the neural network model can predict that the cell will be an ultra-high load cell.

[0103] It should be noted that some existing base station load prediction methods directly use user traffic, number of users and utilization rate to train neural network models and predict base station load. In practice, if only one parameter exceeds the limit, it is impossible to determine whether the base station is overloaded. It is necessary to comprehensively consider the three parameters to judge the base station load situation. Therefore, the present invention uses the load indicators obtained according to these three parameters for prediction, which can make the prediction results more comprehensive and accurate.

[0104] S5. Output a corresponding optimization plan according to the base station load type.

[0105] According to the network optimization rules, the corresponding optimization measures are taken to intervene in the different load types determined by S4 to avoid excessive load and affect user perception. Figure 2 When the curve shown is judged as an ultra-high load cell, the network optimization personnel take optimization measures at time point A to intervene in advance to avoid the situation of ultra-high load such as curve ③ in subsequent base stations.

[0106] Among them, optimization measures include: (1) reducing base station power to reduce coverage and thus reduce the number of access users; (2) adjusting base station parameters to divert traffic to surrounding base stations; (3) increasing base station resource capacity, etc.

[0107] In another specific embodiment, the base station load prediction method may include the following steps:

[0108] 1. After normalizing the traffic, number of users, and utilization, use the formula to define the load index, i.e., load index = traffic*0.5+utilization*0.25+number of users*0.25;

[0109] 2. Divide a certain number of load index change curves over time into equal lengths of 20 minutes each, and form a neural network data set with the corresponding output labels. Divide them into training set, validation set, and test set in a ratio of 6:2:2, train the neural network, and obtain the neural network model weight parameters.

[0110] 3. The base station generates corresponding load index data every 15 minutes, and inputs the load index collected during this period into the trained neural network for processing every 20 minutes to realize the prediction of the input results.

[0111] 4. Take corresponding optimization measures to intervene in advance for different load scenarios determined by the neural network.

[0112] It should be noted that the base station load prediction method provided by the present invention has the following beneficial effects:

[0113] a) Improve prediction accuracy and optimize resource allocation: The existing base station load prediction method mainly downloads reports from the network management side first, and then the network optimization personnel screen out high-load base stations. The whole process takes a long time, requires manual monitoring, consumes manpower, and cannot proactively discover problems in time, and cannot predict the base station load situation. The present invention can automatically learn and identify complex patterns in historical data by simulating the nonlinear processing ability of human brain neurons, thereby realizing base station load prediction. This helps network optimization personnel to more accurately predict the base station load situation, perform resource scheduling and optimization in advance, and ensure the efficient and stable operation of the network.

[0114] b) Enhanced ability to cope with sudden traffic: During holidays, large-scale events or emergencies, the base station load often increases sharply, which puts great pressure on the network. The present invention can analyze network data in real time based on neural network prediction, respond to load changes in advance, and provide timely warning and decision support for network optimizers. This helps network optimizers to quickly adjust resource allocation, effectively cope with sudden traffic, and avoid large-scale user complaints caused by network congestion and service interruption.

[0115] c) Reduce operating costs and improve economic benefits: Through accurate base station load forecasting, operators can more reasonably arrange parameters such as transmission bandwidth and power of base stations, thereby reducing energy consumption and maintenance costs while ensuring service quality. At the same time, the present invention can also help network optimizers more accurately evaluate network expansion needs, avoid over-investment or under-investment, and further improve economic benefits.

[0116] In summary, the present invention can improve prediction accuracy, optimize resource allocation, enhance the ability to cope with sudden traffic, and can also reduce operating costs and improve economic benefits. It is of great significance for optimizing network resource allocation and improving user experience.

[0117] The base station load prediction method provided by the embodiment of the present invention first generates corresponding load index data based on the traffic, utilization rate and number of users collected by the base station every preset time period; then the generated load index data is sequentially input into the trained neural network model, so that the trained neural network model fits the input load index data into a curve, and predicts the load type of the base station according to the curve; then, optimization measures are taken to intervene in advance according to the predicted base station load type. The present invention does not require too much manual intervention, and automatically extracts features from the data through a neural network model, avoiding tedious manual feature engineering, saving manpower and time costs. At the same time, the present invention can predict the load trend of the base station and proactively discover problems in a timely and proactive manner, solving the problem that the existing base station load prediction method requires real-time monitoring by network optimization personnel, which consumes manpower, and cannot proactively discover problems in a timely and proactive manner, and cannot directly obtain the base station load change trend.

[0118] Embodiment 2:

[0119] like Figure 6 As shown, this embodiment provides a base station load prediction device, which is used to execute the above-mentioned base station load prediction method, including:

[0120] The load index generating module 11 is used to generate corresponding load index data according to the flow, utilization rate and number of users collected by the base station every preset time period;

[0121] A load type prediction module 12, connected to the load index generation module 11, is used to input the generated load index data into the trained neural network model in sequence, so that the trained neural network model fits the input load index data into a curve, and predicts the load type of the base station according to the curve;

[0122] The early intervention module 13 is connected to the load type prediction module 12 and is used to take optimization measures to perform early intervention according to the predicted base station load type.

[0123] Optionally, the device further comprises:

[0124] A historical data collection module is used to collect historical load data sets, wherein the historical load data sets include flow, utilization and number of users in different time periods;

[0125] A data set acquisition module, used to obtain a curve of a certain number of load indicators changing over time according to the historical load data set;

[0126] The model training module is used to train the neural network model using the curves of the certain number of load indicators changing with time and the corresponding load type labels to obtain the trained neural network model.

[0127] Optionally, the device further comprises:

[0128] The pre-processing module is used to remove abnormal data in the historical load data set, such as a flow rate of 0, a user number of 0, or a utilization rate greater than 99.99%.

[0129] Optionally, the load types include normal high load, high load and super high load.

[0130] Optionally, the load index generating module 11 specifically includes:

[0131] A normalization unit, used to normalize the traffic, utilization rate and number of users collected by the base station every preset time period;

[0132] The load index generating unit is used to generate corresponding load index data according to the normalized flow, utilization rate and number of users.

[0133] Optionally, the load index data is generated according to the following formula:

[0134] L=D flow *W1+D prb *W2+D rrc *W3

[0135] In the formula, L represents the load index, D f;ow represents the normalized flow rate, D prb represents the utilization rate after normalization, D rrc represents the number of users after normalization, W1 represents the weight coefficient corresponding to the traffic, W2 represents the weight coefficient corresponding to the utilization, and W3 represents the weight coefficient corresponding to the number of users, where the value range of W1 is [0.5, 0.6].

[0136] Optionally, the early intervention module 13 is specifically used for:

[0137] If the predicted base station load type is an overload, optimization measures are taken to intervene in advance;

[0138] The optimization means include at least one of the following: reducing base station power, adjusting base station parameters, and increasing base station resource capacity.

[0139] Embodiment 3:

[0140] refer to Figure 7 This embodiment provides a base station load prediction device, including a memory 21 and a processor 22, the memory 21 stores a computer program, and the processor 22 is configured to run the computer program to execute the base station load prediction method in Example 1.

[0141] The memory 21 is connected to the processor 22. The memory 21 may be a flash memory, a read-only memory or other memory. The processor 22 may be a central processing unit or a single-chip microcomputer.

[0142] Embodiment 4:

[0143] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the base station load prediction method in the above-mentioned embodiment 1 is implemented.

[0144] The computer-readable storage medium includes volatile or non-volatile, removable or non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, computer program modules or other data). Computer-readable storage media include, but are not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable read only memory), flash memory or other memory technology, CD-ROM (Compact Disc Read-Only Memory), digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer.

[0145] In summary, the base station load prediction method, device and readable storage medium provided by the embodiments of the present invention first generate corresponding load index data based on the traffic, utilization rate and number of users collected by the base station every preset time period; then the generated load index data are sequentially input into the trained neural network model, so that the trained neural network model fits the input load index data into a curve, and predicts the load type of the base station according to the curve; then, optimization measures are taken to intervene in advance according to the predicted base station load type. The present invention does not require too much manual intervention, and automatically extracts features from the data through a neural network model, avoiding tedious manual feature engineering, saving manpower and time costs. At the same time, the present invention can predict the load trend of the base station and proactively discover problems in a timely and proactive manner, solving the problem that the existing base station load prediction method requires real-time monitoring by network optimization personnel, which consumes manpower, and cannot proactively discover problems in a timely and proactive manner, and cannot directly obtain the base station load change trend.

[0146] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present invention, but the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A base station load prediction method, characterized in that: The method comprises: Generate corresponding load index data according to the traffic, utilization rate and number of users collected by the base station every preset time period; Inputting the generated load index data into the trained neural network model in sequence, so that the trained neural network model fits the input load index data into a curve, and predicts the load type of the base station according to the curve; Take optimization measures to intervene in advance based on the predicted base station load type.

2. The method according to claim 1, characterized in that Before generating corresponding load index data according to the flow, utilization rate and number of users collected by the base station every preset time period, the method further includes: Collecting historical load data sets, wherein the historical load data sets include flow, utilization, and number of users in different time periods; Obtaining a curve of a certain number of load indicators changing over time according to the historical load data set; The neural network model is trained using the curves of the certain number of load indicators changing with time and the corresponding load type labels to obtain the trained neural network model.

3. The method according to claim 2, characterized in that Before obtaining a certain number of curves of load indexes changing over time according to the historical load data set, the method further includes: Abnormal data with a flow rate of 0, a user number of 0, or a utilization rate greater than 99.99% are eliminated from the historical load data set.

4. The method according to claim 1, characterized in that: The load types include normal high load, high load and super high load.

5. The method according to claim 1, characterized in that The generating corresponding load index data according to the flow, utilization rate and number of users collected by the base station every preset time period specifically includes: Normalize the traffic, utilization and number of users collected by the base station at every preset time period; The corresponding load index data is generated based on the normalized traffic, utilization and number of users.

6. The method according to claim 5, characterized in that The load index data is generated according to the following formula: L=D flow *W1+D prb *W2+D rrc *W3 In the formula, L represents the load index, D flow represents the normalized flow rate, D prb represents the utilization rate after normalization, D rrc represents the number of users after normalization, W1 represents the weight coefficient corresponding to the traffic, W2 represents the weight coefficient corresponding to the utilization, and W3 represents the weight coefficient corresponding to the number of users, where the value range of W1 is [0.5, 0.6].

7. The method according to claim 4, characterized in that The optimization measures are taken according to the predicted base station load type to perform early intervention, specifically including: If the predicted base station load type is an overload, optimization measures are taken to intervene in advance; The optimization means include at least one of the following: reducing base station power, adjusting base station parameters, and increasing base station resource capacity.

8. A base station load prediction device, characterized in that: The device comprises: A load index generation module, used to generate corresponding load index data according to the flow, utilization rate and number of users collected by the base station every preset time period; A load type prediction module, connected to the load index generation module, for sequentially inputting the generated load index data into the trained neural network model, so that the trained neural network model fits the input load index data into a curve, and predicts the load type of the base station according to the curve; The early intervention module is connected to the load type prediction module and is used to take optimization measures to perform early intervention according to the predicted base station load type.

9. A base station load prediction device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to implement the base station load prediction method according to any one of claims 1 to 7.

10. 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 base station load prediction method according to any one of claims 1 to 7 is implemented.

Images