A device health threshold management method, apparatus, device, and medium

Abstract

The present application belongs to the technical field of industrial equipment maintenance, and provides a device health threshold management method, device, equipment and medium, which are used for improving the explainability of rules, the working condition adaptability and the anti-pollution ability to abnormal data. According to the current working condition parameters of the target device, the target rule including the operation parameters required for dynamically calculating the health threshold is determined, after the target window data and the target expansion mode are determined according to the target rule, the target health threshold interval is calculated based on the target window data through the target expansion mode. In this way, the corresponding specific rule is traced back through the determined target rule, the explainability is strong, and through the target rule, the target window data and the target expansion mode are determined, the automatic switching of the rule and the self-correction of the parameters can be realized, the working condition adaptability and the anti-pollution ability to abnormal data are improved.

Description

Technical Field

[0001] This invention relates to the field of industrial equipment maintenance technology, and more specifically, to a method, apparatus, device, and medium for managing equipment health thresholds. Background Technology

[0002] Threshold settings play a crucial role in the healthy operation of equipment. Existing methods for setting equipment health thresholds include data-driven methods, manual setting methods, and hybrid methods.

[0003] Data-driven methods such as K-means clustering, support vector regression, and LSTM time series prediction suffer from poor interpretability due to their black-box model. Manually set methods, such as setting fixed thresholds or segmented thresholds, cannot track performance degradation and changes in operating conditions, resulting in poor adaptability. Hybrid methods, such as those based on simple combinations of rules and statistics, are easily contaminated by anomalous data.

[0004] Therefore, existing methods for setting device health thresholds suffer from poor interpretability and adaptability, and are easily contaminated by abnormal data. Summary of the Invention

[0005] This application aims to provide a method, apparatus, device, and medium for managing device health thresholds, in order to address the shortcomings of existing methods for setting device health thresholds, such as poor interpretability and adaptability, and susceptibility to contamination by abnormal data.

[0006] Firstly, this application provides a method for managing device health thresholds, comprising: The target rules are determined based on the current operating parameters of the target equipment; the target rules include the operating parameters required for dynamically calculating the health threshold. The target window data and target expansion method are determined according to the target rules; the target expansion method includes at least one of percentile weighting algorithm, interface coefficient adaptive algorithm and / or incremental percentile update algorithm. Based on the target window data, the target health threshold range is calculated through target expansion.

[0007] The technical solution provided in this application offers at least the following beneficial effects: It determines target rules, including the operating parameters required for dynamically calculating health thresholds, based on the current operating parameters of the target equipment. After determining the target window data and target expansion method based on the target rules, it calculates the target health threshold range based on the target window data and the target expansion method. Thus, this application traces the corresponding specific rules through the determined target rules, resulting in strong interpretability. Furthermore, by determining the target window data and target expansion method through the target rules, it enables automatic rule switching and parameter self-correction, improving operating condition adaptability and resistance to abnormal data contamination.

[0008] Secondly, this application provides a device for managing device health thresholds, comprising: The acquisition module is used to determine the target rules based on the current operating parameters of the target device; the target rules include the operating parameters required for dynamically calculating the health threshold. The processing module is used to determine the target window data and the target expansion method according to the target rules; the target expansion method includes at least one of percentile weighting algorithm, interface coefficient adaptive algorithm and / or incremental percentile update algorithm. The management module is used to calculate the target health threshold range based on the target window data and through target expansion.

[0009] Thirdly, this application provides an electronic device comprising: a processor and a memory; the memory storing processor-executable instructions; when the processor is configured to execute the instructions, causing the electronic device to implement the method of the first aspect described above.

[0010] Fourthly, this application provides a computer-readable storage medium comprising: computer software instructions; which, when executed in an electronic device, cause the electronic device to implement the method described in the first aspect.

[0011] Fifthly, this application provides a computer program product comprising a computer program that, when run on an electronic device, causes the electronic device to perform the method described in the first aspect.

[0012] The beneficial effects of the second to fifth aspects mentioned above are described in the corresponding description of the first aspect and will not be repeated here. Attached Figure Description

[0013] Figure 1 This application provides a method for managing device health thresholds. Figure 2 A flowchart illustrating an independent window manager provided in an embodiment of this application; Figure 3 A flowchart illustrating an independent window manager provided in an embodiment of this application; Figure 4 This is a schematic diagram illustrating the composition of a device health threshold management apparatus provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0014] The terms “comprising” and “having”, and any variations thereof, used in the description of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0015] It should be noted that in the embodiments of this application, the words "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0016] In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0017] The following description, in conjunction with the accompanying drawings, details a device health threshold management method provided in this application.

[0018] Figure 1 This application provides a method for managing device health thresholds. Figure 2 This application provides an example diagram of a rule base. Figure 3 A flowchart illustrating an independent window manager provided in this application embodiment, combined with... Figures 1 to 3 The following describes a device health threshold management method provided in an embodiment of this application: S101. Determine the target rules based on the current operating parameters of the target equipment.

[0019] The target rules include the runtime parameters required for dynamically calculating the health threshold, which guide the kernel selection, parameter configuration, and response strategy for subsequent threshold calculations.

[0020] Match the corresponding operating conditions in the preset rule base according to the current operating condition parameters, and obtain the associated target rules according to the corresponding operating conditions.

[0021] Among them, such as Figure 2 As shown, the preset rule base stores multiple rules for different devices under different operating conditions; the items in the rules include at least window length, baseline coefficient, percentile weighted enable flag, interface coefficient adaptive enable flag, incremental percentile update enable flag, and action strategy.

[0022] The window length is used to indicate the range of historical data collected from the target device. The baseline coefficient is the IQR (Interquartile Range) coefficient. The percentile weighting enable flag is used to indicate whether the percentile weighting algorithm is started. The interface coefficient adaptive enable flag is used to indicate whether the interface coefficient adaptive algorithm is started. The incremental percentile update enable flag is used to indicate whether the incremental percentile update algorithm is started.

[0023] Specifically, this step matches operating conditions with rules using a pre-set rule base. The pre-set rule base stores multiple rules for different devices and operating conditions. Each rule uses a triplet structure, comprising a condition domain, a parameter domain, and an action domain, as shown in Table 1 below: Table 1

[0024] The condition field is a combination of logical conditions such as equipment operating conditions, environmental parameters, and time windows, such as "wind speed ≥ 3m / s and power ≥ 100kW", used to determine whether the current operating condition triggers the corresponding rule; the parameter field is the operating parameters of dynamic statistical calculation, including at least window length, baseline coefficient (i.e., IQR coefficient), percentile weighting enable flag, interface coefficient adaptive enable flag, incremental percentile update enable flag, baseline offset, etc., used to define the operating mode of the threshold calculation kernel; the action field is the response strategy after exceeding the threshold, such as triggering an early warning, recording but not alarming, or suggesting shutdown, used to define the action to be performed after the threshold is determined.

[0025] For example, the preset rule base can be managed through a graphical configuration interface. Users can select device monitoring parameters through drop-down menus, form condition fields by combining logical operators such as AND and OR, configure the values ​​and switches in the parameter fields through forms or sliders, and select the response strategy of the action fields through radio buttons. The configured rules are stored in the database in a structured form for the rule parsing engine to call in real time.

[0026] During the operating condition matching process, the target device's own status data and external environment data are collected in real time as the current operating condition parameters. The current operating condition parameters are matched with the condition domains of each rule in the rule base, and the rule that perfectly matches the conditions is selected as the target rule.

[0027] Furthermore, to avoid frequent rule changes caused by data packet loss, this embodiment supports a fuzzy matching strategy for operating conditions: when short-term data packet loss occurs and complete operating condition parameters cannot be obtained, the "most recent valid operating condition" continuation strategy is adopted, and the target rule that was successfully matched at the previous moment is used until the operating condition data returns to normal and a new round of matching is completed, thereby ensuring the continuity and stability of threshold monitoring.

[0028] S102. Determine the target window data and target expansion method according to the target rules.

[0029] The target expansion method includes at least one of the percentile weighting algorithm, interface coefficient adaptive algorithm and / or incremental percentile update algorithm.

[0030] Based on the window length in the target rules, target window data is collected from the historical data of the target device.

[0031] Specifically, such as Figure 3 As shown, the independent window manager extracts a data sequence of the corresponding length from the historical normal data of the target device based on the window length in the target rule parameter field, and uses it as the target window data. It should be noted that each rule in the preset rule base is associated with and maintains an independent data window, and different rules do not interfere with each other even if they are for the same physical measurement point. This design can isolate the statistical benchmarks under different operating conditions, avoid statistical benchmark pollution caused by operating condition switching, and ensure the accuracy of threshold calculation under each operating condition.

[0032] Based on the function activation flag in the target rules, determine the corresponding target expansion method: In response to the percentile weighting enable flag, the percentile weighting algorithm is started. The percentile weighting algorithm is used as the target expansion method to reflect the higher weight of recent data and improve the sensitivity of the threshold to the performance degradation trend.

[0033] In response to the interface coefficient adaptive enable flag, the interface coefficient adaptive algorithm is started. The interface coefficient adaptive algorithm is used as the target extension method and is incorporated into the target extension method to dynamically adjust the threshold coefficient according to the data distribution pattern to adapt to the data distribution characteristics under different working conditions.

[0034] In response to the incremental percentile update enable flag, the incremental percentile update algorithm is started. The incremental percentile update algorithm is used as the target extension method to reduce computation and memory overhead and adapt to resource-constrained edge deployment scenarios.

[0035] The three extension methods mentioned above can be enabled independently or in combination. For example, the percentile weighting algorithm and the interface coefficient adaptive algorithm can be enabled at the same time to balance trend responsiveness and distribution adaptability.

[0036] S103. Based on the target window data, calculate the target health threshold range through target expansion.

[0037] This step uses percentile calculation as the unified core and the IQR (interquartile range) threshold model as the standard output interface. Depending on the different combinations of target expansion methods, it executes the corresponding calculation logic and finally outputs a dynamic target health threshold range.

[0038] To facilitate understanding, the basic definition of the standard IQR threshold model is first explained: For a data window W, the 25th percentile Q1 and the 75th percentile Q3 are calculated, the interquartile range IQR = Q3 - Q1, and the standard threshold interval is [Q1 - k × IQR, Q3 + k × IQR], where k is the baseline coefficient. All extension methods adjust the percentile calculation process or coefficient values ​​to ultimately map to this standard interface, ensuring the consistency and interpretability of the threshold output format.

[0039] The specific calculation process is explained below for different combinations of target expansion methods: In scenario one, when the target expansion method only includes percentile weighting algorithm, each data point in the target window is assigned an exponentially decaying weight according to the collection time from recent to distant, to obtain the weighted 25th percentile and 75th percentile.

[0040] The difference between the weighted 75th percentile and the 25th percentile is used as the target interquartile range.

[0041] The starting endpoint of the target health threshold interval is calculated based on the weighted 25th percentile, the target interquartile range, and the benchmark coefficient; the ending endpoint of the target health threshold interval is calculated based on the weighted 75th percentile, the target interquartile range, and the benchmark coefficient.

[0042] For example, in the scenario where the target expansion method only includes a percentile weighting algorithm, to reflect the more important engineering characteristics of recent data, each data point in the target window data is assigned an exponentially decaying weight according to the collection time, from most recent to oldest; the newer the data, the higher the weight, and the older the data, the lower the weight. Based on the weighted data, the weighted 25th percentile Q1w and the weighted 75th percentile Q3w are calculated, and the weighted interquartile range, i.e., the target interquartile range IQRw = Q3w - Q1w, is further calculated.

[0043] Subsequently, the weighted percentiles and weighted interquartile ranges are substituted into the following standard IQR threshold formula: Threshold interval start endpoint = Q1w - k × IQRw; The endpoint of the threshold interval = Q3w + k × IQRw.

[0044] Where k is the baseline coefficient configured in the target rule. The target health threshold range obtained in this way can more quickly track the recent performance change trend of the device and capture early degradation signals in a timely manner.

[0045] In scenario two, when the target expansion method includes percentile weighting algorithm and interface coefficient adaptive algorithm, the upper tail coefficient and lower tail coefficient are dynamically decoupled and generated according to the skewness and kurtosis of the target window data distribution.

[0046] The starting endpoint of the target health threshold interval is calculated based on the weighted 25th percentile, the upper tail coefficient, and the target interquartile range.

[0047] The endpoint of the target health threshold interval is calculated based on the weighted 75th percentile, the lower tail coefficient, and the target interquartile range.

[0048] For example, in a scenario where the target expansion method includes a percentile weighting algorithm and an interface coefficient adaptive algorithm, the weighted 25th percentile Q1w, the weighted 75th percentile Q3w, and the weighted interquartile range, i.e. the target interquartile range IQRw, are obtained first by following the calculation process of the percentile weighting algorithm described above.

[0049] Based on this, adaptive calculation of interface coefficients is performed: according to the distribution characteristics of the target window data, the skewness and kurtosis of the data are calculated, where skewness reflects the asymmetry of the data distribution, and kurtosis reflects the steepness of the data distribution; the formula for skewness is Skewness = E [(X-μ)³] / σ³, and the formula for kurtosis is Kurtosis = E [(X-μ)³]. 4 ] / σ 4 In the formula, μ is the mean of the data and σ is the standard deviation of the data.

[0050] Based on the calculation results of skewness and kurtosis, the upper tail coefficient k is dynamically decoupled and generated. u With the lower tail coefficient k l Achieve asymmetric adjustment within the threshold range: Regarding the effect of skewness: When the data is positively skewed (right-skewed), the mean is greater than the median, and the upper tail data is more dispersed. Therefore, an upper tail coefficient k is set. u Greater than the lower tail coefficient k l To reduce false alarms, the upper tail threshold is relaxed; when the data is negatively skewed (left-skewed), the mean is smaller than the median, and the lower tail data is more dispersed, therefore a lower tail coefficient k is set. l Greater than the upper tail coefficient k u Relax the lower tail threshold.

[0051] Regarding the impact of kurtosis: when the kurtosis is greater than 3, the data is more concentrated near the mean, so the upper and lower tail coefficients should be appropriately reduced to improve monitoring sensitivity; when the kurtosis is less than 3, the data distribution is more dispersed, so the upper and lower tail coefficients should be appropriately increased to avoid false alarms.

[0052] To prevent excessive adjustment of coefficients from causing threshold failure, upper and lower limits for coefficient adjustment are set. For example, the range of values ​​for the upper and lower tail coefficients is constrained to [0.5, 2.0] times the base k value, ensuring that the threshold range is always within a reasonable range for engineering.

[0053] Finally, the target health threshold range is calculated based on asymmetric coefficients: Threshold interval start endpoint = Q1w-k l ×IQRw; Threshold interval end point = Q3w+k u ×IQRw.

[0054] In this way, the threshold range can adapt to changes in the data distribution pattern, and maintain a reasonable early warning sensitivity under different operating conditions.

[0055] Scenario 3: When the target expansion method only includes the incremental percentile update algorithm, it is achieved through P 2 The algorithm updates the 25th and 75th percentiles in real time within the target window data.

[0056] The difference between the 75th percentile and the 25th percentile, updated in real time, is used as the target interquartile range.

[0057] The starting endpoint of the target health threshold interval is calculated based on the real-time updated 75th percentile, target interquartile range, and baseline coefficient.

[0058] The endpoint of the target health threshold interval is calculated based on the real-time updated 25th percentile, target interquartile range, and baseline coefficient.

[0059] Specifically, in resource-constrained edge deployment scenarios, to support long-term monitoring with ultra-long windows without consuming excessive memory and computing power, this scenario adopts an online quantile estimation algorithm (such as the P² algorithm) to achieve incremental percentile updates.

[0060] When there is no need to store the complete target window data, only a fixed number of markers (usually 5-9) are maintained in memory for real-time estimation of the 25th percentile Q1 and the 75th percentile Q3. When new normal data points arrive, the algorithm updates the position and value of the internal markers in O(1) time complexity and outputs the updated Q1 and Q3 estimates immediately, without needing to sort and store the full data.

[0061] Based on the real-time updated Q1 and Q3, the interquartile range (IQR) is calculated as IQR = Q3 - Q1, and then substituted into the standard IQR threshold formula to obtain the target health threshold range. Threshold interval start endpoint = Q1 - k × IQR; The endpoint of the threshold interval = Q3 + k × IQR.

[0062] This approach reduces computational complexity to constant levels and memory usage by more than 99%, enabling long-window adaptive thresholding to run stably on low-end embedded devices.

[0063] Furthermore, to ensure that the statistical benchmark is not contaminated by abnormal data, this embodiment includes a clean update control mechanism that forms a closed-loop coordination with the threshold calculation, specifically including: Three-level status decision-making: The system substitutes real-time collected device data into the current target health threshold range for judgment, outputting a three-level health status: "Normal," "Warning," and "Alarm." The system supports configuring multiple threshold coefficients under the same rule, such as a warning coefficient k_warning=2.0 and an alarm coefficient k_alarm=3.0. Different levels of threshold ranges are generated based on the same statistical benchmark (Q1, Q3, IQR), corresponding to different health states. Each coefficient can be set independently, but is uniformly defined by the rule parameter domain.

[0064] Clean Admission Update: Only when data is determined to be in a "normal" state is the data point allowed to enter the data window of the corresponding rule and participate in the next round of percentile and threshold updates; abnormal data determined to be warning or alarm is directly excluded and does not enter the statistics window.

[0065] Anti-contamination closed-loop effect: Clean window data can output accurate IQR thresholds. Accurate thresholds can precisely distinguish between normal and abnormal data, thereby ensuring that the data entering the window is always normal data. This forms a positive closed loop from clean window, accurate threshold, accurate identification to keeping the window clean, thus preventing abnormal data from contaminating the statistical benchmark.

[0066] In addition, this embodiment also includes cold start and window degradation strategies, covering the entire lifecycle management of windows: Cold start strategy: When a rule is activated for the first time and the corresponding data window is empty, a conservative threshold (such as a fixed threshold or an empirical value based on the device design limit) preset in the rule parameter field is used for monitoring; after the window has collected n normal data that have been initially screened and judged, it automatically switches to the dynamic IQR threshold; a smooth transition strategy of linear interpolation can be configured between the cold start threshold and the dynamic threshold to avoid false alarms caused by sudden changes in monitoring sensitivity.

[0067] Window degradation strategy: Supports window time-to-live (TTL) mechanism. If a data window associated with a rule does not receive any new normal data within the TTL period (such as long-term equipment shutdown), the system will automatically clear the data window and mark the rule status as "awaiting cold start". When the operating conditions rematch the rule, the cold start process will be re-executed to prevent outdated data from polluting the statistical benchmark and ensure the timeliness of the threshold.

[0068] In this embodiment, target rules, including the operating parameters required for dynamically calculating health thresholds, are determined based on the current operating parameters of the target device. After determining the target window data and target expansion method based on the target rules, the target health threshold range is calculated based on the target window data and the target expansion method. Thus, this application traces the corresponding specific rules through the determined target rules, providing strong interpretability. Furthermore, by determining the target window data and target expansion method through the target rules, automatic rule switching and parameter self-correction are achieved, improving adaptability to operating conditions and resistance to abnormal data contamination.

[0069] To more clearly illustrate the implementation process of this method, the following examples are provided using two typical industrial scenarios: Application Example 1: Vibration Monitoring and Early Warning of Wind Turbine Generator Main Bearing This scenario involves real-time health monitoring of the vibration velocity of the main shaft bearing of a 1.5MW wind turbine generator set. The vibration level is significantly affected by operating conditions such as wind speed and power.

[0070] First, configure three rules corresponding to different working conditions in the preset rule base, as shown in Table 2 below: Table 2

[0071] Standby operating condition rules: The condition field is "wind speed < 3m / s and speed < 10rpm", the parameter field is window length 500, base coefficient 3.0, weighting is not enabled, adaptive k is not enabled, incremental update is not enabled, and the action field is record without alarm; Normal operation rules: The condition field is "wind speed ≥ 3m / s and power > 100kW", the parameter field is window length 1000, base coefficient 2.0, weighted average enabled, adaptive k enabled, incremental update disabled, and the action field is three levels: normal / early warning / alarm. Severe weather rules: The condition field is "wind speed > 15m / s or temperature < -20℃", the parameter field is window length 2000, baseline coefficient 2.5, weighted average enabled, adaptive k disabled, incremental update disabled, and the action field is trigger warning only.

[0072] Assuming the current wind speed is 8 m / s and the power is 1200 kW, the system matches the normal operation rule as the target rule. The system loads the independent data window corresponding to this rule (containing 1000 historical normal vibration data), and determines the target expansion method as a percentile weighted algorithm and an interface coefficient adaptive algorithm.

[0073] The calculation process is as follows: Weighted percentile calculation: The data within the window are assigned exponential decay weights over time, and the weighted Q1w = 1.2 mm / s, weighted Q3w = 1.8 mm / s, and weighted IQRw = 0.6 mm / s are calculated. Adaptive coefficient calculation: With a window data skewness of 0.7 (right skew), the upper tail coefficient k is obtained by adjusting according to adaptive logic. u =2.2, lower tail coefficient k l =2.0; Threshold range calculation: lower threshold = 1.2 - 2.0 × 0.6 = 0.0 mm / s, upper threshold = 1.8 + 2.2 × 0.6 = 3.12 mm / s, that is, the current normal vibration range is [0.0, 3.12] mm / s.

[0074] If the real-time vibration value is 2.1 mm / s, it is determined to be in a normal state. This data point is admitted into the data window, while the oldest data point is removed to complete the window update, providing a benchmark for the threshold calculation at the next moment.

[0075] Application Example 2: Spindle Temperature Health Management in CNC Machining Centers This scenario utilizes a resource-constrained embedded edge controller to achieve long-term trend monitoring and overheat warning of the spindle temperature in a machining center.

[0076] Two rules are configured in the default rule base, as shown in Table 3 below: Table 3

[0077] Warm-up rules: The condition field is "Power-on < 30 minutes", the parameter field is baseline offset +5℃, window length 500, baseline coefficient 2.5, and incremental update enabled, and the action field is normal / warning; Heavy cutting rule: The condition field is "feed rate > set value 90%", the parameter field is reference offset + 8℃, window length 800, reference coefficient 1.8, incremental update is not enabled, and the action field is normal / warning / alarm.

[0078] When the device has been powered on for 20 minutes, the warm-up rule is matched as the target rule, and the target expansion method is determined to be an incremental percentile update algorithm. The system uses the P² algorithm, which only needs to maintain 5 marker points to estimate percentiles in real time, without needing to store all 500 data points.

[0079] The calculation process is as follows: Incremental percentile update: When new data arrives, the marker points are updated in O(1) complexity, and Q1=38.2℃ and Q3=41.5℃ are output in real time, and IQR=3.3℃ is calculated; Baseline threshold calculation: Upper limit of baseline threshold = 41.5 + 2.5 × 3.3 = 49.75℃; Offset compensation: The +5℃ baseline offset configured by the superimposed rule is used, and the final upper limit of the threshold is 49.75 + 5 = 54.75℃, that is, the upper limit of the temperature during the warm-up stage is 54.75℃.

[0080] This method has extremely low memory usage and can run stably on edge devices for a long time, enabling adaptive monitoring and overheat warning of spindle temperature.

[0081] Table 4 is a performance comparison report table provided by an embodiment of this application. This report aims to quantitatively illustrate the performance advantages of the "incremental percentile kernel (such as the P² algorithm)" in this invention compared to the traditional "full data sorting method" on edge devices. The test was based on the scenario of application embodiment two (CNC spindle temperature monitoring) and was conducted on a low-end embedded microcontroller.

[0082] Table 4

[0083] The incremental kernel (P² algorithm) used in this invention achieves orders-of-magnitude optimization in computational efficiency and memory usage while ensuring the accuracy of quantile estimation, making it possible to achieve long-term, adaptive health monitoring on low-end edge devices.

[0084] Therefore, in summary, the device health threshold management method provided in this application has at least the following technical effects: a. Extremely lightweight and highly interpretable computation: Introducing incremental quantile updates reduces computational complexity to O(1), making it more suitable for edge deployment. The decision-making process is transparent, making it easy for engineers to understand and trust.

[0085] b. Effective integration and automation of expert experience: Transforming unstructured knowledge into structured rules to achieve knowledge accumulation and automated application.

[0086] c. Reliable adaptive and anti-interference capabilities: The clean update mechanism ensures the purity of the baseline; the weighted IQR improves response agility; and the adaptive k-value enhances the robustness of operating conditions.

[0087] d. Highly adaptable to different scenarios: Monitoring strategies can be flexibly adjusted for different equipment and operating conditions.

[0088] In some embodiments, this application also provides a device health threshold management apparatus. This device health threshold management apparatus may include one or more functional modules for implementing a device health threshold management method according to the above method embodiments.

[0089] For example, Figure 4 This is a schematic diagram illustrating the composition of a device health threshold management apparatus provided in an embodiment of this application. Figure 4 As shown, the health threshold management device includes: an acquisition module 201, a processing module 202, and a management module 203.

[0090] The acquisition module 201 is used to determine the target rules based on the current operating parameters of the target device; the target rules include the operating parameters required for dynamically calculating the health threshold; Processing module 202 is used to determine the target window data and the target expansion method according to the target rules; the target expansion method includes at least one of percentile weighting algorithm, interface coefficient adaptive algorithm and / or incremental percentile update algorithm; The management module 203 is used to calculate the target health threshold range based on the target window data through a target expansion method.

[0091] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device includes: a processor 302, a communication interface 303, and a bus 304. Optionally, the electronic device may also include a memory 301.

[0092] Processor 302 may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 302 may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 302 may also be a combination of functions implementing computing capabilities, such as a combination including CPU0 and CPU1, a DSP, and a microprocessor.

[0093] The communication interface 303 includes a receiving unit and a transmitting unit, and is used to connect with other devices via a communication network. This communication network can be Ethernet, a wireless access network, a wireless local area network (WLAN), etc.

[0094] The memory 301 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0095] As one possible implementation, the memory 301 can exist independently of the processor 302. The memory 301 can be connected to the processor 302 via a bus 304 and is used to store instructions or program code. When the processor 302 calls the instructions or program code stored in 301, it can implement the device health threshold management method provided in this embodiment of the invention.

[0096] In another possible implementation, the memory 301 can also be integrated with the processor 302.

[0097] Bus 304 can be an extended industry standard architecture (EISA) bus, etc. Bus 304 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0098] Through the above description of the implementation methods, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the service calling device can be divided into different functional modules to complete all or part of the functions described above.

[0099] This application also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be executed by computer instructions instructing related hardware. The program can be stored in the aforementioned computer-readable storage medium, and when executed, it can include the processes of the above method embodiments. The computer-readable storage medium can be any of the foregoing embodiments or memory. The aforementioned computer-readable storage medium can also be an external storage device of the aforementioned service invocation device, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the aforementioned service invocation device. Further, the aforementioned computer-readable storage medium can include both internal storage units of the aforementioned service invocation device and external storage devices. The aforementioned computer-readable storage medium is used to store the aforementioned computer program and other programs and data required by the aforementioned service invocation device. The aforementioned computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0100] This application also provides a computer program product, which includes a computer program that, when run on a computer, causes the computer to execute the device health threshold management method provided in the above embodiments.

[0101] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for managing device health thresholds, characterized in that, include: The target rules are determined based on the current operating parameters of the target equipment; the target rules include the operating parameters required for dynamically calculating the health threshold. The target window data and target expansion method are determined according to the target rules; the target expansion method includes at least one of percentile weighting algorithm, interface coefficient adaptive algorithm and / or incremental percentile update algorithm. Based on the target window data, the target health threshold range is calculated using the target expansion method.

2. The device health threshold management method according to claim 1, characterized in that, The step of determining the target rule based on the current operating parameters of the target equipment includes: Match the corresponding operating conditions in the preset rule base according to the current operating condition parameters, and obtain the associated target rules according to the corresponding operating conditions; The preset rule base stores multiple rules for different devices under different operating conditions. The items in the rules include window length, baseline coefficient, percentile weighting enable flag, interface coefficient adaptive enable flag, incremental percentile update enable flag, and action strategy. The window length is used to indicate the range of historical data collected from the target device. The baseline coefficient is the IQR coefficient. The percentile weighting enable flag is used to indicate whether the percentile weighting algorithm is started. The interface coefficient adaptive enable flag is used to indicate whether the interface coefficient adaptive algorithm is started. The incremental percentile update enable flag is used to indicate whether the incremental percentile update algorithm is started.

3. The device health threshold management method according to claim 2, characterized in that, The step of determining the target window data and target expansion method according to the target rules includes: Based on the window length in the target rule, the target window data is collected from the historical data of the target device. In response to the percentile weighting enable flag indicating that the percentile weighting algorithm is initiated, the percentile weighting algorithm is used as the target expansion method; and / or In response to the interface coefficient adaptation enable flag indicating that the interface coefficient adaptation algorithm is started, the interface coefficient adaptation algorithm is used as the target extension method; and / or In response to the incremental percentile update enable flag indication, the incremental percentile update algorithm is started, and the incremental percentile update algorithm is used as the target expansion method.

4. The device health threshold management method according to claim 2, characterized in that, The step of calculating the target health threshold range based on the target window data and through the target expansion method includes: When the target expansion method only includes the percentile weighting algorithm, each data in the target window data is assigned an exponentially decaying weight according to the collection time from recent to distant, to obtain the weighted 25th percentile and 75th percentile. The difference between the weighted 75th percentile and the 25th percentile is taken as the target interquartile range. The starting endpoint of the target health threshold interval is calculated based on the weighted 25th percentile, the target interquartile range, and the benchmark coefficient; the ending endpoint of the target health threshold interval is calculated based on the weighted 75th percentile, the target interquartile range, and the benchmark coefficient.

5. The device health threshold management method according to claim 4, characterized in that, Also includes: When the target expansion method includes percentile weighting algorithm and interface coefficient adaptive algorithm, the upper tail coefficient and lower tail coefficient are dynamically decoupled and generated according to the skewness and kurtosis of the target window data distribution. The starting endpoint of the target health threshold interval is calculated based on the weighted 25th percentile, the upper tail coefficient, and the target interquartile range. The endpoint of the target health threshold interval is calculated based on the weighted 75th percentile, the lower tail coefficient, and the target interquartile range.

6. The device health threshold management method according to claim 4, characterized in that, It also includes, when the target expansion method only includes the incremental percentile update algorithm, using P 2 The algorithm updates the 25th percentile and 75th percentile in real time within the target window data; The difference between the 75th percentile and the 25th percentile, updated in real time, is used as the target interquartile range. The starting endpoint of the target health threshold interval is calculated based on the real-time updated 75th percentile, the target interquartile range, and the benchmark coefficient. The endpoint of the target health threshold interval is calculated based on the real-time updated 25th percentile, the target interquartile range, and the benchmark coefficient.

7. A device for managing equipment health thresholds, characterized in that, include: The acquisition module is used to determine the target rule based on the current operating parameters of the target device; the target rule includes the operating parameters required for dynamically calculating the health threshold; The processing module is used to determine the target window data and the target expansion method according to the target rules; the target expansion method includes at least one of percentile weighting algorithm, interface coefficient adaptive algorithm and / or incremental percentile update algorithm; The management module is used to calculate the target health threshold range based on the target window data and through the target expansion method.

8. An electronic device, characterized in that, It includes a processor and a memory, the processor being coupled to the memory; the memory is used to store computer instructions, which are loaded and executed by the processor to enable the computer device to perform the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes computer-executable instructions that, when executed on a computer, cause the computer to perform the method according to any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when run on an electronic device, causes the electronic device to perform the method as described in any one of claims 1 to 6.

Images