Abnormal early warning system for equipment health monitoring

Abstract

The present application is applicable to the field of computer technology, and provides an abnormal warning system for equipment health monitoring, including: a data acquisition module determines the collection method by obtaining the initial operating data of the equipment to be monitored, which can ensure that the collection process is highly consistent with the characteristics of the equipment, greatly improving the pertinence of data collection, avoiding the loss or redundancy of key data due to improper collection methods, and improving the data quality of the target operating data collected. Afterwards, the data analysis module analyzes the target operating data based on the current working conditions and the current environment, which can comprehensively and truly reflect the health status of the equipment, making the health monitoring results more reliable and practical. At the same time, the data analysis module outputs warning information in a timely manner when it detects that the health monitoring results are abnormal, so that equipment management personnel and maintenance personnel can promptly understand the abnormal situation of the equipment and take corresponding measures to protect the equipment and prevent further deterioration of the fault.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an abnormality warning system for equipment health monitoring. Background Art

[0002] In modern industrial production, mechanical equipment (such as motors, pumps, and fans) is critical for power transmission and fluid transport, and is widely used in numerous industries, including chemical, electric power, metallurgy, and mining. The stable operation of these equipment is directly related to the continuity and stability of the entire production process. Therefore, real-time health monitoring of these equipment is essential. However, existing technologies typically simply input collected equipment data into deep learning models for processing to generate health monitoring results. This is incomplete and fails to meet practical needs. Summary of the Invention

[0003] The embodiment of the present application provides an abnormal warning system for equipment health monitoring, which can solve the problem that the existing technology is not comprehensive enough and cannot meet actual needs.

[0004] In a first aspect, an embodiment of the present application provides an abnormality warning system for equipment health monitoring, comprising:

[0005] a data acquisition module, configured to acquire initial operating data of the device to be monitored, determine an acquisition method for the device to be monitored based on the initial operating data, and acquire target operating data of the device to be monitored based on the acquisition method;

[0006] The data analysis module is connected to the data acquisition module and is used to analyze the target operating data based on the current working conditions and the current environment to obtain the health monitoring results of the equipment to be monitored, and output warning information when the health monitoring results are detected to be abnormal.

[0007] Optionally, the data acquisition module is specifically used to:

[0008] Acquiring the initial operating data and determining abnormal points in the initial operating data;

[0009] Performing anomaly detection on the abnormal point based on a trend recognition algorithm to obtain an abnormality level of the initial operation data; the trend recognition algorithm refers to an algorithm for analyzing the trend of abnormal points changing with set variables;

[0010] determining the collection method based on the abnormality level;

[0011] Data is collected from the device to be monitored based on the collection method to obtain the target operating data.

[0012] Optionally, determining the collection mode based on the abnormality level includes:

[0013] If the abnormality level is the first level, the collection mode is determined to be a numerical value collection mode; the numerical value collection mode refers to collecting only the operating value of the device to be monitored;

[0014] If the abnormality level is the second level, the acquisition mode is determined to be the additional waveform acquisition mode; the additional waveform acquisition refers to additional acquisition of waveform data under the numerical acquisition mode; the abnormality level of the second level is higher than the abnormality level of the first level;

[0015] If the abnormality level is the third level, the acquisition mode is determined to be a multi-point synchronous waveform acquisition mode; the multi-point synchronous waveform acquisition means that under the numerical acquisition mode, multiple measurement points of the device to be monitored are added to simultaneously acquire waveforms; the abnormality degree of the third level is higher than the abnormality degree of the second level.

[0016] Optionally, the data acquisition module is further used to:

[0017] determining a density type of the target operating data;

[0018] Analyze the target operation data based on the density type to obtain indicator data corresponding to the monitoring device;

[0019] The target operation data and the indicator data are sent to the data analysis module.

[0020] Optionally, analyzing the target operation data based on the density type to obtain indicator data corresponding to the monitoring device includes:

[0021] If the density type is the first density type, a traditional waveform indicator corresponding to the device to be monitored is calculated based on the target operating data; the first density type is used to describe that the target operating data is low-density data, and the low density is used to characterize that the target operating data contains a small amount of information per unit time, and the traditional waveform indicator is used to describe signal characteristics;

[0022] If the density type is the second density type, the traditional waveform indicator and the set fault indicator corresponding to the monitored equipment are calculated based on the target operation data; the second density type is used to describe that the target operation data is high-density data, and the high density is used to characterize that the target operation data contains a large amount of information per unit time; the set fault indicator is different from the traditional waveform indicator.

[0023] Optionally, the data analysis module is specifically used to:

[0024] Extracting features from the target operating data based on the current operating condition and the current environment to obtain feature information of the device to be monitored;

[0025] Analyzing the characteristic information based on the constructed equipment performance degradation knowledge graph to obtain performance degradation information of the equipment to be monitored;

[0026] generating the health monitoring result based on the performance degradation information;

[0027] If the health monitoring result is abnormal, the warning information is output.

[0028] Optionally, extracting features from the target operating data based on the current operating condition and the current environment to obtain feature information of the device to be monitored includes:

[0029] determining a dictionary learning algorithm corresponding to the target operating data based on the current environment and the current operating condition;

[0030] Feature extraction is performed on the target operation data based on the dictionary learning algorithm to obtain the feature information.

[0031] Optionally, before analyzing the characteristic information based on the constructed device performance degradation knowledge graph to obtain the performance degradation trend information of the device to be monitored, the method further includes:

[0032] Obtain a historical operation data set of the device to be monitored; the historical operation data set includes historical data of different data types when the device to be monitored is continuously operating under different working conditions and different environments, and historical data of different data types when the device to be monitored fails under different working conditions and different environments;

[0033] Extracting features from each historical data in the historical operation data set to obtain a plurality of historical feature parameters;

[0034] Analyzing the plurality of characteristic parameters to obtain a data degradation trajectory of the device to be monitored;

[0035] Performing correlation analysis on each historical data of the different data types to obtain a degree of interaction between each historical data of the different data types;

[0036] Based on the historical operation data set, the data degradation trajectory and the degree of interaction, the equipment performance degradation knowledge graph is constructed.

[0037] Optionally, the performance degradation information includes a plurality of the data degradation trajectories and a degree of interaction between different operating data; and generating the health monitoring result based on the performance degradation information includes:

[0038] Determining a plurality of health monitoring data based on the device type of the device to be monitored;

[0039] Calculating weight values ​​of the plurality of health monitoring data based on the degree of interaction between the different operating data;

[0040] Determining a target degradation trajectory corresponding to each of the plurality of health monitoring data from the plurality of data degradation trajectories;

[0041] The health monitoring result is generated based on the weight values ​​of the respective plurality of health monitoring data and the target degradation trajectory.

[0042] Optionally, the data analysis module is further configured to:

[0043] Performing time domain analysis on the target operation data to obtain statistical features corresponding to the target operation data;

[0044] performing a correlation analysis on the target operation data and the indicator data to obtain a correlation degree between the target operation data and the indicator data;

[0045] Inputting the statistical features and the indicator data into a health monitoring model for processing respectively, to obtain a first monitoring result corresponding to the target operating data and a second monitoring result of the indicator data;

[0046] Based on the degree of correlation between the target operating data and the indicator data, the current operating conditions and the current environment, the first monitoring result and the second monitoring result are adjusted to obtain the health monitoring result.

[0047] The abnormal warning system for equipment health monitoring provided by the embodiment of the present application has the following beneficial effects compared with the prior art: the data acquisition module determines the acquisition method by obtaining the initial operating data of the equipment to be monitored, which can ensure that the acquisition process is highly consistent with the characteristics of the equipment, greatly improving the pertinence of data acquisition, avoiding the loss or redundancy of key data due to improper acquisition methods, and improving the data quality of the target operating data collected. The data analysis module analyzes the target operating data based on the current working conditions and the current environment, which can comprehensively and truly reflect the health status of the equipment, making the health monitoring results more reliable and practical. At the same time, the data analysis module outputs warning information in a timely manner when it detects that the health monitoring results are abnormal, so that equipment management personnel and maintenance personnel can promptly understand the abnormal conditions of the equipment and take corresponding measures to protect the equipment and prevent further deterioration of the fault. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0049] Figure 1 This is a schematic diagram of the structure of an abnormal warning system for equipment health monitoring provided by an embodiment of the present application;

[0050] Figure 2 This is a structural diagram of an abnormal warning system for equipment health monitoring provided by another embodiment of the present application;

[0051] Figure 3 This is a specific implementation flow chart of the data acquisition module in the abnormal warning system for equipment health monitoring provided by one embodiment of the present application;

[0052] Figure 4 This is a specific implementation flow chart of the data acquisition module in the abnormal warning system for equipment health monitoring provided by another embodiment of the present application;

[0053] Figure 5 This is a flowchart of a specific implementation of a data analysis module in an abnormal warning system for equipment health monitoring provided by an embodiment of the present application;

[0054] Figure 6 This is a specific implementation flowchart of the data analysis module in the abnormal warning system for equipment health monitoring provided by another embodiment of the present application. DETAILED DESCRIPTION

[0055] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0056] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0057] It will also be understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0058] As used in this specification and the appended claims, the term "if" can be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0059] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0060] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present application. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more but not all embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0061] See also Figure 1 , Figure 1 This is a schematic diagram of the structure of an abnormal warning system for equipment health monitoring provided by an embodiment of the present application. For ease of explanation, only the parts related to this embodiment are shown, which are detailed as follows:

[0062] like Figure 1 As shown, the abnormal warning system 1 for equipment health monitoring includes: a data acquisition module 11 and a data analysis module 12, which are communicatively connected to each other. The above communication connection mode can be a wired communication connection or a wireless communication connection, which is not limited here.

[0063] Specifically, the data acquisition module 11 is used to obtain initial operating data of the device to be monitored, determine an acquisition method for the device to be monitored based on the initial operating data, and acquire target operating data of the device to be monitored based on the acquisition method.

[0064] The data analysis module 12 is used to analyze the target operating data based on the current working conditions and the current environment to obtain the health monitoring results of the equipment to be monitored, and output warning information when the health monitoring results are detected to be abnormal.

[0065] It should be noted that the equipment to be monitored includes but is not limited to motors, pumps, fans and other mechanical equipment.

[0066] In the embodiment of the present application, the data acquisition module 11 can continuously and in real time acquire the initial operating data of the monitored device during operation, wherein the initial operating data includes but is not limited to: temperature data, vibration data, sound data and other data.

[0067] In one implementation of the present application, please refer to Figure 2 , Figure 2 This is a structural diagram of an abnormal warning system for equipment health monitoring provided by another embodiment of the present application. Figure 2 As shown, the abnormal warning system 1 for equipment health monitoring may further include a data sensing module 13 . The data sensing module 13 is in communication connection with the data acquisition module 11 .

[0068] In this embodiment, the data sensing module 13 is used to sense in real time the initial operation data generated by the device to be monitored during operation, and send the initial operation data to the data acquisition module 11 .

[0069] Based on this, the data acquisition module 11 can obtain the initial operating data of the device to be monitored in real time through the data perception module 13 set at the device to be monitored.

[0070] It should be noted that the data sensing module 13 can be installed on the bearing seat surface of the device to be monitored or other device surface that needs to be monitored and is easy to install. Among them, the installation methods of the data sensing module 13 include but are not limited to: a method of combined action of magnetism and adhesion, and a method of fixing by punching.

[0071] In practical applications, the data sensing module 13 can be various types of sensors, such as a temperature sensor, a vibration sensor, and a sound sensor.

[0072] In the embodiment of the present application, since the fault types of the equipment to be monitored include but are not limited to slow-changing types, fast-changing types, and sudden changes, in order to accurately obtain the operating data corresponding to each type of fault, the data acquisition module 11 can determine the acquisition frequency of the initial operating data corresponding to each fault type according to each fault type, so that the data acquisition module 11 obtains the corresponding initial operating data based on each acquisition frequency.

[0073] It should be noted that slowly changing faults refer to equipment failures that develop slowly, with characteristic parameters changing gradually over time. These faults typically only become apparent over a long period of time (e.g., months or years). Examples of slowly changing faults include long-term wear and tear and aging of equipment components.

[0074] Fast-changing faults are those that change relatively quickly, with characteristic parameters exhibiting significant changes within a short period of time (e.g., minutes to hours). Examples include intermittent faults within the equipment, sensor failure, or transient anomalies in the control system.

[0075] Sudden failures occur suddenly during equipment operation. Fault characteristic parameters change dramatically in an instant (almost zero time interval), causing the equipment's performance or status to change from normal to abnormal or even complete failure. Examples include severe external impact, sudden breakage of key internal components, and electrical short circuits.

[0076] In an embodiment of the present application, the collection frequency of initial operating data corresponding to a slowly changing type of fault is lower than the collection frequency of initial operating data corresponding to a rapidly changing type of fault, and the collection frequency of initial operating data corresponding to a rapidly changing type of fault is lower than the collection frequency of initial operating data corresponding to a sudden change type of fault.

[0077] In the embodiment of the present application, after acquiring the initial operating data of the device to be monitored, the data acquisition module 11 can calculate the statistical characteristics (such as the mean, standard deviation, and peak value) corresponding to each operating cycle of the initial operating data and determine the operating status of the device to be monitored at that time based on the statistical characteristics. The operating cycle can be determined based on actual needs and is not limited here.

[0078] Specifically, the data collection module 11 can input the corresponding statistical features in each operation cycle into the trained analysis model for processing to obtain data change information corresponding to the initial operation data, wherein the data change information is used to describe the change pattern of the initial operation data.

[0079] It should be noted that the above-mentioned analysis model can be obtained by training a pre-built neural network model based on a preset sample set. Among them, each sample data in the preset sample set includes the statistical characteristics of the sample operation data and the sample data change information corresponding to the statistical characteristics of the sample operation data. When training the pre-built neural network model, the statistical characteristics of the sample operation data in each sample data are used as the input of the neural network model, and the sample data change information corresponding to the statistical characteristics of the sample operation data in each sample data is used as the output of the neural network model. Through training, the neural network model can learn the correspondence between the statistical characteristics of all possible sample operation data and the sample data change information, and use the trained neural network model as the analysis model.

[0080] Afterwards, the data acquisition module 11 may determine the operating status of the device to be monitored based on the data change information.

[0081] Specifically, when the data acquisition module 11 detects that the data fluctuations of the initial operating data are small and the changes are smooth, it can be determined that the operating state of the equipment to be monitored is stable operation; when the data acquisition module 11 detects that the data fluctuations of the initial operating data are large and irregular, it can be determined that the operating state of the equipment to be monitored is unstable operation.

[0082] In the embodiment of the present application, the data collection module 11 can determine the collection method of the operating data of the device to be monitored according to the operating status of the device to be monitored.

[0083] Specifically, when the data acquisition module 11 detects that the operating state of the monitored device is stable, it indicates that the probability of the monitored device experiencing an abnormality is low. Therefore, the data acquisition module 11 can determine that the data acquisition method for the monitored device is to collect data based on a first acquisition frequency. The first acquisition frequency can be determined based on actual needs and is not limited here.

[0084] When the data acquisition module 11 detects that the operating state of the monitored device is unstable, it indicates that the monitored device is likely to be abnormal. Therefore, the data acquisition module 11 can determine that the data acquisition method for the monitored device is to collect data based on the second set frequency. The second collection frequency can be determined based on actual needs and is not limited here.

[0085] It should be noted that the second acquisition frequency is higher than the first acquisition frequency.

[0086] Based on this, the data acquisition module 11 can obtain the target operating data of the device to be monitored according to the determined acquisition mode set, and send the target operating data to the data analysis module 12. The target operating data includes but is not limited to: temperature data, vibration data, and sound data.

[0087] In the embodiment of the present application, after obtaining the target operation data, the data analysis module 12 may perform data preprocessing on the target operation data to improve the data quality. The data preprocessing includes but is not limited to data cleaning and data normalization.

[0088] Afterwards, the data analysis module 12 may analyze the target operating data that has undergone data preprocessing based on the current working conditions and the current environment to obtain the health monitoring results of the equipment to be monitored.

[0089] The current working condition includes but is not limited to the workload and working mode of the equipment to be monitored. The current specifically refers to the moment when the data acquisition module 11 acquires the target operating data.

[0090] In the embodiment of the present application, the data analysis module 12 can input the target operating data into the trained health analysis model for processing to obtain the initial monitoring results of the device to be monitored, wherein the initial monitoring results include but are not limited to normal and abnormal.

[0091] Normal is used to describe that the monitored device is in a healthy state, and abnormal is used to describe that the monitored device is in an abnormal state.

[0092] It should be noted that the health analysis model can be obtained by training a pre-built first deep learning model based on a preset sample set. Each sample data in the preset sample set includes sample operation data and the sample monitoring results corresponding to the sample operation data. When training the pre-built first deep learning model, the sample operation data in each sample data is used as the input of the first deep learning model, and the sample monitoring results corresponding to the sample operation data in each sample data are used as the output of the first deep learning model. Through training, the first deep learning model can learn the correspondence between all possible sample operation data and sample monitoring results, and the trained first deep learning model is used as the health analysis model.

[0093] In actual applications, since the temperature, vibration, and other data ranges of the monitored equipment during normal operation may increase accordingly when the equipment is under high load, in an embodiment of the present application, in order to improve the accuracy of health monitoring of the monitored equipment, the data analysis module 12 can obtain a historical operating data set of the monitored equipment and construct a correlation model between the operating conditions, environment, and health monitoring results based on the historical operating data set, such as a multivariate linear regression model. The historical operating data set includes historical data of different data types and historical monitoring results of the monitored equipment under different operating conditions and environments when the monitored equipment continues to operate under different operating conditions and environments.

[0094] It should be noted that the multiple linear regression model is a statistical analysis method used to study the linear relationship between multiple independent variables and a dependent variable. Its core is to quantify the influence of variables through mathematical models and evaluate the effectiveness of the model based on hypothesis testing.

[0095] Afterwards, the data analysis module 12 can simultaneously input the initial monitoring results, current working conditions and current environment into the above-mentioned correlation model for fusion analysis to obtain predicted monitoring results, and adjust the initial monitoring results based on the predicted monitoring results to obtain the final health monitoring results.

[0096] In the embodiment of the present application, health monitoring results include but are not limited to normal and abnormal.

[0097] Normal is used to describe that the monitored device is in a healthy state, and abnormal is used to describe that the monitored device is in an abnormal state.

[0098] It should be noted that when the health monitoring result is abnormal, the health monitoring result may also carry the abnormality type and abnormality cause. The abnormality type includes but is not limited to: sub-health, failure, and danger.

[0099] Subhealthy is used to describe equipment that is experiencing early to mid-term damage, but is stable and can continue to operate. Faulty is used to describe equipment that is experiencing mid to late-term damage or is rapidly deteriorating, requiring short-term monitoring. Dangerous is used to describe equipment that is severely damaged, creating a risk of continued operation and requiring immediate shutdown for inspection and repair.

[0100] Based on this, when the data analysis module 12 detects that the health monitoring result of the device to be monitored is abnormal, it can output early warning information including the abnormality type and the cause of the abnormality.

[0101] From the above, it can be seen that the abnormal warning system for equipment health monitoring provided by this embodiment determines the collection method by obtaining the initial operating data of the equipment to be monitored through the data collection module, which can ensure that the collection process is highly consistent with the characteristics of the equipment, greatly improves the pertinence of data collection, avoids the loss or redundancy of key data due to improper collection methods, and improves the data quality of the target operating data collected. The data analysis module analyzes the target operating data based on the current working conditions and the current environment, which can comprehensively and truly reflect the health status of the equipment, making the health monitoring results more reliable and practical. Finally, the early warning module outputs early warning information in a timely manner when it detects that the health monitoring results are abnormal, so that equipment management personnel and maintenance personnel can promptly understand the abnormal conditions of the equipment and take corresponding measures to protect the equipment and prevent further deterioration of the fault.

[0102] See also Figure 3 , Figure 3 This is a specific implementation flow chart of the data acquisition module in the abnormal warning system for equipment health monitoring provided by an embodiment of the present application. Figure 3 As shown, the data acquisition module can specifically acquire target operation data by executing steps S101 to S104, as detailed below:

[0103] In S101 , the initial operating data is acquired, and abnormal points of the initial operating data are determined.

[0104] In S102, anomaly detection is performed on the abnormal point based on a trend identification algorithm to obtain an abnormality level of the initial operation data; the trend identification algorithm is an algorithm for analyzing a trend of an abnormal point changing with a set variable.

[0105] In S103, the collection method is determined based on the abnormality level.

[0106] In S104, data is collected from the device to be monitored based on the collection method to obtain the target operation data.

[0107] In this embodiment, after obtaining the initial operating data, the data acquisition module 11 can construct data curves showing the time-varying changes of different types of data based on the data types, since the initial operating data includes multiple different types of data (such as temperature data, vibration data, and sound data). The data acquisition module 11 can then identify data points in the different types of data curves that exceed their corresponding set thresholds or exhibit sudden changes as abnormal points, thereby obtaining the abnormal points corresponding to the initial operating data. The set thresholds for different types of data can be determined based on actual needs and are not limited herein.

[0108] After obtaining anomalies in the initial operating data, the data acquisition module 11 can perform anomaly detection on the anomalies based on a trend recognition algorithm. Specifically, the algorithm determines the trend characteristics of the anomalies, thereby determining the anomaly level of the initial operating data. The trend recognition algorithm is an algorithm used to analyze the trend of anomalies as they change with a set variable. The set variable can be time.

[0109] In practical applications, trend identification algorithms include but are not limited to: exponentially weighted moving average method, Holt-Winters method, and Kalman filter algorithm.

[0110] In this embodiment, the trend feature is used to describe the degree of deviation between an abnormal point and a set value, including but not limited to: mild abnormality (i.e., mild deviation), moderate abnormality (i.e., moderate deviation), and severe abnormality (i.e., severe deviation). The set value can be determined based on actual needs and is not limited here.

[0111] In some possible embodiments, the set value may be determined based on the average value of a data group consisting of a set window size with the outlier as the center point. The set window size may be determined based on actual needs and is not limited here.

[0112] It should be noted that the abnormality levels include but are not limited to the first level, the second level, and the third level. The second level has a higher abnormality level than the first level, and the third level has a higher abnormality level than the second level.

[0113] In this embodiment, the data acquisition module 11 can set the abnormality level of the abnormal point with a trend characteristic of mild abnormality to the first level, the abnormality level of the abnormal point with a trend characteristic of moderate abnormality to the second level, and the abnormality level of the abnormal point with a trend characteristic of severe abnormality to the third level.

[0114] In this embodiment, after determining the abnormality level of the abnormal point, the data collection module 11 may determine a collection method for the operating data of the device to be monitored according to the abnormality level.

[0115] It should be noted that acquisition methods include, but are not limited to, numerical acquisition, waveform acquisition, and simultaneous waveform acquisition at multiple measurement points. The numerical acquisition method involves acquiring only the operating values ​​of the monitored device, i.e., the actual values ​​corresponding to the operating data. The waveform acquisition method involves acquiring additional waveform data from the monitored device under the numerical acquisition method. The simultaneous waveform acquisition method involves acquiring additional waveform data from multiple measurement points of the monitored device under the numerical acquisition method.

[0116] In this embodiment, when the data acquisition module 11 detects that the abnormality level of the abnormal point is the first level, it means that the monitored device has a slight abnormality at this time, but can still continue to run. Therefore, in order to avoid occupying memory, the data acquisition module 11 can determine that the collection method for the monitored device at this time is a numerical collection method.

[0117] When the data acquisition module 11 detects the second level of abnormality, it indicates that the monitored device has a moderate abnormality. Although it can continue to operate, it may be prone to failure. Therefore, in order to improve the accuracy of health monitoring of the monitored device, the data acquisition module 11 can determine that the acquisition method is the waveform acquisition method.

[0118] When the data acquisition module 11 detects the third level of abnormality, it indicates that the monitored equipment has experienced a severe abnormality, and continued operation is risky, requiring prompt shutdown and inspection. Therefore, to improve the accuracy of health monitoring of the monitored equipment and facilitate subsequent maintenance personnel, the data acquisition module 11 can select a multi-point simultaneous waveform acquisition method.

[0119] In this embodiment, after determining the collection method for the operating data of the device to be monitored, the data collection module 11 may directly collect data from the device to be monitored according to the collection method to obtain the final target operating data.

[0120] As can be seen from the above, the data acquisition module provided by this embodiment obtains the initial operating data and determines the abnormal points of the initial operating data; performs abnormality detection on the abnormal points based on the trend recognition algorithm to obtain the abnormality level of the initial operating data; the trend recognition algorithm refers to an algorithm used to analyze the trend of abnormal points changing with set variables; determines the acquisition method based on the abnormality level; and collects data from the monitored equipment based on the acquisition method to obtain the target operating data. This embodiment can dynamically adjust the acquisition method according to the actual status of the equipment by detecting abnormal points in the initial operating data and dividing them into abnormality levels. At the same time, the abnormality level is determined based on the trend recognition algorithm, fully considering the changing trend of abnormal points with set variables. Compared with single data judgment, it can more comprehensively and accurately evaluate the degree of equipment abnormality.

[0121] See also Figure 4 , Figure 4 This is a specific implementation flow chart of the data acquisition module in the abnormal warning system for equipment health monitoring provided by another embodiment of the present application. Figure 4 As shown, after obtaining the target operation data, the data acquisition module may further execute steps S201 to S203 to acquire the target operation data, as detailed below:

[0122] In S201 , the density type of the target operating data is determined.

[0123] In S202, the target operation data is analyzed based on the density type to obtain indicator data corresponding to the monitoring device.

[0124] In S203, the target operation data and the indicator data are sent to the data analysis module.

[0125] It should be noted that the target operating data includes data of various types (such as temperature, vibration, and sound, etc.).

[0126] In this embodiment, the data acquisition module 11 can record the collection frequencies corresponding to data of multiple different data types while collecting the target operation data. Therefore, the data acquisition module 11 can determine the density types of data of multiple different data types in the target operation data based on the collection frequencies of data of multiple different data types in the target operation data.

[0127] The density type includes, but is not limited to, a first density type and a second density type. The first density type is used to describe low-density target operating data, where low density indicates that the target operating data contains a small amount of information per unit time. The second density type is used to describe high-density target operating data, where high density indicates that the target operating data contains a large amount of information per unit time.

[0128] It's understandable that "small amounts of information" refer to data collected at a low frequency, resulting in a relatively small amount of data collected per unit time. This data can only reveal relatively limited details and changes in the operating status of the equipment being monitored, hence the term "low-density data." For example, if data for a certain data type is collected only every 10 minutes, with each collection containing only a simple numerical value or a few parameters, then within a unit of time (e.g., an hour), only six data points of that data type are obtained. Each data point carries limited information, making it impossible to fully and meticulously describe the operating status of the equipment being monitored. This constitutes low-density data containing a small amount of information.

[0129] "Large amounts of information" refer to the large number of data points collected per unit time when the data collection frequency is high. Each data point may contain multi-dimensional parameters or complex content. This data can more comprehensively and accurately reflect the various operating states and changes of the monitored equipment, that is, it contains a large amount of information. For example, for another data type, 100 data points can be collected per second, each containing multiple parameters such as temperature, humidity, and pressure. In one hour, there will be 360,000 data points. This large number of data points can present a variety of information about the target's operation in detail, which is considered high-density data containing a large amount of information.

[0130] For any data type in the target operating data, the data acquisition module 11 may determine that the density type of the data of the data type is a first density type when it detects that the acquisition frequency of the data of the data type is less than or equal to a first threshold. The data acquisition module 11 may determine that the density type of the data of the data type is a second density type when it detects that the acquisition frequency of the data of the data type is greater than the first threshold. The first threshold can be determined based on actual needs and is not limited herein.

[0131] In this embodiment, after determining the density type of the target operating data, the data acquisition module 11 can analyze the target operating data based on the density type to obtain indicator data corresponding to the monitoring device. The indicator data includes but is not limited to traditional waveform indicators and set fault indicators.

[0132] It should be noted that traditional waveform indicators are used to describe signal characteristics, such as peak value, valley value, average value, period, phase, etc.

[0133] The set fault indicators include but are not limited to bearing fault indicators and looseness fault indicators.

[0134] Bearing fault indicators include but are not limited to: vibration indicators (such as acceleration, velocity, and displacement) and temperature indicators (such as bearing temperature and temperature change rate).

[0135] Looseness fault indicators include but are not limited to: vibration amplitude and vibration frequency, etc.

[0136] In this embodiment, when the data acquisition module 11 detects that the density type of the target operating data is the first density type, it means that its collection frequency is low, that is, the probability of an abnormality in the monitored equipment is small at this time, so there is no need to frequently collect the operating data of the monitored equipment. Therefore, in order to improve work efficiency, the data acquisition module 11 only needs to calculate the traditional waveform indicators corresponding to the monitored equipment based on the target operating data.

[0137] When the data acquisition module 11 detects that the density type of the target operating data is the second density type, it means that its collection frequency is relatively high. That is to say, the probability of abnormality in the monitored equipment is relatively high at this time, so it is necessary to frequently collect the operating data of the monitored equipment. Therefore, in order to improve the accuracy of health monitoring, the data acquisition module 11 can calculate the traditional waveform indicators and set fault indicators corresponding to the monitored equipment based on the target operating data.

[0138] In this embodiment, after obtaining the target operating data and indicator data, the data acquisition module 11 can send the target operating data and indicator operating data to the data analysis module 12, so that the data analysis module 12 can obtain more accurate health monitoring results based on the target operating data and indicator data.

[0139] From the above, it can be seen that after obtaining the target operation data, the data acquisition module provided in this embodiment can determine the density type of the target operation data; analyze the target operation data based on the density type to obtain the indicator data corresponding to the monitoring equipment; and send the target operation data and the indicator data to the data analysis module, thereby improving the accuracy of the health monitoring results obtained by the subsequent data analysis module in combination with the target operation data and the indicator data.

[0140] See also Figure 5 , Figure 5 This is a specific implementation flow chart of the data analysis module in the abnormal warning system for equipment health monitoring provided by an embodiment of the present application. Figure 5 As shown, the data analysis module can specifically output warning information by executing steps S301 to S304, which are detailed as follows:

[0141] In S301 , feature extraction is performed on the target operating data based on the current operating condition and the current environment to obtain feature information of the device to be monitored.

[0142] In this embodiment, the data analysis module 12 can determine a feature extraction method that matches the environment of the device to be monitored based on the current environment. The feature extraction methods include, but are not limited to, feature extraction methods based on wavelet transform and feature extraction methods based on short-time Fourier transform.

[0143] The data analysis module 12 can determine, based on the current working condition, a set feature that matches the working condition of the device to be monitored. The set feature can be determined based on actual needs and is not limited here.

[0144] Therefore, in this embodiment, the data analysis module 12 can extract set features that match the current working conditions from the target operating data according to a feature extraction method that matches the current environment, thereby obtaining feature information of the device to be monitored.

[0145] In one embodiment of the present application, the data analysis module may further perform the following steps S301, which are described in detail as follows:

[0146] determining a dictionary learning algorithm corresponding to the target operating data based on the current environment and the current operating condition;

[0147] Feature extraction is performed on the target operation data based on the dictionary learning algorithm to obtain the feature information.

[0148] In this embodiment, the data analysis module 12 pre-stores the correspondence between different operating conditions, different environments, and different dictionary learning algorithms. Therefore, based on the current environment and current operating conditions of the monitored equipment, as well as these correspondences, the data analysis module 12 can determine the dictionary learning algorithm corresponding to the target operating data. Subsequently, the data analysis module 12 can extract features from the target operating data using the dictionary learning algorithm to obtain feature information.

[0149] For example, assuming that the equipment to be monitored is in a situation where the working conditions are changing, the data analysis module 12 may determine a dictionary learning algorithm that can quickly adapt to data changes.

[0150] In practical applications, dictionary learning algorithms include, but are not limited to, online dictionary learning algorithms and sparse coding-based dictionary learning algorithms. Online dictionary learning algorithms, with their ability to update dictionaries in real time, are suitable for scenarios where data is constantly changing or operating conditions are unstable, and can quickly adapt to new data features. Sparse coding-based dictionary learning algorithms emphasize data sparsity and learn dictionaries by minimizing sparse representation errors. They are suitable for scenarios where sparse data features need to be extracted.

[0151] In S302, the characteristic information is analyzed based on the constructed equipment performance degradation knowledge graph to obtain performance degradation information of the equipment to be monitored.

[0152] In this embodiment, the equipment performance degradation knowledge graph includes information such as failure modes, performance indicators, operating environment factors, and their interrelationships. For example, the equipment performance degradation knowledge graph records that steel rolling equipment bearings are prone to wear failures under high temperature and high humidity conditions, and that bearing wear can lead to performance changes such as increased vibration amplitude and temperature rise.

[0153] In one embodiment of the present application, the data analysis module 12 may construct a device performance degradation knowledge graph according to the following steps, as detailed below:

[0154] Obtain a historical operation data set of the device to be monitored; the historical operation data set includes historical data of different data types when the device to be monitored is continuously operating under different working conditions and different environments, and historical data of different data types when the device to be monitored fails under different working conditions and different environments;

[0155] Extracting features from each historical data in the historical operation data set to obtain a plurality of historical feature parameters;

[0156] Analyzing the plurality of characteristic parameters to obtain a data degradation trajectory of the device to be monitored;

[0157] Performing correlation analysis on each historical data of the different data types to obtain a degree of interaction between each historical data of the different data types;

[0158] Based on the historical operation data set, the data degradation trajectory and the degree of interaction, the equipment performance degradation knowledge graph is constructed.

[0159] In this embodiment, after obtaining the historical operation data set, the data analysis module 12 can perform time domain feature extraction on any historical data in the historical operation data set, that is, calculate the time domain feature parameters of any historical data. The time domain feature parameters include but are not limited to: mean and peak value.

[0160] The data analysis module 12 can then convert the target operating data into frequency domain data using frequency domain analysis methods (such as Fourier transform and wavelet transform) and extract the corresponding frequency domain characteristic parameters. The frequency domain characteristic parameters include, but are not limited to, power spectrum density, harmonic frequency amplitude and phase, etc.

[0161] Afterwards, the data analysis module 12 may also obtain characteristic parameters of the target operating data in the time-frequency joint domain according to a time-frequency analysis method (such as short-time Fourier transform, wavelet packet transform, etc.).

[0162] Based on this, the data analysis module 12 may determine the above-mentioned time domain characteristic parameters, frequency domain characteristic parameters, and characteristic parameters of the time-frequency joint domain as a plurality of historical characteristic parameters.

[0163] In this embodiment, the data analysis module 12 can divide the aforementioned multiple historical characteristic parameters according to the collection time, operating conditions, and fault occurrence time corresponding to each historical data of the monitored equipment to obtain a first data set of different equipment states. Among them, the equipment state includes: normal operation, early failure, medium-term failure, and serious failure).

[0164] Exemplarily, the data analysis module 12 may group the historical characteristic parameters of the monitored equipment during a period of stable operation into a group and mark it as "normal operation"; and group the historical characteristic parameters of the monitored equipment during a period of time before a failure occurs into a group and mark it as "early failure".

[0165] Afterwards, for the historical data of the same data type in the first data set of any different device status, the data analysis module 12 can input the characteristic parameters of the same data type into the trained degradation analysis model for analysis to obtain the data degradation trajectory corresponding to the characteristic parameters of the same data type.

[0166] Based on this, the data analysis module 12 can obtain multiple data degradation trajectories corresponding to the historical operation data set, and determine all of the above data degradation trajectories as the data degradation trajectories of the equipment to be monitored.

[0167] It should be noted that the degradation analysis model can be obtained by training a pre-constructed second deep learning model based on a preset sample set. Each sample data in the preset sample set includes a sample feature parameter and a sample degradation trajectory corresponding to the sample feature parameter. When training the pre-constructed second deep learning model, the sample feature parameter in each sample is used as the input of the second deep learning model, and the sample degradation trajectory corresponding to the sample feature parameter in each sample is used as the output of the second deep learning model. Through training, the second deep learning model can learn the correspondence between all possible sample feature parameters and sample degradation trajectories, and the trained second deep learning model is used as the degradation analysis model.

[0168] In this embodiment, the data analysis module 12 may divide the historical operation data set into second data sets of different data types, wherein each second data set only includes historical data of the same data type.

[0169] Afterwards, the data analysis module 12 can perform a correlation analysis on each historical data of different data types, i.e., each second data set of different data types, according to a correlation analysis method to obtain a degree of interaction between each historical data of different data types. The degree of interaction is used to describe the degree of association between each historical data of different data types.

[0170] In practical applications, correlation analysis methods include but are not limited to: Pearson correlation coefficient, Spearman rank correlation coefficient, and mutual information.

[0171] It should be noted that when the correlation analysis method is the Pearson correlation coefficient, the degree of interaction is the actual value corresponding to the Pearson correlation coefficient; when the correlation analysis method is the Spearman rank correlation coefficient, the degree of interaction is the actual value corresponding to the Spearman rank correlation coefficient; when the correlation analysis method is the mutual information, the degree of interaction is the actual value corresponding to the mutual information.

[0172] In this embodiment, the data analysis module 12 can determine the basic framework of the knowledge graph, including nodes and edges. Nodes may include various equipment components of the equipment to be monitored (such as bearings, gears, motors, etc.), failure modes (such as wear, fatigue, fracture, etc.), performance indicators (such as vibration amplitude, current effective value, temperature, etc.), operating environment factors (such as load, ambient temperature, humidity, etc.), and time nodes, etc. Edges are used to represent the relationship between different nodes, such as "leads to" (indicating the causal relationship between the failure mode and the performance indicator), "influence" (indicating the influence relationship between the operating environment factor and the equipment component or performance indicator), "association" (indicating the correlation relationship between different performance indicators), and "occurs in" (indicating the temporal relationship between the failure mode and the time node).

[0173] Afterwards, the data analysis module 12 can extract nodes and edges of the knowledge graph from the historical operating data set of the equipment to be monitored, the data degradation trajectory, and the degree of interaction between each historical data of different data types. For nodes, equipment components, failure modes, performance indicators, working environment factors, etc. are added as entity nodes to the knowledge graph, and each node is assigned a unique identifier and related attributes (such as the model of the equipment component, the description of the failure mode, the numerical range of the performance indicator, etc.). For edges, based on the data degradation trajectory and the degree of interaction, the relationship between different nodes is determined, and the corresponding edges are added. For example, based on the data degradation trajectory, it is found that bearing wear causes an increase in vibration amplitude, then a "cause" edge is created in the knowledge graph from the "bearing wear" node to the "increase in vibration amplitude" node.

[0174] In this embodiment, the data analysis module 12 can match the feature information obtained through feature extraction with the nodes and relationships in the constructed device performance degradation knowledge graph to determine the performance degradation information that matches the feature information. The performance degradation information includes the data degradation trajectory associated with each feature information, as well as the degree of interaction between different operating data associated with different feature information.

[0175] In S303 , the health monitoring result is generated based on the performance degradation information.

[0176] In S304, if the health monitoring result is abnormal, the warning information is output.

[0177] In this embodiment, the data analysis module 12 may input the above performance degradation information into a trained first monitoring model for processing to obtain a health monitoring result of the device to be monitored.

[0178] It should be noted that the first monitoring model can be obtained by training a pre-constructed third deep learning model based on a preset sample set. Each sample data in the preset sample set includes sample performance degradation information and a sample monitoring result corresponding to the sample performance degradation information. When training the pre-constructed third deep learning model, the sample performance degradation information in each sample is used as the input of the third deep learning model, and the sample monitoring result corresponding to the sample performance degradation information in each sample is used as the output of the third deep learning model. Through training, the third deep learning model can learn the correspondence between all possible sample performance degradation information and sample monitoring results, and the trained third deep learning model is used as the first monitoring model.

[0179] In one embodiment of the present application, when the performance degradation information includes multiple data degradation trajectories and the degree of interaction between different operating data, the data analysis module 12 may generate health monitoring results according to the following steps, as detailed below:

[0180] Determining a plurality of health monitoring data based on the device type of the device to be monitored;

[0181] Calculating weight values ​​of the plurality of health monitoring data based on the degree of interaction between the different operating data;

[0182] Determining a target degradation trajectory corresponding to each of the plurality of health monitoring data from the plurality of data degradation trajectories;

[0183] The health monitoring result is generated based on the weight values ​​of the respective plurality of health monitoring data and the target degradation trajectory.

[0184] In actual applications, the health monitoring data corresponding to different types of equipment may not be exactly the same. For example, for rotating equipment (such as motors and fans), vibration, speed, and temperature data are important health monitoring data; for electrical equipment (such as transformers and distribution cabinets), current, voltage, and power factor data are important health monitoring data. Therefore, in this embodiment, the data analysis module 12 can pre-store the corresponding relationship between different equipment types and health monitoring data.

[0185] In this embodiment, after obtaining the performance degradation information, the data analysis module 12 can determine a plurality of health monitoring data corresponding to the device to be monitored according to the device type of the device to be monitored and the above-mentioned pre-stored correspondence relationship.

[0186] It should be noted that the greater the degree of interaction between a certain operating data and another operating data, the deeper the influence between the certain operating data and the other operating data, that is, the more important the certain operating data is to the other operating data.

[0187] Therefore, the data analysis module 12 can calculate the weight values ​​of the multiple health monitoring data based on the degree of interaction between the different operating data. The greater the degree of interaction, the greater the weight value of the corresponding health monitoring data.

[0188] In this embodiment, after obtaining the performance degradation information, the data analysis module 12 may determine a target degradation trajectory corresponding to each of the plurality of health monitoring data from the plurality of data degradation trajectories included in the performance degradation information.

[0189] Afterwards, the data analysis module 12 can perform feature extraction and screening on the obtained target degradation trajectory. Specifically, the data analysis module 12 can extract key parameters in the target degradation trajectory, such as the slope, inflection point, fluctuation amplitude, etc. in the trajectory. Afterwards, the data analysis module 12 can screen out key parameters that are of great significance for the health monitoring of the equipment to be monitored based on the operating characteristics of the equipment to be monitored and the corresponding fault diagnosis requirements. For example, for the target degradation trajectory of vibration data, focus on characteristic parameters such as the growth trend of the vibration amplitude over time and the frequency and amplitude of abnormal fluctuations. These parameters can reflect whether the vibration state of the equipment to be monitored is normal and whether there is a potential fault risk.

[0190] In this embodiment, the data analysis module 12 can perform a weighted summation of the weight values ​​of each health monitoring data and its corresponding key parameters to obtain a final health assessment value. The data analysis module 12 can then obtain the health monitoring results of the monitored device based on the health assessment value and pre-stored health grading standards.

[0191] The health grading standard can include four levels: healthy, sub-healthy, faulty, and dangerous, and each level corresponds to a different assessment value range. Among them, sub-healthy, faulty, and dangerous can be collectively referred to as abnormal.

[0192] In this embodiment, the data analysis module 12 may determine the target range of the health assessment value, and determine the level corresponding to the target range as the health monitoring result of the device to be monitored.

[0193] In this embodiment, health monitoring results include but are not limited to normal and abnormal.

[0194] It should be noted that when the health monitoring result is abnormal, the health monitoring result may also carry the abnormality type and abnormality cause. The abnormality type includes but is not limited to: sub-health, failure, and danger.

[0195] Subhealthy is used to describe equipment that is experiencing early to mid-term damage, but is stable and can continue to operate. Faulty is used to describe equipment that is experiencing mid to late-term damage or is rapidly deteriorating, requiring short-term monitoring. Dangerous is used to describe equipment that is severely damaged, creating a risk of continued operation and requiring immediate shutdown for inspection and repair.

[0196] Based on this, when the data analysis module 12 detects that the health monitoring result of the device to be monitored is abnormal, it can output early warning information including the abnormality type and the cause of the abnormality.

[0197] As can be seen from the above, the data analysis module provided in this embodiment extracts features from the target operating data in combination with the current operating conditions and environment, and can accurately adapt to the actual operating status of the equipment. Subsequently, by analyzing the feature information with the help of the constructed equipment performance degradation knowledge graph, possible situations of equipment performance degradation can be quickly located. Finally, health monitoring results are generated based on the performance degradation information, making the health assessment more targeted.

[0198] See also Figure 6 , Figure 6 This is a specific implementation flow chart of the data analysis module in the abnormal warning system for equipment health monitoring provided by another embodiment of the present application. Figure 6 As shown, after obtaining the target operation data and indicator data, the data analysis module can also perform steps S401 to S404, which are detailed as follows:

[0199] In S401 , a time domain analysis is performed on the target operation data to obtain statistical features corresponding to the target operation data.

[0200] In S402, a correlation analysis is performed on the target operation data and the indicator data to obtain a correlation degree between the target operation data and the indicator data.

[0201] In S403, the statistical features and the indicator data are respectively input into a health monitoring model for processing to obtain a first monitoring result corresponding to the target operation data and a second monitoring result of the indicator data.

[0202] In S404, based on the correlation between the target operating data and the indicator data, the current operating condition and the current environment, the first monitoring result and the second monitoring result are adjusted to obtain the health monitoring result.

[0203] It should be noted that the target operating data includes data of multiple different data types (such as temperature, vibration, and sound, etc.).

[0204] In this embodiment, the data analysis module 12 can perform time domain analysis on the target operation data to obtain statistical features corresponding to the target operation data, wherein the statistical features include but are not limited to: mean value and peak value.

[0205] Afterwards, the data analysis module 12 may analyze the correlation between data of different data types in the target operation data and the indicator data according to a correlation analysis method to obtain the degree of correlation between the target operation data and the indicator data.

[0206] In practical applications, correlation analysis methods include but are not limited to: Pearson correlation coefficient, Spearman rank correlation coefficient, and mutual information.

[0207] It should be noted that when the correlation analysis method is the Pearson correlation coefficient, the degree of interaction is the actual value corresponding to the Pearson correlation coefficient; when the correlation analysis method is the Spearman rank correlation coefficient, the degree of interaction is the actual value corresponding to the Spearman rank correlation coefficient; when the correlation analysis method is the mutual information, the degree of interaction is the actual value corresponding to the mutual information.

[0208] In this embodiment, the data analysis module 12 can input the statistical characteristics and the indicator data into the health monitoring model for processing, and obtain a first monitoring result corresponding to the target operation data and a second monitoring result of the indicator data.

[0209] It should be noted that the health monitoring model can be obtained by training a pre-constructed fourth deep learning model based on a preset sample set. Each sample data in the preset sample set includes sample information (statistical features or indicator data) and the sample monitoring results corresponding to the sample information (the first monitoring results corresponding to the statistical features or the second monitoring results corresponding to the indicator data). When training the pre-constructed fourth deep learning model, the sample information in each sample is used as the input of the fourth deep learning model, and the sample monitoring results corresponding to the sample information in each sample are used as the output of the fourth deep learning model. Through training, the fourth deep learning model can learn the correspondence between all possible sample information and sample monitoring results, and the trained fourth deep learning model is used as the health monitoring model.

[0210] In this embodiment, the data analysis module 12 can determine a first weight corresponding to the first monitoring result based on the correlation between the target operation data and the indicator data. The higher the correlation, the greater the first weight. The data analysis module 12 can then determine a second weight corresponding to the second monitoring result based on the first weight.

[0211] It should be noted that the first weight + the second weight = 1.

[0212] In this embodiment, the data analysis module 12 may process the current working condition and the current environment input information into a quantitative model to obtain adjustment coefficients corresponding to the current working condition and the current environment.

[0213] The information quantification model can be obtained by training a pre-constructed fifth deep learning model based on a preset sample set. Each piece of sample data in the preset sample set includes sample parameters (sample operating conditions and sample environment) and sample coefficients corresponding to the sample parameters. When training the pre-constructed fifth deep learning model, the sample parameters in each sample are used as input to the fifth deep learning model, and the sample coefficients corresponding to the sample parameters in each sample are used as output of the fifth deep learning model. Through training, the fifth deep learning model can learn the correspondence between all possible sample parameters and sample coefficients, and the trained fifth deep learning model is used as the information quantification model.

[0214] In this embodiment, the data analysis module 12 can perform weighted summation on the first monitoring result, the first weight, the second monitoring result, the second weight, and the adjustment coefficient to determine the health score.

[0215] Afterwards, the data analysis module 12 can obtain the health monitoring result of the device to be monitored based on the health score and the pre-stored health grading standard.

[0216] The health grading standard can include four levels: healthy, sub-healthy, faulty, and dangerous, and each level corresponds to a different assessment value range. Among them, sub-healthy, faulty, and dangerous can be collectively referred to as abnormal.

[0217] In this embodiment, the data analysis module 12 may determine the target range of the health score, and determine the level corresponding to the target range as the health monitoring result of the device to be monitored.

[0218] In this embodiment, health monitoring results include but are not limited to normal and abnormal.

[0219] It should be noted that when the health monitoring result is abnormal, the health monitoring result may also carry the abnormality type and abnormality cause. The abnormality type includes but is not limited to: sub-health, failure, and danger.

[0220] Subhealthy is used to describe equipment that is experiencing early to mid-term damage, but is stable and can continue to operate. Faulty is used to describe equipment that is experiencing mid to late-term damage or is rapidly deteriorating, requiring short-term monitoring. Dangerous is used to describe equipment that is severely damaged, creating a risk of continued operation and requiring immediate shutdown for inspection and repair.

[0221] Based on this, when the data analysis module 12 detects that the health monitoring result of the device to be monitored is abnormal, it can output early warning information including the abnormality type and the cause of the abnormality.

[0222] From the above, it can be seen that the data analysis module provided in this embodiment obtains statistical characteristics by performing time domain analysis on the target operation data, and can accurately extract the basic characteristic information of the equipment operation; then, the target operation data and the indicator data are subjected to correlation analysis to clearly present the relationship between the two; finally, the first monitoring result and the second monitoring result are adjusted according to the degree of correlation between the target operation data and the indicator data, the current working conditions and the current environment, so that the final health monitoring result is as close to the actual operation status of the equipment as possible.

[0223] In one embodiment of the present application, the abnormal warning system for equipment health monitoring also supports a dual alarm system of equipment adaptive threshold + manual threshold.

[0224] Specifically, the equipment adaptive threshold alarm system specifically refers to: having the adaptive threshold learning capability for on-site service equipment, that is, after the monitored equipment is communicated with the equipment health monitoring abnormal warning system, the equipment health monitoring abnormal warning system can collect, learn and adaptively match the parameters of the normal operating state of this equipment within 20 to 30 days. When the working value of the equipment shows a trend of rising, the equipment health monitoring abnormal warning system can automatically issue an alarm.

[0225] The manual threshold alarm system opens the function of autonomously adjusting the alarm threshold at the back end when the user needs it, so as to better fit the user's on-site monitoring of the actual operation of each device, and the alarm accuracy can be effectively improved.

[0226] Through the multiple guarantees of adaptive equipment threshold learning + manual threshold setting + automatic alarm, the false alarm rate of previous fixed threshold alarms for on-site service equipment failures has been greatly reduced, the workload of on-site equipment management personnel has been reduced, and the intelligent management and control of on-site equipment has been better assisted.

[0227] In another embodiment of the present application, combined with the dual alarm system of the above-mentioned device adaptive threshold + manual threshold, the data analysis module in the abnormal warning system of the equipment health monitoring can be in a dormant state when the equipment is working normally. When the data acquisition module detects that the value of a certain data type in the monitored equipment shows a trend of increasing, the abnormal warning system of the equipment health monitoring uses the dual alarm system of the equipment adaptive threshold + manual threshold. After detecting that the above-mentioned value reaches the alarm condition, the data analysis module will be awakened. The data analysis module can be combined with the data information corresponding to the above-mentioned value in the database of the abnormal warning system of the equipment health monitoring to realize automatic diagnosis of whether the monitored equipment has fault degradation. After degradation occurs, the data analysis module can automatically push diagnostic information related to the monitored equipment.

[0228] In another embodiment of the present application, the abnormality warning system for equipment health monitoring may also include a three-dimensional equipment digital twin module. This module can implement the three-dimensional equipment digital twin function to display the operating status of the monitored equipment and various monitoring data in real time. The three-dimensional equipment digital twin module can accurately diagnose and locate faulty equipment down to the component level and send the corresponding twin model of the equipment to a display device. The faulty part can also be flashed on the corresponding twin model of the equipment, allowing the user to intuitively locate the fault location of the faulty equipment on the display device.

[0229] In another embodiment of the present application, the abnormal warning system for equipment health monitoring may further include a data monitoring module. The data monitoring module can monitor the data transmission link to ensure the security of data transmission at each node. When the data monitoring module detects a signal interruption between the data sensing module and the data acquisition module, it can accurately capture the communication abnormality node, ensuring convenient troubleshooting and maintenance when system component failure occurs. The specific functions are as follows:

[0230] Monitor the transmission signal between the data perception module and the data acquisition module, and locate the interruption in time;

[0231] Monitor the transmission signal between the data acquisition module and the data analysis module, and locate the interruption in time;

[0232] Monitor the battery level of the data sensing module and issue a timely reminder of low battery to avoid data transmission interruption.

[0233] At the same time, the data perception module, data acquisition module and data analysis module also have built-in "watchdog" circuits, which can automatically restart themselves when a signal interruption failure occurs in the data perception module, data acquisition module or data analysis module, reducing the maintenance workload of maintenance personnel and ensuring safer and more stable operation of the system.

[0234] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0235] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0236] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. An abnormal warning system for equipment health monitoring, characterized in that: include: A data acquisition module is used to determine, according to each fault type of the equipment to be monitored, the acquisition frequency of the initial operating data corresponding to each fault type, and obtain the initial operating data corresponding to each fault type based on the acquisition frequency; Statistical features of the initial operating data in each operating cycle are obtained by calculation, and the statistical features in each operating cycle are input into the trained analysis model for processing to obtain data change information corresponding to the initial operating data, and the operating state of the device to be monitored is determined according to the data change information, and the collection method of the operating data of the device to be monitored is determined according to the operating state of the device to be monitored, and the target operating data of the device to be monitored is obtained based on the collection method; wherein, the fault types include slow-changing type, fast-changing type and sudden change type; the slow-changing type fault refers to a fault in which the development process of the equipment fault is relatively slow, and the fault characteristic parameters gradually change with time, and the fault will only be obviously manifested on a longer time scale; the fast-changing type fault refers to a fault in which the change speed of the fault is relatively fast, and the fault characteristic parameters show a relatively obvious change in a relatively short period of time; the sudden change type fault refers to a fault in which the change speed of the fault is relatively fast, and the fault characteristic parameters show a relatively obvious change in a relatively short period of time; A sudden fault occurs during the operation of the equipment, and the fault characteristic parameters change dramatically in an instant, and the performance or status of the equipment immediately changes from normal to abnormal or even completely fails; the acquisition frequency of the initial operation data corresponding to the slowly changing type of fault is lower than the acquisition frequency of the initial operation data corresponding to the fast changing type of fault, and the acquisition frequency of the initial operation data corresponding to the fast changing type of fault is lower than the acquisition frequency of the initial operation data corresponding to the sudden type of fault; the acquisition mode includes a numerical acquisition mode, an additional waveform acquisition mode and a multi-point synchronous additional waveform acquisition mode. The numerical acquisition mode refers to only acquiring the actual numerical value corresponding to the operating data of the equipment to be monitored, the additional waveform acquisition refers to adding the acquisition of waveform data of the equipment to be monitored under the numerical acquisition mode, and the multi-point synchronous additional waveform acquisition refers to adding waveform acquisition of multiple measuring points of the equipment to be monitored at the same time under the numerical acquisition mode; A data analysis module is connected to the data acquisition module and is used to analyze the target operating data based on the current operating conditions and the current environment to obtain the health monitoring results of the equipment to be monitored, and output early warning information when the health monitoring results are detected to be abnormal; the current operating conditions include the workload and working mode of the equipment to be monitored.

2. The abnormal warning system for equipment health monitoring according to claim 1, characterized in that: The data acquisition module is specifically used for: Acquiring the initial operating data and determining abnormal points in the initial operating data; Performing anomaly detection on the abnormal point based on a trend recognition algorithm to obtain an abnormality level of the initial operation data; The trend identification algorithm refers to an algorithm used to analyze the trend of abnormal points changing with the set variables; determining the collection method based on the abnormality level; Data is collected from the device to be monitored based on the collection method to obtain the target operating data.

3. The abnormal warning system for equipment health monitoring according to claim 2, characterized in that: The determining the collection method based on the abnormality level includes: If the abnormality level is the first level, determining that the collection mode is a numerical collection mode; If the abnormality level is the second level, the acquisition mode is determined to be the additional waveform acquisition mode; the abnormality level of the second level is higher than the abnormality level of the first level; If the abnormality level is the third level, the acquisition mode is determined to be a multi-point synchronous waveform acquisition mode; the abnormality level of the third level is higher than the abnormality level of the second level.

4. The abnormal warning system for equipment health monitoring according to claim 1, characterized in that: The data acquisition module is also used for: determining a density type of the target operating data; Analyze the target operation data based on the density type to obtain indicator data corresponding to the monitoring device; The target operation data and the indicator data are sent to the data analysis module.

5. The abnormal warning system for equipment health monitoring according to claim 4, characterized in that: The analyzing the target operation data based on the density type to obtain indicator data corresponding to the monitoring device includes: If the density type is the first density type, a traditional waveform indicator corresponding to the device to be monitored is calculated based on the target operating data; the first density type is used to describe that the target operating data is low-density data, and the low density is used to characterize that the target operating data contains a small amount of information per unit time, and the traditional waveform indicator is used to describe signal characteristics; If the density type is the second density type, the traditional waveform indicator and the set fault indicator corresponding to the monitored equipment are calculated based on the target operation data; the second density type is used to describe that the target operation data is high-density data, and the high density is used to characterize that the target operation data contains a large amount of information per unit time; the set fault indicator is different from the traditional waveform indicator.

6. The abnormal warning system for equipment health monitoring according to any one of claims 1 to 5, characterized in that: The data analysis module is specifically used for: Extracting features from the target operating data based on the current operating condition and the current environment to obtain feature information of the device to be monitored; Analyzing the characteristic information based on the constructed equipment performance degradation knowledge graph to obtain performance degradation information of the equipment to be monitored; generating the health monitoring result based on the performance degradation information; If the health monitoring result is abnormal, the warning information is output.

7. The abnormal warning system for equipment health monitoring according to claim 6, characterized in that: The extracting features of the target operating data based on the current operating condition and the current environment to obtain feature information of the device to be monitored includes: determining a dictionary learning algorithm corresponding to the target operating data based on the current environment and the current operating condition; Feature extraction is performed on the target operation data based on the dictionary learning algorithm to obtain the feature information.

8. The abnormal warning system for equipment health monitoring according to claim 6, characterized in that: Before analyzing the characteristic information based on the constructed device performance degradation knowledge graph to obtain the performance degradation trend information of the device to be monitored, the method further includes: Obtain a historical operation data set of the device to be monitored; the historical operation data set includes historical data of different data types when the device to be monitored is continuously operating under different working conditions and different environments, and historical data of different data types when the device to be monitored fails under different working conditions and different environments; Extracting features from each historical data in the historical operation data set to obtain a plurality of historical feature parameters; Analyzing the plurality of characteristic parameters to obtain a data degradation trajectory of the device to be monitored; Performing correlation analysis on each historical data of the different data types to obtain a degree of interaction between each historical data of the different data types; Based on the historical operation data set, the data degradation trajectory and the degree of interaction, the equipment performance degradation knowledge graph is constructed.

9. The abnormal warning system for equipment health monitoring according to claim 8, characterized in that: The performance degradation information includes a plurality of the data degradation trajectories and a degree of interaction between different operating data; Generating the health monitoring result based on the performance degradation information includes: Determining a plurality of health monitoring data based on the device type of the device to be monitored; Calculating weight values ​​of the plurality of health monitoring data based on the degree of interaction between the different operating data; Determining a target degradation trajectory corresponding to each of the plurality of health monitoring data from the plurality of data degradation trajectories; The health monitoring result is generated based on the weight values ​​of the respective plurality of health monitoring data and the target degradation trajectory.

10. The abnormal warning system for equipment health monitoring according to claim 4, characterized in that: The data analysis module is also used to: Performing time domain analysis on the target operation data to obtain statistical features corresponding to the target operation data; performing a correlation analysis on the target operation data and the indicator data to obtain a correlation degree between the target operation data and the indicator data; Inputting the statistical features and the indicator data into a health monitoring model for processing respectively, to obtain a first monitoring result corresponding to the target operating data and a second monitoring result of the indicator data; Based on the degree of correlation between the target operating data and the indicator data, the current operating conditions and the current environment, the first monitoring result and the second monitoring result are adjusted to obtain the health monitoring result.

Images