Equipment training safety analysis method and system based on state monitoring

Abstract

The invention relates to the technical field of equipment safety analysis, and particularly discloses an equipment training safety analysis method and system based on state monitoring, and the method comprises the steps: segmenting monitoring data into a feature data group and a label data group; extracting a preset number of data from the feature data set, and performing data expansion based on the extracted data to obtain a virtual feature data set; constructing a sample set from the virtual feature data groups of each extraction quantity to the label data groups, sending the sample set to a model provider, and receiving a neural network model of each extraction quantity fed back by the model provider; calculating the error rate and the safety degree of each extraction number of neural network models; receiving an error rate demand and a safety degree demand input by a user, and matching the neural network model as an analysis model; the invention provides a data extraction and reexpansion scheme, so that a user can autonomously select how much real data is provided, and the information security of equipment is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment safety analysis, and in particular to an equipment training safety analysis method and system based on state monitoring. Background Art

[0002] In the era of smart manufacturing and the Industrial Internet of Things (IIoT), equipment training (such as for intelligent robots, automated machinery, or industrial applications of AI models) relies on vast amounts of operational data. This data typically comes from equipment status monitoring systems, such as sensors, control units, and SCADA (Supervisory Control and Data Acquisition) systems. After acquiring status data and the current output of the equipment, training can be used to establish a relationship between status data and output. For example, neural network models can be used to analyze certain preset status data and predict the equipment's output under those conditions.

[0003] However, the existing model training process is generally completed by a third party, and some third parties specialize in providing model training services. If the model demander provides all the operating data of the equipment, the third party can easily know the specific details of the equipment, which poses a certain risk of information leakage. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for analyzing equipment training safety based on state monitoring, so as to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A device training safety analysis method based on state monitoring, the method comprising: Establish a connection channel with the monitor in the device to obtain monitoring data containing the monitor number and monitoring time in real time; The monitoring data is divided into feature data group and label data group according to the monitor number; Extracting a preset amount of data from the feature data group, performing data expansion based on the extracted data, and obtaining a virtual feature data group; wherein the monitoring time of the virtual feature data group adopts the monitoring time of the corresponding feature data group; Construct a sample set of virtual feature data groups for each extracted number to label data groups, send it to the model provider, and receive the neural network model for each extracted number fed back by the model provider; Calculate the error rate and safety of the neural network model for each number of samples; Receive the error rate requirements and safety requirements input by the user, match the neural network model, and use it as the analysis model.

[0006] As a further solution of the present invention, the step of establishing a connection channel with the monitor in the device and obtaining monitoring data containing the monitor number and monitoring time in real time includes: Query the smart module registered in the device; the smart module includes a sensor and a controller; Obtaining the operating data of each intelligent module, for any intelligent module, comparing the operating data of the intelligent module with the operating data of other intelligent modules within a preset distance range, and calculating the correlation; When the absolute value of the correlation reaches the preset threshold, only one of the intelligent modules of the two comparisons is retained; After comparing and retaining all intelligent modules, they are regarded as monitors and numbered; Establish a connection channel with the monitor in the device and obtain monitoring data containing the monitor number and monitoring time in real time.

[0007] As a further solution of the present invention: the step of dividing the monitoring data into a feature data group and a label data group according to the monitor number includes: Receive the target indicator input by the user based on the preset indicator receiving port; Query the monitor number of the monitor corresponding to the target indicator as the target number, and use the monitor number of the monitor corresponding to the non-target indicator as the non-target number; At any moment, the monitoring data corresponding to all target numbers are counted as a label data group; Count the monitoring data corresponding to all non-target numbers as a feature data group; The data arrangement order in the label data group and the data arrangement order in the feature data group are both based on the numbering order.

[0008] As a further solution of the present invention, the step of extracting a preset amount of data from the feature data set and performing data expansion based on the extracted data to obtain a virtual feature data set includes: Determine the number of draws in sequence within a preset number range; the number of draws is an integer; Select a feature data group, and randomly select a number of data from the feature data group as known data; The data position of the known data is retained and a transit data group is created; the dimension of the transit data group is the same as that of the feature data group; For unknown data in the transfer data group, query the correlation between the unknown data and each known data; the correlation adopts the correlation of the intelligent module corresponding to the data; Query the maximum value in the historical data of known data, and predict unknown data based on the maximum value and correlation; When all unknown data predictions are completed, a virtual feature data set is obtained.

[0009] As a further solution of the present invention, the process of searching for the maximum value in the historical data of the known data and predicting the unknown data based on the maximum value and the correlation includes: ;in, is the predicted value of the unknown data, is the total number of known data, For the Known data, For the The maximum value among the historical data of known data, is the maximum value in the historical data of unknown data; Indicates the The correlation between the known data and the currently predicted unknown data.

[0010] As a further solution of the present invention: the step of calculating the error rate and safety of the neural network model for each number of extractions includes: Get the error rate of the neural network model for each number of samples; Calculate the safety margin based on the inverse ratio of the number of draws; The calculation process of safety is: Where, For safety, is the preset correction factor, Indicates the number of extractions.

[0011] The technical solution of the present invention also provides a device training safety analysis system based on state monitoring, the system comprising: The monitoring data acquisition module is used to establish a connection channel with the monitor in the device and obtain monitoring data containing the monitor number and monitoring time in real time; The data segmentation module is used to segment the monitoring data into feature data groups and label data groups according to the monitor number; A data expansion module is used to extract a preset amount of data from the feature data group, perform data expansion based on the extracted data, and obtain a virtual feature data group; wherein the monitoring time of the virtual feature data group adopts the monitoring time of the corresponding feature data group; The sample construction application module is used to construct a sample set of virtual feature data groups to label data groups for each extraction number, send it to the model provider, and receive the neural network model of each extraction number fed back by the model provider; Parameter operation module, used to calculate the error rate and safety of the neural network model for each number of extractions; The model matching module is used to receive the error rate requirements and safety requirements input by the user and match the neural network model as the analysis model.

[0012] As a further solution of the present invention: the monitoring data acquisition module includes: A data query unit, used to query the registered intelligent module in the device; the intelligent module includes a sensor and a controller; A correlation calculation unit is used to obtain the operating data of each intelligent module, and for any intelligent module, compare the operating data of the intelligent module with the operating data of other intelligent modules within a preset distance range to calculate the correlation; The module elimination unit is used to retain only one of the intelligent modules of both parties when the absolute value of the correlation reaches a preset threshold; A numbering unit is used to compare and retain all intelligent modules, and then use them as monitors and number the monitors; The acquisition execution unit is used to establish a connection channel with the monitor in the device and obtain monitoring data containing the monitor number and monitoring time in real time.

[0013] As a further solution of the present invention: the data segmentation module includes: An indicator receiving unit, configured to receive a target indicator input by a user based on a preset indicator receiving port; a number query unit, configured to query the monitor number of the monitor corresponding to the target indicator as the target number, and the monitor number of the monitor corresponding to the non-target indicator as the non-target number; The first statistical unit is used to count the monitoring data corresponding to all target numbers at any time as a label data group; The second statistical unit is used to count the monitoring data corresponding to all non-target numbers as a feature data group; The data arrangement order in the label data group and the data arrangement order in the feature data group are both based on the numbering order.

[0014] As a further solution of the present invention: the data expansion module includes: A quantity determination unit, configured to sequentially determine the number of draws within a preset quantity range; the number of draws is an integer; A data selection unit is used to select a feature data group and randomly select a number of data from the feature data group as known data; A data retention unit is used to retain the data location of known data and create a transit data group; the dimension of the transit data group is the same as the dimension of the feature data group; A correlation query unit is used to query the correlation between the unknown data and each known data in the transfer data group; the correlation adopts the correlation of the intelligent module corresponding to the data; The prediction execution unit is used to query the maximum value in the historical data of known data and predict unknown data based on the maximum value and correlation; The output unit is used to obtain a virtual feature data set after all unknown data predictions are completed.

[0015] Compared with the existing technology, the beneficial effects of the present invention are: the present invention provides a solution for data extraction and expansion, and performs fuzzy processing on the samples provided by the model demander. The third party cannot know whether the data is completely real data. The third party's job is only to train the model based on the samples. After the model is trained, the model demander can obtain the error rate and security of each extraction quantity based on the real data stored by itself, and then choose how much real data to provide, which greatly improves the information security of the device. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention.

[0017] Figure 1 A flowchart of the safety analysis method for training condition-based monitoring equipment.

[0018] Figure 2 A structural block diagram of the equipment training safety analysis system based on condition monitoring. DETAILED DESCRIPTION

[0019] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] Figure 1 This is a flowchart of a method for analyzing the safety of equipment training based on state monitoring. In an embodiment of the present invention, a method for analyzing the safety of equipment training based on state monitoring includes: Step S100: establishing a connection channel with the monitor in the device and obtaining monitoring data including the monitor number and monitoring time in real time; The modules in the device that contain data acquisition and data transmission functions are called monitors, such as sensors connected to the data transmission bus, or controllers of various components. During the operation of the device, these monitors acquire data and upload it to the management background of the device. When acquiring data, the monitor needs to insert the monitor number and monitoring time into the data. The monitor number is the identity of the monitor and is a predetermined unique label. The monitoring time is the time when the data is acquired.

[0021] Step S200: dividing the monitoring data into a feature data group and a label data group according to the monitor number; The monitoring data of each monitor obtained can be divided into feature data groups and label data groups according to the monitor number; in the model training process, features generally refer to independent variables, and labels generally refer to dependent variables. The above segmentation process is essentially a segmentation of the monitors. Which monitors are regarded as independent variables and which monitors are regarded as dependent variables, and the trained model functions will also be updated accordingly, indicating which monitors perform data predictions on which monitors.

[0022] Generally speaking, when the equipment is production equipment, the monitor serving as a label is mostly the monitor at the equipment outlet, which is used to characterize the final production status of the production equipment. This type of monitor mostly obtains product information, not equipment information. For example, the monitor serving as a feature can be a module in the equipment used to obtain the operating parameters of each component, and the monitor serving as a label is generally a module used to obtain product parameters, including weight, volume, and surface roughness, etc. In terms of data volume, the number of elements in the feature data group can be large, and the number of elements in the label data group is generally small, and most of them are only one.

[0023] Step S300: extracting a preset amount of data from the feature data group, performing data expansion based on the extracted data, and obtaining a virtual feature data group; wherein the monitoring time of the virtual feature data group adopts the monitoring time of the corresponding feature data group; The feature data group is the data obtained by all monitors that serve as labels. It contains too much information. If the data is known to a third party (model provider), the third party can fully know the situation of the device, and the security is actually not high. For this, the present application provides a data virtualization solution, which extracts a preset amount of data from the feature data group, that is, selects a part of the data, and then expands this part of the data to obtain a data group with the same dimension as the feature data group, which is called a virtual feature data group. The virtual feature data group also contains a time tag, and its time tag is the monitoring time of the feature data group.

[0024] An example of the above process is: assuming there are 10 monitors, the feature data set obtained is 10-dimensional. When the number is 3, it means that 3 data are extracted from the feature data set and their positions are retained. For the data at the other 7 positions, it is unknown. The unknown data is simulated based on the 3 extracted data. After the simulation is completed, the data set obtained is a virtual feature data set.

[0025] Step S400: constructing a sample set of virtual feature data sets to label data sets for each extraction quantity, sending the sample set to the model provider, and receiving the neural network model for each extraction quantity fed back by the model provider; For any number of extractions, the data extraction process is random (taking the above example as the basis, the process of extracting data at 3 positions in a 10-dimensional data set is random), but no matter which data are extracted, data expansion will be performed to obtain a virtual feature data set. Therefore, this application directly uses the number of extractions as the index to count the virtual feature data sets for each number of extractions. At the same time, the label data set at the same moment is obtained. The label data is the prediction target and the dependent variable. All data must be retained. After the number of extractions is determined, a sample set from the virtual feature data set to the label data set is constructed, and it is divided into a training set and a test set according to a preset ratio to train the neural network model.

[0026] It should be noted that the premise for the virtual feature data group and the label data group to constitute a sample is that their times need to correspond, that is, the virtual feature data group is the independent variable of the label data group at the same moment; in addition, regarding the number of samples, each moment corresponds to a sample, a device has multiple moments, and the device itself can also have multiple moments, so the number of samples can be large.

[0027] Step S500: Calculate the error rate and safety of each extracted number of neural network models; In the process of training the neural network model, the error rate is used to represent the difference between the real data and the output data. After the model training party provides the model, the model demander can judge the error rate of the model based on its own real data; the larger the number of extractions, the more real the data provided by the model demander, and the greater the error rate. At the same time, this application also calculates the security level once. The larger the number of extractions, the more public the operation status of the equipment, and the lower the security level.

[0028] Step S600: receiving the error rate requirement and safety requirement input by the user, and matching the neural network model as the analysis model; The user is the person who needs to perform analysis and prediction. The system receives the error rate and safety requirements input by the user, compares them with the error rate and safety of the neural network model of each extraction quantity, and obtains the extraction quantity that meets the requirements. The neural network model corresponding to the extraction quantity is read as the analysis model. At this time, the user inputs some operating parameters of the equipment (the prediction stage is generally the simulated operating parameters), and the label data under the operating parameters can be obtained, which provides analysis functions while meeting the safety requirements.

[0029] It is worth mentioning that the user in this application refers to the model demander, who is the party providing data, and the third party is the model provider, who is the party training the model based on the data.

[0030] Regarding step S100, the steps of establishing a connection channel with the monitor in the device and obtaining monitoring data containing the monitor number and monitoring time in real time include: Query the smart module registered in the device; the smart module includes a sensor and a controller; Obtaining the operating data of each intelligent module, for any intelligent module, comparing the operating data of the intelligent module with the operating data of other intelligent modules within a preset distance range, and calculating the correlation; When the absolute value of the correlation reaches the preset threshold, only one of the intelligent modules of the two comparisons is retained; After comparing and retaining all intelligent modules, they are regarded as monitors and numbered; Establish a connection channel with the monitor in the device and obtain monitoring data containing the monitor number and monitoring time in real time.

[0031] In an example of the technical solution of the present invention, a monitor selection scheme is provided. In a device, there are many intelligent modules that can be used as monitors. The intelligent modules registered in the device are queried to obtain the operating data of each intelligent module. For any intelligent module, the operating data of the intelligent module is compared with the operating data of other intelligent modules within a preset distance range, and the correlation is calculated. When the two intelligent modules are sufficiently correlated, one intelligent module is eliminated and only one is retained. This process is continuously executed in a loop, which can optimize the number of monitors, reduce the number of selected monitors, and alleviate the pressure of sample acquisition and analysis, specifically the dimension of the feature data group. After the monitor is selected, the monitor is numbered.

[0032] It is worth mentioning that there are two possibilities for sufficient correlation between two intelligent modules, one is positive correlation and the other is negative correlation. This application regards both situations as sufficiently correlated. Therefore, the way to judge the degree of correlation in the above content is to judge whether the absolute value of the correlation reaches a preset threshold.

[0033] Regarding step S200, the step of dividing the monitoring data into feature data groups and label data groups according to the monitor number includes: Receive the target indicator input by the user based on the preset indicator receiving port; Query the monitor number of the monitor corresponding to the target indicator as the target number, and use the monitor number of the monitor corresponding to the non-target indicator as the non-target number; At any moment, the monitoring data corresponding to all target numbers are counted as a label data group; Count the monitoring data corresponding to all non-target numbers as a feature data group; The data arrangement order in the label data group and the data arrangement order in the feature data group are both based on the numbering order.

[0034] In an example of the technical solution of the present invention, the monitor segmentation process is explained. The user inputs the indicator in the preset indicator receiving port as the target indicator, queries the monitor number of the monitor corresponding to the target indicator as the target number, and uses the monitor number of the monitor corresponding to the non-target indicator as the non-target number. In this way, the monitor is segmented. After the segmentation is completed, the monitoring data corresponding to the target number at the same time is the label data group, and the monitoring data corresponding to all non-target numbers is the feature data group.

[0035] In the above content, no matter it is a tag data set or a feature data set, the corresponding relationship between the element position and the monitor is preset and fixed.

[0036] Regarding step S300, the step of extracting a preset amount of data from the feature data set and performing data expansion based on the extracted data to obtain a virtual feature data set includes: Determine the number of draws in sequence within a preset number range; the number of draws is an integer; Select a feature data group, and randomly select a number of data from the feature data group as known data; The data position of the known data is retained and a transit data group is created; the dimension of the transit data group is the same as that of the feature data group; For unknown data in the transfer data group, query the correlation between the unknown data and each known data; the correlation adopts the correlation of the intelligent module corresponding to the data; Query the maximum value in the historical data of known data, and predict unknown data based on the maximum value and correlation; When all unknown data predictions are completed, a virtual feature data set is obtained.

[0037] The key point of this application is to first extract data from the feature data set and then expand the data. The process is as follows: Determine the number of extractions in sequence within the preset range. If the range is 1 to 10, the number of extractions is determined to be 1, 2, 3, ..., 10. After determining the number of extractions, for any feature data group, randomly select the number of data to be extracted from the feature data group. If the number of extractions is 3, extract 3 data from any feature data group. The extracted data is regarded as known data, and the unextracted data is called unknown data.

[0038] For the unknown data in the transit data group, query the correlation between the unknown data and each known data. When executing step S200, the correlation between the monitors of each data location has been calculated. At this time, it can be read directly, and the maximum value is queried in the historical data of the known data. The unknown data is predicted based on the maximum value and correlation.

[0039] The process of searching for the maximum value in the historical data of known data and predicting unknown data based on the maximum value and the correlation includes: ;in, is the predicted value of the unknown data, is the total number of known data, For the Known data, For the The maximum value among the historical data of known data, is the maximum value in the historical data of unknown data; Indicates the The correlation between the known data and the currently predicted unknown data.

[0040] In an example of the technical solution of the present invention, the prediction process of unknown data is explained. It uses the average of the predicted data of all known data for unknown data. The prediction process of each known data for unknown data is to calculate the ratio of the known data to the maximum value in the historical data. If the correlation is 1, then the ratio is multiplied by the maximum value in the historical data of the unknown data to obtain the predicted value. However, the correlation is not necessarily 1. Therefore, in the multiplication process, the correlation parameter is introduced, and the correlation adopts the Pearson correlation coefficient.

[0041] Regarding step S500, the step of calculating the error rate and safety of each extracted number of neural network models includes: Get the error rate of the neural network model for each number of samples; The safety factor is calculated based on the inverse ratio of the number of draws.

[0042] In one example of the technical solution of the present invention, for each number of neural network models extracted, after the model training party provides the model, the model demander can judge the error rate of the model based on its own real data; on this basis, the safety level is also calculated based on the number of extractions, and the safety level is inversely proportional to the number of extractions; after the processing is completed, the error rate and safety level of the neural network model for each number of extractions are obtained.

[0043] The calculation process of safety is: Where, For safety, is the preset correction factor, Indicates the number of extractions.

[0044] The calculation process of safety is not complicated. The safety is inversely proportional to the number of extractions. The reason for using a composite function with an exponential function and a positive proportional coefficient is that as the number of extractions increases, the reduction in safety becomes smaller. The principle is that the larger the number of extractions, the closer the derivative is to zero.

[0045] Figure 2 The following is a structural block diagram of a device training safety analysis system based on state monitoring. The technical solution of the present invention also provides a device training safety analysis system based on state monitoring. The system 10 includes: The monitoring data acquisition module 11 is used to establish a connection channel with the monitor in the device and obtain monitoring data containing the monitor number and monitoring time in real time; The data segmentation module 12 is used to segment the monitoring data into a feature data group and a label data group according to the monitor number; The data expansion module 13 is used to extract a preset amount of data from the feature data group, perform data expansion based on the extracted data, and obtain a virtual feature data group; wherein the monitoring time of the virtual feature data group adopts the monitoring time of the corresponding feature data group; The sample construction application module 14 is used to construct a sample set of virtual feature data groups to label data groups for each extraction number, send it to the model provider, and receive the neural network model for each extraction number fed back by the model provider; Parameter calculation module 15, used to calculate the error rate and safety of the neural network model for each number of extractions; The model matching module 16 is used to receive the error rate requirement and safety requirement input by the user, and match the neural network model as the analysis model.

[0046] Furthermore, the monitoring data acquisition module 11 includes: A data query unit, used to query the registered intelligent module in the device; the intelligent module includes a sensor and a controller; A correlation calculation unit is used to obtain the operating data of each intelligent module, and for any intelligent module, compare the operating data of the intelligent module with the operating data of other intelligent modules within a preset distance range to calculate the correlation; The module elimination unit is used to retain only one of the intelligent modules of both parties when the absolute value of the correlation reaches a preset threshold; A numbering unit is used to compare and retain all intelligent modules, and then use them as monitors and number the monitors; The acquisition execution unit is used to establish a connection channel with the monitor in the device and obtain monitoring data containing the monitor number and monitoring time in real time.

[0047] Specifically, the data segmentation module 12 includes: An indicator receiving unit, configured to receive a target indicator input by a user based on a preset indicator receiving port; a number query unit, configured to query the monitor number of the monitor corresponding to the target indicator as the target number, and the monitor number of the monitor corresponding to the non-target indicator as the non-target number; The first statistical unit is used to count the monitoring data corresponding to all target numbers at any time as a label data group; The second statistical unit is used to count the monitoring data corresponding to all non-target numbers as a feature data group; The data arrangement order in the label data group and the data arrangement order in the feature data group are both based on the numbering order.

[0048] Furthermore, the data expansion module 13 includes: A quantity determination unit, configured to sequentially determine the number of draws within a preset quantity range; the number of draws is an integer; A data selection unit is used to select a feature data group and randomly select a number of data from the feature data group as known data; A data retention unit is used to retain the data location of known data and create a transit data group; the dimension of the transit data group is the same as the dimension of the feature data group; A correlation query unit is used to query the correlation between the unknown data and each known data in the transfer data group; the correlation adopts the correlation of the intelligent module corresponding to the data; The prediction execution unit is used to query the maximum value in the historical data of known data and predict unknown data based on the maximum value and correlation; The output unit is used to obtain a virtual feature data set after all unknown data predictions are completed.

[0049] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A device training safety analysis method based on state monitoring, characterized in that: The method comprises: Establish a connection channel with the monitor in the device to obtain monitoring data containing the monitor number and monitoring time in real time; The monitoring data is divided into feature data group and label data group according to the monitor number; Extracting a preset amount of data from the feature data group, performing data expansion based on the extracted data, and obtaining a virtual feature data group; wherein the monitoring time of the virtual feature data group adopts the monitoring time of the corresponding feature data group; Construct a sample set of virtual feature data groups for each extracted number to label data groups, send it to the model provider, and receive the neural network model for each extracted number fed back by the model provider; Calculate the error rate and safety of the neural network model for each number of samples; Receive the error rate requirements and safety requirements input by the user, match the neural network model, and use it as the analysis model.

2. The equipment training safety analysis method based on state monitoring according to claim 1 is characterized in that: The steps of establishing a connection channel with the monitor in the device and obtaining monitoring data containing the monitor number and monitoring time in real time include: Query the smart module registered in the device; the smart module includes a sensor and a controller; Obtaining the operating data of each intelligent module, for any intelligent module, comparing the operating data of the intelligent module with the operating data of other intelligent modules within a preset distance range, and calculating the correlation; When the absolute value of the correlation reaches the preset threshold, only one of the intelligent modules of the two comparisons is retained; After comparing and retaining all intelligent modules, they are regarded as monitors and numbered; Establish a connection channel with the monitor in the device and obtain monitoring data containing the monitor number and monitoring time in real time.

3. The equipment training safety analysis method based on state monitoring according to claim 1 is characterized in that: The step of dividing the monitoring data into a feature data group and a label data group according to the monitor number includes: Receive the target indicator input by the user based on the preset indicator receiving port; Query the monitor number of the monitor corresponding to the target indicator as the target number, and use the monitor number of the monitor corresponding to the non-target indicator as the non-target number; At any moment, the monitoring data corresponding to all target numbers are counted as a label data group; Count the monitoring data corresponding to all non-target numbers as a feature data group; The data arrangement order in the label data group and the data arrangement order in the feature data group are both based on the numbering order.

4. The equipment training safety analysis method based on state monitoring according to claim 2 is characterized in that: The step of extracting a preset amount of data from the feature data set and performing data expansion based on the extracted data to obtain a virtual feature data set includes: Determine the number of draws in sequence within a preset number range; the number of draws is an integer; Select a feature data group, and randomly select a number of data from the feature data group as known data; The data position of the known data is retained and a transit data group is created; the dimension of the transit data group is the same as that of the feature data group; For unknown data in the transfer data group, query the correlation between the unknown data and each known data; the correlation adopts the correlation of the intelligent module corresponding to the data; Query the maximum value in the historical data of known data, and predict unknown data based on the maximum value and correlation; When all unknown data predictions are completed, a virtual feature data set is obtained.

5. The equipment training safety analysis method based on state monitoring according to claim 4 is characterized in that: The process of searching for the maximum value in the historical data of known data and predicting unknown data based on the maximum value and the correlation includes: ;in, is the predicted value of the unknown data, is the total number of known data, For the Known data, For the The maximum value among the historical data of known data, is the maximum value in the historical data of unknown data; Indicates the The correlation between the known data and the currently predicted unknown data.

6. The equipment training safety analysis method based on state monitoring according to claim 1 is characterized in that: The step of calculating the error rate and safety of the neural network model for each number of extractions includes: Get the error rate of the neural network model for each number of samples; Calculate the safety margin based on the inverse ratio of the number of draws; The calculation process of safety is: Where, For safety, is the preset correction factor, Indicates the number of extractions.

7. A device training safety analysis system based on condition monitoring, characterized in that: The system comprises: The monitoring data acquisition module is used to establish a connection channel with the monitor in the device and obtain monitoring data containing the monitor number and monitoring time in real time; The data segmentation module is used to segment the monitoring data into feature data groups and label data groups according to the monitor number; A data expansion module is used to extract a preset amount of data from the feature data group, perform data expansion based on the extracted data, and obtain a virtual feature data group; wherein the monitoring time of the virtual feature data group adopts the monitoring time of the corresponding feature data group; The sample construction application module is used to construct a sample set of virtual feature data groups to label data groups for each extraction number, send it to the model provider, and receive the neural network model of each extraction number fed back by the model provider; Parameter operation module, used to calculate the error rate and safety of the neural network model for each number of extractions; The model matching module is used to receive the error rate requirements and safety requirements input by the user and match the neural network model as the analysis model.

8. The equipment training safety analysis system based on condition monitoring according to claim 7 is characterized in that: The monitoring data acquisition module includes: A data query unit, used to query the registered intelligent module in the device; the intelligent module includes a sensor and a controller; A correlation calculation unit is used to obtain the operating data of each intelligent module, and for any intelligent module, compare the operating data of the intelligent module with the operating data of other intelligent modules within a preset distance range to calculate the correlation; The module elimination unit is used to retain only one of the intelligent modules of both parties when the absolute value of the correlation reaches a preset threshold; A numbering unit is used to compare and retain all intelligent modules, and then use them as monitors and number the monitors; The acquisition execution unit is used to establish a connection channel with the monitor in the device and obtain monitoring data containing the monitor number and monitoring time in real time.

9. The equipment training safety analysis system based on condition monitoring according to claim 7, characterized in that: The data segmentation module includes: An indicator receiving unit, configured to receive a target indicator input by a user based on a preset indicator receiving port; a number query unit, configured to query the monitor number of the monitor corresponding to the target indicator as the target number, and the monitor number of the monitor corresponding to the non-target indicator as the non-target number; The first statistical unit is used to count the monitoring data corresponding to all target numbers at any time as a label data group; The second statistical unit is used to count the monitoring data corresponding to all non-target numbers as a feature data group; The data arrangement order in the label data group and the data arrangement order in the feature data group are both based on the numbering order.

10. The equipment training safety analysis system based on condition monitoring according to claim 8, characterized in that: The data expansion module includes: A quantity determination unit, configured to sequentially determine the number of draws within a preset quantity range; the number of draws is an integer; A data selection unit is used to select a feature data group and randomly select a number of data from the feature data group as known data; A data retention unit is used to retain the data location of known data and create a transit data group; the dimension of the transit data group is the same as the dimension of the feature data group; A correlation query unit is used to query the correlation between the unknown data and each known data in the transfer data group; the correlation adopts the correlation of the intelligent module corresponding to the data; A prediction execution unit is used to query the maximum value in the historical data of known data and predict unknown data based on the maximum value and correlation; The output unit is used to obtain a virtual feature data set after all unknown data predictions are completed.

Images