Machine health monitoring method and equipment and storage medium
By using a dynamic threshold interval update mechanism based on multi-source verification at the individual data point level in semiconductor manufacturing, the problem of parameter fluctuations caused by equipment aging and environmental disturbances is solved, achieving highly sensitive and accurate machine health monitoring and reducing false alarm rates.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-13
- Publication Date
- 2026-05-12
AI Technical Summary
Existing technologies in semiconductor manufacturing are increasingly sensitive to parameter fluctuations caused by equipment aging and environmental disturbances, leading to increased difficulty in anomaly detection, high false alarm rates, and a lack of ability to discern the actual impact of anomalies, resulting in increased response costs.
A dynamic threshold interval update mechanism based on multi-source verification is adopted, using a single data point as the basic unit. By acquiring machine data and yield detection results, verification data is retrieved to determine whether to update the current threshold interval of abnormal data points. A multi-dimensional verification data system is constructed, and an allowable update label is generated to adjust the threshold interval.
It effectively reduced the false alarm rate caused by harmless process fluctuations, improved the sensitivity and accuracy of anomaly monitoring, and achieved synergistic optimization and risk balance between sensitivity and accuracy in machine health monitoring.
Smart Images
Figure CN122022778A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of semiconductor manufacturing, and more particularly to a method, device, and storage medium for monitoring machine health. Background Technology
[0002] In semiconductor manufacturing, process stability directly determines chip yield and performance consistency. To ensure production quality, wafer fabs generally need to monitor key process parameters (such as film thickness, critical dimension (CD), doping concentration, and electrical parameters) in real time.
[0003] For example, patent application CN121009451A discloses an anomaly identification method and system combining hierarchical parameter analysis, relating to the field of fault detection technology. The method includes: extracting fault detection information of a target semiconductor device; searching a fault record database to determine a first set of sub-events; extracting key parameters to obtain a set of key parameter value sequences for the first sub-events; performing parameter fluctuation duration integration analysis to obtain the first parameter fluctuation duration; extracting associated devices to obtain a set of K secondary faulty components; obtaining a set of K secondary operating parameters; obtaining a set of abnormal components and a set of abnormal detection items; and performing potential fault detection on the target semiconductor device.
[0004] For example, patent application CN118197960A discloses a semiconductor production data monitoring method and system based on a multimodal intelligent agent. This system is applied to a data monitoring system, which includes a data acquisition intelligent agent, a data fusion intelligent agent, a data analysis intelligent agent, and a process analysis intelligent agent. The method includes: the data acquisition intelligent agent acquiring raw production data during the semiconductor production process and sending it to the data fusion intelligent agent; the data fusion intelligent agent receiving the raw production data, performing data preprocessing to obtain production data, and sending the production data to the data analysis intelligent agent; the data analysis intelligent agent receiving the production data, performing data analysis to generate data analysis results, and sending the data analysis results to the process analysis intelligent agent; and the process analysis intelligent agent receiving the data analysis results, evaluating the semiconductor production process based on the data analysis results, and generating a process evaluation result.
[0005] However, as process nodes continue to shrink and process windows become increasingly narrow, parameter fluctuations during manufacturing become more sensitive. At the same time, non-critical variations introduced by factors such as equipment aging and environmental disturbances have increased significantly, making it significantly more difficult to determine anomalies.
[0006] Traditional solutions typically set fixed upper and lower control limits or specification threshold ranges based on historical data. When the data points collected by the machine exceed these ranges, an anomaly is identified, an alarm is triggered, and response measures such as line shutdown, troubleshooting, or equipment maintenance are initiated. This type of solution lacks the ability to discern the actual impact of anomalies, easily misinterpreting harmless fluctuations as malfunctions, resulting in a high false alarm rate, increased response costs, and low practicality. Summary of the Invention
[0007] The main objective of this application is to provide a method, device, and storage medium for monitoring machine health. To solve the aforementioned technical problems, this application specifically adopts the following technical solution: A first aspect of this application is to provide a method for monitoring machine health, the method comprising: S1, acquire machine data and the yield detection result of the machine data, wherein the machine data includes multiple data points; S2, based on the current threshold range corresponding to each data point, perform anomaly monitoring on the machine data and identify abnormal data points that are suspected to be abnormal; S3, when the yield detection result is qualified, retrieve the verification data based on the target acquisition time, target process step, and process time of the abnormal data point within the process step. S4, based on the verification data, determine whether to allow updating the current threshold range of abnormal data points; S5, if so, then update the current threshold range based on the parameter values of the abnormal data points; The verification data includes at least one of the following: the anomaly rate of similar data points, threshold adjustment records, and the usage time information of the target machine and the machine environment information. S3 further includes: S31, retrieving similar data points in the target process steps executed in the current or preset time window that are within a preset time difference from the abnormal data points, and obtaining the abnormality rate and threshold adjustment records of similar data points; S32, obtaining the usage time information and machine environment information of the target machine at the target acquisition time. The S4 further includes: S41, when any one or more verification data meet the preset triggering criteria, determining that the update is permitted and generating the corresponding abnormal data point's permitted update tag.
[0008] In some embodiments, step S31 includes: S311, acquiring local threshold adjustment records and / or local anomaly rates of similar data points in the target machine as first verification data; S312, constructing a target machine group, the target machine group including multiple machines of the same type as the target machine and performing the same process steps; acquiring group threshold adjustment records and / or group anomaly rates of similar data points in the machine group as second verification data.
[0009] In some embodiments, the second verification data includes third verification data and fourth verification data. S312 includes: acquiring a number of prior machines from the target machine group whose usage time is longer than that of the target machine; extracting prior threshold adjustment records and / or prior anomaly rates of similar data points among the prior machines when they reach the corresponding usage time, based on the usage time information at the target acquisition time, as the third verification data; and / or acquiring a number of concurrent machines from the target machine group whose usage time differs from that of the target machine by no more than a preset duration; extracting concurrent threshold adjustment records and / or concurrent anomaly rates of similar data points among the concurrent machines, as the fourth verification data.
[0010] In some embodiments, S41 includes: if the number of threshold adjustment records corresponding to the same type of data points is greater than a preset number threshold, generating an approval update label for the corresponding abnormal data points; and / or, if the abnormality rate is higher than a preset abnormality rate threshold, generating an approval update label for the corresponding abnormal data points.
[0011] In some embodiments, the machine environment information includes at least one of the following: real-time temperature, humidity, air pressure, gas flow rate, particle concentration, cleanliness level of the cavity where the target machine is located, machine maintenance and upkeep records, process gas purity and pressure, and power supply voltage and current stability; S41 includes: acquiring the parameter values of environmental parameters associated with abnormal data points at the target acquisition time, and / or acquiring the usage duration information of the target machine at the target acquisition time as the fifth verification data; if the rate of change of the parameter values is greater than a preset change threshold and / or if the usage duration information is greater than a preset duration threshold, generating an approval update tag for the corresponding abnormal data point.
[0012] In some embodiments, S5 includes: shifting and adjusting the upper and lower limits of the current threshold interval based on the parameter values of abnormal data points; or recalculating the upper and lower limits of the threshold interval based on the numerical distribution of similar abnormal data points; or querying the upper and lower limits of the current threshold interval according to the usage time information and / or equipment environment information of the target equipment, and a preset mapping rule; wherein the width of the threshold interval remains unchanged during the adjustment of the upper and lower limits of the threshold interval.
[0013] In some embodiments, the method further includes: S6, determining that the current threshold range of the abnormal data point should not be updated based on the verification data, and triggering an anomaly prompt for the abnormal data point; and / or, S7, re-performing anomaly monitoring on the machine data using the updated current threshold range; if the abnormal data point is identified as normal, generating a confidence prompt for the abnormal data point.
[0014] In some embodiments, the confidence level indication is determined based on the number of verification data sources and / or verification data source types used to generate the approval update label; S7 includes: if the approval update label is generated solely based on the first verification data, a low confidence level indication is generated, and the current threshold range after manual review and update is pushed; if the approval update label is generated solely based on the fourth or fifth verification data, a medium confidence level indication is generated, and the abnormal data point is rendered and displayed with an emphasis mark in the visualization interface; if the approval update label is generated jointly based on verification data from at least two different sources, or based on a combination of the fifth verification data and any other verification data, a high confidence level indication is generated, and a machine aging assessment report, key component replacement recommendations, or preventive maintenance reminders are output.
[0015] A second aspect of this application is to provide a computer device, the device comprising: Memory, used to store computer programs; A processor is configured to execute the computer program and, in executing the computer program, implement the steps of the machine health monitoring method provided in any embodiment of this application.
[0016] A third aspect of this application is that a computer-readable storage medium is also provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the machine health monitoring method provided in any embodiment of this application.
[0017] Beneficial technical effects: This application provides a method, device, and storage medium for monitoring machine health. Under the premise of qualified yield, it proposes a dynamic threshold interval update mechanism with a single data point as the basic unit and multi-source verification as support. While maintaining high sensitivity, it carefully initiates threshold evaluation and adjustment, effectively reducing the false alarm rate caused by harmless process fluctuations, and ensuring the safety and reliability of the threshold adaptation process. This achieves synergistic optimization and risk balance between sensitivity and accuracy in abnormal monitoring of machine health.
[0018] First, independent judgment and processing are performed on a single data point basis. That is, for each abnormal data point marked as "suspected anomaly," corresponding verification data is retrieved based on its unique context information, and an independent judgment is made as to whether to allow updating the parameter threshold range corresponding to that data point. Through a fine-grained, point-to-point update mechanism, the impact of adjustments is strictly limited to a single data point, avoiding a "one-size-fits-all" adjustment of entire parameter classes or entire process steps. This enhances the responsiveness of health monitoring to localized process drift or individual equipment differences, and improves the feasibility and adaptability of adaptive threshold adjustments in complex and variable production line environments, reducing the risk of anomalies being missed.
[0019] Secondly, a multi-dimensional and multi-layered verification data system was constructed, including the anomaly rate and historical threshold adjustment records of similar data points on the local machine or a group of machines, as well as real-time environmental parameters, effectively identifying parameter deviations that are repetitive, trend-based, or widespread. Based on this, a flexible logic for generating approval update labels was configured: as long as any verification dimension meets the preset triggering conditions (e.g., anomaly rate exceeding a threshold, excessively high historical adjustment frequency, significant environmental disturbances, etc.), an approval update label can be generated, thereby avoiding the omission of reasonable threshold optimization opportunities due to reliance on a single data source and improving sensitivity.
[0020] Meanwhile, to prevent quality risks caused by threshold adjustments, a differentiated response strategy is implemented based on the number and type of sources from which updated labels are permitted: when only local data is used, a low-confidence alert is triggered and manual review is pushed; when only environmental information or group performance is used, a medium-confidence alert is triggered and abnormal data is highlighted; and when multiple independent verification data support the update, a high-confidence conclusion is output and auxiliary information such as machine health assessment and preventive maintenance recommendations are generated in conjunction. Attached Figure Description
[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. The elements or parts in the drawings are not necessarily drawn to scale. Obviously, the drawings described below are some embodiments of this application; for those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.
[0022] Figure 1 This is a schematic flowchart illustrating a machine health monitoring method provided in an embodiment of this application; Figure 2 This is a schematic flowchart illustrating another method for monitoring machine health provided in the embodiments of this application; Figure 3This is a schematic flowchart illustrating an anomaly monitoring process provided in an embodiment of this application; Figure 4 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation
[0023] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.
[0024] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.
[0025] In this document, suffixes such as “module,” “part,” or “unit” used to denote elements are used only for illustrative purposes and have no specific meaning in themselves. Therefore, “module,” “part,” or “unit” may be used interchangeably.
[0026] In this document, the terms "upper," "lower," "inner," "outer," "front," "rear," "one end," and "the other end," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.
[0027] In this document, unless otherwise explicitly specified and limited, the terms "installed," "equipped with," and "connected," etc., should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, a direct connection, or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.
[0028] In this document, the term “and / or” includes any and all combinations of one or more of the listed related items.
[0029] In this article, the term "multiple" means two or more, that is, it includes two, three, four, five, etc.
[0030] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.
[0031] This application provides a method, device, and storage medium for monitoring machine health. Under the premise of acceptable yield, it proposes a dynamic threshold interval update mechanism based on a single data point and supported by multi-source verification. While maintaining high sensitivity, it carefully initiates threshold evaluation and adjustment, effectively reducing the false alarm rate caused by harmless process fluctuations, and ensuring the safety and reliability of the threshold adaptation process. This achieves synergistic optimization and risk balance between sensitivity and accuracy in equipment health monitoring, and resolves the contradiction of "acceptable yield but parameters exceeding limits" in machine data health monitoring.
[0032] In semiconductor manufacturing, hundreds or even thousands of process steps can be categorized into specific procedures based on physicochemical principles and functional properties, such as photolithography, etching, thin film deposition, ion implantation, chemical mechanical polishing, and cleaning. Furthermore, each process step strictly adheres to a predetermined process sequence throughout the overall manufacturing process.
[0033] In this article, "machine" refers to equipment on a production line that performs specific process steps. For example, in a semiconductor wafer fab, it can specifically refer to lithography machines, etching machines, thin film deposition equipment, ion implanters, or chemical mechanical polishing equipment. Machine data is collected and recorded during the operation of each machine for real-time monitoring, fault diagnosis, yield analysis, and predictive maintenance.
[0034] For example, machine data may include engineering data on the health status of the equipment itself (such as time-series data such as temperature, pressure, gas flow, motor speed, vibration spectrum, etc.), process parameters of the processing (such as formula settings, actual execution values, alarm logs, event records), and contextual information related to production results (such as wafer ID, batch number, process step identifier, process sequence label, timestamp, etc.), etc., without limitation.
[0035] In this paper, the machine data includes multiple data points, each representing the measured value of a key process parameter at a specific moment or process node. Each data point can be associated with clear contextual information, including but not limited to: the data acquisition timestamp (i.e., the target acquisition time), the process step it is in (i.e., the target process step), the process time in that step, and the corresponding product type or batch identifier, enabling it to be accurately located and traced back throughout the entire manufacturing process.
[0036] The target acquisition time refers to the specific timestamp of the abnormal data point recorded by the machine's sensors or measurement system, used to associate the equipment's operating status (such as usage duration information) with external environmental conditions (such as machine environment information). The target process step refers to the standardized manufacturing process name to which the abnormal data point belongs (such as gate oxide deposition), which can be used to locate the specific step where the abnormality occurred; the process time refers to the time position within the target process step (such as 60 seconds after the start of the process step), which can be used to accurately locate the specific step stage where the abnormality occurred.
[0037] In this article, yield test results refer to the quality assessment results obtained after performing electrical tests (such as CP test or FT test) on wafers or chips after they have been processed by the equipment, for example, expressed as a percentage of the number of qualified chips out of the total number of tests.
[0038] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.
[0039] Please see Figures 1 to 2 , Figure 1 This is a schematic flowchart illustrating a machine health monitoring method provided in an embodiment of this application. Figure 2 This is a schematic flowchart illustrating another machine health monitoring method provided in the embodiments of this application, such as... Figure 1 and Figure 2 As shown in the figure, this application provides a method for monitoring machine health.
[0040] S1, acquire machine data and the yield detection results of the machine data, wherein the machine data includes multiple data points.
[0041] Specifically, the machine data generated by the target machine within a preset production cycle (such as 7 days or 1 month) is acquired in real time or in batches, and the yield test results corresponding to the batch of wafers or chips are acquired simultaneously.
[0042] S2, based on the current threshold range corresponding to each data point, perform anomaly monitoring on the machine data and identify abnormal data points that are suspected to be abnormal.
[0043] Specifically, the machine data obtained in step S1 is compared and analyzed based on the current threshold range pre-configured for each type of process parameter.
[0044] For example, such as Figure 2 As shown, data points whose parameter values are within the current threshold range are considered normal and are not processed; data points that exceed the current threshold range are marked as "suspected abnormal" abnormal data points and included in the subsequent verification process.
[0045] For example, data points falling within the current threshold range are considered normal and are not processed; data points exceeding the current threshold range but not exceeding the extended tolerance range are marked as "suspected abnormal" abnormal data points and included in the subsequent verification process; data points exceeding the extended tolerance range are marked as high-risk "real abnormalities", triggering an abnormal alarm normally, and are not included in the subsequent verification process (as in steps S3 to S5 in the embodiments of this application).
[0046] The current threshold range refers to the allowable range for normal operation of a specific process parameter under a specific machine, process step, and product type. It is usually composed of upper and lower limits and is used for anomaly monitoring and identification. The initial value of the current threshold range can be derived from historical data statistics or dynamically configured according to process windows, customer specifications, or engineering experience. It is also dynamically adjusted based on yield feedback, verification data, and adaptive update mechanisms (such as steps S3 to S5 in the embodiments of this application) to continuously adapt to normal fluctuations caused by equipment aging, environmental changes, or process evolution, thereby improving the sensitivity and accuracy of anomaly monitoring in machine health.
[0047] The extended tolerance range refers to a buffer zone formed by expanding the current threshold range by a preset percentage (e.g., ±10%~20%) or a fixed margin. It should be understood that the extended tolerance range is used to define data points that are slightly out of bounds and severely deviated from the target range. It limits suspected anomalies to those that are slightly out of bounds, possibly caused by non-critical disturbances, and have not yet affected yield. This allows for the timely response to high-risk anomalies while capturing potentially harmless ones, balancing sensitivity and safety.
[0048] S3, when the yield detection result is qualified, the verification data is retrieved based on the target acquisition time, target process step, and process time of the abnormal data point within the process step.
[0049] Specifically, such as Figure 2As shown, if the yield test result is confirmed to be qualified, it indicates that the overall quality of the current production batch is under control. Therefore, the suspected abnormal data points meet the prerequisite for entering the threshold update evaluation process. Based on the target acquisition time, target process step, and its operation time within that step associated with the abnormal data point, the context of the data point is constructed. Using this as an index, verification data from multiple sources, including the equipment database, historical operation logs, and machine groups, is retrieved to ensure a high degree of referentiality between the verification data and the abnormal data.
[0050] In some embodiments, such as Figure 2 As shown, if the yield test result is unqualified, an abnormal alarm will be issued for the abnormal data points.
[0051] S4. Based on the verification data, determine whether to allow updating the current threshold range of abnormal data points.
[0052] Specifically, based on the retrieved multi-source verification data, it is determined whether the anomaly is repetitive or trend-like, widespread, and related to equipment status or environmental disturbances. Then, it is determined whether to allow updating the current threshold range corresponding to the anomaly data point. For anomaly data points that are allowed to be updated, an update permission label is generated.
[0053] S5, if so, then update the current threshold range based on the parameter values of the abnormal data points.
[0054] Specifically, if an abnormal data point has an allowable update label, the current threshold range corresponding to the abnormal data point is dynamically updated.
[0055] In this embodiment, independent judgment and processing are performed on a single data point as the basic unit. Specifically, for each abnormal data point marked as "suspected anomaly," corresponding verification data is retrieved based on its unique context information, and an independent judgment is made as to whether to allow updating the parameter threshold range corresponding to that data point. Through a fine-grained, point-to-point update mechanism, the impact of adjustments is strictly limited to a single data point, avoiding a "one-size-fits-all" adjustment of entire parameter classes or entire process steps. This enhances the responsiveness of health monitoring to localized process drift or individual equipment differences, and improves the feasibility and adaptability of adaptive threshold adjustment in complex and variable production line environments, reducing the risk of anomalies being missed due to threshold adjustments.
[0056] In some embodiments, the verification data includes at least one of the following: the anomaly rate of similar data points, threshold adjustment records, and the usage duration information of the target machine and the machine environment information.
[0057] Similar data points refer to a group of historical or contemporaneous data points that are comparable to the current anomalous data point in the critical process and equipment context of semiconductor manufacturing. As the basis for verification analysis, verification data must originate from completely identical process operations; therefore, similar data points must meet the following basic screening criteria: A. Belonging to the same target process step as the anomalous data point; B. Belonging to the same process parameter type as the anomalous data point (e.g., both are gate oxide thickness); C. Belonging to the same or similar product type and process node as the anomalous data point. Based on this, similar data points are further divided into two categories: The first category is data points of the same type as the target machine, that is, data points generated by the target machine itself in other batches or time periods. In addition to meeting the basic screening conditions AC above, data points of the same type as the target machine must also meet the following screening conditions: D. Belong to the same or similar process time as the abnormal data points (such as 30±5 seconds after the start of the step) to eliminate normal fluctuation interference caused by the dynamic process.
[0058] The second category is cross-machine similar data points, which are data points generated by other machines. In addition to meeting the basic screening conditions AC mentioned above, cross-machine similar data points must also meet at least some of the following screening conditions: D. Belong to the same or similar process time as the abnormal data point (e.g., 30±5 seconds after the start of the step); E. Be generated by the same manufacturer or model of machine as the abnormal data point, ensuring that the equipment hardware and control logic are consistent; F. The machine environment information at the time of acquisition is within a preset similarity range to the machine environment information at the target acquisition time; G. Be generated by a reference machine with a similar equipment usage cycle to the target machine.
[0059] In some embodiments, to accurately locate reference machines and obtain cross-machine similar data points with reference value, the method further includes: constructing a machine profile of the machine, the machine profile including at least one of machine type and process technology; and clustering multiple machines into one or more machine groups based on the machine profile. The machine profile may optionally include one or more of the following: basic equipment attributes (such as equipment manufacturer, specific model, core hardware architecture version, etc.), executed process attributes (such as the process node to which it belongs, the specific process steps executed, key process parameter windows, etc.), and operating environment attributes (also known as machine environment information, such as temperature, humidity, etc.).
[0060] In some embodiments, machines are clustered based on their basic equipment attributes, process attributes, and operating environment attributes to obtain different types of machine groups. The machines in a group can be distributed across one or more production lines, and their physical attributes and process functions are highly homogeneous. Furthermore, the inclusion of machines with similar functions but significant differences in internal structure or control logic is avoided, ensuring that the operating performance of each machine within the group has good comparability and reference value.
[0061] It should be understood that the target machine group includes the target machine and several reference machines of the same type as the target machine.
[0062] For example, the reference machines include several concurrent machines whose usage time differs from that of the target machine by no more than a preset duration. These are machines in the same production batch or at the same lifecycle stage as the target machine, such as those all operating between the 30th and 40th day after production commencement. The preset duration can be flexibly set according to the actual application scenario, such as 10 days.
[0063] For example, the reference machine includes a prior machine with a longer service life than the target machine. It should be understood that a prior machine refers to other machines that belong to the same equipment type and process steps as the target machine, and whose cumulative operating years are longer than those of the target machine. Because prior machines have a longer service life, their historical databases contain complete records of threshold updates from new machine commissioning to the current aging stage, which are of reference value.
[0064] In some embodiments, S3 further includes: S31, retrieving similar data points in the target process step executed in the current or preset time window that are within a preset time difference from the process time of the abnormal data point, and obtaining the abnormality rate and threshold adjustment record of the similar data points; S32, obtaining the usage time information and machine environment information of the target machine at the target acquisition time.
[0065] Specifically, from the current production batch or a preset time window (such as the last 7 days), all records that execute the same target process steps are retrieved, and similar data points whose process time and the deviation from the current abnormal data point are within a preset time difference (such as 5 seconds) are selected, such as multiple similar data points on the same machine or multiple similar data points across different machines. The abnormality rate of these similar data points and their corresponding threshold adjustment records are then calculated.
[0066] Furthermore, information on the usage duration of the target machine at the target acquisition time (such as cumulative operating hours) and real-time machine environment information (such as cavity temperature) is obtained to assess the potential impact of equipment aging or environmental disturbances on parameter drift.
[0067] The preset time window and preset time difference can be flexibly determined based on specific process characteristics and engineering experience, and the specific values are not limited here. For example, assuming the target process step is CMP grinding, the data type is grinding pressure, the abnormal data point is located at 120 seconds after the start of this step, and the preset time difference is ±10 seconds, then similar data points include data with the same parameters collected between 110 and 130 seconds in historical batches.
[0068] The anomaly rate refers to the frequency at which multiple similar data points are judged as abnormal, and can be expressed as the ratio of the number of abnormal data points to the total number of data points. In some embodiments, the anomaly rate is calculated based on data from machines with acceptable yield test results, and is used to measure whether a certain type of parameter deviation is a harmless fluctuation rather than a failure that truly affects product quality. In some embodiments, the anomaly rate is calculated based on the number of suspected abnormal data points. In other embodiments, the anomaly rate is calculated based on the number of suspected and actual abnormal data points.
[0069] The threshold adjustment record refers to the modification log of historical threshold ranges for the same data points, including the time of each adjustment, the upper and lower limits of the threshold before and after the adjustment, and the data type of the verification data on which it was based. If a parameter frequently triggers threshold adjustments within a certain time window, and the yield remains acceptable, it may indicate that the original threshold is no longer suitable for the current equipment status or process conditions.
[0070] In some embodiments, the usage time information of the machine is determined based on the cumulative usage time of the machine at the target acquisition time.
[0071] In some embodiments, the equipment environment information includes various environmental parameters that have a direct impact on process stability. The specific type of environmental information collected can be flexibly selected and adjusted according to actual process requirements, equipment configuration, or monitoring focus, and the environmental parameters to be collected can be associated with the corresponding data points. For example, in etching processes, more attention is paid to gas flow rate and chamber pressure, while in thin film deposition, the focus may be on temperature uniformity and particle control.
[0072] For example, the machine environment information includes at least one of the following: real-time temperature, humidity, air pressure, gas flow rate, particle concentration, cleanliness level of the cavity where the target machine is located, machine maintenance and upkeep records, process gas purity and pressure, and power supply voltage and current stability.
[0073] The embodiments of this application construct a multi-dimensional and multi-level verification data system, including first verification data based on the same type of data points on the local machine, second verification data based on the same type of data points across different machines (which is further divided into third verification data based on the prior machine and fourth verification data based on the concurrent machine), and fifth verification data based on environmental parameters.
[0074] In some embodiments, when the data points of the same type are local data points of the same type, step S31 includes: S311, acquiring local threshold adjustment records and / or local anomaly rates of the data points of the same type in the target machine as first verification data.
[0075] In some embodiments, when the data points of the same type are cross-machine type data points, step S31 includes: S312, constructing a target machine group, the target machine group including multiple machines of the same type as the target machine and performing the same process steps; obtaining the group threshold adjustment record and / or group anomaly rate of the data points of the same type in the machine group as the second verification data.
[0076] In some embodiments, the second verification data includes third verification data and fourth verification data.
[0077] In some embodiments, S312 includes: acquiring a number of prior machines from the target machine group whose usage time is longer than that of the target machine; and extracting prior threshold adjustment records and / or prior anomaly rates of similar data points in the prior machines when the prior machines have reached the corresponding usage time, based on the usage time information at the target acquisition time, as third verification data.
[0078] In some embodiments, S312 includes: obtaining from the target machine group a number of concurrent machines whose usage time differs from that of the target machine by no more than a preset time; extracting concurrent threshold adjustment records and / or concurrent anomaly rates of similar data points from the concurrent machines as fourth verification data.
[0079] In some embodiments, S4 further includes: S41, when any one or more verification data meet the preset trigger criteria, determining that the update is permitted and generating an approved update tag for the corresponding abnormal data point.
[0080] If any of the above verification data meets the preset trigger criteria, an allowable update label is generated for that abnormal data point, which serves as the basis for deciding whether to allow adjustment of its current threshold range. This avoids missing reasonable threshold optimization opportunities due to reliance on a single data source and improves sensitivity.
[0081] Among them, preset trigger criteria refer to a set of pre-defined judgment rules or threshold conditions used to evaluate whether various verification data are sufficient to support updating the threshold range for the current abnormal data points. These criteria are configured by process engineers or the system based on historical experience, statistical analysis, and risk control strategies, and can exist in the form of quantitative indicators, which are not limited here.
[0082] In some embodiments, S41 includes: if the number of threshold adjustment records corresponding to the same type of data points is greater than a preset number threshold, generating an approval update tag for the corresponding abnormal data points.
[0083] In some embodiments, S41 includes: if the anomaly rate is higher than a preset anomaly rate threshold, speculating to generate an allowable update label for the corresponding abnormal data point.
[0084] The preset quantity threshold refers to the minimum number of times the threshold adjustment occurs for the same type of data points (e.g., 10 times); the preset anomaly rate threshold refers to the maximum proportion of the same type of data points that are judged as anomalies (e.g., 60%). Specific values can be pre-configured by process engineers based on historical data, process windows, and risk tolerance, and can be dynamically adjusted based on the current sample size of the same type of data points. For example, when the number of similar data points is small, the threshold can be appropriately relaxed; the preset quantity threshold can be set to 5 times.
[0085] It should be understood that by analyzing the number of threshold adjustment records and the anomaly rate of similar data points, it is possible to effectively identify whether the current parameter deviation is repetitive, trend-based, or widespread, thereby distinguishing whether its root cause stems from changes in equipment status (such as component aging), steady-state process migration (such as material batch differences), or is merely caused by occasional noise or transient disturbances. When adjustment records are frequent or the anomaly rate remains consistently high, and the corresponding batch yield remains acceptable, it indicates that the existing threshold range can no longer accurately reflect the normal behavior range of the current process. In this case, generating an "approved update" label to support reasonable correction of the current threshold range not only conforms to the actual production status but also avoids continuously triggering invalid alarms due to adhering to outdated and rigid thresholds, improving the practicality and accuracy of anomaly monitoring in machine health.
[0086] In some embodiments, S41 includes: acquiring the parameter values of environmental parameters associated with abnormal data points at the target acquisition time, as the fifth verification data; if the rate of change of the parameter values is greater than a preset change threshold, generating an approval update tag for the corresponding abnormal data point.
[0087] Specifically, the real-time parameter values of the environmental parameters associated with the abnormal data points are obtained at the target acquisition time. If the environmental parameter values change significantly, it is inferred that the current anomaly may be caused by changes in the external environment. An update label is generated for the abnormal data points to accommodate such interpretable and reasonable offsets and avoid false alarms.
[0088] In some embodiments, S41 includes: acquiring the usage duration information of the target machine at the target acquisition time as the fifth verification data; if the usage duration information is greater than a preset duration threshold, generating an approval update tag for the corresponding abnormal data point.
[0089] The preset duration threshold can be flexibly set according to factors such as the design life of different machines, actual usage intensity, maintenance cycle and specific process requirements. It is used to determine whether the target machine has entered the stage of performance aging or high parameter drift. The specific value is not limited here.
[0090] For example, if equipment has been running for more than 10,000 hours and has entered the aging stage, the deviation in current process parameters is considered to be due to the natural degradation of equipment performance rather than a sudden failure. In this case, an update permission label will be generated for the corresponding abnormal data point to encompass systematic and predictable parameter drift caused by equipment aging, avoiding a large number of harmless false alarms caused by adhering to the initial anomaly identification criteria.
[0091] In some embodiments, S5 includes: the existence of an "update allowed" label, indicating that the current threshold range of the abnormal data point is allowed to be updated; the absence of an "update allowed" label, indicating that the current threshold range of the abnormal data point is not allowed to be updated, and the abnormal data point will subsequently trigger an abnormal alarm normally.
[0092] In some embodiments, S5 includes: shifting and adjusting the upper and lower limits of the current threshold interval based on the parameter values of abnormal data points.
[0093] For example, when the distance between the actual parameter value of an abnormal data point and the upper or lower limit of the current threshold interval is less than a preset tolerance threshold (e.g., 1 measurement unit), it indicates that the offset is marginal and slightly exceeds the limit. Shifting the threshold interval incorporates the parameter value of the abnormal data point into the updated current threshold interval, thereby reclassifying it as a normal data point. In this case, the abnormal data point is the upper or lower limit of the updated current threshold interval. For example, if the target process step is CMP grinding, the data type is grinding pressure (e.g., temperature, voltage, etc.), and the value of the abnormal data point is 8.2 psi; the current threshold interval is [7.0, 8.0] psi; if a shift adjustment is used while keeping the width unchanged, the new current threshold interval is [7.2, 8.2] psi.
[0094] For example, the parameter offset direction is determined based on the parameter value of the abnormal data point, and the entire interval is shifted by a fixed step size or proportion along the parameter offset direction. The entire threshold interval is then shifted slightly along the parameter offset direction (e.g., if it is higher than the upper limit, it is shifted towards the upper limit; if it is lower than the lower limit, it is shifted towards the lower limit). This type of adjustment has a small range and the risk is controllable, avoiding over-response caused by a single edge anomaly. Moreover, the updated interval may not contain the original abnormal data point.
[0095] In some embodiments, S5 includes: recalculating the upper and lower limits of the threshold interval based on the numerical distribution of similar abnormal data points. Specifically, the current threshold interval that can cover most data points can be determined by combining the historical numerical distribution of similar data points or the threshold interval currently used for similar data points. The updated interval may not necessarily include the original abnormal data points. For example, a weighted average or union of multiple threshold intervals currently used for similar data points can be performed to generate a more representative unified interval.
[0096] In some embodiments, S5 includes: querying a preset mapping rule, the upper and lower limits of the current threshold range, based on the target machine's usage time information and / or machine environment information. The preset mapping rule can be constructed based on historical data analysis or process expert experience. For example, when the equipment has been running for more than 10,000 hours and enters the aging stage, or when the particle concentration is higher than a preset level, the upper and lower limits of the threshold range of the corresponding process parameters are increased by 2% overall.
[0097] In some embodiments, the width of the threshold interval remains unchanged during the adjustment of the upper and lower limits. It should be understood that keeping the width constant during threshold interval adjustment helps to adapt to systematic shifts in process parameters (such as equipment aging or environmental changes) while maintaining the detection sensitivity to abnormal fluctuations, and avoids an increased risk of missed detection due to relaxing the anomaly identification threshold.
[0098] like Figure 2 As shown, in some embodiments, the method includes: S6, if it is determined that the current threshold range of the abnormal data point should not be updated (i.e., there is no label that allows updating), then an abnormal alarm is triggered for the abnormal data point. Here, an abnormal alarm refers to a real-time alarm signal issued for high-risk real anomalies, used to immediately alert the user and prompt them to take appropriate measures, such as recording alarm logs, suspending machine operation, initiating equipment diagnostic processes, or notifying process / equipment engineers to intervene.
[0099] Please see Figure 3 , Figure 3 This is a schematic flowchart illustrating the re-execution of anomaly monitoring provided in an embodiment of this application. Figure 3 As shown, S5 is followed by S7.
[0100] S7, re-execute anomaly monitoring on the machine data using the updated current threshold range. For example... Figure 3 As shown, it determines whether the parameter value of the abnormal data point is within the updated current threshold range. If it is, it is updated to a normal data point; otherwise, it is still identified as an abnormal data point.
[0101] S71, if the abnormal data point is identified as normal, a confidence level prompt is generated for the abnormal data point.
[0102] It should be noted that anomaly alarms differ from other auxiliary, non-interrupting alerts. Anomaly alarms target high-risk, genuine anomalies that may disrupt normal production processes (such as automatic shutdown or process step locking) or force manual intervention. They require immediate allocation of engineering resources for investigation and handling, incurring significant manpower and time costs, in order to prevent potential quality defects or equipment damage from escalating further and to ensure production safety and product yield.
[0103] The credibility indicator is primarily used to assist in threshold update decisions. The manual review it triggers mainly assesses whether the current threshold range should be adjusted, rather than addressing equipment or process malfunctions. This type of review can be completed by process engineers during routine inspections or data analysis periods without affecting production line operations. It consumes fewer resources and aims to optimize the monitoring system's own parameters, rather than responding to immediate risks.
[0104] In some embodiments, to prevent quality risks associated with threshold adjustments, the confidence level is determined based on the number of verification data sources and / or verification data source types that generate the approved update label, in order to implement a differentiated response strategy.
[0105] In some embodiments, such as Figure 3 As shown, S7 includes: S711, if the approved update label is generated only based on the first verification data, a low confidence prompt is generated, and the current threshold range after manual review and update is pushed.
[0106] It should be understood that the sample size is limited and there is a lack of external references, which may lead to randomness or individual bias. It is difficult to distinguish between real process drift and isolated noise. Therefore, manual intervention is required to confirm the rationality of the threshold range adjustment and reduce the quality risks brought about by the threshold adjustment.
[0107] S712, if the approved update label is generated only based on the fourth or fifth verification data, a medium confidence prompt is generated, and the abnormal data point is rendered and displayed in the visualization interface with an emphasis mark.
[0108] It should be understood that environmental information or group performance has a certain degree of objectivity (such as synchronous offset of similar devices or environmental changes), but a single dimension is still insufficient to completely eliminate the risk of local anomalies. Emphasis is placed on marking (such as highlighting, flashing borders or color coding) to help operators quickly pay attention but not perform maintenance actions.
[0109] S713, if the approved update tag is generated based on verification data from at least two different sources, or based on a combination of the fifth verification data and any other verification data, a high confidence prompt is generated, and a machine aging assessment report, key component replacement recommendations, or preventive maintenance prompts are output.
[0110] It should be understood that when multiple independent verification data points jointly support an update and / or objective fifth verification data supports an update, the offset is deemed to have multidimensional consistency and interpretability, and is highly likely to reflect actual changes in equipment status. Therefore, a threshold update is adopted, and equipment maintenance-related prompts are synchronized. For example, assuming the target process step is CMP grinding, the data type is grinding head pressure, and the abnormal data point is located 180 seconds after the start of this step, if the threshold for this data is adjusted, it may indicate a problem with the grinding head components, and corresponding maintenance or replacement recommendations for the grinding head components can be output.
[0111] S72, if the abnormal data point is identified as abnormal, then the abnormal data point is sent for manual review.
[0112] If the abnormal data point is still identified as abnormal after the threshold range is updated (i.e. its parameter value still exceeds the current threshold range after the update), it is considered as an offset that cannot be reasonably explained by the threshold adaptive mechanism or a threshold that needs to be adjusted by a large margin. Its abnormal state will be retained and the manual intervention process will be initiated.
[0113] For example, based on the requirement of real-time response, the priority of manual review triggered by abnormal alarm is higher than that of manual review triggered by step S72, and the priority of manual review triggered by step S72 is higher than that of manual review triggered by step S711, so as to avoid leaving real faults and ensure process quality and equipment safety.
[0114] It should be understood that such situations may arise from extreme deviations in parameter values, sudden equipment failures, or insufficient similar data (for example, the first occurrence of a deviation caused by equipment aging, with insufficient historical similar data to support a large adjustment), which limits the range of threshold updates. As a result, due to a conservative strategy, only a limited range of adjustments are made, failing to cover the parameter values of the abnormal data points.
[0115] If subsequent manual verification confirms that the abnormal data point is indeed normal, and the current threshold range is manually configured or calibrated accordingly, then the data point and its associated updated threshold range will be recorded as valid prior data. This type of data can serve as a reference for adaptive threshold adjustments in similar scenarios in the future (such as the same machine model, the same process steps, and similar process times). When similar offsets occur again, the accuracy of threshold range adjustments can be improved based on this prior data, thereby avoiding repeated high-cost manual intervention processes, improving automation response capabilities, and reducing the interference of false alarms on production line efficiency.
[0116] In some embodiments, a time limit (e.g., 30 minutes, one batch cycle, or until the next process step reset) is set for the permission to update the label generated based on the machine environment information. Within the time limit, the updated current threshold range is valid; after the time limit is exceeded, it automatically reverts to the original current threshold range, or the current threshold range is re-evaluated based on steps S3 to S5. It should be understood that environmental disturbances are mostly instantaneous or short-term events, and the parameter offsets they cause are only reasonable within a specific time window. Once the environment returns to stability, continuing to use the threshold range relaxed due to the temporary disturbance may lead to missed detection of subsequent real anomalies.
[0117] The methods and apparatus of this application can be used in a wide variety of general-purpose or special-purpose computing system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer terminal devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc.
[0118] Please see Figure 4 , Figure 4 This is a schematic block diagram illustrating the structure of a computer device according to an embodiment of this application. The computer device may be a terminal device or a server.
[0119] For example, the above method can be implemented as a computer program, which can be used in, for example... Figure 4 It runs on the computer device shown.
[0120] like Figure 4 As shown, the computer device includes a processor, memory, and network interface connected via a system bus, wherein the memory may include non-volatile storage media and internal memory.
[0121] Non-volatile storage media can store operating systems and computer programs. These computer programs include program instructions that, when executed, cause the processor to perform any machine health monitoring method and its specific implementation steps.
[0122] The processor provides computing and control capabilities, supporting the operation of the entire computer device.
[0123] Internal memory provides an environment for the execution of computer programs in non-volatile storage media. When the computer program is executed by the processor, it enables the processor to execute any machine health monitoring method and the specific implementation steps of the method.
[0124] This network interface is used for network communication, such as sending assigned tasks.
[0125] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.
[0126] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: S1, acquire machine data and the yield detection result of the machine data, wherein the machine data includes multiple data points; S2, based on the current threshold range corresponding to each data point, perform anomaly monitoring on the machine data and identify abnormal data points that are suspected to be abnormal; S3, when the yield detection result is qualified, retrieve the verification data based on the target acquisition time, target process step, and process time of the abnormal data point within the process step. S4, based on the verification data, determine whether to allow updating the current threshold range of abnormal data points; S5, if so, then update the current threshold range based on the parameter values of the abnormal data points.
[0127] The verification data includes at least one of the following: anomaly rate of similar data points, threshold adjustment records, and target machine usage duration information and machine environment information. In one embodiment, the processor is used to run a computer program stored in the memory to perform the following steps: S31, retrieve similar data points in the target process steps executed in the current or preset time window that are within a preset time difference from the abnormal data points, and obtain the abnormality rate and threshold adjustment records of similar data points; S32, obtain the usage time information and machine environment information of the target machine at the target acquisition time. S41, when any one or more verification data meet the preset trigger criteria, determine that the update is allowed and generate the corresponding abnormal data point's allow update tag.
[0128] For example, the processor is used to run a computer program stored in the memory, and is also used to implement the steps and specific implementation steps of the machine health monitoring method provided in any embodiment of this application, which will not be repeated here.
[0129] The embodiments of this application also provide a computer-readable storage medium storing a computer program, the computer program including program instructions, and the processor executing the program instructions to implement the steps and specific implementation steps of the machine health monitoring method provided in any of the embodiments of this application.
[0130] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.
[0131] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for monitoring machine health, characterized in that, The method includes: S1, acquire machine data and the yield detection result of the machine data, wherein the machine data includes multiple data points; S2, based on the current threshold range corresponding to each data point, perform anomaly monitoring on the machine data and identify abnormal data points that are suspected to be abnormal; S3, when the yield detection result is qualified, retrieve the verification data based on the target acquisition time, target process step, and process time of the abnormal data point within the process step. S4, based on the verification data, determine whether to allow updating the current threshold range of abnormal data points; S5, if so, then update the current threshold range based on the parameter values of the abnormal data points; The verification data includes at least one of the following: the anomaly rate of similar data points, threshold adjustment records, and the usage time information of the target machine and the machine environment information. S3 further includes: S31, retrieving similar data points in the target process steps executed in the current or preset time window that are within a preset time difference from the abnormal data points, and obtaining the abnormality rate and threshold adjustment records of similar data points; S32, obtaining the usage time information and machine environment information of the target machine at the target acquisition time. The S4 further includes: S41, when any one or more verification data meet the preset triggering criteria, determining that the update is permitted and generating the corresponding abnormal data point's permitted update tag.
2. The method according to claim 1, characterized in that, S31 includes: S311, Obtain the local threshold adjustment records and / or local anomaly rate of similar data points in the target machine as the first verification data; S312, construct a target machine group, which includes multiple machines of the same type as the target machine and performing the same process steps; obtain the group threshold adjustment records and / or group anomaly rate of similar data points in the machine group as the second verification data.
3. The method according to claim 2, characterized in that, The second verification data includes the third verification data and the fourth verification data. S312 includes: From the target machine group, obtain several prior machines whose usage time exceeds that of the target machine; based on the usage time information at the target collection time, extract the prior threshold adjustment records and / or prior anomaly rates of similar data points among the prior machines when they have reached the corresponding usage time, as third verification data; and / or, From the target machine group, obtain several synchronous machines whose usage time differs from that of the target machine by no more than a preset time; extract the synchronous threshold adjustment records and / or synchronous anomaly rates of similar data points from the synchronous machines as the fourth verification data.
4. The method according to any one of claims 1 to 3, characterized in that, S41 includes: If the number of threshold adjustment records corresponding to the same type of data points exceeds the preset threshold, generate an approval update label for the corresponding abnormal data points; and / or, If the anomaly rate is higher than a preset anomaly rate threshold, an approval update label is generated for the corresponding anomaly data point.
5. The method according to any one of claims 1 to 3, characterized in that, The machine environment information includes at least one of the following: real-time temperature, humidity, air pressure, gas flow rate, particle concentration, cleanliness level of the cavity where the target machine is located, machine maintenance and upkeep records, process gas purity and pressure, and power supply voltage and current stability; S41 includes: Obtain the parameter values of the environmental parameters associated with the abnormal data points at the target acquisition time, and / or obtain the usage duration information of the target machine at the target acquisition time as the fifth verification data; If the rate of change of the parameter value is greater than the preset change threshold and / or if the duration information used is greater than the preset duration threshold, generate an approval update label for the corresponding abnormal data point.
6. The method according to claim 1, characterized in that, S5 includes: Based on the parameter values of abnormal data points, the upper and lower limits of the current threshold range are shifted and adjusted. Alternatively, based on the numerical distribution of similar abnormal data points, recalculate the upper and lower limits of the threshold interval; Alternatively, based on the target machine's usage duration information and / or machine environment information, query the preset mapping rules, and the upper and lower limits of the current threshold range; During the adjustment of the upper and lower limits of the threshold interval, the width of the threshold interval remains unchanged.
7. The method according to claim 1, characterized in that, The method further includes: S6, based on the verification data, if it is determined that the current threshold range of the abnormal data point should not be updated, then an exception prompt is triggered for the abnormal data point; and / or, S7, re-execute anomaly monitoring on machine data using the updated current threshold range; if the abnormal data point is identified as normal, generate a confidence level prompt for the abnormal data point.
8. The method according to claim 7, characterized in that, The credibility indicator is determined based on the number of verification data sources and / or verification data source types that generate the approved update label; S7 includes: If the approved update label is generated solely based on the first verification data, a low confidence level prompt is generated, and the current threshold range after manual review and update is pushed to the system. If the approved update label is generated only based on the fourth or fifth verification data, a medium confidence level prompt is generated, and the abnormal data point is rendered and displayed with an emphasis mark in the visualization interface; If the approved update label is generated based on verification data from at least two different sources, or based on a combination of the fifth verification data and any other verification data, a high-confidence prompt is generated, and a machine aging assessment report, key component replacement recommendations, or preventative maintenance prompts are output.
9. A computer device, characterized in that, The device includes: Memory, used to store computer programs; A processor is configured to execute the computer program and, in executing the computer program, implement the machine health monitoring method as described in any one of claims 1 to 8.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to implement the machine health monitoring method as described in any one of claims 1 to 8.