Abnormal data analysis method and device in home broadband

By cleaning and smoothly processing home broadband data, an abnormal data analysis model is built, and real-time data abnormality detection is automated, which solves the problem of inefficiency of traditional methods and realizes efficient abnormal data analysis.

CN117093942BActive Publication Date: 2024-11-19TANGREN COMM TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311072339.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-08-24
Publication Date
2024-11-19
Estimated Expiration
2043-08-24

AI Technical Summary

Technical Problem

Traditional home broadband anomaly data analysis methods are inefficient and cannot effectively process large-scale data, resulting in reduced analysis efficiency.

Method used

Collect historical data from home broadband for cleaning and stationary enhancement processing, build an abnormal data analysis model, configure the model order using statistical feature indicators, and perform real-time data anomaly analysis through fitting the model to realize automated abnormality detection.

Benefits of technology

It improves data quality and analysis efficiency, reduces the need for manual processing, and improves the efficiency of abnormal data analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117093942B_ABST
    Figure CN117093942B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of artificial intelligence technology, and discloses a method and device for analyzing abnormal data in home broadband, the method comprising: collecting historical broadband data of home broadband; performing data cleaning and stationarity enhancement processing on the historical broadband data to obtain stationary data; using a preset abnormal data analysis algorithm to construct an abnormal data analysis model of the home broadband; using the stationary number to perform order configuration and parameter configuration on the abnormal data analysis model to obtain a fitting model of the abnormal data analysis model; acquiring real-time data of the home broadband, and performing abnormal data analysis on the real-time data using the fitting model to obtain abnormal data of the real-time data. The present invention can improve the efficiency of abnormal data analysis in home broadband.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and device for analyzing abnormal data in home broadband. Background Art

[0002] With the development of the Internet of Things, more and more devices and home products need to be connected to broadband networks. The normal operation of home broadband ensures smooth communication and networking between these devices, such as smart TVs, smart speakers, smart appliances, etc., which enables families to enjoy the convenience brought by digital life and smart homes. At the same time, through broadband, users can access the Internet anytime and anywhere, browse the web, watch videos, use social media, etc. The normal operation of broadband ensures that users can carry out online work, study and entertainment activities normally.

[0003] Traditional analysis methods often require manual data processing, cleaning and analysis, which is very time-consuming and error-prone, especially when dealing with large-scale data. The inefficiency of manual processing limits the amount and accuracy of data that analysts can process. With the continuous growth of the number of broadband users and network usage, the scale of data generated by home broadband is also increasing rapidly. Traditional analysis methods may not be able to effectively handle such a large amount of data, resulting in reduced analysis efficiency. Therefore, how to improve the analysis of abnormal data in home broadband has become an urgent problem to be solved. Summary of the invention

[0004] The present invention provides a method and device for analyzing abnormal data in home broadband, the main purpose of which is to solve the problem of low efficiency in analyzing abnormal data in home broadband.

[0005] To achieve the above object, the present invention provides a method for analyzing abnormal data in home broadband, comprising:

[0006] Collect historical broadband data for household broadband;

[0007] Performing data cleaning on the historical broadband data to obtain cleaned data of the historical broadband data, and performing stationarity enhancement processing on the cleaned data to obtain stationary data of the cleaned data;

[0008] The abnormal data analysis model of the home broadband is constructed using a preset abnormal data analysis algorithm, wherein the preset abnormal data analysis algorithm is:

[0009]

[0010] Among them, y t is the predicted value of the abnormal data analysis model of the home broadband at time t, is the autoregressive coefficient of the abnormal data analysis model, θq is the moving average coefficient, ε t is a white noise sequence at time t, t is the time mark of the predicted value, q is the moving order, and p is the autoregressive order;

[0011] Generate a statistical characteristic index of the stationary data, and use the statistical characteristic index to configure the order of the abnormal data analysis model to obtain an initialization analysis model of the abnormal data analysis model;

[0012] Using the stationary data, parameter configuration is performed on the initialization analysis model to obtain a fitting model of the initialization analysis model;

[0013] The real-time data of the home broadband is acquired, and abnormal data analysis is performed on the real-time data using the fitting model to obtain abnormal data of the real-time data.

[0014] Optionally, the collecting of historical broadband data of home broadband includes:

[0015] Acquire a usage log of the home broadband, perform keyword extraction on the usage log, and obtain log keywords of the usage log;

[0016] The historical broadband data of the home broadband is generated according to the log keyword.

[0017] Optionally, the performing data cleaning on the historical broadband data to obtain cleaned data of the historical broadband data includes:

[0018] Eliminating invalid data from the historical broadband data to obtain valid data of the historical broadband data;

[0019] Performing outlier correction on the valid data to obtain corrected data of the valid data;

[0020] Performing noise reduction processing on the corrected data to obtain de-redundant data of the corrected data;

[0021] The de-redundant data is standardized to obtain standard data of the de-redundant data, and the standard data is determined to be cleaned data of the historical broadband data.

[0022] Optionally, performing stationarity enhancement processing on the cleaned data to obtain stationary data of the cleaned data includes:

[0023] generating a time series of the home broadband according to the cleaned data;

[0024] Performing a stationarity test on the data points of the time series one by one, and determining a target data point according to the test result of the stationarity test;

[0025] Performing differential processing on the target data point to obtain a differential data point of the target data point;

[0026] A stationary sequence of the time series is generated according to the differential data points, and the stationary data of the cleaned data is determined using the stationary sequence.

[0027] Optionally, performing a stationarity test on the data points of the time series one by one includes:

[0028] The following test algorithm is used to test the stability of the data points of the time series one by one:

[0029]

[0030] Among them, r a represents the output value at time a, α i is the weight coefficient of the lagged data point, r t-i is the input value representing a data point i periods earlier than time a, w a is a random error term, i is the period identifier of the lag period of the data point, I is the total number of periods of the lag period of the data point, and a is the time identifier of the data point.

[0031] Optionally, the generating of the statistical characteristic index of the stationary data includes:

[0032] The correlation index of the stationary data is generated using a preset correlation algorithm, wherein the preset correlation algorithm is:

[0033]

[0034] Among them, ρ k is the stationary data y b and the stationary data y b-k The correlation index, Cov(y b ,y b-k ) is the stationary data y b and the stationary data y b-k The covariance of y b is the stationary number at time b, Var(y b ) is the stationary data y b The variance of , k is the time difference, which is used to measure the time interval between two said stationary data, and b is the time mark of said stationary data;

[0035] The stationary data is used to generate a partial autocorrelation index of the stationary data, and the correlation index and the partial autocorrelation index are collected as a statistical characteristic index of the stationary data.

[0036] Optionally, the using the statistical characteristic index to configure the order of the abnormal data analysis model to obtain an initialization analysis model of the abnormal data analysis model includes:

[0037] Generate a correlation index graph of the abnormal data analysis model according to the correlation index in the statistical feature index;

[0038] Generate a partial autocorrelation index graph of the abnormal data analysis model according to the partial autocorrelation index in the statistical characteristic index;

[0039] The order of the abnormal data analysis model is configured by using the truncated region of the correlation index graph and the truncated region of the partial autocorrelation index graph to obtain an initialization analysis model of the abnormal data analysis model.

[0040] Optionally, the using the stationary data to configure parameters of the initialization analysis model to obtain a fitting model of the initialization analysis model includes:

[0041] Generating a likelihood function of the initialized analysis model according to the stationary data, and calculating a function maximum value of the likelihood function;

[0042] Parameters of the initialization analysis model are configured according to the maximum value of the function to obtain a fitting model of the initialization analysis model.

[0043] Optionally, performing abnormal data analysis on the real-time data using the fitting model to obtain abnormal data of the real-time data includes:

[0044] Inputting the real-time data into the fitting model to obtain a data output of the real-time data;

[0045] The data output is compared with the real-time data, and abnormal data analysis is performed on the real-time data according to a comparison result of the data comparison and a preset deviation threshold to obtain abnormal data of the real-time data.

[0046] In order to solve the above problems, the present invention also provides a device for analyzing abnormal data in home broadband, the device comprising:

[0047] A data collection module, used to collect historical broadband data of home broadband;

[0048] A data stability enhancement module, used for performing data cleaning on the historical broadband data to obtain cleaned data of the historical broadband data, and performing stability enhancement processing on the cleaned data to obtain stable data of the cleaned data;

[0049] The model building module is used to build the abnormal data analysis model of the home broadband using a preset abnormal data analysis algorithm, wherein the preset abnormal data analysis algorithm is:

[0050]

[0051] Among them, y t is the predicted value of the abnormal data analysis model of the home broadband at time t, is the autoregressive coefficient of the abnormal data analysis model, θ q is the moving average coefficient, ε t is a white noise sequence at time t, t is the time mark of the predicted value, q is the moving order, and p is the autoregressive order;

[0052] A model order configuration module, used to generate statistical characteristic indicators of the stationary data, and use the statistical characteristic indicators to perform order configuration on the abnormal data analysis model to obtain an initialization analysis model of the abnormal data analysis model;

[0053] A model parameter configuration module, used to configure parameters of the initialization analysis model using the stationary data to obtain a fitting model of the initialization analysis model;

[0054] The abnormal data analysis module is used to obtain the real-time data of the home broadband, and perform abnormal data analysis on the real-time data using the fitting model to obtain abnormal data of the real-time data.

[0055] The embodiment of the present invention can obtain a large amount of original data by collecting historical data of home broadband, providing a basis for subsequent abnormal data analysis, thus avoiding the complexity and time-consuming of manual data collection, cleaning the collected historical broadband data, removing invalid data, processing missing values ​​and abnormal values, etc. This step can reduce errors and interference in subsequent analysis, improve the quality and accuracy of data, perform stationarity enhancement processing on the cleaned data, make the data stationary by removing trend and seasonal factors, and process stationary data more easily than non-stationary data, thereby improving analysis efficiency, configure the order of the abnormal data analysis model by using the statistical characteristic indicators of the generated stationary data, further optimize the analysis model, and use the fitting model to perform abnormal data analysis on real-time data, so that abnormal detection can be automatically performed on real-time data, reducing the need for manual processing and improving efficiency. Therefore, the present invention proposes a method and device for analyzing abnormal data in home broadband, which can solve the problem of low efficiency of abnormal data analysis in home broadband. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1A schematic diagram of a flow chart of a method for analyzing abnormal data in home broadband provided by an embodiment of the present invention;

[0057] Figure 2 A schematic diagram of a data cleaning process according to an embodiment of the present invention;

[0058] Figure 3 A schematic diagram of a process flow for data stationarity enhancement provided by an embodiment of the present invention;

[0059] Figure 4 A functional module diagram of an abnormal data analysis device in a home broadband provided by an embodiment of the present invention;

[0060] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0061] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0062] The embodiment of the present application provides a method for analyzing abnormal data in home broadband. The execution subject of the method for analyzing abnormal data in home broadband includes but is not limited to at least one of the electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the method for analyzing abnormal data in home broadband can be executed by software or hardware installed in a terminal device or a server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and big data and artificial intelligence platforms.

[0063] Reference Figure 1 FIG. 1 is a flow chart of a method for analyzing abnormal data in home broadband provided by an embodiment of the present invention. In this embodiment, the method for analyzing abnormal data in home broadband includes:

[0064] S1. Collect historical broadband data of household broadband.

[0065] In an embodiment of the present invention, the historical broadband data of the home broadband refers to the data record of the home broadband network connection in the past period of time, which contains information such as performance indicators and usage of the home broadband network. The historical broadband data may include the following: bandwidth speed, delay, jitter, traffic usage and connection status, wherein the bandwidth speed refers to the upload and download speed of the home broadband network, usually expressed in the amount of data transmitted per second, the delay refers to the time required for data to be transmitted from the sending end to the receiving end, usually in milliseconds, the jitter refers to the instability of data transmission, that is, the time difference between data packets, the traffic usage refers to the data usage of the home broadband network, including the total amount of data uploaded and downloaded, usually in bytes or gigabytes, and the connection status refers to the connection status of the home broadband network, such as connection time, disconnection time and reconnection time.

[0066] In the embodiment of the present invention, the collecting of historical broadband data of home broadband includes:

[0067] Acquire a usage log of the home broadband, perform keyword extraction on the usage log, and obtain log keywords of the usage log;

[0068] The historical broadband data of the home broadband is generated according to the log keyword.

[0069] In detail, the acquisition of home broadband usage logs can be performed by following the steps below: log in to the home router management interface, which can usually be accessed by entering the router's IP address in a browser, select enable logging, set the log level and storage location, and export log files. These log files may contain data records such as home broadband connection information, usage duration, upload and download speeds, etc.

[0070] In detail, the keyword extraction of the usage log refers to extracting words or phrases related to a specific topic or field from the log text of the usage log, that is, extracting keywords related to home broadband usage; the generating of historical broadband data of the home broadband based on the log keywords refers to converting the extracted log keywords into the required historical broadband data, which may include information such as connection duration, bandwidth speed, and usage date.

[0071] S2. Performing data cleaning on the historical broadband data to obtain cleaned data of the historical broadband data, and performing stationarity enhancement processing on the cleaned data to obtain stationary data of the cleaned data.

[0072] In an embodiment of the present invention, the data cleaning of the historical broadband data is to clean and organize the original data to make it more suitable for subsequent analysis and processing, which will improve data quality, reduce the impact on subsequent steps, and help discover patterns and trends in the data.

[0073] In the embodiment of the present invention, Figure 2 As shown, the data cleaning of the historical broadband data to obtain the cleaned data of the historical broadband data includes:

[0074] S21, removing invalid data from the historical broadband data to obtain valid data of the historical broadband data;

[0075] S22, performing outlier correction on the valid data to obtain corrected data of the valid data;

[0076] S23, performing noise reduction processing on the corrected data to obtain de-redundant data of the corrected data;

[0077] S24, performing standardization processing on the de-redundant data to obtain standard data of the de-redundant data, and determining that the standard data is cleaned data of the historical broadband data.

[0078] In detail, the invalid data elimination of the historical broadband data refers to eliminating invalid data in the historical broadband data, such as duplicate values ​​and erroneous data, so as to ensure the accuracy and completeness of the data; the outlier correction of the valid data refers to considering using reasonable methods to process the valid data with outliers, and statistical techniques, such as standard deviation or median, can be used to identify and adjust the outliers; the denoising of the corrected data refers to removing noise or interference in the corrected data to reduce the impact on subsequent analysis; the standardization of the de-redundant data helps to compare and analyze the relationship between different features, and z-score standardization or minimum-maximum standardization can be used.

[0079] In detail, the data cleaning of the historical broadband data is to clean and organize the original data to make it more suitable for subsequent analysis and processing, which will improve data quality, reduce the impact on subsequent steps, and help discover patterns and trends in the data.

[0080] In the embodiment of the present invention, Figure 3 As shown, the step of performing stationarity enhancement processing on the cleaned data to obtain stationary data of the cleaned data includes:

[0081] S31, generating a time series of the home broadband according to the cleaned data;

[0082] S32, performing a stationarity test on the data points of the time series one by one, and determining a target data point according to the test result of the stationarity test;

[0083] S33, performing differential processing on the target data point to obtain a differential data point of the target data point;

[0084] S34. Generate a stationary sequence of the time series according to the differential data points, and use the stationary sequence to determine the stationary data of the cleaned data.

[0085] In detail, generating the time series of the home broadband according to the cleaned data refers to arranging the data in chronological order to generate the time series of the home broadband.

[0086] In detail, the stationarity test is performed on the data points of the time series one by one in order to evaluate the stationarity of the time series, and which data points need to be differentially processed are determined according to the results of the stationarity test; the differential processing is a commonly used stationarity enhancement method, which reduces the trend and seasonality of the time series by calculating the difference between the current data point and the previous data point, and the differential processing can be performed using first-order differences, second-order differences or seasonal differences.

[0087] In detail, the stationary series is generated by difference-processed data points and has stronger stationarity and stability.

[0088] In detail, the performing a stationarity test on the data points of the time series one by one includes:

[0089] The following test algorithm is used to test the stability of the data points of the time series one by one:

[0090]

[0091] Among them, r a represents the output value at time a, α i is the weight coefficient of the lagged data point, r t-i is the input value representing a data point i periods earlier than time a, w a is a random error term, i is the period identifier of the lag period of the data point, I is the total number of periods of the lag period of the data point, and a is the time identifier of the data point.

[0092] In detail, according to the time series that needs to be tested for stationarity, the total number of periods of the lag number is determined, where the lag number refers to the difference in periods before the current time point. For example, if the first-order difference is selected, the total number of periods of the lag number is 1; the selection of the weight coefficient can be determined according to specific needs and methods. Common choices include exponential weighted average, constant weighted average or adaptive weighted average. The purpose of the weight coefficient is to weight the lagged data points to reflect their impact on the current data points.

[0093] S3. Constructing an abnormal data analysis model for the home broadband using a preset abnormal data analysis algorithm.

[0094] In the embodiment of the present invention, the preset abnormal data analysis algorithm is:

[0095]

[0096] Among them, y t is the predicted value of the abnormal data analysis model of the home broadband at time t, is the autoregressive coefficient of the abnormal data analysis model, θ q is the moving average coefficient, ε t is a white noise sequence at time t, t is the time mark of the predicted value, q is the moving order, and p is the autoregressive order.

[0097] In detail, the autoregressive coefficient of the abnormal data analysis model represents the weight between the current prediction value and the past p prediction values, and a larger coefficient indicates that the past prediction values ​​have a greater influence on the current prediction value; the moving average coefficient represents the weight between the current prediction value and the past q white noise items, and a larger coefficient indicates that the past white noise has a greater influence on the current prediction value; the moving order represents the historical order of white noise in the abnormal data analysis model, that is, the influence range of the past q white noises; the autoregressive order represents the historical order of the predicted value in the abnormal data analysis model, that is, the influence range of the past p predicted values.

[0098] S4. Generate statistical characteristic indicators of the stationary data, and use the statistical characteristic indicators to configure the order of the abnormal data analysis model to obtain an initialized analysis model of the abnormal data analysis model.

[0099] In an embodiment of the present invention, the generating of the statistical characteristic index of the stationary data includes:

[0100] The correlation index of the stationary data is generated using a preset correlation algorithm, wherein the preset correlation algorithm is:

[0101]

[0102] Among them, ρ k is the stationary data y b and the stationary data y b-k The correlation index, Cov(y b ,y b-k ) is the stationary data y b and the stationary data y b-k The covariance of y b is the stationary number at time b, Var(y b ) is the stationary data y b The variance of , k is the time difference, which is used to measure the time interval between two said stationary data, and b is the time mark of said stationary data;

[0103] The stationary data is used to generate a partial autocorrelation index of the stationary data, and the correlation index and the partial autocorrelation index are collected as a statistical characteristic index of the stationary data.

[0104] In detail, the correlation index measures the correlation between the stationary data and itself at different time delays; the partial autocorrelation index takes into account the situation where other influencing factors are eliminated first when calculating the correlation.

[0105] In the embodiment of the present invention, the step of configuring the order of the abnormal data analysis model by using the statistical characteristic index to obtain an initialization analysis model of the abnormal data analysis model includes:

[0106] Generate a correlation index graph of the abnormal data analysis model according to the correlation index in the statistical feature index;

[0107] Generate a partial autocorrelation index graph of the abnormal data analysis model according to the partial autocorrelation index in the statistical characteristic index;

[0108] The order of the abnormal data analysis model is configured by using the truncated region of the correlation index graph and the truncated region of the partial autocorrelation index graph to obtain an initialization analysis model of the abnormal data analysis model.

[0109] In detail, the generating of the correlation index graph of the abnormal data analysis model according to the correlation index in the statistical characteristic index refers to converting the correlation index ρ k The values ​​at different time differences k are plotted into a graph, with the horizontal axis representing the time difference k and the vertical axis representing the correlation index ρ k The correlation index diagram shows the correlation between the stationary data.

[0110] In detail, the partial autocorrelation index graph of the abnormal data analysis model is generated according to the partial autocorrelation index in the statistical characteristic index. kThe values ​​at different time differences k are plotted into a graph, with the horizontal axis representing the time difference k and the vertical axis representing the partial autocorrelation index g k The partial autocorrelation index diagram reflects the direct correlation between the stationary data and its past data.

[0111] In detail, the use of the truncated area of ​​the correlation index graph and the truncated area of ​​the partial autocorrelation index graph to configure the order of the abnormal data analysis model refers to observing the graphical features of the correlation index graph and the partial autocorrelation index graph, paying special attention to the correlation and partial autocorrelation values ​​in the truncated area, and judging the applicable model order based on the truncated features of the graph.

[0112] Furthermore, the order of the abnormal data analysis model can be preliminarily judged based on the truncated area of ​​the correlation index graph and the truncated area of ​​the partial autocorrelation index graph. This is because when the order of the correlation index graph of the abnormal data analysis model is p, the value after the truncated area will quickly decay to 0. Therefore, when it is observed on the correlation index graph that the values ​​after the truncated area are all close to 0, it can be preliminarily judged that the order of the abnormal data analysis model is the delay number of the truncation point; the order can be determined based on the truncated area of ​​the partial autocorrelation index graph. When the order of the partial autocorrelation index graph of the abnormal data analysis model is q, the value after the truncated area will also quickly decay to 0. Therefore, when it is observed on the partial autocorrelation index graph that the values ​​after the truncated area are all close to 0, it can be preliminarily judged that the order of the abnormal data analysis model is the delay number of the truncation point.

[0113] S5. Utilize the stationary data to configure parameters of the initialization analysis model to obtain a fitting model of the initialization analysis model.

[0114] In an embodiment of the present invention, the step of configuring parameters of the initialization analysis model using the stationary data to obtain a fitting model of the initialization analysis model includes:

[0115] Generating a likelihood function of the initialized analysis model according to the stationary data, and calculating a function maximum value of the likelihood function;

[0116] Parameters of the initialization analysis model are configured according to the maximum value of the function to obtain a fitting model of the initialization analysis model.

[0117] In detail, the likelihood function is a function used to evaluate the degree of fit between the model parameters and the actual observed data. By maximizing the likelihood function, the optimal parameter combination can be selected; an optimization algorithm is used to search in the parameter space of the likelihood function to find the parameter combination that maximizes the likelihood function. These parameters are the optimal parameters for fitting the model, wherein the optimization algorithm is: gradient descent method or Newton method.

[0118] In detail, configuring the parameters of the initialization analysis model according to the maximum value of the function refers to applying the found optimal parameters to the initialization analysis model and updating the parameters of the model, so that the initialization analysis model becomes a fitting model whose parameters are optimized.

[0119] In detail, the parameters of the fitting model have been optimized and configured to better fit the actual observed data and provide prediction capabilities for future data. The fitting model can be used to further analyze and process abnormal data, as well as predict trends and change patterns of future data.

[0120] S6. Acquire the real-time data of the home broadband, and use the fitting model to perform abnormal data analysis on the real-time data to obtain abnormal data of the real-time data.

[0121] In an embodiment of the present invention, the real-time data of the home broadband may be obtained by monitoring the network transmission status of the home broadband, and the real-time data includes indicators such as upload speed, download speed, and delay.

[0122] In the embodiment of the present invention, the using the fitting model to perform abnormal data analysis on the real-time data to obtain abnormal data of the real-time data includes:

[0123] Inputting the real-time data into the fitting model to obtain a data output of the real-time data;

[0124] The data output is compared with the real-time data, and abnormal data analysis is performed on the real-time data according to a comparison result of the data comparison and a preset deviation threshold to obtain abnormal data of the real-time data.

[0125] In detail, inputting the real-time data into the fitting model to obtain the data output of the real-time data means taking the real-time data as input and using the model to perform calculations and predictions. The fitting model can predict and estimate the real-time data based on the fitting results of historical data.

[0126] In detail, the abnormal data analysis is performed on the real-time data according to the comparison result of the data comparison and the preset deviation threshold value. Obtaining the abnormal data of the real-time data means comparing the data output with the real-time data, and judging according to the preset deviation threshold value. If the difference between the data output and the real-time data exceeds the threshold value, that is, exceeds the normal range, the corresponding real-time data can be regarded as abnormal data.

[0127] The embodiment of the present invention can obtain a large amount of original data by collecting historical data of home broadband, providing a basis for subsequent abnormal data analysis, thus avoiding the complexity and time-consuming of manual data collection, cleaning the collected historical broadband data, removing invalid data, processing missing values ​​and abnormal values, etc. This step can reduce errors and interference in subsequent analysis, improve the quality and accuracy of data, perform stationarity enhancement processing on the cleaned data, make the data stationary by removing trend and seasonal factors, and process stationary data more easily than non-stationary data, thereby improving analysis efficiency, configure the order of the abnormal data analysis model by using the statistical characteristic indicators of the generated stationary data, further optimize the analysis model, and use the fitting model to perform abnormal data analysis on real-time data, so that abnormal detection can be automatically performed on real-time data, reducing the need for manual processing and improving efficiency. Therefore, the present invention proposes an abnormal data analysis method in home broadband, which can solve the problem of low efficiency of abnormal data analysis in home broadband.

[0128] like Figure 4 , which is a functional module diagram of an abnormal data analysis device in home broadband provided by an embodiment of the present invention.

[0129] The abnormal data analysis device 100 in home broadband of the present invention can be installed in an electronic device. According to the functions to be implemented, the abnormal data analysis device 100 in home broadband can include a data acquisition module 101, a data stationarity enhancement module 102, a model construction module 103, a model order configuration module 104, a model parameter configuration module 105 and an abnormal data analysis module 106. The module of the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0130] In this embodiment, the functions of each module / unit are as follows:

[0131] The data collection module 101 is used to collect historical broadband data of home broadband;

[0132] The data stability enhancement module 102 is used to perform data cleaning on the historical broadband data to obtain cleaned data of the historical broadband data, and perform stability enhancement processing on the cleaned data to obtain stable data of the cleaned data;

[0133] The model building module 103 is used to build the abnormal data analysis model of the home broadband using a preset abnormal data analysis algorithm, wherein the preset abnormal data analysis algorithm is:

[0134]

[0135] Among them, y t is the predicted value of the abnormal data analysis model of the home broadband at time t, is the autoregressive coefficient of the abnormal data analysis model, θ q is the moving average coefficient, ε t is a white noise sequence at time t, t is the time mark of the predicted value, q is the moving order, and p is the autoregressive order;

[0136] The model order configuration module 104 is used to generate statistical characteristic indicators of the stationary data, and use the statistical characteristic indicators to perform order configuration on the abnormal data analysis model to obtain an initialization analysis model of the abnormal data analysis model;

[0137] The model parameter configuration module 105 is used to configure the parameters of the initialization analysis model using the stationary data to obtain a fitting model of the initialization analysis model;

[0138] The abnormal data analysis module 106 is used to obtain the real-time data of the home broadband, and perform abnormal data analysis on the real-time data using the fitting model to obtain abnormal data of the real-time data.

[0139] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

Claims

1. A method for analyzing abnormal data in home broadband. It is characterized in that The method comprises: Collect historical broadband data for home broadband; Performing data cleaning on the historical broadband data to obtain cleaned data of the historical broadband data, and performing stationarity enhancement processing on the cleaned data to obtain stationary data of the cleaned data; The abnormal data analysis model of the home broadband is constructed using a preset abnormal data analysis algorithm, wherein the preset abnormal data analysis algorithm is: Among them, y t is the predicted value of the abnormal data analysis model of the home broadband at time t, is the autoregressive coefficient of the abnormal data analysis model, θ q is the moving average coefficient, ε t is a white noise sequence at time t, t is the time mark of the predicted value, q is the moving order, and p is the autoregressive order; The correlation index of the stationary data is generated using a preset correlation algorithm, wherein the preset correlation algorithm is: Among them, ρ k is the stationary data y b and the stationary data y b-k The correlation index, Cov(y b ,y b-k ) is the stationary data y b and the stationary data y b-k The covariance of y b is the stationary number at time b, Var(y b ) is the stationary data y b The variance of the above data is , k is the time difference, which is used to measure the time interval between two stationary data, b is the time mark of the stationary data, the stationary data is used to generate the partial autocorrelation index of the stationary data, the correlation index and the partial autocorrelation index are collected as the statistical characteristic index of the stationary data, the correlation index graph of the abnormal data analysis model is generated according to the correlation index in the statistical characteristic index, the partial autocorrelation index graph of the abnormal data analysis model is generated according to the partial autocorrelation index in the statistical characteristic index, the abnormal data analysis model is configured in order using the truncated area of ​​the correlation index graph and the truncated area of ​​the partial autocorrelation index graph, and the initialization analysis model of the abnormal data analysis model is obtained; Using the stationary data, parameter configuration is performed on the initialization analysis model to obtain a fitting model of the initialization analysis model; The real-time data of the home broadband is acquired, and abnormal data analysis is performed on the real-time data using the fitting model to obtain abnormal data of the real-time data.

2. The method for analyzing abnormal data in home broadband according to claim 1, It is characterized in that The collection of historical broadband data of household broadband includes: Acquire a usage log of the home broadband, perform keyword extraction on the usage log, and obtain log keywords of the usage log; The historical broadband data of the home broadband is generated according to the log keyword.

3. The method for analyzing abnormal data in home broadband according to claim 1, It is characterized in that The step of performing data cleaning on the historical broadband data to obtain cleaned data of the historical broadband data includes: Eliminating invalid data from the historical broadband data to obtain valid data of the historical broadband data; Performing outlier correction on the valid data to obtain corrected data of the valid data; Performing noise reduction processing on the corrected data to obtain de-redundant data of the corrected data; The de-redundant data is standardized to obtain standard data of the de-redundant data, and the standard data is determined to be cleaned data of the historical broadband data.

4. The method for analyzing abnormal data in home broadband according to claim 1, It is characterized in that The step of performing stationarity enhancement processing on the cleaned data to obtain stationary data of the cleaned data includes: generating a time series of the home broadband according to the cleaned data; Performing a stationarity test on the data points of the time series one by one, and determining a target data point according to the test result of the stationarity test; Performing differential processing on the target data point to obtain a differential data point of the target data point; A stationary sequence of the time series is generated according to the differential data points, and the stationary data of the cleaned data is determined using the stationary sequence.

5. The method for analyzing abnormal data in home broadband according to claim 4, It is characterized in that The step of performing a stationarity test on the data points of the time series one by one comprises: The following test algorithm is used to test the stability of the data points of the time series one by one: Among them, r a represents the output value at time a, α i is the weight coefficient of the lagged data point, r t-i is the input value representing a data point i periods earlier than time a, w a is a random error term, i is the period identifier of the lag period of the data point, I is the total number of periods of the lag period of the data point, and a is the time identifier of the data point.

6. The method for analyzing abnormal data in home broadband according to claim 1, It is characterized in that The step of configuring parameters of the initialization analysis model using the stationary data to obtain a fitting model of the initialization analysis model includes: Generating a likelihood function of the initialized analysis model according to the stationary data, and calculating a function maximum value of the likelihood function; Parameters of the initialization analysis model are configured according to the maximum value of the function to obtain a fitting model of the initialization analysis model.

7. The method for analyzing abnormal data in home broadband according to any one of claims 1 to 6, It is characterized in that The using the fitting model to perform abnormal data analysis on the real-time data to obtain abnormal data of the real-time data includes: Inputting the real-time data into the fitting model to obtain a data output of the real-time data; The data output is compared with the real-time data, and abnormal data analysis is performed on the real-time data according to a comparison result of the data comparison and a preset deviation threshold to obtain abnormal data of the real-time data.

8. A device for analyzing abnormal data in home broadband, It is characterized in that The device comprises: A data collection module, used to collect historical broadband data of home broadband; A data stability enhancement module, used for performing data cleaning on the historical broadband data to obtain cleaned data of the historical broadband data, and performing stability enhancement processing on the cleaned data to obtain stable data of the cleaned data; The model building module is used to build the abnormal data analysis model of the home broadband using a preset abnormal data analysis algorithm, wherein the preset abnormal data analysis algorithm is: Among them, y t is the predicted value of the abnormal data analysis model of the home broadband at time t, is the autoregressive coefficient of the abnormal data analysis model, θ q is the moving average coefficient, ε t is a white noise sequence at time t, t is the time mark of the predicted value, q is the moving order, and p is the autoregressive order; The model order configuration module is used to generate the correlation index of the stationary data using a preset correlation algorithm, wherein the preset correlation algorithm is: Among them, ρ k is the stationary data y b and the stationary data y b-k The correlation index, Cov(y b ,y b-k ) is the stationary data y b and the stationary data y b-k The covariance of y b is the stationary number at time b, Var(y b ) is the stationary data y b The variance of the above data is , k is the time difference, which is used to measure the time interval between two stationary data, b is the time mark of the stationary data, the stationary data is used to generate the partial autocorrelation index of the stationary data, the correlation index and the partial autocorrelation index are collected as the statistical characteristic index of the stationary data, the correlation index graph of the abnormal data analysis model is generated according to the correlation index in the statistical characteristic index, the partial autocorrelation index graph of the abnormal data analysis model is generated according to the partial autocorrelation index in the statistical characteristic index, the abnormal data analysis model is configured in order using the truncated area of ​​the correlation index graph and the truncated area of ​​the partial autocorrelation index graph, and the initialization analysis model of the abnormal data analysis model is obtained; A model parameter configuration module, used to configure the parameters of the initialization analysis model using the stationary data to obtain a fitting model of the initialization analysis model; The abnormal data analysis module is used to obtain the real-time data of the home broadband, and perform abnormal data analysis on the real-time data using the fitting model to obtain abnormal data of the real-time data.