Methods for determining the health status of the model device, training methods, and equipment.

Abstract

This application provides a method, training method, and device for determining the health status of a device model. The method includes: obtaining a health status confirmation instruction; wherein the health status confirmation instruction indicates a corresponding device type; determining at least one corresponding sub-health status type based on the device type; obtaining sub-source data corresponding to the device type and sub-health status type, and obtaining a sub-health status model corresponding to the device type and sub-health status type; wherein the sub-health status model is trained based on labeled historical sub-source data; there is a one-to-one correspondence between the sub-source data and the sub-health status model; and determining the health status corresponding to the health status confirmation instruction based on the sub-source data and the sub-health status model corresponding to the sub-source data. This method aims to improve the accuracy and relevance of determining the health status of a device.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, training method and device for determining the health of a model device. Background Technology

[0002] With the development of the oil refining and chemical industries, various types of equipment have emerged. Among them, determining the health status of the equipment is particularly important during its operation.

[0003] In existing technologies, the health of a device is typically assessed based on its operational data.

[0004] However, existing methods can lead to inaccurate and poorly targeted determination of health status. Summary of the Invention

[0005] This application provides a method, training method, and equipment for determining the health status of a device, in order to improve the accuracy and relevance of determining the health status of the device.

[0006] In a first aspect, embodiments of this application provide a model-based method for determining the health of a device, comprising:

[0007] Obtain a health confirmation command; wherein the health confirmation command indicates the corresponding device type;

[0008] Based on the device type, at least one corresponding sub-health type is determined;

[0009] Obtain sub-source data corresponding to the device type and the sub-health type, and obtain sub-health models corresponding to the device type and the sub-health type; wherein, the sub-health model is trained based on labeled historical sub-source data; the sub-source data and the sub-health model are in one-to-one correspondence.

[0010] Based on the sub-source data and the sub-health model corresponding to the sub-source data, the health level corresponding to the health confirmation instruction is determined.

[0011] In one possible implementation, determining the health level corresponding to the health level confirmation instruction based on the sub-source data and the sub-health level model corresponding to the sub-source data includes: inputting the sub-source data into the sub-health level model corresponding to the sub-source data to obtain the corresponding sub-health level; and determining the health level corresponding to the health level confirmation instruction based on the obtained sub-health levels.

[0012] In one possible implementation, the method further includes: generating and presenting corresponding alarm information based on a preset health warning threshold and the obtained health level; or, generating and presenting corresponding alarm information based on a preset sub-health warning threshold, a health warning threshold, the obtained sub-health level, and the obtained health level.

[0013] In one possible implementation, if the device type is a refining device or an atmospheric and vacuum distillation device, then the corresponding at least one sub-health type includes: equipment type sub-health, energy consumption type sub-health, safety type sub-health, environmental protection type sub-health, process type sub-health, raw material type sub-health, and product type sub-health.

[0014] In one possible implementation, if the device type is a refining type device or an atmospheric and vacuum distillation type device, then the process type sub-health includes at least one process sub-process health.

[0015] In one possible implementation, the method further includes: performing time-series alignment processing, outlier handling, missing value handling, time-series resampling processing, and time-series difference processing on the sub-source data.

[0016] Secondly, embodiments of this application provide a method for training a model for determining the health status of a device, comprising:

[0017] Obtain the initial sub-health model and historical sub-source data corresponding to the device type and sub-health type;

[0018] Based on clustering and similarity matching algorithms, the historical sub-source data is labeled to obtain labeled historical resource data.

[0019] Based on the historical resource data after the labeling process, the initial sub-health model is trained to obtain the sub-health model;

[0020] The sub-health model is the model described in the first aspect and / or various possible implementations of the first aspect above.

[0021] In one possible implementation, the historical sub-source data includes multiple historical sub-source sample data; based on clustering algorithms and similarity matching algorithms, the historical sub-source data is labeled to obtain labeled historical resource data, including: clustering the historical sub-source data based on clustering algorithms to obtain clustered historical sub-source data; wherein the clustered historical sub-source data includes multiple cluster data sets; wherein each cluster data set includes at least one of the historical sub-source sample data; in response to a user's labeling operation, labeling is performed on each historical sub-source sample data in at least one of the multiple cluster data sets; and labeling is performed on each historical sub-source sample data in other cluster data sets based on similarity matching algorithms.

[0022] In one possible implementation, the historical sub-source data is clustered using a clustering algorithm to obtain clustered historical sub-source data, including: determining multiple initial cluster center samples from multiple historical sub-source sample data included in the historical sub-source data; performing clustering processing on the multiple historical sub-source sample data included in the historical sub-source data based on the multiple initial cluster center samples to obtain a clustering processing result; updating the initial cluster center samples according to the clustering processing result, and re-clustering the multiple historical sub-source sample data included in the historical sub-source data based on the updated multiple initial cluster center samples, until the clustering processing result no longer changes.

[0023] In one possible implementation, the method further includes: performing time-series alignment processing, outlier processing, missing value processing, time-series resampling processing, and time-series difference processing on the historical sub-source data.

[0024] Thirdly, embodiments of this application provide a model-based device for determining the health status of an apparatus, comprising:

[0025] The first acquisition module is used to acquire a health confirmation command; wherein the health confirmation command indicates the corresponding device type;

[0026] The first determining module is used to determine at least one corresponding sub-health type based on the device type;

[0027] The second acquisition module is used to acquire sub-source data corresponding to the device type and the sub-health type, and to acquire sub-health models corresponding to the device type and the sub-health type; wherein, the sub-health model is trained based on labeled historical sub-source data; the sub-source data and the sub-health model are in one-to-one correspondence.

[0028] The second determining module is used to determine the health level corresponding to the health level confirmation instruction based on the sub-source data and the sub-health level model corresponding to the sub-source data.

[0029] In one possible implementation, the second determining module is used to input the sub-source data into the sub-health model corresponding to the sub-source data to obtain the corresponding sub-health; and determine the health corresponding to the health confirmation instruction based on the obtained sub-health values.

[0030] In one possible implementation, the second determining module is further configured to generate and present corresponding alarm information based on a preset health warning threshold and the obtained health level; or, based on a preset sub-health warning threshold, a health warning threshold, the obtained sub-health level, and the obtained health level, generate and present corresponding alarm information.

[0031] In one possible implementation, if the device type is a refining device or an atmospheric and vacuum distillation device, then the corresponding at least one sub-health type includes: equipment type sub-health, energy consumption type sub-health, safety type sub-health, environmental protection type sub-health, process type sub-health, raw material type sub-health, and product type sub-health.

[0032] In one possible implementation, if the device type is a refining type device or an atmospheric and vacuum distillation type device, then the process type sub-health includes at least one process sub-process health.

[0033] In one possible implementation, the second acquisition module is further configured to perform time-series alignment processing, outlier processing, missing value processing, time-series resampling processing, and time-series difference processing on the sub-source data.

[0034] Fourthly, embodiments of this application provide a training apparatus for a model used to determine the health status of a device, comprising:

[0035] The acquisition module is used to acquire the initial sub-health model and historical sub-source data corresponding to the device type and sub-health type;

[0036] The tagging module is used to perform tagging processing on the historical sub-source data based on clustering algorithms and similarity matching algorithms to obtain historical resource data after tagging processing;

[0037] The training module is used to train the initial sub-health model based on the historical resource data after the labeling process, so as to obtain the sub-health model.

[0038] The sub-health model is the model described in the first aspect and / or various possible implementations of the first aspect above.

[0039] In one possible implementation, the historical sub-source data includes multiple historical sub-source sample data; the labeling module is specifically used to perform clustering processing on the historical sub-source data based on a clustering algorithm to obtain clustered historical sub-source data; wherein the clustered historical sub-source data includes multiple clustered data sets; wherein each clustered data set includes at least one of the historical sub-source sample data; in response to the user's labeling operation, labeling processing is performed on each historical sub-source sample data in at least one of the multiple clustered data sets; and labeling processing is performed on each historical sub-source sample data in other clustered data sets based on a similarity matching algorithm.

[0040] In one possible implementation, the labeling module is further specifically configured to: determine multiple initial cluster center samples from multiple historical sub-source sample data included in the historical sub-source data; perform clustering processing on the multiple historical sub-source sample data included in the historical sub-source data based on the multiple initial cluster center samples to obtain the clustering processing result; update the initial cluster center samples according to the clustering processing result, and re-cluster the multiple historical sub-source sample data included in the historical sub-source data based on the updated multiple initial cluster center samples, until the clustering processing result no longer changes.

[0041] In one possible implementation, the acquisition module is further configured to perform time-series alignment processing, outlier processing, missing value processing, time-series resampling processing, and time-series difference processing on the historical sub-source data.

[0042] Fifthly, embodiments of this application provide a clock-clocking device, including: a memory and a processor;

[0043] The memory stores computer-executed instructions;

[0044] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect, or the second aspect and / or various possible implementations of the second aspect.

[0045] In a sixth aspect, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect, or the second aspect and / or various possible implementations of the second aspect.

[0046] In a seventh aspect, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect, or the second aspect and / or various possible implementations of the second aspect.

[0047] The method, training method, and device for determining the health status of a device provided in this application embodiment obtain a health status confirmation instruction, determine at least one corresponding sub-health status type according to the device type, obtain sub-source data corresponding to the device type and sub-health status type, and obtain sub-health status models corresponding to the device type and sub-health status types. Based on the sub-source data and the sub-health status models corresponding to the sub-source data, the health status corresponding to the health status confirmation instruction is determined. In the process of determining the health status, the corresponding sub-health status type is first determined according to the device type, and then the health status of the device is determined according to the determined at least one sub-health status type. This can solve the problems of inaccurate and poor targeting in determining the health status of the device. Attached Figure Description

[0048] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0049] Figure 1 A flowchart illustrating the model-based device health determination method provided in this application. Figure 1 ;

[0050] Figure 2 A flowchart illustrating the model-based device health determination method provided in this application. Figure 2 ;

[0051] Figure 3 A flowchart illustrating the training method for the model used to determine the health of a device provided in this application. Figure 1 ;

[0052] Figure 4 A flowchart illustrating the training method for the model used to determine the health of a device provided in this application. Figure 2 ;

[0053] Figure 5 A schematic diagram of the structure of the model-based device for determining the health of the device provided in this application;

[0054] Figure 6 A schematic diagram of the structure of the training device for the model used to determine the health of the device provided in this application;

[0055] Figure 7 A schematic diagram of the structure of the electronic device provided in this application.

[0056] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0057] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0058] In existing technologies, the health of a device is typically assessed based on its operational data. However, this approach can lead to inaccurate and poorly targeted assessments of health status.

[0059] The method for determining the health status of a device model provided in this application involves obtaining a health status confirmation instruction, determining at least one corresponding sub-health status type based on the device type, obtaining sub-source data corresponding to the device type and sub-health status type, obtaining a sub-health status model corresponding to the device type and sub-health status type, and determining the health status corresponding to the health status confirmation instruction based on the sub-source data and the sub-health status model corresponding to the sub-source data. In the process of determining the health status, the corresponding sub-health status type is first determined based on the device type, and then the health status of the device is determined based on the determined at least one sub-health status type. This method can solve the problems of inaccurate and poor targeting in determining the health status of the device.

[0060] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0061] Figure 1 A flowchart illustrating the model-based device health determination method provided in this application. Figure 1 ,like Figure 1 As shown, the method includes:

[0062] Step S101: Obtain health confirmation command.

[0063] Specifically, a health confirmation command can be obtained. This health confirmation command is used to confirm the health status of the corresponding device. The health confirmation command indicates the corresponding device type. This application does not limit the type of device indicated by the health confirmation command; any device type can be used as the device type provided in this application. For example, in the chemical industry, the device type indicated by the health confirmation command can be a refining and chemical plant, an atmospheric and vacuum distillation unit, a catalytic cracking unit, or a cracking unit, etc.

[0064] This application does not limit the process of obtaining health confirmation instructions; optionally, health confirmation instructions can be obtained at a preset time period or using a preset acquisition strategy. This application does not limit the form of health confirmation instructions; optionally, they can be any of the following: voice instructions, text instructions, operation instructions, etc.

[0065] Step S102: Determine at least one corresponding sub-health type based on the device type.

[0066] Specifically, based on the device type indicated by the health confirmation instruction obtained in step S101, at least one corresponding sub-health type can be determined. This application does not limit the process of determining at least one corresponding sub-health type based on the device type; optionally, it can determine at least one corresponding sub-health type based on the device type and a preset correspondence between sub-health types. The preset correspondence between sub-health types indicates the correspondence between device types and sub-health types. This application does not limit the correspondence between device types and sub-health types; optionally, it can determine the corresponding sub-health type based on the device characteristics corresponding to the device type. This application does not limit the number of sub-health types corresponding to a device type; the number of sub-health types corresponding to a device type corresponds to the device type. For example, if the device type is a refining type device, then there are seven corresponding sub-health types.

[0067] Step S103: Obtain the sub-source data corresponding to the device type and sub-health type, and obtain the sub-health model corresponding to the device type and sub-health type.

[0068] Specifically, after determining at least one corresponding sub-health type based on the device type, sub-source data corresponding to the device type and sub-health type can be obtained, along with a sub-health model corresponding to the device type and sub-health type. Sub-source data refers to the source data used to determine the corresponding sub-health. This application does not limit the sub-source data; any source data capable of determining the corresponding sub-health can be used as the sub-source data corresponding to the device type and sub-health type provided in this application. Sub-health model refers to the model used to determine the corresponding sub-health. This application does not limit the sub-health model; any model capable of determining the corresponding sub-health can be used as the sub-health model corresponding to the device type and sub-health type provided in this application. The sub-health model is trained based on labeled historical sub-source data. There is a one-to-one correspondence between sub-source data and sub-health models.

[0069] Step S104: Determine the health level corresponding to the health level confirmation command based on the sub-source data and the sub-health level model corresponding to the sub-source data.

[0070] Specifically, based on the sub-source data and the corresponding sub-health model obtained in step S103, the health level corresponding to the health confirmation command can be determined. This application does not limit the process of determining the health level corresponding to the health confirmation command based on the sub-source data and the corresponding sub-health model; any process that can determine the health level corresponding to the health confirmation command based on the sub-source data and the corresponding sub-health model can be used as the process provided in this application for determining the health level corresponding to the health confirmation command based on the sub-source data and the corresponding sub-health model. Optionally, the sub-source data can be first input into the corresponding sub-health model to obtain the corresponding sub-health level; then, the health level corresponding to the health confirmation command can be determined based on the obtained sub-health levels.

[0071] The model-based method for determining the health of a device provided in this application involves obtaining a health confirmation instruction, determining at least one corresponding sub-health type based on the device type, obtaining sub-source data corresponding to the device type and sub-health type, obtaining a sub-health model corresponding to the device type and sub-health type, and determining the health corresponding to the health confirmation instruction based on the sub-source data and the sub-health model corresponding to the sub-source data. In the process of determining the health, the corresponding sub-health type is first determined based on the device type, and then the health of the device is determined based on the determined at least one sub-health type. This method can solve the problems of inaccurate and poor targeting in determining the health of the device.

[0072] In one possible embodiment, if the equipment type is a refining or atmospheric / distillation type equipment, then the corresponding at least one sub-health type includes: equipment type sub-health, energy consumption type sub-health, safety type sub-health, environmental protection type sub-health, process type sub-health, raw material type sub-health, and product type sub-health. Specifically, the equipment type sub-health is the sub-health type corresponding to the equipment itself; the energy consumption type sub-health is the sub-health type corresponding to the energy consumption of the equipment; the safety type sub-health is the sub-health type corresponding to the safety of the equipment; the environmental protection type sub-health is the sub-health type corresponding to the environmental friendliness of the equipment; the process type sub-health is the sub-health type corresponding to the process of the equipment; the raw material type sub-health is the sub-health type corresponding to the raw materials used by the equipment; and the product type sub-health is the sub-health type corresponding to the product produced by the equipment.

[0073] Specifically, according to the description of sub-source data in step S103, if the unit type is a refining / chemical type unit or an atmospheric / distillation type unit, and the corresponding sub-health type is equipment type sub-health, then the corresponding sub-source data is equipment monitoring alarm data. If the unit type is a refining / chemical type unit or an atmospheric / distillation type unit, and the corresponding sub-health type is energy consumption type sub-health, then the corresponding sub-source data is energy consumption monitoring data. If the unit type is a refining / chemical type unit or an atmospheric / distillation type unit, and the corresponding sub-health type is safety type sub-health, then the corresponding sub-source data is combustible gas alarm data. If the unit type is a refining / chemical type unit or an atmospheric / distillation type unit, and the corresponding sub-health type is environmental protection type sub-health, then the corresponding sub-source data is environmental monitoring data. If the unit type is a refining / chemical type unit or an atmospheric / distillation type unit, and the corresponding sub-health type is process type sub-health, then the corresponding sub-source data is process operation data. If the unit type is a refining and chemical processing unit or an atmospheric and vacuum distillation unit, and the corresponding sub-health type is a raw material type sub-health, then the corresponding sub-source data is raw material quality analysis data. If the unit type is a refining and chemical processing unit or an atmospheric and vacuum distillation unit, and the corresponding sub-health type is a product type sub-health, then the corresponding sub-source data is product quality analysis data.

[0074] Specifically, according to the description of the sub-health model in step S103, if the device type is a refining type device or an atmospheric and vacuum distillation type device, and the corresponding sub-health type is equipment type sub-health, then the corresponding sub-health model is a sub-health model trained based on historical equipment monitoring alarm data. If the device type is a refining type device or an atmospheric and vacuum distillation type device, and the corresponding sub-health type is energy consumption type sub-health, then the corresponding sub-health model is a sub-health model trained based on historical energy consumption monitoring data. If the device type is a refining type device or an atmospheric and vacuum distillation type device, and the corresponding sub-health type is safety type sub-health, then the corresponding sub-health model is a sub-health model trained based on historical combustible gas alarm data. If the device type is a refining type device or an atmospheric and vacuum distillation type device, and the corresponding sub-health type is environmental protection type sub-health, then the corresponding sub-health model is a sub-health model trained based on historical environmental protection monitoring data. If the unit type is a refining / chemical unit or an atmospheric / distillation unit, and the corresponding sub-health type is a process-type sub-health, then the corresponding sub-health model is a sub-health model trained based on historical process operation data. If the unit type is a refining / chemical unit or an atmospheric / distillation unit, and the corresponding sub-health type is a raw material-type sub-health, then the corresponding sub-health model is a sub-health model trained based on historical raw material quality analysis data. If the unit type is a refining / chemical unit or an atmospheric / distillation unit, and the corresponding sub-health type is a product-type sub-health, then the corresponding sub-health model is a sub-health model trained based on historical product quality analysis data.

[0075] In one possible embodiment, if the plant type is a refining or atmospheric / vacuum-fired boiler plant, the process type sub-health status includes at least one process sub-process health status. Here, the process sub-process health status is the health status corresponding to a process sub-process within a process, and the process includes at least one process sub-process. This application does not limit the number of process sub-process health statuses included in the process type sub-health status; the number of process sub-process health statuses included in the process type sub-health status corresponds to the plant type. For example, if the plant type is a refining or chemical plant, the process type sub-health status includes six process sub-process health statuses: desalination process sub-process health status, primary distillation process sub-process health status, atmospheric furnace process sub-process health status, atmospheric distillation process sub-process health status, vacuum furnace process sub-process health status, and vacuum distillation process sub-process health status.

[0076] Figure 2 A flowchart illustrating the model-based device health determination method provided in this application. Figure 2 ,like Figure 2 As shown, in this embodiment... Figure 1Based on the embodiments, another model-based method for determining the health of a device is described in detail, the method comprising:

[0077] Step S201: Obtain health confirmation command.

[0078] For a detailed description of this step, please refer to step S101; it will not be repeated here.

[0079] Step S202: Determine at least one corresponding sub-health type based on the device type.

[0080] For a detailed description of this step, please refer to step S102; it will not be repeated here.

[0081] Step S203: Obtain the sub-source data corresponding to the device type and sub-health type, and obtain the sub-health model corresponding to the device type and sub-health type.

[0082] For a detailed description of this step, please refer to step S103; it will not be repeated here.

[0083] Step S204: Input the sub-source data into the sub-health model corresponding to the sub-source data to obtain the corresponding sub-health score.

[0084] Specifically, the sub-source data can be input into the corresponding sub-health model to obtain the corresponding sub-health score. The description of the sub-source data and the sub-source health model can be found in step S103, and will not be repeated here.

[0085] Optionally, if the unit type is a refining and chemical unit or an atmospheric and vacuum distillation unit, the equipment monitoring alarm data is input into a sub-health model trained based on historical equipment monitoring alarm data to obtain the equipment sub-health; the energy consumption monitoring data is input into a sub-health model trained based on historical energy consumption monitoring data to obtain the energy consumption sub-health; the combustible gas alarm data is input into a sub-health model trained based on historical combustible gas alarm data to obtain the energy consumption sub-health; the environmental monitoring data is input into a sub-health model trained based on historical environmental monitoring data to obtain the energy consumption sub-health; the process operation data is input into a sub-health model trained based on historical process operation data to obtain the energy consumption sub-health; the raw material quality analysis data is input into a sub-health model trained based on historical raw material quality analysis data to obtain the energy consumption sub-health; and the product quality analysis data is input into a sub-health model trained based on historical product quality analysis data to obtain the energy consumption sub-health.

[0086] Step S205: Determine the health level corresponding to the health level confirmation command based on the obtained sub-health levels.

[0087] Specifically, based on the sub-health levels obtained in step S201, the health level corresponding to the health level confirmation command can be determined. This application does not limit the process of determining the health level corresponding to the health level confirmation command based on the obtained sub-health levels; any process capable of determining the health level corresponding to the health level confirmation command based on the obtained sub-health levels can be considered as the process provided in this application for determining the health level corresponding to the health level confirmation command based on the obtained sub-health levels. Optionally, a weight set corresponding to the health level confirmation command can be obtained, wherein the weight set includes the weight of each sub-health level among at least one sub-health level; then, a weighted summation is performed based on the obtained sub-health levels and the obtained health level set to obtain the corresponding health level. The formula for the weighted summation is as follows:

[0088]

[0089] in, For health, For the i-th sub-health, Let be the weight of the i-th sub-health score.

[0090] In this application, the weight of each sub-health level included in the weight set is not limited, and the sum of the weights of all sub-health levels included in the weight set is 1. For example, if the device type corresponding to the health level determination instruction is an atmospheric and vacuum distillation device, then the weight of the equipment sub-health level included in the corresponding weight set is 0.1, the weight of the energy consumption sub-health level is 0.1, the weight of the safety sub-health level is 0.1, the weight of the environmental protection sub-health level is 0.1, the weight of the process sub-health level is 0.2, the weight of the raw material sub-health level is 0.1, and the weight of the product sub-health level is 0.3.

[0091] Optionally, if the sub-health score is a process sub-health score, then the health scores of each process sub-process can be obtained, and a weight set of the process sub-process health scores corresponding to the health score confirmation instruction can be obtained. The weight set of process sub-process health scores includes the weight of each process sub-process health score within at least one process sub-process health score. Then, based on the obtained process sub-process health scores and the obtained weight set of process sub-process health scores, a weighted summation is performed to obtain the process sub-health score. This application does not limit the process for obtaining the health scores of each process sub-process. Optionally, the sub-source data corresponding to each process sub-process and a sub-health score model trained based on the historical sub-source data corresponding to the process sub-process can be obtained first. Then, the sub-source data corresponding to each process sub-process can be input into the sub-health score model trained based on the historical sub-source data corresponding to the process sub-process to obtain the corresponding process sub-process health score.

[0092] Step S206: Generate and present corresponding alarm information based on the preset health warning threshold and the obtained health status; or, generate and present corresponding alarm information based on the preset sub-health warning threshold, the health warning threshold, the obtained sub-health status, and the obtained health status.

[0093] Specifically, after determining the health level corresponding to the health confirmation command, a corresponding alarm message can be generated and presented based on a preset health warning threshold and the obtained health level. This application does not limit the preset health warning threshold; any warning threshold capable of providing a health level warning can be used as the health warning threshold provided in this application. This application does not limit the process of generating and presenting the corresponding alarm message based on the preset health warning threshold and the obtained health level. Optionally, if the obtained health level is greater than or equal to the health warning threshold, the corresponding alarm message is generated and presented; if the obtained health level is less than the health warning threshold, no alarm message is generated. The alarm message includes a health level value that is greater than or equal to the health warning threshold.

[0094] Alternatively, after determining the health level corresponding to the health level confirmation command, corresponding alarm information can be generated and presented based on preset sub-health level warning thresholds, health level warning thresholds, and the obtained sub-health level and health level. This application does not limit the preset sub-health level warning thresholds; any warning threshold capable of issuing warnings for sub-health levels can be used as the sub-health level warning threshold provided by this application. This application does not limit the process of generating and presenting corresponding alarm information based on preset sub-health level warning thresholds, health level warning thresholds, and the obtained sub-health level and health level. Optionally, if the obtained health level is greater than or equal to the health level warning threshold, and / or at least one obtained sub-health level is greater than or equal to the corresponding sub-health level warning threshold, then corresponding alarm information is generated and presented; otherwise, no alarm information is generated. The alarm information includes health levels greater than or equal to the health level warning threshold, and / or sub-health levels greater than or equal to the sub-health level warning threshold and their numerical values.

[0095] The model-based method for determining the health of a device provided in this application embodiment, in Figure 1 Based on the illustrated embodiment, by inputting sub-source data into the sub-health model corresponding to the sub-source data, the corresponding sub-health level is obtained. Based on the obtained sub-health levels, the health level corresponding to the health confirmation command is determined. This enables the determination of the health level corresponding to the health confirmation command based on the sub-source data and the sub-health model corresponding to the sub-source data. Figure 1Based on the illustrated embodiment, the accuracy of the obtained health status is further improved. Specifically, alarm information can be generated and presented by using a preset health status warning threshold, or a preset health status warning threshold and a preset sub-health status warning threshold, thereby improving the accuracy of the alarm.

[0096] In one possible embodiment, after acquiring the sub-source data corresponding to the device type and sub-health type, the model-based method for determining the health of the device further includes:

[0097] The sub-source data undergoes time-series alignment, outlier handling, missing value handling, time-series resampling, and time-series differencing. Time-series alignment involves generating new timestamps using interpolation or resampling methods to align the data temporally. Outlier handling involves detecting outliers using the standard deviation method and replacing them with common values ​​such as the mean, median, and mode. Missing value handling can be achieved through near substitution, mean substitution, or interpolation. Time-series resampling transforms the original data from low-frequency to high-frequency data. Time-series differencing can reduce data volume and extract key information.

[0098] In one possible embodiment, after obtaining the resource data corresponding to the device type and sub-health type, the model-based method for determining the health of the device further includes:

[0099] First, statistical feature extraction algorithms are used to collect information such as the mean, maximum, minimum, standard deviation, missing values, and trends of various indicators, reflecting the central trend, dispersion, and distribution of the data, thus completing edge or cloud feature engineering processing for big data. Then, machine learning algorithms (such as threshold-based methods, support vector machines, and random forests) are used to detect abnormal patterns or outliers in the data stream. These algorithms can automatically learn the normal behavior patterns of the data stream and identify data points that deviate from the normal pattern. If outliers are identified, outlier replacement processing is performed. The specific process for outlier replacement can be found in the description above.

[0100] Figure 3 A flowchart illustrating the training method for the model used to determine the health of a device provided in this application. Figure 1 ,like Figure 3 As shown, the method includes:

[0101] Step S301: Obtain the initial sub-health model and historical sub-source data corresponding to the device type and sub-health type.

[0102] Specifically, the initial sub-health model and historical sub-source data corresponding to the device type and sub-health type can be obtained. The description of the historical sub-source data can be found in step S103, where historical sub-source data refers to the sub-source data corresponding to the device's historical usage.

[0103] The initial sub-health model is the initial model used to train the sub-health models. Different sub-health models correspond to different initial sub-health models.

[0104] Step S302: Based on clustering algorithm and similarity matching algorithm, label the historical sub-source data to obtain the labeled historical resource data.

[0105] Specifically, based on clustering algorithms and similarity matching algorithms, the historical sub-source data obtained in step S301 can be labeled to obtain labeled historical resource data. This application does not limit the process of labeling historical sub-source data based on clustering algorithms and similarity matching algorithms to obtain labeled historical resource data. Any process that can label historical sub-source data based on clustering algorithms and similarity matching algorithms to obtain labeled historical resource data can be used as the process provided in this application for labeling historical sub-source data based on clustering algorithms and similarity matching algorithms to obtain labeled historical resource data. Optionally, the historical sub-source data can be clustered based on a clustering algorithm to obtain clustered historical sub-source data. In response to the user's labeling operation, labeling processing can be performed on each historical sub-source sample data in at least one of multiple clustered datasets, and labeling processing can be performed on each historical sub-source sample data in other clustered datasets based on a similarity matching algorithm.

[0106] Step S303: Train the initial sub-health model based on the historical resource data after tagging and marking to obtain the sub-health model.

[0107] Specifically, based on the historical resource data processed by tagging obtained in step S302, the initial sub-health model obtained in step S301 is trained to obtain the sub-health model. This application does not limit the process of training the initial sub-health model based on the historical resource data processed by tagging to obtain the sub-health model. Optionally, the initial sub-health model can be trained based on the natural extended semi-supervised machine learning algorithm, using the historical resource data processed by tagging, to obtain the sub-health model.

[0108] Among them, the sub-health model is Figure 1 or Figure 2 The model in any of the embodiments shown.

[0109] The training method for a model used to determine the health status of a device provided in this application involves acquiring an initial sub-health status model and historical sub-source data corresponding to the device type and sub-health status type. Based on clustering and similarity matching algorithms, the historical sub-source data is labeled to obtain labeled historical resource data. The initial sub-health status model is then trained based on the labeled historical resource data to obtain the sub-health status model. Labeling the historical sub-source data using clustering and similarity matching algorithms improves the accuracy of the labeling process. Furthermore, training the initial sub-health status model based on the labeled historical resource data further improves the accuracy of the sub-health status model training.

[0110] In one possible embodiment, before labeling the historical sub-source data based on clustering and similarity matching algorithms to obtain the labeled historical resource data, the method further includes: performing temporal alignment, outlier handling, missing value handling, temporal resampling, and temporal differencing on the historical sub-source data. The descriptions of the temporal alignment, outlier handling, missing value handling, temporal resampling, and temporal differencing processes on the historical sub-source data can be found above, and will not be repeated here.

[0111] In one possible embodiment, the training process of the model used to determine the health of the device can be conducted visually. Specifically, in a pre-built visualization modeling platform, the graphics corresponding to each model building process can be displayed on the interface by dragging and dropping. The data transmission process between each model building process is presented on the interface by connecting the graphics with arrows, so that the visualization modeling platform can sequentially implement the corresponding model building process according to the order of the graphics presented on the interface. The visualization modeling platform can display the big data processing and analysis process through a graphical interface, using visualization tools and graphical elements to intuitively present complex data structures and data relationships to users, helping them better understand the meaning of the data and extract its value.

[0112] Figure 4 A flowchart illustrating the training method for the model used to determine the health of a device provided in this application. Figure 2 ,like Figure 4 As shown, in this embodiment... Figure 3Based on the embodiments, the process of labeling historical sub-source data using clustering and similarity matching algorithms to obtain labeled historical resource data is described in detail. This method includes:

[0113] Step S401: Perform clustering processing on the historical sub-source data based on the clustering algorithm to obtain the clustered historical sub-source data.

[0114] The description of the historical sub-source data can be found in step S301, and will not be repeated here. Optionally, the historical sub-source data includes multiple historical sub-source sample data.

[0115] Specifically, historical sub-source data can be clustered using a clustering algorithm to obtain clustered historical sub-source data. This application does not limit the process of clustering historical sub-source data using a clustering algorithm to obtain clustered historical sub-source data. Optionally, the process may include: first, determining multiple initial cluster center samples from the multiple historical sub-source sample data included in the historical sub-source data; then, based on the multiple initial cluster center samples, performing clustering processing on the multiple historical sub-source sample data included in the historical sub-source data to obtain the clustering result; then, based on the clustering result, updating the initial cluster center samples, and re-clustering the multiple historical sub-source sample data included in the historical sub-source data based on the updated initial cluster center samples, until the clustering result no longer changes.

[0116] The historical sub-source data after clustering includes multiple cluster datasets. Each cluster dataset includes at least one historical sub-source sample data.

[0117] Specifically, in determining multiple initial cluster center samples from multiple historical sub-source samples included in the historical sub-source data, this application does not limit the specific process. Optionally, multiple initial cluster center samples can be determined randomly from the multiple historical sub-source samples included in the historical sub-source data. For example, if the historical sub-source data includes n historical sub-source samples, then K historical sub-source samples can be randomly selected as initial cluster center samples.

[0118] Specifically, in the process of clustering multiple historical sub-source sample data included in the historical sub-source data based on multiple initial cluster center samples to obtain the clustering results, this application does not limit the specific process. Optionally, the feature vector corresponding to each historical sub-source sample data can be calculated first; then the distance between the feature vector corresponding to each historical sub-source sample data and the feature vector corresponding to each initial cluster center sample can be calculated; then, according to the calculated distance, each historical sub-source sample data can be assigned to the class corresponding to the initial cluster center sample with the closest distance.

[0119] The formula for calculating the distance between the feature vector corresponding to each historical sub-source sample data and the feature vector corresponding to each initial cluster center sample is as follows:

[0120]

[0121] in, The feature vector corresponding to the historical sub-source sample data. Let be the feature vector corresponding to the initial cluster center sample, where , , where m is the number of elements in the feature vector. The k-th element in the feature vector corresponding to the historical sub-source sample data. Let i be the k-th element in the feature vector corresponding to the initial cluster center sample, where i∈{1,2,...,n}, j∈{1,2,...,K}, where n is the number of historical sub-source sample data, and K is the number of initial cluster center samples, i.e., the number of clusters.

[0122] Specifically, in the process of updating the initial cluster center samples based on the results of clustering, and re-clustering the historical sub-source sample data included in the historical sub-source data based on the updated initial cluster center samples until the results of clustering no longer change, this application does not limit the specific process. Optionally, the average value of the feature vectors corresponding to each historical sub-source sample data in the class corresponding to each initial cluster center sample can be calculated; the historical sub-source sample data that is closest to the calculated average value in each historical sub-source sample data is determined as the updated initial cluster center sample.

[0123] The formula for calculating the average value of the feature vectors corresponding to each historical sub-source sample data is shown below:

[0124]

[0125] Among them, For sample feature vectors, Let G be the mean of the sample feature vectors, and |G| be the number of samples in class G.

[0126] If the updated initial cluster center sample data differs from the original initial cluster center sample, then based on multiple updated initial cluster center samples, clustering is performed on multiple historical sub-source sample data included in the historical sub-source data to obtain a new clustering result. Then, based on the new clustering result, the updated initial cluster center sample is updated again until the clustering result no longer changes, that is, the updated initial cluster center sample is the same as the updated initial cluster center sample.

[0127] Step S402: In response to the user's labeling operation, label the historical sub-source sample data in at least one of the multiple clustering datasets.

[0128] Specifically, in response to the user's labeling operation, each historical sub-source sample data in at least one of multiple clustered datasets can be labeled. Here, the historical resource sample data in each labeled clustered dataset is the optimal historical resource sample data determined by the user based on their historical experience with the device's operation.

[0129] Step S403: Based on the similarity matching algorithm, label the historical sub-source sample data in each of the other cluster datasets in the multiple cluster datasets.

[0130] Specifically, based on the similarity matching algorithm, each historical sub-source sample data in other cluster data sets among multiple cluster data sets can be labeled. Here, "other cluster data sets" refers to the cluster data sets other than at least one cluster data set that has already undergone labeling in step S402.

[0131] The formula for labeling each historical sub-source sample data in other cluster datasets within multiple cluster datasets, based on the similarity matching algorithm, is shown below:

[0132]

[0133] in, This indicates a labeled dataset, that is, a clustered dataset that has undergone labeling. This indicates an unlabeled dataset, that is, a clustered dataset that has not undergone labeling. The function representing the corresponding sub-health model. This represents a similarity matching function used to calculate the similarity between unlabeled and labeled data. This method employs a distance similarity algorithm. This represents the loss function for labeled data, which measures the prediction error of the corresponding sub-health model on the labeled data. The loss function represents the loss function for unlabeled data. It compares the unlabeled data with the labels of the most similar labeled data through similarity matching. The Ω function represents the square of the Euclidean distance. This represents the weighting parameter, used to balance the importance of labeled and unlabeled data in the optimization process. Adjustments need to be made based on the corresponding sub-health model to ensure the accuracy of the model.

[0134] Specifically, based on the above calculation process, each historical sub-source sample data in other cluster data sets among multiple cluster data sets can be mapped to each historical sub-source sample data in at least one cluster data set among multiple cluster data sets that have been labeled, so as to perform labeling processing on each historical sub-source sample data in other cluster data sets among multiple cluster data sets.

[0135] This application provides a process for labeling historical sub-source data based on clustering and similarity matching algorithms to obtain labeled historical resource data. The process involves clustering the historical sub-source data using a clustering algorithm to obtain clustered historical sub-source data. In response to user labeling operations, labeling is performed on each historical sub-source sample data in at least one of multiple clustered datasets. Furthermore, labeling is performed on each historical sub-source sample data in other clustered datasets using a similarity matching algorithm. The combination of clustering algorithms and user labeling operations improves the accuracy of the initial labeling process. Further, a secondary labeling process using a similarity matching algorithm further enhances the efficiency and accuracy of the labeling process, thereby improving the accuracy of the sub-health model trained on the labeled historical sub-source sample data.

[0136] Figure 5 A schematic diagram of the health determination device based on the model provided in this application is shown below. Figure 5 As shown, the model-based device for determining the health status of an apparatus provided in this embodiment includes:

[0137] The first acquisition module is used to acquire a health confirmation command; wherein, the health confirmation command indicates the corresponding device type;

[0138] The first determining module is used to determine at least one corresponding sub-health type based on the device type;

[0139] The second acquisition module is used to acquire the sub-source data corresponding to the device type and sub-health type, and to acquire the sub-health model corresponding to the device type and sub-health type; wherein, the sub-health model is trained based on the labeled historical sub-source data; the sub-source data and the sub-health model are in one-to-one correspondence.

[0140] The second determining module is used to determine the health level corresponding to the health level confirmation instruction based on the sub-source data and the sub-health level model corresponding to the sub-source data.

[0141] In one possible implementation, the second determining module is used to input the sub-source data into the sub-health model corresponding to the sub-source data to obtain the corresponding sub-health; and to determine the health corresponding to the health confirmation instruction based on the obtained sub-health.

[0142] In one possible embodiment, the second determining module is further configured to generate and present corresponding alarm information based on a preset health warning threshold and the obtained health level; or, based on a preset sub-health warning threshold, a health warning threshold, the obtained sub-health level, and the obtained health level, generate and present corresponding alarm information.

[0143] In one possible embodiment, if the device type is a refining device or an atmospheric and vacuum distillation device, then the corresponding at least one sub-health type includes: equipment type sub-health, energy consumption type sub-health, safety type sub-health, environmental protection type sub-health, process type sub-health, raw material type sub-health, and product type sub-health.

[0144] In one possible embodiment, if the equipment type is a refining type equipment or an atmospheric and vacuum distillation type equipment, the process type sub-health includes at least one process sub-process health.

[0145] In one possible embodiment, the second acquisition module is further configured to perform temporal alignment processing, outlier processing, missing value processing, temporal resampling processing, and temporal difference processing on the sub-source data.

[0146] The model-based device for determining the health of a device provided in this embodiment can execute the method provided in the above-described method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0147] Figure 6 A schematic diagram of the structure of the training device for the model used to determine the health of the device provided in this application is shown below. Figure 6 As shown, the training device 60 for the model used to determine the health of a device provided in this embodiment includes:

[0148] The acquisition module is used to acquire the initial sub-health model and historical sub-source data corresponding to the device type and sub-health type;

[0149] The tagging module is used to perform tagging processing on historical sub-source data based on clustering and similarity matching algorithms to obtain historical resource data after tagging.

[0150] The training module is used to train the initial sub-health model based on the historical resource data after labeling and processing, so as to obtain the sub-health model.

[0151] The sub-health model is the model provided in the above method embodiments.

[0152] In one possible embodiment, the historical sub-source data includes multiple historical sub-source sample data; a labeling module is specifically used to perform clustering processing on the historical sub-source data based on a clustering algorithm to obtain clustered historical sub-source data; wherein the clustered historical sub-source data includes multiple clustered data sets; wherein each clustered data set includes at least one historical sub-source sample data; in response to the user's labeling operation, labeling processing is performed on each historical sub-source sample data in at least one of the multiple clustered data sets; and labeling processing is performed on each historical sub-source sample data in other clustered data sets based on a similarity matching algorithm.

[0153] In one possible embodiment, the labeling module is further specifically configured to: determine multiple initial cluster center samples from multiple historical sub-source sample data included in the historical sub-source data; perform clustering processing on the multiple historical sub-source sample data included in the historical sub-source data based on the multiple initial cluster center samples to obtain the clustering processing result; update the initial cluster center samples according to the clustering processing result, and re-cluster the multiple historical sub-source sample data included in the historical sub-source data based on the updated multiple initial cluster center samples, until the clustering processing result no longer changes.

[0154] The training device for the model used to determine the health of the device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0155] Figure 7 A schematic diagram of the structure of the electronic device provided in this application. Figure 7As shown, the electronic device 70 provided in this embodiment includes at least one processor 701 and a memory 702. Optionally, the device 70 further includes a communication component 703. The processor 701, memory 702, and communication component 703 are connected via a bus 704.

[0156] In a specific implementation, at least one processor 701 executes computer execution instructions stored in memory 702, causing at least one processor 701 to perform the above-described method.

[0157] The specific implementation process of processor 701 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0158] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0159] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0160] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0161] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0162] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0163] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0164] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0165] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0166] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0167] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0168] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0169] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0170] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for determining the health status of a model-based device, characterized in that, include: Obtain a health confirmation command; wherein the health confirmation command indicates the corresponding device type; Based on the device type, at least one corresponding sub-health type is determined; Obtain sub-source data corresponding to the device type and the sub-health type, and obtain sub-health models corresponding to the device type and the sub-health type; wherein, the sub-health model is trained based on labeled historical sub-source data; the sub-source data and the sub-health model are in one-to-one correspondence. Based on the sub-source data and the sub-health model corresponding to the sub-source data, the health level corresponding to the health confirmation instruction is determined.

2. The method according to claim 1, characterized in that, Based on the sub-source data and the corresponding sub-health model, the health level corresponding to the health confirmation instruction is determined, including: Input the sub-source data into the sub-health model corresponding to the sub-source data to obtain the corresponding sub-health score; Based on the obtained sub-health scores, determine the health score corresponding to the health score confirmation instruction.

3. The method according to claim 2, characterized in that, The method further includes: Based on the preset health warning threshold and the obtained health status, generate and present corresponding alarm information; or, based on the preset sub-health warning threshold, the obtained sub-health status, and the obtained health status, generate and present corresponding alarm information.

4. The method according to claim 1, characterized in that, If the device type is a refining and chemical processing device or an atmospheric and vacuum distillation device, then the corresponding at least one sub-health type includes: equipment type sub-health, energy consumption type sub-health, safety type sub-health, environmental protection type sub-health, process type sub-health, raw material type sub-health, and product type sub-health.

5. The method according to claim 4, characterized in that, If the device type is a refining type device or an atmospheric and vacuum distillation type device, then the process type sub-health includes at least one process sub-process health.

6. The method according to any one of claims 1-5, characterized in that, The method further includes: The sub-source data undergoes time-series alignment, outlier handling, missing value handling, time-series resampling, and time-series difference processing.

7. A method for training a model for determining the health status of a device, characterized in that, include: Obtain the initial sub-health model and historical sub-source data corresponding to the device type and sub-health type; Based on clustering and similarity matching algorithms, the historical sub-source data is labeled to obtain labeled historical resource data. Based on the historical resource data after the labeling process, the initial sub-health model is trained to obtain the sub-health model; The sub-health model is the model described in any one of claims 1-6.

8. The method according to claim 7, characterized in that, The historical sub-source data includes multiple historical sub-source sample data; based on clustering algorithms and similarity matching algorithms, the historical sub-source data is labeled to obtain labeled historical resource data, including: The historical sub-source data is clustered based on a clustering algorithm to obtain clustered historical sub-source data; wherein the clustered historical sub-source data includes multiple clustered data sets; wherein each clustered data set includes at least one of the historical sub-source sample data. In response to the user's labeling operation, labeling processing is performed on each historical sub-source sample data in at least one of the plurality of clustered datasets; Based on the similarity matching algorithm, each historical sub-source sample data in other cluster datasets of the multiple cluster datasets is labeled.

9. The method according to claim 8, characterized in that, The historical sub-source data is clustered using a clustering algorithm to obtain clustered historical sub-source data, including: Multiple initial cluster center samples are determined from the multiple historical sub-source sample data included in the historical sub-source data; Based on the multiple initial cluster center samples, the multiple historical sub-source sample data included in the historical sub-source data are clustered to obtain the clustering result. Based on the results of the clustering process, the initial cluster center samples are updated, and based on the updated initial cluster center samples, the historical sub-source sample data included in the historical sub-source data are re-clustered until the results of the clustering process no longer change.

10. The method according to any one of claims 7-9, characterized in that, The method further includes: The historical sub-source data is subjected to time-series alignment, outlier handling, missing value handling, time-series resampling, and time-series differencing.

11. A model-based device for determining the health status of an apparatus, characterized in that, include: The first acquisition module is used to acquire a health confirmation command; wherein the health confirmation command indicates the corresponding device type; The first determining module is used to determine at least one corresponding sub-health type based on the device type; The second acquisition module is used to acquire sub-source data corresponding to the device type and the sub-health type, and to acquire sub-health models corresponding to the device type and the sub-health type; wherein, the sub-health model is trained based on labeled historical sub-source data; the sub-source data and the sub-health model are in one-to-one correspondence. The second determining module is used to determine the health level corresponding to the health level confirmation instruction based on the sub-source data and the sub-health level model corresponding to the sub-source data.

12. A training device for a model used to determine the health status of a device, characterized in that, include: The acquisition module is used to acquire the initial sub-health model and historical sub-source data corresponding to the device type and sub-health type; The tagging module is used to perform tagging processing on the historical sub-source data based on clustering algorithms and similarity matching algorithms to obtain historical resource data after tagging processing; The training module is used to train the initial sub-health model based on the historical resource data after the labeling process, so as to obtain the sub-health model. The sub-health model is the model described in any one of claims 1-6.

13. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in claims 1-6, or any one of claims 7-10.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in claims 1-6, or any one of claims 7-10.

15. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1-6, or claims 7-10.

Images