A method, system, device, and medium for intelligent diagnostics and precision inspection management of distributed devices based on edge computing.
By processing precision inspection data through edge computing units and comparing it with the local FMEA knowledge base, the problem of inefficient equipment fault diagnosis and maintenance has been solved, enabling rapid and accurate fault location and intelligent maintenance guidance, thereby improving the efficiency of equipment management.
Patent Information
- Application Number
- CN202611149848.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-31
- Publication Date
- 2026-11-17
AI Technical Summary
Existing equipment testing and analysis methods suffer from problems such as poor communication between diagnostic information and maintenance actions, insufficient edge-side analysis capabilities, and delayed response, which affect the timeliness and accuracy of equipment fault diagnosis.
The edge computing unit processes the precision inspection data, extracts the feature pattern sequence, compares it with the local FMEA knowledge base, determines the dominant fault type, and outputs the precision inspection analysis results, including diagnostic conclusions and maintenance recommendations.
It enables rapid location and diagnosis of equipment faults and direct intelligent guidance, improving the efficiency of equipment management and the effectiveness of predictive maintenance strategies.
Smart Images

Figure CN123116048A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power equipment technology, specifically to a method, system, device, and medium for intelligent diagnosis and precision inspection management of distributed equipment based on edge computing. Background Technology
[0002] Currently, the operational status testing and health assessment of industrial equipment typically rely on the collection and analysis of various parameter data (such as vibration and temperature) generated during equipment operation. Through these analyses, the aim is to determine whether the equipment is experiencing wear, whether early signs of failure are appearing, or to predict potential failure risks, thereby providing a basis for condition-based maintenance.
[0003] However, existing equipment testing and analysis methods still face some challenges in practical applications. For example, many analysis processes rely on transmitting large amounts of raw data over long distances to a central platform for processing, which may lead to delays in analysis results and affect the timeliness of fault diagnosis. Meanwhile, the intelligent analysis capabilities at the equipment edge are often insufficient, making it difficult to quickly extract precise fault characteristics from complex data. Furthermore, the diagnostic information obtained is often disconnected from specific maintenance measures, failing to directly and effectively guide on-site repair and handling. This limits the full potential of testing technology in ensuring the continuous and reliable operation of equipment. Summary of the Invention
[0004] In view of the above-mentioned problems, the present invention is proposed.
[0005] Therefore, the technical problem solved by this invention is: how to perform in-depth intelligent analysis of precision inspection data by the edge computing unit of distributed industrial equipment, and link it with a localized fault knowledge base that integrates failure mode effect analysis logic, so as to achieve efficient and accurate testing of equipment operating status, rapid fault location and diagnosis, and direct intelligent guidance for maintenance operations, while overcoming the problems of poor connection between diagnostic information and maintenance actions, insufficient edge-side analysis capabilities, and delayed response in existing testing and analysis methods.
[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a method for intelligent diagnosis and precision inspection management of distributed devices based on edge computing, comprising: an edge computing unit processing precision inspection data acquired by self-monitoring distributed industrial equipment, and extracting a feature pattern sequence to characterize abnormal operation of the distributed industrial equipment; the precision inspection data being a set of equipment operating parameters that change over time; determining the current dominant fault type by comparing the feature pattern sequence with a local FMEA knowledge base; and outputting precision inspection analysis results based on the current dominant fault type and related handling information obtained from the local FMEA knowledge base; wherein, the processing of the precision inspection data acquired by the self-monitoring distributed industrial equipment by the edge computing unit includes the edge computing unit dividing the precision inspection data into multiple data segments, and performing a comprehensive stability determination on any current data segment among the multiple data segments, the comprehensive stability determination being based on a first pre-calculated stability index and a second pre-calculated stability index of the current data segment.
[0007] As a preferred embodiment of the edge computing-based distributed device intelligent diagnosis and precision inspection management method of the present invention, wherein: if both the first pre-calculated stability index and the second pre-calculated stability index meet predetermined conditions, the current data segment is determined to be a stable operating condition data segment; otherwise, the current data segment is marked as exceptional data. The first pre-calculated stability index is the result of comparing the absolute difference between the root mean square energy value of the current data segment and the root mean square energy value of a historical reference data segment selected from the precision inspection data according to predetermined rules and located prior to the current data segment in time, and a first reference fluctuation calculated based on the recent historical segment statistical characteristics of the precision inspection data. The second pre-calculated stability index is the result of comparing the absolute difference between the amplitude of the highest frequency component selected in the current data segment and the amplitude of the corresponding highest frequency component selected in the historical reference data segment, and a second reference fluctuation calculated based on the recent historical segment statistical characteristics of the highest frequency component.
[0008] As a preferred embodiment of the edge computing-based distributed device intelligent diagnosis and precision inspection management method of the present invention, the step of extracting the feature pattern sequence to characterize the abnormal operation of the distributed industrial equipment includes: extracting the time-domain feature parameter set and frequency-domain feature parameter set of the stable operating condition data segment in parallel for the stable operating condition data segment to form an initial candidate feature group; filtering and sorting the initial candidate feature group according to the current operating mode parameter of the distributed industrial equipment and the fault indication capability quantization value of each initial candidate feature to form the feature pattern sequence; wherein, the filtering includes: when the current operating mode parameter is a high-priority protection mode and the fault indication capability quantization value of a certain initial candidate feature is lower than the filtering reference value, then that initial candidate feature is not selected to form the feature pattern sequence; the fault indication capability quantization value is calculated based on the statistical distribution of each initial candidate feature, and the filtering reference value is calculated based on the distribution of the fault indication capability quantization values of all members of the initial candidate feature group.
[0009] As a preferred embodiment of the distributed device intelligent diagnosis and precision inspection management method based on edge computing described in this invention, the local FMEA knowledge base includes: a structured and associated storage of various predefined typical fault modes, failure cause chains corresponding to each typical fault mode, multi-level failure impact quantification levels, and recommended response measures codes.
[0010] As a preferred embodiment of the distributed device intelligent diagnosis and precision inspection management method based on edge computing described in this invention, the step of determining the current dominant fault type by comparing the feature pattern sequence with the local FMEA knowledge base includes: calculating the feature matching degree value between the feature pattern sequence and each typical fault mode in the local FMEA knowledge base; determining whether multiple typical fault modes simultaneously satisfy a first matching confirmation condition and a second matching differentiation condition; the first matching confirmation condition is that the corresponding feature matching degree value exceeds a confirmation reference value generated based on historical matching data; the second matching differentiation condition is that the numerical difference between the feature matching degree values corresponding to multiple typical fault modes that satisfy the first matching confirmation condition is less than the differentiation reference value generated based on historical matching data; if multiple typical fault modes simultaneously satisfy the first matching confirmation condition and the second matching differentiation condition, the edge computing unit determines the multiple typical fault modes that satisfy the conditions as a composite dominant fault type; otherwise, the edge computing unit determines the typical fault mode with the highest feature matching degree value as a single current dominant fault type.
[0011] As a preferred embodiment of the distributed device intelligent diagnosis and precision inspection management method based on edge computing described in this invention, the precision inspection analysis results include a confidence assessment of the diagnostic conclusion. Specifically, the edge computing unit calculates and outputs a confidence level value representing the reliability of the diagnostic conclusion based on the degree of match between the current dominant fault type and the feature pattern sequence, combined with the failure impact level quantification parameters associated with the current dominant fault type obtained from the local FMEA knowledge base.
[0012] As a preferred embodiment of the distributed device intelligent diagnosis and precision inspection management method based on edge computing described in this invention, the precision inspection analysis results also include the execution priority of maintenance recommendations. Specifically, the edge computing unit obtains the recommended maintenance procedure index from the local FMEA knowledge base based on the current dominant fault type, and combines the confidence value of the diagnostic conclusion and the failure impact level quantification parameter to determine and output the priority processing level of the maintenance recommendations through a preset decision matrix.
[0013] Another objective of this invention is to provide a distributed device intelligent diagnostic and precision inspection management system based on edge computing.
[0014] To address the aforementioned technical problems, this invention provides the following technical solution: a distributed equipment intelligent diagnosis and precision inspection management system based on edge computing, comprising: a precision inspection data processing module, used to process precision inspection data acquired by self-monitoring distributed industrial equipment, and extract feature pattern sequences to characterize abnormal operation of the distributed industrial equipment, wherein the precision inspection data is a set of equipment operating parameters that change over time; a fault type determination module, used to determine the current dominant fault type by comparing the feature pattern sequences with a local FMEA knowledge base; and an analysis result output module, used to output precision inspection analysis results based on the current dominant fault type and related handling information obtained from the local FMEA knowledge base.
[0015] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the aforementioned distributed device intelligent diagnosis and precision inspection management method based on edge computing.
[0016] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the aforementioned method for intelligent diagnosis and precision inspection management of distributed devices based on edge computing.
[0017] The beneficial effects of this invention are as follows: By processing precise inspection data in an edge computing unit, extracting characteristic pattern sequences representing equipment anomalies, and comparing these sequences with a fault knowledge base that integrates Failure Mode and Effects Analysis (FMEA) logic locally to determine the dominant fault type, this invention outputs precise inspection analysis results containing diagnostic conclusions and maintenance operation guidance based on the fault type and FMEA logic. This achieves more direct and intelligent online testing and evaluation of the status of distributed industrial equipment. This invention improves the accuracy and response speed of fault diagnosis at the equipment edge and can directly transform complex analysis results into clear and actionable maintenance operation guidance, thereby enhancing the effectiveness of equipment testing and the implementation effect of predictive maintenance strategies, and improving the overall efficiency of equipment management. Attached Figure Description
[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is an overall flowchart of a distributed device intelligent diagnosis and precision inspection management method based on edge computing, provided as an embodiment of the present invention.
[0020] Figure 2 This is a computer device diagram illustrating a distributed device intelligent diagnosis and precision inspection management method based on edge computing, provided as an embodiment of the present invention. Detailed Implementation
[0021] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.
[0022] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a distributed device intelligent diagnosis and precision inspection management method based on edge computing, including: S1: The edge computing unit processes the precision inspection data acquired by the self-monitoring distributed industrial equipment and extracts feature pattern sequences to characterize abnormal operation of the distributed industrial equipment. The precision inspection data is a set of equipment operating parameters that change over time. Specifically, the edge computing unit processes the precision inspection data acquired by the self-monitoring distributed industrial equipment by dividing the precision inspection data into multiple data segments and performing a comprehensive stability assessment on any current data segment. The comprehensive stability assessment is based on a first pre-calculated stability index and a second pre-calculated stability index for the current data segment. S2: Determine the current dominant fault type by comparing the characteristic pattern sequence with the local FMEA knowledge base; S3: Based on the current dominant fault type and the related handling information obtained from the local FMEA knowledge base, output the detailed inspection analysis results.
[0023] It should be noted that in industrial environments such as thermal power plants, the stable operation of a large number of distributed critical auxiliary equipment (such as fans, pumps, and coal mills) is crucial to the safety and efficiency of the entire production system. Traditional equipment condition monitoring and fault diagnosis methods mostly rely on uploading operational data collected on-site to a central server for centralized analysis. This model often faces problems such as high data transmission bandwidth pressure, delayed feedback of analysis results, and difficulty in adapting to individual equipment differences and rapid changes in operating conditions. In addition, even when fault diagnosis information is obtained, it often lacks a direct and intelligent connection with specific maintenance and repair measures, making maintenance decisions still require a lot of human experience, which affects the overall efficiency of inspection work and the level of intelligence in equipment management.
[0024] Therefore, addressing the aforementioned issues of insufficient edge-end analysis capabilities and weak linkage between diagnosis and maintenance in distributed industrial equipment precision inspection analysis, this embodiment utilizes steps S1 to S3 to localize the precision inspection data acquired by the self-monitoring equipment within the edge computing unit. First, in step S1, the edge computing unit intelligently segments and determines the stability of the precision inspection data, extracting feature pattern sequences that effectively characterize equipment malfunctions, providing high-quality input for subsequent precise analysis. Then, in step S2, the edge computing unit compares the feature pattern sequences with a locally stored knowledge base integrating Failure Mode and Effects Analysis (FMEA) logic, achieving rapid and accurate identification of the current dominant fault type. Finally, in step S3, based on the identified dominant fault type and related handling information retrieved from the local FMEA knowledge base, the precision inspection analysis results, containing clear diagnostic conclusions and specific maintenance recommendations, are directly output at the edge. Thus, this method improves the real-time performance and accuracy of fault diagnosis by performing a complete precision inspection and intelligent analysis process at the edge of the equipment, and strengthens the direct guiding role of diagnostic results in actual maintenance work. This helps to optimize maintenance strategies, reduce operation and maintenance costs, and improve the operational reliability of distributed industrial equipment.
[0025] Example 2, refer to Figure 1 As one embodiment of the present invention, based on the previous embodiment, a method for intelligent diagnosis and precision inspection management of distributed devices based on edge computing is provided, including: S1: The edge computing unit processes the precision inspection data acquired by the self-monitoring distributed industrial equipment and extracts the feature pattern sequence used to characterize the abnormal operation of the distributed industrial equipment.
[0026] Specifically, precision inspection data consists of a set of equipment operating parameters that change over time.
[0027] The precision inspection data includes various monitoring parameters generated during equipment operation, such as vibration signals, temperature readings, current changes, speed fluctuations, and pressure parameters. These parameters are continuously collected according to a preset sampling frequency, forming time-series data reflecting the equipment's operating status. For example, in the monitoring of induced draft fans in a thermal power plant, the precision inspection data includes multi-dimensional parameters such as bearing housing vibration acceleration values, motor winding temperatures, stator current RMS values, and impeller speeds collected every second. This data comprehensively reflects the real-time operating status of the induced draft fan.
[0028] Specifically, the edge computing unit processes the precision inspection data acquired by the self-monitoring distributed industrial equipment. This includes the edge computing unit dividing the precision inspection data into multiple data segments and performing a comprehensive stability determination on any current data segment. The comprehensive stability determination is based on a first pre-calculated stability index and a second pre-calculated stability index of the current data segment.
[0029] It should be noted that data segmentation is performed to convert continuous data streams into discrete segments that are easier to analyze. The edge computing unit divides the precision inspection data into multiple non-overlapping data segments according to a preset time window length. Each segment contains all sampled data within a fixed duration. Taking a steel mill rolling mill as an example, the edge computing unit segments the continuously collected vibration data into segments of 60 seconds each. For a vibration sensor with a sampling frequency of 2000Hz, each data segment contains 120,000 vibration amplitude data points.
[0030] Furthermore, if both the first pre-calculated stability index and the second pre-calculated stability index meet the predetermined conditions, the current data segment is determined to be a stable operating condition data segment; otherwise, the current data segment is marked as exceptional data. The purpose of comprehensive stability assessment is to identify data segments where the equipment is in a stable operating state and filter out interference data generated by non-steady-state processes such as equipment start-up and shutdown, load switching, and operating condition adjustments. The predetermined conditions are determined based on the equipment's historical operating characteristics and on-site engineering experience. For example, for a circulating water pump, when the first pre-calculated stability index value is less than 0.2 and the second pre-calculated stability index value is less than 0.15, the system determines that the data segment is a stable operating condition data segment and can be used for subsequent feature extraction and analysis.
[0031] The first pre-calculated stability index is the result of comparing the absolute difference between the root mean square energy value of the current data segment and the root mean square energy value of a historical reference data segment selected from the precision inspection data according to a predetermined rule and located before the current data segment in time, and the first reference fluctuation calculated based on the statistical characteristics of recent historical segments of the precision inspection data. The first pre-calculated stability index assesses stability by comparing the energy level difference between the current data segment and a historical baseline. Specifically, the system first calculates the root mean square (RMS) energy value of all sampling points within the current data segment, and then selects a historical reference data segment to calculate its corresponding RMS energy value. The selection of historical reference data segments follows predetermined rules, typically choosing several recently confirmed stable operating condition data segments. The first reference fluctuation is calculated by analyzing the variation amplitude of the RMS energy values of recent historical segments, reflecting the energy fluctuation range of the equipment during normal stable operation. For example, for a coal mill, the system selects 10 stable operating condition data segments from the past 24 hours as historical reference baselines and calculates the standard deviation of their RMS energy values as the first reference fluctuation.
[0032] The second pre-calculated stability index is the result of comparing the absolute difference between the amplitude of the selected primary frequency component in the current data segment and the amplitude of the corresponding selected primary frequency component in the historical reference data segment, and the second reference fluctuation calculated based on the statistical characteristics of the primary frequency component in recent historical segments.
[0033] The second pre-calculated stability index determines the operating status by analyzing the stability of the equipment's characteristic frequency components. The selected primary frequency components are typically the equipment's inherent characteristic frequencies, such as motor rotation frequency, gear meshing frequency, and blade passage frequency. The system performs frequency domain analysis on the current data segment, extracts the amplitude of the selected primary frequency components, and compares it with the amplitude of the corresponding frequency components in historical reference data segments. The second reference fluctuation is calculated by statistically analyzing the range of amplitude changes of the primary frequency components in recent historical segments. For example, for a centrifugal fan with a rotational speed of 1480 rpm, the system selects 24.7 Hz as the primary frequency component and calculates the second pre-calculated stability index by comparing the amplitude difference between the current data segment and the historical benchmark at that frequency.
[0034] Extract feature pattern sequences to characterize operational anomalies in distributed industrial equipment, including: For stable operating condition data segments, time-domain feature parameter sets and frequency-domain feature parameter sets are extracted in parallel to form initial candidate feature groups; The feature extraction process employs parallel computing to simultaneously perform time-domain and frequency-domain analysis to improve processing efficiency. The time-domain feature parameter set includes various parameters reflecting the signal's statistical characteristics, such as the signal's mean, standard deviation, skewness, kurtosis, peak factor, impulse factor, and margin factor. The frequency-domain feature parameter set is extracted after performing a Fast Fourier Transform on the data segment, including parameters such as power spectral density distribution characteristics, energy distribution in each frequency band, spectral peak frequencies and their amplitudes, spectral centroid, and spectral variance. For example, for vibration monitoring of a water pump, the system extracts 18 time-domain feature parameters and 25 frequency-domain feature parameters in parallel from a 60-second stable operating condition data segment, forming an initial candidate feature set containing 43 features.
[0035] Based on the current working mode parameters of the distributed industrial equipment and the fault indication capability quantification value of each initial candidate feature, the initial candidate feature group is screened and sorted to form a feature pattern sequence; The feature selection process comprehensively considers the equipment's operating mode and the fault identification capability of the features. The current operating mode parameter reflects the equipment's operating state level, such as normal operation mode, full-load operation mode, and high-priority protection mode. The fault indication capability quantification value measures the sensitivity and distinguishing ability of each feature parameter to abnormal equipment states, calculated by analyzing the statistical differences between historical fault cases and normal operation data. The selection process ranks features by importance based on the fault indication capability quantification value, prioritizing feature parameters with high fault identification capability. For example, for a steam turbine, under high-load operation mode, the system detects that the kurtosis value and first octave amplitude of the bearing housing vibration have high fault indication capability quantification values; therefore, these features are prioritized to form the feature pattern sequence.
[0036] The screening process includes: when the current working mode parameter is a high-priority protection mode and the fault indication capability quantification value of a certain initial candidate feature is lower than the screening reference value, then a certain initial candidate feature will not be selected to form a feature pattern sequence. In high-priority protection mode, the system applies stricter criteria to feature selection to ensure the accuracy and reliability of fault diagnosis. When equipment is in high-priority protection mode, it usually means that the safe operation of the equipment is of extremely high importance, and misjudgment or omission is not permissible. At this time, the system only selects feature parameters whose fault indication capability quantification value exceeds the screening reference value, filtering out features with weak fault identification capabilities. For example, in the monitoring of the main feedwater pump of a nuclear power plant, when the system detects that the equipment is in high-priority protection mode, it only selects feature parameters whose fault indication capability quantification value is greater than 0.75, excluding features such as the average vibration signal and temperature change rate below this threshold from the feature pattern sequence.
[0037] The fault indication capability quantification value is calculated based on the statistical distribution of each initial candidate feature, and the screening reference value is calculated based on the distribution of the fault indication capability quantification values of all members of the initial candidate feature group.
[0038] The calculation of fault indication capability quantification is based on statistical analysis of a large amount of historical data. It quantifies fault identification capability by comparing the distribution differences of characteristic parameters under normal and fault conditions. The calculation process involves analyzing the statistical characteristics of each feature under different operating conditions, such as mean, variance, and distribution pattern. The screening reference value is determined by analyzing the distribution of fault indication capability quantification values of all features in the initial candidate feature group, typically set to a certain quantile value, such as the 75th or 90th quantile. This dynamic setting method can adaptively adjust the screening criteria according to the characteristics of different equipment and operating conditions, ensuring that the selected features are both representative and have good fault identification performance.
[0039] S2: Determine the current dominant fault type by comparing the feature pattern sequence with the local FMEA knowledge base.
[0040] Specifically, the local FMEA knowledge base includes: a structured and associated storage of various predefined typical failure modes, corresponding failure cause chains, multi-level failure impact quantification levels, and recommended response code.
[0041] The local FMEA knowledge base is a professional fault diagnosis knowledge system built upon Failure Mode and Effects Analysis (FMEA) theory, employing a structured storage method to organize various fault information. Typical fault modes include various common abnormal states of equipment, such as bearing wear, imbalance, misalignment, looseness, and resonance. Each typical fault mode is associated with and stored in detail the failure cause chain information, describing the complete development path from the initial abnormality to final failure. Multi-level failure impact quantification classifies and quantifies the degree of impact of the fault on the equipment and system, typically categorized into minor impact, general impact, severe impact, and catastrophic impact. Recommended response codes contain specific handling plans and maintenance suggestions for each fault mode.
[0042] For example, in the FMEA knowledge base of a centrifugal pump in a chemical plant, the stored information for the bearing wear failure mode includes: the failure cause chain is "poor lubrication → increased bearing temperature → increased bearing clearance → increased vibration → bearing damage", the failure impact level is "severe impact" (quantification value of 8), and the recommended countermeasures are coded as "M001 - check the lubrication system, M002 - replace the bearing, M003 - correct the shaft alignment". Similarly, the information for the imbalance failure mode includes: the failure cause chain is "impeller fouling → uneven mass distribution → centrifugal force imbalance → increased vibration", the failure impact level is "moderate impact" (quantification value of 5), and the recommended countermeasures are coded as "M004 - clean the impeller, M005 - dynamic balancing".
[0043] By comparing the characteristic pattern sequence with the local FMEA knowledge base, the current dominant fault type is determined, including: Calculate the feature matching degree value between the feature pattern sequence and each typical fault mode in the local FMEA knowledge base; The calculation of the feature matching degree value involves comparing the similarity of the current device's feature pattern sequence with the standard feature patterns of each typical fault mode in the FMEA knowledge base. The calculation method employs similarity measurement techniques in a multi-dimensional feature space, comprehensively considering the proximity of feature values, the consistency of feature change trends, and the matching degree of feature combination patterns. Specifically, the system compares each feature parameter in the feature pattern sequence with the standard value of the corresponding feature in the typical fault modes, calculates the numerical deviation, trend correlation, and pattern fit, and then obtains the overall matching degree value through weighted summation.
[0044] Taking the vibration diagnosis of a wind turbine as an example, the current characteristic mode sequence includes characteristic parameters such as a first harmonic amplitude of 8.2 mm / s, a second harmonic amplitude of 3.1 mm / s, a kurtosis of 4.8, and a peak factor of 3.2. The system compares these characteristics with the standard characteristics of the "imbalance" fault mode in the FMEA knowledge base. It finds that the first harmonic amplitude is close to the standard value of 7.5 mm / s, the second harmonic amplitude is close to the standard value of 2.8 mm / s, and the other characteristic parameters are also within a reasonable range. The comprehensive calculation yields a characteristic matching degree of 0.87 with the "imbalance" fault mode. Similarly, the matching degree with the "bearing wear" fault mode is 0.23, and the matching degree with the "resonance" fault mode is 0.41.
[0045] Determine whether there are multiple typical fault modes that simultaneously meet the first matching confirmation condition and the second matching differentiation condition; the first matching confirmation condition is that the corresponding feature matching degree value exceeds the confirmation reference value generated based on historical matching data; the second matching differentiation condition is that the numerical difference between the feature matching degree values corresponding to multiple typical fault modes that meet the first matching confirmation condition is less than the differentiation reference value generated based on historical matching data. The first matching confirmation criterion is used to filter out typical failure modes that highly match the current device state. The confirmation reference value is determined by analyzing the statistical distribution of a large amount of historical matching data, and is usually set as a certain quantile of the feature matching degree value in historical successful matching cases, such as the 85th quantile. This ensures that only failure modes with a truly high degree of matching can pass the initial screening.
[0046] The second matching distinction condition is used to identify situations where multiple fault modes have similarly high matching degrees, making them difficult to distinguish. The distinction reference value is also determined based on statistical analysis of historical matching data, reflecting the minimum matching degree difference threshold that can reliably distinguish different fault modes in actual diagnosis. When the difference in matching degree values of multiple fault modes is less than the distinction reference value, it indicates that these fault modes may coexist or are difficult to distinguish accurately.
[0047] For example, in a diagnostic case of a water pump, the system calculates that the feature matching degree value for the "bearing wear" fault mode is 0.82, the matching degree value for the "misalignment" fault mode is 0.78, and the matching degree value for the "loosening" fault mode is 0.35. Assuming the confirmation reference value is 0.75 and the differentiation reference value is 0.08, both the "bearing wear" and "misalignment" fault modes satisfy the first matching confirmation condition (0.82>0.75, 0.78>0.75), and the difference in their matching degree values is 0.04, which is less than the differentiation reference value of 0.08. Therefore, they also satisfy the second matching differentiation condition.
[0048] If multiple typical fault modes simultaneously satisfy the first matching confirmation condition and the second matching differentiation condition, the edge computing unit will jointly determine the multiple typical fault modes that meet the conditions as the composite dominant fault type; otherwise, the edge computing unit will determine the typical fault mode with the highest feature matching degree value as the single current dominant fault type.
[0049] The identification of composite dominant fault types reflects the complexity of industrial equipment failures, meaning that equipment may simultaneously exhibit multiple fault modes or be in a state of overlapping fault modes. When the system detects multiple fault modes with similar high matching degrees, it indicates that these fault modes may all contribute to the current equipment state, requiring comprehensive consideration for diagnosis and handling. Identifying composite dominant fault types provides more comprehensive information for subsequent maintenance decisions, avoiding the oversight of other potential problems due to a single fault mode assessment.
[0050] The determination of a single dominant fault type applies when there is a clearly dominant fault mode. When the feature matching value of a fault mode is significantly higher than that of other fault modes, and the distinguishing criteria are met, the system identifies it as the single dominant fault type. This situation usually indicates that the equipment has a relatively clear and prominent specific fault problem.
[0051] Continuing with the above case study of the feedwater pump, since both "bearing wear" and "misalignment" failure modes simultaneously meet the matching confirmation and differentiation criteria, the system identifies it as a composite dominant failure type: "bearing wear + misalignment." This diagnostic result suggests that on-site maintenance personnel need to pay attention to both the bearing condition and the shaft alignment, and develop a comprehensive maintenance plan.
[0052] In another case, the diagnostic results of a compressor showed that the characteristic matching degree value of the "resonance" fault mode was 0.91, while the matching degree values of other fault modes were all below 0.60. The difference was significantly greater than the distinguishing reference value. Therefore, the system identified "resonance" as the single dominant fault type and guided maintenance personnel to focus on the equipment's natural frequency and excitation source.
[0053] S3: Based on the current dominant fault type and the related handling information obtained from the local FMEA knowledge base, output the detailed inspection analysis results.
[0054] The results of the precision inspection analysis include a confidence assessment of the diagnostic conclusions. Specifically, the edge computing unit compares the degree of agreement between the current dominant fault type and the characteristic pattern sequence, and combines the failure impact level quantification parameters associated with the current dominant fault type obtained from the local FMEA knowledge base to calculate and output a confidence value representing the reliability of the diagnostic conclusions.
[0055] The confidence score calculation comprehensively considers both the accuracy and reliability of fault diagnosis. The degree of agreement reflects the closeness of the match between the current equipment's actual characteristic patterns and typical fault modes; a higher value indicates a more accurate diagnosis. The failure impact level quantification parameter reflects the severity of the fault type and the importance of the diagnosis; faults with higher impact levels require a higher diagnostic confidence score. The edge computing unit calculates the comprehensive confidence score by weightedly fusing these two factors and considering the statistical patterns of historical diagnostic cases.
[0056] The specific calculation process includes three main steps: First, the system extracts the final feature matching degree value between the current dominant fault type and the feature pattern sequence as the basic consistency index; second, it obtains the failure impact level quantification parameter of the fault type from the FMEA knowledge base, and corrects the consistency according to the level of impact; finally, it combines the confidence distribution characteristics of historical similar fault diagnoses and obtains the final confidence value through statistical calibration.
[0057] For example, in the diagnosis of the main bearing of a steel mill rolling mill, the system determined that the current dominant fault type was "bearing inner ring wear," and the feature matching degree between this fault mode and the current characteristic pattern sequence was 0.86. The quantification parameter of the failure impact level of "bearing inner ring wear" obtained from the FMEA knowledge base was 9 (out of 10, belonging to the severe impact level). The system adjusted the basic fit based on the high impact level and combined it with the confidence statistical distribution of historical bearing fault diagnosis cases, finally calculating a confidence value of 0.91 for the diagnostic conclusion, indicating that the diagnostic result has high reliability.
[0058] In another case, the diagnosis for a circulating water pump was "slight impeller imbalance," with a feature matching degree of 0.72 and a failure impact level quantification parameter of 4 (belonging to the general impact level). Due to the low impact level and moderate matching degree, the system calculated a confidence level of 0.68, indicating that the diagnosis has moderate reliability. It is recommended to combine this diagnosis with other information for a comprehensive assessment.
[0059] The precision inspection analysis results also include the execution priority of maintenance recommendations. Specifically, the edge computing unit obtains the recommended maintenance procedure index from the local FMEA knowledge base based on the current dominant fault type, and combines the confidence value of the diagnostic conclusion and the quantitative parameters of the failure impact level to determine and output the priority level of the maintenance recommendations through a preset decision matrix.
[0060] Determining the priority of maintenance recommendations is a multi-factor comprehensive decision-making process aimed at providing on-site maintenance personnel with scientific operational guidance and resource allocation suggestions. The recommended maintenance procedure index contains standardized handling procedures and technical requirements for each fault type, such as detailed information on inspection items, repair methods, required tools, and estimated working hours. The edge computing unit first retrieves the corresponding recommended maintenance procedure index from the FMEA knowledge base based on the currently dominant fault type. Then, it uses the confidence level of the diagnostic conclusion and the quantification parameters of the failure impact level as input variables, and performs a comprehensive evaluation through a pre-set decision matrix.
[0061] The pre-defined decision matrix is a two-dimensional or multi-dimensional decision table constructed based on equipment management experience and risk assessment theory. One dimension represents the failure impact level, and the other represents the confidence level. The intersection area corresponds to different priority levels. Priority levels are typically divided into multiple levels such as emergency handling, priority handling, planned handling, and observation and monitoring. Each level corresponds to different response time requirements and resource allocation standards.
[0062] For example, in the maintenance decision-making process of a feedwater pump in a thermal power plant, the system diagnosed a "bearing wear" fault with a confidence level of 0.88 and a failure impact level quantification parameter of 8. The recommended maintenance procedure index obtained from the FMEA knowledge base includes: "R001-Bearing temperature and vibration inspection, R002-Lubricating oil quality analysis, R003-Bearing clearance measurement, R004-Bearing replacement operation." Through decision matrix lookup, the combination of high confidence (above 0.85) and high impact level (above 7) corresponds to the "emergency handling" level. The system outputs a maintenance recommendation prioritizing "emergency handling," requiring on-site inspection by professional technicians within 8 hours and the development of a detailed repair plan within 24 hours.
[0063] In another case, a wind turbine was diagnosed as having a "slight imbalance" with a confidence level of 0.65 and a failure impact level quantification parameter of 3. The corresponding recommended maintenance procedures included: "R005-Impeller Cleaning Inspection, R006-Dynamic Balance Test, R007-Fastener Inspection." A decision matrix query showed that the combination of medium confidence (0.6-0.8) and low impact level (1-4) corresponded to the "Planned Handling" level. The system output suggested that the maintenance priority level be "Planned Handling," recommending that it be addressed during the next planned shutdown maintenance, without requiring immediate shutdown.
[0064] For cases involving multiple dominant fault types, the edge computing unit employs a comprehensive evaluation strategy to determine the execution priority of maintenance recommendations. The system obtains the recommended maintenance procedure index and impact level parameters for each fault mode, and then uses the highest risk principle to merge priorities, that is, determining the overall priority level based on the fault mode with the highest risk level. At the same time, it integrates all relevant maintenance procedure indexes to form a comprehensive maintenance recommendation.
[0065] For example, a compressor is diagnosed with a combined fault type of "bearing wear + misalignment," where the impact level of "bearing wear" is 8 and the impact level of "misalignment" is 6, with an overall confidence level of 0.83. The system adopts the highest risk principle, using the impact level of "bearing wear" (8) as the dominant factor, combined with the high confidence level, and ultimately determines the maintenance recommendation priority level as "emergency handling." Simultaneously, the system integrates the maintenance procedure indexes for both fault types, outputting comprehensive maintenance recommendations including "bearing inspection and replacement + shaft alignment correction," providing on-site technicians with complete guidance on handling the situation.
[0066] The system's precise inspection and analysis results also include detailed text descriptions and numerical information to facilitate quick understanding and execution by on-site personnel. Typical output formats include: fault type description, confidence level, impact level, priority level, recommended maintenance items, estimated processing time, and required technical resources. This structured output method ensures effective integration of diagnostic information and maintenance actions, improving the intelligence level and response efficiency of equipment management.
[0067] Example 3 is an embodiment of the present invention, which provides a distributed device intelligent diagnostic and precision inspection management system based on edge computing, including: The precision inspection data processing module is used to process the precision inspection data acquired by the self-monitoring distributed industrial equipment and extract the feature pattern sequence used to characterize the abnormal operation of the distributed industrial equipment. The precision inspection data is a set of equipment operating parameters that change over time. The fault type determination module is used to determine the current dominant fault type by comparing the characteristic pattern sequence with the local FMEA knowledge base; The analysis results output module is used to output detailed inspection analysis results based on the current dominant fault type and related handling information obtained from the local FMEA knowledge base.
[0068] Example 4, refer to Figure 2 This is the fourth embodiment of the present invention. Based on the first three embodiments, an electronic device is also provided, which is applicable to a distributed device intelligent diagnosis and precision inspection management method based on edge computing. The device includes: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the distributed device intelligent diagnosis and precision inspection management method based on edge computing proposed in the above embodiments.
[0069] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements a distributed device intelligent diagnosis and precision inspection management method based on edge computing as proposed in the above embodiment.
[0070] The storage medium proposed in this embodiment and the method for implementing a distributed device intelligent diagnosis and precision inspection management based on edge computing proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.
[0071] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.
[0072] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. An edge-computing-based distributed equipment intelligent diagnosis and precision point inspection management method, characterized in that: include, edge The computing unit processes the precision inspection data acquired by the self-monitoring distributed industrial equipment and extracts feature pattern sequences to characterize the abnormal operation of the distributed industrial equipment. The precision inspection data is a set of equipment operating parameters that change over time. The dominant fault type is determined by comparing the characteristic pattern sequence with the local FMEA knowledge base; Based on the current dominant fault type and the associated handling information obtained from the local FMEA knowledge base, output the detailed inspection analysis results; The edge computing unit processes the precision inspection data acquired by the self-monitoring distributed industrial equipment, which includes dividing the precision inspection data into multiple data segments and performing a comprehensive stability determination on any current data segment among the multiple data segments. The comprehensive stability determination is based on a first pre-calculated stability index and a second pre-calculated stability index of the current data segment. 2.The edge computing based distributed equipment intelligent diagnosis and precision inspection management method according to claim 1, characterized in that: If both the first pre-calculated stability index and the second pre-calculated stability index meet the predetermined conditions, then the current data segment is determined to be a stable operating condition data segment; otherwise, the current data segment is marked as exceptional data. The first pre-calculated stability index is a comparison between the absolute value of the energy root mean square value of the current data segment and the energy root mean square value of a historical reference data segment selected from the precision inspection data according to a predetermined rule and located before the current data segment in time, and the first reference fluctuation calculated based on the statistical characteristics of recent historical segments of the precision inspection data. The second pre-calculated stability index is the result of comparing the absolute difference between the amplitude of the highest frequency component selected in the current data segment and the amplitude of the corresponding highest frequency component selected in the historical reference data segment, and the second reference fluctuation calculated based on the statistical characteristics of the highest frequency component in recent historical segments. 3.The edge computing based distributed equipment intelligent diagnosis and precision inspection management method of claim 2, wherein: The extraction of the feature pattern sequence used to characterize the abnormal operation of the distributed industrial equipment includes: For the stable operating condition data segment, the time domain feature parameter set and frequency domain feature parameter set of the stable operating condition data segment are extracted in parallel to form an initial candidate feature group; The initial candidate feature group is screened and sorted according to the current working mode parameters of the distributed industrial equipment and the fault indication capability quantification value of each initial candidate feature to form the feature pattern sequence; The screening includes: when the current working mode parameter is a high-priority protection mode and the fault indication capability quantification value of a certain initial candidate feature is lower than the screening reference value, then the certain initial candidate feature is not selected to form the feature pattern sequence. The fault indication capability quantification value is calculated based on the statistical distribution of each of the initial candidate features, and the screening reference value is calculated based on the distribution of the fault indication capability quantification values of all members of the initial candidate feature group.
4. The edge-computing-based distributed equipment intelligent diagnosis and precision inspection management method according to claim 3, characterized in that: The local FMEA knowledge base includes: The system stores a structured association between various predefined typical failure modes and their corresponding failure cause chains, multi-level failure impact quantification levels, and recommended response codes.
5. The edge computing-based distributed equipment intelligent diagnosis and precision inspection management method according to claim 4, characterized in that: The step of determining the current dominant fault type by comparing the feature pattern sequence with the local FMEA knowledge base includes: Calculate the feature matching degree value between the feature pattern sequence and each typical fault mode in the local FMEA knowledge base; Determine whether there are multiple typical fault modes that simultaneously satisfy the first matching confirmation condition and the second matching differentiation condition; the first matching confirmation condition is that the corresponding feature matching degree value exceeds the confirmation reference value generated based on historical matching data; the second matching differentiation condition is that the numerical difference between the feature matching degree values corresponding to multiple typical fault modes that satisfy the first matching confirmation condition is less than the differentiation reference value generated based on historical matching data. If multiple typical fault modes simultaneously satisfy the first matching confirmation condition and the second matching differentiation condition, the edge computing unit will jointly determine the multiple typical fault modes that satisfy the conditions as a composite dominant fault type; otherwise, the edge computing unit will determine the typical fault mode with the highest feature matching degree value as a single current dominant fault type. 6.The edge computing based distributed equipment intelligent diagnosis and precision inspection management method of claim 5, wherein: The results of the precision inspection analysis include a confidence assessment of the diagnostic conclusions, specifically: The edge computing unit calculates and outputs a confidence level value representing the reliability of the diagnostic conclusion based on the degree of match between the current dominant fault type and the feature pattern sequence, and in conjunction with the failure impact level quantification parameters associated with the current dominant fault type obtained from the local FMEA knowledge base.
7. The edge computing-based distributed equipment intelligent diagnosis and precision inspection management method according to claim 6, characterized in that: The precision inspection analysis results also include the execution priority of maintenance recommendations, specifically: The edge computing unit obtains the recommended maintenance procedure index from the local FMEA knowledge base based on the current dominant fault type, and combines the confidence value of the diagnostic conclusion and the failure impact level quantification parameter to determine and output the priority level of maintenance recommendations through a preset decision matrix.
8. A distributed equipment intelligent diagnosis and precision point inspection management system based on edge computing, applying the distributed equipment intelligent diagnosis and precision point inspection management method based on edge computing according to any one of claims 1-7, characterized in that, include: The precision inspection data processing module is used to process the precision inspection data acquired by the self-monitoring distributed industrial equipment and extract the feature pattern sequence used to characterize the abnormal operation of the distributed industrial equipment, wherein the precision inspection data is a set of equipment operating parameters that change over time. The fault type determination module is used to determine the current dominant fault type by comparing the feature pattern sequence with the local FMEA knowledge base; The analysis results output module is used to output precise inspection analysis results based on the current dominant fault type and the related handling information obtained from the local FMEA knowledge base. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. When the processor executes the computer program, it implements the steps of the distributed device intelligent diagnosis and precision inspection management method based on edge computing as described in any one of claims 1 to 7.
10. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the distributed device intelligent diagnosis and precision inspection management method based on edge computing as described in any one of claims 1 to 7.
