A method for improving positioning accuracy based on a GRU neural network model

Abstract

The application relates to a method for improving positioning accuracy based on a GRU neural network model, and relates to the technical field of data processing, which comprises the following steps: S1, acquiring a time sequence corresponding to fault information, outputting a first target set based on the distribution of the time sequence; S2, determining a classification identifier corresponding to the first target set based on the first target set, and determining a fault node corresponding to the first target set based on the classification identifier; S3, determining a change trend of the fault information in the fault node according to the fault node, and outputting a second target set based on the change trend of the fault information; S4, processing the second target set by using a neural network, determining effective fault information in the second target set, and outputting the effective fault information as a third target set; and S5, evaluating the fault information based on the third target set, and outputting a final evaluation result; the method can improve the analysis efficiency of the fault information and improve the positioning accuracy of the fault node.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for improving positioning accuracy based on a GRU neural network model. Background Technology

[0002] A distributed system is a system in which hardware or software components are distributed across different devices and communicate and coordinate with each other only through message passing. Distributed systems are highly reliable, scalable, and offer fast communication, while also facilitating resource sharing among users. However, due to the large scale of devices involved in distributed services within a distributed system, and the complex inter-device and intra-device module calls, it is difficult to locate faults when they occur.

[0003] Existing fault location methods use trace IDs to track calls between devices. After a fault occurs, a global index is created based on the trace ID of the abnormal service, followed by analysis and fault localization. However, in this process, only the faulty device reports the fault information; other devices involved in handling the abnormal service may not. Therefore, the server obtains very limited fault-related information, which is detrimental to subsequent fault location analysis. Furthermore, this fault location method collects a large amount of normal business process data, which is likely unrelated to the fault, leading to unnecessary data analysis costs and increasing the difficulty of analysis.

[0004] For example, Chinese Patent Publication No. CN115460071A discloses a fault location method and apparatus. The method includes: acquiring a set of logs to be analyzed when a fault is detected in a first distributed service; determining the fault time; filtering related logs from the set of logs to be analyzed based on the fault time; extracting node information of multiple nodes from the fault logs and related logs, where each node represents the initiator or executor of a call, and the node information of each node includes the identifier of the first distributed service, the span identifier of the node, and the parent span identifier; constructing a call tree based on the node information of multiple nodes; and determining any node as a fault node if the out-degree and in-degree of any node in the call tree are different.

[0005] However, when identifying faulty nodes, if a large amount of data is processed, a large amount of fault information will be recorded during the identification process. Moreover, the identified fault information will not be screened, which may lead to false identification of faults due to excessive data, thus reducing the effectiveness of fault identification. Summary of the Invention

[0006] This invention addresses the technical problems existing in the prior art by providing a method for improving positioning accuracy based on a GRU neural network model.

[0007] The technical solution of this invention to solve the above-mentioned technical problems is as follows: A method for improving positioning accuracy based on a GRU neural network model, comprising:

[0008] S1, Obtain the time series corresponding to the fault information, and output the first target set based on the distribution of the time series;

[0009] S2, based on the first target set, determine the classification identifier corresponding to the first target set, and based on the classification identifier, determine the corresponding fault node in the first target set;

[0010] S3, Based on the fault node, determine the changing trend of fault information in the fault node, and output the second target set based on the changing trend of fault information;

[0011] S4, use a neural network to process the second target set, determine the valid fault information in the second target set, and output it as the third target set;

[0012] S5 evaluates the fault information based on the third target set and outputs the final evaluation result.

[0013] The beneficial effects of this invention are as follows: By classifying and identifying the fault information in the first target set, the corresponding fault nodes are determined, which helps to classify and locate faults; by analyzing the changing trends of fault information in each fault node, including extreme points and slopes, a second target set is output, which helps to understand the dynamic development of faults; by using a GRU neural network to process the second target set, effective fault information is determined and output as a third target set, which helps to extract and filter key fault information; based on the third target set, the fault information is comprehensively evaluated, and the final evaluation result is output, which helps to fully understand and prioritize faults. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating a method for improving positioning accuracy based on a GRU neural network model according to the present invention.

[0015] Figure 2 This is a flowchart illustrating step S2 of a method for improving positioning accuracy based on a GRU neural network model according to the present invention.

[0016] Figure 3 This is a flowchart illustrating step S3 of a method for improving positioning accuracy based on a GRU neural network model according to the present invention.

[0017] Figure 4 This is a flowchart illustrating step S33 of a method for improving positioning accuracy based on a GRU neural network model according to the present invention.

[0018] Figure 5 This is a flowchart illustrating step S4 of a method for improving positioning accuracy based on a GRU neural network model according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0021] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0022] Example 1

[0023] To improve the efficiency of processing acquired fault data, after acquiring the fault data, based on the time series corresponding to the fault data, the fault data is divided into multiple categories according to the acquisition time. The fault data in each category is then processed centrally to determine the context that needs to be remembered. Based on the final remembered context, the corresponding content is output to ensure that the current fault output information is not affected by past fault information, thus preventing a decrease in the accuracy of the current fault location. This helps to determine the fault information corresponding to the current timeline and assists in subsequent fault processing.

[0024] Specifically, when fault information is obtained, the present invention provides a method for improving positioning accuracy based on a GRU neural network model, comprising:

[0025] S1, obtain the time series corresponding to the fault information, and output the first target set based on the distribution of the time series. The first target set is used to obtain the fault information within a fixed time interval and sort the fault information according to the time series.

[0026] S2, based on the first target set, determine the classification identifier corresponding to the first target set, and based on the classification identifier, determine the corresponding fault node in the first target set;

[0027] S3. Based on the fault node, determine the changing trend of fault information in the fault node, and output a second target set based on the changing trend of fault information. The second target set is used to output the fault equipment information and classification representation corresponding to the extreme points in the changing trend of fault information, so as to determine the actual manifestation state of the fault when the fault occurs, and assist in the location of the fault.

[0028] S4, use a neural network to process the second target set, determine the valid fault information in the second target set, and output it as the third target set;

[0029] S5 evaluates the fault information based on the third target set and outputs the final evaluation result.

[0030] The first target set represents fault information within a fixed time period, the second target set is used to identify fault information with obvious changing trends, and the third target set is used to filter fault information and optimize the extraction of current fault information to obtain more effective or refined fault information.

[0031] By sorting and classifying time series data, fault information within a specific time interval can be quickly located, avoiding the need for extensive searching across the entire dataset. By analyzing the changing trends of fault information and paying particular attention to extreme points, the timing of fault occurrence and its actual manifestation can be determined more accurately. By refining and extracting fault occurrence data, more accurate fault information can be obtained, enabling more reliable fault information to be acquired when calling fault nodes.

[0032] Example 2

[0033] After obtaining the first target set, for the fault information within the first target set, the different fault nodes are classified according to the node, time and number of times the fault occurred, and the corresponding fault nodes are marked to make it easier to locate the fault nodes.

[0034] Specifically, such as Figure 2As shown, the implementation of step S2 also includes,

[0035] S21, obtain the key attributes corresponding to the fault information in the first target set. The key attributes include fault type, fault location, and severity. Based on these key attributes, assign one or more classification labels to each fault information. The classification labels are pre-set classification tags.

[0036] S22, group the fault information in the first target set according to the classification identifier, and classify the fault information with the same classification identifier and similar classification identifier into one category.

[0037] S23, determine the fault node corresponding to the fault information under each category.

[0038] The fault node is an abstract representation of the fault information, which contains the common features and attributes of the fault information. When classifying fault information with similar classification identifiers, fault information with a similarity greater than 80% is classified into one category.

[0039] Example 3

[0040] Based on the acquired fault nodes, the changing trend of fault information is determined. Based on the extreme points and slope of each changing trend in the fault information, the relationship between the fault node and the corresponding faulty equipment is determined, and the actual fault state of the fault node is obtained, thereby enabling rapid location of the fault type in the current fault node.

[0041] Specifically, such as Figure 3 As shown, step S3 is implemented in the following ways:

[0042] S31, For each fault node, analyze the changing trend of fault information in the fault node over time;

[0043] S32, obtain the extreme points and slope of each trend in the fault information, and determine the corresponding status of the fault information corresponding to the trend.

[0044] S33, based on the corresponding situation of the changing trend, determine the correlation between fault information, and output the correlated fault information as the second target set.

[0045] By processing the fault information corresponding to the acquired fault node and judging the trend of the corresponding fault information and the slope of the trend under the corresponding conditions, we can know the changes of the corresponding data when the fault node is invoked, and thus filter out the data that is highly related to the current fault node from the corresponding fault information.

[0046] Specifically, step S32, regarding the corresponding status of the changing trend, also includes:

[0047] Obtain the fault information corresponding to the extreme points of the changing trend, take the fault information at each extreme point as the first change information, calculate the dimension corresponding to each first change information, and divide the first change information according to the dimension;

[0048] Based on the first change information in each dimension, obtain the second change information corresponding to the slope of the change trend, compare the changes in the value of the second change information in different time periods, and output the corresponding status of the change trend.

[0049] The first dimension of change information refers to the classification of fault information by type and system component, which facilitates the verification of fault information at different fault nodes. The second dimension of change information represents the fault information corresponding to the increase and decrease of the slope of the current change trend. Based on these data, the reasons for the current fault trend change can be understood.

[0050] Specifically, such as Figure 4 As shown, the method for obtaining the second target set in step S33 includes:

[0051] S331, Based on the acquired trend of change, determine the fault information corresponding to the trend of change and determine the number of times the fault occurs;

[0052] S332, determine the correlation ratio between the trend of change and the fault information based on the number of times the fault occurs each time the fault occurs;

[0053] S333, count the number of impact ranges for each fault information and determine the probability ratio of each fault information to other fault information;

[0054] S334. For all fault information, determine the final correlation degree between fault information based on the correlation ratio and probability ratio of the fault information, and output the fault information that satisfies the final correlation degree as the second target set.

[0055] In step S332, the method for calculating the correlation ratio between fault information includes:

[0056] Based on the number of times each fault information appears in each version and the number of times each fault appears at different fault nodes, the current fault type is determined. Based on the fault type of the current fault information, the correlation between fault information is calculated, and the correlation ratio of fault information is determined.

[0057] Specifically, the correlation ratio of each fault information is expressed as follows:

[0058] R = (Number of times a certain fault occurs in the trend / Total number of faults in the trend) × 100%

[0059] Where R represents the correlation ratio of each fault information.

[0060] To determine the correlation ratio between each fault information, a weighted average is calculated based on the correlation ratio of each fault information to obtain the corresponding correlation ratio.

[0061] Specifically, the correlation ratio between fault information is expressed as follows:

[0062] C = (R1 + R2 + ... + Rn) / n

[0063] Where R1, R2, ..., Rn represent the correlation ratio of each fault information, n represents the total number of fault information, and C represents the correlation ratio between fault information.

[0064] Specifically, in step S333, the probability of fault information is...

[0065] Formula for calculating probability percentage:

[0066] The probability percentage reflects the likelihood of a particular fault occurring relative to other faults. It is calculated based on the scope of the fault's impact (e.g., the number of devices affected).

[0067] formula:

[0068] D = (Number of affected areas of a specific fault / Total number of affected areas of all faults) × 100%

[0069] Where D represents the probability of fault information;

[0070] By comparing the number of impact areas for each fault with the total number of impact areas for all faults, we can calculate the probability percentage of each fault and thus understand their relative importance.

[0071] To distinguish the currently occurring fault information from other fault information, the probability ratio among the set fault information can also be expressed as:

[0072] D = mx / Σ(mi × wi)) × 100%

[0073] Where D represents the probability of association between fault information x and other fault information, mx represents the number of influence ranges of fault information x, mi represents the number of influence ranges of the i-th fault information, wi represents the weight of the occurrence frequency of the i-th fault information, which can be simply taken as the occurrence frequency itself, or a function of the occurrence frequency (such as a logarithmic function to reduce the influence of extreme values), and Σ represents the summation sign, which sums the product of the number of influence ranges and the weights of all fault information.

[0074] Specifically, the final correlation calculation formula in step S334 includes:

[0075] The final correlation score combines information on correlation percentage and probability percentage to comprehensively assess the correlation between failures. It is usually calculated using a weighted average method.

[0076] formula:

[0077] Final correlation degree = α×C + β×D

[0078] Where α and β are weighting coefficients used to balance the contributions of correlation ratio and probability ratio in the comprehensive evaluation, C is the correlation ratio between fault information, and D is the probability ratio of correlation between fault information.

[0079] Preferably, in step S334, the fault information that satisfies the final correlation degree is output as the second target set, and the method further includes, based on the obtained final correlation degree, outputting the fault information with a final correlation degree greater than the preset prediction, and outputting it as the second target set.

[0080] By acquiring the correlation ratio and probability ratio between fault information, the fault information is processed according to the number of fault occurrences and the scope of fault impact, so that the acquired second target set can reflect the more obvious range of faults and the prominent correlation between faults, thereby improving the fault identification effect.

[0081] Example 4

[0082] After obtaining the second target set, the information in the second target set is filtered, and fault nodes that match the current fault state are selected from the corresponding fault information in the second target set. The corresponding fault nodes are verified, and the judgment is made based on the time, number, duration of the fault occurrence, and the correlation between each fault occurrence to determine whether the current fault is a real fault. The most obvious part of the fault information is extracted to assist in the fault identification based on effective fault information.

[0083] Specifically, such as Figure 5 As shown, step S4 is implemented in the following ways:

[0084] S41, Based on the acquired second target set, determine the fault level corresponding to the fault information;

[0085] S42, Determine the severity of the fault information based on the fault level corresponding to the fault information;

[0086] S43, based on the severity of the fault information, input the fault information into the GRU neural network, and use the output of the GRU neural network as the third target set.

[0087] Suppose the second target set contains multiple fault information entries, each of which may include fault type, fault code, fault description, etc. A fault level table is predefined, listing various possible fault types and their corresponding fault levels. Fault levels can be numbers (e.g., 1, 2, 3...) or descriptive labels (e.g., "low", "medium", "high").

[0088] Iterate through each fault information entry in the second target set. For each entry, look up the corresponding fault level in the fault level table based on the fault type or fault code. Associate the found fault level with the original fault information. This can be done by creating a new data structure (such as a dictionary or object) in memory, where the key is the fault information and the value is the corresponding fault level.

[0089] Similar to fault levels, a severity table is predefined, mapping fault levels to corresponding severity levels. Severity levels can be numbers, color-coded, or descriptive labels. Iterate through each fault message obtained in the previous step and its corresponding fault level. Using the severity table, look up the corresponding severity level based on the fault level. Associate the found severity level with the original fault message, updating the previously created data structure.

[0090] After obtaining the severity of the corresponding fault information, the specific implementation of step S43 is as follows: obtain the hidden state corresponding to the fault information, filter out the valid fault information in the second target set based on the hidden state of the fault information, and output it as the third target set.

[0091] The hidden state of fault information includes the historical context of the fault event, including information such as the type, frequency, and duration of previous faults. This information is encoded in the hidden state and used to help the network more accurately determine whether the current fault is real and to identify its category or severity.

[0092] In this embodiment, the acquired second target set is processed by using a GRU neural network to construct one or more GRU layers. Each layer contains multiple GRU units. Each GRU unit records the fault information of each input in the current time step and calculates the hidden state of the fault information corresponding to the current time step by combining the fault information in the previous time step.

[0093] In each GRU unit, the fault information of the current time step is combined with the hidden state of the fault information of the previous time step through the reset gate. When the output of the reset gate is close to 0, it effectively "forgets" the past hidden state, making the unit mainly rely on the current input. When the output of the reset gate is close to 1, it retains more past information.

[0094] Based on the combination of the fault information of the current time step and the hidden state of the fault information of the previous time step, the candidate hidden state corresponding to the current time step is determined; the hidden state of the fault information of the previous time step is combined with the candidate hidden state to form the final hidden state of the current time step, and the fault information corresponding to the output final hidden state is output as the third target set.

[0095] Specifically, the reset gate determines how to combine the previous state with the current input, while the update gate controls the extent to which information from the previous state is incorporated into the current state. These gates are activated by the sigmoid function and learn how to optimally combine information during training. Based on the outputs of the reset and update gates, the GRU unit computes a new candidate hidden state (usually activated by the tanh function). Then, based on the output of the update gate, it combines the hidden state from the previous time step with the new candidate hidden state to form the final hidden state at the current time step.

[0096] Preferably, the hidden state of the last time step is used as a representation of the entire sequence and passed to a fully connected layer or classifier to output the classification or identification result of the fault.

[0097] In this embodiment, the purpose of the GRU neural network may be to predict whether a currently occurring fault is real and to identify its category or severity based on historical fault data and currently observed fault characteristics (such as time, frequency, duration, etc.). In this way, the network can help automated fault detection systems more accurately identify and handle fault events.

[0098] Example 5

[0099] After obtaining the processed third target set, the weight of the entire system is evaluated based on the fault node to determine the corresponding fault evaluation value when the fault occurs.

[0100] Specifically, the output methods corresponding to the fault assessment in step S5 also include,

[0101] Based on the acquired third target set, the corresponding fault nodes within the third target set are identified. For each fault node, its impact value on the system is determined based on the set weights. The impact values ​​of all fault nodes are summed to obtain the overall assessment value of the current fault. This calculated fault assessment value is compared with a preset threshold to determine whether the current fault requires immediate attention or is within a tolerable range. The assessment results are output in an appropriate format, such as a report, chart, or alarm, so that relevant personnel can understand and take action.

[0102] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0103] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0104] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0105] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0106] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0107] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0108] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for improving positioning accuracy based on a GRU neural network model, characterized in that, include: S1, Obtain the time series corresponding to the fault information, and output the first target set based on the distribution of the time series; S2, based on the first target set, determine the classification identifier corresponding to the first target set, and based on the classification identifier, determine the corresponding fault node in the first target set; S3, Based on the fault node, determine the changing trend of fault information in the fault node, and output the second target set based on the changing trend of fault information: S31, For each fault node, analyze the changing trend of fault information in the fault node in the time series. S32, obtain the extreme points and slope of each trend in the fault information, and determine the corresponding status of the fault information corresponding to the trend; S33, based on the corresponding status of the trend, determine the correlation between the fault information, and output the correlated fault information as the second target set. The method for obtaining the second target set in S33 includes: S331, based on the obtained change trend, determining the fault information corresponding to the change trend and determining the number of times the fault occurs; S332, determining the correlation ratio between the change trend and the fault information based on the number of times the fault occurs each time; S333, counting the number of influence ranges of each fault information and determining the probability ratio of each fault information to other fault information; S334, for all fault information, determining the final correlation degree between fault information based on the correlation ratio and probability ratio corresponding to the fault information, and outputting the fault information that satisfies the final correlation degree as the second target set. S4, use a neural network to process the second target set, determine the valid fault information in the second target set, and output it as the third target set; S5 evaluates the fault information based on the third target set and outputs the final evaluation result.

2. The method for improving positioning accuracy based on a GRU neural network model according to claim 1, characterized in that, The implementation of step S2 also includes, S21, Obtain the key attributes corresponding to the fault information in the first target set. The key attributes include fault type, fault location, and severity. Based on the key attributes, assign a classification label to each fault information. The classification label is a pre-set classification tag. S22, group the fault information in the first target set according to the classification identifier, and classify the fault information with the same classification identifier and similar classification identifier into one category; S23, determine the fault node corresponding to the fault information under each category.

3. The method for improving positioning accuracy based on a GRU neural network model according to claim 1, characterized in that, The corresponding status regarding the trend of change in step S32 also includes: Obtain the fault information corresponding to the extreme points of the changing trend, take the fault information at each extreme point as the first change information, calculate the dimension corresponding to each first change information, and divide the first change information according to the dimension; Based on the first change information in each dimension, obtain the second change information corresponding to the slope of the change trend, compare the changes in the value of the second change information in different time periods, and output the corresponding status of the change trend.

4. The method for improving positioning accuracy based on a GRU neural network model according to claim 1, characterized in that, Step S4 can be implemented in the following ways: S41, Based on the acquired second target set, determine the fault level corresponding to the fault information; S42, Determine the severity of the fault information based on the fault level corresponding to the fault information; S43, based on the severity of the fault information, input the fault information into the GRU neural network, and use the output of the GRU neural network as the third target set.

5. The method for improving positioning accuracy based on a GRU neural network model according to claim 1, characterized in that, In step S332, the method for calculating the correlation ratio between fault information includes: Based on the number of times each fault information appears in each version and the number of times each fault appears at different fault nodes, the current fault type is determined. Based on the fault type of the current fault information, the correlation between fault information is calculated, and the correlation ratio of fault information is determined.

6. The method for improving positioning accuracy based on a GRU neural network model according to claim 5, characterized in that, The correlation ratio of each fault information is represented as follows: R = (Number of times a certain fault occurs in the trend / Total number of faults in the trend) × 100% Where R represents the correlation ratio of each fault information; The correlation ratio between fault information is expressed as follows: C = (R1 + R2 + ... + Rn) / n Where R1, R2, ..., Rn represent the correlation ratio of each fault information, n represents the total number of fault information, and C represents the correlation ratio between fault information.

7. The method for improving positioning accuracy based on a GRU neural network model according to claim 1, characterized in that, The probability ratio among fault information is expressed as follows: D = mx / Σ(mi × wi)) × 100% Where D represents the probability of association between fault information x and other fault information, mx represents the number of influence ranges of fault information x, mi represents the number of influence ranges of the i-th fault information, wi represents the weight of the occurrence frequency of the i-th fault information, and Σ represents the summation symbol, which sums the product of the number of influence ranges and the weights of all fault information.

8. The method for improving positioning accuracy based on a GRU neural network model according to claim 4, characterized in that, The specific implementation methods of step S43 include: The acquired second target set is processed by using a GRU neural network to construct one or more GRU layers, each containing multiple GRU units. Each GRU unit records the fault information of each input in the current time step and calculates the hidden state of the fault information corresponding to the current time step by combining the fault information in the previous time step. Based on the combination of the fault information of the current time step and the hidden state of the fault information of the previous time step, the candidate hidden state corresponding to the current time step is determined; the hidden state of the fault information of the previous time step is combined with the candidate hidden state to form the final hidden state of the current time step, and the fault information corresponding to the output final hidden state is output as the third target set.

Images