A method and device for optimizing communication effectiveness based on machine learning

Abstract

The application provides a communication effect optimization method and device based on machine learning, and relates to the technical field of communication.The method comprises the following steps: calculating the Pearson correlation coefficient of preset communication index data and preset communication quality data; screening the preset communication index data according to the Pearson correlation coefficient and setting weight data; obtaining historical communication quality data, obtaining historical index data corresponding to the historical communication quality data according to type information, and constructing a training set according to the weight data; training an initial prediction model by using the training set to obtain a model library; obtaining target communication quality data of a region to be optimized, environment information of the region to be optimized and actual business types of the region to be optimized, and selecting a target prediction model for the region to be optimized in the model library based on the environment information and the business types; and inputting the target communication quality data into the target prediction model for the region to be optimized to obtain prediction index data.

Description

Technical Field

[0001] This invention relates to the field of communication optimization, and in particular to a method and apparatus for optimizing communication performance based on machine learning. Background Technology

[0002] In real-time voice communication, poor communication metrics can cause audio stuttering, interruptions, and even one-way communication where the caller and receiver cannot hear each other, severely impacting call quality, affecting user experience, and leading to voice user complaints. Operators optimize wireless networks on a cell-by-cell basis. However, traditional methods often lack in-depth analysis of the relationship between communication metrics and communication quality, resulting in poor optimization effects. Furthermore, optimization strategies are often based on analysis and formulation by experienced technical personnel, leading to inefficiency. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a machine learning-based communication performance optimization method. This method solves the problems of poor optimization results and inefficient optimization strategy formulation caused by the lack of in-depth analysis of the relationship between communication metrics and communication quality in existing technologies.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following technical solution.

[0005] Firstly, this application provides a machine learning-based method for optimizing communication performance, comprising the following steps:

[0006] Under different scenario information, preset communication indicator data and preset communication quality data corresponding to the preset communication indicator data for different business types are obtained, and the Pearson correlation coefficient between the preset communication indicator data and the preset communication quality data is calculated under different scenario information for different business types.

[0007] Based on the Pearson correlation coefficient, preset communication indicator data for different business types under different scenario information are filtered to obtain target communication indicator data. The weight data of the target communication indicator data is determined according to the Pearson correlation coefficient and preset rules to obtain weight datasets for different business types under different scenario information. The type information of the target communication indicator data is recorded to obtain type information sets for different business types under different scenario information.

[0008] Obtain historical communication quality data for different business types under different scenario information, obtain historical indicator data corresponding to the historical communication quality data according to the type information set, and construct training sets for different business types under different scenario information according to the weighted dataset based on the historical communication quality data and the historical indicator data for different business types under different scenario information.

[0009] Using training sets for different business types under different scenario information, initial prediction models are trained separately to obtain a model library that includes target prediction models for different business types under different scenario information;

[0010] The system acquires target communication quality data of the area to be optimized, target environmental information of the area to be optimized, and actual service type of the area to be optimized. The target environmental information includes at least the location type and population density data of the area to be optimized. Based on the target environmental information and the actual service type, the system selects a target prediction model for the area to be optimized from the model library.

[0011] The target communication quality data is input into the target prediction model for the region to be optimized to obtain prediction index data, and an optimization strategy is generated based on the prediction index data to optimize the communication effect of the region to be optimized.

[0012] Optionally, the step of obtaining preset communication indicator data for different business types under different scenario information and preset communication quality data corresponding to the preset communication indicator data includes:

[0013] Obtain the location type of the region to which the preset communication indicator data belongs and the population density data of the region to which the preset communication indicator data belongs, and use the location type of the region to which the preset communication indicator data belongs and the population density data of the region to which the preset communication indicator data belongs as the scene information of the preset communication indicator data;

[0014] Obtain the service data of the region to which the preset communication indicator data belongs, and determine the service type of the region to which the preset communication indicator data belongs based on the service data;

[0015] Based on the scenario information and the business type, the preset communication indicator data is classified to obtain preset communication indicator data for different business types under different scenario information.

[0016] Based on the preset communication indicator data, obtain the preset communication quality data corresponding to the preset communication indicator data.

[0017] Optionally, the step of filtering preset communication indicator data for different business types under different scenario information based on the Pearson correlation coefficient to obtain target communication indicator data, and determining the weight data of the target communication indicator data according to the Pearson correlation coefficient and preset rules to obtain weighted datasets for different business types under different scenario information includes:

[0018] Under different scenario information and for different business types, preset communication indicator data with Pearson correlation coefficient greater than or equal to the threshold are used as target communication indicator data;

[0019] According to the preset rules, the target communication indicator data is assigned weights in descending order of the Pearson correlation coefficient, resulting in a weighted dataset of the target communication indicator data for different business types under different scenario information.

[0020] Optionally, the steps of obtaining historical communication quality data for different business types under different scenario information, obtaining historical indicator data corresponding to the historical communication quality data according to the type information set, and constructing training sets for different business types under different scenario information according to the weighted dataset based on the historical communication quality data and the historical indicator data for different business types under different scenario information, include:

[0021] Based on the scene information of the region to which the historical communication quality data belongs and the service type of the region to which the historical communication quality data belongs, select the target weight data corresponding to the historical communication quality data in the weight dataset;

[0022] Based on the scene information of the region to which the historical communication quality data belongs and the service type of the region to which the historical communication quality data belongs, the type information of the target communication indicator data corresponding to the historical communication quality data is searched in the type information set, and the historical indicator data corresponding to the historical communication quality data is obtained according to the type information of the target communication indicator data corresponding to the historical communication quality data.

[0023] The historical indicator data is organized according to the target weight data, and a training set is constructed based on the organized historical indicator data and the historical communication quality data.

[0024] Optionally, the step of obtaining target communication quality data of the area to be optimized, obtaining target environmental information and actual service type of the area to be optimized based on the tag data of the area to be optimized, wherein the target environmental information includes at least the location type and population density data of the area to be optimized, and selecting a target prediction model for the area to be optimized from the model library based on the target environmental information and the actual service type, includes:

[0025] The target environmental information of the region to be optimized is subjected to unsupervised clustering using the OPTICS clustering method to obtain the clustering results.

[0026] The clustering results are then subjected to supervised classification using the random forest classification method to obtain the classification results.

[0027] Based on the classification results, an initial prediction model for the region to be optimized is selected from the model library.

[0028] Based on the actual business type, a target prediction model for the region to be optimized is selected from the initial prediction model.

[0029] Optionally, the method further includes:

[0030] Acquire the predicted indicator data and the optimization data for the region to be optimized based on the target communication quality data by business personnel;

[0031] The predicted index data and the optimized data are matched to obtain similarity data between the predicted index data and the optimized data;

[0032] If the similarity data is less than a preset value, an updated dataset is constructed based on the optimized data and the target communication quality data, and the target prediction model is trained using the updated dataset to obtain an updated target prediction model. The updated target prediction model is used to receive the target communication quality data and obtain prediction index data.

[0033] Optionally, before the step of inputting the target communication quality data into a target prediction model for the region to be optimized to obtain prediction index data, the method further includes:

[0034] When the target environment information of the region to be optimized and the actual business type of the region to be optimized are obtained based on the label data of the region to be optimized, the real-time characteristics of the region to be optimized are obtained.

[0035] The real-time features are matched with the preset feature set, which includes preset features for different business types under different scenarios. The preset features that are consistent with the real-time features in the preset feature set are taken as target features, and the scenario information and business type corresponding to the target features are obtained.

[0036] The scene information corresponding to the target feature and the target environment information are matched. If the scene information corresponding to the target feature and the target environment information are inconsistent, a target prediction model for the region to be optimized is selected from the model library according to the scene information corresponding to the target feature and the business type corresponding to the target feature. The label data of the region to be optimized is updated based on the scene information corresponding to the target feature to obtain the updated label of the region to be optimized.

[0037] If the scene information corresponding to the target feature is consistent with the target environment information, then continue with the step of selecting a target prediction model for the region to be optimized from the model library based on the target environment information and the actual business type.

[0038] On the other hand, a communication performance optimization device based on machine learning is provided, the device comprising:

[0039] The feature extraction module is used to obtain preset communication indicator data and preset communication quality data corresponding to the preset communication indicator data for different business types under different scenario information, and to calculate the Pearson correlation coefficient between the preset communication indicator data and the preset communication quality data for different business types under different scenario information.

[0040] The feature filtering module is used to filter preset communication indicator data for different business types under different scenario information based on the Pearson correlation coefficient to obtain target communication indicator data, and determine the weight data of the target communication indicator data according to the Pearson correlation coefficient and preset rules to obtain weight datasets for different business types under different scenario information, and record the type information of the target communication indicator data to obtain type information sets for different business types under different scenario information.

[0041] The training set construction module is used to acquire historical communication quality data for different business types under different scenario information, acquire historical indicator data corresponding to the historical communication quality data according to the type information set, and construct training sets for different business types under different scenario information according to the weighted dataset based on the historical communication quality data and the historical indicator data for different business types under different scenario information.

[0042] The training module is used to train initial prediction models for different business types using training sets under different scenario information, resulting in a model library that includes target prediction models for different business types under different scenario information.

[0043] The model selection module is used to acquire target communication quality data of the area to be optimized, target environmental information of the area to be optimized, and actual service type of the area to be optimized. The target environmental information includes at least the location type and population density data of the area to be optimized. Based on the target environmental information and the actual service type, the module selects a target prediction model for the area to be optimized from the model library.

[0044] The prediction module is used to input the target communication quality data into the target prediction model for the region to be optimized, obtain prediction index data, and generate an optimization strategy based on the prediction index data to optimize the communication effect of the region to be optimized.

[0045] Thirdly, this application also provides an electronic device, including a memory, a processor, and a first computer program stored in the memory and executable on the processor, wherein the processor, when executing the first computer program, implements the communication effect optimization method based on machine learning as described above.

[0046] Fourthly, this application also provides a computer-readable storage medium storing a second computer program, which, when executed by a processor, implements the machine learning-based communication effect optimization method described above.

[0047] Beneficial effects:

[0048] This application obtains preset communication indicator data and corresponding preset communication quality data for different business types under different scenario information. It calculates the Pearson correlation coefficient between the preset communication indicator data and the preset communication quality data for different business types under different scenario information. Based on the Pearson correlation coefficient, it filters the preset communication indicator data for different business types under different scenario information to obtain target communication indicator data. It then determines the weight data of the target communication indicator data according to the Pearson correlation coefficient and preset rules, obtaining a weighted dataset for different business types under different scenario information. It records the type information of the target communication indicator data, obtaining a type information set for different business types under different scenario information. Finally, it obtains historical communication quality data for different business types under different scenario information, acquires historical indicator data corresponding to the historical communication quality data based on the type information set, and then... The following steps are taken: First, training sets for different service types are constructed based on historical communication quality data and historical indicator data for different service types, according to the weighted dataset. Second, initial prediction models are trained using these training sets for different service types under different scenario information, resulting in a model library containing target prediction models for different service types under different scenario information. Third, target communication quality data, target environment information, and actual service types of the area to be optimized are obtained. The target environment information includes at least the location type and population density data of the area to be optimized. Based on the target environment information and the actual service type, a target prediction model for the area to be optimized is selected from the model library. The target communication quality data is input into the target prediction model for the area to be optimized to obtain prediction indicator data. An optimization strategy is then generated based on the prediction indicator data to optimize the communication performance of the area to be optimized. The Pearson coefficient is used as a statistic to measure the correlation between preset communication indicator data and the preset communication quality data corresponding to the preset communication indicator data, thereby achieving more accurate wireless network optimization. Machine learning technology enables more accurate predictions and smarter optimization decisions, improving the efficiency of optimization strategy formulation. Through refined scenario segmentation and differentiated modeling, more accurate optimization strategies can be generated for different coverage scenarios. Attached Figure Description

[0049] Figure 1 This is a flowchart illustrating the operation of a machine learning-based communication performance optimization method according to an embodiment of the present invention.

[0050] Figure 2 This is a schematic diagram of the structure of a communication performance optimization system based on machine learning in an embodiment of the present invention;

[0051] Figure 3This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application;

[0052] Figure 4 This is a schematic diagram of the structure of a computer-readable storage medium provided in an embodiment of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0054] The embodiments described in this application are merely some, not all, embodiments of the present invention. Based on the spirit of the present invention, all other embodiments obtained by those skilled in the art without inventive effort are within the protection scope of the present invention.

[0055] See Figure 1 As shown, the present invention provides a communication performance optimization method based on machine learning, comprising the following steps:

[0056] S110. Obtain preset communication indicator data and preset communication quality data corresponding to the preset communication indicator data for different business types under different scenario information, and calculate the Pearson correlation coefficient between the preset communication indicator data and the preset communication quality data for different business types under different scenario information.

[0057] In one possible implementation, the step of acquiring preset communication indicator data for different service types under different scenario information and preset communication quality data corresponding to the preset communication indicator data includes:

[0058] Obtain the location type of the region to which the preset communication indicator data belongs and the population density data of the region to which the preset communication indicator data belongs, and use the location type of the region to which the preset communication indicator data belongs and the population density data of the region to which the preset communication indicator data belongs as the scene information of the preset communication indicator data;

[0059] Obtain the service data of the region to which the preset communication indicator data belongs, and determine the service type of the region to which the preset communication indicator data belongs based on the service data;

[0060] Based on the scenario information and the business type, the preset communication indicator data is classified to obtain preset communication indicator data for different business types under different scenario information.

[0061] Based on the preset communication indicator data, obtain the preset communication quality data corresponding to the preset communication indicator data.

[0062] For example, the different scenarios can be understood as environments with different locations and population densities, such as high-population-density urban environments. The preset communication index data are the basic wireless indicators when performing communication optimization. The basic wireless indicators include, but are not limited to, the proportion of weak uplink and downlink coverage in the cell, the CQI good rate, the uplink and downlink PRB utilization rate, uplink interference, ping-pong handover ratio, intra-system handover success rate, and inter-system handover success rate.

[0063] For example, dividing different areas by location and population density yields refined scenarios. By combining these scenarios with different business types, targeted processing can be applied to ensure data accuracy.

[0064] S120. Based on the Pearson correlation coefficient, filter the preset communication indicator data for different business types under different scenario information to obtain target communication indicator data, and determine the weight data of the target communication indicator data according to the Pearson correlation coefficient and preset rules to obtain the weight dataset for different business types under different scenario information. Record the type information of the target communication indicator data to obtain the type information set for different business types under different scenario information.

[0065] In one possible implementation, the step of filtering preset communication indicator data for different business types under different scenario information based on the Pearson correlation coefficient to obtain target communication indicator data, and determining the weight data of the target communication indicator data according to the Pearson correlation coefficient and preset rules to obtain weighted datasets for different business types under different scenario information, includes:

[0066] Under different scenario information and for different business types, preset communication indicator data with Pearson correlation coefficient greater than or equal to the threshold are used as target communication indicator data;

[0067] According to the preset rules, the target communication indicator data is assigned weights in descending order of the Pearson correlation coefficient, resulting in a weighted dataset of the target communication indicator data for different business types under different scenario information.

[0068] For example, by inputting communication quality data and communication indicator data of a total of 61,876 cells under the 5G network, 49,591 cells with an average daily call volume of more than 10 were selected to calculate the Pearson coefficient of the communication quality data and communication indicator data, and the optimization threshold of the communication quality data was obtained through the linear relationship. That is, when the proportion of communication quality data is 97%, the voice user experience is good. The optimization threshold of the communication indicator data is obtained with 97% as the target.

[0069] For example, in the communication quality data and communication indicator data, the Pearson coefficient for the downlink weak coverage ratio is -0.12, and the downlink weak coverage optimization threshold is 13.69%. The Pearson coefficient for the uplink weak coverage ratio is -0.10, and the uplink weak coverage optimization threshold is 16.80%. The Pearson coefficient for the CQI good / good rate is 0.11, and the uplink weak coverage optimization threshold is 91.68%. The Pearson coefficient for uplink interference is -0.07, and the uplink interference optimization threshold is -110. The Pearson coefficient for the ping-pong handover ratio is 0.06, and the ping-pong handover optimization threshold is 7.77%. The Pearson coefficient for the intra-system handover success rate is 0.05, and the intra-system handover success rate optimization threshold is 98.26%. The Pearson coefficient for inter-system handover success rate in communication quality data and communication indicator data is 0.04, and the optimization threshold for inter-system handover success rate is 92.78%. When the correlation threshold is set to 0.03, the Pearson coefficient for uplink PRB utilization in communication quality data and communication indicator data is 0.02, indicating a weak correlation. The Pearson coefficient for downlink PRB utilization in communication quality data and communication indicator data is 0.00, also indicating a weak correlation. Therefore, both uplink and downlink PRB utilization rates in the communication indicator data are removed, completing the filtering process.

[0070] For example, the target communication indicator data is sorted in descending order to obtain the following order: downlink weak coverage > CQI good rate > uplink weak coverage > uplink interference > ping-pong handover > intra-system handover success rate > inter-system handover success rate. Then, weights are set for the target communication indicator data according to the sorting of the target communication indicator data to obtain the weighted dataset of the target communication indicator data for different service types under different scenario information.

[0071] S130. Obtain historical communication quality data for different business types under different scenario information, obtain historical indicator data corresponding to the historical communication quality data according to the type information set, and construct training sets for different business types under different scenario information according to the weighted dataset based on the historical communication quality data for different business types under different scenario information and the historical indicator data.

[0072] In one possible implementation, the steps of acquiring historical communication quality data for different service types under different scenario information, acquiring historical indicator data corresponding to the historical communication quality data according to the type information set, and constructing training sets for different service types under different scenario information according to the weighted dataset based on the historical communication quality data and the historical indicator data for different service types under different scenario information, include:

[0073] Based on the scene information of the region to which the historical communication quality data belongs and the service type of the region to which the historical communication quality data belongs, select the target weight data corresponding to the historical communication quality data in the weight dataset;

[0074] Based on the scene information of the region to which the historical communication quality data belongs and the service type of the region to which the historical communication quality data belongs, the type information of the target communication indicator data corresponding to the historical communication quality data is searched in the type information set, and the historical indicator data corresponding to the historical communication quality data is obtained according to the type information of the target communication indicator data corresponding to the historical communication quality data.

[0075] The historical indicator data is organized according to the target weight data, and a training set is constructed based on the organized historical indicator data and the historical communication quality data.

[0076] For example, target weight data corresponding to the historical communication quality data is selected from the weight dataset. Taking urban and rural scenarios as examples, the corresponding target weight data are shown in Table 1 below:

[0077] Table 1: Target Weight Data Table for Urban and Rural Scenarios

[0078]

[0079] Specifically, after obtaining the actual value of the corresponding weak coverage ratio, when the actual value comes from the urban scene, the actual data needs to be multiplied by the target weight data in the urban scene, that is, the actual value is multiplied by 0.35 to obtain the adjusted weak coverage ratio. Then, a training set is constructed based on the adjusted weak coverage ratio. In this example, the training set constructed can be understood as the training set for the urban scene model.

[0080] S140. Using training sets for different business types under different scenario information, train initial prediction models respectively to obtain a model library including target prediction models for different business types under different scenario information;

[0081] For example, by dividing real-world scenarios and training models for different scenarios using data from those scenarios, targeted model construction can be carried out for various scenarios, avoiding the problem of inaccurate predictions caused by the inability of general models to adapt to scenario differences.

[0082] S150. Obtain target communication quality data of the area to be optimized, target environmental information of the area to be optimized, and actual service type of the area to be optimized. The target environmental information includes at least the location type of the area to be optimized and the population density data of the area to be optimized. Based on the target environmental information and the actual service type, select a target prediction model for the area to be optimized from the model library.

[0083] In one possible implementation, the steps of acquiring target communication quality data of the area to be optimized, acquiring target environmental information and actual service type of the area to be optimized based on the tag data of the area to be optimized, wherein the target environmental information includes at least the location type and population density data of the area to be optimized, and selecting a target prediction model for the area to be optimized from the model library based on the target environmental information and the actual service type, include:

[0084] The target environmental information of the region to be optimized is subjected to unsupervised clustering using the OPTICS clustering method to obtain the clustering results.

[0085] The clustering results are then subjected to supervised classification using the random forest classification method to obtain the classification results.

[0086] Based on the classification results, an initial prediction model for the region to be optimized is selected from the model library.

[0087] Based on the actual business type, a target prediction model for the region to be optimized is selected from the initial prediction model.

[0088] For example, the OPTICS clustering method is a density-based clustering algorithm that discovers the clustering structure of a dataset by building a reachability distance graph between objects. It requires no pre-setting of the number of clusters: Unlike algorithms such as K-means, which require pre-setting the number of clusters, OPTICS can automatically identify the clustering structure in the dataset. It is robust to noise: The OPTICS algorithm has a certain tolerance for noisy data and can filter out noisy points to some extent. It is suitable for discovering clusters of different densities: OPTICS can discover clusters of different densities in the dataset, not just uniformly dense clusters like DBSCAN. It provides more information: OPTICS not only provides clustering results but also provides more information about the clustering structure, such as core distance and reachability distance.

[0089] For example, the random forest classification method exhibits high prediction accuracy. By combining the results of multiple decision trees, the risk of overfitting from a single decision tree can be effectively reduced, improving the generalization ability of the overall model. Furthermore, due to the use of bootstrapping and feature selection methods, random forests can handle common problems such as high-dimensional data and missing values.

[0090] S160. Input the target communication quality data into the target prediction model for the region to be optimized to obtain prediction index data, and generate an optimization strategy based on the prediction index data to optimize the communication effect of the region to be optimized.

[0091] By acquiring preset communication indicator data and corresponding preset communication quality data for different business types under different scenario information, the Pearson correlation coefficient between the preset communication indicator data and the preset communication quality data is calculated for different business types under different scenario information. Based on the Pearson correlation coefficient, the preset communication indicator data for different business types under different scenario information is filtered to obtain target communication indicator data. The weight data of the target communication indicator data is determined according to the Pearson correlation coefficient and preset rules to obtain a weighted dataset for different business types under different scenario information. The type information of the target communication indicator data is recorded to obtain a type information set for different business types under different scenario information. Historical communication quality data for different business types under different scenario information is acquired. Based on the type information set, historical indicator data corresponding to the historical communication quality data is obtained, and based on different scenario information... For historical communication quality data and historical indicator data of different service types, training sets for different service types under different scenario information are constructed according to the weighted dataset. Initial prediction models are trained using the training sets for different service types under different scenario information, resulting in a model library including target prediction models for different service types under different scenario information. Target communication quality data, target environment information, and actual service types of the area to be optimized are obtained. The target environment information includes at least the location type and population density data of the area to be optimized. Based on the target environment information and the actual service type, a target prediction model for the area to be optimized is selected from the model library. The target communication quality data is input into the target prediction model for the area to be optimized to obtain prediction indicator data. An optimization strategy is generated based on the prediction indicator data to optimize the communication performance of the area to be optimized. The Pearson coefficient is used as a statistic to measure the correlation between preset communication indicator data and preset communication quality data corresponding to the preset communication indicator data, thereby achieving more accurate wireless network optimization. Machine learning technology enables more accurate predictions and smarter optimization decisions, improving the efficiency of optimization strategy formulation. Through refined scenario segmentation and differentiated modeling, more accurate optimization strategies can be generated for different coverage scenarios.

[0092] In one possible implementation, the method further includes:

[0093] Acquire the predicted indicator data and the optimization data for the region to be optimized based on the target communication quality data by business personnel;

[0094] The predicted index data and the optimized data are matched to obtain similarity data between the predicted index data and the optimized data;

[0095] If the similarity data is less than a preset value, an updated dataset is constructed based on the optimized data and the target communication quality data, and the target prediction model is trained using the updated dataset to obtain an updated target prediction model. The updated target prediction model is used to receive the target communication quality data and obtain prediction index data.

[0096] For example, to ensure the accuracy of the target prediction model, the predicted indicator data output by the target prediction model and the optimization data for the region to be optimized by business personnel based on the target communication quality data are matched. If the predicted indicator data and the optimization data differ too much, the target prediction model is updated by business personnel based on the optimization data for the region to be optimized according to the target communication quality data, so that the target prediction model outputs more accurate prediction results in the future.

[0097] In one possible implementation, prior to the step of inputting the target communication quality data into a target prediction model for the region to be optimized to obtain prediction index data, the method further includes:

[0098] When the target environment information of the region to be optimized and the actual business type of the region to be optimized are obtained based on the label data of the region to be optimized, the real-time characteristics of the region to be optimized are obtained.

[0099] The real-time features are matched with the preset feature set, which includes preset features for different business types under different scenarios. The preset features that are consistent with the real-time features in the preset feature set are taken as target features, and the scenario information and business type corresponding to the target features are obtained.

[0100] The scene information corresponding to the target feature and the target environment information are matched. If the scene information corresponding to the target feature and the target environment information are inconsistent, a target prediction model for the region to be optimized is selected from the model library according to the scene information corresponding to the target feature and the business type corresponding to the target feature. The label data of the region to be optimized is updated based on the scene information corresponding to the target feature to obtain the updated label of the region to be optimized.

[0101] If the scene information corresponding to the target feature is consistent with the target environment information, then continue with the step of selecting a target prediction model for the region to be optimized from the model library based on the target environment information and the actual business type.

[0102] For example, to conveniently and quickly obtain the target environment information of the area to be optimized, the historical label data of the area to be optimized is often used to obtain the target environment information of the area to be optimized. However, the target environment information of the area to be optimized is often not static. For example, in a university area, the population density data is low during holidays and high during school hours. Therefore, if the target environment information is obtained by relying on historical label data, errors may easily occur (selecting the wrong model). Therefore, obtaining the real-time characteristics of the area to be optimized can ensure that the selected target prediction model is the correct model and improve the prediction accuracy.

[0103] In one possible implementation, such as Figure 2 As shown, a communication performance optimization device based on machine learning is provided. The device includes:

[0104] Feature extraction module 201 is used to acquire preset communication indicator data and preset communication quality data corresponding to the preset communication indicator data for different business types under different scenario information, and to calculate the Pearson correlation coefficient between the preset communication indicator data and the preset communication quality data for different business types under different scenario information.

[0105] The feature filtering module 202 is used to filter preset communication indicator data for different business types under different scenario information based on the Pearson correlation coefficient to obtain target communication indicator data, and determine the weight data of the target communication indicator data according to the Pearson correlation coefficient and preset rules to obtain weight datasets for different business types under different scenario information, and record the type information of the target communication indicator data to obtain type information sets for different business types under different scenario information.

[0106] The training set construction module 203 is used to obtain historical communication quality data for different business types under different scenario information, obtain historical indicator data corresponding to the historical communication quality data according to the type information set, and construct training sets for different business types under different scenario information according to the weighted dataset based on the historical communication quality data and the historical indicator data for different business types under different scenario information.

[0107] Training module 204 is used to train initial prediction models for different business types using training sets under different scenario information, so as to obtain a model library including target prediction models for different business types under different scenario information;

[0108] The model selection module 205 is used to acquire target communication quality data of the area to be optimized, target environmental information of the area to be optimized, and actual service type of the area to be optimized. The target environmental information includes at least the location type and population density data of the area to be optimized. Based on the target environmental information and the actual service type, the module selects a target prediction model for the area to be optimized from the model library.

[0109] The prediction module 206 is used to input the target communication quality data into the target prediction model for the region to be optimized, obtain prediction index data, and generate an optimization strategy based on the prediction index data to optimize the communication effect of the region to be optimized.

[0110] In one possible implementation, such as Figure 3 As shown, this application embodiment provides a terminal device 300, including: a memory 310, a processor 320, and a first computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the first computer program 311, it implements a communication effect optimization method based on machine learning.

[0111] In one possible implementation, such as Figure 4 As shown, this application embodiment provides a computer-readable storage medium 400, on which a second computer program 411 is stored. When the second computer program 411 is executed by a processor, it implements the steps of a communication effect optimization method based on machine learning.

[0112] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0113] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0114] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0115] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0116] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0117] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0118] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0119] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

[0120] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for optimizing communication performance based on machine learning, characterized in that, Includes the following steps: Under different scenario information, preset communication indicator data and preset communication quality data corresponding to the preset communication indicator data for different business types are obtained, and the Pearson correlation coefficient between the preset communication indicator data and the preset communication quality data is calculated under different scenario information for different business types. The step of obtaining preset communication indicator data for different business types under different scenario information and preset communication quality data corresponding to the preset communication indicator data includes: Different regions are divided according to location and population density to obtain refined scenarios. Specifically, the location type and population density data of the region to which the preset communication indicator data belongs are obtained, and the location type and population density data of the region to which the preset communication indicator data belongs are used as the scenario information of the preset communication indicator data. Obtain the service data of the region to which the preset communication indicator data belongs, and determine the service type of the region to which the preset communication indicator data belongs based on the service data; Based on the scenario information and the business type, the preset communication indicator data is classified to obtain preset communication indicator data for different business types under different scenario information. Based on the preset communication index data, obtain the preset communication quality data corresponding to the preset communication index data; Based on the Pearson correlation coefficient, preset communication indicator data for different business types under different scenario information are filtered to obtain target communication indicator data. The weight data of the target communication indicator data is determined according to the Pearson correlation coefficient and preset rules to obtain weight datasets for different business types under different scenario information. The type information of the target communication indicator data is recorded to obtain type information sets for different business types under different scenario information. The step of filtering preset communication indicator data for different business types under different scenario information based on the Pearson correlation coefficient to obtain target communication indicator data, and determining the weight data of the target communication indicator data according to the Pearson correlation coefficient and preset rules to obtain weighted datasets for different business types under different scenario information includes: Under different scenario information and for different business types, preset communication indicator data with Pearson correlation coefficient greater than or equal to the threshold are used as target communication indicator data; According to the preset rules, the target communication indicator data is weighted in descending order of the Pearson correlation coefficient to obtain the weighted dataset of the target communication indicator data under different scenario information and for different business types. Obtain historical communication quality data for different business types under different scenario information, obtain historical indicator data corresponding to the historical communication quality data according to the type information set, and construct training sets for different business types under different scenario information according to the weighted dataset based on the historical communication quality data and the historical indicator data for different business types under different scenario information. Using training sets for different business types under different scenario information, initial prediction models are trained separately to obtain a model library that includes target prediction models for different business types under different scenario information; The system acquires target communication quality data of the area to be optimized, target environmental information of the area to be optimized, and actual service type of the area to be optimized. The target environmental information includes at least the location type and population density data of the area to be optimized. Based on the target environmental information and the actual service type, the system selects a target prediction model for the area to be optimized from the model library. The steps of obtaining target communication quality data for the area to be optimized, obtaining target environmental information and actual service type of the area to be optimized based on the tag data of the area to be optimized, wherein the target environmental information includes at least the location type and population density data of the area to be optimized, and selecting a target prediction model for the area to be optimized from the model library based on the target environmental information and the actual service type, include: The target environmental information of the region to be optimized is subjected to unsupervised clustering using the OPTICS clustering method to obtain the clustering results. The clustering results are then subjected to supervised classification using the random forest classification method to obtain the classification results. Based on the classification results, an initial prediction model for the region to be optimized is selected from the model library. Based on the actual business type, a target prediction model for the area to be optimized is selected from the initial prediction model. The target communication quality data is input into the target prediction model for the region to be optimized to obtain prediction index data, and an optimization strategy is generated based on the prediction index data to optimize the communication effect of the region to be optimized.

2. The communication performance optimization method based on machine learning according to claim 1, characterized in that, The steps of acquiring historical communication quality data for different business types under different scenario information, acquiring historical indicator data corresponding to the historical communication quality data according to the type information set, and constructing training sets for different business types under different scenario information according to the weighted dataset based on the historical communication quality data and the historical indicator data for different business types under different scenario information, include: Based on the scene information of the region to which the historical communication quality data belongs and the service type of the region to which the historical communication quality data belongs, select the target weight data corresponding to the historical communication quality data in the weight dataset; Based on the scene information of the region to which the historical communication quality data belongs and the service type of the region to which the historical communication quality data belongs, the type information of the target communication indicator data corresponding to the historical communication quality data is searched in the type information set, and the historical indicator data corresponding to the historical communication quality data is obtained according to the type information of the target communication indicator data corresponding to the historical communication quality data. The historical indicator data is organized according to the target weight data, and a training set is constructed based on the organized historical indicator data and the historical communication quality data.

3. The communication performance optimization method based on machine learning according to claim 1, characterized in that, The method further includes: Acquire the predicted indicator data and the optimization data for the region to be optimized based on the target communication quality data by business personnel; The predicted index data and the optimized data are matched to obtain similarity data between the predicted index data and the optimized data; If the similarity data is less than a preset value, an updated dataset is constructed based on the optimized data and the target communication quality data, and the target prediction model is trained using the updated dataset to obtain an updated target prediction model. The updated target prediction model is used to receive the target communication quality data and obtain prediction index data.

4. The communication performance optimization method based on machine learning according to claim 1, characterized in that, Before the step of inputting the target communication quality data into the target prediction model for the region to be optimized to obtain the prediction index data, the method further includes: When the target environment information of the region to be optimized and the actual business type of the region to be optimized are obtained based on the label data of the region to be optimized, the real-time characteristics of the region to be optimized are obtained. The real-time features are matched with a preset feature set, which includes preset features for different business types under different scenarios. The preset features that are consistent with the real-time features in the preset feature set are taken as target features, and the scenario information and business type corresponding to the target features are obtained. The scene information corresponding to the target feature and the target environment information are matched. If the scene information corresponding to the target feature and the target environment information are inconsistent, a target prediction model for the region to be optimized is selected from the model library according to the scene information corresponding to the target feature and the business type corresponding to the target feature. The label data of the region to be optimized is updated based on the scene information corresponding to the target feature to obtain the updated label of the region to be optimized. If the scene information corresponding to the target feature is consistent with the target environment information, then continue with the step of selecting a target prediction model for the region to be optimized from the model library based on the target environment information and the actual business type.

5. A communication performance optimization device based on machine learning, characterized in that, The device includes: The feature extraction module is used to acquire preset communication indicator data and corresponding preset communication quality data for different business types under different scenario information, and to calculate the Pearson correlation coefficient between the preset communication indicator data and the preset communication quality data for different business types under different scenario information. The step of acquiring preset communication indicator data and corresponding preset communication quality data for different business types under different scenario information includes: dividing different regions by location and population density to obtain refined scenario segments; specifically, acquiring the location type and population density data of the region to which the preset communication indicator data belongs, and using these data as the scenario information of the preset communication indicator data; acquiring business data of the region to which the preset communication indicator data belongs, and determining the business type based on the business data; classifying the preset communication indicator data based on the scenario information and the business type to obtain preset communication indicator data for different business types under different scenario information; and acquiring the corresponding preset communication quality data based on the preset communication indicator data. The feature filtering module is used to filter preset communication indicator data for different business types under different scenario information based on the Pearson correlation coefficient to obtain target communication indicator data, and to determine the weight data of the target communication indicator data according to the Pearson correlation coefficient and preset rules to obtain a weighted dataset for different business types under different scenario information. It also records the type information of the target communication indicator data to obtain a type information set for different business types under different scenario information. The step of filtering preset communication indicator data for different business types under different scenario information based on the Pearson correlation coefficient to obtain target communication indicator data, and determining the weight data of the target communication indicator data according to the Pearson correlation coefficient and preset rules to obtain a weighted dataset for different business types under different scenario information includes: using preset communication indicator data with a Pearson correlation coefficient greater than or equal to a threshold as target communication indicator data under different scenario information for different business types; and setting weights for the target communication indicator data according to the preset rules in descending order of the Pearson correlation coefficient to obtain a weighted dataset of the target communication indicator data for different business types under different scenario information. The training set construction module is used to acquire historical communication quality data for different business types under different scenario information, acquire historical indicator data corresponding to the historical communication quality data according to the type information set, and construct training sets for different business types under different scenario information according to the weighted dataset based on the historical communication quality data and the historical indicator data for different business types under different scenario information. The training module is used to train initial prediction models for different business types using training sets under different scenario information, resulting in a model library that includes target prediction models for different business types under different scenario information. The model selection module is used to acquire target communication quality data of the region to be optimized, target environmental information of the region to be optimized, and actual service type of the region to be optimized. The target environmental information includes at least the location type and population density data of the region to be optimized. Based on the target environmental information and the actual service type, the module selects a target prediction model for the region to be optimized from the model library. The steps of acquiring target communication quality data of the region to be optimized, acquiring target environmental information and actual service type of the region to be optimized based on the tag data of the region to be optimized, and selecting a target prediction model for the region to be optimized from the model library based on the target environmental information and the actual service type include: performing unsupervised clustering on the target environmental information of the region to be optimized using the OPTICS clustering method to obtain clustering results; performing supervised classification on the clustering results using the random forest classification method to obtain classification results; selecting an initial prediction model for the region to be optimized from the model library based on the classification results; and filtering out a target prediction model for the region to be optimized from the initial prediction models according to the actual service type. The prediction module is used to input the target communication quality data into the target prediction model for the region to be optimized, obtain prediction index data, and generate an optimization strategy based on the prediction index data to optimize the communication effect of the region to be optimized.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the machine learning-based communication performance optimization method as described in any one of claims 1 to 4.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the machine learning-based communication performance optimization method as described in any one of claims 1 to 4.

Images