Fault diagnosis methods, devices, storage media and electronic equipment

Abstract

This application provides a fault diagnosis method, apparatus, storage medium, and electronic device, relating to the field of fault diagnosis. Specifically, after an initial alarm occurs in the target device, instead of directly confirming the fault, it first acquires the operating information within a preset time window prior to the alarm occurrence. Based on this, it selects comparable reference state information from the device's historical operating information, and then compares the current operating information with these historical reference state information for judgment. In this way, it can identify spurious sudden changes in indicators caused by speed deviations, noise interference, or improper threshold settings, and also highlight the subtle evolution of fault characteristics, thereby overcoming the limitation of sensors that can only process single-moment, single-slice data, and improving the reliability of fault judgment.

Description

Technical Field

[0001] This application relates to the field of fault diagnosis, and more specifically, to a fault diagnosis method, apparatus, storage medium, and electronic device. Background Technology

[0002] In industrial production, moving equipment is a core component ensuring the stable operation of the production line. However, as equipment service life continues to extend, mechanical wear and performance degradation intensify, significantly increasing the risk of sudden equipment failures and hidden damage. This not only threatens production safety but also directly impacts operating efficiency and maintenance costs. Therefore, ensuring equipment operates in a healthy condition has become a critical technical issue that manufacturing enterprises urgently need to address.

[0003] Therefore, the industry generally adopts fault diagnosis systems based on online monitoring, which dynamically assess and provide early warnings of equipment health status by real-time monitoring, analysis and prediction of equipment operating status, thereby supporting timely intervention and precise maintenance, and improving the overall reliability and operation and maintenance efficiency of the system.

[0004] Research revealed that current monitoring methods primarily employ static or semi-dynamic numerical comparison logic. This involves directly comparing single-moment, single-slice, and single-dimensional feature indicators during equipment operation with pre-set or periodically updated thresholds. An alarm is triggered once the feature indicator exceeds the threshold. Therefore, current equipment status warnings are highly dependent on the accuracy of these feature indicators. However, limited by sensor hardware performance, the accuracy of single-moment, single-slice, and single-dimensional feature indicators in real-time monitoring is insufficient, leading to missed and false alarms during equipment monitoring. Summary of the Invention

[0005] In order to overcome at least one of the shortcomings of the prior art, one of the objectives of this application is to provide a fault diagnosis method, device, storage medium and electronic device that can identify the false sudden changes in indicators caused by speed deviation, noise interference or improper threshold setting, and can highlight the subtle evolution of fault characteristics, thereby overcoming the limitation that the sensor end can only process single-moment and single-slice data and improving the reliability of fault judgment.

[0006] In a first aspect, this application provides a fault diagnosis method, the method comprising: In response to a preliminary alarm from the target device, the operating information of the target device within a preset time window prior to the occurrence of the preliminary alarm is obtained, wherein the preliminary alarm indicates that at least one device state of the target device exceeds a corresponding state threshold. Based on the operational information, reference status information is selected from the historical operational information of the target device; Based on the operational information and the reference status information, it is determined whether the target device has a fault.

[0007] Secondly, this application provides a fault diagnosis device, the device comprising: An alarm response module is used to respond to a preliminary alarm from a target device and obtain the operating information of the target device within a preset time window before the preliminary alarm occurs, wherein the preliminary alarm indicates that at least one device state of the target device exceeds a corresponding state threshold. The information optimization module is used to select reference status information from the historical operating information of the target device based on the operating information; The fault confirmation module is used to determine whether the target device has a fault based on the operating information and the reference status information.

[0008] Thirdly, this application provides a storage medium storing a computer program that, when executed by a processor, implements the fault diagnosis method.

[0009] Fourthly, this application provides an electronic device, which includes a processor and a memory, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the fault diagnosis method.

[0010] Compared with the prior art, this application has the following beneficial effects: The fault diagnosis method, apparatus, storage medium, and electronic equipment provided in this application do not directly confirm the fault after the target equipment experiences an initial alarm. Instead, they first acquire the operating information within a preset time window before the alarm occurred, and then select comparable reference state information from the equipment's historical operating information. Finally, they compare the current operating information with these historical reference state information for judgment. In this way, they can identify false spikes in indicators caused by speed deviation, noise interference, or improper threshold settings, and highlight the subtle evolution of fault characteristics. This overcomes the limitation of sensors that can only process single-moment, single-slice data, thus improving the reliability of fault judgment. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A flowchart illustrating the fault diagnosis method provided in the embodiments of this application; Figure 2A schematic diagram illustrating the principle of the fault diagnosis method provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the fault diagnosis device provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments of this application (hereinafter referred to as "the embodiments") clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0014] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0015] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0016] In the description of this application, it should be noted that the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0017] Based on the above statement, as introduced in the background section, current equipment status early warning results highly depend on the accuracy of feature indicators. However, limited by sensor hardware performance, single-moment, single-slice, and single-dimensional feature indicators in real-time monitoring suffer from insufficient accuracy, leading to missed and false alarms during equipment monitoring. Specifically, a single-slice feature indicator refers to a segment of raw time-domain signal continuously acquired and captured within a fixed time window. Its characteristics include isolation, static nature, lack of contextual temporal correlation, and reflection of only the local dynamic response of the equipment under that instantaneous operating condition.

[0018] In related technologies, a dynamic self-learning early warning threshold can be calculated during normal operation using a dynamic self-learning early warning threshold algorithm. The trend data of characteristic indicators can be filtered, and an early warning can be issued when the real-time filtered trend data of the unit exceeds the dynamic self-learning early warning threshold space.

[0019] For example, a non-invasive signal acquisition scheme can be used to collect vibration, magnetic field, temperature, and sound signals during the operation of normal and faulty equipment. Key features from these signals can be extracted as baseline state parameters of the equipment. Based on the feature vectors under different fault states, a fault determination model can be constructed. Then, using the currently collected equipment operating parameter features as model input, multi-dimensional state determination during equipment operation can be achieved.

[0020] However, the aforementioned technologies all directly compare the feature indicators of a single moment, a single slice, and a single dimension during the operation of the equipment with pre-set or periodically updated thresholds. Therefore, they are easily affected by the accuracy of data acquisition, resulting in missed and false alarms in practical applications.

[0021] It should be noted that the defects in the solutions in the prior art are the result of practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed by the embodiments of this application in the following text should be regarded as contributions to this application in the process of invention and creation, and should not be understood as technical content known to those skilled in the art.

[0022] Based on the discovery of the above-mentioned technical problems, this embodiment provides a fault diagnosis method. For example... Figure 1 As shown, the method includes: S1, in response to the initial alarm of the target device, obtains the operating information of the target device within a preset time window before the initial alarm occurs.

[0023] The initial alarm indicates that at least one device state of the target device exceeds the corresponding state threshold.

[0024] S2, Select reference status information from the historical operating information of the target device based on the operating information.

[0025] S3 determines whether the target device has a fault based on the operating information and reference status information.

[0026] This embodiment can be understood as follows: after the target device experiences an initial alarm, instead of directly confirming the fault, it first acquires the operating information within a preset time window prior to the alarm, and then selects comparable reference state information from the device's historical operating information. Finally, it compares the current operating information with this reference state information for judgment. In this way, it can identify spurious sudden changes in indicators caused by speed deviations, noise interference, or improper threshold settings, and also highlight the subtle evolution of fault characteristics. This overcomes the limitation of sensors only being able to process single-moment, single-slice data, thus improving the reliability of fault determination.

[0027] Furthermore, the fault diagnosis method provided in this embodiment can be implemented using electronic devices such as mobile terminals, tablet computers, laptop computers, desktop computers, and servers. The server can be a single server or a group of servers. The server group can be centralized or distributed (e.g., the servers can be a distributed system). In some embodiments, the server can be local or remote relative to the user terminal. In some embodiments, the server can be implemented on a cloud platform; as an example only, the cloud platform can include private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, inter-cloud, multi-cloud, etc., or any combination thereof. In some embodiments, the server can be implemented on an electronic device having one or more components.

[0028] To make the solution provided in this embodiment clearer, the following uses the server as an example, and combines it with... Figure 1 Each step of the method is described in detail. However, it should be understood that the operations in the flowchart may not be implemented in sequence, and steps without logical contextual relationships may be reversed in order or performed simultaneously. Furthermore, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowchart, or remove one or more operations from the flowchart. See also... Figure 1 The method includes: S1, in response to the initial alarm of the target device, obtains the operating information of the target device within a preset time window before the initial alarm occurs.

[0029] The initial alarm indicates that at least one device state of the target device exceeds the corresponding state threshold.

[0030] In this embodiment, when the server responds to the initial alarm of the target device, it can obtain the operating information of the target device within a preset time window before the initial alarm occurs. The initial alarm indicates that at least one device state of the target device exceeds a corresponding state threshold, which is a pre-set numerical boundary used to determine whether the device is abnormal. The operating information specifically includes at least one characteristic indicator on which the alarm is based. These indicators include a trend curve of change over a period of time before the alarm, as well as the original waveform segment (e.g., vibration waveform) corresponding to the alarm time, its FFT-transformed spectrum, envelope spectrum, and other time-frequency data.

[0031] Based on the above operational information, the following will continue to... Figure 1 Step S2 will be explained below: S2, Select reference status information from the historical operating information of the target device based on the operating information.

[0032] It should be noted that in current online monitoring systems for industrial equipment, upon receiving an initial alarm, subsequent diagnostic processes are typically initiated directly. However, this initial alarm may itself originate from non-fault factors such as sensor loosening, electromagnetic interference, or instantaneous fluctuations in operating conditions, thus constituting a false alarm. Because existing technology lacks a pre-emptive mechanism to verify the authenticity of initial alarms, the server must perform high-computational secondary diagnostics on all false alarms, resulting in a waste of computing resources.

[0033] In view of this, the server also calls the pre-trained pre-detection model to process the operation information to obtain the pre-identification result; if the pre-identification result indicates that the preliminary alarm is not a false alarm, then the execution condition for selecting reference status information from the historical operation information of the target device is based on the operation information.

[0034] In practical applications, the server can call a pre-trained pre-detection model to process the operational information to obtain the pre-identification result. This pre-detection model can be a judgment model built based on expert experience, used to verify whether the operational information meets the statistical distribution characteristics that a normal signal should possess, such as amplitude fluctuation range, spectral energy concentration, and time-domain waveform periodic stability; or it can be an identification model trained based on machine learning methods, whose training data includes a large number of samples labeled "normal" and "false alarm," and the input data covers multi-dimensional information such as the original vibration waveform and corresponding spectrum. Since the server does not have the performance and energy consumption limitations of edge sensors, the identification model can be scaled up in terms of depth, parameter magnitude, and computational complexity, thus enabling accurate identification of false alarms caused by sensor loosening, external electromagnetic interference, or hardware defects.

[0035] Based on this, the server will only perform the operation of selecting reference status information from historical operation information if the preliminary identification results indicate that the initial alarm is not a false alarm, so as to carefully diagnose the initial alarm and ensure that all subsequent diagnoses are based on real and valid alarms.

[0036] It should also be noted that in current industrial equipment fault diagnosis, after receiving an initial alarm, the server typically judges the fault directly based solely on real-time data collected in a single time period, lacking targeted comparison with historical operating states. In this case, the reference state information selected by the server often uses a fixed time point (such as before the most recent shutdown) or the average value of all historical data, failing to adapt to the specific abnormal characteristics reflected in this alarm. Since the key monitoring indicators corresponding to different fault types are different—for example, bearing spalling faults are mainly associated with envelope spectrum characteristics, while bearing cage faults are more significantly reflected in changes in the envelope spectrum—this embodiment provides the following optional implementation of step S2: S2-1, Match the set of rules to be verified that are associated with the runtime information from the diagnostic rule base based on expert experience.

[0037] In practical applications, the server can construct a many-to-many mapping matrix between alarm indicators and diagnostic rule groups. This matrix is ​​formed based on the long-term experience accumulated by diagnostic experts and a large number of historical fault cases. Each set of diagnostic rules consists of two or more feature indicators combined according to logical relationships.

[0038] For example, consider a motor bearing cage failure. It should be understood that in motor fault diagnosis, the original vibration signal usually contains a high-frequency carrier excited by mechanical impact and a low-frequency modulated signal with typical fault characteristic frequencies. The envelope of the signal (i.e., the trend of the impact amplitude) can be extracted first, and then a Fourier transform can be performed on the envelope. The result is the envelope spectrum.

[0039] It should also be understood that in motor bearings, the cage corresponds to a theoretical fault characteristic frequency, and the frequency range selected based on this characteristic frequency is called the cage characteristic frequency band. The so-called energy value of the envelope spectrum in this characteristic frequency band (simply referred to as the envelope spectrum cage energy) refers to the summation or integration of the amplitude or the square of the amplitude of the envelope spectrum within this frequency range, which characterizes the response intensity of the current operating information in the cage fault-related frequency region.

[0040] Velocity spectrum frequency harmonics refer to the spectral line components that appear at integer multiples of the rotational speed frequency in the velocity spectrum of a vibration signal (i.e., the frequency domain diagram obtained by performing a Fourier transform on the vibration velocity signal). Their amplitude variations can be used to reflect defects in the bearing cage.

[0041] Based on the above explanation, the following combination rules can be set: "The energy of the envelope spectrum cage increases, and periodic impacts occur in the time-domain waveform cage interval." "Envelope spectrum cage energy increases + spectral noise energy rises" "The envelope spectrum cage characteristics disappear, while the velocity spectrum frequency conversion harmonic characteristics appear." The server can match the rule group to be verified from the diagnostic rule base based on the list of feature indicators contained in the operation information, thereby narrowing down the scope of rules to be verified.

[0042] For example, when the operation information includes the indicator "appearance of speed spectrum frequency conversion harmonic characteristics", the associated diagnostic rule groups such as "increase in total speed value + appearance of speed spectrum frequency conversion harmonic characteristics" and "disappearance of envelope spectrum cage characteristics + appearance of speed spectrum frequency conversion harmonic characteristics" are activated, thereby limiting the specific set of rules for subsequent verification.

[0043] In this way, the diagnostic path most likely related to the current alarm can be located from the diagnostic rule base, avoiding the waste of resources caused by blindly traversing all rules.

[0044] Based on the set of rules to be verified as described in the above implementation, step S2 further includes: S2-2, Based on the feature indicators in the rule group to be verified, select reference status information corresponding to the feature indicators from historical operation information.

[0045] The reference status information includes multiple startup status data closest to the initial alarm, multiple reference limit indicators corresponding to the characteristic indicators, and multiple reference steady-state indicators corresponding to the characteristic indicators.

[0046] It should be understood here that startup status data refers to the data when the device is powered on, that is, the various data collected in the tens of seconds before the device starts from a standstill and enters stable operation, which represents the actual performance of the device under startup shock in a good state.

[0047] As an optional implementation, the server can select multiple startup status data closest to the initial alarm from historical operation information based on the alarm time of the initial alarm; determine historical indicator sequences of the same type as the feature indicators from historical operation information, and select multiple local maxima from them as multiple reference limit indicators; determine historical indicator sequences of the same type as the feature indicators from historical operation information, and extract sequence segments within the preset indicator fluctuation range from them; and extract multiple reference steady-state indicators from the sequence segments.

[0048] In practical applications, to ensure the scientific validity of comparisons under the same working conditions, under the constraints of time span and sample size, we can analyze the historical trend of each characteristic indicator over a period of time based on the characteristic indicators contained in the converged set of rules to be verified, and select the above three types of reference samples accordingly.

[0049] Among them, the multiple startup status data sets most recent to the initial alarm refer to the data selected by the server from historical operational information based on the alarm time of the initial alarm. The status data corresponding to the most recent startup time is used to analyze and confirm the status level of the device at the time of the alarm.

[0050] Multiple reference limit indicators refer to the multiple local maxima selected by the server from the historical indicator sequence of the same type as the feature indicator after determining it from the historical operation information. These maxima are used to assess whether the current alarm has exceeded the safety limit that the device has experienced in the past.

[0051] Multiple reference steady-state indicators refer to the server further identifying historical indicator sequences of the same type as the characteristic indicators from historical operational information, extracting sequence segments within the preset indicator fluctuation range, and then extracting [the relevant data] from these sequence segments. Each sample serves as a reference steady-state index, i.e., a benchmark steady-state reference point, to establish a health baseline for the equipment, and residual analysis is used to quantify the degree of deviation of the current operating state relative to this baseline.

[0052] For example, when the server detects that the vibration frequency of a motor suddenly exceeds the limit during operation and triggers a preliminary alarm, the preliminary alarm indicates that at least one device state of the target device exceeds the corresponding state threshold.

[0053] At this point, the server first identifies the key characteristic indicator (e.g., twice the frequency conversion amplitude in the speed spectrum) upon the set of diagnostic rules activated by the alarm (e.g., the appearance of frequency conversion harmonic characteristics in the speed spectrum); then, based on the time of the alarm occurrence, it retrieves the three most recent startup status data sets (i.e., the three sets of data stored in the historical database that contain historical operation information) for that specific indicator. (Each alarm status reference point includes the impact waveform and corresponding spectrum data after three normal startups, used to determine whether the current alarm originates from the repeated reproduction of the inherent impact during the startup process.)

[0054] Secondly, for the same characteristic indicator ("2 times the frequency amplitude in the speed spectrum"), the server retrieves its historical indicator sequence over the past 30 days and identifies four local maxima points (e.g., the peak values ​​on April 12, April 15, April 17, and April 19) as local extreme reference points, forming a set of safety upper limits that the device has actually reached in this indicator dimension.

[0055] Finally, the server filters out stable periods with fluctuations of less than ±5% within a continuous 120-minute timeframe (e.g., 10:00–12:00 on April 16th) from the historical sequence of the same characteristic indicator, and randomly selects 5 time points from these periods according to time weight (i.e. (A reference steady-state reference point) is used as the reference steady-state reference point to form the steady-state reference information of the equipment under healthy operating conditions.

[0056] Based on the reference state information described in the above embodiments, the following will continue to... Figure 1 Step S3 will be explained below: S3 determines whether the target device has a fault based on the operating information and reference status information.

[0057] It should be noted that the fault diagnosis systems currently widely deployed in industrial sites usually only perform single feature extraction based on real-time collected operating information. The resulting features are mostly static instantaneous amplitudes, spectral peaks, or simple trend indicators, which are difficult to reflect the dynamic process of a fault from its initial onset to its later deterioration.

[0058] Therefore, these static features are extremely sensitive to non-fault disturbances such as speed fluctuations, sensor loosening, and electromagnetic interference, and are prone to false alarms due to single-point data distortion. At the same time, when the equipment is in the early stage of a minor fault, its single waveform or spectrum change is often not significant. Existing methods cannot capture the evolution signs of feature quantity drift, decay, or periodic abrupt changes in the time dimension, which leads to an increased risk of missed alarms.

[0059] Therefore, this embodiment provides the following optional implementation methods for step S3: S3-1, extract features from the running information and reference status information to obtain a refined feature set.

[0060] This embodiment can be understood as follows: by utilizing the computing performance and storage capacity of the server, which far exceeds that of the edge device, more complex feature calculations are performed than those of the edge sensor, under the premise of significantly reduced data volume, thereby obtaining a fine feature set with physical interpretability and diagnostic support.

[0061] In practical applications, the server does not extract features from the running information and the reference state information separately, but rather performs comparative analysis so that the extracted information represents the degree of change in the current state relative to the historical state.

[0062] For example, the energy value of the envelope spectrum of the operation information in the characteristic frequency band of the cage is calculated, and then the mean and standard deviation of the energy of each reference steady-state index in the same frequency band are calculated simultaneously, thereby deriving the normalized residual as a fine feature. In addition, the time-domain waveform of the operation information can be compared and analyzed with the corresponding waveform of each start-up state data to extract the similarity information and confidence level between them; and the velocity spectrum of the operation information and the spectrum of each reference limit index are respectively quantified with structural difference to generate a spectral envelope offset sequence.

[0063] In this way, the refined feature set carries a clear physical meaning, thereby improving the accuracy of fault identification and the interpretability of the conclusions.

[0064] Based on the description of the fine feature set in the above embodiments, step S3 further includes: S3-2, analyze the evolution of fine features in the fine feature set over time to obtain the evolution feature set corresponding to the fine feature set.

[0065] In this embodiment, the server uses time as the main line to sort the values ​​of the fine features in the fine feature set at different time points, forming a dynamic sequence with a chronological order and direction of change. From this sequence, an indicator that can reflect the evolution trend of the device state is extracted, namely the evolution feature set.

[0066] In practical applications, the server performs first-order or higher-order difference operations on each fine feature based on its value at each time point. For example, it calculates the change or rate of change of the feature value between two adjacent time points to obtain the amplitude growth rate; or it calculates the curvature change at multiple consecutive time points to identify abrupt inflection points; or it performs sliding window fitting on the feature value sequence to extract the trend slope and acceleration.

[0067] The server further transforms the explicit physical meaning of fine features into semantically clear and reasonable evolutionary features. For example, it quantifies the difference in spectral structure between the current alarm time and the benchmark steady-state reference point as spectral envelope similarity. The lower the value, the more severe the current spectral distortion. Another example is modeling the sequential changes of bearing fault characteristic frequencies at multiple reference steady-state index points as drift trajectories. When this trajectory shows accelerated deviation or sudden disappearance, it constitutes a strong criterion pointing to bearing spalling or cage breakage.

[0068] For example, if one of the fine features extracted when an alarm is triggered is the bearing outer ring fault characteristic frequency value, this value itself is a primary indicator; while the change in the bearing fault characteristic frequency is a secondary indicator, and this indicator can reflect certain bearing faults.

[0069] Based on the above embodiments regarding the evolutionary feature set, step S3 further includes: S3-3 determines whether the target equipment has a fault based on the fine feature set, the evolution feature set, and historical operation information.

[0070] It should be noted that current industrial equipment fault diagnosis systems often rely solely on a single path to output diagnostic conclusions after an initial alarm, lacking both cross-validation of the fault location and explanation of the judgment basis. In practical applications, the server can generate a set of rules to be verified based on expert experience and identify the first location and cause of the fault, but this result only reflects causal inference at the experience level; the server can also call a fault diagnosis model to output the second location and confidence level, but this result only reflects statistical correlation at the data level. As an optional implementation, this embodiment combines the two to provide a dual-engine structure for fault diagnosis. Based on this dual-engine structure, step S3-3 may include: S3-3-1 uses the rule set to be verified to identify the fine feature set, evolution feature set and historical operation information to obtain the first part of the target equipment and the cause of the failure.

[0071] In practical applications, the server can invoke a set of rules to be verified. This set of rules is a generalized set of rules summarized by diagnostic experts based on long-term practice, reflecting the correlation between equipment faults and characteristic indicators of monitoring data. However, it should be noted that experts only provide qualitative descriptions of fault modes; for example, what characteristic combinations might accompany typical faults such as imbalance, misalignment, and bearing spalling. Therefore, the server inputs trend-changing characteristic indicators detected from the refined feature set, the evolving feature set, and historical operating information, and outputs a set of rules that logically match these indicators. By combining primary alarm indicators with secondary processing indicators, the server can pinpoint the specific fault type and further determine the primary location and cause of the fault. Therefore, the diagnostic conclusions obtained based on the set of rules to be verified possess a traceable chain of evidence.

[0072] S3-3-2 uses a pre-trained fault diagnosis model to identify the fine feature set, evolution feature set, and historical operation information to obtain the second part of the target equipment and the confidence level.

[0073] This embodiment can be understood as identifying the fault location based on historical fault cases. In practical applications, the server can call a pre-trained fault diagnosis model, which is a deep neural network trained using a historical fault case database. However, it should be noted that the data-driven part does not operate independently of expert knowledge. Instead, it uses rule sets extracted from expert experience, combined with past fault case data, to quantitatively model the specific values ​​or changes of relevant feature indicators. The model's input is historical fault cases from the database, and its output is the specific values ​​or changes of the feature indicators corresponding to each rule set. Therefore, the server takes the refined feature set, the evolutionary feature set, and historical operating information as input, and after processing by the model, identifies the second location and its confidence level. This ensures that the diagnostic conclusion is supported by statistical data patterns.

[0074] S3-3-3, combining the first part and the cause of the fault with the second part and the confidence level, the fault diagnosis result of the target equipment is obtained.

[0075] As an optional implementation, if the first part and the second part are the same target part of the target device and the confidence level is greater than the threshold, then it is determined that the target part is faulty; If the first part and the second part are different parts, then obtain the first chain of evidence for the cause of the failure, and the second chain of evidence for the failure of the second part. Based on the strength of the evidence in the first and second chains of evidence, the target part with the malfunction is determined from the first and second parts.

[0076] This embodiment can be understood as comparing and evaluating the strength of evidence by combining the qualitative judgment formed by expert experience with the quantitative confidence level output by the data-driven model, thereby ensuring that the final diagnostic conclusion not only conforms to the objective laws of equipment operation, but also withstands the verification of engineering practice.

[0077] It should be noted that most current AI-based fault diagnosis methods adopt an end-to-end modeling approach, which directly uses the original vibration sampling points or unanalyzed time-frequency diagrams as input. The model itself is an uninterpretable black box, resulting in output conclusions that cannot correspond to specific frequency components, amplitude changes, or time-series evolution characteristics, making it difficult for field engineers to judge their rationality.

[0078] Therefore, in this embodiment, when the first part and the second part are identical and the confidence level exceeds a preset threshold, the server can confirm that the part is faulty; when they are inconsistent, the server further extracts a first chain of evidence supporting the judgment of the first part and a second chain of evidence supporting the judgment of the second part, wherein each chain of evidence consists of several traceable physical indicators. For example, the first chain of evidence may include a 42% increase in the energy of the cage characteristic frequency in the envelope spectrum at the alarm time.

[0079] The second chain of evidence consists of one or more input feature indicators and their corresponding values ​​that the second part and confidence level output by the fault diagnosis model actually depend on. For example, the network spectrum energy that is strongly correlated with the second part in the fine feature set, the rate of change that characterizes its evolution in the evolution feature set, and the sequence of similar historical indicators used for comparison in historical operation information.

[0080] Ultimately, the server can quantitatively assess the overall evidentiary strength of the two chains of evidence based on dimensions such as measurement certainty, historical consistency, and operational condition matching of each piece of evidence, and determine the target part that better reflects the actual state of the equipment.

[0081] Based on the fault diagnosis results obtained from the above embodiments, in practical applications, if the server confirms that a certain part has a fault, it can automatically trigger the rights confirmation enhancement logic, update the alarm threshold and feature label of the corresponding part in the primary alarm, and generate a structured closed-loop report containing the three elements of "evidence chain - conclusion - countermeasures" for on-site personnel to quickly access and verify.

[0082] If the server determines that the alarm is a false alarm, it can immediately execute the false alarm suppression logic, clear the pending state of the alarm and refresh the cached data, so as to avoid repeated alarms caused by transient interference or fluctuations in operating conditions from disrupting the operation and maintenance rhythm.

[0083] In this way, the server not only completes the output of a diagnostic conclusion, but also realizes the dynamic calibration and continuous optimization of the entire monitoring and diagnostic process, thereby improving the robustness and reliability of fault diagnosis methods in complex industrial sites.

[0084] In the above embodiments, each step of the fault diagnosis method provided in this application has been described in detail. The complete execution process of this method can be simplified as follows: Figure 2 To make the entire methodology clearer and more coherent, the following section combines... Figure 2 The method is described in its entirety.

[0085] When the server receives a preliminary alarm and operational information within a preset time window prior to the alarm's occurrence, it first determines whether the preliminary alarm is a false alarm. If the determination is yes, the alarm is reset. Otherwise, if the determination is no, it matches a set of rules to be verified against the preliminary alarm and operational information from a diagnostic rule base based on expert experience. Then, based on the feature indicators in the set of rules to be verified, it selects corresponding historical reference points from the target device's historical operational information. Next, it reads the historical reference points and calculates refined feature indicators. Then, it performs secondary feature indicator processing on the refined feature indicators to obtain an evolutionary feature set. Finally, it inputs the refined feature indicators, the evolutionary feature set, and the historical operational information into the set of rules to be verified to complete the rule group application, and finally outputs a secondary diagnostic conclusion. Finally, it updates the preliminary alarm based on the secondary diagnostic conclusion.

[0086] In summary, in this embodiment, after the initial alarm is issued, it is first determined whether the initial alarm is a false alarm. If it is confirmed not to be a false alarm, the set of rules to be verified is matched, the corresponding historical reference point is selected, the waveform is read and fine-grained feature indicators are calculated, and the evolution feature set is processed to obtain the final evolution feature set. Then, the set of rules to be verified is used to comprehensively analyze the fine-grained feature indicators, the evolution feature set, and historical operational information, finally outputting a secondary diagnostic conclusion. This avoids misjudgment of single-point thresholds and achieves dual verification of expert experience and data models.

[0087] Based on the same inventive concept as the fault diagnosis method provided in this embodiment, this embodiment also provides a fault diagnosis device, which includes at least one software functional module that can be stored in a memory or embedded in an electronic device. The processor in the electronic device executes the executable module stored in the memory. For example, the software functional module and computer program included in this device. Please refer to... Figure 3 Functionally, fault diagnosis devices can include: Alarm response module 11 is used to respond to the initial alarm of the target device and obtain the operating information of the target device within a preset time window before the initial alarm occurs, wherein the initial alarm indicates that at least one device state of the target device exceeds the corresponding state threshold. Information optimization module 12 is used to select reference status information from the historical operating information of the target device based on the operating information; The fault confirmation module 13 is used to determine whether the target device has a fault based on the operating information and reference status information.

[0088] In this embodiment, the alarm response module 11 is used to implement Figure 1 In step S1, the information optimization module 12 is used to implement Figure 1 In step S2, the fault confirmation module 13 is used to implement... Figure 1Step S3 in the above process. Therefore, for a detailed description of each of the above modules, please refer to the specific methods described in the corresponding steps above.

[0089] Optionally, the information optimization module 12 selects reference status information from the historical operating information of the target device based on the operating information in the following ways: Match the groups of rules to be verified that are associated with the operational information from the diagnostic rule base based on expert experience; Based on the feature indicators in the rule group to be verified, select reference status information corresponding to the feature indicators from historical operation information.

[0090] Optionally, the fault confirmation module 13 determines whether the target device has a fault based on the operating information and reference status information, including: By extracting features from the operational information and reference status information, a refined feature set is obtained. By analyzing the evolution of fine features in the fine feature set over time, we obtain the evolution feature set corresponding to the fine feature set. Based on the detailed feature set, evolutionary feature set, and historical operation information, determine whether the target equipment has a fault.

[0091] Optionally, the fault confirmation module 13 determines whether the target device has a fault based on the fine feature set, the evolution feature set, and historical operating information, including: By using the set of rules to be verified to identify the fine feature set, the evolution feature set, and historical operation information, the first part of the target equipment and the cause of the failure can be obtained. By using a pre-trained fault diagnosis model to identify the fine feature set, evolution feature set, and historical operation information, the second part of the target equipment and the confidence level are obtained. By combining the first part and the cause of the fault with the second part and the confidence level, the fault diagnosis result of the target equipment is obtained.

[0092] Optionally, the fault confirmation module 13 obtains the fault diagnosis result of the target device by combining the first part and the fault cause with the second part and the confidence level, including: If the first part and the second part are the same target part of the target device, and the confidence level is greater than the threshold, then it is determined that there is a fault in the target part; If the first part and the second part are different parts, then obtain the first chain of evidence for the cause of the failure, and the second chain of evidence for the failure of the second part. Based on the strength of the evidence in the first and second chains of evidence, the target part with the malfunction is determined from the first and second parts.

[0093] Optionally, the reference status information includes multiple startup status data closest to the initial alarm, multiple reference limit indicators corresponding to the characteristic indicators, and multiple reference steady-state indicators corresponding to the characteristic indicators; The information optimization module 12 selects reference status information corresponding to the feature indicators in the rule group to be verified from historical operation information in the following ways: Based on the alarm time of the initial alarm, select the most recent start-up status data from the historical operation information; Identify historical indicator sequences of the same type as the feature indicators from historical operational information, and select multiple local maxima from them as multiple reference limit indicators; Identify historical indicator sequences of the same type as the characteristic indicators from historical operational information, and extract sequence segments within the preset indicator fluctuation range from them; Multiple reference steady-state indices are extracted from sequence fragments.

[0094] Optionally, the alarm response module 11 is also used to call a pre-trained pre-detection model to process the running information in order to obtain the pre-identification result; Based on the operational information, the execution condition for selecting reference status information from the historical operational information of the target device is to indicate that the preliminary alarm is not a false alarm.

[0095] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0096] It should also be understood that if the above embodiments are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application.

[0097] Therefore, this embodiment also provides a storage medium, which is a computer-readable storage medium. The storage medium stores a computer program, which, when executed by a processor, implements the fault diagnosis method provided in this embodiment. The storage medium can be any medium capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0098] Please refer to Figure 4 The electronic device provided in this embodiment may include a processor 22 and a memory 21. The memory 21 stores a computer program, and the processor implements the fault diagnosis method provided in this embodiment by reading and executing the computer program in the memory 21 corresponding to the above-described embodiments.

[0099] See also Figure 4 The electronic device also includes a communication unit 23. The memory 21, processor 22 and communication unit 23 are electrically connected to each other directly or indirectly through system bus 24 to realize data transmission or interaction.

[0100] The memory 21 can be an information recording device based on any electronic, magnetic, optical, or other physical principles, used to record execution instructions, data, etc. In some embodiments, the memory 21 can be, but is not limited to, volatile memory, non-volatile memory, memory drive, etc.

[0101] In some embodiments, the volatile memory may be random access memory (RAM); in some embodiments, the non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory, etc.; in some embodiments, the storage drive may be a disk drive, solid-state drive, any type of storage disk (such as optical disc, DVD, etc.), or similar storage media, or a combination thereof.

[0102] The communication unit 23 is used to send and receive data over a network. In some embodiments, the network may include a wired network, a wireless network, a fiber optic network, a telecommunications network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a public switched telephone network (PSTN), a Bluetooth network, a ZigBee network, or a near field communication (NFC) network, or any combination thereof. In some embodiments, the network may include one or more network access points. For example, the network may include wired or wireless network access points, such as base stations and / or network switching nodes, through which one or more components of the service request processing system can connect to the network to exchange data and / or information.

[0103] The processor 22 may be an integrated circuit chip with signal processing capabilities, and may include one or more processing cores (e.g., a single-core processor or a multi-core processor). By way of example only, the processor described above may include a Central Processing Unit (CPU), an Application Specific Integrated Circuit (ASIC), an Application Specific Instruction-set Processor (ASIP), a Graphics Processing Unit (GPU), a Physics Processing Unit (PPU), a Digital Signal Processor (DSP), a Field Programmable Gate Array (FPGA), a Programmable Logic Device (PLD), a controller, a microcontroller unit, a Reduced Instruction Set Computing (RISC) computer, or a microprocessor, or any combination thereof.

[0104] Understandable. Figure 4The structure shown is for illustrative purposes only. Electronic devices may also have more advanced features. Figure 4 Showing more or fewer components, or having with Figure 4 The different configurations shown. Figure 4 The components shown can be implemented using hardware, software, or a combination thereof.

[0105] It should be understood that the apparatus and methods disclosed in the above embodiments can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0106] The above descriptions are merely various embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A fault diagnosis method, characterized in that, The method includes: In response to a preliminary alarm from the target device, the operating information of the target device within a preset time window prior to the occurrence of the preliminary alarm is obtained, wherein the preliminary alarm indicates that at least one device state of the target device exceeds a corresponding state threshold. Based on the operational information, reference status information is selected from the historical operational information of the target device; Based on the operational information and the reference status information, it is determined whether the target device has a fault.

2. The fault diagnosis method according to claim 1, characterized in that, Based on the operational information, reference status information is selected from the historical operational information of the target device, including: Match the set of rules to be verified that are associated with the operational information from the diagnostic rule base based on expert experience; Based on the feature indicators in the rule group to be verified, reference status information corresponding to the feature indicators is selected from the historical operation information.

3. The fault diagnosis method according to claim 2, characterized in that, Based on the operational information and the reference status information, determining whether the target device has a fault includes: The running information and the reference state information are used to extract features to obtain a refined feature set; Analyze the evolution of the fine features in the fine feature set over time to obtain the evolution feature set corresponding to the fine feature set; Based on the refined feature set, the evolutionary feature set, and the historical operation information, it is determined whether the target device has a fault.

4. The fault diagnosis method according to claim 3, characterized in that, Based on the refined feature set, the evolutionary feature set, and the historical operational information, determining whether the target device has a fault includes: The fine feature set, the evolution feature set, and the historical operation information are identified using the set of rules to be verified to obtain the first part of the target device and the cause of the fault. The fine feature set, the evolution feature set, and the historical operation information are identified using a pre-trained fault diagnosis model to obtain the second part of the target device and its confidence level. By combining the first part and the cause of the fault with the second part and the confidence level, the fault diagnosis result of the target device is obtained.

5. The fault diagnosis method according to claim 4, characterized in that, By combining the first location and the cause of the fault with the second location and the confidence level, the fault diagnosis result of the target device is obtained, including: If the first part and the second part are the same target part of the target device, and the confidence level is greater than the threshold, then it is determined that the target part is faulty; If the first part and the second part are different parts, then obtain the first chain of evidence in the cause of the failure, and the second chain of evidence that caused the second part to fail. Based on the strength of the evidence in the first and second chains of evidence, the target part with the fault is determined from the first and second parts.

6. The fault diagnosis method according to claim 2, characterized in that, The reference status information includes multiple startup status data closest to the initial alarm, multiple reference limit indicators corresponding to the feature indicator, and multiple reference steady-state indicators corresponding to the feature indicator; Based on the feature indicators in the rule group to be verified, reference status information corresponding to the feature indicators is selected from the historical operation information, including: Based on the alarm time of the initial alarm, select multiple startup status data sets closest to the initial alarm from the historical operation information; From the historical operational information, determine the historical indicator sequence of the same type as the feature indicator, and select multiple local maxima from them as the multiple reference limit indicators; Determine the historical indicator sequence of the same type as the feature indicator from the historical operation information, and extract the sequence segment within the preset indicator fluctuation range from it; Extract the plurality of reference steady-state indices from the sequence fragments.

7. The fault diagnosis method according to claim 1, characterized in that, The method further includes: The pre-trained pre-detection model is invoked to process the runtime information to obtain the pre-identification result; Based on the operational information, the execution condition for selecting reference status information from the historical operational information of the target device is that the preliminary identification result indicates that the initial alarm is not a false alarm.

8. A fault diagnosis device, characterized in that, The device includes: An alarm response module is used to respond to a preliminary alarm from a target device and obtain the operating information of the target device within a preset time window before the preliminary alarm occurs, wherein the preliminary alarm indicates that at least one device state of the target device exceeds a corresponding state threshold. The information optimization module is used to select reference status information from the historical operating information of the target device based on the operating information; The fault confirmation module is used to determine whether the target device has a fault based on the operating information and the reference status information.

9. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the fault diagnosis method according to any one of claims 1-7.

10. An electronic device, characterized in that, The electronic device includes a processor and a memory, the memory storing a computer program, which, when executed by the processor, implements the fault diagnosis method according to any one of claims 1-7.

Images