Data Acquisition Methods and Predictive Maintenance System for Edge Gateways

Abstract

This invention relates to the field of edge gateway health management and predictive maintenance technology, specifically to a data acquisition method and predictive maintenance system for edge gateways. The method includes acquiring multi-dimensional status data of the edge gateway, including operating condition and environmental data, as well as degradation data of key components; calculating the overall degradation rate of capacitors and determining capacitor and flash memory health; calculating the basic health level; determining the environmental penalty factor; determining the overall gateway health score through a preset multiplicative penalty model; estimating the instantaneous degradation rate through a preset state-space model to determine the remaining service life; and calculating the corrected data acquisition frequency through load adaptive closed-loop control logic and outputting adaptive control commands. This invention transforms traditional passive response into proactive predictive maintenance, effectively avoiding production line downtime and data interruption caused by unexpected failures.

Description

Technical Field

[0001] This invention relates to the field of edge gateway health management and predictive maintenance technology, specifically to an edge gateway data acquisition method and predictive maintenance system. Background Technology

[0002] As the core device for industrial data acquisition, the operational stability of the edge gateway is crucial to ensuring the continuity of industrial processes. Edge gateways are usually deployed in harsh working conditions such as high temperature, vibration, and electrical stress fluctuations, and face the risk of chronic degradation of key components such as capacitors and flash memory, as well as acute shocks from the external environment.

[0003] Existing maintenance strategies are mostly reactive or rely on single-dimensional status monitoring. These methods struggle to integrate multi-source heterogeneous data, cannot accurately assess the overall health status of gateways under multiple stress couplings, and cannot effectively predict their remaining lifespan. Therefore, when gateways experience unexpected failures, it can easily lead to data interruptions and production line downtime, posing serious security risks and economic losses. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a data acquisition method and a predictive maintenance system for edge gateways. Specifically, the technical solution of this invention is as follows:

[0005] Data acquisition methods for edge gateways include:

[0006] S1 collects multi-dimensional status data from the edge gateway, including operating conditions and environment data, as well as degradation data of key components;

[0007] S2, based on operating condition and environmental data and a preset physical failure model, calculates the overall degradation rate of the capacitor and determines the capacitor health; and based on the degradation data of key components and a preset data-driven model, determines the flash memory health.

[0008] S3 combines capacitor health and flash memory health, and calculates the basic health based on preset component importance weights; and determines the environmental penalty factor based on operating condition and environmental data and a pre-trained anomaly detection model.

[0009] S4 integrates the basic health score and environmental penalty factor through a preset multiplicative penalty model to determine the gateway's overall health score; and based on the time series of the gateway's overall health score, estimates the instantaneous degradation rate through a preset state-space model to determine the remaining service life.

[0010] S5, based on real-time onboard temperature, preset safe temperature threshold, preset load-temperature control coefficient, preset minimum performance factor, and original data acquisition frequency from operating condition and environmental data, calculates and corrects the data acquisition frequency through load adaptive closed-loop control logic and outputs adaptive control commands.

[0011] Preferably, the operating conditions and environmental data include: CPU utilization; network throughput; onboard temperature; ambient humidity; triaxial vibration acceleration; and power input voltage fluctuation.

[0012] Preferably, the degradation data of key components includes: the cumulative number of erase / write cycles of NAND Flash; the number of ECC error corrections; the number of bad blocks; and the output voltage ripple of the power module.

[0013] Preferably, the overall degradation rate of the capacitor is calculated using a physical failure model, including:

[0014] The thermal aging rate of the base is calculated based on the Arrhenius model and the plate temperature.

[0015] The overall degradation rate of the capacitor is calculated by combining the basic thermal aging rate, CPU utilization, and output voltage ripple, and by introducing preset load and voltage stress sensitivity coefficients.

[0016] Preferably, the environmental penalty factor is determined, including:

[0017] Using triaxial vibration acceleration, ambient humidity, and power input voltage fluctuation as inputs to the anomaly detection model, the real-time vibration anomaly score, real-time humidity anomaly score, and real-time power anomaly score are calculated respectively.

[0018] By combining the real-time vibration anomaly score, the real-time humidity anomaly score, the real-time power supply anomaly score, and the preset environmental pressure penalty weight, the environmental penalty factor is determined through minimum amplitude processing.

[0019] Preferably, determining the remaining useful life includes:

[0020] Using the time series of the gateway's overall health score as observations, the instantaneous degradation rate is estimated by applying a state-space model with a particle filter or Kalman filter framework.

[0021] The remaining lifespan is calculated by linear extrapolation based on the instantaneous degradation rate, the current gateway's overall health score, and the preset failure threshold.

[0022] Preferred options also include:

[0023] In response to a gateway’s overall health score being less than a preset health threshold or its remaining lifespan being less than a preset time threshold, a predictive maintenance alert is triggered, and a maintenance work order is automatically generated.

[0024] The predictive maintenance system for edge gateways includes:

[0025] The data acquisition module is used to collect multi-dimensional status data of the edge gateway, including operating condition and environmental data, as well as degradation data of key components.

[0026] The predictive maintenance module includes:

[0027] The health assessment unit is used to calculate the overall degradation rate of capacitors and determine the health of capacitors based on operating condition and environmental data and a preset physical failure model; and to determine the health of flash memory based on degradation data of key components and a preset data-driven model.

[0028] The fusion evaluation unit combines capacitor health and flash memory health, calculates basic health based on preset component importance weights, and determines environmental penalty factors based on operating condition and environmental data and a pre-trained anomaly detection model.

[0029] The integrated processing unit is used to combine the basic health score and environmental penalty factors through a preset multiplicative penalty model to determine the overall health score of the gateway.

[0030] The lifetime prediction unit is used to estimate the instantaneous degradation rate and determine the remaining lifetime based on the time series of the gateway's comprehensive health score and a preset state-space model.

[0031] The closed-loop control unit is used to calculate and correct the data acquisition frequency based on the real-time onboard temperature, preset safe temperature threshold, preset load-temperature control coefficient, preset minimum performance factor and original data acquisition frequency in the operating condition and environmental data, and output adaptive control commands through load adaptive closed-loop control logic.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. This invention innovatively combines physical failure models and data-driven models to quantify the degradation state of capacitors and flash memory respectively. Through two-layer fusion processing, a comprehensive health status that takes into account both chronic aging and acute impact is obtained, resulting in a more comprehensive and accurate assessment.

[0034] 2. This invention utilizes an anomaly detection model to quantify acute shocks from external environments such as vibration, humidity, and voltage fluctuations, forming an environmental penalty factor. This factor is then integrated with the basic health status through a multiplicative penalty model, accurately reflecting the true state of the gateway under harsh operating conditions.

[0035] 3. This invention uses a state-space model to estimate the instantaneous degradation rate, rather than directly extrapolating the health score. This method can extract a more realistic degradation trend from noisy observations, making the remaining useful life prediction results smoother, more robust and more sensitive.

[0036] 4. This invention automatically triggers predictive maintenance alarms and work orders by setting dual thresholds for health status and remaining lifespan, transforming traditional passive response into proactive predictive maintenance, effectively avoiding production line downtime and data interruption caused by unexpected failures. Attached Figure Description

[0037] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0038] Figure 1 This is a flowchart of the method of the present invention;

[0039] Figure 2 This is a structural diagram of the system of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0041] Example 1:

[0042] Please see Figure 1 Data acquisition methods for edge gateways include:

[0043] S1 collects multi-dimensional status data from the edge gateway, including operating conditions and environment data, as well as degradation data of key components;

[0044] S2, based on operating condition and environmental data and a preset physical failure model, calculates the overall degradation rate of the capacitor and determines the capacitor health; and based on the degradation data of key components and a preset data-driven model, determines the flash memory health.

[0045] S3 combines capacitor health and flash memory health, and calculates the basic health based on preset component importance weights; and determines the environmental penalty factor based on operating condition and environmental data and a pre-trained anomaly detection model.

[0046] S4 integrates the basic health score and environmental penalty factor through a preset multiplicative penalty model to determine the gateway's overall health score; and based on the time series of the gateway's overall health score, estimates the instantaneous degradation rate through a preset state-space model to determine the remaining service life.

[0047] S5, based on real-time onboard temperature, preset safe temperature threshold, preset load-temperature control coefficient, preset minimum performance factor, and original data acquisition frequency from operating condition and environmental data, calculates and corrects the data acquisition frequency through load adaptive closed-loop control logic and outputs adaptive control commands.

[0048] This invention provides a data acquisition method for an edge gateway. This method is the core of the technical solution of this invention and aims to achieve accurate health assessment, remaining lifetime prediction and adaptive operation control of the edge gateway through multi-dimensional data fusion, hybrid driving modeling and closed-loop control.

[0049] In one specific embodiment, the method includes:

[0050] Collect multidimensional status data from the edge gateway; the purpose of this step is to provide comprehensive, multidimensional, real-time data input for subsequent health assessment and degradation modeling.

[0051] Multidimensional state data refers to a data set that, under the specific technical environment of this invention, is divided into two key subsets: operating condition and environmental data and key component degradation data;

[0052] Operating condition and environmental data are used to characterize the real-time operating load of the gateway and the external environmental conditions it is in; in this embodiment, it may specifically include, but is not limited to, data collected by integrated sensors, such as: CPU utilization, network throughput, onboard temperature, ambient humidity, triaxial vibration acceleration, and power input voltage fluctuation.

[0053] Key component degradation data is used to characterize the physical wear and deterioration of core components inside the gateway, such as flash memory and power modules, and is a direct indicator reflecting their physical degradation. In this embodiment, it may specifically include, but is not limited to: the cumulative number of erase and write cycles of NAND Flash, the number of ECC error correction cycles, the number of bad blocks, and the output voltage ripple of the power module.

[0054] Based on operating condition and environmental data and a preset physical failure model, the overall degradation rate of the capacitor is calculated and the capacitor health is determined; and based on the degradation data of key components and a preset data-driven model, the flash memory health is determined. The purpose of this step is to quantify the health status of the two most critical and vulnerable components, namely the capacitor and the flash memory, through a hybrid-driven modeling method.

[0055] For capacitors, degradation is affected by a combination of temperature, load, and electrical stress. This embodiment uses a pre-defined Physical Failure (PoF) model for calculation, which includes two levels:

[0056] To quantify the fundamental accelerating effect of temperature on capacitor aging, the fundamental thermal aging rate was calculated based on the Arrhenius model. :

[0057]

[0058] in: Based on the thermal aging rate, in units of It is calculated using this formula; Real-time onboard temperature, unit: It is provided by the working condition and environmental data collected in the previous steps; The preset material and geometric scale factor, in units of ; The material constants related to the activation energy, in units of . ; and This calibration is based on the capacitor model's datasheet or obtained through accelerated life testing (ALT) data calibration. To illustrate the calibration process, those skilled in the art can obtain a set of calibration data points, each containing a specific constant ambient temperature. and the corresponding mean time to failure. Through the By performing methods such as least squares regression analysis, the parameters can be fitted and determined. and ;

[0059] To simulate the nonlinear accelerated aging effect caused by the coupling of high temperature, high load, and high voltage ripple, the operating load and electrical stress are introduced as acceleration factors to calculate the overall degradation rate of the capacitor. :

[0060]

[0061] in: The overall degradation rate of the capacitor, in units of It is calculated using this formula; The base thermal aging rate is calculated in the previous step; This represents the CPU utilization rate, ranging from 0 to 1, and is provided by the operating conditions and environmental data collected in the preceding steps. Real-time voltage ripple is provided by degradation data of key components acquired in previous steps; The preset rated reference ripple; and The preset load and voltage stress sensitivity coefficients, , and These are preset parameters based on hardware design, such as heat dissipation capacity, power margin, and experimental data calibration, used to adjust the weight of the impact of different operating conditions on aging.

[0062] To illustrate the calibration process, those skilled in the art can obtain a set of calibration data through controlled variable experiments, for example, at a constant temperature. and constant ripple Below, measure different CPU utilization rates. corresponding Change, and then constancy and constant Below, measure different corresponding The changes were finally determined by fitting the data using multiple regression analysis. and ;

[0063] Through the The total damage is obtained by integrating over time. And finally determine the health of the capacitor. :

[0064]

[0065] in: It is a preset capacitor failure damage threshold, which can be determined based on historical failure data statistics or accelerated aging experiments.

[0066] For flash memory, its health is determined by both inherent wear and real-time error correction status; this embodiment uses a preset data-driven model to determine the flash memory health. The technical motivation behind this model is to transform multiple discrete SMART attributes, i.e., key component degradation data, into a single, quantified flash health score.

[0067]

[0068] in: for The flash memory health status at any given time, with a value ranging from 0 to 1, is calculated using this formula; The cumulative number of erase / write cycles is provided by degradation data of key components collected in previous steps; The preset maximum number of erase / write cycles threshold; The number of real-time ECC error corrections is provided by the degradation data of key components collected in the preceding steps; The preset failure threshold for the number of ECC cycles; The real-time bad block growth number is provided by the degradation data of key components collected in the preceding steps; The preset failure threshold for the number of bad blocks; and These are the weighting coefficients; , and All thresholds are preset according to the NAND Flash chip specifications, such as TLC / QLC type; weights and satisfy Based on historical failure data analysis, FMEA calibration is used to balance the different impacts of correctable errors and permanent bad blocks on health.

[0069] Combining capacitor health and flash memory health, and based on preset component importance weights, a base health score is calculated. An environmental penalty factor is determined based on operating condition and environmental data and a pre-trained anomaly detection model. The purpose of this step is to integrate the component-level health scores obtained in previous steps with the risk of instantaneous environmental impacts to assess the overall state of the gateway. To incorporate the chronic aging of internal components, a base health score is calculated. :

[0070]

[0071] in: The basic health score, ranging from 0 to 1, is calculated using this formula. and These are the capacitor health and flash memory health calculated in the previous steps, respectively; and The component importance weights are preset to satisfy... It is calibrated based on the degree to which component failure affects the fundamental value of the system, such as production continuity;

[0072] To quantify the acute impact of the external environment, based on the operating conditions and environmental data collected in previous steps, specifically... , , The real-time vibration anomaly score is calculated using a pre-trained anomaly detection model, such as an LSTM model. Real-time humidity anomaly score and real-time power supply anomaly score The pre-trained anomaly detection model is trained using data from the gateway under normal service conditions as a baseline training set, enabling it to identify abnormal patterns that deviate from the norm. These scores range from 0 (normal) to 1 (severe anomaly). An environmental penalty factor is determined. :

[0073]

[0074] in: The environmental penalty factor, with a value range of 0-1, is calculated using this formula. , , The above are the real-time anomaly scores; , , The environmental stress penalty weight is determined based on the sensitivity of the gateway deployment scenario to specific environmental stresses; The process ensured that the penalty factor remained in the 0-1 range.

[0075] The basic health score and environmental penalty factor are fused together using a preset multiplicative penalty model to determine the gateway's overall health score. Based on the time series of the gateway's overall health score, the instantaneous degradation rate is estimated using a preset state-space model to determine the remaining service life. This step is the core output of this embodiment, providing the final health status assessment and future service life prediction.

[0076] To obtain a final score that reflects both chronic aging and acute shocks, a pre-defined multiplicative penalty model is used to calculate the gateway's overall health score. The technical motivation for adopting the multiplication model is to ensure... It always stays in the 0-1 range and fixes the defect that the subtraction model may result in negative values;

[0077]

[0078] in: The gateway's overall health score, ranging from 0 to 1, is calculated using this formula. The baseline health level calculated for the preceding steps; The environmental penalty factor calculated for the preceding steps; when the environment is harsh. Increase Will The punishment was reduced on the basis of this;

[0079] To predict future lifespan, this embodiment will Using historical time series data as observations, a pre-defined state-space model, such as a particle filter or Kalman filter framework, is applied to estimate a key hidden state in real time, namely the instantaneous degradation rate. ;

[0080] The instantaneous degradation rate is expressed in units of 1000 m / s. It is estimated by the state-space model and reflects The current rate of deterioration; the estimation process of this state-space model relies on pre-defined process noise and observation noise covariance matrices, whose parameters are based on historical data. The dataset was optimized; based on this rate, the remaining lifetime was calculated via linear extrapolation. :

[0081]

[0082] in: for The remaining useful life predicted at any time, in units of It is calculated using this formula; This represents the current overall health score of the gateway. The estimated instantaneous degradation rate; To ensure robustness of the calculation, a preset failure threshold is used as the denominator. Form, in which It is a preset, extremely small positive number, for example This measure can effectively prevent issues caused by unstable equipment status. It avoids division-by-zero errors caused by data approaching zero and can prevent errors caused by data noise. When a negative value occurs, a negative lifetime with no physical meaning is calculated, such as 0.15. This threshold is a failure judgment standard preset based on historical data and operational requirements. This formula maintains consistency in dimensions; the difference in dimensionless health scores is divided by... The rate, to obtain The remaining lifetime of a dimension;

[0083] Based on the real-time onboard temperature, preset safe temperature threshold, preset load-temperature control coefficient, preset minimum performance factor, and original data acquisition frequency in the operating condition and environmental data, the corrected data acquisition frequency is calculated through load adaptive closed-loop control logic, and adaptive control commands are output. The purpose of this step is to build a closed-loop control system that actively adjusts the gateway operating conditions to ensure its long-term survival and the fundamental value of the system when harsh environments such as high temperatures are detected.

[0084] In this embodiment, the load adaptive closed-loop control logic achieves graceful performance degradation through the following formula:

[0085]

[0086] in: The corrected data acquisition frequency, calculated using this formula, is used as the output of the adaptive control command. This is the original data acquisition frequency, which is a preset value; Real-time onboard temperature is provided by operating condition and environmental data collected in previous steps; For example, a preset safe temperature threshold. The maximum rated operating temperature of the gateway hardware, especially the capacitors, is preset. This is the preset load-temperature control coefficient, in units of... or ,make sure It is a dimensionless value; This is the preset minimum performance factor, with a value ranging from 0 to 1, for example, 0.3; and The calibration is based on the tolerance of the industrial scenario to the real-time performance of the data, ensuring that even after the frequency is reduced, the collected data can still meet the basic needs of the upper-level system.

[0087] when hour, Will be less than 1, It will be reduced but will not be lower than This reduces CPU load. This reduces Slow down the rate of capacitor degradation ;

[0088] The method described in this embodiment collects comprehensive multidimensional state data and innovatively combines the Physical Failure Model (PoF) and a data-driven model to accurately model the degradation of capacitors and flash memory, respectively. This method employs a two-layer fusion approach—inter-component fusion and environmental penalty fusion—and utilizes a multiplicative penalty model to obtain a comprehensive gateway health score that simultaneously reflects both chronic aging and acute impact. To further improve prediction performance, this scheme utilizes a state-space model from... Extracting instantaneous degradation rate from sequence This results in a more robust remaining service life. Prediction; This solution also constructs a load-adaptive closed-loop control logic, enabling the gateway to proactively cope with harsh environments and achieve graceful performance degradation; In summary, this invention provides a complete technical solution from data acquisition, hybrid modeling, fusion evaluation, lifetime prediction to adaptive closed-loop control, significantly improving the accuracy, robustness, and intelligence level of edge gateway health management, transforming traditional passive maintenance into proactive predictive maintenance, and maximizing the continuity of industrial processes and asset security.

[0089] Example 2:

[0090] Operating and environmental data include: CPU utilization; network throughput; onboard temperature; ambient humidity; triaxial vibration acceleration; and power input voltage fluctuation.

[0091] This embodiment, based on the method described in Embodiment 1, specifically defines the working conditions and environmental data in the preceding steps;

[0092] Operating and environmental data include: CPU utilization; network throughput; onboard temperature; ambient humidity; triaxial vibration acceleration; and power input voltage fluctuation.

[0093] CPU utilization and onboard temperature It is to calculate the overall degradation rate of the capacitor. The core inputs are the thermal stress and load stress in the physical failure model; triaxial vibration acceleration. Ambient humidity and power input voltage fluctuation It is to determine the environmental penalty factor The core inputs are the network throughput, as they are the direct basis for pre-trained anomaly detection models to assess acute shocks from the external environment; It is another important indicator reflecting the gateway's operating status and can be used for more refined load modeling or anomaly detection;

[0094] This embodiment ensures the completeness of key information required for subsequent hybrid-driven modeling and fusion evaluation by clearly defining the specific composition of the operating condition and environmental data; this specific data combination has a clear technical orientation: and It supports the physical failure model, and , and This invention supports an anomaly detection model, enabling it to simultaneously detect chronic degradation caused by internal operating conditions and acute shocks caused by the external environment, thereby significantly improving the overall health score of the gateway. The accuracy and comprehensiveness of the information.

[0095] Example 3:

[0096] Key component degradation data include: cumulative write / erase cycles of NAND Flash; number of ECC error corrections; number of bad blocks; and output voltage ripple of the power module.

[0097] This embodiment, based on the method described in Embodiment 1, specifically defines the degradation data of key components in the preceding steps;

[0098] Key component degradation data includes: cumulative write / erase cycles of NAND Flash; number of ECC error corrections; number of bad blocks; and output voltage ripple of the power module.

[0099] Total number of erase / write cycles for NAND Flash It is a flash data-driven model Key parameters used to measure long-term wear and tear, i.e., the first failure mode; ECC error correction count. and the number of bad blocks yes Key parameters used to measure transient state degradation, i.e., the second failure mode; output voltage ripple of the power module. It is a capacitor physical failure model Key parameters used to quantify the acceleration effect of electrical stress;

[0100] This embodiment ensures that the two core branches of the hybrid driving model, namely the capacitor PoF model and the flash data driving model, obtain the most direct and critical physical state observations by clearly defining the specific composition of the degradation data of key components. , , The combination enables flash memory health assessment It can balance long-term wear and tear with instantaneous conditions, and The introduction of this reduces the rate of capacitor degradation. It can quantify the impact of electrical stress; therefore, this embodiment greatly enhances the physical realism and accuracy of component-level health calculation.

[0101] Example 4:

[0102] The overall degradation rate of the capacitor is calculated using a physical failure model, including:

[0103] The thermal aging rate of the base is calculated based on the Arrhenius model and the plate temperature.

[0104] The overall degradation rate of the capacitor is calculated by combining the basic thermal aging rate, CPU utilization, and output voltage ripple, and by introducing preset load and voltage stress sensitivity coefficients.

[0105] This embodiment, based on the method described in Embodiment 1, specifically defines the process of calculating the overall degradation rate of the capacitor using a physical failure model in the preceding steps;

[0106] The process specifically includes:

[0107] The thermal aging rate of the base is calculated based on the Arrhenius model and the plate temperature.

[0108] This embodiment uses the Arrhenius equation to quantify the fundamental effect of temperature on aging: in This is the onboard temperature; As a baseline rate, it reflects the aging rate under thermal stress only;

[0109] The overall degradation rate of the capacitor is calculated by combining the basic thermal aging rate, CPU utilization and output voltage ripple, and introducing preset load and voltage stress sensitivity coefficients.

[0110] As described in the implementation of Example 1, this example is in Based on this, the operating load was further coupled. and electrical stress The influence of load and voltage stress sensitivity coefficients is mitigated by introducing preset load and voltage stress sensitivity coefficients. , The overall degradation rate of the capacitor was calculated. :

[0111]

[0112] This design simulates the nonlinear accelerated aging effect on capacitor lifespan caused by the coupling of high temperature, high CPU load, and high voltage ripple under real-world operating conditions; those skilled in the art will understand that this design... The linear relationship is an effective approximation and preferred simplified implementation of the acceleration effect. In scenarios requiring higher accuracy, this linear term can be replaced with other nonlinear functions, such as power-law functions. or exponential function This allows for a more precise characterization of the dose-effect relationship between stress and aging rate.

[0113] This embodiment provides a more accurate method for modeling capacitance degradation; compared to the traditional Arrhenius model that relies solely on temperature, this scheme has a significant gain effect: it achieves this through... A physically reasonable degradation baseline was established, and innovatively, a multiplication factor was used. and Two key non-thermal stress sources, CPU utilization and output voltage ripple, were introduced; this coupled model calculates... It more accurately reflects the actual aging rate of the gateway under real, harsh, and multi-stress coupling conditions, thereby greatly improving capacitor health. The accuracy of the assessment.

[0114] Example 5:

[0115] Identify environmental penalty factors, including:

[0116] Using triaxial vibration acceleration, ambient humidity, and power input voltage fluctuation as inputs to the anomaly detection model, the real-time vibration anomaly score, real-time humidity anomaly score, and real-time power anomaly score are calculated respectively.

[0117] By combining the real-time vibration anomaly score, the real-time humidity anomaly score, the real-time power supply anomaly score, and the preset environmental pressure penalty weight, the environmental penalty factor is determined through minimum amplitude processing.

[0118] This embodiment, based on the method described in Embodiment 1, specifically defines the process of determining the environmental penalty factor in the preceding steps;

[0119] The process specifically includes:

[0120] Using triaxial vibration acceleration, ambient humidity, and power input voltage fluctuation as inputs to the anomaly detection model, the real-time vibration anomaly score, real-time humidity anomaly score, and real-time power anomaly score are calculated respectively.

[0121] This embodiment will collect data from the preceding steps. That is, triaxial vibration acceleration, That is, ambient humidity, That is, the power supply input voltage fluctuation is used as the input to a pre-trained anomaly detection model such as LSTM. This model is trained with normal operating condition data and can output a normalized real-time vibration anomaly score. Real-time humidity anomaly score and real-time power supply anomaly score ;

[0122] By combining the real-time vibration anomaly score, the real-time humidity anomaly score, the real-time power supply anomaly score, and the preset environmental pressure penalty weight, the environmental penalty factor is determined through minimum amplitude processing.

[0123] This embodiment calculates the result by weighted fusion of the three outlier scores mentioned above. And it specifically adopts minimum amplitude limiting processing, that is :

[0124]

[0125] in , , It is a preset environmental pressure penalty weight;

[0126] This embodiment provides a robust and interpretable method for quantifying environmental risks; its advantage lies in that it uses an anomaly detection model to convert raw sensor data with varying physical dimensions into quantifiable data. , , It was converted into a unified real-time anomaly score in the 0-1 range. , , This solves the problem of multi-source heterogeneous data fusion; furthermore, it determines the environmental penalty factor through minimum amplitude limiting processing. This ensures that even when multiple environmental factors simultaneously exhibit extreme anomalies, causing the weighted sum to exceed 1, It will also be limited to 1; this avoids calculations in subsequent steps. The occurrence of negative values ​​ensures the overall health score of the gateway. The mathematical stability and physical meaning of the value are always in the 0-1 range, making the penalty model more robust and reliable.

[0127] Example 6:

[0128] Determine the remaining useful life, including:

[0129] Using the time series of the gateway's overall health score as observations, the instantaneous degradation rate is estimated by applying a state-space model with a particle filter or Kalman filter framework.

[0130] The remaining lifespan is calculated by linear extrapolation based on the instantaneous degradation rate, the current gateway's overall health score, and the preset failure threshold.

[0131] This embodiment, based on the method described in Embodiment 1, specifically defines the process of determining the remaining service life in the preceding steps;

[0132] The process specifically includes:

[0133] Using the time series of the gateway's overall health score as observations, the instantaneous degradation rate is estimated by applying a state-space model with a particle filter or Kalman filter framework.

[0134] This embodiment does not directly use Instead of extrapolating the time series data, this model uses it as observations for a state-space model; particle filtering or Kalman filtering frameworks familiar to those skilled in the art are used for this purpose; the technical motivation of this model is... This is a noisy observation, while the system's true instantaneous degradation rate is... It is a hidden state; through filtering algorithms, it can be extracted from noisy states. More accurate estimates ;

[0135] The remaining service life is calculated by linear extrapolation based on the instantaneous degradation rate, the current gateway's overall health score, and the preset failure threshold.

[0136] To obtain more reliable After estimating the value, this embodiment uses a linear extrapolation method based on the current gateway's overall health score. Instantaneous degradation rate and preset failure threshold To calculate the remaining useful life :

[0137]

[0138] This embodiment provides a more robust and sensitive RUL prediction method; compared to directly... The sequence is fitted and extrapolated. The gain effect of this scheme lies in the two-step estimation-extrapolation method: it uses state-space models such as particle filtering or Kalman filtering to extrapolate from noisy observations. Extracting a smoother instantaneous degradation rate that better reflects the true degradation trend. ; then use this more reliable one Linear extrapolation is performed; this decoupling process, namely the decoupling of state estimation and extrapolation calculation, enables RUL prediction. right The changes, namely the changes in the rate of deterioration, are more sensitive, while avoiding [the following]. This eliminates transient noise interference, resulting in more accurate and stable remaining service life predictions.

[0139] Example 7:

[0140] This method also includes:

[0141] In response to a gateway’s overall health score being less than a preset health threshold or its remaining lifespan being less than a preset time threshold, a predictive maintenance alert is triggered, and a maintenance work order is automatically generated.

[0142] This embodiment adds a subsequent decision-making and action step to the method described in Embodiment 1;

[0143] In response to the gateway's overall health score being less than a preset health threshold or its remaining service life being less than a preset time threshold, a predictive maintenance alarm is triggered, and a maintenance work order is automatically generated.

[0144] This embodiment sets up two parallel trigger conditions, i.e., OR logic:

[0145] Gateway overall health score The health level is calculated to be less than the preset health threshold by the preceding steps. For example, 0.3; this condition is used to detect situations where the current state is already at a dangerous level;

[0146] Remaining service life The time threshold calculated from the preceding steps is less than the preset time threshold. For example, 20 days; this condition is used to capture situations where the current state is acceptable, but the rate of degradation is too fast and the system is about to fail.

[0147] and The setting is based on the maintenance window, which ensures that the warning time must be greater than the average cycle required for spare parts procurement, logistics and engineer dispatch, so as to ensure that replacement is completed before unplanned downtime occurs;

[0148] Once any condition is met, the system will automatically trigger a predictive maintenance alert and automatically generate a maintenance work order, which can further specify the primary risk points, such as the power module health Hc being below the threshold.

[0149] This embodiment uses the abstract data calculated in the preceding steps. and This translates into specific, executable maintenance actions, forming a closed loop from prediction to maintenance. By employing dual-threshold triggering logic based on health status or RUL, this solution ensures the timeliness and completeness of early warnings, capturing both poor condition caused by chronic degradation and rapid speed caused by acute shocks. The function of automatically generating maintenance work orders automates the predictive maintenance process, ensuring that maintenance decisions can be executed before unplanned downtime occurs, significantly improving operational efficiency and system reliability.

[0150] Example 8:

[0151] Please see Figure 2 A predictive maintenance system for edge gateways, including:

[0152] The data acquisition module is used to collect multi-dimensional status data of the edge gateway, including operating condition and environmental data, as well as degradation data of key components.

[0153] The predictive maintenance module includes:

[0154] The health assessment unit is used to calculate the overall degradation rate of capacitors and determine the health of capacitors based on operating condition and environmental data and a preset physical failure model; and to determine the health of flash memory based on degradation data of key components and a preset data-driven model.

[0155] The fusion evaluation unit combines capacitor health and flash memory health, calculates basic health based on preset component importance weights, and determines environmental penalty factors based on operating condition and environmental data and a pre-trained anomaly detection model.

[0156] The integrated processing unit is used to combine the basic health score and environmental penalty factors through a preset multiplicative penalty model to determine the overall health score of the gateway.

[0157] The lifetime prediction unit is used to estimate the instantaneous degradation rate and determine the remaining lifetime based on the time series of the gateway's comprehensive health score and a preset state-space model.

[0158] The closed-loop control unit is used to calculate and correct the data acquisition frequency based on the real-time onboard temperature, preset safe temperature threshold, preset load-temperature control coefficient, preset minimum performance factor and original data acquisition frequency in the operating condition and environmental data, and output adaptive control commands through load adaptive closed-loop control logic.

[0159] This embodiment provides a predictive maintenance system for an edge gateway, configured to perform the method described in any one of embodiments 1-7 above; in a preferred embodiment, the system includes:

[0160] The data acquisition module, used to execute the preceding S1 step, is configured to collect multi-dimensional status data from the edge gateway; as described in the previous embodiment, the multi-dimensional status data includes: operating condition and environmental data, such as... , , , , and critical component degradation data such as , , , ;

[0161] The predictive maintenance module, the core processing unit of the system, is configured to execute steps S2 to S5, and can be further divided into multiple functional units:

[0162] Health assessment unit: Its purpose is to perform step S2; this unit is configured to perform the assessment based on operating condition and environmental data and a preset physical failure model, such as... Formulas are used to calculate the overall degradation rate of capacitors and determine their health. ; and based on the degradation data of key components and the preset data-driven model, such as Formula to determine flash memory health ;

[0163] Fusion Evaluation Unit: Its purpose is to perform step S3; this unit is configured to incorporate capacitor health. With flash memory health And based on the preset component importance weights , Calculate basic health status like Formula; and based on operating conditions and environmental data such as , , Using a pre-trained anomaly detection model, determine the environmental penalty factor. like formula;

[0164] The comprehensive processing unit is designed to execute the first part of S4; this unit is configured to process the basic health status. Environmental penalty factors Through a pre-defined multiplication penalty model, such as Formula fusion processing determines the gateway's overall health score. ;

[0165] Lifetime prediction unit: Its purpose is to perform the second part of S4; this unit is configured to perform based on the gateway's overall health score. The instantaneous degradation rate is estimated from the time series using a pre-defined state-space model such as particle filtering. Determine the remaining service life like formula;

[0166] Closed-loop control unit: Its purpose is to execute step S5; this unit is configured to perform the operation based on real-time onboard temperature from operating condition and environmental data. Preset safe temperature threshold Preset load-temperature control coefficient Preset minimum performance factor Compared with the frequency of raw data acquisition Through load adaptive closed-loop control logic, such as Formula for calculating the corrected data acquisition frequency It also outputs adaptive control commands;

[0167] This embodiment provides a complete and modular edge gateway predictive maintenance system architecture. The data acquisition module provides the data foundation. The predictive maintenance module, through its five internal logical units—health calculation, fusion assessment, comprehensive processing, lifetime prediction, and closed-loop control—achieves a complete functional mapping of the methods described in embodiments 1-7 at the system level. This clear modularization and unitization decouples the complex multidimensional data analysis, hybrid modeling, fusion assessment, lifetime prediction, and adaptive control processes into explicit functional entities. This not only makes the system logic clear, easy to implement and maintain, but also creates an end-to-end intelligent system from data input (data acquisition module) to the first four units of intelligent analysis and prediction (predictive maintenance module) and then to proactive closed-loop response (closed-loop control unit). It has extremely high scalability and robustness, and can effectively realize predictive maintenance and autonomous health management of the edge gateway.

[0168] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A data acquisition method for an edge gateway, characterized in that, include: S1 collects multi-dimensional status data from the edge gateway, including operating conditions and environment data, as well as degradation data of key components; S2, based on operating condition and environmental data and a preset physical failure model, calculates the overall degradation rate of the capacitor and determines the capacitor health; and based on the degradation data of key components and a preset data-driven model, determines the flash memory health. S3 combines capacitor health and flash memory health, and calculates the basic health based on preset component importance weights; and determines the environmental penalty factor based on operating condition and environmental data and a pre-trained anomaly detection model. S4 integrates the basic health score and environmental penalty factor through a preset multiplicative penalty model to determine the gateway's overall health score; and based on the time series of the gateway's overall health score, estimates the instantaneous degradation rate through a preset state-space model to determine the remaining service life. S5, based on real-time onboard temperature, preset safe temperature threshold, preset load-temperature control coefficient, preset minimum performance factor and original data acquisition frequency in the operating condition and environmental data, calculates the corrected data acquisition frequency through load adaptive closed-loop control logic and outputs adaptive control commands. Key component degradation data includes: cumulative write / erase cycles of NAND Flash; number of ECC error corrections; number of bad blocks; and output voltage ripple of the power module. The overall degradation rate of the capacitor is calculated using a physical failure model, including: The thermal aging rate of the base is calculated based on the Arrhenius model and the plate temperature. Combined with basic thermal aging rate CPU utilization With output voltage ripple And introduce preset load and voltage stress sensitivity coefficients. , Calculate the overall degradation rate of the capacitor. : in, This is the preset rated reference voltage ripple.

2. The data acquisition method for an edge gateway according to claim 1, characterized in that, Operating and environmental data include: CPU utilization; network throughput; onboard temperature; ambient humidity; triaxial vibration acceleration; and power input voltage fluctuation.

3. The data acquisition method for an edge gateway according to claim 1, characterized in that, Identify environmental penalty factors, including: Using triaxial vibration acceleration, ambient humidity, and power input voltage fluctuation as inputs to the anomaly detection model, the real-time vibration anomaly score, real-time humidity anomaly score, and real-time power anomaly score are calculated respectively. By combining the real-time vibration anomaly score, the real-time humidity anomaly score, the real-time power supply anomaly score, and the preset environmental pressure penalty weight, the environmental penalty factor is determined through minimum amplitude processing.

4. The data acquisition method for an edge gateway according to claim 1, characterized in that, Determine the remaining useful life, including: Using the time series of the gateway's overall health score as observations, the instantaneous degradation rate is estimated by applying a state-space model with a particle filter or Kalman filter framework. The remaining service life is calculated by linear extrapolation based on the instantaneous degradation rate, the current gateway's overall health score, and the preset failure threshold.

5. The data acquisition method for an edge gateway according to claim 1, characterized in that, Also includes: In response to a gateway’s overall health score being less than a preset health threshold or its remaining lifespan being less than a preset time threshold, a predictive maintenance alarm is triggered and a maintenance work order is automatically generated.

6. A predictive maintenance system for an edge gateway, based on the data acquisition method for an edge gateway according to any one of claims 1-5, characterized in that, include: The data acquisition module is used to collect multi-dimensional status data of the edge gateway, including operating condition and environmental data, as well as degradation data of key components. The predictive maintenance module includes: The health assessment unit is used to calculate the overall degradation rate of capacitors and determine the health of capacitors based on operating condition and environmental data and a preset physical failure model; and to determine the health of flash memory based on degradation data of key components and a preset data-driven model. The fusion evaluation unit combines capacitor health and flash memory health, calculates basic health based on preset component importance weights, and determines environmental penalty factors based on operating condition and environmental data and a pre-trained anomaly detection model. The integrated processing unit is used to combine the basic health score and environmental penalty factors through a preset multiplicative penalty model to determine the overall health score of the gateway. The lifetime prediction unit is used to estimate the instantaneous degradation rate and determine the remaining lifetime based on the time series of the gateway's comprehensive health score and a preset state-space model. The closed-loop control unit is used to calculate and correct the data acquisition frequency based on the real-time onboard temperature, preset safe temperature threshold, preset load-temperature control coefficient, preset minimum performance factor and original data acquisition frequency in the operating condition and environmental data, and output adaptive control commands through load adaptive closed-loop control logic.

Images