Vehicle fault similar knowledge base acquisition method and device, vehicle, and storage medium

Abstract

The present disclosure relates to a vehicle fault homogeneous knowledge base acquisition method and device, a vehicle, and a storage medium. The method comprises: acquiring a to-be-merged log group, wherein the to-be-merged log group contains a plurality of fault characterization information; based on the fault characterization information, a key fault knowledge set is screened out from an existing fault knowledge set; the importance of the keywords in each log in the key fault knowledge set is calculated to obtain the keyword weight of each log; based on the keyword weight, the feature similarity weight between each log in the key fault knowledge set is determined; based on the feature similarity weight, a target fault knowledge set is screened out from the key fault knowledge set, and the target fault knowledge set is taken as the homogeneous knowledge base of the to-be-merged log group. Through the complete process of "fault characterization information screening + keyword importance weight calculation + feature similarity weight acquisition", the present disclosure avoids the misjudgment caused by pure string matching or single-dimensional matching, and further obtains an accurate homogeneous knowledge set.

Description

Technical Field

[0001] This disclosure relates to the field of fault log processing technology, specifically to methods, devices, vehicles, and storage media for acquiring similar knowledge bases for vehicle faults. Background Technology

[0002] With the rapid improvement of the intelligence and connectivity of new energy vehicles, the cockpit domain has become the core interactive hub of the entire vehicle, integrating multiple electronic control units such as the in-vehicle infotainment system, digital instrument cluster, head-up display, and in-vehicle communication module. It achieves unified and coordinated control through high-performance cockpit domain controllers, in-vehicle Ethernet, and CAN bus. Industry data shows that electronic components in the intelligent cockpit domain account for more than 35% of the entire vehicle's electronic system. The amount of critical log data generated during its operation is growing exponentially, with a single vehicle generating thousands of fault-related logs daily, covering multi-dimensional information such as fault system, fault type, log template, variable parameters, fault codes, root cause descriptions, and fault phenomena.

[0003] These log data are the core basis for fault diagnosis, after-sales maintenance, and product iteration optimization. However, due to the complexity of the cockpit domain system architecture (multi-protocol compatibility, multi-module interaction) and the diversity of fault triggering conditions (such as ambient temperature, user operating habits, and software version differences), similar faults often generate log sets with similar forms but different details. For example, the same "CDC module communication timeout" fault may form multiple seemingly independent log entries due to different timeout parameters (500ms / 300ms) and triggering scenarios (cold start / warm start), thus creating a large number of redundant entries in the knowledge base. Therefore, it is necessary to merge a large number of redundant entries to reduce resource consumption.

[0004] However, when merging similar logs, logs of the same type of fault are sometimes misjudged as dissimilar, while logs with different root causes but similar semantics are also misjudged as the same type. This results in insufficient merging accuracy and fails to meet the precision requirements for cockpit domain fault diagnosis. Summary of the Invention

[0005] In view of this, this disclosure provides a method, device, vehicle, and storage medium for obtaining a knowledge base of similar vehicle faults, in order to solve the problem in related technologies that the accuracy of merging similar logs is insufficient and cannot meet the accuracy requirements of cockpit domain fault diagnosis.

[0006] Firstly, this disclosure provides a method for obtaining a knowledge base of similar vehicle faults, the method comprising: Obtain the log group to be merged, which contains multiple fault characterization information; Based on fault characterization information, a set of key fault knowledge is selected from the existing set of fault knowledge. The importance of keywords in each log in the critical fault knowledge set is calculated to obtain the keyword weight of each log. The feature similarity weights between each log within the critical fault knowledge set are determined based on keyword weights. Based on feature similarity weights, a target fault knowledge set is selected from the key fault knowledge set, and the target fault knowledge set is used as a similar knowledge base for the log group to be merged.

[0007] In one optional implementation, the fault characterization information includes root factor information and fault manifestation sub-information. Based on fault characterization information, a key fault knowledge set is selected from the existing fault knowledge set, including: Based on the fault manifestation sub-information, the first fault knowledge set is selected from the existing fault knowledge set; The key fault knowledge set is selected from the first fault knowledge set based on root factor information.

[0008] In one optional implementation, a first fault knowledge set is selected from an existing fault knowledge set based on fault manifestation sub-information, including: The fault manifestation sub-information is vectorized to obtain the first vector information; Based on the first vector information, a first mapping table is matched to filter out multiple fault types that have a similarity greater than a first threshold with the first vector information. The first mapping table is used to characterize the mapping relationship between fault type and fault manifestation. Based on the fault type, the first fault knowledge set is selected from the existing fault knowledge set.

[0009] In one optional implementation, a key fault knowledge set is selected from the first fault knowledge set based on root factor information, including: The root factor information is vectorized to obtain the second vector information; The second vector information is matched with the third vector information of each log in the first fault knowledge set to obtain a key fault knowledge set whose similarity is greater than the second threshold. The third vector information is the vectorized information of the root factor information of each log in the first fault knowledge set.

[0010] In one optional implementation, the feature similarity weight includes keyword similarity weight and time similarity weight in the log. The time similarity weight is obtained from the similarity interval corresponding to the difference between the time of failure recorded in the log and the actual time of failure. The feature similarity weights between each log within the critical fault knowledge set are determined based on keyword weights, including: Key logs are selected from the critical fault knowledge set based on keyword weights; Calculate the similarity of keywords in key logs to obtain keyword similarity weights; Obtain the first time of failure and the second time when the failure actually occurred from each log record in the critical failure knowledge set; Based on the first and second time points, determine the time difference between the fault and the actual occurrence of each log record in the critical fault knowledge set; Based on the time difference, the time similarity weight is obtained.

[0011] In one optional implementation, a target fault knowledge set is selected from the key fault knowledge set based on feature similarity weights, including: The keyword similarity weight and time similarity weight are combined to obtain the combined weight; Obtain the reproducibility weight of each log in the critical fault knowledge set; Based on the fused weights and the reproducible weights, the overall similarity score of the key fault knowledge set is obtained; Select target fault knowledge sets from the key fault knowledge set, whose total similarity score is greater than a preset threshold. In one alternative implementation, the method further includes: If the target fault knowledge set is not present in the critical fault knowledge set, other fault knowledge sets in the critical fault knowledge set whose total similarity score is less than or equal to a preset threshold are added to the knowledge graph of the log group to be merged, resulting in a new knowledge graph.

[0012] Secondly, this disclosure provides an apparatus for acquiring a knowledge base of similar vehicle faults, the apparatus comprising: The first acquisition module is used to acquire the log group to be merged, wherein the log group to be merged contains multiple fault characterization information; The first filtering module is used to filter out key fault knowledge sets from the existing fault knowledge set based on fault characterization information. The calculation module is used to calculate the importance of keywords in each log in the critical fault knowledge set and obtain the keyword weight of each log. The determination module is used to determine the feature similarity weights between each log within the critical fault knowledge set based on keyword weights. The second filtering module is used to filter out the target fault knowledge set from the key fault knowledge set based on feature similarity weights, and to use the target fault knowledge set as a similar knowledge base for the log group to be merged.

[0013] Thirdly, this disclosure provides a vehicle, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to perform the method for acquiring a vehicle fault similar knowledge base according to the first aspect or any corresponding embodiment described above.

[0014] Fourthly, this disclosure provides a computer-readable storage medium storing computer instructions for causing a computer to execute the method for acquiring a vehicle fault similar knowledge base according to the first aspect or any corresponding embodiment described above.

[0015] Fifthly, this disclosure provides a computer program product, including computer instructions for causing a computer to execute the method for acquiring a vehicle fault similar knowledge base as described in the first aspect or any corresponding embodiment.

[0016] In this embodiment, a log group to be merged is obtained, which contains multiple fault characterization information. Then, based on the fault characterization information, a key fault knowledge set is selected from an existing fault knowledge set. Next, the importance of keywords in each log within the key fault knowledge set is calculated to obtain the keyword weight for each log. Based on the keyword weight, the feature similarity weight between each log within the key fault knowledge set is determined. Based on the feature similarity weight, a target fault knowledge set is selected from the key fault knowledge set and used as the same type of knowledge base for the log group to be merged. Thus, this embodiment, through a complete process of "fault characterization information filtering + keyword importance weight calculation + feature similarity weight acquisition" to obtain a vehicle fault similarity knowledge base, avoids misjudgments caused by pure string matching or single-dimensional matching, significantly improving the accuracy of subsequent merging of similar faults. This effectively solves the problems of severe knowledge base redundancy and high maintenance costs, reduces duplicate storage of similar fault entries, reduces storage resource consumption, and meets the accuracy requirements of cockpit domain fault diagnosis. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the specific embodiments of this disclosure or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a method for obtaining a vehicle fault knowledge base according to an embodiment of the present disclosure; Figure 2 This is a flowchart illustrating a method for obtaining a vehicle fault knowledge base according to another embodiment of the present disclosure; Figure 3 This is a flowchart illustrating a method for obtaining a vehicle fault knowledge base according to yet another embodiment of this disclosure; Figure 4This is a schematic diagram illustrating the complete process of acquiring a vehicle fault knowledge base according to an embodiment of this disclosure; Figure 5 This is a structural block diagram of a device for acquiring a vehicle fault similar knowledge base according to an embodiment of the present disclosure; Figure 6 This is a structural block diagram of a vehicle according to an embodiment of the present disclosure. Detailed Implementation

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

[0020] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0021] 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 technical features indicated. Therefore, a feature defined as "first" may explicitly or implicitly include one or more of that feature.

[0022] In the field of intelligent new energy vehicles, the intelligent cockpit domain of the vehicle will generate a large number of critical logs during operation. A single vehicle can generate thousands of fault-related logs per day, covering multi-dimensional information such as fault system, fault type, log template, variable parameters, fault code, root cause description, and fault phenomenon.

[0023] These log data are the core basis for fault diagnosis, after-sales maintenance, and product iteration optimization. However, due to the complexity of the cockpit domain system architecture and the diversity of fault triggering conditions, similar faults often generate log sets with similar forms but different details. For example, the same "CDC module communication timeout" fault may generate multiple seemingly independent log entries due to different timeout parameters (500ms / 300ms) and triggering scenarios (cold start / warm start), thus creating a large number of redundant entries in the knowledge base. Therefore, it is necessary to merge a large number of redundant entries to reduce resource consumption.

[0024] However, there are two key misjudgments when merging similar logs: First, pure string matching schemes fail to accurately identify the structured characteristics of logs ("template + parameters"), easily misjudging similar logs with "same template but different parameters" (such as "CDC timeout 500ms" and "CDC timeout 300ms") as different types, or misjudging dissimilar logs with "different templates but overlapping fixed fields" as similar types. Second, some technical implementations overemphasize parameter differences and surface semantics, misjudging logs of similar faults as dissimilar, and misjudging logs with different root causes but similar semantics as similar types, resulting in insufficient merging accuracy and failing to meet the accuracy requirements of cockpit domain fault diagnosis.

[0025] To address the aforementioned problems, according to an embodiment of this disclosure, a method for acquiring a vehicle fault knowledge base is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0026] This embodiment provides a method for acquiring a knowledge base of similar vehicle faults, which can be used in the vehicle's MCU (Microcontroller Unit). Figure 1 This is a flowchart of a method for obtaining a vehicle fault similarity knowledge base according to an embodiment of this disclosure, such as... Figure 1 As shown, the process includes the following steps: Step S101: Obtain the log group to be merged, wherein the log group to be merged contains multiple fault characterization information.

[0027] Optionally, in the fault data management scenario of the intelligent cockpit domain of new energy vehicles, the fault log groups to be merged are first obtained from the log storage modules of the vehicle's hardware systems such as the cockpit domain controller (CDC), in-vehicle Ethernet, and CAN / LIN bus, as well as the database of the fault diagnosis platform. These log groups are not single fault logs, but rather contain fault characterization information such as root cause, fault code, fault symptoms, and log entries. For example, root causes could include network link congestion or poor contact at the CDC module hardware interface. Fault code: P1234 (CDC communication timeout general code); Fault symptoms: The central control screen is black, the vehicle Ethernet data transmission is interrupted, and the cabin domain controller is unresponsive after a cold start; Log entries include entries such as "2025-10-01 08:05 CDC communication timeout 500ms (cold start)" and "2025-10-01 14:30 CDC communication timeout 300ms (warm start)" (each log entry contains a millisecond-level timestamp).

[0028] Step S102: Based on the fault characterization information, select the key fault knowledge set from the existing fault knowledge set.

[0029] Optionally, in this embodiment of the disclosure, taking fault characterization information including root factor information and fault manifestation sub-information as an example, the screening of the fault knowledge set is carried out. Specifically: Fault manifestation sub-information refers to the external and intuitive characteristics of a fault, such as specific phenomena of cockpit domain faults like "CDC module communication timeout" or "in-vehicle infotainment system black screen". It is the primary dimension of fault screening, and the first fault knowledge set is selected from a large number of existing fault knowledge sets through fault manifestation sub-information.

[0030] Furthermore, such as Figure 2 Step S102 includes: Step S1021: Based on the fault manifestation sub-information, select the first fault knowledge set from the existing fault knowledge set.

[0031] Optionally, fault manifestation sub-information refers to the external, intuitive characteristics of the fault, such as specific phenomena of cockpit domain faults like "CDC module communication timeout" or "in-vehicle infotainment system black screen," which are the primary dimensions for fault screening.

[0032] Fault manifestation sub-information is the easiest feature to extract and match. Currently, we can first select the first fault knowledge set from the existing fault knowledge set based on the fault manifestation sub-information. This can greatly reduce the amount of subsequent calculations, avoid complex calculations on the full existing fault knowledge set, and improve the screening efficiency.

[0033] Specifically, in step a1, the fault manifestation sub-information is vectorized to obtain the first vector information; Step a2: Based on the first vector information, match the first mapping table to filter out multiple fault types with a similarity greater than the first vector information. The first mapping table is used to characterize the mapping relationship between fault type and fault manifestation. Step a3: Based on the fault type, select the first fault knowledge set from the existing fault knowledge set.

[0034] Optionally, a preset first mapping table is invoked (this mapping table is constructed manually based on cockpit domain fault diagnosis experience, and is used to represent the one-to-one or many-to-many mapping relationship between fault types and fault manifestations, such as fault manifestation "communication timeout" corresponding to fault types such as "network failure" and "module offline"), and the first vector information is used to calculate the cosine similarity with the vector information of each fault type in the mapping table. The value ranges from 0 to 1, and the closer the value is to 1, the higher the similarity.

[0035] Filter out all fault types whose similarity to the first vector information is greater than the first threshold (e.g., set to 0.8, which is the optimal value verified by combining massive cockpit domain fault data statistics and algorithms, and can balance the accuracy and coverage of the filtering). Then, based on these fault types, extract the corresponding associated log sets from the existing fault knowledge set as the first fault knowledge set.

[0036] Step S1022: Select the key fault knowledge set from the first fault knowledge set based on the root factor information.

[0037] Optionally, root factor information, such as "network link interruption", "hardware module failure", "software version incompatibility", etc., is the core dimension of fault screening.

[0038] The same fault manifestation sub-information may be caused by different root causes. For example, "communication timeout" may be caused by network failure or hardware failure. Filtering by fault manifestation sub-information alone will result in a lot of redundancy. It is necessary to further narrow down the scope by identifying the root cause.

[0039] Specifically, in step b1, the root factor information is vectorized to obtain the second vector information; Step b2: Perform similarity matching between the second vector information and the third vector information of each log in the first fault knowledge set to obtain a key fault knowledge set where the similarity is greater than the second threshold. The third vector information is the vectorized information of the root factor information of each log in the first fault knowledge set.

[0040] Optionally, consistent with the fault performance sub-information processing logic, the root factor information in the log group to be merged is vectorized, and the same text embedding model is used to generate the second vector information to ensure the consistency of vector representation.

[0041] Root factor information is extracted from all logs in the first fault knowledge set and pre-vectorized to obtain third vector information. Cosine similarity matching is then performed between the second vector information and all third vector information, calculating the pairwise similarity values. Log sets with similarity values ​​greater than a second threshold are selected as the key fault knowledge set. The second threshold can be the same as the first threshold or fine-tuned according to the accuracy requirements of root cause matching. The core purpose is to eliminate dissimilar fault logs with "same symptoms but different root causes," further narrowing the selection range.

[0042] Step S103: Calculate the importance of keywords in each log in the critical fault knowledge set to obtain the keyword weight of each log.

[0043] Optionally, after two levels of screening using fault manifestation sub-information and root factor information, the resulting key fault knowledge set is a set of highly relevant fault logs. This step determines the importance weight of the keywords in each log in the key fault knowledge set by searching and scoring the keywords, thereby completing the final accurate screening of key logs in the key fault knowledge set.

[0044] The specific implementation process is as follows: ① Keyword extraction: All logs in the key fault knowledge set are processed by word segmentation. For example, using the NLTK (Natural Language Toolkit) word segmentation tool, stop words can be filtered and core keywords can be extracted. For example, core keywords such as "CDC", "communication timeout", and "500ms" can be extracted from "CDC module communication timeout 500ms".

[0045] ② Search scoring: The keywords in each log entry of the key fault knowledge set are scored to obtain the keyword weight of each log entry. The scoring rule is base score × 10 + keyword occurrence frequency × 2. The higher the word frequency, the higher the score, which can reflect the overlap and importance of keywords.

[0046] The algorithm formula for calculating the base score of each keyword in each log entry within the critical fault knowledge set is as follows: ; Where α is the global smoothing constant; The total number of any log in the critical fault knowledge set; : The number of times the keyword's position changed in the corresponding log; The total number of keywords in each log entry; Position fluctuation correction factor ( ), The standard deviation of the number of positional changes. This represents the average length of a single log entry. Keyword frequency correction factor ( ), For keyword frequency, This is the frequency threshold (which can be 3 to 5). Semantic importance correction factor (2.0 for core keywords, adjusted by length for non-core keywords): ), This represents the length of the keyword.

[0047] Step S104: Determine the feature similarity weights between each log in the critical fault knowledge set based on keyword weights.

[0048] Optionally, feature similarity weight is the core quantitative indicator for determining the correlation between each log in the critical fault knowledge set and the log group to be merged. Since the similarity of fault logs is reflected not only in content but also in temporal correlation, only logs within a certain time range before and after the fault have actual correlation value. Therefore, in this embodiment, it is split into keyword similarity weight and time similarity weight, which evaluate similarity from two dimensions: textual features and temporal features, respectively. This aligns with the trend of multi-dimensional feature fusion in fault diagnosis technology.

[0049] Furthermore, the feature similarity weight includes keyword similarity weight and time similarity weight in the log. The time similarity weight is obtained from the similarity interval corresponding to the difference between the time of the fault recorded in the log and the actual time of the fault.

[0050] like Figure 3 The above step S104 includes: Step S1041: Select key logs from the key fault knowledge set based on keyword weights.

[0051] Step S1042: Calculate the similarity of keywords in the key logs to obtain keyword similarity weights.

[0052] Step S1043: Obtain the first time of each fault record in the critical fault knowledge set and the second time when the fault actually occurred.

[0053] Step S1044: Based on the first time and the second time, determine the time difference between each log record in the critical fault knowledge set and the actual occurrence of the fault.

[0054] Step S1045: Obtain the time similarity weight based on the time difference.

[0055] Optionally, keyword similarity weighting assesses the similarity of logs in the critical fault knowledge set from the perspective of log text content. Its core is the calculation of keyword importance: different keywords have different representational value for faults. For example, the importance of core keywords such as "CDC" and "fault code P001" is much higher than that of general keywords such as "module" and "occurrence". Different weights need to be assigned through importance calculation to avoid interference from general keywords in similarity determination.

[0056] The specific implementation process is as follows: based on the selection rule of the top 5 scores, the set of logs that meet the score criteria are selected from the critical fault knowledge set and referred to as critical logs.

[0057] Cosine similarity is calculated between keywords in key logs. The result is the keyword similarity weight. The closer the weight value is to 1, the higher the similarity of the text content.

[0058] Temporal similarity weighting assesses the similarity of logs across a critical fault knowledge set from the perspective of temporal correlation. Its core is the analysis of the time difference between the fault recording time and the actual fault time. The principle behind this is that after a cockpit domain fault occurs, only logs recorded within a reasonable timeframe are directly related to the fault. If the time interval between the log recording time and the actual fault occurrence is too far, the correlation is extremely low, and the influence of such logs needs to be weakened through temporal similarity weighting, conforming to the temporal distribution pattern of fault logs.

[0059] The specific implementation process is as follows: ①Time information extraction: Extract the timestamp of the log record of the fault from the metadata of each log record in the critical fault knowledge set as the first time; at the same time, extract the timestamp of the actual occurrence of the fault from the fault diagnosis platform as the second time. Both types of time are uniformly absolute timestamps to avoid calculation errors caused by inconsistent time formats. ② Time difference calculation: Calculate the time difference based on the first and second times using the following formula: ,in, Time of failure (i.e., the second time). Log time (i.e., the first time).

[0060] Simultaneously, a global smoothing term ε (with a value of 10) is introduced. -5 ~10 -4 This avoids calculation errors caused by a time difference of 0.

[0061] ③ Segmented decay calculation of similarity weight: Based on the dynamic time interval to which the time difference belongs, the time similarity weight is calculated using a non-linear decay method. The algorithm formula is as follows:

[0062] in, i For the first i Calculation of relative time difference for each log entry: ; This refers to the maximum reasonable time interval between the log entry and the time the fault occurred, for example, set to 10. min ; Time range: ,Replenish Avoid starting the interval with a value of 0; Time-based smoothing term (value range: This avoids calculation errors caused by a time difference of 0. : No. i The preceding interval value of each time interval (default) ), : No. i Critical values ​​for each time interval ( ;default ); : No. i The attenuation coefficient for each interval (value range: 0.5~2.0, the larger the interval, the higher the attenuation coefficient). The larger the value, the faster the decay. : Interval correction factor (value range: 0.9~1.1), used to calibrate the relative time difference accuracy of different intervals.

[0063] As can be seen from the above, the time difference ≤ The logs were judged to have a similarity of 1, and the time difference exceeded [a certain threshold]. m +120 min Logs that are directly identified as unrelated have a similarity of 0. Other cases are identified using a formula. The similarity is calculated, and the final result is the time similarity weight. The closer the weight value is to 1, the higher the time correlation.

[0064] Since not every log entry and fault occurrence time interval is fixed, some log entries may have a normal time interval of 30 minutes, while others may have a time interval of 10 minutes or 5 minutes. Therefore, we currently need intelligent log occurrence time intervals to calculate dynamic and reasonable time intervals.

[0065] Specifically, the average interval between the original log entry time and the fault occurrence time in the knowledge base is utilized. Replace the above formula Combined with the interval of new logs Calculate the new baseline value ; Based on this, the dynamic time interval is determined as [ +ε, +10min, +30min, +60min, +120min].

[0066] in, The interval between the new log time and the time of the failure, used to combine historical benchmarks. This enables dynamic updates of new baseline values, adapting to the timeliness requirements of new logs.

[0067] The actual number of occurrences of this log within its respective key knowledge set is used as a weighted fusion factor to balance the results. (Historical data) and The higher the frequency of (new data) occurrences, the greater the weighting of new data. The higher the weight, the better.

[0068] Time decay coefficient (value range: 0.9~0.98). The number of days between the log time and the current time; both constitute the time decay term. Its core function is to reflect the timeliness of logs: the newer the log, the better. The smaller the value, the weaker the attenuation effect. The closer the value is to the actual interval, the more timely the benchmark value becomes.

[0069] Occurrence frequency threshold (range: 2~4), combined with nonlinear transformation term This constitutes a frequency correction term, used to prevent log occurrences from being too frequent or too infrequent. Deviation, ensuring the stability and rationality of the new benchmark value, and reducing The resulting volatility.

[0070] Global smoothing term, range of values , used to avoid To improve robustness and maintain consistency with the overall algorithm's smoothing logic, the calculation process is designed to handle cases where the denominator is 0 or extreme values ​​are abnormal.

[0071] The new time-related baseline value replaces... m .

[0072] Step S105: Based on feature similarity weights, select the target fault knowledge set from the key fault knowledge set and use the target fault knowledge set as a similar knowledge base for the log group to be merged.

[0073] Optionally, the feature similarity weights complete the separate evaluation of the keyword text and time dimensions. Currently, the total similarity score of the key fault knowledge set can be obtained based on the keyword similarity weights and time similarity weights.

[0074] Specifically, the obtained keyword similarity weights and time similarity weights are weighted and fused according to a preset fixed ratio. For example, the total similarity score = keyword similarity weight × 0.8 + time similarity weight × 0.2.

[0075] The algorithm formula for calculating the reproducibility weight of each log entry in the critical fault knowledge set is as follows: ,in, The total number of a certain log entry under the critical fault knowledge set; The total number of logs within the critical fault knowledge set; Global smoothing constant (range of values: This is used to avoid denominators of 0 or extreme values ​​in all algorithms, thereby improving overall robustness; : Inevitable stability correction coefficient (value range: 0.8~1.2); This log represents the actual number of times the event occurred within the critical fault knowledge set. This is a frequency correction term to avoid the impact of extreme frequency values.

[0076] The core logic of reproducibility weight is that logs that occur more frequently and have a higher reproducibility rate among similar faults have higher weights and better reflect the core characteristics of the fault.

[0077] The merged weights are then multiplied and summed with the reproducible weights to obtain the total similarity score for each log entry in the critical fault knowledge set. The formula is as follows: If the overall similarity score is the total score of the critical fault knowledge set, then the scores of all logs are weighted averaged or summed to obtain the final comprehensive score.

[0078] in, The similarity is obtained from the root factor information. This represents the weighted sum of keyword similarity and time similarity.

[0079] ; For keyword similarity, For time similarity, i For the first i A log. For each log i The must-appear weight, This is the weighting coefficient, which can be equal to 0.6. This is the weighting coefficient, which can be equal to 0.4.

[0080] Set a preset threshold for the total similarity score, such as 0.85. This threshold can be adjusted according to the accuracy requirements of actual fault diagnosis. Alternatively, a TopN filtering rule, such as Top10, can be used to select all log sets with a total similarity score greater than the preset threshold from the key fault knowledge set and define them as the target fault knowledge set.

[0081] The target fault knowledge set is used as a similar knowledge base for the log groups to be merged. This knowledge base contains similar fault logs that share the same root cause, the same manifestation, high keyword similarity, and high temporal correlation with the log groups to be merged. It is the core basis for merging similar log groups. At the same time, the output of the similar knowledge base also provides accurate reference for subsequent fault diagnosis and after-sales maintenance, which meets the actual business needs of fault management in the intelligent cockpit domain.

[0082] In this embodiment of the disclosure, a log group to be merged is obtained, wherein the log group to be merged contains multiple fault characterization information; then, based on the fault characterization information, a key fault knowledge set is selected from the existing fault knowledge set; then, the importance of keywords in each log in the key fault knowledge set is calculated to obtain the keyword weight of each log; based on the keyword weight, the feature similarity weight between each log in the key fault knowledge set is determined; based on the feature similarity weight, a target fault knowledge set is selected from the key fault knowledge set, and the target fault knowledge set is used as a similar knowledge base for the log group to be merged.

[0083] In this way, the present embodiment of the invention obtains a complete knowledge base of vehicle faults by means of "fault characterization information screening + keyword importance weight calculation + feature similarity weight acquisition". This avoids misjudgment caused by pure string matching or single-dimensional matching, and significantly improves the accuracy of similar faults in subsequent merging. It effectively solves the problems of serious knowledge base redundancy and high maintenance costs, reduces the duplicate storage of similar fault entries, reduces storage resource consumption, and meets the accuracy requirements of cockpit domain fault diagnosis.

[0084] Based on the above embodiments, the method further includes: If the target fault knowledge set is not present in the critical fault knowledge set, other fault knowledge sets in the critical fault knowledge set whose total similarity score is less than or equal to a preset threshold are added to the knowledge graph of the log group to be merged, resulting in a new knowledge graph.

[0085] Optionally, if no log set with a total similarity score greater than a preset threshold is found in the key fault knowledge set, i.e., there is no target fault knowledge set, the log group to be merged is determined to be a new fault type or a low-relevance fault type.

[0086] At this point, other fault knowledge sets with a total similarity score less than or equal to a preset threshold are selected from the critical fault knowledge set. Although the similarity of such knowledge sets with the log group to be merged does not meet the standard, they are still relatively relevant fault knowledge and potential associated knowledge of new fault types.

[0087] The selected fault knowledge sets are added to the original knowledge graph of the log group to be merged, along with the other fault knowledge sets. The associations of nodes and edges in the knowledge graph are then updated to construct a new knowledge graph. Simultaneously, keyword similarity, time similarity, and log reproducibility weights are calculated for the other fault knowledge sets, and the results are stored. This new knowledge graph enables incremental merging of fault knowledge, avoiding duplicate storage of similar faults, improving the knowledge base's coverage of new fault types, and reducing the maintenance cost of the knowledge base.

[0088] This disclosure provides an effective knowledge processing solution for scenarios where there is no target fault knowledge set in the critical fault knowledge set, avoiding the omission of new types of fault knowledge and ensuring the integrity of fault knowledge.

[0089] like Figure 4 As shown, Figure 4 This is a schematic diagram illustrating the complete process of obtaining a vehicle fault knowledge base according to an embodiment of this disclosure. The specific process is as follows: Enter the log group information to be merged in the input layer; Fault phenomenon analysis and root cause analysis are performed in the similar knowledge base screening layer. Key fault knowledge sets are selected from the existing fault knowledge sets. Then, keywords are extracted from the logs of the key fault knowledge sets and the importance of the keywords is scored. Based on the importance scores of the keywords, key logs are selected, and keyword similarity is calculated for the key logs in the key fault knowledge sets. Temporal similarity is calculated for each log in the key fault knowledge sets. Then, based on the keyword similarity and temporal similarity, the total similarity score of the key fault knowledge sets is obtained. Finally, based on the total similarity score, the target fault knowledge sets with a total similarity score greater than a preset threshold (such as 0.85) are selected from the key fault knowledge sets as similar knowledge sets of the same type as the log group information to be merged.

[0090] In the merging layer, the target fault knowledge set and the log group information to be merged are merged. If no similar knowledge set of the same type as the log group information to be merged is found, a subset of the currently obtained key fault knowledge set, such as other fault knowledge sets whose total similarity score is less than a preset threshold but greater than another set threshold (e.g., 0.6), is merged into the knowledge graph where the log group information to be merged is located. Keyword similarity, time similarity, and log recurrence weight are calculated for other fault knowledge sets, and these calculation results are stored. They can be optionally stored in the knowledge graph for updating the knowledge set and parameters.

[0091] In addition, after obtaining the target fault knowledge set, it is also necessary to calculate keyword similarity, time similarity, and log reproducibility weight, and store these calculation results. You can choose to store them in the knowledge graph to update the knowledge set and parameters.

[0092] This embodiment also provides a device for acquiring a vehicle fault knowledge base. This device is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0093] This embodiment provides a device for acquiring a knowledge base of similar vehicle faults, such as... Figure 5 As shown, it includes: The first acquisition module 501 is used to acquire the log group to be merged, wherein the log group to be merged contains multiple fault characterization information; The first screening module 502 is used to screen out a set of key fault knowledge from an existing set of fault knowledge based on fault characterization information. The calculation module 503 is used to calculate the importance of keywords in each log in the critical fault knowledge set and obtain the keyword weight of each log. Module 504 is used to determine the feature similarity weights between each log in the critical fault knowledge set based on keyword weights. The second filtering module 505 is used to filter out the target fault knowledge set from the key fault knowledge set based on feature similarity weights, and to use the target fault knowledge set as a similar knowledge base for the log group to be merged.

[0094] In this embodiment, a log group to be merged is obtained, which contains multiple fault characterization information. Then, based on the fault characterization information, a key fault knowledge set is selected from an existing fault knowledge set. Next, the importance of keywords in each log within the key fault knowledge set is calculated to obtain the keyword weight for each log. Based on the keyword weight, the feature similarity weight between each log within the key fault knowledge set is determined. Based on the feature similarity weight, a target fault knowledge set is selected from the key fault knowledge set and used as the same type of knowledge base for the log group to be merged. Thus, this embodiment, through a complete process of "fault characterization information filtering + keyword importance weight calculation + feature similarity weight acquisition" to obtain a vehicle fault similarity knowledge base, avoids misjudgments caused by pure string matching or single-dimensional matching, significantly improving the accuracy of subsequent merging of similar faults. This effectively solves the problems of severe knowledge base redundancy and high maintenance costs, reduces duplicate storage of similar fault entries, reduces storage resource consumption, and meets the accuracy requirements of cockpit domain fault diagnosis.

[0095] In some optional implementations, the fault characterization information includes root factor information and fault manifestation sub-information; the first screening module 502 is used to screen out a first fault knowledge set from the existing fault knowledge set based on the fault manifestation sub-information; and to screen out a key fault knowledge set from the first fault knowledge set based on the root factor information.

[0096] In some optional implementations, the first filtering module 502 is used to vectorize the fault manifestation sub-information to obtain first vector information; based on the first vector information, a first mapping table is matched to filter out multiple fault types with a similarity greater than a first threshold with the first vector information, wherein the first mapping table is used to characterize the mapping relationship between fault type and fault manifestation; based on the fault type, a first fault knowledge set is selected from the existing fault knowledge set.

[0097] In some optional implementations, the first filtering module 502 is used to vectorize the root factor information to obtain the second vector information; and to perform similarity matching between the second vector information and the third vector information of each log in the first fault knowledge set to obtain a key fault knowledge set whose similarity is greater than the second threshold, wherein the third vector information is the information after vectorization of the root factor information of each log in the first fault knowledge set.

[0098] In some optional implementations, the feature similarity weight includes keyword similarity weight and time similarity weight in the logs. The time similarity weight is obtained from the similarity interval corresponding to the difference between the time of the fault occurrence recorded in the log and the actual time of the fault occurrence. The determination module 504 is used to select key logs from the key fault knowledge set based on the keyword weight; calculate the similarity of the keywords in the key logs to obtain the keyword similarity weight; obtain the first time of the fault and the second time of the actual occurrence of the fault for each log record in the key fault knowledge set; determine the time difference between the fault of each log record in the key fault knowledge set and the actual occurrence of the fault based on the first time and the second time; and obtain the time similarity weight based on the time difference.

[0099] In some optional implementations, the second filtering module 505 is used to merge the keyword similarity weight and the time similarity weight to obtain the merged weight; obtain the reproducibility weight of each log in the key fault knowledge set; obtain the total similarity score of the key fault knowledge set based on the merged weight and the reproducibility weight; and filter out the target fault knowledge set whose total similarity score is greater than a preset threshold from the key fault knowledge set.

[0100] In some alternative embodiments, the device further includes: The addition module is used to add other fault knowledge sets in the critical fault knowledge set whose total similarity score is less than or equal to a preset threshold to the knowledge graph of the log group to be merged, when the target fault knowledge set does not exist in the critical fault knowledge set, thus obtaining a new knowledge graph.

[0101] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0102] In this embodiment, the vehicle navigation and positioning device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0103] This disclosure also provides a vehicle having the above-described features. Figure 6 The device shown is for vehicle navigation and positioning.

[0104] Please see Figure 6 , Figure 6 This is a structural block diagram of a vehicle provided in an optional embodiment of this disclosure, such as... Figure 6As shown, the vehicle includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 6 Take a processor 10 as an example.

[0105] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0106] The memory 20 stores instructions executable by at least one processor 10 to cause at least one processor 10 to perform the method shown in the above embodiments.

[0107] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0108] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0109] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0110] This disclosure also provides a computer-readable storage medium in which the methods described in this disclosure can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium after being downloaded over a network. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium may be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium may also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code that, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0111] Although embodiments of the present disclosure have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present disclosure, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for acquiring a knowledge base of similar vehicle faults, characterized in that, The method includes: Obtain the log group to be merged, wherein the log group to be merged contains multiple fault characterization information; Based on the fault characterization information, a key fault knowledge set is selected from the existing fault knowledge set. The importance of keywords in each log in the critical fault knowledge set is calculated to obtain the keyword weight of each log. The feature similarity weights between each log in the critical fault knowledge set are determined based on the keyword weights. Based on the feature similarity weights, a target fault knowledge set is selected from the key fault knowledge set, and the target fault knowledge set is used as a similar knowledge base for the log group to be merged.

2. The method according to claim 1, characterized in that, The fault characterization information includes root factor information and fault manifestation sub-information. The step of selecting a key fault knowledge set from an existing fault knowledge set based on the fault characterization information includes: Based on the fault manifestation sub-information, a first fault knowledge set is selected from the existing fault knowledge set; The key fault knowledge set is selected from the first fault knowledge set based on the root factor information.

3. The method according to claim 2, characterized in that, The step of filtering a first fault knowledge set from the existing fault knowledge set based on the fault manifestation sub-information includes: The fault manifestation sub-information is vectorized to obtain the first vector information; Based on the first vector information, a first mapping table is matched to filter out multiple fault types that have a similarity greater than a first threshold with the first vector information. The first mapping table is used to characterize the mapping relationship between fault type and fault manifestation. Based on the fault type, the first fault knowledge set is selected from the existing fault knowledge set.

4. The method according to claim 2, characterized in that, The step of filtering the key fault knowledge set from the first fault knowledge set based on the root factor information includes: The root factor information is vectorized to obtain the second vector information; The second vector information is matched with the third vector information of each log in the first fault knowledge set to obtain a key fault knowledge set whose similarity is greater than the second threshold. The third vector information is the vectorized information of the root factor information of each log in the first fault knowledge set.

5. The method according to claim 2, characterized in that, The feature similarity weight includes keyword similarity weight and time similarity weight in the log. The time similarity weight is obtained from the similarity interval corresponding to the difference between the time of failure recorded in the log and the actual time of failure. Based on the keyword weights, the feature similarity weights between each log within the critical fault knowledge set are determined, including: Key logs are selected from the key fault knowledge set based on the keyword weights. The similarity of keywords in the key logs is calculated to obtain the keyword similarity weights; Obtain the first time of each fault recorded in the key fault knowledge set and the second time when the fault actually occurred. Based on the first time and the second time, determine the time difference between the fault and the actual occurrence of each log record in the critical fault knowledge set; The time similarity weight is obtained based on the time difference.

6. The method according to claim 5, characterized in that, The step of selecting a target fault knowledge set from the key fault knowledge set based on the feature similarity weights includes: The keyword similarity weight and the time similarity weight are combined to obtain the combined weight; Obtain the reproducibility weight of each log in the critical fault knowledge set; Based on the fused weights and the necessarily recurring weights, the total similarity score of the key fault knowledge set is obtained; Target fault knowledge sets with a total similarity score greater than a preset threshold are selected from the key fault knowledge set.

7. The method according to claim 1, characterized in that, The method further includes: If the target fault knowledge set is not present in the critical fault knowledge set, other fault knowledge sets in the critical fault knowledge set whose total similarity score is less than or equal to a preset threshold are added to the knowledge graph of the log group to be merged, and a new knowledge graph is obtained.

8. A device for acquiring a knowledge base of similar vehicle faults, characterized in that, The device includes: The first acquisition module is used to acquire the log group to be merged, wherein the log group to be merged contains multiple fault characterization information; The first screening module is used to screen out a set of key fault knowledge from an existing set of fault knowledge based on the fault characterization information. The calculation module is used to calculate the importance of keywords in each log in the key fault knowledge set and obtain the keyword weight of each log. The determination module is used to determine the feature similarity weights between each log in the critical fault knowledge set based on the keyword weights. The second filtering module is used to filter out the target fault knowledge set from the key fault knowledge set based on the feature similarity weight, and to use the target fault knowledge set as the same type of knowledge base for the log group to be merged.

9. A vehicle, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the method for acquiring a vehicle fault similar knowledge base as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the method for acquiring a vehicle fault similar knowledge base as described in any one of claims 1-7.

Images