A method for diagnosing faults of HPLC

Abstract

The application relates to an HPLC fault diagnosis method and relates to the field of HPLC fault identification, which comprises the following steps: obtaining historical data information corresponding to a fault signal in response to the fault signal; inputting the historical data information into a fault analysis model to obtain a fault probability corresponding to a fault cause; outputting an error fault signal when all the fault probabilities do not exceed a fault probability threshold value; defining the corresponding fault probability as a suspected fault probability and defining a fault cause corresponding to the suspected fault probability as a suspected fault cause when the fault probability exceeds the fault probability threshold value; obtaining a fault analysis scheme from a fault analysis scheme library based on the suspected fault cause; executing the fault analysis scheme to obtain a credibility parameter; and outputting the suspected fault cause corresponding to the credibility parameter when the credibility parameter is higher than a credibility threshold value. The application improves the diagnosis efficiency and scientificity through a full-automatic process, ensures the reliability of a diagnosis result by relying on a bidirectional verification mechanism, and realizes efficient and accurate fault diagnosis effect.

Description

Technical Field

[0001] This invention relates to the field of HPLC fault identification, and in particular to an HPLC fault diagnosis method. Background Technology

[0002] Currently, HPLC (High-Speed ​​Power Line Carrier Communication) is a communication method that uses power lines to transmit data. Compared to traditional power line communication, it significantly improves data transmission efficiency due to its high transmission rate, strong anti-interference capability, and wide coverage. However, HPLC communication units may experience data transmission anomalies during operation due to the complex power line environment and line aging. Therefore, timely fault diagnosis of the HPLC is necessary to quickly restore communication, reduce economic losses, provide a basis for preventative maintenance, and lower operation and maintenance costs.

[0003] Currently, HPLC fault diagnosis mainly relies on manual on-site troubleshooting. Technicians need to bring specialized equipment, such as signal analyzers and multimeters, to the fault area to test terminal communication quality, check line connection status, verify equipment parameter configuration, and analyze key indicators such as RSSI, SNR, and BER. Personnel locate the fault point by comparing the differences between normal and abnormal data, and rely on experience to judge noise interference, line aging, or equipment compatibility issues.

[0004] Regarding the aforementioned technologies, since technicians can only analyze faults based on the limited data available, the fault analysis relies on the individual skill level of the technicians and lacks scientific fault judgment, which may result in the HPLC not being able to receive timely and accurate maintenance. Summary of the Invention

[0005] In order to diagnose HPLC faults in a timely and scientific manner and reduce the impact of individual technicians' skill levels on HPLC fault repair, this invention provides an HPLC fault diagnosis method.

[0006] This invention provides an HPLC fault diagnosis method, which adopts the following technical solution:

[0007] An HPLC fault diagnosis method, characterized in that it includes:

[0008] Step 1: In response to a fault signal, obtain historical data information on the line corresponding to the fault signal;

[0009] Step 2: Input the historical data information into the preset fault analysis model to obtain the fault probability corresponding to all preset fault causes;

[0010] Step 20: Output a preset error fault signal when all the aforementioned fault probabilities do not exceed a preset fault probability threshold;

[0011] Step 21: When the failure probability exceeds the failure probability threshold, the corresponding failure probability is defined as the suspected failure probability, and the failure cause corresponding to the suspected failure probability is defined as the suspected failure cause.

[0012] Step 210: Obtain a fault analysis solution from a preset fault analysis solution library based on the suspected fault cause;

[0013] Step 211: Execute the fault analysis scheme to obtain the confidence parameters;

[0014] Step 212: When the confidence parameter is higher than the preset confidence threshold, output the suspected fault cause corresponding to the confidence parameter.

[0015] Optional, also includes:

[0016] Step 22: Define the number of suspected fault probabilities as the number of suspected faults;

[0017] Step 220: If the number of suspected faults is equal to 1, obtain a fault analysis solution from the fault analysis solution library based on the suspected fault cause;

[0018] Step 221: If the number of suspected faults is greater than 1, combine all suspected fault causes based on the number of suspected faults to obtain a suspected fault cause group;

[0019] Step 2210: Input the historical signal strength information into a preset multi-source fault analysis model to obtain the joint fault probability;

[0020] Step 2211: Take the joint failure probability with the highest value as the suspected joint failure probability;

[0021] Step 2212: When the suspected fault cause corresponding to the suspected joint fault probability is the same as the suspected fault cause corresponding to the suspected fault cause group, a fault analysis scheme is obtained from the fault analysis scheme library based on the joint fault probability.

[0022] Step 2213: Continue with step 211;

[0023] Step 2214: When the suspected fault cause corresponding to the suspected joint fault probability is different from the suspected fault cause corresponding to the suspected fault probability, the fault cause corresponding to the fault probability with the largest value is output as the suspected fault cause.

[0024] Optionally, it also includes a method for obtaining a fault analysis scheme from the fault analysis scheme library even when the fault probability does not exceed the fault probability threshold, the method comprising:

[0025] Step 222: Obtain the corresponding standard fault phenomenon from the preset fault analysis database based on the fault cause;

[0026] Step 223: Perform correlation analysis on all the standard fault phenomena to obtain similarity parameters between all the fault causes, whereby the similarity parameters represent the similarity of the standard fault phenomena between different fault causes;

[0027] Step 224: When the similarity parameter is greater than the preset similarity threshold, the corresponding fault causes are integrated into a fault cause group with the same phenomenon;

[0028] Step 225: Accumulate the failure probabilities corresponding to all the failure causes within the same failure cause group to obtain the accumulated failure probability;

[0029] Step 2250: If the cumulative failure probability is greater than the preset cumulative probability threshold, all the failure causes in the same phenomenon failure cause group are regarded as the suspected failure causes and step 210 is continued.

[0030] Step 2251: If there is no cumulative fault probability greater than the cumulative probability threshold, proceed to step 20.

[0031] Optionally, it also includes a method for obtaining a fault analysis scheme from the fault analysis scheme library even when the cumulative fault probability is not greater than the cumulative probability threshold, the method comprising:

[0032] Step 22510: Based on the historical data information, analyze the actual feature parameter change trends corresponding to all preset feature information types;

[0033] Step 22511: Based on the standard fault phenomenon, obtain the changing trends of all standard characteristic parameters corresponding to the fault cause;

[0034] Step 22512: Traverse all the fault causes to obtain a multi-source fault cause group. The variation trends of all the standard characteristic parameters in the multi-source fault cause group are superimposed and are the same as the variation trends of the actual characteristic parameters.

[0035] Step 22513: Obtain a fault analysis scheme from the fault analysis scheme library based on all the fault causes in the multi-source fault cause group and output it.

[0036] Optionally, it also includes a fault analysis method when the multi-source fault cause group does not exist, the method comprising:

[0037] Step 225120: Obtain the generation time, generation location, and implementation environment parameters of the fault signal;

[0038] Step 225121: Search the preset historical fault cause table for historical fault events, historical fault causes, and fault occurrence frequencies under the conditions of the occurrence time and the occurrence location;

[0039] Step 225122: Retrieve the historical environmental parameters corresponding to the historical fault events;

[0040] Step 225123: Perform a similarity analysis between the implemented environmental parameters and the historical environmental parameters to obtain environmental similarity parameters;

[0041] Step 225124: When the environmental similarity parameter is greater than the preset environmental similarity threshold, the corresponding historical fault event is defined as a suspected fault event;

[0042] Step 225125: Output the historical fault cause corresponding to the suspected fault event with the highest frequency of occurrence of the fault as the suspected fault cause.

[0043] Optionally, it also includes a method for obtaining and outputting the credibility parameter, the method comprising:

[0044] Step 2110: Obtain the transmission data information of the route to which the fault signal belongs;

[0045] Step 21100: When the transmitted data information is not obtained, the fault signal is defined as a data acquisition abnormal signal;

[0046] Step 211001: Based on the preset data cause table, find the change trends of all historical feature parameters corresponding to the abnormal data acquisition signal;

[0047] Step 211002: Perform a similarity analysis between the historical characteristic parameter change trends and the actual characteristic parameter change trends to obtain trend similarity parameters;

[0048] Step 211003: Define the trend similarity parameter with the largest value as the trend most similar parameter;

[0049] Step 211004: Obtain the anomaly correction parameter from the preset collection anomaly correction parameter table according to the trend most similar parameter;

[0050] Step 211005: Obtain the anomaly correction confidence parameter based on the confidence parameter and the anomaly correction parameter, and output it as the confidence parameter;

[0051] Step 21101: When the transmitted data information is obtained, the fault signal is defined as a data integrity abnormality signal;

[0052] Step 211010: Obtain complete data information from the preset backup line;

[0053] Step 211011: Based on the complete data information and the transmitted data information, obtain the cause of data loss from the preset data loss principle table;

[0054] Step 211012: Obtain integrity correction parameters from the preset data integrity parameter table based on the reasons for data loss and the suspected causes of failure;

[0055] Step 211013: Obtain the integrity correction confidence parameter based on the confidence parameter and the integrity correction parameter, and output it as the confidence parameter.

[0056] Optionally, methods for obtaining and outputting the credibility parameter also include:

[0057] Step 21102: When the data integrity error signal and the data acquisition error signal are received alternately during a preset historical time period, the fault signal is defined as a mixed error signal;

[0058] Step 211020: Analyze the proportion of the data integrity anomaly signal and the data acquisition anomaly signal in the historical time in the mixed anomaly signal, and define it as the mixed anomaly proportion;

[0059] Step 211021: Based on the simulated fault results corresponding to the suspected fault causes of all the fault analysis schemes, obtain simulated abnormal signals;

[0060] Step 211022: Analyze the proportion of the data integrity anomaly signal and the data acquisition anomaly signal in the simulated anomaly signals during the historical time period, and define it as the simulated anomaly proportion;

[0061] Step 211023: Obtain the hybrid correction parameter from the preset hybrid correction database based on the ratio of the hybrid anomaly ratio to the simulated anomaly ratio;

[0062] Step 211024: Obtain the mixed correction confidence parameter based on the confidence parameter and the mixed correction parameter, and continue to execute step 212 using the mixed correction confidence parameter as the confidence parameter.

[0063] Optionally, a diagnostic method for when the confidence parameter is lower than the confidence threshold, the method comprising:

[0064] Step 2120: Sort the suspected fault causes based on the fault probability to obtain a suspected fault cause sorting table;

[0065] Step 2121: Traverse the suspected fault cause sorting table and define the next suspected fault cause as the next suspected fault cause;

[0066] Step 21210: When the confidence parameter of the fault analysis scheme corresponding to the suspected fault cause is lower than the confidence threshold, the next suspected fault cause is used as the suspected fault cause and step 2121 is continued.

[0067] Step 21211: When the credibility parameter of the fault analysis scheme corresponding to the suspected fault cause is higher than the credibility threshold, output the suspected fault cause corresponding to the credibility parameter.

[0068] Optionally, a diagnostic method for when there is no confidence parameter of the fault analysis scheme corresponding to the suspected fault cause that is higher than the confidence threshold, the method comprising:

[0069] Step 212100: Locate all the preset fault device points in the aforementioned fault analysis scheme;

[0070] Step 212101: Based on integrating the fault analysis schemes corresponding to the fault device points, obtain a fault analysis scheme group;

[0071] Step 212102: Sum all the confidence parameters of the fault analysis scheme group to obtain the device confidence level;

[0072] Step 212103: When the confidence level of the device point exceeds a preset device confidence level threshold, the possible causes of failure of the faulty device point corresponding to the confidence level of the device point are obtained.

[0073] Step 212104: Output the possible causes of failure as the suspected causes of failure.

[0074] Optionally, a method for maintaining the fault analysis scheme library, the method comprising:

[0075] Step 3: Upon receiving a preset fault repair signal, obtain the fault cause and fault phenomenon corresponding to the fault signal;

[0076] Step 30: Input the fault signal into the fault analysis scheme library to obtain the corresponding fault ID;

[0077] Step 31: Input the fault cause, the fault phenomenon, and the historical data information corresponding to the fault phenomenon into the fault analysis scheme library according to the fault ID.

[0078] In summary, this application includes at least one of the following beneficial technical effects:

[0079] By implementing a fully automated design that automatically collects historical data after a fault signal is triggered, calls a pre-trained model to calculate the fault probability, dynamically adjusts the ranking of suspected causes, and outputs results based on credibility parameters, the need for manual intervention is significantly reduced.

[0080] The method innovatively integrates multi-level analysis logic such as independent analysis of single faults, joint probability calculation of multiple faults, clustering of fault groups with the same phenomenon, and trend matching of multi-source features, which fully covers complex scenarios such as single faults, concurrent multiple faults, and hidden faults, and breaks through the limitations of traditional single-dimensional diagnosis.

[0081] The reliability of the results is ensured through a dual mechanism of forward analysis and reverse verification. Attached Figure Description

[0082] Figure 1 This is a flowchart of an HPLC fault diagnosis method in an embodiment of this application.

[0083] Figure 2 This is a flowchart of a diagnostic method for when the confidence parameter is lower than the confidence threshold in an embodiment of this application. Detailed Implementation

[0084] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0085] This application discloses an HPLC fault diagnosis method. (Refer to...) Figure 1 An HPLC fault diagnosis method includes:

[0086] Step 1: In response to the fault signal, obtain historical data information on the line corresponding to the fault signal.

[0087] Fault signals are abnormal event identifiers used in HPLC systems to trigger fault diagnosis processes; they originate from proactive reporting by the HPLC terminal.

[0088] Historical data refers to relevant data on the faulty signal line within a preset time period before the fault occurs, which includes key parameters such as bit error rate (BER), received signal strength (RSSI), and signal-to-noise ratio (SNR).

[0089] By extracting communication logs from the local storage of the HPLC terminal device, historical data information of a preset time window is traced back based on the timestamp of the fault signal. For example, if the fault occurs at 10:00:00, data from 09:50:00 to 10:05:00 is extracted.

[0090] Step 2: Input historical data into the preset fault analysis model to obtain the fault probabilities corresponding to all preset fault causes.

[0091] Fault analysis models are derived from historical data and used to determine the causes and probabilities of current faults. These models are constructed using machine learning methods: a sample set is formed by obtaining three core parameters—bit error rate, received signal strength, and signal-to-noise ratio—from past fault occurrences, combined with fault labels. The dataset is then divided into a 70% training set, 15% test set, and 15% validation set, and the data is standardized to eliminate dimensional differences. Algorithms suitable for small-sample classification, such as decision trees and random forests, are selected to learn the mapping relationship between different fault causes and parameter features based on the training set, and the model's hyperparameters are optimized using the validation set. Finally, the model's generalization ability is evaluated using the test set, and the optimal model with a balance between precision and recall is selected as the final analysis model.

[0092] The causes of failures are predefined specific factors that may trigger HPLC communication anomalies, typically derived from expert experience, historical failure cases, and theoretical analysis methods. These include environmental interference, physical link problems, terminal equipment failures, and multipath effects in complex networks.

[0093] Failure probability is a confidence quantification of each preset failure cause by the failure analysis model, representing the likelihood that the model believes that the cause will lead to the current failure.

[0094] Step 20: Output a preset error fault signal when all fault probabilities do not exceed the preset fault probability threshold.

[0095] The failure probability threshold is a preset numerical threshold for the probability of failure, used to determine whether the failure causes output by the failure analysis model have sufficient credibility. The failure probability threshold is obtained by staff through extensive experiments. By evaluating the degree of agreement between the failure probability and the actual occurrence of failure, the lowest value that can prove the failure occurrence probability value matches the actual failure status is taken as the failure probability threshold.

[0096] An error fault signal is a signal automatically generated by the system when the probability of all preset fault causes in the fault diagnosis system does not exceed a preset threshold. It is used to indicate that the current fault analysis process has failed to locate a credible fault cause.

[0097] If all fault probabilities do not exceed the preset fault probability threshold, it indicates that the existing fault analysis model cannot find a typical pattern matching the current fault signal in the preset fault cause database. This may be due to insufficient correlation between historical data characteristics and fault causes, leading to inaccurate analysis, or it may be a false alarm of the fault signal. Therefore, the system determines that the current diagnostic process cannot obtain a reliable conclusion and triggers manual intervention or expands the investigation scope by outputting an erroneous fault signal.

[0098] Step 21: When the failure probability exceeds the failure probability threshold, the corresponding failure probability is defined as the suspected failure probability, and the failure cause corresponding to the suspected failure probability is defined as the suspected failure cause.

[0099] The suspected fault probability refers to the probability value corresponding to a fault cause that exceeds a preset fault probability threshold, as output by the fault analysis model. The suspected fault cause refers to a fault cause whose fault probability exceeds a preset threshold, as output by the fault analysis model.

[0100] A failure probability exceeding the failure probability threshold indicates that the model, based on historical data characteristics such as the changing trends of parameters like bit error rate (BER), received signal strength (RSSI), and signal-to-noise ratio (SNR), has identified a potential root cause with a high degree of matching to the current failure signal. This means the confidence level has reached a preset standard and the model is used as a credible clue in subsequent analysis processes.

[0101] Step 210: Obtain a fault analysis solution from the preset fault analysis solution library based on the suspected fault cause.

[0102] The fault analysis solution library is a pre-configured collection of verification solutions for various suspected fault causes. It stores the detection methods, data acquisition methods, and judgment rules that correspond one-to-one with the fault causes. The fault analysis library is obtained by staff through analyzing a large number of historical actual fault problems. Staff link the causes of faults with changes in environmental and equipment parameters, and then obtain detection solutions for acquiring environmental and equipment parameters based on the operating scenario. Ultimately, it establishes the correlation between fault causes, environmental parameters, equipment parameters, and detection solutions, and stores this correlation in the fault analysis solution library.

[0103] Fault analysis solutions are verification plans retrieved from a database for specific suspected fault causes. By inputting the code or keywords of suspected fault causes into the fault analysis solution database, the system can quickly match and extract the corresponding detection solutions.

[0104] Step 211: Execute the fault analysis scheme to obtain confidence parameters.

[0105] The credibility parameter is a quantitative indicator generated after executing the fault analysis plan, used to measure the degree of matching between the suspected fault cause and the actual situation. According to the requirements of the fault analysis plan, the system collects or retrieves relevant environmental parameters in real time within the equipment operating scenario, such as temperature, humidity, and electromagnetic interference, and equipment parameters, such as voltage, current, and signal strength. Then, it compares the actually collected parameters with the expected normal value range or typical fault characteristic values ​​defined in the plan to quantify the degree of matching between the suspected cause and the actual situation, ultimately generating the credibility parameter to measure the probability that the fault cause is valid. For example, when the suspected fault cause is determined to be temperature exceeding the sensor calibration threshold, the system calculates the cosine similarity between the temperature parameter under the current conditions and the preset typical temperature parameter to generate the credibility parameter, which is used to measure the credibility of the suspected fault cause.

[0106] Step 212: When the confidence parameter is higher than the preset confidence threshold, output the suspected fault cause corresponding to the confidence parameter.

[0107] A credibility threshold is a preset numerical threshold used to determine whether the credibility parameters obtained after the execution of a fault analysis plan meet the credibility standard. The credibility threshold is set directly by the staff.

[0108] When the confidence parameter is higher than the confidence threshold, it indicates that the cause of the failure has been verified as credible. Real-time data or evidence obtained through the failure analysis scheme highly matches the typical characteristics of the suspected failure cause, significantly reducing the probability of model misjudgment. Therefore, the system believes that the authenticity of the current failure cause has been verified through multiple dimensions and can be output as the failure cause.

[0109] Also includes:

[0110] Step 22: Define the number of suspected fault probabilities as the number of suspected faults.

[0111] The number of suspected faults refers to the total number of fault causes whose probability exceeds a preset threshold, as identified by the fault analysis model.

[0112] Step 220: If the number of suspected faults is equal to 1, obtain a fault analysis solution from the fault analysis solution library based on the suspected fault cause.

[0113] When the number of suspected faults equals 1, it indicates that the model has converged to the only high-confidence fault cause. At this point, the corresponding solution can be obtained directly from the fault analysis solution library by looking up the table.

[0114] Step 221: If the number of suspected faults is greater than 1, combine all suspected fault causes based on the number of suspected faults to obtain suspected fault cause groups.

[0115] When the number of suspected faults is greater than 1, it indicates that there may be multiple independent fault sources in the system, such as power failure and hardware damage occurring simultaneously, or a single fault causing multi-dimensional anomalies, such as a loose antenna leading to signal attenuation and increased bit error rate.

[0116] A suspected fault cause group refers to a set of possible fault combinations formed by combining all suspected fault causes when the number of suspected faults is greater than 1.

[0117] Step 2210: Input the historical signal strength information into the preset multi-source fault analysis model to obtain the joint fault probability.

[0118] The combined failure probability refers to the overall failure probability value calculated when multiple suspected failure causes occur simultaneously or interact with each other.

[0119] The multi-source fault analysis model is constructed using machine learning methods. Specifically, it involves obtaining the changing trends of three core parameters—bit error rate, received signal strength, and signal-to-noise ratio—during past fault occurrences, and combining these with fault labels to form a sample set. During sample set construction, considering the nonlinear coupling characteristics of multiple fault parameters, typical cases of multiple faults occurring simultaneously are prioritized when selecting fault causes. Examples include scenarios where equipment aging is combined with electromagnetic interference, or line loss is accompanied by signal attenuation. Deep learning algorithms are then used to uncover the implicit correlation rules between different fault parameters.

[0120] Step 2211: Take the joint failure probability with the highest value as the suspected joint failure probability.

[0121] The suspected joint failure probability refers to the highest probability value among the joint probability parameters corresponding to all possible combinations of failure causes calculated by the multi-source failure analysis model in a multi-failure concurrent analysis scenario. This parameter represents the comprehensive probability of the most likely combination of multiple failures and is the optimal solution selected by the system from various potential failure correlations, used to guide subsequent failure verification and repair decisions. The model construction method is the same as in step 2 and will not be repeated here.

[0122] Step 2212: When the suspected fault cause corresponding to the suspected joint fault probability is the same as the suspected fault cause corresponding to the suspected fault cause group, a fault analysis scheme is obtained from the fault analysis scheme library based on the joint fault probability.

[0123] When the suspected fault cause corresponding to the suspected joint fault probability is the same as the suspected fault cause corresponding to the suspected fault probability group, it indicates that the current fault signal is very likely caused by multiple fault causes. Furthermore, both the fault analysis model and the multi-source fault analysis model yield the same result, indicating high reliability. Therefore, all suspected fault causes corresponding to the joint fault probability can be analyzed as possible causes. Both the suspected fault cause corresponding to the suspected joint fault probability and the suspected fault cause corresponding to the suspected fault cause group are sets of suspected roadblock causes consisting of multiple suspected fault causes. When comparing them, both the number of causes must be consistent, and the specific causes must be completely matched.

[0124] Step 2213: Continue with step 211.

[0125] Step 2214: When the suspected cause of the suspected joint fault probability is different from the suspected cause of the suspected fault probability, the fault cause corresponding to the fault probability with the largest value is output as the suspected fault cause.

[0126] When the suspected cause of a joint fault is inconsistent with the suspected cause of a fault, it indicates that the model cannot determine whether the fault is caused by a single cause or multiple causes concurrently. Therefore, the system first selects the fault cause corresponding to the single fault probability with the largest value as the simplified decision logic for the complex scenario of the suspected fault cause, and prioritizes the analysis of the cause with the highest probability to ensure timely handling of the fault.

[0127] It also includes a method for obtaining a fault analysis solution from a fault analysis solution library even when the probability of no fault exceeds a fault probability threshold. This method includes:

[0128] Step 222: Obtain the corresponding standard fault phenomenon from the preset fault analysis database based on the cause of the fault.

[0129] The fault analysis database, compiled by staff through extensive analysis, contains fault causes and their corresponding classic fault phenomena. Staff collected fault causes encountered during past operations and compiled the corresponding fault phenomena for each cause. For example, by analyzing numerous external electromagnetic interference fault causes, staff determined that under these conditions, the bit error rate would significantly increase, RSSI would fluctuate and decrease, the signal-to-noise ratio would drop sharply, and the noise level would rise. These key parameter changes were used as standard fault phenomena corresponding to the fault cause. After establishing the correspondence between fault causes and standard fault phenomena, the data was stored in the fault analysis database.

[0130] Standard fault phenomena refer to the observable characteristics corresponding to typical faults that have been verified in the historical operation of equipment or systems, such as abnormal temperature or response delay. These phenomena serve as a reference for current fault diagnosis. By inputting the fault cause into the fault analysis database, the standard fault phenomena corresponding to the fault cause can be obtained.

[0131] Step 223: Perform correlation analysis on all standard fault phenomena to obtain similarity parameters among all fault causes. The similarity parameters represent the similarity of standard fault phenomena among different fault causes.

[0132] Correlation analysis refers to calculating the similarity between standard fault phenomena corresponding to fault causes using statistical or algorithmic methods. By obtaining the parameter change characteristics of all preset key parameters corresponding to the standard fault phenomena, key parameters with identical changes between two standard fault phenomena are recorded as similar parameters. The similarity parameter can be obtained by dividing the number of similar parameters by the number of key environmental and equipment parameters. For example, key environmental and equipment parameters include: Bit Error Rate (BER), Received Signal Strength (RSS), and Signal-to-Noise Ratio (SNR). Fault A is wire interference, Fault B is impedance mismatch, and Fault C is external electromagnetic interference. Fault A results in increased BER, decreased RSSI, and decreased SNR. Fault B has no impact on BER, decreased RSSI, and decreased SNR; Fault C results in increased BER, decreased RSSI, and decreased SNR. Faults A and B share RSSI and SNR, with a similarity parameter of 2 / 3, indicating relatively similar phenomena, both affecting RSSI and SNR. Faults A and C share BER, RSSI, and SNR, with a similarity parameter of 1, indicating completely identical phenomena. This is only an example; specific key environmental and equipment parameters need to be determined based on the on-site environmental conditions and equipment usage.

[0133] The similarity parameter is a parameter that quantitatively describes the degree of similarity between different fault causes at the level of standard fault phenomena. The larger the similarity parameter, the more similar the fault phenomena are.

[0134] Step 224: When the similarity parameter is greater than the preset similarity threshold, the corresponding fault causes are integrated into the same phenomenon fault cause group.

[0135] The similarity threshold is determined by staff through analysis of a large number of historical failure phenomena. It is the lowest value that represents a high degree of similarity between the failure phenomena caused by two different failure causes. When the similarity parameter exceeds the preset threshold, it indicates that the standard failure phenomena caused by these failure causes, such as changes in parameters like BER, RSSI, and SNR, highly overlap in key characteristics, possibly triggered by similar physical mechanisms or external interference. If each failure cause is analyzed individually, its probability may be "equally distributed" due to the similarity of the phenomena.

[0136] A fault cause group refers to a group of fault causes whose similarity parameters are greater than a preset similarity threshold. When multiple fault causes lead to highly similar standard fault phenomena, these fault causes are difficult to distinguish directly at the phenomenological level, but their impact on the system is consistent, so they need to be integrated into a group for joint analysis.

[0137] Step 225: Accumulate the failure probabilities corresponding to all failure causes within the same failure phenomenon group to obtain the accumulated failure probability.

[0138] The cumulative failure probability refers to the total probability value obtained by summing the failure probabilities of all failure causes within a group of causes of a certain phenomenon. By accumulating the independent probabilities of each failure cause within the group, the probability of a single cause being underestimated due to similar phenomena is avoided, thus making the overall credibility of the phenomenon more reliable.

[0139] Step 2250: If the cumulative failure probability is greater than the preset cumulative probability threshold, all failure causes in the same failure phenomenon group are regarded as suspected failure causes and step 210 is continued.

[0140] The cumulative probability threshold is obtained by staff through the analysis of a large number of actual historical faults. It is the minimum parameter that can prove that the fault phenomenon is actually credible. The staff defines this parameter as the cumulative probability threshold.

[0141] The fact that the cumulative failure probability exceeds the preset cumulative probability threshold indicates that the standard failure phenomena caused by all failure causes within the same failure phenomenon group are not accidental. This phenomenon arises because when multiple failure causes produce highly similar standard failure phenomena, the determination of the failure cause is affected. However, since the cumulative failure probability exceeds the threshold, the failure phenomenon is considered credible. Therefore, to identify the failure cause, all failure causes need to be treated as potential causes for further verification. For example, strong external electromagnetic interference and aging power lines can both lead to a decrease in RSSI and an increase in BER. The calculated failure probabilities are 30% and 35%, respectively. However, in reality, the probability misjudgment is due to the similarity of the two phenomena. The actual probability of the "decreased RSSI and increased BER" ​​failure phenomenon reaches 65%, thus requiring careful analysis.

[0142] Step 2251: If there is no cumulative fault probability greater than the cumulative probability threshold, proceed to step 20.

[0143] When the cumulative failure probability corresponding to a group of non-existent fault phenomena exceeds the cumulative probability threshold, it indicates a low correlation between the currently observed fault phenomenon and known fault causes. This suggests that the probabilities of all fault causes are low, and the issue is not due to the dispersion of probabilities for a single cause caused by similar fault phenomena. In this case, the system determines that in-depth cause analysis of the existing fault phenomenon group is unnecessary and directly triggers step 20.

[0144] A method for obtaining a fault analysis solution from a fault analysis solution library when there is no cumulative fault probability greater than a cumulative probability threshold, the method includes:

[0145] Step 22510: Analyze the actual feature parameter change trends corresponding to all preset feature information types based on historical data information.

[0146] Feature information type refers to the category of observable parameters that describe fault phenomena, usually key indicators of equipment operating status. Examples include bit error rate, received signal strength, signal-to-noise ratio, and latency jitter.

[0147] The actual characteristic parameter change trend refers to the direction of parameter change corresponding to the characteristic information type in a real fault scenario. For example: BER increases, RSSI remains unchanged, and SNR decreases.

[0148] The system extracts time-series data of target feature parameters from historical data, arranges them chronologically, removes outliers, iterates through data points, and counts the increase or decrease between adjacent points: if the current value is greater than the previous value, it is recorded as "increasing"; otherwise, it is recorded as "decreasing"; if they are equal, it is recorded as "no change". If the proportion of "increasing" exceeds a set threshold, such as 70%, and there is no significant reverse fluctuation, it is considered monotonically increasing; similarly, if the proportion of "decreasing" exceeds the threshold, it is considered monotonically decreasing. If the proportion of "no change" exceeds the threshold, such as 80%, it is considered stable. If the increases and decreases alternate frequently without a dominant direction, such as the increase and decrease ratios both being close to 50%, it is considered fluctuating. The proportion thresholds can vary depending on the actual application environment; this is just an example.

[0149] Step 22511: Obtain the changing trends of all standard characteristic parameters corresponding to the fault causes based on standard fault phenomena.

[0150] The trend of standard characteristic parameters refers to the qualitative or quantitative change pattern of characteristic parameters corresponding to a fault cause, summarized by analyzing the correlation between typical fault phenomena and fault causes. For example, when a certain type of fault occurs, specific parameters may show a trend of monotonically increasing, decreasing, fluctuating, or no significant change. For instance, since power line carrier signals rely on voltage amplitude for transmission, a decrease in voltage will lead to a significant decrease in signal strength, manifested as a sharp drop in RSSI values.

[0151] Step 22512: Traverse all fault causes to obtain multi-source fault cause groups. The variation trends of all standard characteristic parameters in the multi-source fault cause group are superimposed and are the same as the variation trends of the actual characteristic parameters.

[0152] A multi-source fault cause group refers to a set of two or more fault causes. The changing trends of the standard characteristic parameters corresponding to all fault causes within this group, such as increasing, decreasing, or fluctuating, result in an overall effect that perfectly matches the observed trend of characteristic parameter changes after superposition or interaction. For example, in an HPLC communication system, if the actual manifestation is a decrease in RSSI and an increase in BER, it may be a multi-source fault group composed of two fault causes: "unstable power supply voltage" and "line interference." The former causes a decrease in RSSI, while the latter causes an increase in BER. The combined trend of the two perfectly matches the actual phenomenon.

[0153] Step 22513: Obtain the fault analysis solution from the fault analysis solution library based on all fault causes in the multi-source fault cause group and output it.

[0154] By inputting all the fault causes within a multi-source fault cause group into the fault analysis solution library, a fault analysis solution corresponding to each fault cause can be obtained. Then, the fault analysis solution can be used to determine whether the analyzed fault cause is credible.

[0155] It also includes a fault analysis method for situations where there is no multi-source fault cause group, the method comprising:

[0156] Step 225120: Retrieve the generation time, generation location, and implementation environment parameters of the fault signal.

[0157] The occurrence time refers to the specific time point when the fault occurred. Only the exact moment when the fault occurred is recorded. For example, if the fault occurred at 12:35, the date is not recorded.

[0158] The location of the fault refers to the physical or logical location where the fault occurs, which can be the specific area where the equipment is deployed or a network node. For example, the location of a fault in a power distribution automation terminal is "No. 3 distribution cabinet of XX substation in XX city, XX province", or the location information of an HPLC communication fault is "A phase power line, GPS coordinates (longitude 120.12°, latitude 30.56°)".

[0159] The implementation environmental parameters refer to the real-time or historical conditions of the environment in which the equipment is located when a fault occurs, including parameters that directly affect the operation of the equipment, such as temperature, humidity, voltage, current, electromagnetic interference, and weather conditions.

[0160] Step 225121: Find historical fault events, historical fault causes, and fault occurrence frequencies from the preset historical fault cause table under the conditions of occurrence time and location.

[0161] The historical fault cause table is a structured data table built by staff after analyzing a large number of historical fault cases. It contains key information about historical fault events, such as the time, location, and environmental parameters of occurrence, as well as the fault cause type and frequency of occurrence for each historical fault event. Staff search for historical fault events in the system, link the key information corresponding to the fault with the fault event, and accumulate the number of faults in chronological order. The fault frequency can be obtained by dividing the number of fault causes by the total number of faults.

[0162] Simply input the occurrence time and location into the historical fault cause table, and the system will obtain the historical fault times that have occurred at that time and location under historical conditions, and find the corresponding historical fault cause and fault frequency based on the historical fault cause table.

[0163] Step 225122: Retrieve the historical environmental parameters corresponding to the historical fault events.

[0164] Historical environmental parameters refer to quantitative data on the external environment or equipment operating status related to historical failure events, used to describe the objective conditions at the time of the failure. These include physical environment parameters such as temperature, humidity, and air pressure, and electrical environment parameters such as voltage fluctuations, abnormal current, and electromagnetic interference intensity.

[0165] By inputting the ID of a historical fault event into the system, the historical environmental parameters corresponding to that fault time can be retrieved from the system.

[0166] Step 225123: Perform a similarity analysis between the implemented environmental parameters and historical environmental parameters to obtain environmental similarity parameters.

[0167] Environmental similarity parameters refer to numerical indices reflecting the similarity between current fault-related environmental parameters (such as temperature, humidity, voltage, and electromagnetic interference) and historical fault-related environmental parameters, calculated by quantifying the differences between these parameters. For example: Current environmental parameters: temperature 45℃, humidity 30%, voltage fluctuation ±8%. Historical environmental parameters: temperature 43℃, humidity 28%, voltage fluctuation ±10%. Assumed parameter range: temperature 0 to 100℃, humidity 0 to 100%, voltage fluctuation 0 to 20%.

[0168] After normalization, the current temperature is normalized to 45 / 100 = 0.45, and the historical temperature is normalized to 43 / 100 = 0.43; the current humidity is normalized to 30 / 100 = 0.3, and the historical humidity is normalized to 28 / 100 = 0.28; the current voltage is normalized to 8 / 20 = 0.4, and the historical voltage is normalized to 10 / 20 = 0.5.

[0169] Substitute the above values ​​into the Euclidean distance formula to obtain the Euclidean distance, and then substitute the distance into the similarity calculation formula: Substituting the values ​​yields a similarity of 0.94. This indicates that the similarity parameter between the current environment and the historical environment is 94%.

[0170] Step 225124: When the environmental similarity parameter is greater than the preset environmental similarity threshold, the corresponding historical fault event is defined as a suspected fault event.

[0171] A suspected failure event refers to a failure event that may have the same cause as a historical failure event.

[0172] The environmental similarity threshold, obtained by operators through analysis of numerous past fault events, represents the lowest value that demonstrates the similarity between current environmental parameters and historical fault parameters. When the environmental similarity parameter exceeds the preset threshold, it indicates a high degree of match between the current environment and the historical fault environment. This proves that the current operating environment of the equipment, such as temperature, humidity, and voltage fluctuations, is sufficiently close to the environmental conditions at the time of a historical fault, thus potentially triggering the same or similar fault mechanisms. Therefore, the corresponding historical fault event can be considered a suspected fault event.

[0173] Step 225125: Output the historical fault causes corresponding to the suspected fault events with the highest frequency of fault occurrence as suspected fault causes.

[0174] By statistically analyzing the frequency of suspected fault events in historical data, the system selects the historical fault causes corresponding to the most frequent events as the output. The core logic lies in leveraging the statistical patterns of historical data to prioritize the output of the fault causes with the highest probability of occurrence.

[0175] It also includes a method for obtaining and outputting credibility parameters, which includes:

[0176] Step 2110: Obtain the transmission data information of the route to which the fault signal belongs.

[0177] Transmitted data information refers to the raw data on the communication lines or data transmission paths related to the fault signal. The archived transmission data of this route before the fault occurred can be retrieved by querying historical databases or storage systems.

[0178] Step 21100: When no transmitted data information is obtained, the fault signal is defined as a data acquisition abnormal signal.

[0179] A data acquisition anomaly signal refers to an abnormal event marked by the system during fault diagnosis due to the failure to acquire transmitted data. This signal indicates that the problem is not with the data content itself, but with the acquisition process or the data source.

[0180] The failure to obtain transmitted data information indicates a defect in the data acquisition process, which may be due to equipment failure or human error, causing the system to be unable to extract the required transmitted data from the specified route.

[0181] Step 211001: Based on the preset data cause table, find the change trend of all historical characteristic parameters corresponding to the abnormal data acquisition signal.

[0182] The data cause table is a pre-set structured database or knowledge base used to store the mapping relationship between abnormal data acquisition signals and the corresponding historical characteristic parameter change trends. The data cause table can be obtained by staff by searching a large amount of historical fault data. Staff associate the change trends of parameters such as bit error rate, received signal strength, signal-to-noise ratio, and delay jitter during the data acquisition failure process and store them in the data cause table, and label them with abnormal data acquisition signals.

[0183] By inputting the labels of abnormal data acquisition signals into the data cause table, the changing trends of various historical characteristic parameters can be obtained for all faults caused by failure to acquire transmitted data information.

[0184] Step 211002: Perform similarity analysis on the historical trend of change of all characteristic parameters and the actual trend of change of characteristic parameters to obtain trend similarity parameters.

[0185] The trend similarity parameter is a quantitative indicator used to measure the degree of matching between the historical trend of characteristic parameters and the actual trend of characteristic parameters. This parameter comprehensively analyzes the changing patterns of multiple characteristic parameters (such as signal strength, bit error rate, delay, etc.), combined with the shape, slope, periodicity, and other features of the time series, and finally outputs a normalized similarity value.

[0186] Similarity analysis is a method that assesses the consistency or difference between two or more objects by quantitatively comparing the degree of matching between them. By comparing the consistency between historical and actual trends in feature parameters, and calculating the proportion of parameters with the same trend, the degree of similarity can be quantitatively assessed. Specific calculation methods can be found in step 223. Other similarity calculation methods, such as cosine similarity, will not be elaborated upon here.

[0187] Step 211003: Define the trend similarity parameter with the largest value as the trend most similar parameter.

[0188] The most similar parameter in terms of trend refers to the feature parameter with the highest similarity value among all historical feature parameter change trends and actual feature parameter change trends.

[0189] Step 211004: Obtain the anomaly correction parameter from the preset collection anomaly correction parameter table based on the most similar trend parameter.

[0190] The anomaly correction parameter table is a pre-configured mapping table that stores parameters with different trends and their corresponding correction parameters. By analyzing a large number of historical faults, staff can statistically determine the correspondence between the most similar parameter and the anomaly correction parameter. The larger the statistically most similar parameter, the larger the correction parameter, indicating a more reliable parameter. By inputting the most similar parameter into the anomaly correction parameter table, the corresponding anomaly correction parameter can be found.

[0191] Step 211005: Obtain the anomaly correction confidence parameter based on the confidence parameter and the anomaly correction parameter, and output it as the confidence parameter.

[0192] The anomaly correction confidence parameter is a dynamic value calculated by combining the original confidence parameter and the collected anomaly correction parameter. It is used to quantify the reliability of the corrected conclusion.

[0193] The anomaly correction confidence parameter can be obtained by multiplying the original confidence parameter by the anomaly correction parameter.

[0194] Step 21101: When the transmitted data information is obtained, the fault signal is defined as a data integrity abnormal signal.

[0195] A data integrity anomaly signal refers to a situation where, during data transmission, the system detects that data is lost, corrupted, or structurally incomplete, resulting in the received data failing to meet preset integrity requirements. Triggering conditions for this signal include: data packet loss, incorrect checksum, missing critical fields, or abnormal values.

[0196] Step 211010: Obtain complete data information from the preset backup line.

[0197] A backup line is a pre-configured redundant transmission channel or backup path used to provide reliable and complete data supplementation when data transmission from the main line experiences integrity issues. Complete data refers to the data that would have been received had the fault not occurred.

[0198] When the data transmitted on the main line is detected as incomplete, the system automatically or manually switches to the backup line to obtain complete data information.

[0199] Step 211011: Based on the complete data information and the transmitted data information, obtain the cause of data loss from the preset data loss principle table.

[0200] The data loss principle table is a pre-defined mapping table used to store the association rules between different data loss patterns and their corresponding root causes. By analyzing a large number of historical failure events, staff identify the correspondence between different loss types and their causes under data integrity anomaly signals, and then store this correspondence in the data loss principle table.

[0201] By analyzing the types of missing data between complete and transmitted data, such as field-level missing data, packet-level missing data, and checksum-level anomalies, and substituting the corresponding missing type into the data loss principle table, the possible causes of data loss for that missing type can be obtained.

[0202] Step 211012: Obtain integrity correction parameters from the preset data integrity parameter table based on the reasons for data loss and suspected fault causes.

[0203] The data integrity parameter table is a pre-defined mapping table used to store integrity correction parameters corresponding to the causes of data loss and suspected fault causes. By analyzing a large number of fault events, staff compare the data loss causes obtained through table lookup with the suspected fault causes obtained through machine learning analysis, constructing a correspondence between the similarity of the two causes and the integrity correction parameters. This allows the integrity correction parameters to represent the similarity of the fault causes obtained by the two analysis methods.

[0204] For example, when the cause of data loss is exactly the same as the suspected cause of failure, the staff sets the integrity correction parameter to 1.2. When the cause of data loss corresponds to the same device as the suspected cause of failure, the staff sets the integrity correction parameter to 1.1. When the cause of data loss is completely unrelated to the suspected cause of failure, the staff sets the integrity correction parameter to 0.8. The above data is for demonstration purposes only; in actual use, the range and parameters of the integrity correction parameter can be set according to the actual conditions.

[0205] Step 211013: Obtain the integrity correction credibility parameter based on the credibility parameter and the integrity correction parameter, and output it as the credibility parameter.

[0206] The integrity correction confidence parameter refers to the parameter value calculated based on the confidence parameter and the integrity correction parameter.

[0207] The integrity correction confidence level can be obtained by multiplying the confidence level parameter by the integrity correction parameter.

[0208] Methods for obtaining and outputting credibility parameters also include:

[0209] Step 21102: When data integrity error signals and data acquisition error signals are received alternately during a preset historical time period, the fault signal is defined as a mixed error signal.

[0210] Hybrid anomaly signals refer to a composite anomaly state in which the system alternately detects data integrity anomaly signals and data acquisition anomaly signals within a preset historical time window, and the two are temporally correlated.

[0211] When the system alternately detects data integrity and data acquisition anomaly signals within a historical time window, it indicates that a fault may simultaneously cause problems with both data integrity and acquisition capability. For example, when using HPLC technology to remotely acquire data from smart meters, some meters may experience missing voltage and current data during peak electricity consumption periods, which constitutes a data integrity anomaly; simultaneously, the concentrator may display "communication interruption," which constitutes a data acquisition anomaly.

[0212] Step 211020: Analyze the proportion of data integrity anomaly signals and data acquisition anomaly signals in the historical time of the mixed anomaly signals, and define it as the mixed anomaly proportion.

[0213] The mixed anomaly ratio refers to the proportion of the frequency of alternating occurrences of data integrity anomaly signals and data acquisition anomaly signals within a preset historical time window. It is used to quantify the degree of mixing between the two types of anomalies. The mixed anomaly ratio includes the proportion of data integrity anomaly signals and the proportion of data acquisition anomaly signals in the historical time period.

[0214] The proportion of mixed anomalies includes: {the proportion of data integrity anomaly signals, and the proportion of data acquisition anomaly signals}. The proportion of data integrity anomaly signals can be obtained by dividing the number of data integrity anomaly signals by the number of samples taken in the historical time window, and the proportion of data acquisition anomaly signals can be obtained by dividing the number of data acquisition anomaly signals by the number of samples taken in the historical time window.

[0215] Step 211021: Based on the suspected fault causes corresponding to all fault analysis schemes, simulate the fault results to obtain simulated abnormal signals.

[0216] The simulated abnormal signal is a theoretical abnormality pattern derived from the suspected fault cause, used to verify whether the actual abnormality is caused by the fault. A physical model is established based on the suspected fault cause to explain the resulting abnormality. When the suspected fault is a transmission link impairment, the bit error rate (BER) increases exponentially due to signal attenuation, the received signal strength (RSSI) decreases linearly with path loss, and the signal-to-noise ratio (SNR) decreases logarithmically due to environmental noise. If it is an interference-related fault, the BER exhibits pulse-like abrupt changes in the interference frequency band, the RSSI fluctuation variance increases, and the SNR drops sharply during the interference period. By matching parameter sensitivity coefficients using a fault mechanism library, the multi-dimensional parameter change trends are time-axis aligned and amplitude normalized, ultimately outputting a composite simulated abnormal signal that matches the fault characteristics.

[0217] Step 211022: Analyze the proportion of data integrity anomaly signals and data acquisition anomaly signals in the simulated anomaly signals over historical time, and define it as the simulated anomaly proportion.

[0218] The simulated anomaly percentage refers to the relative proportion of data integrity anomalies and data acquisition anomalies within a preset historical time window among the anomaly signals generated through fault simulation.

[0219] For simulated abnormal signals, the number of data integrity anomalies and the number of data acquisition anomalies are statistically analyzed on a unified time scale. The proportion of simulated anomalies can be obtained by comparing it with the number of samples taken in historical time windows. The calculation method for the proportion of simulated anomalies is the same as in steps 211020, and will not be repeated here.

[0220] Step 211023: Obtain the mixed correction parameters from the preset mixed correction database based on the ratio of the mixed anomaly ratio to the simulated anomaly ratio.

[0221] The hybrid correction parameter is used to quantify the degree of matching between actual anomaly patterns and simulated fault patterns. The hybrid correction database is a pre-defined mapping table that stores the hybrid correction parameters corresponding to different ranges of the ratio between the proportion of hybrid anomalies and the proportion of simulated anomalies. By analyzing a large amount of historical fault data, staff correlated the proportion of hybrid anomalies caused by different faults with the hybrid correction parameters, enabling the hybrid correction parameters to accurately reflect the similarity of anomaly proportions between different faults, and thus reflect the similarity between different faults.

[0222] Step 211024: Obtain the mixed correction confidence parameter based on the confidence parameter and the mixed correction parameter, and continue to execute step 212 using the mixed correction confidence parameter as the confidence parameter.

[0223] The hybrid correction confidence parameter is a parameter obtained by multiplying the confidence parameter and the hybrid correction parameter. It is used to quantify the degree of matching between the suspected fault cause and the actual application.

[0224] The mixed correction confidence parameter can be obtained by multiplying the confidence parameter by the mixed correction parameter.

[0225] Reference Figure 2 A diagnostic method for when the confidence parameter is below a confidence threshold, the method comprising:

[0226] Step 2120: Sort the suspected fault causes based on the fault probability to obtain a suspected fault cause sorting table.

[0227] The suspected fault cause ranking table is a list that prioritizes potential fault causes based on their fault probability. The system sorts the fault probabilities from highest to lowest, and the corresponding suspected fault causes are also ranked accordingly.

[0228] Step 2121: Traverse the suspected fault cause sorting table and define the next suspected fault cause as the next suspected fault cause.

[0229] Secondary suspected causes refer to candidate causes that follow the current analysis object and are ranked next in order of probability in the list of suspected causes. Their role is to serve as alternative analysis targets in the fault diagnosis process when the confidence level of the current cause does not reach a threshold.

[0230] Step 21210: When the confidence parameter of the fault analysis scheme corresponding to the suspected fault cause is lower than the confidence threshold, the next suspected fault cause is taken as the suspected fault cause and step 2121 is continued.

[0231] When the confidence parameter of the fault analysis scheme corresponding to the suspected fault cause is lower than the threshold, it indicates that the hypothesis of the current fault cause lacks sufficient evidence to support it. Therefore, the next in line with a slightly lower fault probability is selected as the suspected fault cause for verification.

[0232] Step 21211: When the credibility parameter of the fault analysis scheme corresponding to the suspected fault cause is higher than the credibility threshold, output the suspected fault cause corresponding to the credibility parameter.

[0233] When the credibility parameter of the fault analysis scheme corresponding to the suspected fault cause is higher than the credibility threshold, it indicates that the matching degree between the simulated abnormal signal of the fault cause and the actual observation data has reached the preset standard. The system terminates the traversal process and outputs the fault cause and its associated fault probability as the final diagnosis result, and at the same time ends the current fault analysis process.

[0234] A diagnostic method for cases where the confidence parameter of a fault analysis solution for which no suspected fault cause exists is higher than a confidence threshold, the method comprising:

[0235] Step 212100: Locate the preset fault device points in all fault analysis schemes.

[0236] Preset fault points refer to specific hardware components, connection nodes, or functional modules in an HPLC system that are prone to communication failures due to their physical structure, electrical characteristics, or environmental interactions. Examples include line connectors and connection points, power lines themselves, and filters.

[0237] Step 212101: Based on the integration of the fault analysis schemes corresponding to the fault device points, obtain a fault analysis scheme group.

[0238] A fault analysis scheme group is a collection of multiple fault analysis schemes targeting the same faulty device point. A fault analysis scheme group is obtained by finding the locations of devices that may cause the fault within each fault analysis scheme, and then integrating the fault analysis schemes corresponding to the same potential faulty device locations.

[0239] Step 212102: Sum up all the confidence parameters of the fault analysis scheme group to obtain the device point confidence.

[0240] The reliability of a device point is a comprehensive value obtained by summing the reliability parameters of all relevant failure analysis schemes for a certain faulty device point, and is used to measure the probability of the device point failing.

[0241] Step 212103: When the confidence level of a device point exceeds a preset device confidence level threshold, the possible causes of failure of the faulty device point corresponding to the device point confidence level are obtained.

[0242] The device reliability threshold is a preset numerical threshold used to determine whether the reliability of a faulty device point meets the standard requiring an alarm or action. The device reliability threshold was obtained by staff through extensive testing and represents the lowest threshold indicating a high probability of a current faulty device point malfunctioning.

[0243] If the confidence level of a device exceeds a preset device confidence level threshold, the device is considered to have a risk of failure.

[0244] Possible causes of failure refer to the set of specific failure hypotheses directly related to a device point, determined by combining the confidence parameters of all failure analysis schemes corresponding to that device point when the device point confidence exceeds a preset threshold.

[0245] Step 212104: Output the possible causes of failure as suspected causes of failure.

[0246] The method for maintaining the fault analysis solution library includes:

[0247] Step 3: After receiving the preset fault repair signal, obtain the fault cause and fault phenomenon corresponding to the fault signal.

[0248] The fault repair signal is a pre-set trigger mechanism used to indicate that a fault has been repaired or eliminated. The fault repair signal can be manually sent by maintenance personnel through the user interface after fault repair is completed, or it can be automatically generated by the system after sensor or logic-based judgment that signal indicators have normalized.

[0249] The cause of failure refers to the fundamental factor that causes equipment or system abnormalities, such as hardware damage (e.g., aging power module), software vulnerabilities, signal interference, or environmental parameters exceeding the standard (e.g., excessively high temperature).

[0250] Fault phenomena refer to the abnormal state exhibited by equipment or systems when a fault occurs, such as communication interruption, increased signal noise, equipment unresponsiveness, and data packet loss.

[0251] Step 30: Input the fault signal into the fault analysis solution library to obtain the corresponding fault ID.

[0252] The fault ID is a code used to uniquely identify a fault in the fault analysis solution library. After the staff enters the key data such as the equipment status, alarm information, and operation records at the time of each fault event into the fault analysis solution library, the system will automatically generate a unique fault ID according to the preset coding rules. This code serves as the digital identity of the fault in the system and is used to establish a fault file index, associate historical data, and facilitate subsequent analysis and tracing.

[0253] Step 31: Input the fault cause, fault symptoms and corresponding historical data information into the fault analysis solution database according to the fault ID.

Claims

1. A method for diagnosing a fault in an HPLC, characterized by, The method comprises the following steps: Step 1: obtaining historical data information on a line corresponding to a fault signal in response to the fault signal; Step 2: inputting the historical data information into a preset fault analysis model to obtain a fault probability corresponding to all preset fault causes; Step 20: outputting a preset false fault signal when all the fault probabilities do not exceed a preset fault probability threshold; Step 21: defining a corresponding fault probability as a suspected fault probability and a suspected fault cause corresponding to the suspected fault probability as the suspected fault cause when the fault probability exceeds the fault probability threshold; Step 210: obtaining a fault analysis scheme from a preset fault analysis scheme library based on the suspected fault cause; Step 211: executing the fault analysis scheme to obtain a credibility parameter; Step 212: outputting the suspected fault cause corresponding to the credibility parameter when the credibility parameter is higher than a preset credibility threshold; After step 21, the method further comprises the following steps: Step 22: defining a number of suspected fault probabilities as a suspected fault number; Step 220: obtaining a fault analysis scheme from the fault analysis scheme library based on the suspected fault cause if the suspected fault number is equal to 1; Step 221: combining all suspected fault causes based on the suspected fault number to obtain a suspected fault cause group if the suspected fault number is greater than 1; Step 2210: inputting the historical data information into a preset multi-source fault analysis model to obtain a joint fault probability; Step 2211: taking the joint fault probability with the highest value as a suspected joint fault probability; Step 2212: obtaining a fault analysis scheme from the fault analysis scheme library based on the suspected joint fault probability when the suspected joint fault probability corresponds to the suspected fault cause and the suspected fault cause corresponding to the suspected fault cause group is the same; Step 2213: continuing to execute step 211; Step 2214: defining the fault cause corresponding to the fault probability with the largest value as the suspected fault cause and continuing to execute step 210 when the suspected joint fault probability corresponds to the suspected fault cause and the suspected fault cause corresponding to the suspected fault cause group is different; The method further comprises a method for obtaining a fault analysis scheme from the fault analysis scheme library when there is no fault probability exceeding the fault probability threshold, which comprises the following steps: Step 222: obtaining a standard fault phenomenon corresponding to all the fault causes from a preset fault analysis database; Step 223: performing a correlation analysis on all the standard fault phenomena to obtain a similarity parameter between all the fault causes, the similarity parameter representing the similarity of the standard fault phenomena between different fault causes; Step 224: integrating corresponding fault causes into a same-phenomenon fault cause group when the similarity parameter is greater than a preset similarity threshold; Step 225: accumulating the fault probabilities corresponding to all the fault causes in the same-phenomenon fault cause group to obtain an accumulated fault probability; Step 2250: If the cumulative failure probability is greater than the preset cumulative probability threshold, all the failure causes in the same phenomenon failure cause group are regarded as the suspected failure causes, and step 210 is continued. Step 2251: If there is no cumulative fault probability greater than the cumulative probability threshold, proceed to step 20.

2. The HPLC fault diagnosis method according to claim 1, characterized in that, It also includes a method for obtaining a fault analysis scheme from the fault analysis scheme library even when the cumulative fault probability is not greater than the cumulative probability threshold, the method comprising: Step 22510: Based on the historical data information, analyze the actual feature parameter change trends corresponding to all preset feature information types; Step 22511: Based on the standard fault phenomena, obtain the changing trends of all standard characteristic parameters corresponding to all fault causes; Step 22512: Traverse all the fault causes to obtain a multi-source fault cause group. The variation trends of all the standard characteristic parameters in the multi-source fault cause group are superimposed and are the same as the variation trends of the actual characteristic parameters. Step 22513: Obtain a fault analysis scheme from the fault analysis scheme library based on all the fault causes in the multi-source fault cause group and output it.

3. The HPLC fault diagnosis method according to claim 2, characterized in that, It also includes a fault analysis method when the multi-source fault cause group does not exist, the method comprising: Step 225120: Obtain the generation time, generation location, and implementation environment parameters of the fault signal; Step 225121: Search the preset historical fault cause table for historical fault events, historical fault causes, and fault occurrence frequencies under the conditions of the occurrence time and the occurrence location; Step 225122: Retrieve the historical environmental parameters corresponding to the historical fault events; Step 225123: Perform a similarity analysis between the implemented environmental parameters and the historical environmental parameters to obtain environmental similarity parameters; Step 225124: When the environmental similarity parameter is greater than the preset environmental similarity threshold, the corresponding historical fault event is defined as a suspected fault event; Step 225125: Define the historical fault cause corresponding to the suspected fault event with the highest frequency of occurrence as the suspected fault cause.

4. The HPLC fault diagnosis method according to claim 2, characterized in that, It also includes a method for obtaining and outputting the credibility parameter, the method comprising: Step 2110: Obtain the transmission data information of the line to which the fault signal belongs; Step 21100: When the transmitted data information is not obtained, the fault signal is defined as a data acquisition abnormal signal; Step 211001: Based on the preset data cause table, find the change trends of all historical feature parameters corresponding to the abnormal data acquisition signal; Step 211002: Perform a similarity analysis between the historical characteristic parameter change trends and the actual characteristic parameter change trends to obtain trend similarity parameters; Step 211003: Define the trend similarity parameter with the largest value as the trend most similar parameter; Step 211004: Obtain the anomaly correction parameter from the preset collection anomaly correction parameter table according to the trend most similar parameter; Step 211005: Obtain the anomaly correction confidence parameter based on the confidence parameter and the anomaly correction parameter, and output it as the confidence parameter; Step 21101: When the transmitted data information is obtained, the fault signal is defined as a data integrity abnormality signal; Step 211010: Obtain complete data information from the preset backup line; Step 211011: Based on the complete data information and the transmitted data information, obtain the cause of data loss from the preset data loss principle table; Step 211012: Obtain integrity correction parameters from the preset data integrity parameter table based on the reasons for data loss and the suspected causes of failure; Step 211013: Obtain the integrity correction confidence parameter based on the confidence parameter and the integrity correction parameter, and output it as the confidence parameter.

5. The HPLC fault diagnosis method according to claim 4, characterized in that, Methods for obtaining and outputting credibility parameters also include: Step 21102: When the data integrity error signal and the data acquisition error signal are received alternately during a preset historical time period, the fault signal is defined as a mixed error signal; Step 211020: Analyze the proportion of the data integrity anomaly signal and the data acquisition anomaly signal in the historical time in the mixed anomaly signal, and define it as the mixed anomaly proportion; Step 211021: Based on the simulated fault results corresponding to the suspected fault causes of all the fault analysis schemes, obtain simulated abnormal signals; Step 211022: Analyze the proportion of the data integrity anomaly signal and the data acquisition anomaly signal in the simulated anomaly signals during the historical time period, and define it as the simulated anomaly proportion; Step 211023: Obtain the hybrid correction parameter from the preset hybrid correction database based on the ratio of the hybrid anomaly ratio to the simulated anomaly ratio; Step 211024: Obtain the mixed correction confidence parameter based on the confidence parameter and the mixed correction parameter, and use the mixed correction confidence parameter as the confidence parameter, then continue to execute step 212.

6. The HPLC fault diagnosis method of claim 1, wherein, A diagnostic method for when the confidence parameter is lower than the confidence threshold, the method comprising: Step 2120: Sort all the suspected fault causes based on the fault probability to obtain a suspected fault cause sorting table; Step 2121: Traverse the suspected fault cause sorting table and define the next suspected fault cause as the next suspected fault cause; Step 21210: When the confidence parameter of the fault analysis scheme corresponding to the suspected fault cause is lower than the confidence threshold, the next suspected fault cause is taken as the suspected fault cause, and step 2121 is continued. Step 21211: When the credibility parameter of the fault analysis scheme corresponding to the suspected fault cause is higher than the credibility threshold, output the suspected fault cause corresponding to the credibility parameter.

7. The HPLC fault diagnosis method according to claim 6, characterized in that, A diagnostic method for when the confidence parameter of the fault analysis scheme corresponding to the suspected fault cause is higher than the confidence threshold, the method comprising: Step 212100: Locate all the preset fault device points in the aforementioned fault analysis scheme; Step 212101: Based on integrating the fault analysis schemes corresponding to the fault device points, obtain a fault analysis scheme group; Step 212102: Sum all the confidence parameters of the fault analysis scheme group to obtain the device confidence level; Step 212103: When the confidence level of the device point exceeds a preset device confidence level threshold, the possible causes of failure of the faulty device point corresponding to the confidence level of the device point are obtained. Step 212104: Output the possible causes of failure as the suspected causes of failure.

8. The HPLC fault diagnosis method of claim 1, wherein, A method for maintaining the fault analysis scheme library, the method comprising: Step 3: Upon receiving a preset fault repair signal, obtain the fault cause and fault phenomenon corresponding to the fault signal; Step 30: Input the fault signal into the fault analysis scheme library to obtain the corresponding fault ID; Step 31: Input the fault cause, the fault phenomenon, and the historical data information corresponding to the fault phenomenon into the fault analysis scheme library according to the fault ID.

Images