A communication equipment fault prediction and diagnosis method and system based on large model

Through a large-model-based method, the neural network model is used to analyze and predict the operation data of communication equipment, which solves the problem of lack of early prediction of fault detection in the prior art, and realizes efficient diagnosis and processing of communication equipment failures.

CN118740598BActive Publication Date: 2024-11-15BEIJING TUODIE KAICHENG TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411230361.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-04
Publication Date
2024-11-15
Estimated Expiration
2044-09-04

AI Technical Summary

Technical Problem

The existing communication equipment fault detection process is mainly post-event inspection and detection, which lacks the ability to predict risks in advance, resulting in inefficient fault handling.

Method used

Using a large model-based method, by obtaining the operating data of communication equipment, establishing a neural network model, analyzing the amplitude of data changes in real time, adjusting the data acquisition frequency, and building a data table based on the preset module arrangement sequence, training the neural network to assist in fault parameter detection .

Benefits of technology

It realizes the advance prediction and diagnosis of communication equipment faults, significantly optimizes the fault handling efficiency, and improves the accuracy and efficiency of fault handling.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118740598B_ABST
    Figure CN118740598B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of fault detection, and specifically discloses a method and system for predicting and diagnosing faults of communication equipment based on a large model. The method comprises acquiring the operating data of each equipment module in real time based on a transmission channel; obtaining a data table by counting the normalized data of each equipment module at each moment according to a preset module arrangement order; when a fault parameter is detected, intercepting a data subtable before a preset time period, taking the data subtable as input and the fault parameter as output to construct a sample set, and training a neural network model based on the sample set; assisting the detection process of the fault parameter based on the neural network model; on the basis of traditional fault detection, the present invention records the operating data, establishes a neural network model between the operating data and the fault detection result, and in actual use, performs fault prediction with the help of the trained neural network model, which complements the existing fault detection scheme and greatly optimizes the fault handling efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of fault detection, and in particular to a method and system for predicting and diagnosing faults of communication equipment based on a large model. Background Art

[0002] Communication equipment is used to transmit data, voice, video and other information between two or more locations. They are the core components of modern communication networks and are widely used in various communication systems, including telephone networks, the Internet, radio and television, satellite communications, mobile communications, etc.; the importance of communication equipment is very high, so fault detection is required.

[0003] The fault detection process of existing communication equipment is mostly a post-event troubleshooting solution. Once a problem occurs, the built-in modules and algorithms are used to troubleshoot the problem, which has a very high accuracy rate, but does not provide advance warning. If a solution is provided that can predict risks in advance, the fault handling progress of communication equipment can be greatly optimized. Summary of the invention

[0004] The purpose of the present invention is to provide a communication equipment fault prediction and diagnosis method and system based on a large model to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A communication equipment fault prediction and diagnosis method based on a large model, the method comprising:

[0007] Acquire a device module with a transmission function in the communication device, and establish a transmission channel with the device module;

[0008] Based on the transmission channel, the operation data of each equipment module is acquired in real time; the operation data is a time-based sequence and is discrete data;

[0009] Normalizing the operation data, recording the data change range, and adjusting the data acquisition frequency of each equipment module according to the data change range;

[0010] According to the preset module arrangement order, the normalized data of each device module at each time is counted to obtain a data table;

[0011] When a fault parameter is detected, the data sub-table before the preset period is intercepted, the data sub-table is used as input, and the fault parameter is used as output to construct a sample set, and the neural network model is trained based on the sample set;

[0012] The detection process of fault parameters is assisted by neural network model.

[0013] Furthermore, the step of obtaining a device module with a transmission function in the communication device and establishing a transmission channel with the device module includes:

[0014] Obtain the registered component information of the communication equipment and query the equipment modules with transmission function;

[0015] Acquire the connection relationship between the equipment modules, and classify the equipment modules according to the connection relationship; the classification rule is that any equipment module of the same type is connected to at least one equipment module of all other equipment modules;

[0016] A device module with the largest amount of remaining resources is selected from among the same type of device modules, and a transmission channel is established with the device module.

[0017] Furthermore, the step of normalizing the operation data, recording the data variation range, and adjusting the data acquisition frequency of each device module according to the data variation range includes:

[0018] updating the maximum value and the minimum value in the operation data in real time, and converting each operation data based on the maximum value and the minimum value to obtain a value from zero to one;

[0019] Obtaining a preset number of converted values ​​from the tail of the running data, calculating the standard deviation of the obtained values, and using the standard deviation as the data variation range;

[0020] The data acquisition frequency of the equipment module corresponding to the operating data is adjusted according to the data change amplitude; the data acquisition frequency is proportional to the data change amplitude.

[0021] Furthermore, the step of obtaining a data table by counting the normalized data of each device module at each time according to the preset module arrangement order includes:

[0022] Read the preset module arrangement sequence, read the device modules in sequence based on the module arrangement sequence, and determine the corresponding relationship between the modules and each column;

[0023] Determine the time point according to the maximum reading frequency of the device module, and determine the corresponding relationship between each time point and each row according to the time sequence;

[0024] The normalized data of each device module at each time is read, the filling position of the data is determined according to the determined corresponding relationship, and the data is inserted into the corresponding filling position to obtain a data table.

[0025] Furthermore, the step of obtaining a data table by counting the normalized data of each device module at each time according to the preset module arrangement order also includes:

[0026] Calculate the correlation between any two columns of data;

[0027] For the missing data in each column of data, select another column of existing data with the largest correlation to fit the data;

[0028] The calculation process of correlation is:

[0029] For any two columns of data, extract the rows that have data in both columns of data, obtain the sub-column data corresponding to the two columns of data, calculate the distance between the sub-column data, and calculate the correlation between the two columns of data based on the similarity;

[0030] The specific calculation process is:

[0031] ; In the formula, is the relevance, is the preset correction factor, is the total number of elements in the sub-column data, and are the first individual data;

[0032] The process of data fitting is:

[0033] ; In the formula, The data in a column that needs to be fitted. is the corresponding data of another column with the greatest correlation, is the mean of the column data that needs to be fitted. is the mean of another column of data with the greatest correlation; yes The adjacent data.

[0034] Furthermore, the steps of the neural network model-assisted fault parameter detection process include:

[0035] Based on the current moment, the data before the preset period is intercepted in real time in the data table to obtain a data sub-table;

[0036] Input the data sub-table into the trained neural network model to obtain the predicted fault parameters;

[0037] The existing fault parameter detection process is assisted according to the predicted fault parameters; wherein the assistance process is to adjust the priority of each step in the existing fault parameter detection process.

[0038] Furthermore, a communication equipment fault prediction and diagnosis system based on a large model is provided, the system comprising:

[0039] A channel establishment module is used to obtain a device module with a transmission function in a communication device and establish a transmission channel with the device module;

[0040] An operation data acquisition module, used to acquire the operation data of each device module in real time based on the transmission channel; the operation data is a time-based sequence and is discrete data;

[0041] A frequency adjustment module, used to normalize the operation data, record the data change amplitude, and adjust the data acquisition frequency of each device module according to the data change amplitude;

[0042] Running a data statistics module to count the normalized data of each device module at each time according to a preset module arrangement order to obtain a data table;

[0043] A sample construction application module is used to intercept the data sub-table before the preset period when the fault parameter is detected, use the data sub-table as input and the fault parameter as output to construct a sample set, and train the neural network model based on the sample set;

[0044] The auxiliary detection module is used to assist the detection process of fault parameters based on the neural network model.

[0045] Furthermore, the channel establishment module includes:

[0046] A module query unit, used to obtain the registered component information of the communication equipment and query the equipment module with transmission function;

[0047] A module classification unit is used to obtain the connection relationship between the equipment modules and classify the equipment modules according to the connection relationship; the classification rule is that any equipment module of the same type is connected to at least one equipment module of all other equipment modules;

[0048] An execution unit is established to select a device module with the largest amount of remaining resources from similar device modules and establish a transmission channel with the device module.

[0049] Furthermore, the frequency adjustment module includes:

[0050] A data conversion unit, used for updating the maximum value and the minimum value in the operation data in real time, and converting each operation data based on the maximum value and the minimum value to obtain a value from zero to one;

[0051] an amplitude calculation unit, for acquiring a preset number of converted values ​​from the tail of the running data, calculating a standard deviation of the acquired values, and using the standard deviation as the data variation amplitude;

[0052] The adjustment execution unit is used to adjust the data acquisition frequency of the equipment module corresponding to the operating data according to the data change amplitude; the data acquisition frequency is proportional to the data change amplitude.

[0053] Furthermore, the operation data statistics module includes:

[0054] A column number determination unit, used to read a preset module arrangement sequence, read device modules in sequence based on the module arrangement sequence, and determine the corresponding relationship between the modules and each column;

[0055] A row number determination unit, used to determine a time point according to a maximum reading frequency of a device module, and determine a corresponding relationship between each time point and each row according to a time sequence;

[0056] The filling unit is used to read the normalized data of each device module at each moment, determine the filling position of the data according to the determined corresponding relationship, insert the data into the corresponding filling position, and obtain a data table.

[0057] Compared with the prior art, the beneficial effects of the present invention are: on the basis of traditional fault detection, the present invention records operating data and establishes a neural network model between the operating data and the fault detection results. In actual use, fault prediction is performed with the help of the trained neural network model, which complements the existing fault detection scheme and greatly optimizes the fault handling efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0059] Figure 1 This is a flowchart of the communication equipment fault prediction and diagnosis method based on a large model.

[0060] Figure 2 This is the structural diagram of the communication equipment fault prediction and diagnosis system based on the large model. DETAILED DESCRIPTION

[0061] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0062] Figure 1 The flowchart of the method for predicting and diagnosing a communication device fault based on a large model is shown in FIG. In an embodiment of the present invention, a method for predicting and diagnosing a communication device fault based on a large model includes:

[0063] Step S100: obtaining a device module with a transmission function in a communication device, and establishing a transmission channel with the device module;

[0064] The structure of the communication device is known, and its components are known. The various components in the communication device are queried to obtain a device module with a transmission function. Having a transmission function means that it can upload data. Then, a transmission channel is established between the device module with the transmission function to obtain the data generated by the device module during operation.

[0065] Step S200: acquiring the operation data of each device module in real time based on the transmission channel; the operation data is a time-based sequence and is discrete data;

[0066] With the help of the transmission channel, the platform executing the method can obtain the operating data of each device module; the operating data is the data of a certain device module at different times, which is discrete data and contains a time tag, that is, a time series.

[0067] Step S300: normalizing the operation data, recording the data variation range, and adjusting the data acquisition frequency of each device module according to the data variation range;

[0068] The operating data of different equipment modules are different. This difference is not only in numerical value, but also in data format. Therefore, the operating data needs to be standardized. The most important function of standardization is to unify the data format. In the process of standardizing the operating data, the data change amplitude is also recorded, and the data acquisition frequency of each equipment module is adjusted according to the data change amplitude. The higher the data change amplitude, the more unstable the data is, and the greater the data acquisition frequency is.

[0069] Step S400: Counting the normalized data of each device module at each time according to the preset module arrangement order to obtain a data table;

[0070] The normalized data of all equipment modules at different times are counted in a table to obtain a data table.

[0071] Step S500: when a fault parameter is detected, intercept the data sub-table before the preset period, use the data sub-table as input and the fault parameter as output to construct a sample set, and train the neural network model based on the sample set;

[0072] The fault detection process is actually carried out. In the existing architecture, there are many accurate and stable fault detection schemes. When a fault occurs, the fault parameters can be obtained at the first time. On this basis, the data of the previous period is obtained, that is, the data table before the preset time period is intercepted, which is called the data sub-table. The data sub-table is used as input and the fault parameters are used as output to construct a sample set. The neural network model is trained based on the sample set, thereby obtaining a fault prediction model based on a period of operating data.

[0073] Step S600: Assisting the detection process of fault parameters based on the neural network model;

[0074] After the neural network model is generated, when it is actually used, the current operating data is analyzed in real time with the help of the neural network model, so that the fault parameters can be predicted. The advantage of the prediction process is that it can make the existing fault detection process more efficient, because the existing fault detection process is a troubleshooting detection process. If the type of fault is predicted in advance, the corresponding fault can be checked first during troubleshooting, which is extremely efficient.

[0075] Regarding step S100, the step of obtaining a device module with a transmission function in the communication device and establishing a transmission channel with the device module includes:

[0076] Obtain the registered component information of the communication equipment and query the equipment modules with transmission function;

[0077] Acquire the connection relationship between the equipment modules, and classify the equipment modules according to the connection relationship; the classification rule is that any equipment module of the same type is connected to at least one equipment module of all other equipment modules;

[0078] A device module with the largest amount of remaining resources is selected from among the same type of device modules, and a transmission channel is established with the device module.

[0079] The above content refines the device module selection process. The simplest way is the global selection mode. This method obtains more data and the output of the trained neural network model is more accurate. However, the cost will also be very high, including the data acquisition stage and the data processing stage.

[0080] In order to reduce costs, the present application first classifies the device modules according to the connection relationship. The connection relationship refers to the data intercommunication between two device modules. For the same type of device modules, any device module in the same type of device modules is connected to at least one device module among all other device modules. After the classification is completed, the device module with the largest amount of remaining resources is selected from each type of device module, and a transmission channel is established with it to carry out data transmission.

[0081] For step S300, the step of normalizing the operation data, recording the data variation range, and adjusting the data acquisition frequency of each device module according to the data variation range includes:

[0082] updating the maximum value and the minimum value in the operation data in real time, and converting each operation data based on the maximum value and the minimum value to obtain a value from zero to one;

[0083] Obtaining a preset number of converted values ​​from the tail of the running data, calculating the standard deviation of the obtained values, and using the standard deviation as the data variation range;

[0084] The data acquisition frequency of the equipment module corresponding to the operating data is adjusted according to the data change amplitude; the data acquisition frequency is proportional to the data change amplitude.

[0085] For each equipment module, obtain the operating data, query and record the maximum and minimum values ​​in the operating data, calculate the difference between the maximum and minimum values, and then calculate the ratio of each operating data to the difference to get a value from zero to one. After conversion, all values ​​are formatted and counted without units.

[0086] Then, obtain a preset number of converted values ​​(also corresponding to a period of time) from the tail of the operating data forward and calculate the standard deviation. The larger the standard deviation, the greater the data change. At this time, it is necessary to increase the data acquisition frequency of the corresponding equipment module to obtain more data. The advantage of doing so is that it can reduce the data acquisition frequency of equipment modules with relatively stable data changes, thereby reducing costs.

[0087] For step S400, the step of obtaining a data table by counting the normalized data of each device module at each time according to the preset module arrangement order includes:

[0088] Read the preset module arrangement sequence, read the device modules in sequence based on the module arrangement sequence, and determine the corresponding relationship between the modules and each column;

[0089] Determine the time point according to the maximum reading frequency of the device module, and determine the corresponding relationship between each time point and each row according to the time sequence;

[0090] The normalized data of each device module at each time is read, the filling position of the data is determined according to the determined corresponding relationship, and the data is inserted into the corresponding filling position to obtain a data table.

[0091] The module arrangement order is preset. The preset module arrangement order is read as the column order, so that each column corresponds to each device module one by one. Then, the time point is determined according to the maximum reading frequency of the device module, and the most dense time point is determined. Then, the time points are arranged in time sequence as the row order. In this way, the position of any device module at any time can be determined, which is called the filling position. The data is inserted into the filling position to obtain a data table.

[0092] Furthermore, the step of obtaining a data table by counting the normalized data of each device module at each time according to the preset module arrangement order also includes:

[0093] Calculate the correlation between any two columns of data;

[0094] For the missing data in each column of data, select another column of existing data with the largest correlation to fit the data;

[0095] The calculation process of correlation is:

[0096] For any two columns of data, extract the rows that have data in both columns of data, obtain the sub-column data corresponding to the two columns of data, calculate the distance between the sub-column data, and calculate the correlation between the two columns of data based on the similarity;

[0097] The specific calculation process is:

[0098] ; In the formula, is the relevance, is the preset correction factor, is the total number of elements in the sub-column data, and are the first individual data;

[0099] The process of data fitting is:

[0100] ; In the formula, The data in a column that needs to be fitted. is the corresponding data of another column with the greatest correlation, is the mean of the column data that needs to be fitted. is the mean of another column of data with the greatest correlation; yes The adjacent data.

[0101] Since the data acquisition frequency of each device module is different, there may be missing data in the data table. For the missing data, data fitting is required.

[0102] For the correlation calculation process, due to the different data acquisition frequencies, the dimensions of the two columns of data are different. Only when there is data in both positions can the comparison calculation be performed. By extracting the rows with data in both columns of data, sub-column data can be obtained, and the distance of the sub-column data is calculated. The larger the distance, the smaller the correlation. The more rows with data in both columns of data, the closer the data acquisition frequency is, and the higher the correlation between the two.

[0103] Furthermore, there are two situations in the data fitting process. One is that the other most relevant column has a large enough correlation with the column to be fitted. In this case, the other column is used for fitting, and the ratio of the corresponding row data to the mean is calculated, and then multiplied by the mean of the current column to get the fitted data. If the correlation is small, the adjacent data can be used as the fitting data.

[0104] For step S500, the steps of the detection process of the fault parameters assisted by the neural network model include:

[0105] Based on the current moment, the data before the preset period is intercepted in real time in the data table to obtain a data sub-table;

[0106] Input the data sub-table into the trained neural network model to obtain the predicted fault parameters;

[0107] The existing fault parameter detection process is assisted according to the predicted fault parameters; wherein the assistance process is to adjust the priority of each step in the existing fault parameter detection process.

[0108] As time goes by, the current moment keeps changing. Based on the current moment, we can reverse a period of time, obtain the operating data and convert it to get a data sub-table. By inputting the data sub-table into the trained neural network model, we can get the predicted fault parameters. According to the predicted fault parameters, we can adjust the priority of each step in the existing fault parameter detection process to further improve efficiency and reduce costs.

[0109] Figure 2 : is a structural diagram of a communication equipment fault prediction and diagnosis system based on a large model. In an embodiment of the present invention, a communication equipment fault prediction and diagnosis system based on a large model, the system 10 includes:

[0110] The channel establishing module 11 is used to obtain a device module with a transmission function in the communication device and establish a transmission channel with the device module;

[0111] An operation data acquisition module 12 is used to acquire the operation data of each device module in real time based on the transmission channel; the operation data is a time-based sequence and is discrete data;

[0112] The frequency adjustment module 13 is used to normalize the operation data, record the data change amplitude, and adjust the data acquisition frequency of each device module according to the data change amplitude;

[0113] The data statistics module 14 is used to count the normalized data of each device module at each time according to the preset module arrangement order to obtain a data table;

[0114] The sample construction application module 15 is used to intercept the data sub-table before the preset period when the fault parameter is detected, use the data sub-table as input and the fault parameter as output to construct a sample set, and train the neural network model based on the sample set;

[0115] The auxiliary detection module 16 is used to assist the detection process of fault parameters based on the neural network model.

[0116] Furthermore, the channel establishing module 11 includes:

[0117] A module query unit, used to obtain the registered component information of the communication equipment and query the equipment module with transmission function;

[0118] A module classification unit is used to obtain the connection relationship between the equipment modules and classify the equipment modules according to the connection relationship; the classification rule is that any equipment module of the same type is connected to at least one equipment module of all other equipment modules;

[0119] An execution unit is established to select a device module with the largest amount of remaining resources from similar device modules and establish a transmission channel with the device module.

[0120] Specifically, the frequency adjustment module 13 includes:

[0121] A data conversion unit, used for updating the maximum value and the minimum value in the operation data in real time, and converting each operation data based on the maximum value and the minimum value to obtain a value from zero to one;

[0122] an amplitude calculation unit, for acquiring a preset number of converted values ​​from the tail of the running data, calculating a standard deviation of the acquired values, and using the standard deviation as the data variation amplitude;

[0123] The adjustment execution unit is used to adjust the data acquisition frequency of the equipment module corresponding to the operating data according to the data change amplitude; the data acquisition frequency is proportional to the data change amplitude.

[0124] Furthermore, the operation data statistics module 14 includes:

[0125] A column number determination unit, used to read a preset module arrangement sequence, read device modules in sequence based on the module arrangement sequence, and determine the corresponding relationship between the modules and each column;

[0126] A row number determination unit, used to determine a time point according to a maximum reading frequency of a device module, and determine a corresponding relationship between each time point and each row according to a time sequence;

[0127] The filling unit is used to read the normalized data of each device module at each moment, determine the filling position of the data according to the determined corresponding relationship, insert the data into the corresponding filling position, and obtain a data table.

[0128] 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 and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A communication equipment fault prediction and diagnosis method based on a large model, characterized in that: The method comprises: Acquire a device module with a transmission function in the communication device, and establish a transmission channel with the device module; Based on the transmission channel, the operation data of each equipment module is acquired in real time; the operation data is a time-based sequence and is discrete data; Normalizing the operation data, recording the data change range, and adjusting the data acquisition frequency of each equipment module according to the data change range; According to the preset module arrangement order, the normalized data of each equipment module at each time is counted to obtain a data table; When a fault parameter is detected, the data sub-table before the preset period is intercepted, the data sub-table is used as input, and the fault parameter is used as output to construct a sample set, and the neural network model is trained based on the sample set; The detection process of fault parameters assisted by neural network model; The step of obtaining a data table by counting the normalized data of each device module at each time according to the preset module arrangement order comprises: Read the preset module arrangement order, read the device modules in sequence based on the module arrangement order, and determine the corresponding relationship between them and each column; Determine the time point according to the maximum reading frequency of the device module, and determine the corresponding relationship between each time point and each row according to the time sequence; Read the normalized data of each device module at each time, determine the filling position of the data according to the determined corresponding relationship, insert the data into the corresponding filling position, and obtain a data table; The step of obtaining a data table by counting the normalized data of each device module at each time according to the preset module arrangement order further includes: Calculate the correlation between any two columns of data; For the missing data in each column of data, select another column of existing data with the largest correlation to fit the data; The calculation process of correlation is: For any two columns of data, extract the rows that have data in both columns of data, obtain the sub-column data corresponding to the two columns of data, calculate the distance between the sub-column data, and calculate the correlation between the two columns of data based on the similarity; The specific calculation process is: ; In the formula, is the relevance, is the preset correction factor, is the total number of elements in the sub-column data, and are the first individual data; The process of data fitting is: ; In the formula, The data in a column that needs to be fitted. is the corresponding data of another column with the greatest correlation, is the mean of the column data that needs to be fitted. is the mean of another column of data with the greatest correlation; yes Adjacent data of The steps of the neural network model-assisted fault parameter detection process include: Based on the current moment, the data before the preset period is intercepted in real time in the data table to obtain a data sub-table; Input the data sub-table into the trained neural network model to obtain the predicted fault parameters; The existing fault parameter detection process is assisted according to the predicted fault parameters; wherein the assistance process is to adjust the priority of each step in the existing fault parameter detection process.

2. The method for predicting and diagnosing communication equipment faults based on a large model according to claim 1, characterized in that: The step of obtaining a device module with a transmission function in the communication device and establishing a transmission channel with the device module comprises: Obtain the registered component information of the communication equipment and query the equipment modules with transmission function; Acquire the connection relationship between the equipment modules, and classify the equipment modules according to the connection relationship; the classification rule is that any equipment module of the same type is connected to at least one equipment module of all other equipment modules; A device module with the largest amount of remaining resources is selected from among similar device modules, and a transmission channel is established with the device module.

3. The method for predicting and diagnosing communication equipment faults based on a large model according to claim 1, characterized in that: The step of normalizing the operation data, recording the data variation range, and adjusting the data acquisition frequency of each device module according to the data variation range includes: updating the maximum value and the minimum value in the operation data in real time, and converting each operation data based on the maximum value and the minimum value to obtain a value from zero to one; Obtaining a preset number of converted values ​​from the tail of the running data, calculating the standard deviation of the obtained values, and using the standard deviation as the data variation range; The data acquisition frequency of the equipment module corresponding to the operating data is adjusted according to the data change amplitude; the data acquisition frequency is proportional to the data change amplitude.

4. A communication equipment fault prediction and diagnosis system based on a large model, characterized in that: The system is used to implement the communication equipment fault prediction and diagnosis method based on a large model as described in any one of claims 1 to 3, and the system includes: A channel establishment module is used to obtain a device module with a transmission function in a communication device and establish a transmission channel with the device module; An operation data acquisition module, used to acquire the operation data of each device module in real time based on the transmission channel; the operation data is a time-based sequence and is discrete data; A frequency adjustment module, used to normalize the operation data, record the data change amplitude, and adjust the data acquisition frequency of each device module according to the data change amplitude; Running a data statistics module to count the normalized data of each device module at each time according to a preset module arrangement order to obtain a data table; A sample construction application module is used to intercept the data sub-table before the preset period when the fault parameter is detected, use the data sub-table as input and the fault parameter as output to construct a sample set, and train the neural network model based on the sample set; The auxiliary detection module is used to assist the detection process of fault parameters based on the neural network model.

5. The communication equipment fault prediction and diagnosis system based on large model according to claim 4 is characterized in that: The channel establishment module comprises: A module query unit, used to obtain the registered component information of the communication equipment and query the equipment module with transmission function; A module classification unit is used to obtain the connection relationship between the equipment modules and classify the equipment modules according to the connection relationship; the classification rule is that any equipment module of the same type is connected to at least one equipment module of all other equipment modules; An execution unit is established to select a device module with the largest amount of remaining resources from similar device modules and establish a transmission channel with the device module.

6. The communication equipment fault prediction and diagnosis system based on large model according to claim 4 is characterized in that: The frequency adjustment module comprises: A data conversion unit, used for updating the maximum value and the minimum value in the operation data in real time, and converting each operation data based on the maximum value and the minimum value to obtain a value from zero to one; an amplitude calculation unit, for acquiring a preset number of converted values ​​from the tail of the running data, calculating a standard deviation of the acquired values, and using the standard deviation as the data variation amplitude; The adjustment execution unit is used to adjust the data acquisition frequency of the equipment module corresponding to the operating data according to the data change amplitude; the data acquisition frequency is proportional to the data change amplitude.

7. The communication equipment fault prediction and diagnosis system based on a large model according to claim 4 is characterized in that: The operation data statistics module includes: A column number determination unit, used to read a preset module arrangement sequence, read device modules in sequence based on the module arrangement sequence, and determine the corresponding relationship between the modules and each column; A row number determination unit, used to determine a time point according to a maximum reading frequency of a device module, and determine a corresponding relationship between each time point and each row according to a time sequence; The filling unit is used to read the normalized data of each device module at each moment, determine the filling position of the data according to the determined corresponding relationship, insert the data into the corresponding filling position, and obtain a data table.