Equipment multi-data fusion method and system with dial testing function

Abstract

The invention relates to the technical field of network testing devices, in particular to a multi-data fusion method and system for equipment with a dial testing function. The method comprises the following steps: acquiring a network time delay index according to the difference of dial test durations of each feature data point and other feature data points, and acquiring a description capability index of each feature dimension by combining the difference of network test data of the same feature dimension of the feature data points; obtaining an optimization priority index of each feature dimension according to the correlation degree of the network test data of each feature dimension and the network time delay index and the network time delay index; and performing data dimension reduction on the feature data points based on the optimization priority indexes to obtain low-dimensional data points, and performing data fusion on the network test data after dimension reduction of the feature dimensions of the low-dimensional data points. According to the method, the dimension of the data is reduced by utilizing the optimization priority reflecting the real-time degree of the data, so that the information obtained by fusing the data subjected to the dimension reduction can remarkably reflect the real-time characteristic of the network.

Description

Technical Field

[0001] The present invention relates to the technical field of network testing device analysis, and in particular to a method and system for fusing multiple data of equipment with a dialing test function. Background Art

[0002] Devices with dial-up testing functions are mainly used to monitor and evaluate network performance and availability. With the development of electronic information technology and network communication technology, the amount of data, data diversity, and complexity of data relationships in equipment systems are all relatively large. In order to ensure the timeliness and accuracy of data analysis, it is necessary to perform data fusion on multiple types of data. Data fusion aims to generate more accurate and comprehensive information or knowledge. Because the amount of data and features in the equipment system are large, directly fusing the network parameter data generated by the equipment system results in high computational costs and a lot of redundant information, so data dimensionality reduction is required before data fusion processing.

[0003] Traditional data dimensionality reduction methods, such as the principal component analysis (PCA) algorithm, can reduce the dimension of data and usually focus on reducing the complexity of data transmission and processing. However, when analyzing networks with high real-time requirements, the real-time performance of network analysis is usually ignored during the data dimensionality reduction process, making it difficult to meet the needs of scenarios with low network latency requirements. Therefore, existing data dimensionality reduction methods do not fully consider the real-time performance of network analysis and cannot provide efficient dimensionality reduction that meets the real-time requirements of the network, resulting in weak network real-time characteristics after the fusion of the reduced data, resulting in the loss of network real-time information for devices with dial-up testing functions. Summary of the invention

[0004] In order to solve the technical problem that the dimension reduction method does not fully consider the real-time nature of network analysis, resulting in the weak network real-time characteristics of the information after fusion of the dimension-reduced data, the purpose of the present invention is to provide a device multi-data fusion method and system with a dialing function, and the technical solution adopted is as follows:

[0005] In a first aspect, an embodiment of the present invention provides a method for fusing multiple data of a device with a dialing function, the method comprising:

[0006] Control the dialing test tool to perform at least two dialing tests on the dialing test object, and obtain characteristic data points of each dialing test and the dialing test duration of the corresponding characteristic data, wherein the characteristic data points include network test data of different characteristic dimensions;

[0007] According to the difference between the dialing time of each characteristic data point and the other characteristic data points, the network delay index of each characteristic data point is obtained;

[0008] Obtaining a description capability index for each feature dimension based on differences in network test data of the same feature dimension of all feature data points and the network delay index;

[0009] According to the correlation degree between the network test data of each characteristic dimension of the characteristic data point and the network delay index, and the description capability index, the optimization priority index of each characteristic dimension is obtained;

[0010] The characteristic data points are subjected to data dimensionality reduction according to the optimization priority index to obtain low-dimensional data points; and data fusion is performed on the network test data after the dimensionality reduction of the characteristic dimensions of the low-dimensional data points.

[0011] Furthermore, the method for obtaining the network delay index of each characteristic data point includes:

[0012] The start time of each dial test on the dial test object is obtained, and recorded as the dial test time of the characteristic data point of each dial test; based on the interval between the dial test time of each characteristic data point and the other characteristic data points, the adjacent data points of each characteristic data point are obtained; the average of the difference between the dial test time of each characteristic data point and the adjacent data points is used as the network fluctuation value of the corresponding characteristic data point;

[0013] Obtaining a concentrated value of the duration according to the dialing duration of all characteristic data points; taking the difference between the dialing duration of each characteristic data point and the concentrated value of the duration as the delay duration of the corresponding characteristic data point;

[0014] The network delay index of each characteristic data point is obtained according to the network fluctuation value and the delay duration.

[0015] Furthermore, the method for obtaining the description capability index of each feature dimension includes:

[0016] The network test data is standardized to obtain standard test data of each feature dimension of all feature data points; one feature dimension is selected as the analysis dimension, and a standard difference value of the analysis dimension of each feature data point is obtained according to the difference between the standard test data of the analysis dimension of each feature data point and the mean of the standard test data of the analysis dimension of all feature data points;

[0017] Obtain a comprehensive description value of the analysis dimension of each characteristic data point based on the network delay index and the standard difference value; and take the cumulative sum of the comprehensive description values ​​of the analysis dimension of all characteristic data points as the description capability index of the analysis dimension.

[0018] Furthermore, the method for obtaining the optimization priority index of each feature dimension includes:

[0019] The absolute value of the correlation coefficient between the standard test data of each feature dimension of all feature data points and the network delay index is used as the delay correlation of each feature dimension; and the optimization priority index of each feature dimension is obtained based on the delay correlation and the description capability index.

[0020] Furthermore, the method for obtaining low-dimensional data points includes:

[0021] The network test data of the same characteristic dimension of the characteristic data points are subjected to zero-mean processing to obtain the zero-mean data of each characteristic dimension of each characteristic data point; the product of the zero-mean data of each characteristic dimension of each characteristic data point and the optimization priority index is used as the optimized data of each characteristic dimension of each characteristic data point; an optimization matrix is ​​constructed according to the optimized data of the characteristic dimensions of all characteristic data points; the principal component analysis method is used to perform data dimensionality reduction on the optimization matrix to obtain a reduced dimension matrix; and low-dimensional data points are formed by all elements in each row of the reduced dimension matrix.

[0022] Furthermore, the correlation coefficient is a Pearson correlation coefficient.

[0023] Furthermore, the number of rows of the optimization matrix is ​​equal to the total number of characteristic dimensions of the characteristic data points, and the number of columns is equal to the total number of characteristic dimensions of the characteristic data points.

[0024] Furthermore, the method for obtaining a concentrated value of duration based on the dialing durations of all characteristic data points includes: taking an average of the dialing durations of all characteristic data points as the concentrated value of duration.

[0025] Furthermore, the method for standardizing the network test data is a range standardization method.

[0026] In a second aspect, another embodiment of the present invention provides a device multi-data fusion system with a dialing test function, the system comprising:

[0027] A data acquisition module is used to control the dial test tool to perform at least two dial tests on the dial test object, and obtain characteristic data points of each dial test and the dial test duration of the corresponding characteristic data, wherein the characteristic data points include network test data of different characteristic dimensions;

[0028] The delay analysis module is used to obtain the network delay index of each characteristic data point based on the difference between the dialing time of each characteristic data point and the other characteristic data points;

[0029] A description analysis module, used to obtain a description capability index for each feature dimension based on the difference between the network test data of the same feature dimension of all feature data points and the network delay index;

[0030] A priority analysis module, used to obtain an optimization priority index for each feature dimension based on the correlation between the network test data of each feature dimension of the feature data point and the network delay index, and the description capability index;

[0031] The data fusion module is used to perform data dimensionality reduction on the characteristic data points according to the optimization priority index to obtain low-dimensional data points; and perform data fusion on the network test data after the dimensionality reduction of the characteristic dimensions of the low-dimensional data points.

[0032] An electronic device comprises a processor, a memory and a program or instruction stored in the memory and running on the processor, wherein the program or instruction implements the steps of the impurity detection method as described above when executed by the processor.

[0033] The present invention has the following beneficial effects:

[0034] In an embodiment of the present invention, the network delay index is analyzed by utilizing the difference in the dialing test time of feature data points of different dialing tests; the difference in the network test data of the same feature dimension of the feature data point reflects the network link state of the network test data of each feature dimension, and further reflects the description capability of the feature dimension to the network link state, and the description capability index of the feature dimension is obtained in combination with the network delay index; the degree of correlation between the network test data of each feature dimension of the feature data point and the network delay index reflects the significance of the feature dimension to the delay feature of the overall dialing process, the description capability index reflects the expression capability of the feature dimension to the network link feature, and the optimization priority index is comprehensively obtained, and the feature dimension with higher significance of the delay feature of the dialing process and the expression capability of the network link feature should be retained in the data dimensionality reduction process and has a higher priority; data dimensionality reduction is performed on the feature data points based on the optimization priority index. Compared with the traditional data dimensionality reduction method, the characteristic dimensions of the low-dimensional data points obtained after data dimensionality reduction can more significantly reflect the network real-time performance of the dial-up test process. Data fusion is performed on the network test data after dimensionality reduction of the characteristic dimensions of the low-dimensional data points, which effectively improves the real-time performance of network analysis, reduces the loss of network real-time information caused by data dimensionality reduction, reduces the data fusion calculation cost, and improves the efficiency of network status analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1A flowchart of a method for fusing multiple data of a device with a dialing function provided by an embodiment of the present invention;

[0037] Figure 2 A flowchart of the steps of a method for obtaining a network delay index of a characteristic data point provided by an embodiment of the present invention;

[0038] Figure 3 A structural diagram of a multi-data fusion system of a device with a dialing and testing function provided by an embodiment of the present invention;

[0039] Figure 4 A schematic diagram of a computer device of a multi-data fusion device with a dialing function provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the multi-data fusion method and system of a device with a dialing function proposed by the present invention, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0041] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0042] The specific scenario targeted by the present invention is that a device with a dialing test function will generate multiple types of data during the test process, such as performance indicator data and traffic consumption data. During the dialing test process, these data are crucial to the evaluation of the quality of the network connection. However, a single data may not be sufficient to provide a complete network health status, so it is necessary to merge these data for a more comprehensive analysis. In order to improve the real-time performance of network status analysis, it is also necessary to optimize the data fusion process.

[0043] The specific scheme of a device multi-data fusion method and system with a dialing test function provided by the present invention is described in detail below with reference to the accompanying drawings.

[0044] Embodiment 1:

[0045] The present invention proposes a method for fusion of multiple data of a device with dialing function, please refer to Figure 1 , which shows a flowchart of a method for fusion of multiple data of a device with a dialing function provided by an embodiment of the present invention, the method comprising:

[0046] Step S1: Control the dialing test tool to perform at least two dialing tests on the dialing test object, and obtain characteristic data points of each dialing test and the dialing test duration of the corresponding characteristic data, wherein the characteristic data points include network test data of different characteristic dimensions.

[0047] Dial testing refers to testing a system, application or website to determine whether it is working properly and to monitor the performance of key interfaces. The main purpose of dial testing is to ensure software quality and stability, detect possible problems by simulating user access requests and network connections, and thus help the development team discover and solve problems in a timely manner.

[0048] A device with a dialing test function is used as a dialing test tool. The device with a dialing test function simulates real user behavior to perform regular or on-demand access, interaction and data transmission on Internet services. The dialing test tool is a device that can connect to the Internet and can receive and send data, such as a mobile phone and a computer. The dialing test tool is used to perform multiple dialing tests on the dialing test object. The dialing test object can be a server, a router, and various network applications. A set of network test data is obtained for each dialing test, and the start time of each dialing test and the time when the network test data is received are obtained. In the embodiment of the present invention, each set of network test data includes: packet loss rate, jitter, and bandwidth. Each type of data corresponds to a feature dimension. There is only one data under each feature dimension in a set of network test data. The network test data of all feature dimensions obtained from each dialing test constitute a feature data point; the length of the time period between the start time of each dialing test and the time when the network test data is received is taken as the dialing test duration.

[0049] It should be noted that, in this embodiment, the dialing test frequency takes an empirical value of 10 minutes, that is, the dialing test object is tested once every 10 minutes using the dialing test tool. The implementer can set the dialing test frequency according to the actual situation.

[0050] Step S2: Obtain the network delay index of each characteristic data point according to the difference between the dialing time of each characteristic data point and the other characteristic data points.

[0051] For network analysis modules with high real-time requirements, the PCA algorithm is used to directly reduce the dimension of the network test data generated by the dial test device before data fusion, which is easy to affect the real-time performance of network analysis and is difficult to meet the needs of scenarios with low network delay requirements. In order to improve the real-time performance of the data fusion process of the dial test device, this solution optimizes the process of reducing the dimension of the network test data of the dial test process through the PCA algorithm.

[0052] The dial test duration is the time required from the start of the dial test to the return of the dial test data. It reflects the delay of the network link and the quality of the network connection. If the network fluctuates, the time required from the start of the dial test to the return of the dial test data is longer, and the delay characteristics of the network test data in the network analysis process are more obvious. Therefore, the dial test duration is used as the main analysis feature of the network delay during each dial test. The greater the difference between the dial test duration of each characteristic data point and the other characteristic data points, the greater the degree of network anomaly corresponding to the characteristic data point in the dial test, and the more obvious the network delay in the dial test process. Therefore, the network delay index is obtained based on the difference between the dial test duration of each characteristic data point and the other characteristic data points.

[0053] See also Figure 2 , which shows a flowchart of the steps of a method for obtaining a network delay index of a characteristic data point provided by an embodiment of the present invention, the method comprising:

[0054] Step S210: Obtain the start time of each dial test on the dial test object, and record it as the dial test time of the characteristic data point of each dial test; based on the difference in the dial test time between each characteristic data point and the other characteristic data points, obtain the adjacent data points of each characteristic data point; and use the average of the difference in dial test time between each characteristic data point and the adjacent data points as the network fluctuation value of the corresponding characteristic data point.

[0055] The dialing duration of characteristic data points of multiple dialing tests with a close interval between the dialing times corresponding to each characteristic data point can accurately reflect the network fluctuation in the adjacent time period of the dialing test. Therefore, the adjacent data points of each characteristic data point are determined based on the interval between the dialing times of the characteristic data points.

[0056] Preferably, in an embodiment of the present invention, a method for acquiring neighboring points of a feature data point includes: based on the dialing time of the feature data point, arranging the feature data points in time series to obtain a feature sequence, and taking the remaining feature data points except the feature data point within a preset window of each feature data point in the feature sequence as neighboring data points of each feature data point.

[0057] It should be noted that the accuracy of local delay analysis is highest when the number of neighboring data points before and after the dialing time of each characteristic data point is the same. In this embodiment of the present invention, the characteristic data point is located at the center of its preset window, and the length of the preset window takes the empirical value of 3 and the width is 1. The implementer can set the length of the preset window according to the specific situation; it should be noted that the neighboring data point of the first characteristic data point in the characteristic sequence is the second characteristic data point in the characteristic sequence, and the neighboring data point of the last characteristic data point is the second to last characteristic data point in the characteristic sequence. In another embodiment of the present invention, the characteristic data point can be located at the beginning or the last position in its preset window.

[0058] In other embodiments of the present invention, any one characteristic data point is selected as the analysis data point, and the remaining characteristic data points, except for the analysis data point, corresponding to the dialing time in the preset adjacent time period of the dialing time of the analysis data point are used as the adjacent data points of the analysis data point. It should be noted that the analysis data point is located in the middle of its preset adjacent time period, and the duration of the preset adjacent time period is an empirical value of 20 minutes.

[0059] The greater the difference in dialing time between a feature data point and each adjacent data point, the greater the probability that network fluctuations will occur in the dialing process corresponding to the feature data point under the network delay corresponding to the dialing of each adjacent data point. Therefore, the average of the differences in dialing time between each feature data point and its adjacent data points is taken as the network fluctuation value of the corresponding feature data point, so that the network fluctuation value can more accurately present the network fluctuation situation of the dialing process corresponding to each feature data point.

[0060] Step S220: obtaining a concentrated value of the duration according to the dialing durations of all characteristic data points; and taking the difference between the dialing duration of each characteristic data point and the concentrated value of the duration as the delay duration of the corresponding characteristic data point.

[0061] The duration concentration value is the concentration trend of the dialing duration of all dialing processes, reflecting the overall characteristics of the network delay of all dialing. In an embodiment of the present invention, the method for obtaining the duration concentration value includes: taking the mean of the dialing duration of all characteristic data points as the duration concentration value. In other embodiments of the present invention, the median or mode of the dialing duration of the characteristic data points can also be taken as the duration concentration value, and the median and mode can also reflect the centralized distribution trend of a group of data.

[0062] The difference between the dialing duration of each characteristic data point and the concentrated value of the duration reflects the overall characteristics of the network delay compared to the dialing process, and the network abnormality during a single dialing test. If the difference between the dialing duration of the characteristic data point of each dialing test and the concentrated value of the duration is larger, the long delay in the dialing test is more obvious, and there may be problems such as network congestion and high load.

[0063] Step S230: Obtain the network delay index of each characteristic data point according to the network fluctuation value and the delay duration.

[0064] The network fluctuation value analyzes the delay of the characteristic data point of each dial test locally, and the delay duration analyzes the delay of the characteristic data point of each dial test as a whole; if the network fluctuation value and delay duration of the characteristic data point of each dial test are larger, the probability of network fluctuation in this dial test is greater, and the network delay time is longer, then the network fluctuation value and delay duration are both positively correlated with the network delay index. In an embodiment of the present invention, the network fluctuation value is normalized to obtain a normalized network fluctuation value, and the product of the normalized network fluctuation value of each characteristic data point and the delay duration is normalized to obtain the network delay index of each characteristic data point.

[0065] In the embodiment of the present invention, the correlation between the network fluctuation value and the delay duration and the network delay index, such as the sum value, can also be constructed through other basic mathematical operations, which are not limited or elaborated here.

[0066] It should be noted that in the embodiment of the present invention, the Sigmoid function is used to normalize the product of the normalized network fluctuation value and the delay duration. Other normalization methods may also be selected in the embodiment of the present invention, such as function transformation, maximum and minimum normalization, etc., which are not limited here.

[0067] Step S3: According to the difference of network test data of the same feature dimension of all feature data points and the network delay index, the description capability index of each feature dimension is obtained.

[0068] The network test data generated during the dial test process will change with the changes in the actual network link status. The network test data of each feature dimension have different angles of reflecting the network link status. When optimizing the network test data of different feature dimensions during the data dimensionality reduction process, it is necessary to analyze the optimization priority indicators of the network test data of each feature dimension in combination with the focus on the changes in the network link status.

[0069] To analyze the priority of network test data of different feature dimensions, it is necessary to analyze the expressive ability of a certain feature of the network link state in combination with the network test data under different dialing conditions. Specifically, if the network fluctuates, the network test data of different feature dimensions of the feature data point will also change within a certain range. Therefore, the difference of the network test data of the same feature dimension of all feature data points reflects the network link state of the network test data of each feature dimension. In the process of optimizing the dimensionality reduction process of multidimensional data, it is necessary to highlight the data changes that can reflect the network link state as much as possible, so that the network connection situation can be accurately analyzed through the optimized multidimensional data. The delay characteristics of different data points are different. The feature data points with larger delay characteristics can better reflect the fluctuations in the network link process. Therefore, the feature data points with larger network delay indicators have stronger description capabilities for the network link state. In summary, based on the difference of the network test data of the same feature dimension of all feature data points and the network delay indicator, the description capability index of each feature dimension is obtained.

[0070] Step S310: Standardize the network test data to obtain standard test data for each feature dimension of all feature data points.

[0071] Since the dimensions of network test data of different feature dimensions are different, it is necessary to standardize the network test data of the feature dimensions of the feature data points before analysis. In the embodiment of the present invention, the range standardization method is selected for standardization. In other embodiments of the present invention, other methods can also be used for standardization, such as Z-score standardization, decimal calibration standardization, and sum normalization standardization.

[0072] Step S320: select any one feature dimension as the analysis dimension, and obtain the standard difference value of the analysis dimension of each feature data point based on the difference between the standard test data of the analysis dimension of each feature data point and the mean of the standard test data of the analysis dimension of all feature data points.

[0073] To analyze the ability of network test data of different feature dimensions to describe network fluctuations, we can analyze it from the purpose of dialing test. The purpose of dialing test is to simulate the link status from users to service access points in real usage environment and provide the best access point for service scheduling system. For different devices with dialing test function, dialing test will simulate the situation of users connecting to service access points in various situations. The difference of network link status when users connect to service points in different situations is reflected through network test data. Therefore, for network test data of different feature dimensions, especially in the case of multiple dialing test devices, in order to reflect the network characteristics of these devices when simulating different real users' access, the distribution of network test data of each feature dimension should be relatively dispersed, that is, the greater the degree of dispersion of network test data of the same feature dimension of feature data point, the greater the standard difference value of the analysis dimension of feature data point, so as to reflect the difference of dialing test of different devices and better analyze the network link status.

[0074] In an embodiment of the present invention, the square value of the difference between the standard test data of the analysis dimension of each feature data point and the mean of the standard test data of the analysis dimension of all feature data points is used as the standard difference value of the analysis dimension of each feature data point. In another embodiment of the present invention, the absolute value of the difference between the standard test data of the analysis dimension of each feature data point and the mean of the standard test data of the analysis dimension of all feature data points can be used as the standard difference value of the analysis dimension of each feature data point; in other embodiments of the present invention, the mean of the standard test data of the analysis dimension of all feature data points can also be replaced by the median or mode that represents the trend of the data set.

[0075] Step S330: Obtain the comprehensive description value of the analysis dimension of each characteristic data point based on the network delay index and the standard difference value; and take the cumulative sum of the comprehensive description values ​​of the analysis dimension of all characteristic data points as the description capability index of the analysis dimension.

[0076] The larger the standard difference value, the more the network test data of the analysis dimension of a single feature data point can reflect the network characteristics of the device with dialing function simulating different real users when accessing, and improve the expression ability of the analysis dimension of a single feature data point on the network connection characteristics; the feature data point with a larger network delay index can better reflect the fluctuations in the network connection process, and the stronger the ability to describe the network connection status, the network delay index and the standard difference value are positively correlated with the comprehensive description value. In an embodiment of the present invention, the network delay index is normalized, and the product of the standard difference value of the analysis dimension of each feature data point and the normalized network delay index is used as the comprehensive description value of the analysis dimension of each feature data point.

[0077] If the analysis dimension is the jth feature dimension, taking the jth feature dimension as an example for analysis, in the embodiment of the present invention, the calculation formula for obtaining the description capability index of the jth feature dimension is as follows:

[0078] In the formula, S j is the descriptive ability index of the jth feature dimension, n is the total number of feature data points, τ i is the network delay index of the i-th characteristic data point, x i,j is the standard test data of the jth feature dimension of the i-th feature data point, is the mean of the standard test data of the jth feature dimension of all feature data points, is the standard difference value of the j-th feature dimension of the i-th feature data point, is the comprehensive description value of the jth feature dimension of the ith feature data point, and Norm is the normalization function.

[0079] It should be noted that the description capability index can be understood as obtaining the discrete degree of the standard test data of the j-th feature dimension of all feature data points, and in the process of obtaining the discrete degree, the network delay index is used to weight the standard test data. is the degree to which the standard test data of the jth feature dimension of the ith feature data point deviates from the standard test data center of this dimension of all feature data points, and the above deviation degrees of the standard test data of this dimension of all feature data points are accumulated to obtain the overall deviation degree of the jth feature dimension of all feature data points; when The larger the value is, the greater the deviation of the standard test data of the j-th feature dimension is. The network test data of the j-th feature dimension can better reflect the network characteristics of the device with dial-up test function when simulating different real users access. The stronger the ability of this feature dimension to express the network link characteristics is, the better the description capability index S is. j The larger the network delay index τ i When is larger, the i-th feature data point can better reflect the fluctuation of the network connection process, thereby reflecting the data characteristics when the network status is poor, and the feature data point has a stronger ability to describe the network connection status.

[0080] In another embodiment of the present invention, a method for obtaining a descriptive capability index for each feature dimension includes: normalizing the network delay index, taking the product of the standard test data of each feature dimension of each feature data point and the normalized network delay index as the weighted index for each feature dimension of each feature data point; and taking the variance of the weighted index for each feature dimension of all feature data points as the descriptive capability index for each feature dimension. In other embodiments of the present invention, the standard deviation, coefficient of variation or range of the weighted index for each feature dimension of all feature data points may also be taken as the descriptive capability index for the corresponding feature dimension.

[0081] According to the method for obtaining the descriptive capability index of the analysis dimension, the descriptive capability index of each characteristic dimension is obtained. It should be noted that the method for obtaining the descriptive capability index of the analysis dimension and each characteristic dimension is the same, which is equivalent to replacing the analysis dimension with each characteristic dimension.

[0082] Step S4: According to the correlation between the network test data of each feature dimension of the feature data point and the network delay index, and the description capability index, the optimization priority index of each feature dimension is obtained.

[0083] In order to achieve the purpose of real-time network analysis, after obtaining the overall descriptive features of different feature dimensions, it is also necessary to combine the network test data of different feature dimensions in different dialing processes to analyze the optimization priority of feature dimensions. The greater the correlation between the network test data of each feature dimension of the feature data point and the network delay index, the more the network test data of this feature dimension can reflect the delay characteristics of the overall dialing process, and the network connection status can be analyzed more accurately through the delay angle corresponding to this feature dimension. The more the above feature dimensions need to be retained in the data dimensionality reduction process, the higher the priority; the descriptive ability index reflects the ability of the feature dimension to express the characteristics of the network link. The feature dimension that can express the characteristics of the network link can improve the accuracy of the analysis of the network link situation. Therefore, according to the correlation between the network test data of each feature dimension of the feature data point and the network delay index, as well as the descriptive ability index, the optimization priority index of each feature dimension is obtained.

[0084] Preferably, in an embodiment of the present invention, the method for obtaining the optimization priority index includes: taking the absolute value of the correlation coefficient between the standard test data of each feature dimension of all feature data points and the network delay index as the delay correlation of each feature dimension; and obtaining the optimization priority index of each feature dimension based on the delay correlation and the description capability index.

[0085] It should be noted that, in the embodiment of the present invention, the correlation coefficient between the standard test data of each feature dimension of the feature data point and the network delay index can be a Pearson correlation coefficient, or a Spearman correlation coefficient, a Kendall rank correlation coefficient, or a typical correlation coefficient. Among them, the Pearson correlation coefficient is a well-known technology and will not be described in detail here.

[0086] When the absolute value of the Pearson correlation coefficient is closer to 1, there is a strong correlation between the standard test data of each feature dimension of all feature data points and the network delay index. The greater the delay correlation, the more accurate the analysis of the network connection status through the delay angle corresponding to the feature dimension, the more the feature dimension needs to be retained in the data dimension reduction process, the higher the priority, and the larger the optimization priority index. When the descriptive ability index of each feature dimension is larger, the expressive ability of the feature dimension for the network link characteristics is stronger. By increasing the accuracy of the feature dimension in analyzing the network link situation, the accuracy of the delay angle analysis of the network connection status corresponding to the feature dimension is improved. Therefore, the delay correlation and the descriptive ability index are both positively correlated with the optimization priority index. In an embodiment of the present invention, the product of the delay correlation and the descriptive ability index of each feature dimension is normalized to obtain the optimization priority index of each feature dimension.

[0087] In the embodiment of the present invention, the correlation between the delay correlation and the description capability index and the optimization priority index, such as the sum value, can also be constructed through other basic mathematical operations, which are not limited or elaborated here.

[0088] It should be noted that in the embodiment of the present invention, the Sigmoid function is used to normalize the product of the network fluctuation value and the delay duration. In the embodiment of the present invention, other normalization methods may also be selected, such as function transformation, maximum and minimum normalization, etc., which are not limited here.

[0089] Step S5: Perform data dimensionality reduction on the characteristic data points according to the optimization priority index to obtain low-dimensional data points; perform data fusion on the network test data after the dimensionality reduction of the characteristic dimensions of the low-dimensional data points.

[0090] After obtaining the optimization priority index of the network test data of each feature dimension, in order to improve the real-time performance of network analysis, the network test data of different feature dimensions of the feature data points are subjected to dimensionality reduction processing based on the optimization priority index.

[0091] Preferably, the method for obtaining low-dimensional data points in this embodiment includes: performing zero-mean processing on the network test data of the same characteristic dimension of the characteristic data point to obtain the zero-mean data of each characteristic dimension of each characteristic data point; multiplying the zero-mean data of each characteristic dimension of each characteristic data point by the optimization priority index as the optimized data of each characteristic dimension of each characteristic data point; constructing an optimization matrix based on the optimized data of the characteristic dimensions of all characteristic data points; performing data dimensionality reduction on the optimization matrix using the principal component analysis method to obtain a reduced dimension matrix; and forming a low-dimensional data point from all elements in each row of the reduced dimension matrix. Among them, the PCA algorithm is a well-known technology for those skilled in the art and will not be described in detail here.

[0092] It should be noted that the network test data of the same feature dimension of the feature data point is zero-mean processed, specifically: subtract the network test data of each feature dimension of each feature data point from the mean of the network test data of the feature dimension of all feature data points, as the zero-mean data of each feature dimension of each feature data point. In order to retain the feature dimension features that have better expression ability of the current network connection status in each feature dimension during the dial test process as much as possible during data dimensionality reduction, it is necessary to adjust the zero-mean data in combination with the optimization priority of different feature dimensions. The specific operation is: the product of the zero-mean data of each feature dimension of each feature data point and the optimization priority index of the feature dimension is used as the optimized data of each feature dimension of each feature data point, and the degree of discreteness between the zero-mean data of the same feature dimension of the feature data point is further adjusted using the optimization priority index. The greater the optimization priority of each feature dimension, the better the expression ability of the feature dimension for the current network connection status, the greater the variance of the zero-mean data of the feature dimension of the feature data point, and the easier it is to retain the feature dimension. As an example, if an optimization matrix is ​​constructed based on the optimized data of the feature dimensions of three feature data points, Then the row vector (y11 y12 y13) is composed of the optimized data of the first feature dimension of the three feature data points, and the column vector It is composed of the optimized data of all characteristic dimensions of the first characteristic data point; the number of rows of the optimized matrix is ​​equal to the total number of characteristic dimensions of the characteristic data points, and the number of columns is equal to the total number of characteristic dimensions of the characteristic data points. Obtain the covariance matrix C of the optimized matrix Y, and calculate the eigenvalues ​​and eigenvectors of the covariance matrix C, and arrange the eigenvectors into a matrix from top to bottom by row according to the corresponding eigenvalues ​​from large to small, and take the first k rows to form a matrix P, where the first k rows are determined by the cumulative contribution rate of the eigenvalues. In this embodiment, the cumulative contribution rate threshold is set to 85% according to the empirical value. Construct the target matrix X according to the network test data of the characteristic dimensions of all characteristic data points, and finally use the matrix P to multiply the matrix X to obtain the dimension reduction matrix W. All elements in each row of the dimension reduction matrix W constitute low-dimensional data points.

[0093] The meanings of the number of rows and columns in the target matrix and the optimization matrix are the same, the difference is that each element in the optimization matrix represents optimization data, while each element in the target matrix represents network test data.

[0094] It should be noted that the feature dimensions of the low-dimensional data points are included in the feature dimensions of the feature data points, and the feature dimensions of the low-dimensional data points are equivalent to selecting part of all the feature dimensions of the feature data points; the network test data after dimensionality reduction of each feature dimension of each low-dimensional data point is the element of the low-dimensional data point in the corresponding row and column position of the dimensionality reduction matrix W.

[0095] Data fusion is performed on the network test data after the feature dimension of the low-dimensional data points is reduced. Lai Weiping, Lin Bixing. Smart grid data fusion method based on PCA-MP-BP [J]. Microcomputer Applications, 2022, 38(01): 198-201.

[0096] All characteristic dimensions of low-dimensional data points can more significantly reflect the network real-time performance of the dialing process. After fusing multiple device data, the real-time performance of network analysis is effectively improved, the loss of network real-time information caused by data dimensionality reduction is reduced, the data fusion calculation cost is reduced, and the efficiency of network status analysis is improved.

[0097] In other embodiments of the present invention, data fusion methods such as comprehensive averaging method, Kalman filtering method, Bayesian estimation method and statistical decision theory can be used to perform data fusion on the network test data after the dimension reduction of the feature dimensions of the low-dimensional data points.

[0098] So far, the present invention is completed.

[0099] In summary, in an embodiment of the present invention, a network delay index is obtained based on the difference in dialing time between each feature data point and the other feature data points, and a description capability index of each feature dimension is obtained in combination with the difference in network test data of the same feature dimension of the feature data point; an optimization priority index of each feature dimension is obtained based on the degree of correlation between the network test data of each feature dimension and the network delay index, and the network delay index; based on the optimization priority index, the feature data points are subjected to data dimension reduction to obtain low-dimensional data points, and data fusion is performed on the network test data after dimension reduction of the feature dimension of the low-dimensional data points. The present invention utilizes the optimization priority that reflects the degree of real-time data to reduce the dimension of the data, so that the information after fusion of the reduced-dimensional data can significantly reflect the real-time characteristics of the network.

[0100] Embodiment 2:

[0101] The present invention proposes a device multi-data fusion system with a dialing test function, please refer to Figure 3, which shows a structure diagram of a device multi-data fusion system with a dialing test function provided by an embodiment of the present invention, the system includes:

[0102] The data collection module 610 is used to control the dialing test tool to perform at least two dialing tests on the dialing test object, and obtain the characteristic data points of each dialing test and the dialing test duration of the corresponding characteristic data, wherein the characteristic data points include network test data of different characteristic dimensions.

[0103] The delay analysis module 620 is used to obtain the network delay index of each characteristic data point according to the difference between the dialing time of each characteristic data point and the other characteristic data points.

[0104] The description analysis module 630 is used to obtain the description capability index of each feature dimension based on the difference of network test data of the same feature dimension of all feature data points and the network delay index.

[0105] The priority analysis module 640 is used to obtain the optimization priority index of each feature dimension according to the correlation between the network test data of each feature dimension of the feature data point and the network delay index, as well as the description capability index.

[0106] The data fusion module 650 is used to perform data dimensionality reduction on the characteristic data points according to the optimization priority index to obtain low-dimensional data points; and perform data fusion on the network test data after the dimensionality reduction of the characteristic dimensions of the low-dimensional data points.

[0107] It should be noted that the device provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the multi-data fusion system of a device with a dialing function and the multi-data fusion method of a device with a dialing function provided in the above embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0108] Embodiment 3:

[0109] Figure 4 A schematic diagram of a computer device of a multi-data fusion device with a dialing function provided by an embodiment of the present invention. For example, Figure 4 As shown, the computer device includes: a memory 701, a processor 702, and a computer program 703 stored in the memory 701 and running on the processor 702, wherein when the processor 702 executes the computer program 703, the computer device can execute any one of the multi-data fusion methods for a device with a dialing function introduced above.

[0110] In addition, an embodiment of the present application also protects a device, which may include a memory and a processor, wherein the memory stores an executable program code, and the processor is used to call and execute the executable program code to execute a multi-data fusion method for a device with a dialing function provided in an embodiment of the present application.

[0111] In this embodiment, the functional modules of the device can be divided according to the above method example. For example, each functional module can be corresponded, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0112] It should be understood that the device provided in this embodiment is used to execute the above-mentioned method for fusing multiple data of a device with a dialing function, and thus can achieve the same effect as the above-mentioned implementation method.

[0113] In the case of an integrated unit, the device may include a processing module and a storage module. When the device is applied to a device, the processing module may be used to control and manage the actions of the device. The storage module may be used to support the device to execute mutual program codes, etc.

[0114] The processing module may be a processor or a controller, which may implement or execute various exemplary logic blocks, modules and circuits disclosed in conjunction with the present application. The processor may also be a combination that implements a computing function, such as a combination of one or more microprocessors, a combination of a digital signal processing (DSP) and a microprocessor, etc. The storage module may be a memory.

[0115] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0116] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0117] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for fusion of multiple data of a device with a dialing test function, characterized in that: The method includes: Control the dialing test tool to perform at least two dialing tests on the dialing test object, and obtain characteristic data points of each dialing test and the dialing test duration of the corresponding characteristic data, wherein the characteristic data points include network test data of different characteristic dimensions; According to the difference between the dialing time of each characteristic data point and the other characteristic data points, the network delay index of each characteristic data point is obtained; Obtaining a description capability index for each feature dimension based on differences in network test data of the same feature dimension of all feature data points and the network delay index; According to the correlation degree between the network test data of each characteristic dimension of the characteristic data point and the network delay index, and the description capability index, the optimization priority index of each characteristic dimension is obtained; The characteristic data points are subjected to data dimensionality reduction according to the optimization priority index to obtain low-dimensional data points; and data fusion is performed on the network test data after the dimensionality reduction of the characteristic dimensions of the low-dimensional data points.

2. According to the method for fusion of multiple data of a device with dialing function in claim 1, it is characterized in that: The method for obtaining the network delay index of each characteristic data point includes: Obtain the start time of each dial test for the dial test object, and record it as the dial test time of the characteristic data point of each dial test; obtain the adjacent data points of each characteristic data point based on the difference between the dial test time of each characteristic data point and the remaining characteristic data points; and take the average of the differences between the dial test time of each characteristic data point and the adjacent data points as the network fluctuation value of the corresponding characteristic data point; Obtaining a concentrated value of the duration according to the dialing duration of all characteristic data points; taking the difference between the dialing duration of each characteristic data point and the concentrated value of the duration as the delay duration of the corresponding characteristic data point; The network delay index of each characteristic data point is obtained according to the network fluctuation value and the delay duration.

3. The method for fusion of multiple data of a device with a dialing function according to claim 1, characterized in that: The method for obtaining the description capability index of each feature dimension includes: Standardizing the network test data to obtain standard test data of each feature dimension of all feature data points; Select any one feature dimension as the analysis dimension, and obtain the standard difference value of the analysis dimension of each feature data point according to the difference between the standard test data of the analysis dimension of each feature data point and the mean of the standard test data of the analysis dimension of all feature data points; Obtain a comprehensive description value of the analysis dimension of each characteristic data point based on the network delay index and the standard difference value; and take the cumulative sum of the comprehensive description values ​​of the analysis dimension of all characteristic data points as the description capability index of the analysis dimension.

4. The method for fusion of multiple data of a device with a dialing function according to claim 3 is characterized in that: The method for obtaining the optimization priority index of each feature dimension includes: The absolute value of the correlation coefficient between the standard test data of each feature dimension of all feature data points and the network delay index is used as the delay correlation of each feature dimension; and the optimization priority index of each feature dimension is obtained based on the delay correlation and the description capability index.

5. The method for fusion of multiple data of a device with a dialing function according to claim 1, characterized in that: The method for obtaining low-dimensional data points comprises: The network test data of the same characteristic dimension of the characteristic data points are subjected to zero-mean processing to obtain the zero-mean data of each characteristic dimension of each characteristic data point; the product of the zero-mean data of each characteristic dimension of each characteristic data point and the optimization priority index is used as the optimized data of each characteristic dimension of each characteristic data point; an optimization matrix is ​​constructed according to the optimized data of the characteristic dimensions of all characteristic data points; the principal component analysis method is used to perform data dimensionality reduction on the optimization matrix to obtain a reduced dimensionality matrix; and low-dimensional data points are formed by all elements in each row of the reduced dimensionality matrix.

6. The method for fusion of multiple data of a device with a dialing function according to claim 4 is characterized in that: The correlation coefficient is the Pearson correlation coefficient.

7. The method for fusion of multiple data of a device with a dialing function according to claim 5, characterized in that: The number of rows of the y optimization matrix is ​​equal to the total number of characteristic dimensions of the characteristic data points, and the number of columns is equal to the total number of characteristic dimensions of the characteristic data points.

8. The method for fusion of multiple data of a device with a dialing function according to claim 2 is characterized in that: The method for obtaining a concentrated value of duration based on the dialing durations of all characteristic data points includes: taking the average of the dialing durations of all characteristic data points as the concentrated value of duration.

9. The method for fusion of multiple data of a device with a dialing function according to claim 3 is characterized in that: The method for standardizing the network test data is a range standardization method.

10. A device multi-data fusion system with dialing and testing function, characterized in that: The system includes: A data acquisition module is used to control the dial test tool to perform at least two dial tests on the dial test object, and obtain characteristic data points of each dial test and the dial test duration of the corresponding characteristic data, wherein the characteristic data points include network test data of different characteristic dimensions; The delay analysis module is used to obtain the network delay index of each characteristic data point based on the difference between the dialing time of each characteristic data point and the other characteristic data points; A description analysis module, used to obtain a description capability index for each feature dimension based on the difference between the network test data of the same feature dimension of all feature data points and the network delay index; A priority analysis module, used to obtain an optimization priority index for each feature dimension based on the correlation between the network test data of each feature dimension of the feature data point and the network delay index, and the description capability index; The data fusion module is used to perform data dimensionality reduction on the characteristic data points according to the optimization priority index to obtain low-dimensional data points; and perform data fusion on the network test data after the dimensionality reduction of the characteristic dimensions of the low-dimensional data points.

Images