An embedded instrument fault positioning method, server, medium and product

Abstract

An embedded instrument fault positioning method, server, medium and product, relate to the technical field of data processing. The method comprises: controlling a standard test source to output a signal to a to-be-tested embedded instrument, and acquiring data when the to-be-tested embedded instrument receives the signal; when it is detected that the to-be-tested embedded instrument has a fault, and the fault type belongs to a preset fault type set, positioning an original fault position and original fault data; screening an abnormal operation data set associated with the original fault data; if the abnormal operation data set is not empty, finding a fault position combination set corresponding to the original fault data; acquiring actual abnormal operation data sets corresponding to each fault position combination; calculating the confidence of each fault position combination; and when there is a confidence higher than a preset confidence threshold, selecting a fault position combination corresponding to the highest confidence as a target fault position combination. The above technical solution improves the accuracy of embedded instrument fault positioning.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, and particularly relates to an embedded instrument fault positioning method, a server, a medium and a product. BACKGROUND

[0002] As a key component in modern industrial control, aerospace, medical equipment and consumer electronics, etc., embedded instruments are increasingly complex and highly integrated. These instruments are usually designed to perform specific monitoring, control or communication tasks and are deeply integrated into a larger electromechanical system.

[0003] At present, in the process of detecting the embedded instrument, when it is detected that the embedded instrument has a fault, a rule-based fault diagnosis method is usually used to locate the fault of the embedded instrument. By matching and analyzing the collected instrument operation data through a pre-set fault rule library, the fault type and position are determined.

[0004] However, as a highly integrated complex system, the faults between the internal components of the embedded instrument often have complex coupling relationships. Different faults may produce new fault data through this coupling effect, resulting in a fault location obtained by rule-based reasoning that may not be the real fault location, but a fault location that is superimposed with other faults to produce fault data, thereby producing the final fault result. It may not have a fault itself, or the fault location may have a fault, but the severity of the fault is not enough to cause the final fault result alone, thereby affecting the accuracy of fault positioning. SUMMARY

[0005] The present application provides an embedded instrument fault positioning method, a server, a medium and a product, which can improve the accuracy of embedded instrument fault positioning.

[0006] In a first aspect, the application provides a method for locating faults of an embedded instrument, the method comprising: controlling a standard test source to output corresponding signals to a to-be-tested embedded instrument according to test data in a preset test case, and obtaining instrument data, running data and system logs of the to-be-tested embedded instrument when the to-be-tested embedded instrument receives the signals; when it is detected that the to-be-tested embedded instrument has a fault and the fault type belongs to a preset fault type set, locating an original fault position and original fault data according to a preset fault diagnosis case based on the instrument data, the running data and the system logs; screening out, based on a preset system causal relationship graph, a set of abnormal running data associated with generation of the original fault data from the running data except the original fault data; if the set of abnormal running data is not empty, finding a set of fault position combinations corresponding to the original fault data in a preset fault feature library, the fault feature library storing fault position combinations composed of single or multiple fault positions and fault data generated by each fault position combination under different preset running data; merging the running data corresponding to each fault position in each fault position combination to obtain actual abnormal running data sets of each fault position combination; calculating a similarity between a preset running data set corresponding to each fault position combination and the actual abnormal running data set to obtain a confidence of each fault position combination; and when there is a confidence higher than a preset confidence threshold, selecting a fault position combination corresponding to the highest confidence as a target fault position combination of the to-be-tested embedded instrument.

[0007] By outputting signals by the standard test source and synchronously obtaining the instrument data, the running data and the system logs of the to-be-tested embedded instrument, comprehensive and synchronous original information support is provided for fault location. When the fault types all belong to the preset set, the original fault position and data located based on the preset fault diagnosis case ensure the normativity and benchmarking of preliminary location. With the preset system causal relationship graph, the set of abnormal running data associated with the original fault data is screened out, which can mine associated influencing factors behind the fault and avoid one-sidedness of positioning based on single fault data. If there is a set of abnormal data, corresponding fault position combinations are obtained by searching the fault feature library, and the similarity between actual abnormal data of each combination and preset data is calculated to determine the confidence, and finally the combination with the highest confidence is selected as the target, which takes into account the complex coupling relationship between faults in the embedded instrument, reduces the misjudgment caused by single fault reasoning, and improves the accuracy of fault location.

[0008] In some embodiments of the first aspect, in some embodiments, based on the preset system causal relationship graph, the abnormal running data set associated with the original fault data is screened out from the running data, specifically comprising: based on the preset system causal relationship graph and the running data, the upstream nodes of the node corresponding to the original fault data in the causal chain related to the original fault data and the corresponding node running data are screened out; based on the actual running parameter data, the actual running environment data and the preset running rule of each upstream node, the normal running data baseline of each upstream node is generated, which is a condition set for judging whether the running data of the upstream node is normal; based on the normal running data baseline, the abnormal node running data in all node running data is screened out to obtain the abnormal running data set.

[0009] By adopting the above technical solution, the upstream nodes and running data related to the original fault data are screened out based on the preset system causal relationship graph, which can comprehensively trace the preceding factors that may affect the generation of the first fault from the perspective of the causal chain, avoiding omission of fault-related factors. The normal running data baseline is generated based on the actual running parameters, environment data and preset rules of each upstream node, which comprehensively considers real-time working conditions and design specifications, making the judgment standard of normal running more in line with the actual scene, rather than relying on fixed thresholds, thereby improving the flexibility and accuracy of abnormal data judgment. Based on the baseline, the abnormal node running data is screened out to form a set, ensuring that the collected abnormal data has causal correlation with the first fault and deviates from the normal running state, providing a reliable and accurate abnormal data basis for subsequent fault position combination matching, further enhancing the accuracy of fault positioning, avoiding misjudgment or missed judgment of abnormal data due to unreasonable normal running baseline, and thus more effectively identifying the abnormal running data set that truly causes the fault.

[0010] In some embodiments of the first aspect, in some embodiments, based on the actual running parameter data, the actual running environment data and the preset running rule of each upstream node, the normal running data baseline of each upstream node is generated, specifically comprising: obtaining the preset standard running data range, the actual running environment data and the actual running parameter data of each upstream node; based on the actual running environment, the actual running parameter data and the preset mapping relationship between working conditions and running data, the theoretical running data range of each upstream node is calculated; based on the parameter coordination relationship defined in the preset running rule and the actual running parameter data, the theoretical running data range is corrected to obtain a corrected running data range; based on the preset standard running data range, the theoretical running data range and the corrected running data range corresponding to each upstream node, the normal running data baseline of each upstream node is generated.

[0011] The technical scheme is adopted, the preset standard running data range, the actual running environment data and the actual running parameter data of each upstream node are acquired, multi-dimensional reference basis is provided for subsequent baseline generation, and it is ensured that the baseline can meet the design specification and adapt to the actual running environment. The theoretical running data range is calculated based on the actual running environment, the parameter data and the preset mapping relationship, the theoretical range can dynamically reflect the real-time working condition change, and the disconnection between the static standard and the dynamic working condition is avoided. The theoretical range is corrected by using the parameter cooperation relationship defined in the preset running rule, the mutual influence between different parameters is fully considered, the corrected range is more consistent with the cooperation characteristics of the actual system running, and the judgment error caused by ignoring the parameter association is reduced. Finally, the normal running data baseline is generated by fusing the three types of ranges of the preset standard, the theoretical calculation and the correction, the multiple influences of the static standard, the dynamic working condition and the parameter cooperation are comprehensively considered, the baseline has normativity and can adapt to complex and changeable actual running conditions, thereby the accuracy of abnormal data judgment is improved, a more reliable benchmark is provided for subsequent fault positioning, and fault misjudgment or omission caused by unreasonable normal running baseline is reduced.

[0012] In some embodiments of the first aspect, after the steps of the control standard test source outputting corresponding signals to the embedded instrument under test according to the test data in the preset test case, and acquiring instrument data, running data and system logs of the embedded instrument under test when the signals are received, the method further includes: when it is detected that the embedded instrument under test has a fault, and the fault type includes an unknown fault type not belonging to the preset fault type set, generating a to-be-tested fault feature set based on abnormal data in the instrument data, the running data and the system logs; selecting two or more preset fault types from the preset fault type set to generate a candidate fault combination set; acquiring a preset combination feature set corresponding to each candidate fault combination in a preset composite fault feature library; the preset composite fault feature library pre-stores combinations composed of two or more preset fault types, and a preset combination feature set that each combination should theoretically exhibit; sequentially performing similarity calculation on the to-be-tested fault feature set and each preset combination feature set to obtain a matching degree of each candidate fault combination; when there is a matching degree higher than a preset matching threshold, selecting the candidate fault combination with the highest matching degree as a target fault combination; finding target fault positions corresponding to each preset fault type in the target fault combination, and combining all target fault positions to obtain a target fault position combination of the embedded instrument under test.

[0013] According to the technical scheme, when the unknown fault type is detected, the generated to-be-tested fault feature set is compared with the preset combination feature set corresponding to the combination of the multiple known fault types in the preset composite fault feature library, the similarity of the two is calculated, and it is judged whether the unknown fault is formed by multiple known faults. If the unknown fault is formed by multiple known faults, based on the identified known fault types, the target fault positions corresponding to each known fault type in the preset fault position library are found, the target fault positions are combined to form a target fault position combination containing all associated fault positions, the fault position of the unknown fault formed by multiple known faults is positioned, and the comprehensiveness and accuracy of fault positioning are improved.

[0014] In combination with some embodiments of the first aspect, in some embodiments, when the to-be-tested embedded instrument is detected to have a fault, and the fault type contains an unknown fault type not belonging to the preset fault type set, after the step of generating the to-be-tested fault feature set based on the instrument data, the operation data, and the abnormal data in the system log, the method further includes: when the fault type contains a known fault type belonging to the preset fault type set, generating a fault feature set of the known fault type in the current operation environment based on a preset fault representation generation rule; if a target feature identifier does not exist in the fault feature set, adding the identifier of the target feature identifier and the corresponding feature data to the unknown fault feature set, the target feature identifier being any of all feature identifiers in the to-be-tested fault feature set; if the target feature identifier exists in the to-be-tested fault feature set, performing residual error calculation on the target feature identifier corresponding to the to-be-tested feature data and the fault feature data to obtain residual error data of the target feature identifier; adding the identifier of the target feature identifier and the corresponding residual error data to the unknown fault feature set; and updating the to-be-tested fault feature set based on the unknown fault feature set.

[0015] By adopting the technical scheme, when the fault types include both known and unknown fault types, the fault feature set of the known fault type in the current environment is generated based on the preset rule, so that the feature range that the known fault should present can be accurately defined. By comparing the existence of the target feature identifier in the to-be-tested set and the known fault feature set, the unmatched features or residual data are added to the unknown fault feature set respectively, so that the feature information of the unknown fault can be accurately separated, and the interference of the known fault features on the unknown fault features is avoided. The to-be-tested set is updated based on the unknown fault feature set, so that the to-be-tested set is more focused on the features of the unknown fault, a more accurate feature basis is provided for subsequent matching with the candidate fault combination, the positioning accuracy of the mixed fault including multiple fault types is improved, the interference of the known fault on the positioning of the unknown fault is reduced, the system can more accurately identify and locate the fault positions in the complex fault scenario, and the comprehensiveness and accuracy of fault positioning are improved.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after the step of acquiring the preset combination feature set corresponding to each candidate fault combination in the preset composite fault feature library, the method further includes: screening out real-time working condition parameter data corresponding to each feature in the preset combination feature set from the operation data and the environment data; based on the real-time working condition parameter data, correcting the preset combination feature set according to a preset influence rule of different working condition parameters on each feature to obtain a corrected combination feature set; and updating the preset combination feature set based on the corrected combination feature set.

[0017] By screening the real-time working condition parameter data corresponding to each feature in the preset combination feature set from the operation data and the environment data, the direct correlation between the real-time data used for correction and the features is ensured, and accurate basic data is provided for subsequent correction. Based on the real-time working condition parameter data and the preset influence rule, the preset combination feature set is corrected, the standard value, the fluctuation threshold and other parameters of the feature can be dynamically adjusted, the feature set is more suitable for the current actual working condition, and the problem that the theoretical feature is inconsistent with the real-time working condition is avoided. The preset set is updated based on the corrected combination feature set, so that the similarity calculation between the to-be-tested fault feature and the preset feature is based on a more actual benchmark, the feature mismatch error caused by the change of the working condition is reduced, the accuracy of the candidate fault combination matching degree calculation is improved, and the fault positioning deviation caused by the inconsistency between the static feature benchmark and the dynamic working condition is reduced.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, after the step of selecting the fault location combination corresponding to the highest confidence level as the target fault location combination of the embedded instrument under test when the confidence level is higher than a preset confidence threshold, the method further includes: when there is no preset fault location in the target fault location combination, searching for control instructions in a preset fault masking instruction library corresponding to each target fault location in the target fault location combination, and sending the control instructions to the embedded instrument under test; when the preset fault location exists in the target fault location combination, the embedded instrument under test cannot execute the control instructions; when the embedded instrument under test has executed the control instructions, searching for target test data corresponding to the target fault location combination in a preset test database; controlling the standard test source to output corresponding signals to the embedded instrument under test according to the target test data, and obtaining new operating data when the embedded instrument under test receives the signals; if there is new abnormal operating data in the new operating data, relocating the new fault location, and adding the new fault location to the target fault location combination.

[0019] By employing the above technical solution, when no preset fault location exists in the target fault location combination, the fault location can be masked by sending corresponding control commands. This allows the embedded instrument under test to continue operating without the influence of identified faults, avoiding continuous interference from the fault location to subsequent tests. This creates conditions for further verification of fault handling effectiveness and discovery of potential faults. After the instrument executes the control commands, it searches for the corresponding target test data and controls the output signal of the standard test source to acquire new operating data. This process ensures the continuity and relevance of the test and effectively verifies the effectiveness of the fault masking measures. If new abnormal operating data is found in the new operating data, the new fault location is repositioned and added to the target fault location combination. This achieves dynamic tracking and supplementary fault location, avoiding the omission of potential faults and thus improving the completeness and comprehensiveness of fault location.

[0020] In a second aspect, embodiments of this application provide a server, including: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the server to perform the method described in the first aspect and any possible implementation thereof.

[0021] Thirdly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a server, cause the server to perform the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, this application provides a computer program product that, when run on a server, causes the server to perform the method described in the first aspect and any possible implementation thereof.

[0023] Understandably, the server provided in the second aspect, the storage medium provided in the third aspect, and the computer program product provided in the fourth aspect are all used to execute the method provided in this application. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0025] 1. This application locates the original fault location and original fault data through preset fault diagnosis test cases. It then uses a preset system causal relationship graph to filter abnormal operating data sets related to the original fault data. This allows for the discovery of underlying influencing factors behind the fault, avoiding the one-sidedness of locating faults based solely on single fault data. If an abnormal data set exists, the corresponding fault location combinations are obtained by searching a fault feature library. The similarity between the actual abnormal data of each combination and the preset data is calculated to determine the confidence level. Finally, the combination with the highest confidence level is selected as the target. This approach considers the complex coupling relationships between faults in embedded instruments, reduces misjudgments caused by reasoning based on a single fault, and thus improves the accuracy of fault location.

[0026] 2. This application determines whether an unknown fault is formed by multiple known faults by comparing the generated set of features of the fault to be tested with the preset set of features corresponding to combinations of multiple known fault types in a preset composite fault feature library one by one, and calculating the similarity between the two. If the unknown fault is formed by multiple known faults, based on the identified known fault types, the target fault location corresponding to each known fault type in the preset fault location library is found, and these target fault locations are merged to form a target fault location combination that includes all associated fault locations. This enables the location of the unknown fault formed by the superposition of multiple known faults, improving the comprehensiveness and accuracy of fault location.

[0027] 3. This application enables the embedded instrument under test to continue operating while eliminating the influence of identified faults by sending corresponding control commands to mask the fault location. This avoids the continuous interference of the fault location on subsequent tests, creating conditions for further verification of the fault handling effect and discovery of potential faults. After the instrument executes the control command, the control standard test source outputs a signal to acquire new operating data. If new abnormal operating data is found in the new operating data, the new fault location is relocated and added to the target fault location combination. This achieves dynamic tracking and supplementary fault location, avoiding the omission of potential faults and thus improving the completeness and comprehensiveness of fault location. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of a system architecture to which the embedded instrument fault location method in the embodiments of this application can be applied;

[0029] Figure 2 This is a flowchart illustrating an embedded instrument fault location method in an embodiment of this application.

[0030] Figure 3 This is another flowchart illustrating the embedded instrument fault location method in this application embodiment;

[0031] Figure 4 This is a schematic diagram of an exemplary hardware structure of the server in an embodiment of this application. Detailed Implementation

[0032] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0033] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0034] Figure 1 This is a schematic diagram of a system architecture to which the embedded instrument fault location method in this application embodiment can be applied.

[0035] Please see Figure 1The embedded instrument automated testing system includes embedded instruments, standard test sources, and servers.

[0036] The server, as the core component of the system, is used to analyze and process test data transmitted by the embedded instruments and to send control commands to the embedded instruments and standard test sources. The embedded instruments receive and execute the control commands transmitted by the server and transmit test data during the testing process back to the server. The standard test sources receive the control commands transmitted by the server and send test signals to the embedded instruments according to the control commands.

[0037] Through the above system architecture, the embedded instrument automated testing system can automate the entire process of board-level testing and whole-system testing of embedded instruments. Based on preset test cases, the server sends control commands to a standard test source via a communication module, causing it to output specified analog voltage, current, and other test signals to the embedded instrument. After receiving the test signals, the embedded instrument transmits the measured values, status feedback, and other test data to the server via a communication link. The data processing module parses the data, calculates measurement errors (such as voltage and current accuracy), and determines whether the test results meet expectations. If an anomaly is detected, the server combines system logs, signal status, and other multi-source data to locate the faulty module (such as issues with screen display, communication links, or power supply paths) through logical reasoning, distinguishing between process-related, hardware, or software problems, and generating repair suggestions. After testing, the system automatically organizes test data, diagnostic results, and system logs to form a test report, supporting quality management and data traceability. This achieves efficient, accurate, and intelligent testing and fault location, improving testing efficiency, reducing labor costs, and increasing the reliability of test results.

[0038] In related technologies, during the automated testing of embedded instruments, when a fault is detected, rule-based reasoning-based fault diagnosis methods are often used to locate the fault. This involves matching and analyzing the collected instrument operating data using a pre-defined fault rule base to determine the fault type and location. However, embedded instruments are highly integrated and complex systems, and the faults between their internal components often exhibit complex coupling relationships. Different faults may generate new fault data through this coupling, meaning the fault location obtained through rule-based reasoning may not be the true fault location. Instead, it may be the result of multiple faults superimposed on each other, leading to the final fault result. The fault itself may not actually be faulty, or the fault location may exist, but its severity may not be sufficient to independently cause the final fault result, thus affecting the accuracy of fault location.

[0039] For example, suppose an embedded energy meter repeatedly displays a "communication module response timeout" fault during automated testing. A rule-based diagnostic system, based on its pre-defined fault rule base, might include a rule like: "IF 'communication module response timeout' appears in the system log AND communication error counter > 10 THEN fault location = communication module". According to this rule, the system would directly pinpoint the fault to the "communication module" and recommend repair or replacement.

[0040] However, in complex coupled systems, the real situation may be far more complex than this, and the following two scenarios may lead to misjudgment:

[0041] In the first scenario, the identified "communication module" itself might not be faulty. The real source of the problem could lie in two seemingly unrelated components. For example, the power module's capacitors might be slightly aged, causing a sudden voltage drop under high load; simultaneously, the main control MCU firmware might have a software defect that causes a sudden surge in power consumption when processing specific data frames. When the test source sends a data frame that triggers this defect, the MCU's power consumption spikes, causing the power supply voltage to drop. This happens to cause the communication module to disconnect due to insufficient power, generating a response timeout log, which the rule-based reasoning then misjudges as a communication module failure.

[0042] In the second scenario, the located "communication module" does indeed have a fault, but its severity is insufficient to trigger the final, serious failure on its own. For example, the communication module's RF antenna might have a slight loose connection (which, when alone, only reduces the communication distance by 5%-10% and does not cause a response timeout). Meanwhile, the system's clock module experiences periodic frequency drift (approximately 2 seconds per hour). When this clock drift accumulates to a certain level, it causes a time synchronization error between the communication module and external devices. The loose connection in the antenna amplifies the signal transmission delay caused by this synchronization error. The combined effect causes the communication module's interactions with external devices to frequently exceed the preset response time, ultimately triggering a "response timeout" fault. Rule-based reasoning only locates the communication module based on the fault symptoms. However, if only the communication module is replaced without fixing the clock module's frequency drift issue, the new module will still repeatedly experience timeout faults due to time synchronization discrepancies.

[0043] The embedded instrument fault location method in this application, when a fault is detected in the instrument, first locates the original fault location and original fault data using preset fault diagnosis test cases. Then, it uses a preset system causal relationship graph to filter the set of abnormal operating data related to the original fault data. If the set of abnormal operating data is not empty, it indicates that the fault may be caused by the superposition of multiple faults. At this time, the corresponding fault location combination is obtained by searching the fault feature library, and the similarity between the actual abnormal data of each combination and the preset data is calculated to determine the confidence level. Finally, the combination with the highest confidence level is selected as the target fault location, reducing misjudgments caused by reasoning about a single fault, thereby improving the accuracy of fault location.

[0044] The following is combined with Figure 2 The method of the embodiments of this application will be described below.

[0045] Please see Figure 2 This is a flowchart illustrating an embedded instrument fault location method in an embodiment of this application.

[0046] S201. The control standard test source outputs corresponding signals to the embedded instrument under test according to the test data in the preset test cases, and obtains the instrument data, running data and system log of the embedded instrument under test when it receives the signal.

[0047] Specifically, the server first establishes a connection with the standard test source via the communication module. Based on supported communication protocols such as RS485 and Modbus RTU, it sends a model query command to the standard test source to obtain the specific model information of the tester. Then, based on the obtained model, it matches and loads the corresponding driver from a preset driver library (such as an ElectricSource driver adapted to a specific model). Next, it obtains preset test cases corresponding to the model of the embedded instrument under test, parses the test data (including specific parameters such as voltage and current, and output timing requirements), and calls the standard interface provided by the loaded driver to bind and convert the test data according to the protocol format specified by the driver. This generates an instruction template containing an instruction header and parameter placeholders. After protocol encapsulation and checksum addition, the control instructions from the standard test source are obtained. Finally, the control instructions are transmitted to the standard test source via the communication module, controlling the test source to output corresponding signals (such as analog voltage signals, current signals, etc.) to the embedded instrument under test according to the test data in the test cases.

[0048] When the standard test source outputs a signal, the server's internal communication module establishes a communication link with the embedded instrument under test (DUT). Through this link, it receives in real-time instrument data, operational data, and system logs generated by the DUT upon receiving the signal. Instrument data includes displayed values, measurement results, and the on / off status of indicator lights. Operational data covers the operating voltage, operating current, real-time temperature, data processing speed, and communication response time of each module within the instrument. The system log contains records of the instrument's startup process, inter-module interactions, error codes, and abnormal interruption messages.

[0049] S202. When a fault is detected in the embedded instrument under test, and the fault types all belong to the preset fault type set, the original fault location and original fault data are located based on the instrument data, running data and system logs, according to the preset fault diagnosis test cases.

[0050] Specifically, the server first analyzes the acquired instrument data, operating data, and system logs. By comparing the measured values ​​in the instrument data with the preset standard range, checking whether the parameters of each module in the operating data exceed the normal threshold, and filtering the error identifiers (such as error codes and abnormal interruption records) in the system logs, it determines whether the embedded instrument under test has a fault.

[0051] If a fault is detected in the embedded instrument under test, fault feature information is first extracted from the instrument data, operating data, and system logs. This includes abnormal values ​​in the instrument data, module operating parameters that exceed the normal range in the operating data (such as voltage fluctuations in a circuit), error codes recorded in the system logs, and fault occurrence timestamps. Then, the extracted fault features are matched with the fault features corresponding to each fault type in a preset fault type set to identify the specific fault type.

[0052] When all identified fault types are detected to belong to a preset fault type set, a preset fault diagnosis test case is invoked. This test case contains rules or historical cases corresponding to fault phenomena, fault locations, and fault data. The server compares the extracted instrument data, operating data, and key information from the system logs with the rule conditions or case features in the fault diagnosis test case one by one. If the conditions of a certain rule are fully met or the feature matching degree with a certain case exceeds a threshold, the fault location corresponding to that rule or case is determined as the original fault location, and the abnormal data directly associated with that fault location (such as abnormal operating parameters of the corresponding module and error log details) is determined as the original fault data.

[0053] S203. Based on the preset system causal relationship map, filter out the set of abnormal operating data other than the original fault data that are related to the generation of the original fault data in the operating data.

[0054] The preset system causal relationship graph presents the causal relationships between various modules and data items of the embedded instrument under test, with nodes representing modules or data and directed edges representing causal influence relationships.

[0055] Specifically, based on a pre-defined system causal relationship graph, starting from the node corresponding to the original fault data, all upstream nodes in the causal chain are traced, i.e., the preceding nodes that may affect the original fault data. The corresponding operational data for these upstream nodes is then extracted from the operational data. When tracing upstream nodes, the tracing depth can be set according to actual needs, such as tracing only directly related upstream nodes or including multi-level upstream nodes with indirect relationships.

[0056] Then, for each upstream node, its operating data is compared with a preset normal operating baseline to determine whether it is abnormal data. The normal operating baseline includes a set of conditions for judging whether the operating data is normal, such as the normal fluctuation range of voltage, the threshold range of temperature, and the upper limit of module response time. If the operating data of an upstream node exceeds the condition range defined by the normal operating baseline, the data is determined to be abnormal node operating data.

[0057] In some embodiments, to improve the accuracy of anomaly detection, the normal operating baseline can be dynamically adjusted according to actual conditions. First, operational data, environmental data (such as temperature, humidity, and air pressure), and actual operational parameter data (such as load rate and operating mode) of each upstream node are acquired. Then, based on the acquired data and a preset dynamic adjustment rule, the normal operating baseline is updated.

[0058] The dynamic adjustment rules include environmental parameter compensation rules and time series prediction rules.

[0059] Environmental parameter compensation rules are rules that establish a mapping relationship between environmental parameters and normal operating data for upstream nodes that are significantly affected by environmental factors (such as temperature, humidity, air pressure, etc.), and dynamically correct the normal operating baseline through real-time environmental data.

[0060] The time series forecasting rules are based on the historical time series characteristics of the upstream node's operational data. Using time series forecasting models such as ARIMA and LSTM, the models are trained on the historical operational data of each upstream node, enabling them to capture the trend, periodicity, and randomness of the data. During real-time operation, the model predicts the confidence interval for the operational data of the next period based on the operational data of the most recent period, and this interval is used as part of the dynamic baseline. If the actual operational data continuously exceeds the predicted interval, it is judged as an anomaly. Simultaneously, the server re-calls the model at preset intervals to generate new prediction intervals, achieving rolling updates of the baseline to adapt to slow drift or sudden fluctuations in the operational data.

[0061] S204. If the abnormal operation data set is not empty, search for the set of fault location combinations in the preset fault feature library that corresponds to the original fault data.

[0062] The fault feature library stores fault location combinations consisting of single or multiple fault locations, as well as fault data generated by each fault location combination under different preset operating data (including various parameter data, status data, and time sequence data during equipment operation).

[0063] Specifically, if the abnormal operation data set is not empty, the original fault data is converted into a first fault feature vector (e.g., converting abnormal values, duration, and impact range into numerical components). Then, the cosine similarity between the original fault data feature vector and the standard feature vector corresponding to each fault location combination in the preset fault feature library is calculated, and all fault combinations with similarity higher than a preset threshold are selected to obtain the fault location combination set.

[0064] If the abnormal operation data set is empty, it means that there are no other fault locations in the instrument besides the original fault location. The original fault location is taken as the target fault location of the embedded instrument under test and displayed on the display interface.

[0065] S205. Merge the operational data corresponding to each fault location in each fault location combination to obtain the actual abnormal operational data set of each fault location combination.

[0066] Specifically, the specific fault locations contained in each fault location combination are first analyzed, and all operational data generated during operation are associated with each fault location, including the operating voltage, current, real-time temperature, response time, data transmission rate, and abnormal status indicators of each fault location.

[0067] Then, for each fault location combination, the abnormal operation data corresponding to all fault locations within the combination are summarized and merged. During the merging process, key metadata of each data point is retained, such as the fault location identifier to which the data belongs, the timestamp of the data generation, the data type (e.g., voltage, temperature), and the specific abnormal value or status description (e.g., "communication timeout"). The merging method can adopt a structured data format (e.g., a table or JSON array), where the column fields of the table include fault location, data type, abnormal value, timestamp, etc., and each row corresponds to one abnormal operation data point; or, subsets can be constructed according to fault location, and then integrated into the overall actual abnormal operation data set for that fault location combination.

[0068] S206. Calculate the similarity between the preset operating data set and the actual abnormal operating data set corresponding to each fault location combination to obtain the confidence level of each fault location combination.

[0069] Specifically, the data in the preset operational data set and the actual abnormal operational data set are first classified by type. For numerical data, the deviation rate from the preset range is calculated, and for state-type data, consistency is determined using preset state matching rules (such as exact match or fuzzy match). Then, weights are assigned to different data types, and a single-dimensional similarity is calculated based on the deviation rate and state matching results. Finally, the overall similarity between the two sets is obtained by weighted summation. This similarity is the confidence level of the fault location combination; a higher confidence level indicates a higher degree of matching between the combination and the actual fault.

[0070] For data containing time-series features (such as parameter change curves over time), the server aligns two time-series curves using the Dynamic Time Warping (DTW) algorithm, calculates the distance between the curves, and the smaller the distance, the higher the time-series similarity. This distance is then included in the overall confidence calculation. Meanwhile, for missing data, it is handled according to preset rules (such as assigning a basic similarity) to ensure that all dimensions participate in the confidence calculation, ultimately obtaining the confidence values ​​for each fault location combination.

[0071] S207. When there is a confidence level higher than the preset confidence level threshold, select the fault location combination corresponding to the highest confidence level as the target fault location combination of the embedded instrument under test.

[0072] Specifically, when there is a confidence level higher than the preset confidence level threshold, all fault location combinations with confidence levels higher than the preset confidence level threshold are sorted from high to low, and the combination with the highest confidence level is selected as the target fault location combination of the embedded instrument under test.

[0073] During the sorting process, if multiple fault location combinations have similar confidence values ​​(e.g., the difference between the confidence values ​​of two combinations is within a preset fluctuation range) and both are higher than the preset confidence threshold, the server will further refer to the historical matching success rate of these combinations in the preset fault feature library (i.e., the proportion of times the combination has been verified as correct in past fault diagnosis) and select the combination with the higher historical matching success rate as the target fault location combination.

[0074] When there is no confidence level higher than the preset confidence threshold, it indicates that there may be other unknown fault combinations that affect the original fault location and generate the original fault data.

[0075] At this point, the locations of all modules or data points involved in the abnormal operation data set are marked as potential fault sources. Based on these potential fault sources, new fault location combinations that do not belong to the fault location combination set are generated. During the generation of new fault combinations, to avoid too many combinations, the server uses a preset system causal relationship graph to perform preliminary pruning, excluding combinations that are logically or physically impossible to jointly cause the original fault data.

[0076] Next, simulation verification based on a causal model is performed for each newly generated fault location combination. The server utilizes a more refined system causal relationship graph (which not only includes correlations but also embeds simplified behavioral models or transfer functions) to simulate the transmission and evolution of the system state when the new fault location combination fails simultaneously. Specifically, this includes: firstly, extracting the associated nodes and causal transmission paths of each fault location in the new fault location combination from the preset system causal relationship graph, clarifying the interaction relationships (such as series influence, parallel superposition, etc.) and influence weights between each fault location. Then, the actual values ​​of each data point in the abnormal operation dataset are used as initial input parameters and substituted into the behavioral models corresponding to each fault location (such as the voltage drop model of the power module, the signal attenuation model of the communication module, etc.). The output state variables at each fault location are obtained through model calculation. These state variables are then transmitted sequentially according to the causal propagation path. The state variables are converted by combining the physical characteristic models of each node (such as voltage conversion efficiency formula and signal attenuation coefficient) (such as converting voltage fluctuation values ​​into module operating frequency offset). The state variables transmitted through multiple paths are superimposed according to fault coupling rules (such as linear superposition to calculate total delay and nonlinear amplification to calculate error rate) to simulate the stepwise diffusion process of the fault in the system. Finally, the response threshold model of the original fault location (such as communication timeout judgment formula) is substituted into the model to obtain the fault data generated by the original fault location under the coupling influence of the new fault location combination.

[0077] Then, the simulated fault data is compared and scored with the original fault data to calculate the similarity between the simulated fault data and the original fault data. If the similarity exceeds a preset threshold, the new fault location combination corresponding to the highest similarity is selected as the target fault location combination; if no similarity exceeds the preset threshold, a manual intervention process is triggered, and a complete report containing all analysis processes (including failed inference attempts) is submitted to the engineer for the engineer to locate the fault.

[0078] In this embodiment, after locating the original fault location and original fault data through preset fault diagnosis test cases, a preset system causal relationship graph is used to filter the abnormal operation data set related to the original fault data. This allows for the discovery of the associated influencing factors behind the fault, avoiding the one-sidedness of locating faults based solely on a single fault data point. If the abnormal data set is not empty, the corresponding fault location combination is obtained by searching the fault feature library, and the similarity between the actual abnormal data of each combination and the preset data is calculated to determine the confidence level. Finally, the combination with the highest confidence level is selected as the target. This approach considers the complex coupling relationships between faults in embedded instruments, reduces misjudgments caused by reasoning based on a single fault, and thus improves the accuracy of fault location.

[0079] The following is combined with Figure 3The methods of the embodiments of this application will be further explained below.

[0080] Please see Figure 3 This is another flowchart illustrating the embedded instrument fault location method in this application embodiment.

[0081] S301. The control standard test source outputs corresponding signals to the embedded instrument under test according to the test data in the preset test cases, and obtains the instrument data, running data and system log of the embedded instrument under test when it receives the signal.

[0082] S302. When a fault is detected in the embedded instrument under test, and the fault types all belong to the preset fault type set, the original fault location and original fault data are located based on the instrument data, running data and system logs, according to the preset fault diagnosis test cases.

[0083] Steps S301-S302 and Figure 2 Steps S201-S202 in the illustrated embodiment are similar and can be found in the descriptions of steps S201-S202, which will not be repeated here.

[0084] S303. Filter out the upstream nodes and corresponding node operation data of the nodes corresponding to the original fault data in the causal chain related to the original fault data.

[0085] Based on the preset system causal relationship map and operation data, the upstream nodes and corresponding node operation data of the nodes corresponding to the original fault data are selected from the causal chain related to the original fault data.

[0086] Specifically, the server starts from the node corresponding to the original fault data and traverses backwards along the directed edges of the preset system causal relationship graph, identifying all direct and indirect upstream nodes layer by layer. During the traversal, a level label is set for each node (e.g., the direct upstream node is level 1, and the upstream of the direct upstream node is level 2), and the traversal terminates according to a preset tracing depth threshold. After the traversal is completed, the server extracts the running data of all nodes corresponding to these upstream nodes from the running data, generating an upstream node running data set.

[0087] S304. Obtain the preset standard operating data range, actual operating environment data, and actual operating parameter data of each upstream node.

[0088] Specifically, the server first retrieves the preset standard operating data range of each upstream node from the preset database. This range is determined based on the design parameters of the upstream nodes, historical normal operating data, and expert experience, and covers the normal thresholds of various parameters such as voltage, current, temperature, and response time.

[0089] At the same time, the communication module acquires actual operating environment data collected by the sensor module of the embedded instrument under test or external environmental monitoring equipment, including but not limited to external factors such as ambient temperature, humidity, air pressure, and electromagnetic interference intensity that may affect the operating status of upstream nodes.

[0090] In addition, actual operating parameter data of each upstream node is extracted from the operating data. These data are real-time operating indicators of the upstream nodes under the current operating conditions, including operating voltage, load rate, data transmission rate, number of module start-stops, etc.

[0091] S305. Calculate the theoretical operating data range of each upstream node.

[0092] Based on the actual operating environment, actual operating parameter data, and the preset mapping relationship between operating conditions and operating data, the theoretical operating data range of each upstream node is calculated.

[0093] Specifically, the server first obtains a preset mapping relationship between operating conditions and operational data. This mapping relationship is constructed based on historical and experimental data, and includes standard ranges of operational data for each upstream node under different operating conditions (such as high load, low power consumption, standby, etc.), as well as association rules between operating conditions and actual operating environment data (such as temperature and humidity) and actual operating parameter data (such as load rate and operating frequency). For example, high load conditions typically correspond to higher processor temperatures and power supply voltage ranges, while low power consumption conditions correspond to lower current ranges.

[0094] Next, the server inputs the acquired actual operating environment data and actual operating parameter data into the mapping relationship model, and determines the specific operating condition of the current system through rule matching or algorithm calculation. If the load rate is ≥70% and the ambient temperature is ≥20℃ in the actual operating parameters, then it is matched with "high load normal temperature operating condition".

[0095] Then, based on the defined operating conditions, the basic theoretical operating data range of each upstream node under that operating condition is extracted from the mapping relationship. At the same time, the basic range is corrected by combining actual operating environment data: for nodes that are significantly affected by environmental factors (such as temperature sensors), the range boundary is adjusted by a preset environmental compensation coefficient; for nodes that are not affected by the environment or are only slightly affected (such as memory modules), the basic range corresponding to the operating condition is directly adopted.

[0096] Finally, by integrating and correcting the operational data range, the theoretical operational data range of each upstream node under the current actual operating environment and conditions is obtained.

[0097] S306. Correct the theoretical operating data range to obtain the corrected operating data range.

[0098] Based on the parameter coordination relationships defined in the preset operating rules and the actual operating parameter data, the theoretical operating data range is corrected to obtain the corrected operating data range.

[0099] Specifically, the server first parses the parameter coordination relationships defined in the preset operating rules. These relationships describe the correlation between operating parameters of different upstream nodes (e.g., "the power module voltage is positively correlated with the processor temperature; theoretically, the temperature increases by 2°C for every 0.1V increase in voltage"), the constraints on different parameters of the same node (e.g., "the transmission rate of the communication module is negatively correlated with the bit error rate; the bit error rate must be below 0.1% when the rate exceeds 1Mbps"), and the parameter linkage rules across modules (e.g., "when the sensor sampling frequency increases, the storage capacity of the data cache module needs to be expanded by 20% synchronously"). These coordination relationships are stored in the form of a rule base, containing information such as parameter type, correlation direction, and influence coefficient.

[0100] Then, the actual operating parameter data of each upstream node is extracted and matched with the parameter coordination relationship. For parameter combinations with positive or negative correlation (such as power supply voltage and cooling fan speed), the deviation between the actual parameter value and the theoretical value of the coordination relationship is calculated (e.g., if the actual power supply voltage is 5.2V, the fan speed should theoretically be ≥3000rpm according to the coordination relationship; if the actual speed is 2800rpm, the deviation is -200rpm). Based on the magnitude of the deviation, the theoretical operating data range is adjusted accordingly: if the deviation is within the allowable range (e.g., ≤±5%), the range boundary is finely adjusted (e.g., the theoretical upper limit of the fan speed is increased by 50rpm); if the deviation exceeds the allowable range, the tolerance range of the theoretical range is expanded (e.g., the speed range is expanded from 2800-3200rpm to 2700-3300rpm) to adapt to the coordination characteristics of the actual parameters.

[0101] For cross-module parameter linkage scenarios (such as linkage between the accuracy of the A / D conversion module and the stability of the reference voltage module), the server verifies whether the actual operating parameters meet the linkage rules. If a module parameter deviates from the linkage expectation (such as reference voltage fluctuation exceeding 0.05V), the theoretical operating data range of the associated module (A / D conversion module) is simultaneously widened (e.g., the conversion accuracy error range is expanded from ±0.1% to ±0.2%). After correction, all adjusted ranges are integrated to generate the corrected operating data range for each upstream node.

[0102] S307. Generate normal operating data baselines for each upstream node.

[0103] Based on the preset standard operating data range, theoretical operating data range, and corrected operating data range corresponding to each upstream node, a normal operating data baseline for each upstream node is generated.

[0104] Among them, the normal operation data baseline is a set of conditions for judging whether the operation data of the upstream node is normal.

[0105] Specifically, the preset standard operating data range, theoretical operating data range, and modified operating data range are fused to generate a normal operating data baseline. During the fusion process, for numerical parameters, the intersection of the three ranges is used as the judgment condition; for state parameters, the state descriptions corresponding to the three ranges are combined. If there is a conflict, the state description of the modified operating data range prevails, ensuring that the baseline conforms to the actual parameter coordination relationship. The final generated normal operating data baseline is a set containing multiple conditions, each condition corresponding to a judgment logic.

[0106] Among them, the preset standard operating data range is a static threshold determined based on design specifications and historical normal data, the theoretical operating data range is a dynamic range calculated in combination with the actual operating environment and working conditions, and the corrected operating data range is the range adjusted after considering the parameter coordination relationship.

[0107] S308. Filter out the abnormal node running data from all node running data to obtain the abnormal running data set.

[0108] Based on the baseline of normal operation data, abnormal node operation data is filtered out from all node operation data to obtain an abnormal operation data set.

[0109] Specifically, the server compares the actual operating data of each upstream node with the condition set in the normal operating data baseline one by one to determine whether the data is abnormal. For numerical parameters, it checks whether the conditions are met; if not, it is determined to be abnormal data. For state parameters, it directly compares whether the actual state is consistent with the state description defined in the baseline; if they are inconsistent, it is determined to be abnormal. For cross-parameter association conditions, it verifies whether the actual parameter combination satisfies the cooperative relationship; if not, it is determined to be abnormal.

[0110] Finally, the runtime data of all nodes that are identified as abnormal are added to the abnormal runtime data set.

[0111] S309. If the abnormal operation data set is not empty, search the set of fault location combinations in the preset fault feature library that corresponds to the original fault data.

[0112] S310. Merge the operating data corresponding to each fault location in each fault location combination to obtain the actual abnormal operating data set of each fault location combination.

[0113] S311. Calculate the similarity between the preset operating data set and the actual abnormal operating data set corresponding to each fault location combination to obtain the confidence level of each fault location combination.

[0114] S312. When there is a confidence level higher than the preset confidence level threshold, select the fault location combination corresponding to the highest confidence level as the target fault location combination of the embedded instrument under test.

[0115] Steps S309-S312 and Figure 2 Steps S204-S207 in the illustrated embodiment are similar and can be found in the descriptions of steps S204-S207, which will not be repeated here.

[0116] S313. When a fault is detected in the embedded instrument under test, and the fault type includes an unknown fault type that does not belong to the preset fault type set, a fault feature set to be tested is generated.

[0117] A set of fault characteristics to be tested is generated based on instrument data, operational data, and abnormal data in system logs.

[0118] Specifically, when a fault is detected in the embedded instrument under test, and the fault type includes an unknown fault type that does not belong to the preset fault type set, the server extracts all abnormal data from the instrument data, operating data, and system logs. Numerical abnormal data is standardized (e.g., normalized to the [0,1] interval), and textual abnormal data is converted into word vectors (e.g., using the Word2Vec model). Then, the K-means clustering algorithm is used to cluster the processed data, grouping the closest abnormal data into the same feature group. After clustering, the center vector (numerical) or keyword (textual) of each feature group is extracted and combined to form the feature set of the fault under test.

[0119] S314. When the fault type includes a known fault type belonging to the preset fault type set, generate a fault feature set of the known fault type in the current operating environment.

[0120] Based on preset fault characterization generation rules, a set of fault features for known fault types in the current operating environment is generated.

[0121] Specifically, when the fault type includes a known fault type belonging to a preset fault type set, the identified known fault type is obtained, and the preset fault characterization generation rules are retrieved. The preset fault characterization generation rules include two types of generation rules, corresponding to the feature generation paths of the parametric physical model and case reasoning, respectively.

[0122] For generation rules using parametric physical models, the server loads mathematical models associated with known fault types from the rule base. The model input is the current operating environment data, and the output is the theoretical value or range of characteristic parameters for that fault type. For example, for a "power module ripple exceeding the standard fault," the model "ripple voltage = 0.02 × load current + 0.01 × (ambient temperature - 25) + 0.05" is called. Substituting the current load current of 2.5A and ambient temperature of 60℃ into the calculation, the theoretical ripple voltage is obtained as 0.45V. Simultaneously, based on the model error range, the characteristic value interval is determined to be 0.45V ± 0.05V. For cases where the model output contains multiple parameters (such as ripple voltage and frequency characteristics), the server calculates the theoretical values ​​of all parameters using the same logic, forming a numerical feature subset.

[0123] For the generation path that employs case-based reasoning and adaptive rules, the server first retrieves historical cases from the fault case library that match the currently known fault type and have the highest similarity to the operating environment data (e.g., by calculating the Euclidean distance between the current ambient temperature, load rate, and historical cases). Then, it calls preset adaptive rules to correct the fault characteristics in the cases to compensate for the differences between the current environment and the base case environment. For example, an adaptive rule could be defined as "if the current temperature is 10°C higher than the base case, then increase the expected leakage current characteristic value by 5%".

[0124] Finally, the server integrates the features generated in the two ways, removes duplicate features (such as numerical features based on the results of the parameterized model and time-series features based on the results of case reasoning), supplements feature metadata, and forms a fault feature set containing numerical features, state features, and time-series features.

[0125] S315. If the target feature identifier does not exist in the fault feature set, add the identifier of the target feature identifier and the corresponding feature data to the unknown fault feature set.

[0126] The target feature identifier is any one of all feature identifiers in the set of fault features to be tested.

[0127] Specifically, the server iterates through each feature identifier in the set of fault features to be tested, using it as the target feature identifier, and compares it one by one with the feature identifiers in the set of fault features. The comparison method is exact matching, that is, checking whether the string, number, or hash value of the target feature identifier is completely consistent with any identifier in the set of fault features.

[0128] If the target feature identifier does not exist in the fault feature set, the server adds the target feature identifier and its corresponding feature data (such as specific abnormal values, text descriptions, time series information, etc.) to the unknown fault feature set. During the addition process, the server records the source information (such as whether the feature is taken from instrument data, operational data, or system logs), extraction timestamp, and data reliability (such as numerical reliability based on sensor accuracy, and log record integrity score) for each entry in the unknown fault feature set. For cases where the same target feature identifier appears multiple times (such as the same feature being extracted at different times), the server retains all instances to reflect the temporal changes of the feature, ensuring that the unknown fault feature set can comprehensively reflect abnormal features not covered by known features.

[0129] S316. If the target feature identifier exists in the set of fault features to be tested, perform residual calculation on the test feature data corresponding to the target feature identifier and the fault feature data to obtain the residual data of the target feature identifier.

[0130] Specifically, if the target feature identifier does not exist in the fault feature set, the test feature data of the target feature identifier and the fault feature data are preprocessed to make the test feature data and the fault feature data match in terms of data structure and dimension.

[0131] Then, based on the data types of the test feature data and the fault feature data, different residual calculation methods are performed to obtain the residual data of the target feature identifier. For numerical scalar data (such as voltage 3.8V and standard value 3.3V), the algebraic difference is directly calculated to obtain the residual data (0.5V); for vector data (such as aligned time-series waveforms), element-wise subtraction is performed to generate residual vectors of the same dimension (such as a vector composed of the current difference of each sampling point); for text data, differential keywords are extracted by comparing bag-of-words models or TF-IDF vectors (such as the test text containing "overload" while the standard text does not), to obtain the residual data.

[0132] S317. Add the identifier of the target feature identifier and the corresponding residual data to the unknown fault feature set.

[0133] Add the target feature identifier and its corresponding residual data to the unknown fault feature set.

[0134] S318. Update the set of fault features to be tested.

[0135] Based on the unknown fault feature set, all data in the fault feature set to be tested are replaced with data from the unknown fault feature set to obtain the updated fault feature set to be tested.

[0136] S319. Select two or more preset fault types from the preset fault type set to generate a candidate fault combination set.

[0137] Specifically, the fault association rule base is first retrieved. This rule base stores the association relationships between all preset fault types in matrix form, along with a description of the association reasons.

[0138] For binary combination generation, the server traverses the association matrix and filters out all fault type pairs with an association degree greater than the strong association threshold to obtain binary candidate fault combinations. If the number of strongly associated pairs is less than the preset lower limit, the strong association threshold is lowered to supplement combinations, ensuring basic coverage of binary combinations.

[0139] For combinations of ternary faults or higher, the server first selects the fault type with the highest correlation from strongly correlated pairs as the core fault. Then, it retrieves the secondary related faults of the core fault and constructs a correlation chain of "core fault → primary correlation → secondary correlation" to obtain multiple candidate fault combinations of ternary faults or higher. If different correlation chains generate the same set of fault types, the first generated combination is retained and the source of the duplication is marked.

[0140] Finally, all candidate fault combinations are integrated to obtain a candidate fault combination set, which is stored according to the combination size (binary, ternary, quaternary and above). Each combination is accompanied by complete association chain information and association degree information.

[0141] S320. Obtain the preset combination feature set corresponding to each candidate fault combination in the preset composite fault feature library.

[0142] The preset composite fault feature library stores combinations consisting of two or more preset fault types, as well as the preset combination feature set that each combination should theoretically exhibit.

[0143] Specifically, for each candidate fault combination, the corresponding preset combination feature set is retrieved from the preset composite fault feature library. The preset composite fault feature library is stored in key-value pair format, where the key is the fault combination identifier and the value is the corresponding preset combination feature set. Each feature set contains multiple feature entries, and each entry contains metadata such as feature identifier, feature type (e.g., numerical, textual, time-series), standard value or value range, and feature weight.

[0144] S321. Filter out the real-time operating condition parameter data that corresponds to each feature in the preset combination feature set from the operating data and environmental data.

[0145] Specifically, for each feature in the preset set of combined features, the feature identifier of the feature is first identified, and then the preset operating parameters and preset environmental parameters corresponding to the feature identifier are found in the table corresponding to the preset feature identifier and real-time operating parameters. The real-time operating parameters include operating parameters and environmental parameters, etc.

[0146] Then, based on the data types of the preset operating parameters and preset environmental parameters, different filtering methods are executed to obtain the parameter data corresponding to each preset operating parameter and preset environmental parameter. For numerical parameters, the latest preset number of sampled values ​​are extracted from real-time data, and statistical features (such as mean, peak value, and root mean square value) are calculated. For textual parameters (such as error codes), the latest entries containing specific keywords are extracted from system logs, and keyword matching and semantic parsing are performed. For time-series parameters, a data segment of the most recent complete cycle (such as one cycle of a current waveform) is extracted, and the sampling points are aligned to the standard time axis through linear interpolation. If a feature does not have a directly corresponding parameter in the operating data or environmental data, the server attempts to derive it through relevant parameters or extract substitute values ​​under similar operating conditions from historical data.

[0147] Finally, the parameter data of each preset operating parameter and each preset environmental parameter obtained from the screening are integrated to obtain the real-time operating condition parameter data of each feature.

[0148] S322. Based on real-time operating condition parameter data, the preset combined feature set is modified according to the preset influence rules of different operating condition parameters on each feature to obtain the modified combined feature set.

[0149] Specifically, the server first parses a pre-defined influence rule library, which stores the influence relationship between each feature and the operating condition parameter in key-value pairs.

[0150] Then, each feature in the preset combined feature set is traversed, and all associated influence rules are extracted. For numerical influence rules, real-time operating condition parameter data is substituted into the preset rule formula to calculate the corrected feature value. For logical influence rules, it is determined whether the real-time operating condition parameters meet the rule conditions. If they do, the fluctuation threshold or boundary conditions of the feature are adjusted according to the rules; otherwise, the original feature parameters remain unchanged. For composite influence rules (such as multiple operating condition parameters jointly affecting a feature), a weighted calculation method is used to calculate the individual influence value of each operating condition parameter on the target feature, and then the final corrected value is obtained by weighted summation according to the preset weight coefficients. If the influence of a certain operating condition parameter on the feature has a threshold condition, it is first determined whether the parameter meets the threshold requirement. Parameters that meet the condition participate in the weighted calculation, while those that do not meet the condition are considered to have no influence (influence value is 0).

[0151] Finally, the corrected data corresponding to each feature are integrated to obtain the corrected combined feature set.

[0152] S323. Update the preset combination feature set.

[0153] Based on the modified combined feature set, all data in the preset combined feature set are replaced with data in the modified combined feature set to obtain the updated preset combined feature set.

[0154] S324. The similarity calculation is performed sequentially between the set of features of the fault to be tested and each preset combination feature set to obtain the matching degree of each candidate fault combination.

[0155] Specifically, the server first preprocesses the preset combined feature set and the fault feature set to be tested to ensure consistency in feature dimensions. For features present in the fault feature set to be tested but missing in the preset combined feature set, feature mapping or generalization rules are used to supplement them (e.g., mapping "specific model error code" to "general equipment fault"). For features present in the preset combined feature set but missing in the test set, if the feature weight is below a threshold, it is ignored; otherwise, its value is set to a default value (e.g., numerical types are set to the standard value, and text types are set to "no anomaly").

[0156] Then, different similarity calculation methods are applied to different types of features. For numerical features, the relative error between the test value and the preset standard value is calculated, and then the error is mapped to a similarity score using a Gaussian function (the smaller the error, the higher the score). For textual features, edit distance or semantic similarity algorithms (such as Word Mover's Distance) are used to calculate the similarity between the test text and the preset text. For time-series features, the Dynamic Time Warping (DTW) algorithm is used to calculate the similarity between two curves, or the statistical features of the curves (such as rise time and fluctuation frequency) are extracted for comparison.

[0157] Next, for each candidate fault combination, the similarity scores of each feature are weighted and summed according to the feature weights in the preset feature set to obtain the matching degree of each candidate fault combination.

[0158] S325. When there is a matching degree higher than the preset matching threshold, select the candidate fault combination with the highest matching degree as the target fault combination.

[0159] Specifically, when there is a matching degree higher than the preset matching threshold, all fault location combinations with matching degrees higher than the preset matching threshold are sorted from high to low according to the matching degree, and the combination ranked first is selected as the target fault combination.

[0160] During the ranking process, if multiple candidate fault combinations have similar matching scores (e.g., the difference between the confidence scores of two combinations is within a preset fluctuation range) and both are higher than the preset matching threshold, the completeness of feature matching in these combinations is compared. That is, the percentage of features in each combination that completely match the feature set of the fault to be tested is counted, and the combination with the higher percentage is ranked first. If the feature matching completeness is the same, the matching quality of key features is further compared. Key features are determined by a preset weight table, and the average similarity of key features is calculated. The combination with the higher average similarity is ranked first. If the matching quality of key features is still consistent, the historical fault case database is retrieved, and the actual occurrence frequency of these candidate combinations in the historical records is counted. The combination with the higher occurrence frequency is ranked first.

[0161] S326. Locate the target fault location corresponding to each preset fault type in the target fault combination, and merge all target fault locations to obtain the target fault location combination of the embedded instrument under test.

[0162] Specifically, for each preset fault type in the target fault combination, the steps S302-S312 described above are executed to obtain the target fault location corresponding to each preset fault type in the target fault combination.

[0163] Then, the target fault locations corresponding to each preset fault type are merged to obtain the target fault location combination of the embedded instrument under test.

[0164] S327. When there is no preset fault location in the target fault location combination, search for the control command corresponding to each target fault location in the target fault location combination in the preset fault masking command library, and send the control command to the embedded instrument under test.

[0165] When a preset fault location exists in the target fault location combination, the embedded instrument under test cannot execute control commands.

[0166] Specifically, first, the preset fault locations corresponding to each fault location in the target fault location combination are obtained. If no preset fault location exists in the target fault location combination, the control commands corresponding to each target fault location in the target fault location combination are searched in the preset fault masking command library, and the control commands are sent to the embedded instrument under test through the communication module. After receiving the command, the embedded instrument under test executes the corresponding operation to mask the fault location corresponding to the command, allowing the instrument to continue automated testing without the influence of the fault location.

[0167] S328. After the embedded instrument under test has executed the control command, search the preset test database for the target test data corresponding to the target fault location combination.

[0168] Specifically, the execution status is verified by receiving the command execution results returned by the instrument. If all execution statuses are "completed", it means that the embedded instrument under test has completed the control command.

[0169] After detecting that the embedded instrument under test has completed the control command, the target test data corresponding to the target fault location combination is searched in the preset test database.

[0170] S329. The control standard test source outputs the corresponding signal to the embedded instrument under test according to the target test data, and obtains the new operating data of the embedded instrument under test when it receives the signal.

[0171] Specifically, the standard interface provided by the loaded driver is first invoked to bind and convert the target test data according to the protocol format specified by the driver, generating an instruction template containing an instruction header and parameter placeholders. After protocol encapsulation and checksum addition, the control instructions of the standard test source are obtained. Next, the control instructions are transmitted to the standard test source through the communication module, and the control test source outputs corresponding signals (such as analog voltage signals, current signals, etc.) to the embedded instrument under test according to the test data in the test case.

[0172] When the standard test source outputs a signal, the communication module inside the server establishes a communication link with the embedded instrument under test (EUT) and receives new operating data generated by the EUT in real time through the communication link.

[0173] S330. If new abnormal operating data exists in the new operating data, the new fault location is relocated and added to the target fault location combination.

[0174] Specifically, the server first performs anomaly detection on the new running data, and filters out the new abnormal running data through preset anomaly judgment rules (such as numerical data exceeding the normal threshold range, time series data showing sudden changes, and new error codes appearing in text logs).

[0175] If the new abnormal operation data is not empty, repeat steps S302-S329 to obtain the new fault location, and add the new fault location to the target fault location combination.

[0176] In this embodiment, when an unknown fault type is detected, the generated set of features to be tested is compared one by one with the preset set of features corresponding to combinations of multiple known fault types in a preset composite fault feature library. The similarity between the two is calculated to determine whether the unknown fault is formed by multiple known faults. If the unknown fault is formed by multiple known faults, based on the identified known fault types, the target fault location corresponding to each known fault type in the preset fault location library is found. These target fault locations are merged to form a target fault location combination that includes all associated fault locations. This enables the location of the unknown fault formed by the superposition of multiple known faults, improving the comprehensiveness and accuracy of fault location. After the location is completed, the identified fault locations are masked, and the fault test is performed again. This enables dynamic tracking and supplementary location of the fault, avoiding the omission of potential faults, thereby improving the completeness and comprehensiveness of fault location.

[0177] The above describes the embedded instrument fault location method in the embodiments of this application. The server in the embodiments of this application will be described in detail below in conjunction with the above embedded instrument fault location method.

[0178] Please see Figure 4 This is a schematic diagram of an exemplary hardware structure of the server in an embodiment of this application.

[0179] In some embodiments, the server 400 includes a computer device, which may be a terminal device. The computer device includes a processor 401, a memory 402, a communication module 403, an input device 404, and an output device 405 connected via a system bus. The processor 401 provides computing and control capabilities. The memory 402 includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores data. The communication module 404 transmits instrument test data and sends control commands to standard test sources and instruments. The input device 405 receives test data transmitted by instruments and receives commands issued by users. The output device 406 displays test reports. When the computer program is executed by the processor 401, it implements the embedded instrument fault location method in this embodiment.

[0180] Those skilled in the art will understand that Figure 4The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0181] In some embodiments of this application, a computer-readable storage medium is provided, including instructions that, when executed on a server 400, cause the server 400 to perform the embedded instrument fault location method of this application.

[0182] In some embodiments of this application, a computer program product is also provided, which, when run on server 400, causes server 400 to execute the embedded instrument fault location method of this application.

[0183] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0184] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0185] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0186] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for locating faults in an embedded instrument, characterized in that, include: The control standard test source outputs corresponding signals to the embedded instrument under test according to the test data in the preset test cases, and obtains the instrument data, running data and system log of the embedded instrument under test when it receives the signal; When a fault is detected in the embedded instrument under test, and the fault types all belong to the preset fault type set, the original fault location and original fault data are located according to the instrument data, the running data and the system log, based on the preset fault diagnosis test cases. Based on a preset system causal relationship graph, a set of abnormal operating data (excluding the original fault data) related to the generation of the original fault data is selected from the operating data. Specifically, this includes: based on the preset system causal relationship graph and the operating data, selecting the upstream nodes and corresponding node operating data of the nodes corresponding to the original fault data in the causal chain related to the original fault data; generating normal operating data baselines for each upstream node based on the actual operating parameter data, actual operating environment data, and preset operating rules of each upstream node, wherein the normal operating data baselines are a set of multiple conditions for judging whether the operating data of the upstream nodes is normal; and selecting abnormal node operating data from all node operating data based on the normal operating data baselines to obtain a set of abnormal operating data. If the abnormal operation data set is not empty, search the set of fault location combinations corresponding to the original fault data in the preset fault feature library. The fault feature library stores fault location combinations consisting of a single or multiple fault locations, as well as fault data that each fault location combination will generate under different preset operation data. The operational data corresponding to each fault location in each fault location combination are merged to obtain the actual abnormal operational data set of each fault location combination. Calculate the similarity between the preset operating data set corresponding to each fault location combination and the actual abnormal operating data set to obtain the confidence level of each fault location combination; When the confidence level is higher than the preset confidence threshold, the fault location combination corresponding to the highest confidence level is selected as the target fault location combination of the embedded instrument under test.

2. The method according to claim 1, characterized in that, The process of generating a normal operating data baseline for each upstream node based on actual operating parameter data, actual operating environment data, and preset operating rules specifically includes: Obtain the preset standard operating data range, actual operating environment data, and actual operating parameter data of each upstream node; Based on the actual operating environment, the actual operating parameter data, and the preset mapping relationship between operating conditions and operating data, the theoretical operating data range of each upstream node is calculated. Based on the parameter coordination relationship defined in the preset operation rules and the actual operation parameter data, the theoretical operation data range is corrected to obtain the corrected operation data range; Based on the preset standard operating data range, the theoretical operating data range, and the corrected operating data range corresponding to each upstream node, a normal operating data baseline for each upstream node is generated.

3. The method according to claim 1, characterized in that, After the steps of the control standard test source outputting corresponding signals to the embedded instrument under test according to the test data in the preset test cases, and acquiring the instrument data, operating data, and system log of the embedded instrument under test when receiving the signals, the method further includes: When a fault is detected in the embedded instrument under test, and the fault type includes an unknown fault type that does not belong to the preset fault type set, a fault feature set to be tested is generated based on the instrument data, the operating data, and the abnormal data in the system log. From the preset fault type set, select two or more preset fault types to generate a candidate fault combination set, the candidate fault combination set including one or more candidate fault combinations; Obtain the preset combination feature set corresponding to each candidate fault combination in the preset composite fault feature library; the preset composite fault feature library pre-stores combinations consisting of two or more preset fault types, as well as the preset combination feature set that each combination should theoretically present; The similarity between the set of features of the fault to be tested and each preset combination feature set is calculated in turn to obtain the matching degree of each candidate fault combination. When the matching degree is higher than the preset matching threshold, the candidate fault combination with the highest matching degree is selected as the target fault combination. Find the target fault location corresponding to each preset fault type in the target fault combination, and merge all target fault locations to obtain the target fault location combination of the embedded instrument under test.

4. The method according to claim 3, characterized in that, After the step of generating a set of fault features based on the instrument data, the operating data, and the abnormal data in the system log when a fault is detected in the embedded instrument under test, and the fault type includes an unknown fault type that does not belong to the preset fault type set, the method further includes: When the fault type includes a known fault type belonging to a preset fault type set, a fault feature set of the known fault type in the current operating environment is generated based on the preset fault characterization generation rules. If the target feature identifier does not exist in the fault feature set, the identifier of the target feature identifier and the corresponding feature data are added to the unknown fault feature set, where the target feature identifier is any one of all feature identifiers in the fault feature set to be tested. If the target feature identifier exists in the set of fault features to be tested, the residual data of the fault feature data corresponding to the target feature identifier is calculated by performing residual calculation on the fault feature data. The identifier of the target feature identifier and the corresponding residual data are added to the unknown fault feature set; The set of features of the unknown faults is updated based on the set of features of the unknown faults.

5. The method according to claim 3, characterized in that, After the step of obtaining the preset combination feature set corresponding to each candidate fault combination in the preset composite fault feature library, the method further includes: Filter out the real-time operating condition parameter data from the operating data and environmental data that correspond to each feature in the preset combination feature set; Based on the real-time operating condition parameter data, the preset combined feature set is modified according to the preset influence rules of different operating condition parameters on each feature to obtain the modified combined feature set. Based on the modified combined feature set, update the preset combined feature set.

6. The method according to claim 1, characterized in that, After the step of selecting the fault location combination corresponding to the highest confidence level as the target fault location combination of the embedded instrument under test when the confidence level is higher than a preset confidence threshold, the method further includes: When there is no preset fault location in the target fault location combination, the control command corresponding to each target fault location in the target fault location combination is searched in the preset fault masking command library, and the control command is sent to the embedded instrument under test. When the preset fault location exists in the target fault location combination, the embedded instrument under test cannot execute the control command. After the embedded instrument under test has executed the control command, the target test data corresponding to the target fault location combination is searched in the preset test database. The standard test source is controlled to output a corresponding signal to the embedded instrument under test according to the target test data, and the new operating data of the embedded instrument under test when it receives the signal is acquired. If new abnormal operating data is found in the new operating data, the new fault location is relocated and added to the target fault location combination.

7. A server, characterized in that, include: One or more processors and memory; The memory is coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the server to perform the method as described in any one of claims 1-6.

8. A computer-readable storage medium storing computer instructions, characterized in that, When the computer instructions are executed on the server, the server causes the server to perform the method as described in any one of claims 1-6.

9. A computer program product, characterized in that, When the computer program product is run on the server, the server performs the method as described in any one of claims 1-6.

Images