Network quality prediction method and device, equipment and storage medium

Abstract

Disclosed are a network quality prediction method, apparatus and device, and a storage medium, which relate to the technical field of communications, are used for improving the accuracy of predicting the network quality of a communication link, and comprise: obtaining a plurality of basic parameters of a target network at the current moment, the plurality of basic parameters affecting the network quality of the target network; performing normalization processing on the plurality of basic parameters to obtain a target parameter, the normalization processing being used for unifying a dimension and a value range corresponding to each basic parameter in the plurality of basic parameters; and inputting the target parameter into a preset network quality prediction model, and determining a quality prediction result of the target network, the preset network quality prediction model being obtained by training based on a plurality of training data and an elastic network model, and the quality prediction result being indicated by at least one of the following items: throughput, time delay and packet loss rate. The method and the device are applied to a scene of predicting the network communication quality.

Description

Technical Field

[0001] This application relates to the field of communication technologies, and in particular, to a method, apparatus, device, and storage medium for predicting network quality. Background Art

[0002] With the continuous development of the fifth-generation mobile communication technology (5G), 5G networks are widely used in fields such as environmental monitoring, intelligent transportation, and intelligent furniture. However, during the process of transmitting data, 5G networks are interfered by noise in the environment, resulting in poor communication quality of the network, data loss, and further reduction of network performance due to the retransmission of more messages. Therefore, it is necessary to predict the network communication quality to select a communication link with higher quality for data transmission, ensure the normal transmission of data, reduce data loss and data retransmission, and improve the performance and throughput of the network.

[0003] Currently, determining the network quality of a communication link based on link characteristics depends on network devices, but network devices have calibration errors, and the prediction of the network quality of the communication link is not accurate enough; determining the network quality of a communication link based on probability estimation requires sending a large amount of probe packet data, resulting in additional communication data on the communication link of the network, which also affects the prediction of the network quality of the communication link; determining the network quality of a communication link based on machine learning is evaluated by one influencing factor and cannot comprehensively evaluate multiple influencing factors, which also affects the prediction of the network quality of the communication link. Therefore, the accuracy of predicting the network quality of a communication link is relatively low. Summary of the Invention

[0004] This application provides a method, apparatus, device, and storage medium for predicting network quality, which is used to improve the accuracy of predicting the network quality of a communication link.

[0005] To achieve the above object, this application adopts the following technical solutions:

[0006] In a first aspect, a method for predicting network quality is provided. The method includes: obtaining a plurality of basic parameters of a target network at the current moment, where the plurality of basic parameters affect the network quality of the target network; performing normalization processing on the plurality of basic parameters to obtain target parameters, and the normalization processing is used to unify the dimension and value range corresponding to each basic parameter in the plurality of basic parameters; inputting the target parameters into a preset network quality prediction model to determine the quality prediction result of the target network, where the preset network quality prediction model is trained based on a plurality of training data and an elastic net model, and the quality prediction result is indicated by at least one of the following: throughput, latency, and packet loss rate.

[0007] In a possible implementation, the method further includes: obtaining multiple groups of first parameters corresponding to the target network, where the multiple groups of first parameters are the network parameters of the target network at multiple historical moments, and one group of first parameters includes multiple basic parameters and multiple performance parameters corresponding to one historical moment; for any one group of the multiple groups of first parameters, normalizing the multiple basic parameters included in any one group of first parameters to obtain multiple second parameters corresponding to the historical moment of any one group of first parameters, where one second parameter is obtained by normalizing one basic parameter; determining at least one third parameter from the multiple second parameters as one group of third parameters, where the at least one third parameter includes second parameters with an influence coefficient greater than a first preset threshold, and the influence coefficient is used to indicate the correlation between the second parameter and the multiple performance parameters; training an elastic network model based on the multiple groups of third parameters determined from the multiple groups of first parameters to obtain a preset network quality prediction model.

[0008] In a possible implementation, determining at least one third parameter from the multiple second parameters as one group of third parameters includes: for any one second parameter among the multiple second parameters, determining the influence coefficient corresponding to any one second parameter based on a first preset algorithm, where the first preset algorithm is used to determine the correlation between any two parameters; determining at least one third parameter with an influence coefficient greater than the first preset threshold from the multiple second parameters as one group of third parameters.

[0009] In a possible implementation, training an elastic network model based on the multiple groups of third parameters determined from the multiple groups of first parameters to obtain a preset network quality prediction model includes: based on the multiple groups of third parameters determined from the multiple groups of first parameters, determining the comprehensive influence coefficient between any one third parameter in the at least one third parameter and any one performance parameter in the multiple performance parameters through a second preset algorithm, where the second preset algorithm is used to determine the correlation between any two parameters in the time dimension; determining a fourth parameter with a comprehensive influence coefficient less than a second preset threshold from the at least one third parameter and inputting the fourth parameter into the elastic network model for training to obtain a preset network quality prediction model.

[0010] In a possible implementation, for any one second parameter among the multiple second parameters, determining the influence coefficient corresponding to any one second parameter based on a first preset algorithm includes: for any one second parameter among the multiple second parameters, determining the correlation between any one second parameter and any one performance parameter in the multiple performance parameters based on a first preset algorithm; summing the correlations between any one second parameter and each performance parameter in the multiple performance parameters as the influence coefficient corresponding to any one second parameter.

[0011] In a possible implementation, the method further includes: training a preset verification model by using multiple second parameters and multiple performance parameters corresponding to each of multiple historical moments through a target network to obtain a trained verification model, where the trained verification model is used to verify the accuracy of a preset network quality prediction model; inputting target parameters corresponding to the target network at the current moment into the trained verification model to obtain a quality verification result; and determining the accuracy of the preset network quality prediction model based on the quality verification result and the quality prediction result.

[0012] In a second aspect, a network quality prediction device is provided. The network quality prediction device includes: an acquisition unit and a processing unit; the acquisition unit is configured to acquire multiple basic parameters of a target network at the current moment, where the multiple basic parameters affect the network quality of the target network; the processing unit is configured to perform normalization processing on the multiple basic parameters to obtain target parameters, and the normalization processing is used to unify the dimension and value range corresponding to each of the multiple basic parameters; the processing unit is further configured to input the target parameters into a preset network quality prediction model to determine a quality prediction result of the target network, where the preset network quality prediction model is trained based on multiple training data and an elastic network model, and the quality prediction result is indicated by at least one of the following: throughput, latency, and packet loss rate.

[0013] In a possible implementation, the acquisition unit is further configured to acquire multiple groups of first parameters corresponding to the target network, where the multiple groups of first parameters are network parameters corresponding to the target network at multiple historical moments, and one group of first parameters includes multiple basic parameters and multiple performance parameters corresponding to one historical moment; the processing unit is further configured to, for any one group of first parameters among the multiple groups of first parameters, perform normalization processing on the multiple basic parameters included in the any one group of first parameters to obtain multiple second parameters corresponding to the historical moment of the any one group of first parameters, where one second parameter is obtained by performing normalization processing on one basic parameter; the processing unit is further configured to determine at least one third parameter from the multiple second parameters as a group of third parameters, where the at least one third parameter includes a second parameter whose influence coefficient is greater than a first preset threshold, and the influence coefficient is used to indicate the correlation between the second parameter and the multiple performance parameters; the processing unit is further configured to train an elastic network model based on multiple groups of third parameters determined from the multiple groups of first parameters to obtain a preset network quality prediction model.

[0014] In a possible implementation, the processing unit is specifically configured to, for any one second parameter among the multiple second parameters, determine an influence coefficient corresponding to the any one second parameter based on a first preset algorithm, where the first preset algorithm is used to determine the correlation between any two parameters; the processing unit is specifically configured to determine at least one third parameter whose influence coefficient is greater than the first preset threshold from the multiple second parameters as a group of third parameters.

[0015] In a possible implementation, the processing unit is specifically configured to determine, based on multiple sets of third parameters determined by multiple sets of first parameters, a comprehensive influence coefficient between any one of at least one third parameter and any one of multiple performance parameters through a second preset algorithm, where the second preset algorithm is used to determine the correlation degree between any two parameters in the time dimension; the processing unit is specifically configured to determine a fourth parameter with a comprehensive influence coefficient less than a second preset threshold from at least one third parameter, and input the fourth parameter into an elastic net model for training to obtain a preset network quality prediction model.

[0016] In a possible implementation, the processing unit is specifically configured to, for any one of multiple second parameters, determine the correlation degree between any one of the second parameters and any one of multiple performance parameters based on a first preset algorithm; the processing unit is specifically configured to sum the correlation degrees between any one of the second parameters and each of the multiple performance parameters as the influence coefficient corresponding to any one of the second parameters.

[0017] In a possible implementation, the processing unit is further configured to train a preset verification model through multiple second parameters and multiple performance parameters corresponding to each historical moment among multiple historical moments of the target network to obtain a trained verification model, where the trained verification model is used to verify the accuracy of the preset network quality prediction model; the processing unit is further configured to input the target parameters corresponding to the target network at the current moment into the trained verification model to obtain a quality verification result; the processing unit is further configured to determine the accuracy of the preset network quality prediction model based on the quality verification result and the quality prediction result.

[0018] In a third aspect, an electronic device includes: a processor and a memory; wherein, the memory is used to store one or more programs, and the one or more programs include computer execution instructions. When the electronic device runs, the processor executes the computer execution instructions stored in the memory to enable the electronic device to execute a network quality prediction method as in the first aspect.

[0019] In a fourth aspect, a computer-readable storage medium storing one or more programs is provided, where the one or more programs include instructions that, when executed by a computer, cause the computer to execute a network quality prediction method as in the first aspect.

[0020] The present application provides a network quality prediction method, apparatus, device and storage medium, which are applied to the scenario of predicting network communication quality. First, a plurality of basic parameters affecting network quality in the target network at the current moment are obtained, and then the dimension and value range corresponding to each basic parameter among the plurality of basic parameters are uniformly processed to obtain normalized target parameters. Then, the target parameters are input into a preset network quality prediction model to obtain a quality prediction result of the target network including throughput, latency and packet loss rate, so as to indicate the quality prediction result through throughput, latency and packet loss rate. Through the above method, based on a preset network quality prediction model obtained by pre-training, the dimension and value range corresponding to each basic parameter are unified, which can improve the data processing efficiency of the model and the accuracy of predicting the network quality of the target network. Thus, the efficiency and accuracy of predicting the network quality of the communication link are effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 FIG. is a schematic structural diagram of a network quality prediction system provided by an embodiment of the present application;

[0022] Figure 2 FIG. is a schematic flow chart of a network quality prediction method provided by an embodiment of the present application Figure 1 ;

[0023] Figure 3 FIG. is a schematic flow chart of a network quality prediction method provided by an embodiment of the present application Figure 2 ;

[0024] Figure 4 FIG. is a schematic flow chart of a network quality prediction method provided by an embodiment of the present application Figure 3 ;

[0025] Figure 5 FIG. is a schematic flow chart of a network quality prediction method provided by an embodiment of the present application Figure 4 ;

[0026] Figure 6 FIG. shows the determination of network quality prediction model parameters provided by an embodiment of the present application Figure 1 ;

[0027] Figure 7 FIG. shows the determination of network quality prediction model parameters provided by an embodiment of the present application Figure 2 ;

[0028] Figure 8 FIG. is a schematic flow chart of a network quality prediction method provided by an embodiment of the present application Figure 5 ;

[0029] Figure 9 FIG. is a schematic flow chart of a network quality prediction method provided by an embodiment of the present applicationFigure 6 ;

[0030] Figure 10 A schematic structural diagram of a network quality prediction device provided for an embodiment of the present application;

[0031] Figure 11 A schematic structural diagram of an electronic device provided for an embodiment of the present application. Detailed implementation manners

[0032] Next, the technical solutions in the embodiments of the present application will be described with reference to the accompanying drawings in the embodiments of the present application.

[0033] In the description of the present application, unless otherwise specified, " / " means "or". For example, A / B may represent A or B. The "and / or" herein is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, "at least one" and "multiple" refer to two or more. The terms such as "first" and "second" do not limit the quantity and execution order, and the terms such as "first" and "second" do not necessarily limit to be different.

[0034] Currently, predicting the network quality of a network communication link and then selecting a network communication link with higher network quality to ensure high-quality data transmission has become the top priority of network communication.

[0035] However, using the method based on link characteristics for network quality prediction depends on network devices, but there are calibration errors in network devices, and the network quality prediction of communication links is not accurate enough.

[0036] Using the method based on probability estimation for network quality prediction requires sending a large number of probe packets, which not only increases the additional communication overhead, but also the probe packets are not sensitive enough to the network link conditions.

[0037] Using the link quality prediction method based on machine learning for network quality prediction is to transform the network link quality prediction problem into a time series prediction problem, use machine learning or deep learning methods to mine the features of training samples, deeply learn the potential relationship between the model input and output, and use the obtained model to predict the network link quality at the next moment. Although the method of using machine learning can effectively mine the features of training samples and has higher accuracy and stability, currently the algorithms based on machine learning can only output the quality prediction level of a single influencing factor, and cannot associate multiple influencing factors of communication quality, that is, the input and output data are both quality data or only use a classification model to classify quality data, and cannot comprehensively consider multiple influencing factors and cannot comprehensively examine communication quality.

[0038] A network quality prediction method provided by an embodiment of the present application can be applicable to a network quality prediction system. Figure 1 A schematic structural diagram of the network quality prediction system is shown. As Figure 1 shown, the network quality prediction system 20 includes: an electronic device 21 and a network device 22.

[0039] Among them, the network device 22 can be a network device corresponding to an operator, such as a base station, a mobile base station, etc., which is used to provide network services for users and can provide 5G networks, wireless sensor networks, etc. for users.

[0040] The electronic device 21 can obtain multiple basic parameters of the network, perform normalization processing on the multiple basic parameters, so as to obtain target parameters with unified dimensions and value ranges, and then input the target parameters into a preset network quality prediction model, and then the network quality of the network can be predicted to obtain the quality prediction result of the network.

[0041] Next, a network quality prediction method provided by an embodiment of the present application will be described with reference to the accompanying drawings. As Figure 2 shown, a network quality prediction method provided by an embodiment of the present application includes S201 - S203:

[0042] S201. Obtain multiple basic parameters of the target network at the current moment.

[0043] Among them, the multiple basic parameters affect the network quality of the target network.

[0044] Optionally, the multiple basic parameters may include: communication distance, transmission power, networking mode, traffic volume, load volume, mobile communication, whether it is a time-sensitive network, receiving sensitivity, etc.

[0045] It can be understood that by collecting multiple basic parameters that affect network quality, the network quality during network transmission can be predicted, so as to screen out network links with higher quality and improve the data transmission efficiency.

[0046] Specifically, the topological structure of the network in the target network, the network devices that make up the network, and the link information connecting the network can be collected, so as to obtain multiple basic parameters of the target network at the current moment.

[0047] S202. Perform normalization processing on the multiple basic parameters to obtain target parameters.

[0048] Among them, the normalization processing is used to unify the dimensions and value ranges corresponding to each basic parameter among the multiple basic parameters.

[0049] It should be noted that the dimension refers to the basic attribute of a physical quantity. Different physical quantities have different dimensions and use different units. Therefore, it is necessary to unify the units of multiple basic parameters to provide a standard for further processing of multiple basic parameters in the future.

[0050] Optionally, the International System of Units can be adopted to process multiple basic parameters. For example, length (meter), mass (kilogram), time (second), electric current (ampere), thermodynamic temperature (kelvin), amount of substance (mole), and luminous intensity (candela).

[0051] S203. Input the target parameter into a preset network quality prediction model to determine the quality prediction result of the target network.

[0052] Among them, the preset network quality prediction model is trained based on multiple training data and an elastic network model, and the quality prediction result is indicated by at least one of the following: throughput, delay, and packet loss rate.

[0053] Specifically, throughput refers to the amount of data effectively transmitted in a network channel per unit time. If the number of data frames successfully transmitted per unit time is M and the frame length of the data frame is L, the current throughput S is as shown in Formula 1 below:

[0054] S = M × L Formula 1

[0055] Specifically, the delay can be divided into propagation delay and transmission delay. Propagation delay refers to the time required for data to propagate in the channel medium and is mainly related to the distance; transmission delay refers to the time required to push all bits in the data packet onto the link. The size of the delay determines whether nodes in the network can receive information sent by other nodes in a timely manner.

[0056] Specifically, the packet loss rate refers to the ratio of the number of lost data packets to the total number of sent data packets during data transmission. The packet loss rate is related to the length of the transmitted data packets and the sending frequency of the data packets. The packet loss rate can be calculated by the following Formula 2:

[0057]

[0058] It can be understood that the throughput, delay, and packet loss rate of the target network can be predicted based on the target parameter and the preset network quality prediction model, so as to realize the prediction of the network quality of the target network.

[0059] In the embodiments of the present application, first, multiple basic parameters affecting the network quality in the target network at the current moment are obtained. Then, the dimensions and value ranges corresponding to each of the multiple basic parameters are uniformly processed to obtain normalized target parameters. Next, the target parameters are input into a preset network quality prediction model to obtain a quality prediction result of the target network including throughput, latency, and packet loss rate, so as to indicate the quality prediction result through throughput, latency, and packet loss rate. Through the above method, based on the preset network quality prediction model obtained by pre-training, the dimensions and value ranges corresponding to each basic parameter are unified, which can improve the model's data processing efficiency and the accuracy of predicting the network quality of the target network. Thus, the efficiency and accuracy of predicting the network quality of the communication link are effectively improved.

[0060] In one design, as Figure 3 shown, a network quality prediction method provided by an embodiment of the present application may specifically further include steps S301 - S304:

[0061] S301. Obtain multiple sets of first parameters corresponding to the target network.

[0062] Among them, the multiple sets of first parameters are network parameters corresponding to the target network at multiple historical moments. One set of first parameters includes multiple basic parameters and multiple performance parameters corresponding to one historical moment.

[0063] Specifically, in combination with the communication scenario of the target network, the network parameters of the target network at multiple historical moments can be collected, and a sample set can be constructed in chronological order, so as to obtain multiple sets of first parameters of the target network.

[0064] Exemplarily, eight basic parameters of the target network at multiple historical moments are obtained, including communication distance, transmission power, networking mode, traffic volume, load volume, mobile communication, whether it is a time-sensitive network, and receiving sensitivity, denoted as x i (i = 1, 2, ··· 8); and three performance parameters of the target network at multiple historical moments are obtained, including throughput, latency, and packet loss rate, denoted as y j (j = 1, 2, 3). And the eight basic parameters and three performance parameters included in one set of first parameters are stored in the data set {x i , y j}.

[0065] S302. For any one set of first parameters among the multiple sets of first parameters, normalize the multiple basic parameters included in the any one set of first parameters to obtain multiple second parameters corresponding to the historical moment of the any one set of first parameters.

[0066] Among them, one second parameter is obtained by normalizing one basic parameter.

[0067] Specifically, normalize the multiple basic parameters included in any group of first parameters to ensure the unity of the dimension and value range corresponding to each basic parameter among the multiple basic parameters included in any group of first parameters, thereby obtaining multiple second parameters at the historical moment corresponding to any group of first parameters.

[0068] Exemplarily, normalize the eight basic parameters x i (i = 1, 2, ··· 8) included in any group of first parameters obtained, and obtain multiple second parameters as x i ’ (i = 1, 2, ··· 8).

[0069] S303. Determine at least one third parameter from the multiple second parameters as a group of third parameters.

[0070] Among them, at least one third parameter includes a second parameter whose influence coefficient is greater than a first preset threshold, and the influence coefficient is used to indicate the correlation between the second parameter and multiple performance parameters.

[0071] Specifically, through a first preset algorithm, first determine the correlation between each second parameter among the multiple second parameters and each performance parameter among the multiple performance parameters, and then determine the influence coefficient corresponding to each second parameter. Further, from the multiple second parameters, screen out at least one third parameter whose influence coefficient is greater than the first preset threshold and use it as a group of third parameters.

[0072] It can be understood that by screening the multiple second parameters, at least one third parameter whose influence coefficient between the performance parameter and the network quality is greater than the first preset threshold is determined, and then the network is predicted through the at least one third parameter, thereby improving the prediction accuracy of the network quality.

[0073] S304. Train an elastic network model based on multiple groups of third parameters determined from multiple groups of first parameters to obtain a preset network quality prediction model.

[0074] Specifically, through a second preset algorithm, based on multiple groups of third parameters determined from multiple groups of first parameters, determine the comprehensive influence coefficient between any third parameter among at least one third parameter and any performance parameter among the multiple performance parameters, and screen out a fourth parameter less than a second preset threshold from the multiple third parameters. Then, input the fourth parameter into the elastic network model for training to obtain a preset network quality prediction model.

[0075] It can be understood that by screening among at least one third parameter, a fourth parameter with a relatively high correlation between the third parameter and the performance parameter in the time dimension is further determined, thereby improving the accuracy of the prediction result of the network quality.

[0076] In the embodiments of the present application, by screening multiple basic parameters included in any one of multiple groups of first parameters of the target network obtained, parameters with a relatively high correlation with the performance parameters are determined. Further, through training with an elastic net model, a preset network quality prediction model is obtained. Furthermore, the network quality can be predicted through multiple basic parameters of the target network, and the accuracy of the prediction result can be ensured.

[0077] In one design, as Figure 4 shown, a network quality prediction method provided by an embodiment of the present application, the method in step S303 above may specifically include steps S401 - S402:

[0078] S401. For any one of multiple second parameters, based on a first preset algorithm, determine the influence coefficient corresponding to any one of the second parameters.

[0079] Wherein, the first preset algorithm is used to determine the correlation degree between any two parameters.

[0080] Optionally, the first preset algorithm may be a Pearson correlation coefficient algorithm, or a Granger causality algorithm, or other algorithms that can determine the correlation degree between any two parameters.

[0081] Specifically, according to the first preset algorithm, determine the correlation degree between any one of multiple second parameters and each performance parameter among multiple performance parameters. Then, add the correlation degrees between any one of the second parameters and multiple performance parameters to obtain the influence coefficient corresponding to any one of the second parameters.

[0082] Exemplarily, according to the Pearson correlation coefficient algorithm, based on eight second parameters x i ’ (i = 1, 2, ··· 8) and three performance parameters y j (j = 1, 2, 3), the correlation degree between any one of the second parameters and any one of the performance parameters can be determined, and then the influence coefficient corresponding to any one of the second parameters can be determined.

[0083]

[0084] S402. Determine at least one third parameter whose influence coefficient is greater than a first preset threshold from multiple second parameters as a group of third parameters.

[0085] Optionally, in addition to determining at least one third parameter when the influence coefficient is greater than the first preset threshold, the influence coefficients corresponding to multiple second parameters can also be sorted in descending order, and the second parameters with higher sorting can be selected as a group of third parameters.

[0086] It can be understood that when the influence coefficient is greater than the first preset threshold, it indicates that the correlation between the corresponding second parameter and the performance parameter is relatively large, that is, the second parameter has a greater influence on the performance parameter and also on the network quality. Therefore, selecting it can more accurately predict the target network and reduce the amount of calculation.

[0087] Exemplarily, the influence coefficients corresponding to multiple second parameters are sorted in descending order, and the second parameters corresponding to the top n influence coefficients are selected as the third parameter x i ’ (i = 1, 2, ··· n), where n is less than 8.

[0088] In the embodiments of the present application, through the first preset algorithm, the correlation between the second parameter and the performance parameter is analyzed and compared horizontally to determine at least one third parameter, ensuring the accuracy of the network quality prediction, reducing the amount of calculation of irrelevant factors, and improving the calculation efficiency of the preset network quality prediction model.

[0089] In one design, as Figure 5 shown, in a network quality prediction method provided by the embodiments of the present application, the method in step S304 above may specifically include steps S501 - S502:

[0090] S501. Based on multiple groups of third parameters determined by multiple groups of first parameters, through a second preset algorithm, determine the comprehensive influence coefficient between any one of the at least one third parameter and any one of the multiple performance parameters.

[0091] Among them, the second preset algorithm is used to determine the correlation degree between any two parameters in the time dimension.

[0092] It should be noted that since there is a correlation and coupling between multiple third parameters, only screening parameters horizontally will result in an unstable constructed model, thus affecting the prediction effect of the model. Therefore, it is also necessary to screen multiple parameters longitudinally in the time dimension to effectively extract parameters, reduce the data dimension, improve the model operation speed, and improve the model prediction accuracy.

[0093] Optionally, the second preset algorithm can use the Granger causality test. Statistically, the sum of squared residuals is usually used to represent the prediction error. Then a regression equation can be established, and the method of hypothesis testing (F - test) is used to test whether the coefficient is zero.

[0094] Exemplarily, based on the Granger causality test, a regression equation at time t can be determined according to multiple sets of third parameters determined from multiple sets of first parameters at multiple historical moments, and multiple performance parameters of the obtained first parameter, as shown in Formula 3:

[0095] y j,t = a 0 + a 1 y j,t-1 + ··· + a p y j,t-p + b 0 + b 1 x i,t-1 ’ + ··· + b p x i,t-p ’ + ε t Formula 3

[0096] Where y j,t is the performance parameter at time t, x i,t-1 ’ (i = 1, 2, ··· n) is the third parameter at time t - 1, a and b are constant coefficients, and b 1 to b p is the comprehensive influence coefficient between any third parameter in at least one set of third parameters and any performance parameter in multiple performance parameters, and ε t is the error term at time t.

[0097] It should be noted that there is a set of comprehensive influence coefficients between each performance parameter y j (j = 1, 2, 3) and each third parameter x i ’ (i = 1, 2, ··· n).

[0098] S502. Determine a fourth parameter with a comprehensive influence coefficient less than a second preset threshold from at least one third parameter, and input the fourth parameter into an elastic net model for training to obtain a preset network quality prediction model.

[0099] It should be noted that in the case where the comprehensive influence coefficient of any third parameter is less than the second preset threshold, it indicates that there is an obvious causal relationship between the third parameter and multiple performance parameters. Therefore, it can be determined as the fourth parameter and input into the elastic net model for further training to obtain a preset network quality prediction model.

[0100] Specifically, the elastic net model is a linear regression model similar to the Least Absolute Shrinkage and Selection Operator (Lasso) and ridge regression algorithms, which combines the characteristics of the Lasso regression algorithm and the ridge regression algorithm. The main difference between the Lasso regression algorithm and the ridge regression algorithm lies in the penalty terms. The Lasso regression algorithm uses the sum of squared coefficients as the penalty term, that is, the L1 regularization term, while the ridge regression algorithm uses the sum of absolute values of coefficients as the penalty term, that is, the L2 regularization term. The advantage of the ridge regression algorithm is to improve the prediction accuracy, but it cannot discard any features to make the regression coefficient zero, and it is relatively stable compared with the Lasso regression algorithm. The Lasso regression algorithm can shrink some related feature coefficients to zero. Therefore, it is beneficial for feature selection and greatly improves the interpretability of the model. The elastic net is a compromise between the two methods. It can automatically perform feature selection while continuously shrinking the coefficients. It can select groups of related features or select more features than the number of samples to saturate the features. It is suitable for models with multiple features that are related to each other, achieving better stability, higher accuracy, and better generalization and interpretability. The elastic net algorithm has been relatively maturely applied and has good effects in fields such as wind speed prediction, stock prediction, and load prediction.

[0101] Exemplarily, in combination with Formula 3, if for any third parameter x i ’ that composes each performance parameter y j (j = 1, 2, 3), the comprehensive influence coefficients are not significantly zero, then this third parameter x i ’ (i = 1, 2, ··· n) is the selected fourth parameter x i ’ (i = 1, 2, ··· m), where m is less than or equal to n.

[0102] Specifically, taking the selected fourth parameter as the input quantity and the performance parameter as the output quantity, and inputting them into the loss function of the elastic net model as shown in Formula 4 below:

[0103]

[0104] where α is the mixing parameter, γ is the complexity parameter, x i ′ is the independent variable (the fourth parameter), ‖β i ‖ 1 is the Lasso term, is the ridge regression term.

[0105] It should be noted that α and γ are two hyperparameters. The mixing parameter α (0 ≤ α ≤ 1) characterizes the degree of the Lasso regression algorithm and the ridge regression algorithm. For example, when α is 0, it is the ridge regression algorithm, and when α is 1, it is the Lasso regression algorithm; the complexity parameter γ characterizes the degree of compression (penalty).

[0106] Optionally, as Figure 6 shown, it is the fitting degree of the elastic net model for the mixing parameter α within the range of 0 to 1, so that the accurate value of the hyperparameter (mixing parameter) α in the elastic net model can be determined. Further, a preset network quality prediction model is obtained.

[0107] Optionally, as Figure 7 shown, it is the fitting degree of the elastic net model for the complexity parameter γ within the range of 0 to 4, so that the accurate value of the complexity parameter γ in the elastic net model can be determined. Further, a preset network quality prediction model is obtained.

[0108] In the embodiments of the present application, by screening multiple groups of third parameters in the time dimension, a fourth parameter causally related to the performance parameter is obtained and input into the elastic net model for training, so as to obtain the hyperparameters of the model, and further obtain a preset network quality prediction model.

[0109] In one design, as Figure 8 shown, in a network quality prediction method provided by the embodiments of the present application, the method in the above step S401 may specifically include steps S601 - S602:

[0110] S601. For any one of the multiple second parameters, based on the first preset algorithm, determine the correlation degree between any one of the second parameters and any one of the multiple performance parameters.

[0111] Exemplarily, based on the Pearson correlation coefficient algorithm, taking eight second parameters x i ’ (i = 1, 2, ··· 8) as variables and three performance parameters y j (j = 1, 2, 3) as target quantities, the correlation degree between any one of the second parameters and any one of the multiple performance parameters can be determined, that is, the correlation degree between the second parameter x i ’ and the three performance parameters y j is r ij . For example, the correlation degrees between the second parameter x 1 ’ and the three performance parameters y 1 , y 2 , y 3 are r 11 , r 12 , r 13。

[0112] S602. Sum the correlation degrees between any second parameter and each of the multiple performance parameters as the influence coefficient corresponding to the any second parameter.

[0113] Exemplarily, for any second parameter x i ’ and the correlation degrees r i1 、r i2 、r i3 between it and each of the multiple performance parameters, sum them to obtain the influence coefficient corresponding to the any second parameter.

[0114]

[0115] It can be understood that according to the obtained influence coefficient corresponding to any second parameter, compare it with the first preset threshold. When the influence coefficient of a certain second parameter is greater than the first preset threshold, determine this second parameter as the third parameter, and then determine multiple groups of third parameters to train the elastic net model.

[0116] In the embodiments of the present application, through the first preset algorithm, determine the correlation degree between any second parameter and any one of the multiple performance parameters, and then add the correlation degrees as the influence coefficient corresponding to the any second parameter, which can further discriminate the influence coefficient and screen out the third parameter from the second parameters.

[0117] In one design, as Figure 9 shown, a network quality prediction method provided by the embodiments of the present application may further specifically include steps S701 - S703:

[0118] S701. Train a preset verification model with the multiple second parameters and multiple performance parameters corresponding to each historical moment among multiple historical moments of the target network to obtain the trained verification model.

[0119] Wherein, the trained verification model is used to verify the accuracy of the preset network quality prediction model.

[0120] Optionally, the preset verification model may be a Back Propagation (BP) neural network model.

[0121] Exemplarily, a BP neural network model is used as a preset verification model. By learning multiple second parameters and multiple performance parameters corresponding to each historical moment among multiple historical moments, a mapping relationship between input and output patterns is obtained to train the model. This process does not involve learning the mathematical equations describing its mapping relationship, but rather, through the total network error and using the gradient descent algorithm, continuously adjusts the thresholds of each node in the network and the weights between nodes, thereby determining the trained verification model. Generally, a BP neural network consists of an input layer, a hidden layer, and an output layer. The input layer corresponds to several attributes of each piece of data; the hidden layer can be one or more layers, with several nodes in each layer; the output layer can classify different results output by several nodes. Each node itself has a threshold and has weights with each node in the next layer. These parameters are finally determined through multiple iterations of the input sample data, so that new data can be predicted.

[0122] S702. Input the target parameters corresponding to the target network at the current moment into the trained verification model to obtain a quality verification result.

[0123] It can be understood that through the trained verification model, based on the target parameters corresponding to the target network at the current moment, the trained verification model can predict the network quality of the target network to obtain a quality verification result, so that it can be further compared with the quality prediction result obtained by the preset network quality prediction model, thereby verifying the accuracy of the preset network quality prediction model.

[0124] S703. Determine the accuracy of the preset network quality prediction model based on the quality verification result and the quality prediction result.

[0125] Optionally, the mean square error, average error, and average relative error can be used to evaluate the quality verification result and the quality prediction result, thereby determining the accuracy of the preset network quality prediction model.

[0126] Optionally, the preset network quality prediction model can be corrected according to the quality verification result, so that the preset network quality prediction model is more convergent and the quality prediction result of the network is more accurate.

[0127] In the embodiments of the present application, by training the preset verification model, a quality verification result is obtained, and further judged with the quality prediction result of the preset network quality prediction model to determine the accuracy of the preset network quality prediction model, so that the preset network quality prediction model can be continuously corrected to ensure the accuracy of the network quality prediction result.

[0128] The present application provides a network quality prediction method. First, multiple basic parameters affecting the network quality in the target network at the current moment are obtained. Then, the dimension and value range corresponding to each of the multiple basic parameters are uniformly processed to obtain normalized target parameters. Next, the target parameters are input into a preset network quality prediction model to obtain a quality prediction result of the target network including throughput, latency, and packet loss rate, so as to indicate the quality prediction result through throughput, latency, and packet loss rate. Through the above method, based on the preset network quality prediction model obtained by pre-training, the dimension and value range corresponding to each basic parameter are unified, which can improve the data processing efficiency of the model and the accuracy of predicting the network quality of the target network. Thus, the efficiency and accuracy of predicting the network quality of the communication link are effectively improved.

[0129] The above mainly introduces the solution provided by the embodiments of the present application from the perspective of the method. To implement the above functions, it includes the corresponding hardware structure and / or software module for executing each function. Those skilled in the art should easily realize that, combining the units and algorithm steps of each example described in the embodiments disclosed herein, the embodiments of the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed in the form of hardware or computer software driving the hardware depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0130] The embodiments of the present application can divide the functional modules of a network quality prediction device according to the above method examples. For example, each functional module can be divided corresponding to each function, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware or in the form of a software functional module. Optionally, the division of modules in the embodiments of the present application is illustrative, merely a logical functional division, and there may be other division methods in actual implementation.

[0131] Figure 10 It is a schematic structural diagram of a network quality prediction device provided by an embodiment of the present application. As Figure 10 shown, a network quality prediction device 100 is used to improve the accuracy of predicting the network quality of the communication link, for example, for executing Figure 2 the network quality prediction method shown. The network quality prediction device 100 includes: an acquisition unit 1001 and a processing unit 1002;

[0132] The acquisition unit 1001 is configured to acquire multiple basic parameters of the target network at the current moment, and the multiple basic parameters affect the network quality of the target network;

[0133] A processing unit 1002 is used to perform normalization processing on the multiple basic parameters to obtain a target parameter, wherein the normalization processing is used to unify the dimension and value range corresponding to each basic parameter in the multiple basic parameters;

[0134] The processing unit 1002 is also used to input the target parameters into a preset network quality prediction model to determine the quality prediction result of the target network. The preset network quality prediction model is obtained by training based on multiple training data and an elastic network model. The quality prediction result is indicated by at least one of the following: throughput, latency, and packet loss rate.

[0135] In a possible implementation, in a network quality prediction device 100 provided in an embodiment of the present application, the acquisition unit 1001 is further used to acquire multiple groups of first parameters corresponding to the target network, where the multiple groups of first parameters are network parameters corresponding to the target network at multiple historical moments, and a group of first parameters includes multiple basic parameters and multiple performance parameters corresponding to one historical moment;

[0136] The processing unit 1002 is further configured to normalize a plurality of basic parameters included in any one of the plurality of first parameter groups to obtain a plurality of second parameters at a historical moment corresponding to the first parameter group, wherein one second parameter is obtained by normalizing one basic parameter.

[0137] The processing unit 1002 is further configured to determine at least one third parameter from the plurality of second parameters as a set of third parameters, the at least one third parameter including a second parameter having an influence coefficient greater than a first preset threshold, the influence coefficient being used to indicate a correlation between the second parameter and the plurality of performance parameters;

[0138] The processing unit 1002 is further configured to train the elastic network model based on the multiple sets of third parameters determined by the multiple sets of first parameters to obtain a preset network quality prediction model.

[0139] In a possible implementation, in a network quality prediction device 100 provided in an embodiment of the present application, the processing unit 1002 is specifically configured to determine, for any second parameter among the multiple second parameters, an influence coefficient corresponding to any second parameter based on a first preset algorithm, where the first preset algorithm is used to determine the correlation between any two parameters;

[0140] The processing unit 1002 is specifically configured to determine at least one third parameter having an influence coefficient greater than a first preset threshold from the plurality of second parameters as a group of third parameters.

[0141] In a possible implementation, in a network quality prediction device 100 provided in an embodiment of the present application, the processing unit 1002 is specifically configured to determine, based on the multiple groups of third parameters determined by the multiple groups of first parameters, a comprehensive influence coefficient between any third parameter of at least one third parameter and any performance parameter of the multiple performance parameters through a second preset algorithm, wherein the second preset algorithm is configured to determine the correlation between any two parameters in the time dimension;

[0142] The processing unit 1002 is specifically used to determine a fourth parameter whose comprehensive influence coefficient is less than a second preset threshold from at least one third parameter, and input the fourth parameter into the elastic network model for training to obtain a preset network quality prediction model.

[0143] In a possible implementation, in a network quality prediction device 100 provided in an embodiment of the present application, the processing unit 1002 is specifically configured to determine, for any second parameter among the multiple second parameters, a correlation between any second parameter and any performance parameter among the multiple performance parameters based on a first preset algorithm;

[0144] The processing unit 1002 is specifically configured to sum the correlation between any second parameter and each performance parameter in the multiple performance parameters as the influence coefficient corresponding to the any second parameter.

[0145] In a possible implementation, in a network quality prediction device 100 provided in an embodiment of the present application, the processing unit 1002 is further used to train a preset verification model through multiple second parameters and multiple performance parameters corresponding to each historical moment of the target network in multiple historical moments to obtain a trained verification model, and the trained verification model is used to verify the accuracy of the preset network quality prediction model;

[0146] The processing unit 1002 is further used to input the target parameters corresponding to the target network at the current moment into the trained verification model to obtain a quality verification result;

[0147] The processing unit 1002 is further configured to determine the accuracy of a preset network quality prediction model based on the quality verification result and the quality prediction result.

[0148] In the case of implementing the functions of the above-mentioned integrated modules in the form of hardware, the embodiment of the present application provides another possible structural diagram of the electronic device involved in the above-mentioned embodiment. Figure 11 As shown, an electronic device 110 is used to improve the accuracy of predicting the network quality of a communication link, for example, to perform Figure 2A network quality prediction method is shown. The electronic device 110 includes a processor 1101, a memory 1102 and a bus 1103. The processor 1101 and the memory 1102 may be connected via the bus 1103.

[0149] The processor 1101 is the control center of the communication device, which can be a processor or a general term for multiple processing elements. For example, the processor 1101 can be a general-purpose central processing unit (CPU) or other general-purpose processors. Among them, the general-purpose processor can be a microprocessor or any conventional processor.

[0150] As an embodiment, the processor 1101 may include one or more CPUs, such as Figure 11 CPU 0 and CPU 1 are shown in .

[0151] The memory 1102 may be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM) or other type of dynamic storage device that can store information and instructions, an electrically erasable programmable read-only memory (EEPROM), a disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0152] As a possible implementation, the memory 1102 may exist independently of the processor 1101, and the memory 1102 may be connected to the processor 1101 via the bus 1103 for storing instructions or program codes. When the processor 1101 calls and executes the instructions or program codes stored in the memory 1102, a network quality prediction method provided in an embodiment of the present application can be implemented.

[0153] In another possible implementation, the memory 1102 may also be integrated with the processor 1101 .

[0154] The bus 1103 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 11 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0155] It should be pointed out that Figure 11 The structure shown does not constitute a limitation on the electronic device 110. Figure 11 In addition to the components shown, the electronic device 110 may include more or fewer components than shown, or combine certain components, or arrange the components differently.

[0156] As an example, combining Figure 10 The functions implemented by the acquisition unit 1001 and the processing unit 1002 in the electronic device are similar to Figure 11 The function of processor 1101 in is the same.

[0157] Optional, such as Figure 11 As shown, the electronic device 110 provided in the embodiment of the present application may further include a communication interface 1104 .

[0158] The communication interface 1104 is used to connect with other devices through a communication network. The communication network may be Ethernet, wireless access network, wireless local area network (WLAN), etc. The communication interface 1104 may include a receiving unit for receiving data and a sending unit for sending data.

[0159] In one design, in the electronic device provided in the embodiment of the present application, the communication interface can also be integrated into the processor.

[0160] Through the description of the above implementation methods, those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above functional units is used as an example. In practical applications, the above functions can be assigned to different functional units as needed, that is, the internal structure of the device can be divided into different functional units to complete all or part of the functions described above. The specific working process of the system, device and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0161] An embodiment of the present application also provides a computer-readable storage medium, in which instructions are stored. When a computer executes the instructions, the computer executes each step in the method flow shown in the above method embodiment.

[0162] An embodiment of the present application provides a computer program product including instructions. When the instructions are executed on a computer, the computer is enabled to execute a network quality prediction method in the above method embodiment.

[0163] Among them, the computer readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk. Random Access Memory (RAM), Read-Only Memory (ROM), Erasable Programmable Read Only Memory (EPROM), registers, hard disks, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or the above in a suitable combination, or any other form of computer readable storage medium known in the art.

[0164] An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be an integral part of the processor. The processor and the storage medium can be located in an Application Specific Integrated Circuit (ASIC).

[0165] In the embodiments of the present application, a computer-readable storage medium may be any tangible medium that contains or stores a program, which may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0166] Since the electronic device, computer-readable storage medium, and computer program product in the embodiments of the present application can be applied to the above-mentioned method, the technical effects that can be obtained can also refer to the above-mentioned method embodiments, and the embodiments of the present application will not be repeated here.

[0167] The above are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present application should be included in the protection scope of the present application.

Claims

1. A network quality prediction method, characterized in that, the method includes: Obtain multiple basic parameters of the target network at the current moment, where the multiple basic parameters affect the network quality of the target network; Perform normalization processing on the multiple basic parameters to obtain target parameters, and the normalization processing is used to unify the dimension and value range corresponding to each basic parameter in the multiple basic parameters; Input the target parameters into a preset network quality prediction model to determine the quality prediction result of the target network. The preset network quality prediction model is trained based on multiple training data and an elastic net model, and the quality prediction result is indicated by at least one of the following: throughput, latency, and packet loss rate.

2. The method according to claim 1, characterized in that, the method further includes: Obtain multiple groups of first parameters corresponding to the target network. The multiple groups of first parameters are network parameters corresponding to the target network at multiple historical moments. One group of first parameters includes multiple basic parameters and multiple performance parameters corresponding to one historical moment; For any group of first parameters in the multiple groups of first parameters, perform normalization processing on the multiple basic parameters included in the any group of first parameters to obtain multiple second parameters corresponding to the historical moment of the any group of first parameters. One second parameter is obtained by performing normalization processing on one basic parameter; Determine at least one third parameter from the multiple second parameters as a group of third parameters. The at least one third parameter includes second parameters with an influence coefficient greater than a first preset threshold, and the influence coefficient is used to indicate the correlation between the second parameter and the multiple performance parameters; Train the elastic net model based on the multiple groups of third parameters determined from the multiple groups of first parameters to obtain the preset network quality prediction model.

3. The method according to claim 2, characterized in that, the determining at least one third parameter from the multiple second parameters as a group of third parameters includes: For any second parameter in the multiple second parameters, based on a first preset algorithm, determine the influence coefficient corresponding to the any second parameter. The first preset algorithm is used to determine the correlation between any two parameters; Determine at least one third parameter with an influence coefficient greater than the first preset threshold from the multiple second parameters as a group of third parameters.

4. The method according to claim 2, characterized in that, the training the elastic net model based on the multiple groups of third parameters determined from the multiple groups of first parameters to obtain the preset network quality prediction model includes: Based on the multiple groups of third parameters determined from the multiple groups of first parameters, through a second preset algorithm, determine the comprehensive influence coefficient between any third parameter in the at least one third parameter and any performance parameter in the multiple performance parameters. The second preset algorithm is used to determine the correlation between any two parameters in the time dimension; Determine a fourth parameter with a comprehensive influence coefficient less than a second preset threshold from the at least one third parameter, and input the fourth parameter into the elastic net model for training to obtain the preset network quality prediction model.

5. The method according to claim 3, wherein, for any one of the multiple second parameters, based on a first preset algorithm, determining an influence coefficient corresponding to the any one of the second parameters includes: for any one of the multiple second parameters, based on a first preset algorithm, determining the correlation between the any one of the second parameters and any one of the multiple performance parameters; Summing the correlations between the any one of the second parameters and each of the multiple performance parameters as the influence coefficient corresponding to the any one of the second parameters.

6. The method according to any one of claims 2-5, wherein, the method further includes: Training a preset verification model with the multiple second parameters and multiple performance parameters corresponding to each historical moment among multiple historical moments of the target network to obtain a trained verification model, and the trained verification model is used to verify the accuracy of the preset network quality prediction model; Inputting the target parameter corresponding to the target network at the current moment into the trained verification model to obtain a quality verification result; Determining the accuracy of the preset network quality prediction model based on the quality verification result and the quality prediction result.

7. A network quality prediction device, wherein, the network quality prediction device includes: an acquisition unit and a processing unit; The acquisition unit is configured to acquire multiple basic parameters of the target network at the current moment, and the multiple basic parameters affect the network quality of the target network; The processing unit is configured to perform normalization processing on the multiple basic parameters to obtain a target parameter, and the normalization processing is used to unify the dimension and value range corresponding to each of the multiple basic parameters; The processing unit is further configured to input the target parameter into a preset network quality prediction model to determine a quality prediction result of the target network, and the preset network quality prediction model is trained based on multiple training data and an elastic net model, and the quality prediction result is indicated by at least one of the following: throughput, latency, and packet loss rate.

8. The network quality prediction device according to claim 7, wherein, the acquisition unit is further configured to acquire multiple groups of first parameters corresponding to the target network, and the multiple groups of first parameters are network parameters corresponding to the target network at multiple historical moments, and one group of first parameters includes multiple basic parameters and multiple performance parameters corresponding to one historical moment; The processing unit is further configured to perform normalization processing on the multiple basic parameters included in any one of the multiple groups of first parameters to obtain multiple second parameters corresponding to the historical moment of the any one of the multiple groups of first parameters, and one second parameter is obtained by performing normalization processing on one basic parameter. The processing unit is further configured to determine at least one third parameter from the multiple second parameters as a set of third parameters, where the at least one third parameter includes a second parameter whose influence coefficient is greater than a first preset threshold, and the influence coefficient is used to indicate the correlation between the second parameter and the multiple performance parameters; The processing unit is further configured to train the elastic network model based on multiple sets of third parameters determined from the multiple sets of first parameters to obtain the preset network quality prediction model.

9. The network quality prediction device according to claim 8, wherein, The processing unit is specifically configured to, for any one of the multiple second parameters, determine the influence coefficient corresponding to the any one of the second parameters based on a first preset algorithm, and the first preset algorithm is used to determine the correlation between any two parameters; The processing unit is specifically configured to determine at least one third parameter whose influence coefficient is greater than the first preset threshold from the multiple second parameters as a set of third parameters.

10. The network quality prediction device according to claim 8, wherein, The processing unit is specifically configured to, based on multiple sets of third parameters determined from the multiple sets of first parameters, determine a comprehensive influence coefficient between any one of the at least one third parameter and any one of the multiple performance parameters through a second preset algorithm, and the second preset algorithm is used to determine the correlation between any two parameters in the time dimension; The processing unit is specifically configured to determine a fourth parameter whose comprehensive influence coefficient is less than a second preset threshold from the at least one third parameter, and input the fourth parameter into the elastic network model for training to obtain the preset network quality prediction model.

11. The network quality prediction device according to claim 9, wherein, The processing unit is specifically configured to, for any one of the multiple second parameters, determine the correlation between the any one of the second parameters and any one of the multiple performance parameters based on a first preset algorithm; The processing unit is specifically configured to sum the correlations between the any one of the second parameters and each of the multiple performance parameters as the influence coefficient corresponding to the any one of the second parameters.

12. The network quality prediction device according to any one of claims 8-11, wherein, The processing unit is further configured to train a preset verification model through the multiple second parameters and multiple performance parameters corresponding to each historical moment among multiple historical moments of the target network to obtain a trained verification model, and the trained verification model is used to verify the accuracy of the preset network quality prediction model; The processing unit is further configured to input the target parameter corresponding to the target network at the current moment into the trained verification model to obtain a quality verification result; The processing unit is further configured to determine the accuracy of the preset network quality prediction model based on the quality verification result and the quality prediction result.

13. An electronic device, wherein, comprising: A processor and a memory; wherein, the memory is used to store one or more programs, and the one or more programs include computer-executable instructions. When the electronic device runs, the processor executes the computer-executable instructions stored in the memory, so that the electronic device executes a network quality prediction method according to any one of claims 1-6.

14. A computer-readable storage medium storing one or more programs, characterized in that the one or more programs include instructions which, when executed by a computer, cause the computer to execute a network quality prediction method according to any one of claims 1-6.

Images