Fault analysis method and device of FTTR equipment, equipment and medium

Abstract

The invention provides a fault analysis method and device of FTTR equipment, equipment and a medium, and relates to the technical field of artificial intelligence. The method comprises the following steps: firstly, acquiring a plurality of pieces of fault related data obtained by performing data acquisition on target FTTR equipment; secondly, semantic mining is conducted on the multiple pieces of fault related data, and target fault semantic representation is formed; and then, based on the target fault semantic representation, performing fault analysis on the target FTTR equipment, and outputting fault prediction data corresponding to the target FTTR equipment. Based on the above content, the problem that it is difficult to effectively analyze the LAN port fault in the prior art can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to a fault analysis method and apparatus, equipment and medium for FTTR equipment. Background Art

[0002] FTTR (Fiber to the Room) equipment is a network access technology that lays optical fiber directly to the user's room, with the aim of improving network bandwidth and stability. FTTR equipment usually involves hardware such as fiber optic modems (ONUs) and wireless routers. In FTTR networks, LAN ports (Local Area Network ports) are key interfaces for connecting local devices (such as computers, TVs, printers, etc.). If the FTTR equipment causes damage to the LAN port, a series of effects will occur. Therefore, it is necessary to analyze or predict damage or failure of the LAN port so that corresponding maintenance can be carried out in a timely manner to ensure the effective operation of the equipment. However, in the prior art, corresponding maintenance is generally carried out after a LAN port fails. Therefore, there is a problem of difficulty in effectively analyzing LAN port failures. Summary of the invention

[0003] In view of this, the purpose of the present application is to provide a fault analysis method and apparatus, equipment and medium for FTTR equipment, so as to improve the problem in the prior art that it is difficult to effectively analyze LAN port faults.

[0004] To achieve the above objectives, this application adopts the following technical solutions: A fault analysis method for FTTR equipment, comprising: Acquire multiple fault-related data obtained by collecting data from a target FTTR device, wherein the fault-related data refers to data that contributes to the failure of a LAN port of the target FTTR device, and there is at least one time series data among the multiple fault-related data; Performing semantic mining on the multiple fault-related data to form a target fault semantic representation, wherein, in the process of semantic mining, for each of the time series data, the data fluctuation semantics in the time series data are mined from at least two time change directions; Based on the target fault semantic representation, a fault analysis is performed on the target FTTR device, and fault prediction data corresponding to the target FTTR device is output, wherein the fault prediction data is used to reflect whether a LAN port of the target FTTR device will fail.

[0005] In a preferred embodiment of the present application, in the fault analysis method of the FTTR device, the step of performing semantic mining on the plurality of fault-related data to form a semantic representation of the target fault includes: For each time series data in the plurality of fault-related data, mining the data fluctuation semantics in the time series data from at least two time change directions, and outputting the corresponding local fault semantic representation; When there is at least one fault-related data that does not belong to time series data among the plurality of fault-related data, for each fault-related data among the at least one fault-related data, semantic mining is performed on the fault-related data, and a corresponding local fault semantic representation is output; The local fault semantic representation corresponding to each of the fault-related data is fused to form a target fault semantic representation, wherein the target fault semantic representation is used to characterize the global semantic information possessed by the multiple fault-related data.

[0006] In a preferred embodiment of the present application, in the fault analysis method of the FTTR device, the step of mining the data fluctuation semantics in the time series data from at least two time change directions for each time series data in the multiple fault-related data and outputting the corresponding local fault semantic representation comprises: Loading the time series data into corresponding semantic mining branches, wherein each time series data corresponds to a semantic mining branch; For each sequence sub-data in the time series data, semantic space mapping is performed on the sequence sub-data to form a corresponding first fault semantic representation, and taking the sequence sub-data as a starting point, through the target window carried by the semantic mining branch, in the time series data according to a first time change direction, a first sequence data segment corresponding to the sequence sub-data is intercepted, and taking the sequence sub-data as a starting point, through the target window, in the time series data according to a second time change direction, a second sequence data segment corresponding to the sequence sub-data is intercepted, wherein the size of the target window is formed during training as a network parameter of the semantic mining branch; Performing semantic space mapping on the first sequence data segments and the second sequence data segments respectively to form corresponding second fault semantic representations and third fault semantic representations; The first fault semantic representation, the second fault semantic representation and the third fault semantic representation corresponding to each sequence sub-data are fused to form a corresponding local fault semantic representation.

[0007] In a preferred embodiment of the present application, in the fault analysis method of the FTTR device, the step of performing semantic space mapping on the first sequence data segments and the second sequence data segments respectively to form corresponding second fault semantic representations and third fault semantic representations includes: Performing a discreteness calculation on the first sequence data segments to obtain a first discreteness, and performing a discreteness calculation on the second sequence data segments to obtain a second discreteness, wherein the time series data is time numerical series data, including temperature, voltage or network traffic; Semantic space mapping is performed on the first discreteness and the second discreteness respectively to form a corresponding second fault semantic representation and a third fault semantic representation, wherein the semantic space mapping includes word embedding processing.

[0008] In a preferred embodiment of the present application, in the fault analysis method of the FTTR device, the step of fusing the first fault semantic representation, the second fault semantic representation and the third fault semantic representation corresponding to each sequence sub-data to form a corresponding local fault semantic representation includes: For each sequence sub-data, concatenate the first fault semantic representation, the second fault semantic representation, and the third fault semantic representation corresponding to the sequence sub-data to form a concatenated fault semantic representation corresponding to the sequence sub-data; The spliced ​​fault semantic representation corresponding to each of the sequence sub-data is spliced ​​to form a spliced ​​fault semantic representation corresponding to the time series data, and the spliced ​​fault semantic representation corresponding to the time series data is convolved and pooled respectively to realize different semantic feature extraction to form corresponding convolution fault semantic representation and pooling fault semantic representation, wherein the spliced ​​fault semantic representation corresponding to each of the sequence sub-data, the convolution fault semantic representation and the pooling fault semantic representation have the same size; For each sequence sub-data, based on the spliced ​​fault semantic representation corresponding to the sequence sub-data, the convolution fault semantic representation and the pooled fault semantic representation are respectively focused and mined to form a corresponding first focused semantic representation and a second focused semantic representation, and the first focused semantic representation and the second focused semantic representation are superimposed to form a corresponding fused focused semantic representation; The fused focused semantic representation corresponding to each of the sequence sub-data is spliced ​​to form a spliced ​​focused semantic representation, and a local fault semantic representation is determined based on the spliced ​​focused semantic representation.

[0009] In a preferred embodiment of the present application, in the fault analysis method of the FTTR device, when there is at least one fault-related data that does not belong to time series data among the plurality of fault-related data, for each of the at least one fault-related data, the step of performing semantic mining on the fault-related data and outputting a corresponding local fault semantic representation includes: When there is at least one fault-related data that does not belong to the time series data among the plurality of fault-related data, for each of the at least one fault-related data, semantic space mapping is performed on the fault-related data to obtain a corresponding fault mapping semantic representation; When the number of obtained fault mapping semantic representations is greater than a preset number, clustering the fault mapping semantic representations to form at least one corresponding semantic representation cluster; For each of the fault mapping semantic representations, focused mining is performed on the fault mapping semantic representation according to the cluster center of the semantic representation cluster cluster corresponding to the fault mapping semantic representation to form a local fault semantic representation corresponding to the fault mapping semantic representation.

[0010] In a preferred embodiment of the present application, in the above-mentioned method for analyzing the failure of the FTTR device, the method for analyzing the failure of the FTTR device further comprises: Using the semantic mining unit included in the candidate fault analysis network, semantic mining is performed on multiple training fault related data to form a training fault semantic representation, wherein the candidate fault analysis network belongs to a neural network, and the multiple training fault related data include at least one training time series data. In the process of semantic mining, for each of the training time series data, the data fluctuation semantics in the training time series data is mined from at least two time change directions; Utilizing the semantic analysis unit included in the candidate fault analysis network, performing fault analysis based on the training fault semantic representation, and outputting training fault prediction data; According to the training loss index between the training fault prediction data and the fault label data corresponding to the plurality of training fault-related data, the network parameters of the candidate fault analysis network are updated to form a target fault analysis network.

[0011] The present application also provides a fault analysis device for FTTR equipment, comprising: A fault-related data acquisition module is used to acquire a plurality of fault-related data obtained by collecting data from a target FTTR device, wherein the fault-related data refers to data that contributes to the failure of a LAN port of the target FTTR device, and there is at least one time series data among the plurality of fault-related data; A data semantic mining module is used to perform semantic mining on the multiple fault-related data to form a target fault semantic representation, wherein, in the process of semantic mining, for each of the time series data, the data fluctuation semantics in the time series data is mined from at least two time change directions; A fault analysis module is used to perform fault analysis on the target FTTR device based on the target fault semantic representation, and output fault prediction data corresponding to the target FTTR device, wherein the fault prediction data is used to reflect whether a LAN port of the target FTTR device will fail.

[0012] Based on the above, the present application also provides an electronic device, including: Memory for storing computer programs; The processor connected to the memory is used to execute the computer program stored in the memory to implement the above-mentioned fault analysis method of the FTTR equipment.

[0013] On the basis of the above, the present application further provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is run, each step of the above-mentioned FTTR device fault analysis method is executed.

[0014] The fault analysis method and apparatus, equipment and medium of the FTTR equipment provided in the present application firstly obtain multiple fault-related data obtained by data collection of the target FTTR equipment; secondly, semantic mining is performed on the multiple fault-related data to form a target fault semantic representation; then, based on the target fault semantic representation, fault analysis is performed on the target FTTR equipment to output fault prediction data corresponding to the target FTTR equipment. Based on the above content, on the one hand, since semantic mining is performed on multiple fault-related data, the semantic information carried by the formed target fault semantic representation can be richer; on the other hand, since in the process of semantic mining, for each time series data, the data fluctuation semantics is mined from at least two time change directions, the representation reliability of the mined semantic information is higher, thereby ensuring the reliability of the fault prediction data output based on the target fault semantic representation, and then improving the problem of difficulty in effectively analyzing LAN port faults in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings.

[0016] Figure 1 A structural block diagram of an electronic device provided in an embodiment of the present application.

[0017] Figure 2A schematic flow chart of a fault analysis method for FTTR equipment provided in an embodiment of the present application.

[0018] Figure 3 A schematic diagram of the fusion of semantic representations of each sequence sub-data provided in an embodiment of the present application.

[0019] Figure 4 A block diagram of a fault analysis device for FTTR equipment provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0021] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0022] like Figure 1 As shown, an embodiment of the present application provides an electronic device, wherein the electronic device may include a memory, a processor, and a fault analysis device of an FTTR device.

[0023] In detail, the memory and the processor are electrically connected directly or indirectly to achieve data transmission or interaction. For example, the memory and the processor can be electrically connected through one or more communication buses or signal lines. The fault analysis device of the FTTR device includes at least one software function module stored in the memory in the form of software or firmware. The processor is used to execute an executable computer program stored in the memory, for example, the software function module and computer program included in the fault analysis device of the FTTR device, so as to implement the fault analysis method of the FTTR device provided in the embodiment of the present application.

[0024] Optionally, the memory may be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), etc.

[0025] Furthermore, the processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0026] Understandably, Figure 1 The structure shown is for illustration only, and the electronic device may also include Figure 1 More or fewer components as shown, or with Figure 1 The different configurations shown, for example, may also include a communication unit for exchanging information with other devices.

[0027] Combination Figure 2 The present application also provides a method for analyzing the failure of an FTTR device applicable to the above electronic device. The method steps defined in the process related to the method for analyzing the failure of the FTTR device can be implemented by the electronic device. Figure 2 The specific process shown is explained in detail.

[0028] Step S110, acquiring a plurality of fault-related data obtained by collecting data from a target FTTR device.

[0029] In an embodiment of the present application, the electronic device can obtain multiple fault-related data obtained by collecting data from the target FTTR device. Among them, the fault-related data refers to data that may contribute to the failure of the LAN port of the target FTTR device (that is, some factors that may cause the failure of the LAN port), and there is at least one time series data in the multiple fault-related data, for example, some of the fault-related data in the multiple fault-related data belong to time series data, or all of the fault-related data in the multiple fault-related data belong to time series data. In addition, time series data refers to a set of data collected in chronological order, such as data at time point 1, data at time point 2, data at time point 3, etc. In addition, the specific application scenario of the target FTTR device can be FTTR-B (Business FTTR), that is, a full-optical networking solution for business enterprises, a Wi-Fi solution specially designed for enterprises for business enterprise scenarios. Using the full-optical networking solution of optical fiber access + optoelectronic composite cable + Wi-Fi6, Wi-Fi is covered in every corner of the enterprise, providing network services for the networking work of enterprise employees, such as live broadcasts, online meetings and other activities.

[0030] Step S120: performing semantic mining on the plurality of fault-related data to form a semantic representation of a target fault.

[0031] In an embodiment of the present application, after obtaining the multiple fault-related data, the electronic device can perform semantic mining on the multiple fault-related data to form a target fault semantic representation. Wherein, in the process of semantic mining, for each of the time series data, the data fluctuation semantics in the time series data is mined from at least two time change directions. In addition, semantic mining refers to mining potential semantic information from the multiple fault-related data and representing it in the form of a vector or matrix, that is, obtaining the corresponding semantic representation.

[0032] Step S130: Based on the target fault semantic representation, perform fault analysis on the target FTTR device, and output fault prediction data corresponding to the target FTTR device.

[0033] In an embodiment of the present application, after obtaining the target fault semantic representation, the electronic device can perform a fault analysis on the target FTTR device based on the target fault semantic representation, and output the fault prediction data corresponding to the target FTTR device. The fault prediction data is used to reflect whether the LAN port of the target FTTR device will fail. Exemplarily, the target fault semantic representation can be fully connected to obtain a corresponding fully connected fault semantic representation, and then the fully connected fault semantic representation can be processed by a classification function (such as softmax, etc.) to obtain a corresponding probability distribution, such as (a, b), where a can refer to the probability that the LAN port of the target FTTR device fails, and b can refer to the probability that the LAN port of the target FTTR device will not fail. Then, the type with a larger probability value can be determined as the corresponding fault prediction data. If a is greater than b, a fault will occur.

[0034] Based on the above content, on the one hand, due to the semantic mining of multiple fault-related data, the semantic information carried by the formed target fault semantic representation can be richer. On the other hand, in the process of semantic mining, for each time series data, the data fluctuation semantics are mined from at least two time change directions, so that the representation reliability of the mined semantic information is higher, thereby ensuring the reliability of the fault prediction data output based on the target fault semantic representation, and then improving the problem of difficulty in effectively analyzing LAN port faults in the prior art.

[0035] It should be noted that for step S120, the specific method of performing semantic mining on the plurality of fault-related data is not limited and can be selected accordingly according to actual needs.

[0036] For example, in an alternative implementation, in order to improve the efficiency of semantic mining, the multiple fault-related data can be separately mapped into semantic spaces (such as word embedding), and then the semantic space mapping results corresponding to the respective fault-related data can be spliced. Then, convolution, pooling, activation, attention and other processing can be performed to obtain the corresponding target fault semantic information.

[0037] For example, in another alternative implementation, in order to improve the accuracy of semantic mining so that the obtained target fault semantic information can represent more semantic information, the above-mentioned step S120 can further include step S121, step S122 and step S123, the specific contents of which are described as follows.

[0038] Step S121 , for each time series data in the plurality of fault-related data, data fluctuation semantics in the time series data are mined from at least two time change directions, and a corresponding local fault semantic representation is output.

[0039] In an embodiment of the present application, for each time series data in the multiple fault-related data, the data fluctuation semantics in the time series data can be mined from at least two time change directions, and the corresponding local fault semantic representation can be output. It should be noted that for time series data, the fluctuation of data generally plays an important role, at least for equipment failure, it has an important characterization role, so corresponding semantic mining can be performed.

[0040] Step S122, when there is at least one fault-related data that does not belong to time series data in the multiple fault-related data, semantic mining is performed on each fault-related data in the at least one fault-related data, and a corresponding local fault semantic representation is output.

[0041] In an embodiment of the present application, when there is at least one fault-related data that does not belong to time series data among the multiple fault-related data, semantic mining is performed on each of the at least one fault-related data, and the corresponding local fault semantic representation is output, that is, the semantic information of the fault-related data itself is mined.

[0042] Step S123: fusing the local fault semantic representation corresponding to each of the fault-related data to form a target fault semantic representation.

[0043] In an embodiment of the present application, after obtaining the local fault semantic representation corresponding to each of the fault-related data (including the local fault semantic representation corresponding to each time series data), the local fault semantic representation corresponding to each of the fault-related data can be fused to form a target fault semantic representation. The target fault semantic representation is used to characterize the global semantic information possessed by the multiple fault-related data. Exemplarily, the local fault semantic representation corresponding to each of the fault-related data can be spliced ​​to obtain the corresponding spliced ​​semantic representation, and then the spliced ​​semantic representation can be convolved, pooled, and activated to obtain the target fault semantic information that can characterize the global semantic information of the multiple fault-related data.

[0044] It can be understood that in the above-mentioned embodiment, the implementation method of step S121 is not limited, that is, the specific method of mining the data fluctuation semantics in time series data is not limited. For example, in an alternative embodiment, in order to fully mine the data fluctuation semantic information and ensure the semantic richness of the obtained local fault semantic representation, the above-mentioned step S121 can further include the following steps S121a, step S121b, step S121c and step S121d.

[0045] Step S121a, loading the time series data into the corresponding semantic mining branch.

[0046] In an embodiment of the present application, the time series data can be loaded into a corresponding semantic mining branch. Each of the time series data corresponds to a semantic mining branch, and the semantic mining branch is part of a corresponding neural network model, such as a part of the network structure of the target fault analysis network described later, for semantic mining.

[0047] Step S121b, for each sequence sub-data in the time series data, perform semantic space mapping on the sequence sub-data to form a corresponding first fault semantic representation, and take the sequence sub-data as the starting point, through the target window carried by the semantic mining branch, in the time series data according to the first time change direction, cut out the first sequence data segment corresponding to the sequence sub-data, and, take the sequence sub-data as the starting point, through the target window, in the time series data according to the second time change direction, cut out the second sequence data segment corresponding to the sequence sub-data.

[0048] In an embodiment of the present application, for each sequence sub-data (such as data at a time point) in the time series data, a semantic space mapping (such as word embedding processing) is performed on the sequence sub-data to form a corresponding first fault semantic representation, and the sequence sub-data is used as a starting point, and the target window carried by the semantic mining branch is used to intercept the first sequence data segment corresponding to the sequence sub-data in the time series data according to the first time change direction (such as the direction from early to late time), and the sequence sub-data is used as a starting point, and the target window is used to intercept the second sequence data segment corresponding to the sequence sub-data in the time series data according to the second time change direction (such as the direction from late to early time). The size of the target window is formed during training as a network parameter of the semantic mining branch. For example, when the size of the target window is equal to 4, it means that the number of sequence sub-data included in the first sequence data segment is less than or equal to 4, and the number of sequence sub-data included in the second sequence data segment is less than or equal to 4. For example, for the first sequence sub-data, the first sequence data segment includes the first sequence sub-data, the second sequence sub-data, the third sequence sub-data and the fourth sequence sub-data, and the second sequence data segment includes the first sequence sub-data. For the second sequence sub-data, the first sequence data segment includes the second sequence sub-data, the third sequence sub-data, the fourth sequence sub-data and the fifth sequence sub-data, and the second sequence data segment includes the first sequence sub-data and the second sequence sub-data. For the fourth sequence sub-data, the first sequence data segment includes the first sequence sub-data, the second sequence sub-data, the third sequence sub-data and the fourth sequence sub-data, and the second sequence data segment includes the fourth sequence sub-data, the fifth sequence sub-data, the sixth sequence sub-data and the seventh sequence sub-data. The same is true for other sequence data segments, which are not given examples one by one.

[0049] Step S121c: perform semantic space mapping on the first sequence data segments and the second sequence data segments respectively to form corresponding second fault semantic representation and third fault semantic representation.

[0050] In an embodiment of the present application, after obtaining the first sequence data segments and the second sequence data segments, semantic space mapping can be performed on the first sequence data segments and the second sequence data segments respectively to form corresponding second fault semantic representations and third fault semantic representations.

[0051] Step S121d: merge the first fault semantic representation, the second fault semantic representation and the third fault semantic representation corresponding to each sequence sub-data to form a corresponding local fault semantic representation.

[0052] In an embodiment of the present application, after obtaining the first fault semantic representation, the second fault semantic representation and the third fault semantic representation corresponding to each sequence sub-data, the first fault semantic representation, the second fault semantic representation and the third fault semantic representation corresponding to each sequence sub-data can be fused to form a corresponding local fault semantic representation. In this way, in the local fault semantic representation, it carries not only the semantic information of the sequence sub-data itself, but also the semantic information of the fluctuation situation with the previous sequence sub-data, and also the semantic information of the fluctuation situation with the subsequent sequence sub-data, and therefore has better semantic representation ability. In addition, since the target window is formed during training, an adapted target window can be determined for different time series data, thereby achieving reliable capture of potential semantic information in fluctuation situations.

[0053] It can be understood that, in the above-mentioned implementation, the implementation of step S121c is not limited, that is, the specific manner of performing semantic space mapping on the first sequence data segments and the second sequence data segments is not limited. For example, in an alternative implementation, in order to reliably capture the potential semantic information in the corresponding fluctuation situation, the above-mentioned step S121c may further include the following contents: First, the first sequence data segment can be subjected to a discreteness calculation to obtain a first discreteness, and the second sequence data segment can be subjected to a discreteness calculation to obtain a second discreteness, wherein the time series data belongs to time numerical series data, including temperature, voltage or network flow, such as the temperature at time point 1, the temperature at time point 2, the temperature at time point 3, etc.; in addition, the discreteness calculation method can be, in the first step, calculating the mean of each sequence sub-data in the sequence data segment, then calculating the absolute difference between each sequence sub-data and the mean, and finally calculating the mean of each absolute difference to obtain the corresponding discreteness; Secondly, semantic space mapping is performed on the first discreteness and the second discreteness respectively to form corresponding second fault semantic representation and third fault semantic representation, wherein the semantic space mapping includes word embedding processing, that is, word embedding processing is performed on the corresponding discreteness to obtain corresponding semantic representation, and the word embedding processing can be implemented by a corresponding word embedding model.

[0054] It can be understood that in the above-mentioned implementation, the implementation of step S121d is not limited, that is, the specific method of forming the corresponding local fault semantic representation is not limited. For example, in an alternative implementation, in order to achieve full integration of multiple semantic representations of each sequence sub-data, the above-mentioned step S121d may further include the following contents (combined with Figure 3 shown): First, for each sequence sub-data, the first fault semantic representation, the second fault semantic representation and the third fault semantic representation corresponding to the sequence sub-data are concatenated to form a concatenated fault semantic representation corresponding to the sequence sub-data; Secondly, the spliced ​​fault semantic representation corresponding to each of the sequence sub-data is spliced ​​to form the spliced ​​fault semantic representation corresponding to the time series data, and the spliced ​​fault semantic representation corresponding to the time series data is convolved and pooled respectively (which can be implemented by corresponding convolutional networks and pooling networks respectively) to achieve different semantic feature extraction, forming corresponding convolutional fault semantic representation and pooled fault semantic representation, wherein the spliced ​​fault semantic representation corresponding to each of the sequence sub-data, the convolutional fault semantic representation and the pooled fault semantic representation have the same size, that is, the important semantic features are extracted respectively by convolution and pooling, and the semantic representation is compressed, so that the size is the same as the spliced ​​fault semantic representation corresponding to each of the sequence sub-data, so that the important semantic features can be captured and the subsequent processing can be facilitated; Then, for each sequence sub-data, based on the spliced ​​fault semantic representation corresponding to the sequence sub-data, the convolution fault semantic representation and the pooled fault semantic representation are focused and mined respectively to form a corresponding first focused semantic representation and a second focused semantic representation, and the first focused semantic representation and the second focused semantic representation are superimposed to form a corresponding fused focused semantic representation; that is, on the one hand, semantic information associated with the convolution fault semantic representation can be mined from the spliced ​​fault semantic representation corresponding to the sequence sub-data, and on the other hand, semantic information associated with the pooled fault semantic representation can be mined from the spliced ​​fault semantic representation corresponding to the sequence sub-data. In this way, since the convolution fault semantic representation and the pooled fault semantic representation are obtained by convolving and pooling the spliced ​​fault semantic representation that can represent the global semantic information, the convolution fault semantic representation and the pooled fault semantic representation can also focus on different semantic features when representing the global semantic information. Therefore, focusing and mining respectively can realize the mining of different associated semantic features, that is, mining more associated semantic features, so that the semantics of the fused focused semantic representation formed by superposition is richer; Secondly, the fused focused semantic representation corresponding to each of the sequence sub-data can be spliced ​​to form a spliced ​​focused semantic representation, and the local fault semantic representation can be determined based on the spliced ​​focused semantic representation; exemplarily, the spliced ​​focused semantic representation can be directly used as the corresponding local fault semantic representation, or the spliced ​​focused semantic representation can be convolved, pooled and activated to obtain the corresponding local fault semantic representation.

[0055] It can be understood that, in the above-mentioned implementation, the implementation method of step S122 is not limited, that is, the specific method of semantic mining of fault-related data is not limited. For example, in an alternative implementation, in order to make the semantic representation ability of the mined local fault semantic representation better, the above-mentioned step S122 may further include the following contents: First, when there is at least one fault-related data that does not belong to time series data among the multiple fault-related data, for each fault-related data in the at least one fault-related data, semantic space mapping is performed on the fault-related data to obtain a corresponding fault mapping semantic representation. Exemplarily, word embedding processing can be performed through a word embedding model to obtain a corresponding semantic space mapping. In addition, the fault-related data may refer to the number of times electrostatic protection is not performed when performing port connection operations, such as not wearing an anti-static wristband. The reason is that electrostatic discharge usually occurs when connecting or operating a device. When the LAN port of the device contacts a static power source, it may cause hardware damage to a certain extent; or, the fault-related data may also refer to electromagnetic interference conditions, such as whether it is in a strong magnetic environment, such as being placed near high-power electrical appliances (such as microwave ovens, air conditioners, etc.). The reason is that frequent electromagnetic interference will affect the operation of the LAN port and may cause damage; Secondly, when the number of obtained fault mapping semantic representations is greater than a preset number (such as 3, 5, 7, 10, etc.), clustering processing is performed on each fault mapping semantic representation (the specific clustering algorithm is not limited, such as KNN (K-Nearest Neighbors) and other clustering algorithms), forming at least one corresponding semantic representation cluster (that is, obtaining at least one cluster center); Then, for each of the fault mapping semantic representations, focused mining is performed on the fault mapping semantic representation according to the cluster center of the semantic representation cluster cluster corresponding to the fault mapping semantic representation to form a local fault semantic representation corresponding to the fault mapping semantic representation.

[0056] Based on this, by clustering first and then focusing mining, the accuracy of focused mining can be guaranteed, and the problem of low accuracy of mined semantic representations caused by focusing mining based on irrelevant semantic representations can be avoided. In addition, the specific implementation method of focused mining can adopt cross attention.

[0057] It should also be noted that, in order to ensure the reliable execution of the above steps S120 and S130, they can be implemented through corresponding neural network models, such as through a target fault analysis network. Based on this, the fault analysis method of the FTTR device can also include the step of training to form the target fault analysis network, which is as follows: First, a semantic mining unit included in the candidate fault analysis network can be used to perform semantic mining on multiple training fault related data to form a training fault semantic representation, wherein the candidate fault analysis network belongs to a neural network, and the multiple training fault related data include at least one training time series data. In the process of semantic mining, for each of the training time series data, the data fluctuation semantics in the training time series data is mined from at least two time change directions, and the relevant explanation of step S120 can be referred to in the previous text; Secondly, the semantic analysis unit included in the candidate fault analysis network can be used to perform fault analysis based on the training fault semantic representation, and output training fault prediction data, and reference can be made to the relevant explanation of step S130 above; Then, based on the training loss index (such as cross entropy loss) between the training fault prediction data and the fault label data corresponding to the multiple training fault-related data (i.e., the labeled information representing whether a fault will occur, such as manual labeling or identification using other neural network models), the network parameters of the candidate fault analysis network can be updated to form a target fault analysis network. For example, the network parameters can be updated in a direction that reduces the training loss index until the training loss index converges, thereby forming a target fault analysis network.

[0058] Combination Figure 4 The present application also provides a fault analysis device for a FTTR device applicable to the above electronic device. The fault analysis device for the FTTR device may include a fault-related data acquisition module, a data semantic mining module, and a fault analysis module.

[0059] In detail, the fault-related data acquisition module can be used to acquire multiple fault-related data obtained by collecting data from the target FTTR device, wherein the fault-related data refers to data that may contribute to the failure of the LAN port of the target FTTR device, and there is at least one time series data in the multiple fault-related data. In the embodiment of the present application, the fault-related data acquisition module can be used to execute Figure 2 As shown in step S110, the relevant contents of the fault-related data acquisition module can refer to the above description of step S110.

[0060] In detail, the data semantic mining module can be used to perform semantic mining on the multiple fault-related data to form a target fault semantic representation, wherein, in the process of semantic mining, for each of the time series data, the data fluctuation semantics in the time series data is mined from at least two time change directions. In the embodiment of the present application, the data semantic mining module can be used to perform Figure 2As shown in step S120, the relevant contents of the data semantic mining module can refer to the description of step S120 in the previous text.

[0061] In detail, the fault analysis module can be used to perform fault analysis on the target FTTR device based on the target fault semantic representation, and output fault prediction data corresponding to the target FTTR device, wherein the fault prediction data is used to reflect whether the LAN port of the target FTTR device will fail. In the embodiment of the present application, the fault analysis module can be used to perform Figure 2 As shown in step S130, the relevant contents of the fault analysis module can refer to the above description of step S130.

[0062] In the embodiment of the present application, corresponding to the above-mentioned fault analysis method of the FTTR device applied to the electronic device, a computer-readable storage medium is also provided, in which a computer program is stored, and when the computer program is run, each step of the fault analysis method of the FTTR device is executed. Among them, each step executed when the aforementioned computer program is run will not be described one by one here, and reference can be made to the above explanation of the fault analysis method of the FTTR device.

[0063] In summary, the fault analysis method and apparatus, equipment and medium of the FTTR equipment provided in the present application firstly obtain multiple fault-related data obtained by data collection of the target FTTR equipment; secondly, semantic mining is performed on the multiple fault-related data to form a target fault semantic representation; then, based on the target fault semantic representation, fault analysis is performed on the target FTTR equipment to output fault prediction data corresponding to the target FTTR equipment. Based on the above content, on the one hand, since semantic mining is performed on multiple fault-related data, the semantic information carried by the formed target fault semantic representation can be richer; on the other hand, since in the process of semantic mining, for each time series data, the data fluctuation semantics are mined from at least two time change directions, the representation reliability of the mined semantic information is higher, thereby ensuring the reliability of the fault prediction data output based on the target fault semantic representation, and then improving the problem of difficulty in effectively analyzing LAN port faults in the prior art.

[0064] In several embodiments provided in the embodiments of the present application, it should be understood that the disclosed device and method can also be implemented in other ways. The device and method embodiments described above are merely schematic, for example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the device, method and computer program product according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0065] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0066] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, electronic device, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code. It should be noted that in this article, the term "include", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such process, method, article or device. Without more constraints, an element defined by the phrase "comprising a..." does not exclude the existence of other identical elements in the process, method, article or apparatus comprising the element.

[0067] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for analyzing a fault of an FTTR device, characterized in that: include: Acquire multiple fault-related data obtained by collecting data from a target FTTR device, wherein the fault-related data refers to data that contributes to the failure of a LAN port of the target FTTR device, and there is at least one time series data among the multiple fault-related data; Performing semantic mining on the multiple fault-related data to form a target fault semantic representation, wherein, in the process of semantic mining, for each of the time series data, the data fluctuation semantics in the time series data are mined from at least two time change directions; Based on the target fault semantic representation, a fault analysis is performed on the target FTTR device, and fault prediction data corresponding to the target FTTR device is output, wherein the fault prediction data is used to reflect whether a LAN port of the target FTTR device will fail.

2. The fault analysis method of FTTR equipment according to claim 1, characterized in that: The step of performing semantic mining on the plurality of fault-related data to form a semantic representation of a target fault includes: For each time series data in the plurality of fault-related data, mining the data fluctuation semantics in the time series data from at least two time change directions, and outputting the corresponding local fault semantic representation; When there is at least one fault-related data that does not belong to time series data among the plurality of fault-related data, for each of the at least one fault-related data, semantic mining is performed on the fault-related data, and a corresponding local fault semantic representation is output; The local fault semantic representation corresponding to each of the fault-related data is fused to form a target fault semantic representation, wherein the target fault semantic representation is used to characterize the global semantic information possessed by the multiple fault-related data.

3. The fault analysis method of FTTR equipment according to claim 2, characterized in that: The step of mining the data fluctuation semantics in each time series data in the plurality of fault-related data from at least two time change directions and outputting the corresponding local fault semantic representation comprises: Loading the time series data into corresponding semantic mining branches, wherein each time series data corresponds to a semantic mining branch; For each sequence sub-data in the time series data, semantic space mapping is performed on the sequence sub-data to form a corresponding first fault semantic representation, and taking the sequence sub-data as a starting point, through the target window carried by the semantic mining branch, in the time series data according to a first time change direction, a first sequence data segment corresponding to the sequence sub-data is intercepted, and taking the sequence sub-data as a starting point, through the target window, in the time series data according to a second time change direction, a second sequence data segment corresponding to the sequence sub-data is intercepted, wherein the size of the target window is formed during training as a network parameter of the semantic mining branch; Performing semantic space mapping on the first sequence data segments and the second sequence data segments respectively to form corresponding second fault semantic representations and third fault semantic representations; The first fault semantic representation, the second fault semantic representation and the third fault semantic representation corresponding to each sequence sub-data are merged to form a corresponding local fault semantic representation.

4. The fault analysis method of FTTR equipment according to claim 3, characterized in that: The step of performing semantic space mapping on the first sequence data segments and the second sequence data segments respectively to form corresponding second fault semantic representations and third fault semantic representations includes: Performing a discreteness calculation on the first sequence data segments to obtain a first discreteness, and performing a discreteness calculation on the second sequence data segments to obtain a second discreteness, wherein the time series data is time numerical series data, including temperature, voltage or network traffic; Semantic space mapping is performed on the first discreteness and the second discreteness respectively to form a corresponding second fault semantic representation and a third fault semantic representation, wherein the semantic space mapping includes word embedding processing.

5. The fault analysis method of FTTR equipment according to claim 3, characterized in that: The step of fusing the first fault semantic representation, the second fault semantic representation and the third fault semantic representation corresponding to each sequence sub-data to form a corresponding local fault semantic representation includes: For each sequence sub-data, concatenate the first fault semantic representation, the second fault semantic representation, and the third fault semantic representation corresponding to the sequence sub-data to form a concatenated fault semantic representation corresponding to the sequence sub-data; The spliced ​​fault semantic representations corresponding to each of the sequence sub-data are spliced ​​to form a spliced ​​fault semantic representation corresponding to the time series data, and the spliced ​​fault semantic representations corresponding to the time series data are convolved and pooled respectively to realize different semantic feature extraction to form corresponding convolution fault semantic representations and pooled fault semantic representations, wherein the spliced ​​fault semantic representations corresponding to each of the sequence sub-data, the convolution fault semantic representations, and the pooled fault semantic representations have the same size; For each sequence sub-data, based on the spliced ​​fault semantic representation corresponding to the sequence sub-data, the convolution fault semantic representation and the pooled fault semantic representation are respectively focused and mined to form a corresponding first focused semantic representation and a second focused semantic representation, and the first focused semantic representation and the second focused semantic representation are superimposed to form a corresponding fused focused semantic representation; The fused focused semantic representation corresponding to each of the sequence sub-data is spliced ​​to form a spliced ​​focused semantic representation, and a local fault semantic representation is determined based on the spliced ​​focused semantic representation.

6. The fault analysis method of FTTR equipment according to claim 2, characterized in that: When there is at least one fault-related data that does not belong to time series data among the plurality of fault-related data, for each of the at least one fault-related data, performing semantic mining on the fault-related data and outputting a corresponding local fault semantic representation step includes: When there is at least one fault-related data that does not belong to the time series data among the plurality of fault-related data, for each of the at least one fault-related data, semantic space mapping is performed on the fault-related data to obtain a corresponding fault mapping semantic representation; When the number of obtained fault mapping semantic representations is greater than a preset number, clustering the fault mapping semantic representations to form at least one corresponding semantic representation cluster; For each of the fault mapping semantic representations, focused mining is performed on the fault mapping semantic representation according to the cluster center of the semantic representation cluster cluster corresponding to the fault mapping semantic representation to form a local fault semantic representation corresponding to the fault mapping semantic representation.

7. The method for analyzing the failure of FTTR equipment according to any one of claims 1 to 6, characterized in that: The fault analysis method of the FTTR device also includes: Using the semantic mining unit included in the candidate fault analysis network, semantic mining is performed on multiple training fault related data to form a training fault semantic representation, wherein the candidate fault analysis network belongs to a neural network, and the multiple training fault related data include at least one training time series data. In the process of semantic mining, for each of the training time series data, the data fluctuation semantics in the training time series data is mined from at least two time change directions; Utilizing the semantic analysis unit included in the candidate fault analysis network, performing fault analysis based on the training fault semantic representation, and outputting training fault prediction data; According to the training loss index between the training fault prediction data and the fault label data corresponding to the plurality of training fault-related data, the network parameters of the candidate fault analysis network are updated to form a target fault analysis network.

8. A fault analysis device for FTTR equipment, characterized in that: include: A fault-related data acquisition module is used to acquire a plurality of fault-related data obtained by collecting data from a target FTTR device, wherein the fault-related data refers to data that contributes to the failure of a LAN port of the target FTTR device, and there is at least one time series data among the plurality of fault-related data; A data semantic mining module is used to perform semantic mining on the multiple fault-related data to form a target fault semantic representation, wherein, in the process of semantic mining, for each of the time series data, the data fluctuation semantics in the time series data is mined from at least two time change directions; A fault analysis module is used to perform fault analysis on the target FTTR device based on the target fault semantic representation, and output fault prediction data corresponding to the target FTTR device, wherein the fault prediction data is used to reflect whether a LAN port of the target FTTR device will fail.

9. An electronic device, characterized in that: include: Memory for storing computer programs; A processor connected to the memory is used to execute the computer program stored in the memory to implement the fault analysis method of the FTTR device according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is run, the fault analysis method for the FTTR device according to any one of claims 1 to 7 is executed.

Images