A real-time fault diagnosis method and system for a minced garlic chili sauce production line

Abstract

The present application relates to the technical field of intelligent monitoring, and discloses a real-time fault diagnosis method and system for a garlic chili sauce production line, the method comprising: standardizing real-time operation data of equipment components in the garlic chili sauce production line to obtain standard operation data; quantifying the deviation of historical performance parameters of the equipment components from the standard operation data to obtain component deterioration rates of the equipment components; comparing and calibrating the component deterioration rates with preset health thresholds to obtain fault performance indicators of the equipment components; combining time domain features and frequency domain features in the fault performance indicators into feature vectors of the equipment components, and performing pattern matching between the feature vectors and a preset fault feature library to obtain fault types; constructing a fault propagation directed graph according to the connection relationship of the equipment components; and performing traceability positioning analysis on the fault types and the fault propagation directed graph to obtain fault root components; the present application can improve the efficiency of real-time fault diagnosis for the garlic chili sauce production line.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology, and in particular to a real-time fault diagnosis method and system for a garlic chili sauce production line. Background Technology

[0002] A garlic chili sauce production line consists of multiple functionally interconnected equipment components. During operation, these components continuously generate real-time operational data, which is a crucial foundation for fault diagnosis. However, current technologies lack efficient preprocessing mechanisms for handling this type of real-time data. This makes it difficult to effectively remove noise interference from the data and to achieve standardized scaling for data of different types and sources. Consequently, the basic data upon which the diagnostic process relies is biased, failing to accurately reflect the actual operating status of the equipment components. This results in insufficient timeliness and accuracy of fault warnings, making it difficult to meet the basic real-time diagnostic requirements of the production line.

[0003] Existing technologies also have significant limitations in the in-depth analysis stage of fault diagnosis. They cannot systematically measure and analyze the deviation between real-time operating data and historical performance parameters of equipment components, making it difficult to scientifically calculate component degradation rates and thus failing to accurately calibrate the risk level and fault performance indicators of the equipment. Furthermore, existing technologies lack effective modeling methods for the fault propagation relationships between equipment components in the production line, failing to clearly present fault propagation paths. After a fault occurs, the root cause component cannot be quickly located, often only addressing surface-level faults. This not only results in low diagnostic efficiency but also easily leads to recurring faults due to failure to eradicate the root cause, prolonging production line downtime and adversely affecting production continuity and efficiency. Therefore, improving the efficiency of real-time fault diagnosis in garlic chili sauce production lines has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a real-time fault diagnosis method and system for a garlic chili sauce production line to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a real-time fault diagnosis method for a garlic chili sauce production line, comprising:

[0006] S1. Standardize the real-time operating data of the equipment components in the garlic chili sauce production line to obtain the standard operating data of the garlic chili sauce production line;

[0007] S2. Measure the deviation between the historical performance parameters of the equipment component and the standard operating data to obtain the component degradation rate of the equipment component;

[0008] S3. The degradation rate of the component is compared with a preset health threshold to obtain the failure performance index of the equipment component.

[0009] S4. Combine the time-domain features and frequency-domain features in the fault performance indicators into a feature vector of the equipment component, and perform pattern matching between the feature vector and a preset fault feature library to obtain the fault type of the garlic chili sauce production line.

[0010] S5. Based on the connection relationship of the equipment components, construct a directed graph of fault propagation for the garlic chili sauce production line;

[0011] S6. Perform source tracing and localization analysis on the fault type and the directed graph of fault propagation to obtain the root cause component of the fault in the garlic chili sauce production line.

[0012] In a preferred embodiment, the standardization processing of the real-time operating data of the equipment components in the garlic chili sauce production line to obtain the standard operating data of the garlic chili sauce production line includes:

[0013] Collect real-time operating data of equipment components in the garlic chili sauce production line;

[0014] The real-time operating data is cleaned to obtain the clean operating data of the garlic chili sauce production line;

[0015] The cleanroom operation data is standardized to obtain the standard operation data of the garlic chili sauce production line.

[0016] In a preferred embodiment, the step of measuring the deviation between the historical performance parameters of the device component and the standard operating data to obtain the component degradation rate of the device component includes:

[0017] Multimodal feature evolution is performed on the multidimensional feature parameters of the device component to obtain the baseline feature matrix of the device component;

[0018] Principal component analysis is performed on the baseline feature matrix to obtain the feature projection vector and variance contribution rate of the device components;

[0019] The difference between the baseline feature matrix and the feature projection vector is calculated to obtain the deviation of the device component, wherein the formula for calculating the deviation is as follows:

[0020] ;

[0021] In the formula, This indicates the degree of deviation of the device component. This represents the number of principal components in the feature projection vector. The first element in the feature projection vector represents the first element. The variance contribution rate of each principal component The first element in the feature projection vector represents the first element. The variance contribution rate of each principal component Represent a natural constant. The first element in the feature projection vector represents the first element. The time decay of each feature This represents the preset characteristic decay time constant. The first element in the feature projection vector represents the first element. The projected values ​​of each principal component. The first element in the feature projection vector represents the first element. Historical mean of each principal component The first element in the feature projection vector represents the first element. The variance of each principal component This represents the size of the time window for the feature projection vector;

[0022] Based on the deviation, calculate the initial degradation rate of the device component;

[0023] Adaptive calibration is performed on the initial degradation rate to obtain the calibrated component degradation rate of the device component.

[0024] In a preferred embodiment, the initial degradation rate is calculated using the following formula:

[0025] ;

[0026] In the formula, This indicates the initial degradation rate of the device component. Represent a natural constant. This represents the preset steepness coefficient of the degradation curve. This indicates the degree of deviation of the device component. This indicates the reference deviation threshold of the device component. This represents the size of the time window for the feature projection vector. Indicates the device component number Deviation from a given time unit.

[0027] In a preferred embodiment, the step of comparing the component degradation rate with a preset health threshold to obtain the failure performance index of the device component includes:

[0028] The component degradation rate is compared with a preset health threshold one by one to determine the risk level of the equipment component;

[0029] Based on the risk level, assess the risk of the component's degradation rate to obtain the risk assessment value of the equipment component;

[0030] Based on the risk assessment value, a correlation analysis is performed on the real-time operating parameters to obtain the fault performance indicators of the equipment components.

[0031] In a preferred embodiment, the step of combining the time-domain and frequency-domain features of the fault performance indicators into a feature vector of the equipment component, and performing pattern matching between the feature vector and a preset fault feature library to obtain the fault type of the garlic chili sauce production line, includes:

[0032] Extract the time-domain and frequency-domain features from the fault performance indicators to obtain the time-domain and frequency-domain feature data of the fault performance indicators;

[0033] The time-domain feature data and the frequency-domain feature data are dimensionally aligned to obtain the aligned feature data of the device component.

[0034] The aligned feature data is matched with a preset fault feature library to obtain the fault type of the garlic chili sauce production line.

[0035] In a preferred embodiment, constructing a directed fault propagation graph of the garlic chili sauce production line based on the device component connection relationships includes:

[0036] Obtain the device component connection relationships in the configuration information of the garlic chili sauce production line to obtain the component topology data of the garlic chili sauce production line;

[0037] Based on the component topology data, the connection nodes between the equipment components are identified to obtain the node relationship data of the garlic chili sauce production line;

[0038] Based on the component topology data and the node relationship data, construct a component directed graph of the garlic chili sauce production line;

[0039] The fault propagation direction of the directed graph of the components is marked to generate the fault propagation directed graph of the garlic chili sauce production line.

[0040] In a preferred embodiment, the step of performing source tracing and localization analysis on the fault type and the directed fault propagation graph to obtain the root cause components of the garlic chili sauce production line includes:

[0041] Based on the fault type of the garlic chili sauce production line, identify the fault propagation path of the directed fault propagation graph;

[0042] By performing reverse tracing analysis on the fault propagation path, candidate root cause nodes of the garlic chili sauce production line are obtained;

[0043] The failure probability of the candidate root cause nodes is evaluated to obtain the failure occurrence probability of the candidate root cause nodes.

[0044] Based on the probability of failure, the root cause component of the failure in the garlic chili sauce production line is determined.

[0045] In a preferred embodiment, the step of evaluating the failure probability of the candidate root cause nodes to obtain the probability of failure of the candidate root cause nodes includes:

[0046] Obtain the node operation status data and associated node status data of the candidate fault root cause node;

[0047] By analyzing the frequency of abnormal features in the node operation status data, the node abnormality frequency data of the node operation status data is obtained.

[0048] Extract the abnormal propagation features from the state data of the associated nodes to obtain the abnormal propagation feature data in the state data of the associated nodes;

[0049] By comprehensively evaluating the node anomaly frequency data and the anomaly propagation characteristic data, the probability of failure of the candidate fault root cause node is obtained.

[0050] To address the above problems, the present invention also provides a real-time fault diagnosis system for a garlic chili sauce production line, the system comprising:

[0051] The data standardization module is used to standardize the real-time operating data of the equipment components in the garlic chili sauce production line to obtain the standard operating data of the garlic chili sauce production line.

[0052] The degradation rate calculation module is used to measure the deviation between the historical performance parameters of the equipment component and the standard operating data to obtain the component degradation rate of the equipment component.

[0053] The risk calibration module is used to compare and calibrate the component degradation rate with a preset health threshold to obtain the failure performance index of the equipment component.

[0054] The fault diagnosis module is used to combine the time-domain features and frequency-domain features in the fault performance indicators into a feature vector of the equipment component, and to perform pattern matching between the feature vector and a preset fault feature library to obtain the fault type of the garlic chili sauce production line.

[0055] The propagation graph construction module is used to construct a directed graph of fault propagation for the garlic chili sauce production line based on the connection relationships of the equipment components.

[0056] The root cause analysis module is used to perform source tracing and localization analysis on the fault type and the directed graph of fault propagation to obtain the root cause components of the fault in the garlic chili sauce production line.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] 1. This invention eliminates data bias and ensures reliable diagnostic foundation by standardizing and cleaning up noise and unifying the scale of real-time operating data of equipment components. Then, it calculates the deviation degree and component degradation rate through multimodal feature evolution, principal component analysis and specific formulas to accurately quantify the degree of equipment performance degradation. Subsequently, it constructs vectors by combining the time domain and frequency domain features of fault performance indicators and matches them with a preset fault feature library to identify fault types. The entire process is based on accurate data analysis and real-time processing, which effectively avoids the misjudgment problem of traditional diagnosis and greatly improves the accuracy and real-time efficiency of fault diagnosis.

[0059] 2. This invention constructs a directed fault propagation graph based on the connection relationships of equipment components, clarifies the fault propagation path by combining the identified fault types, identifies candidate fault root cause nodes through reverse tracing, and evaluates the fault probability by comprehensively considering the abnormal frequency of nodes and the propagation characteristics of associated nodes, ultimately accurately locating the root cause component. This process avoids the limitations of traditional diagnostics that only address surface faults, effectively prevents repeated fault occurrences, shortens production line downtime, and ensures production continuity. Attached Figure Description

[0060] Figure 1 This is a flowchart illustrating a real-time fault diagnosis method for a garlic chili sauce production line according to an embodiment of the present invention.

[0061] Figure 2 This is a functional module diagram of a real-time fault diagnosis system for a garlic chili sauce production line provided in an embodiment of the present invention;

[0062] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0063] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0064] This application provides a real-time fault diagnosis method for a garlic chili sauce production line. The execution subject of this real-time fault diagnosis method for a garlic chili sauce production line includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the real-time fault diagnosis method for a garlic chili sauce production line can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0065] Reference Figure 1 The diagram shown is a flowchart illustrating a real-time fault diagnosis method for a garlic chili sauce production line according to an embodiment of the present invention. In this embodiment, the real-time fault diagnosis method for a garlic chili sauce production line includes:

[0066] S1. Standardize the real-time operating data of the equipment components in the garlic chili sauce production line to obtain the standard operating data of the garlic chili sauce production line;

[0067] In this embodiment of the invention, the standardization processing of the real-time operating data of the equipment components in the garlic chili sauce production line to obtain the standard operating data of the garlic chili sauce production line includes:

[0068] Collect real-time operating data of equipment components in the garlic chili sauce production line;

[0069] The real-time operating data is cleaned to obtain the clean operating data of the garlic chili sauce production line;

[0070] The cleanroom operation data is standardized to obtain the standard operation data of the garlic chili sauce production line.

[0071] Specifically, the entire content revolves around the processing of operational data of the equipment components in the garlic chili sauce production line. It covers three core steps in sequence: real-time operational data acquisition, noise removal of real-time operational data to obtain clean operational data, and standardization of clean operational data to obtain standard operational data. Each step clearly defines the specific operation method and the corresponding data product generated in the end.

[0072] Furthermore, in the real-time operational data acquisition stage, the core equipment components that need to collect data are first identified, including garlic mixing tanks, chili crushing units, sauce cooking pots, automatic filling machines, and conveying pumps. Then, matching sensors are installed on each equipment component, such as temperature and speed sensors for the garlic mixing tanks and current and vibration sensors for the chili crushers. After that, all sensors are connected to an industrial data acquisition card via signal lines. The acquisition card is then connected to a local data server via an industrial Ethernet network. The frequency of collecting parameters once per second is set, and the collected parameters are transmitted to the server for storage in real time according to a specific format, ultimately obtaining real-time operational data.

[0073] Furthermore, in the noise cleaning process, it is first determined that the noise in the real-time operating data comes from instantaneous fluctuations of sensors and electromagnetic interference from the production line. The moving average method is used for processing. Taking a set of data from a certain equipment component as an example, data from 5 consecutive moments are selected as the processing window. The arithmetic mean of the data within the window is calculated to replace the original data at the middle moment of the window. The operation is repeated by moving the window one by one. This process is used to process each real-time operating data of all equipment components to obtain clean operating data.

[0074] Furthermore, in the standardization process, the scale differences in cleanroom operation data due to different parameter units and numerical ranges are first analyzed. A linear transformation method based on the data range is used to transform the data to the 0-1 range. All cleanroom operation data are processed in this way to obtain standard operation data.

[0075] In summary, by identifying the core equipment components of the garlic chili sauce production line, installing corresponding sensors, establishing connections between the sensors and data acquisition cards and servers, setting the acquisition frequency, and storing the data in a specific format, the real-time operating data of the equipment components was collected, ultimately yielding the real-time operating data of the equipment components in the garlic chili sauce production line.

[0076] In summary, by identifying the sources of noise in the real-time operating data, the moving average method was used to calculate the average value of each real-time operating data item by moving the processing window one by one and replacing the original data, thus completing the noise removal and finally obtaining the clean operating data of the garlic chili sauce production line.

[0077] In summary, by analyzing the scale differences in cleanroom operation data, and using a linear transformation method based on data range to find the maximum and minimum values ​​and perform numerical transformations for each cleanroom operation data item, scale unification was achieved, and the standard operation data of the garlic chili sauce production line was finally obtained.

[0078] S2. Measure the deviation between the historical performance parameters of the equipment component and the standard operating data to obtain the component degradation rate of the equipment component;

[0079] In this embodiment of the invention, the step of measuring the deviation between the historical performance parameters of the device component and the standard operating data to obtain the component degradation rate of the device component includes:

[0080] Multimodal feature evolution is performed on the multidimensional feature parameters of the device component to obtain the baseline feature matrix of the device component;

[0081] Principal component analysis is performed on the baseline feature matrix to obtain the feature projection vector and variance contribution rate of the device components;

[0082] The difference between the baseline feature matrix and the feature projection vector is calculated to obtain the deviation of the device component, wherein the formula for calculating the deviation is as follows:

[0083] ;

[0084] In the formula, This indicates the degree of deviation of the device component. This represents the number of principal components in the feature projection vector. The first element in the feature projection vector represents the first element. The variance contribution rate of each principal component The first element in the feature projection vector represents the first element. The variance contribution rate of each principal component Represent a natural constant. The first element in the feature projection vector represents the first element. The time decay of each feature This represents the preset characteristic decay time constant. The first element in the feature projection vector represents the first element. The projected values ​​of each principal component. The first element in the feature projection vector represents the first element. Historical mean of each principal component The first element in the feature projection vector represents the first element. The variance of each principal component This represents the size of the time window for the feature projection vector;

[0085] Based on the deviation, calculate the initial degradation rate of the device component;

[0086] Adaptive calibration is performed on the initial degradation rate to obtain the calibrated component degradation rate of the device component.

[0087] The formula for calculating the initial degradation rate is as follows:

[0088] ;

[0089] In the formula, This indicates the initial degradation rate of the device component. Represent a natural constant. This represents the preset steepness coefficient of the degradation curve. This indicates the degree of deviation of the device component. This indicates the reference deviation threshold of the device component. This represents the size of the time window for the feature projection vector. Indicates the device component number Deviation from a given time unit.

[0090] Specifically, the entire process revolves around determining the degradation rate of equipment components. First, the multi-dimensional characteristic parameters of the equipment components are subjected to multi-modal feature evolution to obtain the baseline feature matrix. Then, principal component analysis is used to obtain the feature projection vector and variance contribution rate.

[0091] Specifically, the deviation is obtained by calculating the difference between the baseline feature matrix and the feature projection vector. Then, the initial degradation rate is calculated based on the deviation. Finally, the initial degradation rate is adaptively calibrated to obtain the calibrated component degradation rate, which is obtained by combining the formulas for deviation and initial degradation rate.

[0092] Furthermore, when performing multimodal feature evolution, it is clarified that the multidimensional feature parameters cover three types of modal data: time domain, frequency domain, and physical state. First, the feature parameters of the three types of modalities are standardized by 0-1 so that the values ​​of all feature parameters are between 0 and 1.

[0093] Furthermore, based on the time sequence of continuous equipment operation, the standardized feature parameters at each time point are integrated into a feature vector, and the feature vectors at all time points are arranged in chronological order into a two-dimensional data table to obtain the baseline feature matrix.

[0094] Furthermore, when performing principal component analysis on the baseline feature matrix, the feature parameters in each column are first centered to eliminate the influence of dimensions. Then, the direction with the largest variance in the data distribution is identified as the first feature projection vector. Next, the direction perpendicular to the first vector and with the second largest variance is identified as the second feature projection vector. In this way, several main feature projection vectors are extracted.

[0095] Furthermore, the variance of each feature projection vector is calculated and divided by the sum of the variances of all extracted vectors to obtain the variance contribution rate of each vector. Finally, a set of main feature projection vectors and a list of corresponding variance contribution rates are compiled.

[0096] Furthermore, when calculating the deviation, the steps are as follows: the feature vector at each time point in the baseline feature matrix is ​​compared with each feature projection vector one by one. The elements at corresponding positions are subtracted and the absolute difference is summed to obtain a single set of difference values. For the feature vector at each time point, the average of the single set of difference values ​​with all feature projection vectors is calculated to obtain the comprehensive difference value. Then, the arithmetic mean of the comprehensive difference values ​​at all time points is calculated to obtain the deviation.

[0097] Furthermore, looking at the formula, the parameters... It is the number of principal components extracted by principal component analysis. , It is the variance contribution rate of the principal components. It is obtained from the decay rate per unit time based on the statistical characteristics of historical data. Determined by fitting the device characteristics. It is the result of projecting the data onto the principal component vectors. It is the average value of the principal component components during normal operation. It is the variance of the principal component components during normal operation. The formula is set according to the equipment's operating conditions. It integrates the variance contribution of multiple principal components, the characteristic time decay, and the difference from historical normal data to quantify the degree of deviation of the equipment from the normal benchmark. When calculating, the deviation of each principal component is integrated into an overall index. The trend is that the more the equipment's characteristic projection value deviates from the historical mean, the more the time decay increases, and the greater the variance contribution rate of the principal components, the greater the degree of deviation.

[0098] Furthermore, when calculating the initial degradation rate, the steps are as follows: first, obtain the deviation values ​​of the equipment under normal and fault conditions from the historical operation database, calculate the effective deviation range, then calculate the difference between the current deviation and the normal deviation value, and divide it by the effective deviation range to obtain the initial degradation rate.

[0099] Furthermore, in the formula, It is a natural constant. Determined by fitting the correspondence between historical degradation data and faults. Calculated using the deviation formula, This is the statistical upper limit of the deviation from normal operation. Deviation Formula Consistent, This is the deviation result at the corresponding historical time point; the formula combines the current and historical deviations, the steepness of the degradation curve to quantify the initial degradation level, and integrates the influence of the current and historical deviations during calculation, as well as the trend. Greater than hour, The larger the value and the greater the historical deviation, the higher the initial degradation rate. The larger the value, Exceed The faster the initial degradation rate increases.

[0100] Furthermore, when performing adaptive calibration on the initial degradation rate, the historical degradation records of the equipment over the past three years are first collected. The records include the deviation of each test, the initial degradation rate, and the subsequent actual degradation. The correspondence between the initial degradation rate and the actual degradation is analyzed, and corresponding calibration coefficients are set for different intervals: if the initial degradation rate in a certain interval is generally lower than the actual rate, a coefficient greater than 1 is set; if it is generally higher, a coefficient less than 1 is set.

[0101] Furthermore, a calibration coefficient is selected based on the current initial degradation rate range, and multiplied by the current initial degradation rate to obtain the calibrated component degradation rate, which can more accurately reflect the actual degradation state of the equipment.

[0102] In summary, by identifying three types of modal data, standardizing them, and integrating the vectors according to the time series, the baseline feature matrix of the device components was finally obtained.

[0103] In summary, by centering the feature parameters, extracting the vector of the direction with the maximum variance, and calculating the variance proportion, the feature projection vector and variance contribution rate of the device components are finally obtained.

[0104] In summary, by comparing vectors at the step level, calculating the difference value, and then averaging it, and by combining factors such as multiple principal components and time decay at the formula level, the deviation of the equipment components was finally obtained.

[0105] In summary, by using historical deviation values ​​to calculate the effective range and difference, and combining the current deviation from the historical values ​​with the degradation steepness at the formula level, the initial degradation rate of the equipment components was finally obtained.

[0106] In summary, by collecting historical degradation records, analyzing the corresponding relationships to set calibration coefficients, and performing multiplication operations, the degradation rate of the equipment components after calibration was finally obtained.

[0107] S3. The degradation rate of the component is compared with a preset health threshold to obtain the failure performance index of the equipment component.

[0108] In this embodiment of the invention, the step of comparing and calibrating the component degradation rate with a preset health threshold to obtain the failure performance index of the device component includes:

[0109] The component degradation rate is compared with a preset health threshold one by one to determine the risk level of the equipment component;

[0110] Based on the risk level, assess the risk of the component's degradation rate to obtain the risk assessment value of the equipment component;

[0111] Based on the risk assessment value, a correlation analysis is performed on the real-time operating parameters to obtain the fault performance indicators of the equipment components.

[0112] Specifically, the entire content revolves around the risk analysis of equipment components, and includes three core steps: comparing the component degradation rate with preset health thresholds one by one to determine the risk level of the equipment component; assessing the risk of the component degradation rate based on the risk level to obtain the risk assessment value of the equipment component; and performing correlation analysis on real-time operating parameters based on the risk assessment value to obtain the failure performance index of the equipment component. Each step clearly defines the specific operation method and the corresponding final product.

[0113] Furthermore, in the process of determining the risk level of equipment components, the first step is to combine the design life of the equipment components, the degradation rate data of the past three years of trouble-free operation, and the health operation standards of similar components in the industry to set a unique preset health threshold for each equipment component. For example, different health thresholds are set for garlic mixing tanks and chili crushing units.

[0114] Furthermore, the component degradation rate of all equipment components is extracted from the data storage system. According to the serial number of the equipment components, the degradation rate of each component is compared with the preset health threshold corresponding to that component one by one. The risk level is divided according to the comparison results: degradation rate less than the threshold is low risk, equal to the threshold is medium risk, and greater than the threshold is high risk. Finally, the risk level of each equipment component is determined.

[0115] Furthermore, in the step of obtaining the risk assessment value of equipment components, a fixed base score is first set for different risk levels. Low risk corresponds to a base score, medium risk corresponds to another fixed base score, and high risk corresponds to a higher fixed base score.

[0116] Furthermore, for each equipment component, the deviation between its component degradation rate and the corresponding preset health threshold is calculated. The base score is adjusted according to the deviation range under different risk levels. For example, the score is increased when the deviation in a low-risk component reaches a certain level. Medium-risk and high-risk components also have different scoring rules according to their respective deviation ranges. The adjusted score is the risk assessment value of the equipment component. Finally, the risk assessment values ​​of all equipment components are obtained.

[0117] Furthermore, in the step of obtaining the failure performance index of equipment components, the real-time operating parameters of all equipment components, including temperature, speed, current, and vibration values, are first extracted from the real-time operating data repository. Then, based on the risk assessment values ​​of the equipment components, components whose risk assessment values ​​meet specific standards are selected as key analysis objects. For each key analysis object, its real-time operating parameter data over a recent period is extracted, and the changing patterns of these parameters over time are observed. Key parameters consistent with the trend of component degradation rate are identified, and these key parameters are combined to form an index reflecting the component's failure status; this index is the failure performance index of the equipment component.

[0118] In summary, by combining the design life of equipment components, past fault-free data, and industry standards to set preset health thresholds, and then comparing the component degradation rate with the corresponding thresholds one by one according to the component number and classifying them into levels, the risk level of the equipment components was determined, and the risk level of the equipment components was finally obtained.

[0119] In summary, by setting fixed base scores for different risk levels and then adjusting the base scores according to the deviation between the component degradation rate and the preset health threshold, the risk assessment of the component degradation rate was completed, and the risk evaluation value of the equipment component was finally obtained.

[0120] In summary, by extracting real-time operating parameters, selecting key analysis objects, observing the changing patterns of parameters, and identifying related key parameters, the correlation analysis of real-time operating parameters was completed, and the fault performance indicators of equipment components were finally obtained.

[0121] S4. Combine the time-domain features and frequency-domain features in the fault performance indicators into a feature vector of the equipment component, and perform pattern matching between the feature vector and a preset fault feature library to obtain the fault type of the garlic chili sauce production line.

[0122] In this embodiment of the invention, the step of combining the time-domain and frequency-domain features in the fault performance indicators into a feature vector of the equipment component, and performing pattern matching between the feature vector and a preset fault feature library to obtain the fault type of the garlic chili sauce production line, includes:

[0123] Extract the time-domain and frequency-domain features from the fault performance indicators to obtain the time-domain and frequency-domain feature data of the fault performance indicators;

[0124] The time-domain feature data and the frequency-domain feature data are dimensionally aligned to obtain the aligned feature data of the device component.

[0125] The aligned feature data is matched with a preset fault feature library to obtain the fault type of the garlic chili sauce production line.

[0126] Specifically, the entire process revolves around fault diagnosis of the garlic chili sauce production line. First, time-domain and frequency-domain features are extracted from the fault performance indicators of the equipment components, forming corresponding time-domain feature data and frequency-domain feature data, respectively.

[0127] Specifically, by adjusting the number of features and integrating them in sequence, the two types of feature data are dimensionally aligned to obtain aligned feature data. Finally, the aligned feature data is compared with a preset fault feature library on a feature-by-feature basis to find a matching feature set, thereby determining the fault type of the production line.

[0128] Furthermore, when obtaining fault performance indicators, it is necessary to collect performance values ​​corresponding to continuous time points during equipment operation to form complete time series data.

[0129] Furthermore, to calculate the mean of the time-domain characteristics, the performance values ​​at all time points are summed, and then the sum is divided by the total number of time points. To calculate the peak value, the largest value is selected directly from all performance values. To calculate the root mean square, each performance value is squared, all squared results are summed and divided by the total number of time points, and finally the square root of the result is taken. Summarizing these calculated time-domain parameters gives the time-domain characteristic data of the fault performance index.

[0130] Furthermore, when processing frequency domain features, the time series data is first divided into multiple continuous data segments according to fixed time intervals. The Fourier transform of the values ​​of each data segment is performed to convert the information in the time dimension into information in the frequency dimension, thus obtaining frequency domain data. The largest amplitude among all frequencies in the frequency domain data is then identified as the dominant frequency amplitude.

[0131] Furthermore, a preset frequency range is determined, and the amplitudes corresponding to all frequencies within this range are added together. The sum obtained is the frequency band energy. The frequency domain parameters such as the main frequency amplitude and frequency band energy are summarized to form the frequency domain characteristic data of the fault performance index.

[0132] Furthermore, when performing dimension alignment, the number of features contained in the time-domain feature data and the frequency-domain feature data are first counted separately to determine their respective dimensions. If the dimensions of the time-domain feature data and the frequency-domain feature data are different, additional features need to be added to adjust them. For example, if the time-domain feature data has three features while the frequency-domain feature data only has two features, a new feature is added to the frequency-domain feature data. During the addition process, the correlation between frequency and amplitude in the frequency-domain data is analyzed.

[0133] Furthermore, the center frequency of the frequency band is calculated by multiplying each frequency value within the preset frequency range by its corresponding amplitude. The sum of all products is then divided by the sum of all amplitudes within that frequency range. The result is the center frequency of the frequency band. This new feature is added to the frequency domain feature data to ensure consistency in the dimensions of the two types of feature data. Then, following the order of "time domain features first, frequency domain features second," all features of the two types of feature data are arranged sequentially to form aligned feature data with a unified dimension.

[0134] Furthermore, before conducting pattern matching, it is necessary to clarify the contents of the preset fault feature library. The library stores feature sets corresponding to all possible fault types in the garlic chili sauce production line. Each fault type has a unique feature set, and the dimensions and feature types of each feature set are completely consistent with the aligned feature data. Each feature is marked with a clear numerical range.

[0135] Furthermore, during the comparison, each feature value in the aligned feature data is checked one by one against the corresponding feature value in each fault type feature set in the fault feature library. For example, the mean value in the aligned feature data must be checked against the mean value in a certain fault type feature set to see if the former is within the preset value range of the latter. Other features such as peak value, root mean square, main frequency amplitude, frequency band energy, and frequency band center frequency are also checked in the same way. Only when all feature values ​​in the aligned feature data meet the corresponding feature value range of a certain fault type feature set can the two be determined to be successfully matched. The fault type that is successfully matched is the fault type that actually exists on the production line.

[0136] In summary, starting from the time series data of fault performance indicators, time-domain parameters are obtained by calculating the mean, peak value, and root mean square. Frequency-domain parameters are obtained by transforming the data through Fourier transform and extracting the main frequency amplitude and frequency band energy, thus forming the time-domain characteristic data and frequency-domain characteristic data of fault performance indicators, respectively.

[0137] In summary, the dimensions of time-domain and frequency-domain feature data are determined by the number of statistical features. If necessary, features such as the center frequency of the frequency band are added to adjust the consistency of the dimensions. Then, the two types of features are integrated in a fixed order to finally obtain the aligned feature data of the device components.

[0138] In summary, by comparing each feature value of the aligned feature data with the corresponding feature values ​​of each fault type in the preset fault feature library one by one, the fault type in which all features match is found, and finally the fault type of the garlic chili sauce production line is obtained.

[0139] S5. Based on the connection relationship of the equipment components, construct a directed graph of fault propagation for the garlic chili sauce production line;

[0140] In this embodiment of the invention, constructing a directed graph of fault propagation for the garlic chili sauce production line based on the connection relationships of the equipment components includes:

[0141] Obtain the device component connection relationships in the configuration information of the garlic chili sauce production line to obtain the component topology data of the garlic chili sauce production line;

[0142] Based on the component topology data, the connection nodes between the equipment components are identified to obtain the node relationship data of the garlic chili sauce production line;

[0143] Based on the component topology data and the node relationship data, construct a component directed graph of the garlic chili sauce production line;

[0144] The fault propagation direction of the directed graph of the components is marked to generate the fault propagation directed graph of the garlic chili sauce production line.

[0145] Specifically, the entire process revolves around generating a directed graph of fault propagation in the garlic chili sauce production line. First, the connection relationships of equipment components are extracted from the production line configuration information to obtain component topology data. Then, based on this data, the connection nodes between components are identified to obtain node relationship data. Next, the component topology data and node relationship data are combined to construct a component directed graph. Finally, the fault propagation direction is marked on the component directed graph, and the fault propagation directed graph of the production line is finally generated.

[0146] Furthermore, when collecting production line configuration information, it is necessary to cover the name, model, installation location, and connection method of all equipment components, such as the specifications of the connecting pipes between the grinder and the mixing tank, and the docking method between the filling machine and the conveyor belt.

[0147] Furthermore, when sorting out this configuration information, it is necessary to record the upstream and downstream components directly corresponding to each component one by one. First, it should be clear that the upstream of the grinder is the chili raw material conveyor belt and the downstream is the mixing tank. Then, the upstream and downstream connection relationships of all components should be organized into a structured table containing "component name", "upstream connected component", "downstream connected component" and "connection medium". This table is the component topology data.

[0148] Furthermore, connection nodes are defined as specific interfaces or connection points for functional transfer between components, such as the connection between the grinder outlet and the conveying pipeline, or the connection between the conveying pipeline and the mixing tank inlet. When examining the connection relationships of each group of upstream and downstream components in the component topology data, the specific connection nodes must be identified. For example, from the connection relationship of "grinder - conveying pipeline - mixing tank", the "grinder outlet node" and the "mixing tank inlet node" are identified. At the same time, the names and function types of the upstream and downstream components corresponding to each node are recorded. Then, this node information is organized into a table containing "node name", "upstream corresponding component", "downstream corresponding component", and "node function type". This table is the node relationship data.

[0149] Furthermore, the constituent elements of the component-directed graph are determined, with equipment components as vertices, connections between components as directed edges, and connecting nodes as edge labeling information. First, all component names are labeled as independent vertices on the drawing platform. Then, directed edges are drawn based on the upstream and downstream relationships in the component topology data, such as drawing an arrow from the "grinding machine" vertex to the "mixing tank" vertex. Finally, combining the node relationship data, the corresponding connecting node names are labeled on each directed edge. For example, the arrow pointing from "grinding machine" to "mixing tank" is labeled with "grinding machine outlet node" and "mixing tank inlet node." The resulting graph is the component-directed graph.

[0150] Furthermore, by analyzing the logic of fault propagation, it is clarified that faults will be transmitted from upstream components to downstream components through connection nodes along the functional transmission direction of the components. Faults in downstream components will not be transmitted back to upstream components. For example, a fault in the grinder will be transmitted to the mixing tank, but a fault in the mixing tank will not be transmitted to the grinder. On each directed edge of the component directed graph, the direction of fault propagation is confirmed by the arrow direction. If the original arrow direction already matches the propagation direction, it is retained. Then, a text description of "fault propagation direction" is added next to each directed edge. For example, the arrow pointing from "grinder" to "mixing tank" is labeled "fault propagation: grinder → mixing tank". The graph after these operations is the directed fault propagation graph.

[0151] In summary, by collecting production line configuration information and sorting out the upstream and downstream connections of components, these relationships were organized into a structured table, ultimately yielding the component topology data of the garlic chili sauce production line.

[0152] In summary, by defining the attributes of the connection nodes, identifying the corresponding nodes for each group of components in combination with the component topology data, and recording the node information, this information is organized into a structured table, ultimately yielding the node relationship data of the garlic chili sauce production line.

[0153] In summary, by determining the constituent elements of the component directed graph, labeling component vertices on the carrier, drawing directed edges, and labeling connection nodes, the component directed graph of the garlic chili sauce production line was finally constructed.

[0154] In summary, by analyzing the logic of fault propagation, confirming the direction of arrows on the directed graph of the components, and supplementing the description of the propagation direction, a directed graph of fault propagation for the garlic chili sauce production line was finally generated.

[0155] S6. Perform source tracing and localization analysis on the fault type and the directed graph of fault propagation to obtain the root cause components of the fault in the garlic chili sauce production line;

[0156] In this embodiment of the invention, the step of performing source tracing and localization analysis on the fault type and the directed graph of fault propagation to obtain the root cause components of the garlic chili sauce production line includes:

[0157] Based on the fault type of the garlic chili sauce production line, identify the fault propagation path of the directed fault propagation graph;

[0158] By performing reverse tracing analysis on the fault propagation path, candidate root cause nodes of the garlic chili sauce production line are obtained;

[0159] The failure probability of the candidate root cause nodes is evaluated to obtain the failure occurrence probability of the candidate root cause nodes.

[0160] Based on the probability of failure, the root cause component of the failure in the garlic chili sauce production line is determined.

[0161] The step of evaluating the failure probability of the candidate root cause nodes to obtain the probability of failure of the candidate root cause nodes includes:

[0162] Obtain the node operation status data and associated node status data of the candidate fault root cause node;

[0163] By analyzing the frequency of abnormal features in the node operation status data, the node abnormality frequency data of the node operation status data is obtained.

[0164] Extract the abnormal propagation features from the state data of the associated nodes to obtain the abnormal propagation feature data in the state data of the associated nodes;

[0165] By comprehensively evaluating the node anomaly frequency data and the anomaly propagation characteristic data, the probability of failure of the candidate fault root cause node is obtained.

[0166] Specifically, the entire process revolves around determining the root cause components of the garlic chili sauce production line failure. First, based on the identified failure type, the failure propagation path is identified in the directed failure propagation graph. Then, these paths are traced back and analyzed to obtain candidate root cause nodes. Next, the failure probability of the candidate nodes is evaluated to obtain their failure probability. Finally, the root cause components of the production line failure are determined based on the failure probability.

[0167] Furthermore, when identifying the fault propagation path, first verify the detailed description of the fault type, clarify the specific functional abnormalities of the affected equipment components, confirm the non-standard performance of the component in the production process, and the impact of such abnormalities on subsequent production stages.

[0168] Furthermore, the vertex corresponding to the affected component is located in the directed graph of fault propagation. This vertex is marked with a labeling tool to clearly identify the starting point for tracing. Then, the associated nodes and components are searched layer by layer along the reverse direction of the directed edges. For each node found, the node name, the actual installation location of the node in the production line, and the upstream and downstream components that the node connects to are recorded in detail to ensure that the relationships between each node are clear and traceable.

[0169] Furthermore, the process continues to trace back in reverse until it reaches the upstream component from which it can no longer extend upwards. Each complete link from the upstream component through all intermediate nodes to the affected component is recorded line by line in a preset order to ensure that the information of all components and nodes in the link is accurate. These complete links are the fault propagation paths.

[0170] Furthermore, when tracing back and analyzing the fault propagation path, each fault propagation path is first compiled into a separate list. The list is arranged in the order from the affected component to the upstream component, listing all the nodes contained in the path in sequence to ensure that no node is missed.

[0171] Furthermore, for each node in the list, the first step is to refer to the production line equipment operation manual, browse through the equipment function chapters related to that node, clarify the specific role the node plays in the production process, and understand the standard state that the node should achieve when operating normally; the second step is to check the production line's fault maintenance records, retrieve all past fault records for that node, including the specific manifestations of the fault, the handling measures taken during the maintenance process, the recovery status of the equipment after maintenance, and whether the node's past faults have caused problems of the same type as the current fault.

[0172] Furthermore, when determining whether a node has the potential to independently cause a failure, if a node's function malfunctions and causes subsequent affected components to exhibit the current failure type without requiring other nodes or components to malfunction, then the node is deemed to have this potential. All nodes that meet this condition are then selected as candidate root cause nodes for failure.

[0173] Furthermore, when assessing the probability of failure of candidate root cause nodes, a fixed assessment period is first determined to be the entire past production cycle. This period covers the high-load operation phase, normal operation phase, and low-load operation phase experienced by the production line, which can comprehensively reflect the operating status of nodes under different working conditions and avoid inaccurate assessment results due to short-term data bias.

[0174] Furthermore, when collecting fault records, three types of key records for each candidate node are retrieved from the equipment management system of the production line: fault report records, maintenance records, and daily inspection records, to ensure that no abnormal situation related to the node is missed.

[0175] Furthermore, when counting the total number of failures at each node, if the same failure spans multiple time periods from its reporting to its completion, it is still counted as one failure to avoid duplicate counting. When calculating the basic failure frequency, the total number of failures at that node is divided by the total duration of the assessment period. If a node does not experience any failures during the entire assessment period, its basic failure frequency is recorded as zero.

[0176] Furthermore, when adjusting the basic fault frequency based on the node's most recent maintenance time, first check the preset maintenance cycle of each candidate node. This cycle is determined comprehensively based on the wear rate of node components described in the equipment manual and the degree of corrosion of the nodes by the materials used in the production line. If the node's most recent maintenance time has exceeded the preset maintenance cycle, the basic fault frequency is increased by a certain percentage, because the aging rate of node components will accelerate after the maintenance cycle is exceeded, and the risk of failure will increase significantly. If the preset maintenance cycle has not been exceeded, the basic fault frequency is kept unchanged. The adjusted basic fault frequency is the probability of failure of the candidate fault root cause node.

[0177] Furthermore, when determining the root cause component of the fault, the failure probability values ​​of all candidate nodes are first precisely compared. By comparing the values ​​bit by bit, the candidate node with the highest probability value is identified to ensure that there is no judgment bias due to similar values. Then, the node relationship data is consulted, and the "core component to which it belongs" information corresponding to the node with the highest probability is found in a special field of the node relationship data. This field clearly records which core device component each node belongs to.

[0178] Furthermore, to verify the accuracy of the correlation, it is necessary to examine the structural design drawings of the core component. The drawings will clearly indicate all the components of the component, the specific location of the node in the component structure, and the connection relationship between the node and other parts of the component. Once it is confirmed that the fault of the node is essentially a problem with a key component of the core component, it can be determined that the core component is the root cause of the fault in the garlic chili sauce production line.

[0179] Specifically, the entire process revolves around determining the probability of failure of candidate root cause nodes. First, the node's operating status data and the status data of associated nodes are obtained. Then, the node's operating status data is parsed to obtain node anomaly frequency data. Next, the status data of associated nodes is extracted to obtain anomaly propagation characteristic data. Finally, the two types of data are comprehensively evaluated to determine the probability of failure of candidate root cause nodes.

[0180] Furthermore, when acquiring data, it is necessary to first clarify that the node operation status data must include the real-time parameters, parameter change trends, and start-stop status feedback of the candidate fault root cause nodes. The real-time parameters cover key equipment operation indicators such as temperature, pressure, speed, and flow rate.

[0181] Furthermore, by installing dedicated sensors on candidate nodes to collect these parameters, the sensors will continuously record data at fixed time intervals. At the same time, a dedicated person will conduct manual inspections of the candidate nodes at a fixed frequency to record whether there are any abnormal physical conditions such as abnormal noises, leaks, or loose parts. The electronic data collected by the sensors and the paper data recorded manually will be integrated to form complete node operation status data.

[0182] Furthermore, when determining associated nodes, based on the list of node connection relationships, nodes that are directly connected to candidate nodes are selected as associated nodes. Using the same sensor acquisition and manual inspection methods, the real-time parameters, change trends and physical status of associated nodes are collected and then organized to form associated node status data.

[0183] Furthermore, when analyzing the abnormal frequency of nodes, the normal operating range of each parameter is first determined based on the production line equipment operation manual. If a parameter exceeds the range, it is judged as an abnormal feature. In the manual inspection records, abnormal noises, leaks and other phenomena are also classified as abnormal features.

[0184] Furthermore, the node operation status data is divided into multiple statistical periods with fixed durations. Within each period, all data records are checked one by one, and the total number of times abnormal features appear is counted. The frequency of abnormal features within each period is obtained by dividing the total number of abnormal features in each period by the total duration of that period.

[0185] Furthermore, the statistical results of all periods are organized into a table in chronological order. The table contains information such as the statistical period, the total number of abnormal features, and the frequency of abnormal occurrences. This table is the node abnormality frequency data.

[0186] Furthermore, when extracting abnormal propagation features, the judgment logic for abnormal propagation is first set: after a candidate node exhibits abnormal features, if the associated node also exhibits abnormalities within a preset time interval, and the two types of abnormalities are causally related, such as when a candidate node is blocked and causes abnormal traffic, the associated node will also exhibit abnormal traffic due to insufficient material supply, then it is judged as abnormal propagation.

[0187] Furthermore, the time records of node operation status data and associated node status data are compared segment by segment. The time points when candidate nodes become abnormal are marked. The status data of associated nodes within a preset time interval after that time point are viewed. The abnormality of associated nodes, the type of abnormality, the time difference between the abnormality of the associated nodes and that of the candidate nodes, and the causal relationship are recorded.

[0188] Furthermore, all records that conform to the judgment logic are organized into a set, which includes the propagation time, candidate node anomaly type, associated node anomaly type, time difference, causal relationship description, etc. This set is the anomaly propagation characteristic data.

[0189] Furthermore, during the comprehensive evaluation, two evaluation dimensions are first set: node anomaly frequency and anomaly propagation characteristics. In the node anomaly frequency dimension, the levels are divided according to the anomaly frequency value: "high" is the frequency value above the preset high-frequency threshold, "medium" is the frequency value between the high-frequency threshold and the low-frequency threshold, and "low" is the frequency value below the low-frequency threshold. In the anomaly propagation characteristics dimension, the levels are divided according to the number of propagations and the strength of causal relationships: "strong" is the frequency value with a high number of propagations and a clear causal relationship in each propagation, "medium" is the frequency value with a medium number of propagations and a clear causal relationship, and "weak" is the frequency value with a low number of propagations or an unclear causal relationship.

[0190] Furthermore, the probability of failure is determined by combining the levels of the two dimensions: "high frequency + strong propagation" corresponds to "extremely high", "high frequency + medium propagation" or "medium frequency + strong propagation" corresponds to "high", "medium frequency + medium propagation" or "high frequency + weak propagation" corresponds to "medium", and the remaining combinations correspond to "low". The final evaluation result is the probability of failure of the candidate root cause node.

[0191] In summary, by identifying the functional abnormalities of the affected components, tracing back and recording the complete component-node association links in the directed fault propagation graph, the fault propagation path of the garlic chili sauce production line was finally obtained.

[0192] In summary, by compiling a list of path nodes, clarifying node functions in conjunction with equipment manuals, and reviewing maintenance records to confirm past fault associations, nodes with the potential to independently cause faults were selected, ultimately yielding candidate root cause nodes for the garlic chili sauce production line.

[0193] In summary, by fixing the complete production cycle as the evaluation cycle, comprehensively collecting fault records of the three types of nodes, calculating the basic fault frequency and adjusting it in conjunction with the maintenance cycle, the probability of failure of the candidate root cause nodes of the garlic chili sauce production line was finally obtained.

[0194] In summary, by accurately comparing probability values ​​to find the node with the highest value, and combining node relationship data with component structure drawings to confirm the associated core components, the root cause component of the garlic chili sauce production line failure was finally determined.

[0195] In summary, by combining sensor-collected parameters with manual inspection records, the status information of candidate fault root cause nodes and associated nodes is collected, ultimately yielding node operation status data of candidate fault root cause nodes and status data of associated nodes.

[0196] In summary, by establishing anomaly standards based on the equipment manual, dividing the statistical period to count the number of anomalies, calculating the frequency, and compiling the data into tables, the node anomaly frequency data of the node operating status data was finally obtained.

[0197] In summary, by setting up anomaly propagation judgment logic, comparing the time and anomaly correlation of the two types of data, recording propagation information and organizing it into a set, the anomaly propagation characteristic data of the associated node status data was finally obtained.

[0198] In summary, by setting two evaluation dimensions and classifying them into levels, and combining the level combinations to determine the probability of occurrence, the probability of failure of the candidate root cause node was finally obtained.

[0199] like Figure 2 The diagram shown is a functional block diagram of a real-time fault diagnosis system for a garlic chili sauce production line provided in an embodiment of the present invention.

[0200] The real-time fault diagnosis system 100 for a garlic chili sauce production line described in this invention can be installed in an electronic device. Depending on the functions implemented, the real-time fault diagnosis system 100 for a garlic chili sauce production line may include a data standardization module 101, a degradation rate calculation module 102, a risk calibration module 103, a fault diagnosis module 104, a propagation graph construction module 105, and a root cause analysis module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0201] In this embodiment, the functions of each module / unit are as follows:

[0202] The data standardization module 101 is used to standardize the real-time operating data of the equipment components in the garlic chili sauce production line to obtain the standard operating data of the garlic chili sauce production line.

[0203] The degradation rate calculation module 102 is used to measure the deviation between the historical performance parameters of the equipment component and the standard operating data to obtain the component degradation rate of the equipment component.

[0204] The risk calibration module 103 is used to compare and calibrate the component degradation rate with a preset health threshold to obtain the failure performance index of the equipment component.

[0205] The fault diagnosis module 104 is used to combine the time-domain features and frequency-domain features in the fault performance indicators into a feature vector of the equipment component, and to perform pattern matching between the feature vector and a preset fault feature library to obtain the fault type of the garlic chili sauce production line.

[0206] The propagation graph construction module 105 is used to construct a directed graph of fault propagation for the garlic chili sauce production line based on the connection relationship of the equipment components.

[0207] The root cause analysis module 106 is used to perform source tracing and localization analysis on the fault type and the fault propagation directed graph to obtain the root cause components of the fault in the garlic chili sauce production line.

[0208] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0209] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0210] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0211] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0212] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0213] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended 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.

Claims

1. A real-time fault diagnosis method for a garlic chili sauce production line, characterized in that, The method includes: S1. Standardize the real-time operating data of the equipment components in the garlic chili sauce production line to obtain the standard operating data of the garlic chili sauce production line; S2. Measure the deviation between the historical performance parameters of the equipment component and the standard operating data to obtain the component degradation rate of the equipment component, including: Multimodal feature evolution is performed on the multidimensional feature parameters of the device component to obtain the baseline feature matrix of the device component; Principal component analysis is performed on the baseline feature matrix to obtain the feature projection vector and variance contribution rate of the device components; The deviation of the device component is obtained by calculating the difference between the reference feature matrix and the feature projection vector, wherein the formula for calculating the deviation is as follows: ； In the formula, This indicates the degree of deviation of the device component. This represents the number of principal components in the feature projection vector. In the feature projection vector, the first... The variance contribution rate of each principal component In the feature projection vector, the first... The variance contribution rate of each principal component Describe a natural constant. In the feature projection vector, the first... The time decay of each feature This represents the preset characteristic decay time constant. In the feature projection vector, the first... The projected values ​​of each principal component. In the feature projection vector, the first... Historical mean of each principal component In the feature projection vector, the first... The variance of each principal component This represents the size of the time window for the feature projection vector; Based on the deviation, the initial degradation rate of the device component is calculated, wherein the initial degradation rate is calculated using the following formula: ； In the formula, This indicates the initial degradation rate of the device component. Describe a natural constant. This represents the preset steepness coefficient of the degradation curve. This indicates the degree of deviation of the device component. This indicates the reference deviation threshold of the device component. This represents the size of the time window for the feature projection vector. Indicates the device component number Deviation from a given time unit; Adaptive calibration is performed on the initial degradation rate to obtain the calibrated component degradation rate of the equipment component; S3. The degradation rate of the component is compared with a preset health threshold to obtain the failure performance index of the equipment component. S4. Combine the time-domain features and frequency-domain features in the fault performance indicators into a feature vector of the equipment component, and perform pattern matching between the feature vector and a preset fault feature library to obtain the fault type of the garlic chili sauce production line. S5. Based on the connection relationship of the equipment components, construct a directed graph of fault propagation for the garlic chili sauce production line; S6. Perform source tracing and localization analysis on the fault type and the directed graph of fault propagation to obtain the root cause component of the fault in the garlic chili sauce production line.

2. The real-time fault diagnosis method for a garlic chili sauce production line as described in claim 1, characterized in that, The standardization process for the real-time operating data of the equipment components in the garlic chili sauce production line, resulting in standard operating data for the garlic chili sauce production line, includes: Collect real-time operating data of equipment components in the garlic chili sauce production line; The real-time operating data is cleaned to obtain the clean operating data of the garlic chili sauce production line; The cleanroom operation data is standardized to obtain the standard operation data of the garlic chili sauce production line.

3. The real-time fault diagnosis method for a garlic chili sauce production line as described in claim 1, characterized in that, The step of comparing and calibrating the component degradation rate with a preset health threshold to obtain the failure performance index of the device component includes: The component degradation rate is compared with a preset health threshold one by one to determine the risk level of the equipment component; Based on the risk level, assess the risk of the component's degradation rate to obtain the risk assessment value of the equipment component; First, extract the real-time operating parameters of all equipment components from the real-time operating data repository, including temperature, speed, current, vibration value, etc. Based on the risk assessment value, a correlation analysis is performed on the real-time operating parameters to obtain the fault performance indicators of the equipment components.

4. The real-time fault diagnosis method for a garlic chili sauce production line as described in claim 1, characterized in that, The step involves combining the time-domain and frequency-domain features of the fault performance indicators into a feature vector for the equipment component, and then performing pattern matching between the feature vector and a preset fault feature library to obtain the fault types of the garlic chili sauce production line, including: Extract the time-domain and frequency-domain features from the fault performance indicators to obtain the time-domain and frequency-domain feature data of the fault performance indicators; The time-domain feature data and the frequency-domain feature data are dimensionally aligned to obtain the aligned feature data of the device component. The aligned feature data is matched with a preset fault feature library to obtain the fault type of the garlic chili sauce production line.

5. The real-time fault diagnosis method for a garlic chili sauce production line as described in claim 1, characterized in that, The step of constructing a directed graph of fault propagation for the garlic chili sauce production line based on the connection relationships of the equipment components includes: Obtain the device component connection relationships in the configuration information of the garlic chili sauce production line to obtain the component topology data of the garlic chili sauce production line; Based on the component topology data, the connection nodes between the equipment components are identified to obtain the node relationship data of the garlic chili sauce production line; Based on the component topology data and the node relationship data, construct a component directed graph of the garlic chili sauce production line; The fault propagation direction of the directed graph of the components is marked to generate the fault propagation directed graph of the garlic chili sauce production line.

6. The real-time fault diagnosis method for a garlic chili sauce production line as described in claim 1, characterized in that, The source tracing and localization analysis of the fault type and the directed graph of fault propagation yields the root cause components of the fault in the garlic chili sauce production line, including: Based on the fault type of the garlic chili sauce production line, identify the fault propagation path of the directed fault propagation graph; By performing reverse tracing analysis on the fault propagation path, candidate root cause nodes of the garlic chili sauce production line are obtained. The failure probability of the candidate root cause nodes is evaluated to obtain the failure occurrence probability of the candidate root cause nodes. Based on the probability of failure, the root cause component of the failure in the garlic chili sauce production line is determined.

7. The real-time fault diagnosis method for a garlic chili sauce production line as described in claim 6, characterized in that, The step of evaluating the failure probability of the candidate root cause nodes to obtain the probability of failure of the candidate root cause nodes includes: Obtain the node operation status data and associated node status data of the candidate fault root cause node; By analyzing the frequency of abnormal features in the node operation status data, the node abnormality frequency data of the node operation status data is obtained. Extract the abnormal propagation features from the state data of the associated nodes to obtain the abnormal propagation feature data in the state data of the associated nodes; By comprehensively evaluating the node anomaly frequency data and the anomaly propagation characteristic data, the probability of failure of the candidate fault root cause node is obtained.

8. A real-time fault diagnosis system for a garlic chili sauce production line, characterized in that, The system includes: The data standardization module is used to standardize the real-time operating data of the equipment components in the garlic chili sauce production line to obtain the standard operating data of the garlic chili sauce production line. The degradation rate calculation module is used to measure the deviation between the historical performance parameters of the equipment component and the standard operating data to obtain the component degradation rate of the equipment component, including: Multimodal feature evolution is performed on the multidimensional feature parameters of the device component to obtain the baseline feature matrix of the device component; Principal component analysis is performed on the baseline feature matrix to obtain the feature projection vector and variance contribution rate of the device components; The deviation of the device component is obtained by calculating the difference between the reference feature matrix and the feature projection vector, wherein the formula for calculating the deviation is as follows: ； In the formula, This indicates the degree of deviation of the device component. This represents the number of principal components in the feature projection vector. In the feature projection vector, the first... The variance contribution rate of each principal component In the feature projection vector, the first... The variance contribution rate of each principal component Describe a natural constant. In the feature projection vector, the first... The time decay of each feature This represents the preset characteristic decay time constant. In the feature projection vector, the first... The projected values ​​of each principal component. In the feature projection vector, the first... Historical mean of each principal component In the feature projection vector, the first... The variance of each principal component This represents the size of the time window for the feature projection vector; Based on the deviation, the initial degradation rate of the device component is calculated, wherein the initial degradation rate is calculated using the following formula: ； In the formula, This indicates the initial degradation rate of the device component. Describe a natural constant. This represents the preset steepness coefficient of the degradation curve. This indicates the degree of deviation of the device component. This indicates the reference deviation threshold of the device component. This represents the size of the time window for the feature projection vector. Indicates the device component number Deviation from a given time unit; Adaptive calibration is performed on the initial degradation rate to obtain the calibrated component degradation rate of the equipment component; The risk calibration module is used to compare and calibrate the component degradation rate with a preset health threshold to obtain the failure performance index of the equipment component. The fault diagnosis module is used to combine the time-domain features and frequency-domain features in the fault performance indicators into a feature vector of the equipment component, and to perform pattern matching between the feature vector and a preset fault feature library to obtain the fault type of the garlic chili sauce production line. The propagation graph construction module is used to construct a directed graph of fault propagation for the garlic chili sauce production line based on the connection relationships of the equipment components. The root cause analysis module is used to perform source tracing and localization analysis on the fault type and the directed graph of fault propagation to obtain the root cause components of the fault in the garlic chili sauce production line.

Images