PINN-based computer structure health monitoring method and system

Abstract

The invention discloses a PINN-based computer structure health monitoring method and system, and relates to the technical field of computer structure health monitoring, and the PINN-based computer structure health monitoring method mainly comprises the steps: collecting structure health parameters, and obtaining a health assessment data packet according to the structure health parameters; according to the health assessment data packet, performing assessment by using a PINN model to obtain a component health assessment level and a case state assessment level; and giving an alarm according to the component health assessment level and the case state assessment level. By implementing the PINN-based computer structure health monitoring method and system provided by the invention, normal parameter changes under different working conditions can be adapted, and the accuracy and pertinence of computer structure health monitoring are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of computer structure health monitoring, and more specifically, to a computer structure health monitoring method and system based on PINN. Background Art

[0002] With the widespread application of computers in various fields, the stability and health of computer hardware structures play a vital role in ensuring their normal operation and data security. However, the computer structure health monitoring system in the prior art still has the following shortcomings in the actual application process: because the normal operating parameter range of each component will change under different load conditions, the monitoring system in the prior art cannot analyze the specific working conditions of the computer and determine the health status of the computer components according to the specific working conditions. It is difficult to dynamically adapt to the differences in working conditions and is easy to misjudge the status of components when the working conditions change. In addition, when monitoring the health status of the computer, it is relatively simple and cannot comprehensively analyze the temperature, current and vibration data of the computer during operation. Therefore, there is an urgent need for a computer structure health monitoring system based on physical information neural networks (Physics-Informed Neural Networks, PINN). Summary of the invention

[0003] The purpose of the present invention is to provide a computer structure health monitoring method and system based on PINN, which can adapt to normal parameter changes under different working conditions and improve the accuracy and pertinence of computer structure health monitoring.

[0004] The present invention provides a computer structure health monitoring method based on PINN, comprising the following steps: S1: collecting structure health parameters, and obtaining a health assessment data packet according to the structure health parameters; S2: evaluating the health assessment data packet using a PINN model to obtain a component health assessment level and a chassis status assessment level; S3: issuing an alarm according to the component health assessment level and the chassis status assessment level.

[0005] Further, step S2 specifically includes: S21: based on the health assessment data packet, using the PINN model, obtaining the usage conditions within the time window of the health assessment time interval corresponding to the current computer; S22: based on the usage conditions and the health assessment data packet, using the PINN model, obtaining the component health assessment index; S23: based on the usage conditions and the health assessment data packet, using the PINN model, obtaining the chassis status assessment index; S24: based on the component health assessment index and the chassis status assessment index, obtaining the component health assessment level and the chassis status assessment level.

[0006] Furthermore, step S21 specifically includes: extracting performance parameters within the time window of the health assessment time interval corresponding to the current computer from the health assessment data packet and parsing them to obtain CPU usage data and memory usage data within the time window of the health assessment time interval corresponding to the current computer; averaging the CPU usage data and memory usage data within the time window of the health assessment time interval corresponding to the current computer, respectively, to obtain the average CPU usage and average memory usage within the time window of the health assessment time interval corresponding to the current computer; obtaining a CPU load score and a memory load score based on the average CPU usage and the average memory usage; and obtaining a comprehensive load value based on the CPU load score and the memory load score.

[0007] Further, step S22 specifically includes: according to the temperature data of each component of the current computer in the health assessment data packet, using the standard deviation method, obtaining the discrete temperature performance value of each component within the set time window; according to the discrete temperature performance value and the reference temperature performance value of each component within the set time window, obtaining the fluctuating performance component and the stable performance component; according to the fluctuating performance component and the stable performance component, obtaining the health temperature evaluation value of each component; according to the use condition and the health temperature evaluation value of each component, obtaining the maximum allowable value of the health temperature evaluation value of each component corresponding to each use condition; according to the health temperature evaluation value of each component and the maximum allowable value of the health temperature evaluation value of each component corresponding to each use condition, obtaining the health temperature ratio of each component of the computer within the set time window; according to the use condition, setting a reference fluctuation range and a maximum allowable value of the current data of each component corresponding to each use condition; according to the current data of each component of the current computer in the health assessment data packet and a reference fluctuation range and a maximum allowable value of the current data of each component corresponding to each use condition, obtaining the duration of the current data of each component outside the reference fluctuation range in the set time window; obtaining the duration of the current data of each component of the computer within the set time window The maximum current data of each component of the computer is obtained by calculating the ratio of the maximum current data of each component of the computer to the corresponding maximum allowable value, and obtaining the current ratio of each component of the computer within the set time window; according to the interval of each group of time lengths corresponding to the duration, the influence coefficient corresponding to the interval of each group of time lengths is obtained; the duration of the current data of each component in the set time window being outside the reference fluctuation range is matched with the influence coefficient corresponding to the interval of each group of time lengths, and the influence coefficient of each component of the computer within the set time window is obtained; the influence coefficient of each component of the computer within the set time window is multiplied by the current ratio of each component of the computer within the set time window, and the health flow ratio of each component of the computer within the set time window is obtained; the health temperature ratio of each component of the computer within the set time window and the health flow ratio of each component of the computer within the set time window are used as the length value and width value of the rectangle respectively, and the area of ​​the rectangle is used as the component performance evaluation index of each component of the computer within the set time window; the weight coefficients of different components of the computer are multiplied by the component performance evaluation index of each component of the computer within the set time window, and then summed to obtain the component health evaluation index of the computer within the set time window before the arrival time point.

[0008] Further, step S23 specifically includes: obtaining the chassis vibration data of the computer within a set time window according to the chassis vibration data of each component of the current computer in the health assessment data packet, and setting an allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition according to the use condition; comparing the chassis vibration data of the computer within the set time window with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition, and obtaining the cumulative time length of the chassis vibration data of the computer within the set time window being higher than the allowable amplitude threshold value as the high vibration time length; according to the health assessment data packet when The chassis vibration data of each component of the computer before the arrival are extracted, the highest chassis vibration data of the computer within the set time window is extracted, and the ratio is calculated with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each corresponding usage condition, so as to obtain the vibration performance value of the computer within the set time window; according to the intervals of each group of time lengths corresponding to the high vibration duration, the additional coefficient corresponding to the interval of each group of time lengths is obtained; the additional coefficient corresponding to the interval of each group of time lengths is multiplied by the vibration performance value of the computer within the set time window, so as to obtain the chassis status evaluation index of the computer within the set time window before the arrival time point.

[0009] The present invention also provides a computer structure health monitoring system based on PINN, comprising the following modules: a data processing module is configured to collect structure health parameters and obtain a health assessment data packet according to the structure health parameters; a PINN model assessment module is configured to perform an assessment using a PINN model according to the health assessment data packet to obtain a component health assessment level and a chassis status assessment level; an alarm module is configured to issue an alarm according to the component health assessment level and the chassis status assessment level.

[0010] Furthermore, the above-mentioned PINN model evaluation module is specifically configured as follows: based on the health assessment data packet, using the PINN model, obtain the usage conditions within the time window of the health assessment time interval corresponding to the current computer; based on the usage conditions and the health assessment data packet, using the PINN model, obtain the component health assessment index; based on the usage conditions and the health assessment data packet, using the PINN model, obtain the chassis status assessment index; based on the component health assessment index and the chassis status assessment index, obtain the component health assessment level and the chassis status assessment level.

[0011] Furthermore, the above-mentioned use of the PINN model based on the health assessment data packet to obtain the usage conditions within the time window of the health assessment time interval corresponding to the current computer specifically includes: extracting the performance parameters within the time window of the health assessment time interval corresponding to the current computer from the health assessment data packet and parsing them to obtain the CPU usage data and memory usage data within the time window of the health assessment time interval corresponding to the current computer; averaging the CPU usage data and memory usage data within the time window of the health assessment time interval corresponding to the current computer, respectively, to obtain the average CPU usage and average memory usage within the time window of the health assessment time interval corresponding to the current computer; obtaining the CPU load score and memory load score based on the CPU average usage and the memory average usage; and obtaining a comprehensive load value based on the CPU load score and the memory load score.

[0012] Furthermore, the above-mentioned component health assessment index is obtained by using the PINN model according to the use conditions and the health assessment data package, which specifically includes: according to the temperature data of each component of the current computer in the health assessment data package, using the standard deviation method, obtaining the discrete temperature performance value of each component within the set time window; according to the discrete temperature performance value and the reference temperature performance value of each component within the set time window, obtaining the fluctuating performance component and the stable performance component; according to the fluctuating performance component and the stable performance component, obtaining the health temperature assessment value of each component; according to the use conditions and the health temperature assessment value of each component, obtaining the temperature corresponding to each use condition The maximum allowable value of the temperature evaluation value of each component; according to the temperature evaluation value of each component and the maximum allowable value of the temperature evaluation value of each component corresponding to each use condition, the temperature ratio of each component of the computer within the set time window is obtained; according to the use condition, a reference fluctuation range and a maximum allowable value of the current data of each component are set to correspond to each use condition; according to the current data of each component of the current computer in the health assessment data packet and a reference fluctuation range and a maximum allowable value of the current data of each component corresponding to each use condition, the duration of the current data of each component in the set time window being outside the reference fluctuation range is obtained; The highest current data of each component of the computer within the set time window is taken, and the ratio of the highest current data of each component of the computer to the corresponding maximum allowable value is calculated to obtain the current ratio of each component of the computer within the set time window; according to the interval of each group of time lengths corresponding to the duration, the influence coefficient corresponding to the interval of each group of time lengths is obtained; the duration of the current data of each component in the set time window being outside the reference fluctuation range is matched with the influence coefficient corresponding to the interval of each group of time lengths to obtain the influence coefficient of each component of the computer within the set time window; the influence coefficient of each component of the computer within the set time window is multiplied by the current ratio of each component of the computer within the set time window to obtain the health flow ratio of each component of the computer within the set time window; the health temperature ratio of each component of the computer within the set time window and the health flow ratio of each component of the computer within the set time window are used as the length value and width value of the rectangle respectively, and the area of ​​the rectangle is used as the component performance evaluation index of each component of the computer within the set time window; the weight coefficients of different components of the computer are multiplied by the component performance evaluation index of each component of the computer within the set time window, and then summed to obtain the component health evaluation index of the computer within the set time window before the arrival time point.

[0013] Furthermore, the above-mentioned chassis status evaluation index is obtained by using the PINN model according to the use conditions and the health assessment data packet, specifically including: obtaining the chassis vibration data of the computer within a set time window according to the chassis vibration data of each component of the current computer in the health assessment data packet, and setting an allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition according to the use conditions; comparing the chassis vibration data of the computer within the set time window with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition, and obtaining the cumulative time length of the computer chassis vibration data within the set time window that is higher than the allowable amplitude threshold, as High vibration duration; according to the chassis vibration data of each component of the current computer in the health assessment data package, the highest chassis vibration data of the computer within the set time window is extracted, and the ratio is calculated with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each corresponding usage condition, so as to obtain the vibration performance value of the computer within the set time window; according to the intervals of each group of durations corresponding to the high vibration duration, the additional coefficient corresponding to the interval of each group of durations is obtained; the additional coefficient corresponding to the interval of each group of durations is multiplied by the vibration performance value of the computer within the set time window, so as to obtain the chassis status evaluation index of the computer within the set time window before the arrival time point.

[0014] The implementation of the computer structure health monitoring method and system based on PINN provided by the present invention has the following beneficial effects: In order to solve the problems pointed out in the background technology, the present invention proposes a computer structure health monitoring system based on PINN; the present invention determines the computer usage condition by analyzing the CPU usage rate and memory usage rate of the computer, and sets different reference thresholds of parameters such as component temperature, current, chassis vibration, etc. for different working conditions, that is, the maximum value and reference fluctuation range are allowed, so that the evaluation is more in line with the actual operation situation and avoids misjudgment caused by changes in working conditions; when evaluating component health and chassis status, the working condition factors are fully considered, and the evaluation criteria are adjusted according to the working conditions, so that the evaluation results can accurately reflect the real health status of the computer structure under different working conditions, that is, when evaluating the health of the hard disk, according to the working conditions, the evaluation criteria are adjusted according to the working conditions. The present invention adjusts the evaluation criteria of its temperature and current according to the conditions, adapts to the normal parameter changes under different working conditions, and improves the accuracy and pertinence of computer structure health monitoring; the present invention uses the pre-built PINN model through the PINN model evaluation module, combines the real-time collected data and computer usage conditions, comprehensively evaluates the component health and chassis status, and outputs quantitative evaluation indexes and grades, overcomes the limitation of the prior art that only relies on a single indicator or experience judgment, and improves the evaluation accuracy; the present invention timely discovers abnormal parameter changes and issues fault warnings in advance through real-time monitoring and analysis of multiple parameters such as component temperature, current and chassis vibration, combined with the set working condition-related thresholds.

[0015] The present invention determines the computer usage condition by analyzing the CPU usage rate and memory usage rate of the computer, and sets different reference thresholds of parameters such as component temperature, current, chassis vibration, etc. for different working conditions, that is, the maximum value and reference fluctuation range are allowed, so that the evaluation is more in line with the actual operation situation and misjudgment caused by changes in working conditions is avoided; when evaluating the health of components and the status of the chassis, the working condition factors are fully considered, and the evaluation standard is adjusted according to the working condition, so that the evaluation result can accurately reflect the real health status of the computer structure under different working conditions, that is, when evaluating the health of the hard disk, the evaluation standard of its temperature and current is adjusted according to the working condition, adapting to the normal parameter changes under different working conditions, thereby improving the accuracy and pertinence of computer structure health monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which: Figure 1 It is a flow chart of a computer structure health monitoring method based on PINN provided by the present invention; Figure 2 It is a schematic diagram of a computer structure health monitoring system based on PINN provided by the present invention. DETAILED DESCRIPTION

[0017] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, specific embodiments of the present invention are now described in detail with reference to the accompanying drawings.

[0018] Figure 1 A schematic diagram of a computer structure health monitoring method based on PINN in this embodiment is shown. In this embodiment, the computer structure health monitoring method based on PINN includes the following steps: S1: Collect structural health parameters and obtain health assessment data package based on the structural health parameters; In an exemplary embodiment, step S1 specifically includes: setting a health assessment time interval corresponding to the current computer, confirming that the health assessment time interval corresponding to the current computer has been reached, and collecting structural health parameters within a time window of the health assessment time interval corresponding to the current computer; the structural health parameters include temperature data, current data, chassis vibration data, CPU usage data, and memory usage data of various components of the current computer; the various components of the current computer include a CPU, a graphics card, a motherboard, and a hard disk; pre-processing the structural health parameters within the time window of the health assessment time interval corresponding to the current computer to obtain a health assessment data packet; In an exemplary embodiment, preprocessing is performed on the structural health parameters within the time window of the health assessment time interval corresponding to the current computer, including data cleaning to remove outliers and noise filtering; S2: Based on the health assessment data packet, the PINN model is used to perform an assessment to obtain the component health assessment level and the chassis status assessment level; In an exemplary embodiment, step S2 specifically includes: S21: according to the health assessment data packet, using the PINN model, obtaining the usage condition within the time window of the health assessment time interval corresponding to the current computer; In an exemplary embodiment, the use conditions include light load conditions, medium load conditions, and heavy load conditions; In an exemplary embodiment, step S21 specifically includes: Extracting performance parameters within a time window of a health assessment time interval corresponding to the current computer from the health assessment data packet and parsing them to obtain CPU usage data and memory usage data within a time window of a health assessment time interval corresponding to the current computer; The CPU usage data and the memory usage data in the time window of the health assessment time interval corresponding to the current computer are averaged to obtain the average CPU usage and the average memory usage in the time window of the health assessment time interval corresponding to the current computer; According to the average CPU usage and the average memory usage, the CPU load score and the memory load score are obtained; According to the CPU load score and the memory load score, the comprehensive load value is obtained; As an exemplary embodiment, in step S21, the average rate intervals corresponding to the CPU average usage rate and the memory average usage rate are preset; the average rate interval corresponding to each group of CPU average usage rates is set to correspond to a CPU load score, and the average rate interval corresponding to each group of memory average usage rates is set to correspond to a memory load score; in this embodiment, the ranges of the CPU load score and the memory load score are set to 1-10, and the higher the CPU average usage rate and the memory average usage rate of the computer within the set time window, the higher the corresponding matched CPU load score and memory load score; the CPU average usage rate and the memory average usage rate of the computer within the set time window are converted into the CPU load score and the memory load score respectively, and after the conversion is completed, the CPU load score obtained after the conversion is calculated according to the weight coefficients set for the CPU usage rate and the memory usage rate. The CPU and memory load scores are multiplied by the corresponding weight coefficients respectively, and then the sum is obtained to obtain the comprehensive load value of the computer within the set time window; the intervals of the three groups of load values ​​corresponding to the preset comprehensive load value are set, and each group of load value intervals are set to correspond to a computer usage condition; the higher the comprehensive load value, the higher the possibility of the heavy load condition in the corresponding matching usage condition; the comprehensive load value of the computer within the set time window is matched with the intervals of the corresponding three groups of load values, so as to determine the computer usage condition within the set time window; it should be noted that by performing the above-mentioned comprehensive analysis on the CPU usage data and the memory usage data within the set time window, according to the typical characteristics of the usage rate under different working conditions, the load condition of the computer within this time window can be determined more accurately, thereby providing an important basis for subsequent comprehensive computer structure health assessment and other work; S22: Obtaining a component health assessment index using the PINN model according to the use conditions and the health assessment data package; In an exemplary embodiment, step S22 specifically includes: According to the temperature data of each component of the current computer in the health assessment data package, the discrete temperature performance value of each component within the set time window is obtained by using the standard deviation method; According to the discrete temperature performance value and the reference temperature performance value of each component within the set time window, a fluctuating performance component and a stable performance component are obtained; According to the fluctuating performance components and the stable performance components, the healthy temperature evaluation value of each component is obtained; According to the use conditions and the thermal assessment values ​​of each component, the maximum allowable value of the thermal assessment value of each component corresponding to each use condition is obtained; According to the temperature evaluation value of each component and the maximum allowable value of the temperature evaluation value of each component corresponding to each use condition, the temperature ratio of each component of the computer within the set time window is obtained; According to the use conditions, set a reference fluctuation range and a maximum allowable value of the current data of each component for each use condition; According to the current data of each component of the current computer in the health assessment data package and each use condition, a reference fluctuation range and a maximum allowable value of the current data of each component are respectively corresponded, and the duration of the current data of each component being outside the reference fluctuation range in the set time window is obtained; Acquire the highest current data of each component of the computer within a set time window, and calculate the ratio of the highest current data of each component of the computer to the corresponding maximum allowable value to obtain the current ratio of each component of the computer within the set time window; According to the intervals of each group of durations corresponding to the duration, the influence coefficient corresponding to the interval of each group of durations is obtained; Match the duration of the current data of each component outside the reference fluctuation range in the set time window with the influence coefficient corresponding to the interval of each group of durations, and obtain the influence coefficient of each component of the computer in the set time window; The influence coefficient of each computer component in the set time window is multiplied by the current ratio of each computer component in the set time window to obtain the current ratio of each computer component in the set time window; The temperature ratio of each computer component in a set time window and the current ratio of each computer component in a set time window are used as the length and width of the rectangle respectively, and the area of ​​the rectangle is used as the component performance evaluation index of each computer component in the set time window; Multiply the weight coefficients of different components of the computer by the component performance evaluation index of each component of the computer within a set time window, and then sum them up to obtain the component health evaluation index of the computer within the set time window before reaching the time point; It should be noted that in step S22, the computer components are analyzed and evaluated by combining the two key dimensions of temperature data and current data, and the health status of the components is no longer limited to a single indicator. The temperature can reflect the heat dissipation of the component and whether there is an abnormal heat source, and the current data reflects the power supply stability, load changes, and potential electrical faults of the component. Combining the two can more comprehensively and accurately grasp the actual operating health status of the component, avoiding the one-sidedness and misjudgment that may occur when relying on only one parameter. S23: according to the use condition and the health assessment data packet, using the PINN model, obtain the chassis status assessment index; In an exemplary embodiment, step S23 specifically includes: According to the chassis vibration data of each component of the current computer in the health assessment data packet, the chassis vibration data of the computer within a set time window is obtained, and according to the use conditions, an allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition is set; Compare the chassis vibration data of the computer within the set time window with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition, and obtain the cumulative time length that the chassis vibration data of the computer within the set time window is higher than the allowable amplitude threshold value as the high vibration time length; According to the chassis vibration data of each component of the current computer in the health assessment data package, the highest chassis vibration data of the computer within the set time window is extracted, and the ratio is calculated with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each corresponding use condition, so as to obtain the vibration performance value of the computer within the set time window; According to the intervals of each group of durations corresponding to the high vibration duration, the additional coefficient corresponding to the interval of each group of durations is obtained; The additional coefficient corresponding to each group of duration intervals is multiplied by the vibration performance value of the computer within the set time window to obtain the chassis status evaluation index of the computer within the set time window before the arrival time point; It should be noted that through a series of calculation steps, the chassis vibration condition, which is relatively abstract and difficult to compare intuitively, is converted into a specific chassis condition evaluation index for quantitative expression; this allows a clear numerical measurement standard for the chassis health status; S24: Obtaining a component health assessment level and a chassis status assessment level according to the component health assessment index and the chassis status assessment index; In an exemplary embodiment, step S24 specifically includes: According to the component health assessment index, the chassis status assessment index and the age of the computer, the component health assessment index and the chassis status assessment index of the computer within the set time window are compared with the component health threshold index and the chassis status threshold index respectively to obtain the component health assessment level and the chassis status assessment level; As an exemplary embodiment, a component health assessment index and a chassis state assessment index of a computer within a set time window are extracted; according to the service life of the computer, the component health assessment index and the chassis state assessment index of the computer are preset to correspond to the component health threshold index and the chassis state threshold index respectively; the component health assessment index and the chassis state assessment index of the computer within the set time window are compared with the corresponding preset threshold indexes respectively, if the component health assessment index is less than the component health threshold index, the component health assessment level is determined to be the component health level, otherwise it is determined to be the component abnormal level; if the chassis state assessment index is less than the chassis state threshold index, the chassis state assessment level is determined to be the chassis normal level, otherwise it is determined to be the chassis abnormal level; S3: Issue an alarm based on the component health assessment level and chassis status assessment level; As an exemplary embodiment, in step S3, according to the component health assessment level and the chassis status assessment level, if the component health assessment level is a component abnormality level, M1 is executed; M1: Get the temperature value of the computer's surrounding environment within the current set time window. If the temperature value of the surrounding environment is higher than the set normal ambient temperature value, trigger an environmental abnormality signal and send it to the technician. If the temperature value of the surrounding environment is lower than the set normal ambient temperature value, execute M2. M2: Send the component abnormality level to the technician, and obtain the maintenance date of the computer from the current time point to the most recent maintenance date, and calculate the time difference between the maintenance date and the current time point to obtain the abnormality interval, and compare the abnormality interval with the maintenance time interval of the current computer. If the abnormality interval is less than the maintenance time interval, replace the abnormality interval as the new maintenance time interval of the computer, otherwise do not replace it; If the chassis status assessment level is the chassis abnormality level, the location of the computer is taken as the starting point, and the location of each maintenance personnel is taken as the end point. The distance between the starting point and each end point is obtained, and the maintenance personnel with the shortest distance is selected as the handler of the chassis abnormality level, and the chassis abnormality level is sent to the handler.

[0019] In an exemplary embodiment, a computer structure health monitoring system based on PINN includes the following modules: a data processing module is configured to collect structure health parameters and obtain a health assessment data packet based on the structure health parameters; a PINN model assessment module is configured to evaluate the health assessment data packet using the PINN model to obtain a component health assessment level and a chassis status assessment level; an alarm module is configured to issue an alarm based on the component health assessment level and the chassis status assessment level. Figure 2 Shown is a schematic diagram of a computer structure health monitoring system based on PINN.

[0020] In an exemplary embodiment, the above-mentioned PINN model evaluation module is specifically configured as follows: based on the health assessment data packet, using the PINN model, obtain the usage conditions within the time window of the health assessment time interval corresponding to the current computer; based on the usage conditions and the health assessment data packet, using the PINN model, obtain the component health assessment index; based on the usage conditions and the health assessment data packet, using the PINN model, obtain the chassis status assessment index; based on the component health assessment index and the chassis status assessment index, obtain the component health assessment level and the chassis status assessment level.

[0021] In an exemplary embodiment, the above-mentioned PINN model is used according to the health assessment data packet to obtain the usage conditions within the time window of the health assessment time interval corresponding to the current computer, specifically including: extracting the performance parameters within the time window of the health assessment time interval corresponding to the current computer from the health assessment data packet and parsing them to obtain the CPU usage data and memory usage data within the time window of the health assessment time interval corresponding to the current computer; averaging the CPU usage data and memory usage data within the time window of the health assessment time interval corresponding to the current computer, respectively, to obtain the average CPU usage and average memory usage within the time window of the health assessment time interval corresponding to the current computer; obtaining the CPU load score and memory load score based on the CPU average usage and the memory average usage; and obtaining a comprehensive load value based on the CPU load score and the memory load score.

[0022] In an exemplary embodiment, the component health assessment index is obtained by using the PINN model according to the use conditions and the health assessment data packet, specifically including: according to the temperature data of each component of the current computer in the health assessment data packet, using the standard deviation method, obtaining the discrete temperature performance value of each component within a set time window; according to the discrete temperature performance value and the reference temperature performance value of each component within the set time window, obtaining the fluctuating performance component and the stable performance component; according to the fluctuating performance component and the stable performance component, obtaining the health temperature assessment value of each component; according to the use conditions and the health temperature assessment value of each component, obtaining the maximum allowable value of the health temperature assessment value of each component corresponding to each use condition; according to the health temperature assessment value of each component and the maximum allowable value of the health temperature assessment value of each component corresponding to each use condition, obtaining the health temperature ratio of each component of the computer within the set time window; according to the use conditions, setting a reference fluctuation range and a maximum allowable value of the current data of each component corresponding to each use condition; according to the current data of each component of the current computer in the health assessment data packet and a reference fluctuation range and a maximum allowable value of the current data of each component corresponding to each use condition, obtaining the duration of the current data of each component in the set time window being outside the reference fluctuation range length; obtain the highest current data of each component of the computer within the set time window, and calculate the ratio of the highest current data of each component of the computer to the corresponding maximum allowable value to obtain the current ratio of each component of the computer within the set time window; according to the intervals of each group of durations corresponding to the duration, obtain the influence coefficient corresponding to the interval of each group of durations; match the duration of each component current data outside the reference fluctuation range in the set time window with the influence coefficient corresponding to the interval of each group of durations to obtain the influence coefficient of each component of the computer within the set time window; The response coefficient is multiplied by the current ratio of each computer component within the set time window to obtain the health current ratio of each computer component within the set time window; the health temperature ratio of each computer component within the set time window and the health current ratio of each computer component within the set time window are used as the length and width of the rectangle respectively, and the area of ​​the rectangle is used as the component performance evaluation index of each computer component within the set time window; the weight coefficients of different computer components and the component performance evaluation index of each computer component within the set time window are multiplied, and then summed to obtain the component health evaluation index of the computer within the set time window before the arrival time point.

[0023] In an exemplary embodiment, the chassis status evaluation index is obtained by using the PINN model according to the use conditions and the health assessment data packet, specifically including: obtaining the chassis vibration data of the computer within a set time window according to the chassis vibration data of each component of the current computer in the health assessment data packet, and setting an allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition according to the use conditions; comparing the chassis vibration data of the computer within the set time window with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition, and obtaining the cumulative time when the chassis vibration data of the computer within the set time window is higher than the allowable amplitude threshold. length, as the high vibration duration; according to the chassis vibration data of each component of the current computer in the health assessment data packet, the highest chassis vibration data of the computer in the set time window is extracted, and the ratio is calculated with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each corresponding use condition, so as to obtain the vibration performance value of the computer in the set time window; according to the intervals of each group of durations corresponding to the high vibration duration, the additional coefficient corresponding to the interval of each group of durations is obtained; the additional coefficient corresponding to the interval of each group of durations is multiplied by the vibration performance value of the computer in the set time window, so as to obtain the chassis status evaluation index of the computer in the set time window before the arrival time point.

[0024] As an exemplary embodiment, the above-mentioned computer structure health monitoring system based on PINN can be implemented in the following manner. In this embodiment, the computer structure health monitoring system based on PINN includes a data processing module, a PINN model evaluation module and an alarm module; In an exemplary embodiment, the data processing module is configured as follows: setting a health assessment time interval corresponding to the current computer, and after reaching the set health assessment time interval, collecting the structural health parameters of the computer within the set time window after reaching the time point; wherein the structural health parameters include temperature data and current data of each component, chassis vibration data, and computer CPU usage and memory usage; each component includes a CPU, a graphics card, a motherboard, and a hard disk, etc.; taking the structural health parameters of the computer within the set time window after reaching the time point as a health assessment data packet, and sending it to the PINN model module; It should be noted that the preprocessing operation of the collected structural health parameters mainly includes data cleaning to remove outliers and noise filtering to improve data quality, so that the preprocessed data can be better used by the PINN model, improving the accuracy and stability of model training and prediction; In an exemplary embodiment, the PINN model evaluation module is configured to: receive a health evaluation data packet of the computer within a set time window before the arrival time point, and input it into a pre-built PINN model for comprehensive evaluation, determine the use condition of the computer within the set time window before the arrival time point, wherein the use condition includes a light load condition, a medium load condition, and a heavy load condition; combine the use condition of the computer within the set time window before the arrival time point; determine the component health evaluation index of the computer within the set time window before the arrival time point, and determine the computer and chassis status evaluation index within the set time window before the arrival time point; based on the component health evaluation index zjk and chassis status evaluation index zjy of the computer within the set time window before the arrival time point, output the component health evaluation level and chassis status evaluation level of the computer within the set time window before the arrival time point, and send the component health evaluation level and chassis status evaluation level of the computer output by the PINN model to the alarm module; It should be noted that the relevant physical principles and constraints are determined based on the specific structural health parameters received. For example, the temperature data of each component involves the physical principle of heat conduction. According to the heat transfer situation inside the computer (such as heat exchange between different components, heat dissipation boundary conditions, etc.), the corresponding heat conduction equation and the constraints embodied in its discretized form should be clarified. For the chassis vibration data, it is necessary to consider the dynamic equations followed by the vibration of the chassis as a whole and the internal components, such as the relevant equations of the multi-degree-of-freedom vibration system and its specific constraint expressions under the current actual situation. According to the determined constraints, the reference values ​​and reference ranges of the computer's temperature, current, and vibration under different working conditions are set. By setting the parameters under different working conditions according to the physical constraints, The reference threshold of the number can more accurately judge and warn the health status of the computer structure; for example, when the actual parameter value monitored is close to or exceeds the corresponding threshold, an alarm message can be issued in time to prompt the operation and maintenance personnel to pay attention to possible problems with the computer; compared with setting thresholds based solely on experience or historical statistical data, threshold setting based on physical constraints is more scientific and reliable, because it takes into account the actual physical changes of various components inside the computer under different working conditions, and can discover potential fault hazards that conform to physical laws in advance, avoiding misjudgment (such as misjudging normal parameter changes under high load as faults) or missed judgment (failure to promptly detect abnormal parameter conditions that actually violate physical laws) caused by unreasonable threshold settings; As an exemplary embodiment, the above determination of the usage condition of the computer within a set time window before reaching the time point is specifically as follows: extracting and parsing the performance parameters within the set time window from the computer's health assessment data packet to obtain the CPU usage data and memory usage data of the computer within the set time window; averaging the CPU usage data and memory usage data of the computer at each time point within the set time window, thereby determining the CPU average usage and memory average usage of the computer within the set time window; presetting the average rate intervals corresponding to the CPU average usage and the memory average usage; setting the average rate interval corresponding to each group of CPU average usage to correspond to a CPU load score, and the average rate interval corresponding to each group of memory average usage to correspond to a memory load score; As an exemplary embodiment, the ranges of the CPU load score and the memory load score are both set to 1-10. The higher the average CPU usage rate and the average memory usage rate of the computer within the set time window, the higher the corresponding matched CPU load score and memory load score; the average CPU usage rate and the average memory usage rate of the computer within the set time window are respectively converted into the CPU load score and the memory load score. After the conversion is completed, according to the weight coefficients set for the CPU usage rate and the memory usage rate, the CPU load score and the memory load score obtained after the conversion are respectively multiplied by the corresponding set weight coefficients, and then the sum is obtained to obtain the comprehensive load value of the computer within the set time window; the intervals of the three groups of load values ​​corresponding to the preset comprehensive load value are set, and each group of load value intervals are set to correspond to a computer usage condition; It should be noted that the higher the comprehensive load value, the higher the possibility of a heavy load condition in the corresponding matching use condition; Matching the comprehensive load value of the computer within the set time window with the corresponding intervals of the three sets of preset load values, thereby determining the use condition of the computer within the set time window; It should be noted that by performing the above comprehensive analysis on the CPU usage data and memory usage data within the set time window, according to the typical characteristics of the usage under different working conditions, the load condition of the computer within this time window can be determined more accurately, thereby providing an important basis for subsequent comprehensive computer structure health assessment and other work; In an exemplary embodiment, the above determination of the component health assessment index within a set time window before the computer reaches the time point is specifically: The components in the computer that require health assessment are numbered, and the number is represented by a, where a=1, 2, ..., b, and b is the total number of components that require health assessment; the temperature data of each component of the computer at each time point in the set time window is read, and the temperature data of each component at each time point is calculated using the standard deviation formula to obtain the discrete temperature performance value F1 of each component in the set time window; the reference temperature performance value F2 corresponding to the discrete temperature performance value F1 of each component is preset, and the discrete temperature performance value F1 of each component is compared with the corresponding reference temperature performance value F2, and the component with the comparison result of F1>F2 is marked as a fluctuating performance component, and the component with the comparison result of F1<F2 is marked as a stable performance component; for the temperature data of the stable performance component at each time point, the highest temperature data is taken as the health temperature assessment value of the stable performance component; for the temperature data of the fluctuating performance component at each time point, the average value is taken as the health temperature assessment value of the fluctuating performance component; according to the use condition of the computer in the set time window, each use condition is set to correspond to a maximum allowable value of the health temperature assessment value of each component; As an exemplary embodiment, the maximum allowable value of the light load condition is less than the maximum allowable value of the medium load condition and less than the maximum allowable value of the heavy load condition; Calculate the ratio of the temperature evaluation value of each component of the computer within the set time window to the corresponding maximum allowable value, that is, the temperature evaluation value of each component / the maximum allowable value, to obtain the temperature ratio of each component of the computer within the set time window; According to the use condition of the computer within the set time window, a reference fluctuation range and a maximum allowable value of the current data of each component are set to correspond to each use condition; As an exemplary embodiment, the reference fluctuation range of the light load condition is less than the reference fluctuation range of the medium load condition and less than the reference fluctuation range of the heavy load condition, and the maximum allowable value of the light load condition is less than the maximum allowable value of the medium load condition and less than the maximum allowable value of the heavy load condition; Read the current data of each component of the computer at each time point within the set time window, and input it into the corresponding reference fluctuation range. After the input is completed, identify the time segment when the current data of each component in the set time window is outside the reference fluctuation range, and accumulate the time segments to obtain the duration of the current data of each component in the set time window being outside the reference fluctuation range; Acquire the highest current data of each component of the computer within a set time window, and calculate the ratio of the highest current data of each component of the computer to the corresponding maximum allowable value to obtain the current ratio of each component of the computer within the set time window; The preset durations correspond to the intervals of each group of durations, and each interval of each group of durations corresponds to an influence coefficient; As an exemplary embodiment, the influence coefficient range is set to 1-1.187, and when the duration is 0, the value is 1, and the longer the duration is, the greater the corresponding matching influence coefficient; Match the duration of each computer component with the interval of the corresponding duration, so as to determine the influence coefficient of each computer component within the set time window, and multiply the influence coefficient determined by each component with the health flow ratio to obtain the health flow ratio of each computer component within the set time window; The temperature ratio and current ratio of each computer component in a set time window are used as the length and width of the rectangle, respectively, to construct a rectangular model, and the area of ​​the rectangle is used as the component performance evaluation index of each computer component in the set time window; According to the importance of the components, weight coefficients of different components of the computer are set, the component performance evaluation index of each component of the computer is multiplied by the corresponding set weight coefficient, and then the sum is obtained to obtain the component health evaluation index of the computer within the set time window before the arrival time point; It should be noted that by combining the two key dimensions of temperature data and current data to analyze and evaluate each computer component, it is no longer limited to a single indicator to judge the health of the component; temperature can reflect the heat dissipation of the component and whether there is an abnormal heat source, and current data reflects the power supply stability, load changes, and potential electrical failures of the component. Combining the two can more comprehensively and accurately grasp the actual operating health status of the component, avoiding the one-sidedness and misjudgment that may occur when relying on only one parameter. In an exemplary embodiment, the above determination of the sum of the chassis status evaluation index within the time window set by the computer before reaching the time point is specifically: Read the chassis vibration data of the computer within a set time window, and set an allowable amplitude threshold of the chassis vibration amplitude corresponding to each operating condition according to the operating condition of the computer within the set time window; As an exemplary embodiment, the allowable amplitude threshold of the light load condition is less than the allowable amplitude threshold of the medium load condition and less than the allowable amplitude threshold of the heavy load condition; Compare the chassis vibration data of the computer within the set time window with the corresponding allowable amplitude threshold, and identify the cumulative duration of the chassis vibration data of the computer within the set time window being higher than the allowable amplitude threshold as the high vibration duration; Extract the highest chassis vibration data of the computer within the set time window, and calculate the ratio with the corresponding allowable amplitude threshold, that is, obtain the vibration performance value of the computer within the set time window through the highest chassis vibration data / highest chassis vibration data; The intervals of each group of durations corresponding to the high vibration duration are preset, and an additional coefficient is set for each group of duration intervals; As an exemplary embodiment, the influence coefficient range is set to 1-1.239, and when the high vibration duration is 0, the value is 1, and the longer the high vibration duration is, the larger the corresponding matching additional coefficient is; Match the high vibration duration of the computer with the interval of the corresponding duration, so as to determine the additional coefficient of the computer in the set time window, and multiply the additional coefficient of the computer in the set time window with the vibration performance value, so as to determine the chassis state evaluation index zjy of the computer in the set time window before reaching the time point; It should be noted that, through a series of calculation steps, the relatively abstract and difficult to compare state of chassis vibration is converted into a specific chassis state evaluation index zjy for quantitative expression; this enables a clear numerical measurement standard for the health of the chassis; In an exemplary embodiment, the above outputs the component health assessment level and chassis status assessment level of the computer within the time window set before the arrival time point based on the component health assessment index zjk and the chassis status assessment index zjy within the time window set before the computer arrives at the time point, specifically: Extract the component health evaluation index zjk and chassis status evaluation index zjy of the computer within the set time window; According to the age of the computer, the component health threshold index and chassis status threshold index corresponding to the computer component health assessment index zjk and chassis status assessment index zjy are preset respectively; The component health assessment index zjk and chassis status assessment index zjy of the computer within the set time window are respectively compared with the corresponding preset threshold indexes. If the component health assessment index zjk is less than the component health threshold index, the component health assessment level is determined to be the component health level, otherwise it is determined to be the component abnormal level; if the chassis status assessment index zjy is less than the chassis status threshold index, the chassis status assessment level is determined to be the chassis normal level, otherwise it is determined to be the chassis abnormal level; In an exemplary embodiment, the alarm module is configured to: receive the component health assessment level and chassis status assessment level of the computer output by the PINN model, and if the component health assessment level is a component abnormality level, first execute M1; M1: Get the temperature value of the computer's surrounding environment within the current set time window. If the temperature value of the surrounding environment is higher than the set normal ambient temperature value, trigger an environmental abnormality signal and send it to the technician. If the temperature value of the surrounding environment is lower than the set normal ambient temperature value, execute M2. M2: Send the component abnormality level to the technician, and obtain the maintenance date of the computer from the current time point to the most recent maintenance date, and calculate the time difference between the maintenance date and the current time point to obtain the abnormality interval, and compare the abnormality interval with the maintenance time interval of the current computer. If the abnormality interval is less than the maintenance time interval, replace the abnormality interval as the new maintenance time interval of the computer, otherwise do not replace it; If the chassis status assessment level is the chassis abnormality level, the location of the computer is taken as the starting point, the location of each maintenance personnel is taken as the end point, the distance between the starting point and each end point is obtained, the maintenance personnel with the shortest distance is selected as the processing personnel of the chassis abnormality level, and the chassis abnormality level is sent to the processing personnel; It should be noted that by taking different response steps according to the component health assessment level and the chassis status assessment level respectively, it is possible to accurately classify and handle abnormal situations in different parts of the computer; there are clear processing logics for component abnormalities and chassis abnormalities, which avoids treating all fault conditions in a general way, making subsequent maintenance operations more targeted, helping to quickly locate the root cause of the problem and implement effective solutions, thereby improving the efficiency of fault repair.

[0025] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A computer structure health monitoring method based on PINN, characterized in that: The following steps are involved: S1: Collecting structural health parameters, and obtaining a health assessment data packet according to the structural health parameters; S2: According to the health assessment data packet, the PINN model is used to perform an assessment to obtain a component health assessment level and a chassis status assessment level; S3: issuing an alarm according to the component health assessment level and the chassis status assessment level.

2. The computer structure health monitoring method based on PINN according to claim 1, characterized in that: Step S2 specifically includes: S21: According to the health assessment data packet, using the PINN model, obtaining the usage condition within the time window of the health assessment time interval corresponding to the current computer; S22: Obtaining a component health assessment index using a PINN model according to the use condition and the health assessment data packet; S23: obtaining a chassis status evaluation index using a PINN model according to the use condition and the health evaluation data packet; S24: Obtain a component health assessment level and a chassis status assessment level according to the component health assessment index and the chassis status assessment index.

3. The computer structure health monitoring method based on PINN according to claim 2, characterized in that: Step S21 specifically includes: extracting the performance parameters within the time window of the health assessment time interval corresponding to the current computer from the health assessment data packet and parsing them to obtain the CPU usage data and memory usage data within the time window of the health assessment time interval corresponding to the current computer; averaging the CPU usage data and memory usage data within the time window of the health assessment time interval corresponding to the current computer, respectively, to obtain the average CPU usage and average memory usage within the time window of the health assessment time interval corresponding to the current computer; obtaining a CPU load score and a memory load score based on the average CPU usage and the average memory usage; and obtaining a comprehensive load value based on the CPU load score and the memory load score.

4. The computer structure health monitoring method based on PINN according to claim 2, characterized in that: Step S22 specifically includes: According to the temperature data of each component of the current computer in the health assessment data packet, a discrete temperature performance value of each component within a set time window is obtained by using a standard deviation method; According to the discrete temperature performance value and the reference temperature performance value of each component within the set time window, a fluctuating performance component and a stable performance component are obtained; According to the fluctuation performance component and the stable performance component, a healthy temperature evaluation value of each component is obtained; According to the use conditions and the thermal assessment values ​​of the components, the maximum allowable value of the thermal assessment value of each component corresponding to each use condition is obtained; According to the thermal evaluation value of each component and the maximum allowable value of the thermal evaluation value of each component corresponding to each use condition, the thermal ratio of each component of the computer within a set time window is obtained; According to the use conditions, a reference fluctuation range and an allowable maximum value of the current data of each component are set to correspond to each use condition; According to the current data of each component of the current computer in the health assessment data packet and each use condition respectively corresponding to a reference fluctuation range and an allowable maximum value of the current data of each component, the duration of the current data of each component being outside the reference fluctuation range in the set time window is obtained; Acquire the highest current data of each component of the computer within a set time window, and calculate the ratio of the highest current data of each component of the computer to the corresponding maximum allowable value to obtain the current ratio of each component of the computer within the set time window; According to the intervals of each group of durations corresponding to the duration, the influence coefficient corresponding to the interval of each group of durations is obtained; Match the duration of the current data of each component outside the reference fluctuation range in the set time window with the influence coefficient corresponding to the interval of each group of durations, and obtain the influence coefficient of each component of the computer in the set time window; The influence coefficient of each computer component in the set time window is multiplied by the current ratio of each computer component in the set time window to obtain the current ratio of each computer component in the set time window; The temperature ratio of each computer component in a set time window and the current ratio of each computer component in a set time window are used as the length and width of the rectangle respectively, and the area of ​​the rectangle is used as the component performance evaluation index of each computer component in the set time window; The weight coefficients of different computer components and the component performance evaluation index of each computer component within a set time window are multiplied and then summed to obtain the component health evaluation index of the computer within the set time window before reaching the time point.

5. The computer structure health monitoring method based on PINN according to claim 2, characterized in that: Step S23 specifically includes: According to the chassis vibration data of each component of the current computer in the health assessment data packet, the chassis vibration data of the computer within a set time window is obtained, and according to the use conditions, an allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition is set; Compare the chassis vibration data of the computer within the set time window with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition, and obtain the cumulative time length that the chassis vibration data of the computer within the set time window is higher than the allowable amplitude threshold value as the high vibration time length; According to the chassis vibration data of each component of the current computer in the health assessment data packet, extract the highest chassis vibration data of the computer within a set time window, and calculate the ratio with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each corresponding use condition, so as to obtain the vibration performance value of the computer within the set time window; According to the intervals of each group of durations corresponding to the high vibration duration, the additional coefficient corresponding to the interval of each group of durations is obtained; The additional coefficient corresponding to the interval of each group of duration is multiplied by the vibration performance value of the computer within the set time window to obtain the chassis status evaluation index of the computer within the set time window before reaching the time point.

6. A computer structure health monitoring system based on PINN, characterized in that: The PINN-based computer structure health monitoring system includes the following modules: The data processing module is configured to: collect structural health parameters, and obtain a health assessment data packet based on the structural health parameters; The PINN model evaluation module is configured to: perform evaluation using the PINN model according to the health evaluation data packet to obtain a component health evaluation level and a chassis status evaluation level; The alarm module is configured to generate an alarm according to the component health assessment level and the chassis status assessment level.

7. The computer structure health monitoring system based on PINN according to claim 6, characterized in that: The specific configuration of the PINN model evaluation module is: According to the health assessment data packet, using the PINN model, obtaining the usage condition within the time window of the health assessment time interval corresponding to the current computer; According to the use condition and the health assessment data packet, using the PINN model, obtain a component health assessment index; According to the use condition and the health assessment data packet, using the PINN model, a chassis status assessment index is obtained; According to the component health assessment index and the chassis status assessment index, a component health assessment grade and a chassis status assessment grade are obtained.

8. The computer structure health monitoring system based on PINN according to claim 6, characterized in that: The method of obtaining the usage condition within the time window of the health assessment time interval corresponding to the current computer based on the health assessment data packet and utilizing the PINN model specifically includes: extracting the performance parameters within the time window of the health assessment time interval corresponding to the current computer from the health assessment data packet and parsing them to obtain the CPU usage data and the memory usage data within the time window of the health assessment time interval corresponding to the current computer; averaging the CPU usage data and the memory usage data within the time window of the health assessment time interval corresponding to the current computer, respectively, to obtain the average CPU usage and the average memory usage within the time window of the health assessment time interval corresponding to the current computer; obtaining a CPU load score and a memory load score based on the average CPU usage and the average memory usage; and obtaining a comprehensive load value based on the CPU load score and the memory load score.

9. The computer structure health monitoring system based on PINN according to claim 6, characterized in that: The obtaining of the component health assessment index by using the PINN model according to the use condition and the health assessment data packet specifically includes: According to the temperature data of each component of the current computer in the health assessment data packet, a discrete temperature performance value of each component within a set time window is obtained by using a standard deviation method; According to the discrete temperature performance value and the reference temperature performance value of each component within the set time window, a fluctuating performance component and a stable performance component are obtained; According to the fluctuation performance components and the stable performance components, the temperature health evaluation value of each component is obtained; according to the use conditions and the temperature health evaluation value of each component, the maximum allowable value of the temperature health evaluation value of each component corresponding to each use condition is obtained; According to the thermal evaluation value of each component and the maximum allowable value of the thermal evaluation value of each component corresponding to each use condition, the thermal ratio of each component of the computer within a set time window is obtained; According to the use conditions, a reference fluctuation range and an allowable maximum value of the current data of each component are set to correspond to each use condition; According to the current data of each component of the current computer in the health assessment data packet and each use condition respectively corresponding to a reference fluctuation range and an allowable maximum value of the current data of each component, the duration of the current data of each component being outside the reference fluctuation range in the set time window is obtained; Acquire the highest current data of each component of the computer within a set time window, and calculate the ratio of the highest current data of each component of the computer to the corresponding maximum allowable value to obtain the current ratio of each component of the computer within the set time window; According to the intervals of each group of durations corresponding to the duration, the influence coefficient corresponding to the interval of each group of durations is obtained; Match the duration of the current data of each component outside the reference fluctuation range in the set time window with the influence coefficient corresponding to the interval of each group of durations, and obtain the influence coefficient of each component of the computer in the set time window; The influence coefficient of each computer component in the set time window is multiplied by the current ratio of each computer component in the set time window to obtain the current ratio of each computer component in the set time window; The temperature ratio of each computer component in a set time window and the current ratio of each computer component in a set time window are used as the length and width of the rectangle respectively, and the area of ​​the rectangle is used as the component performance evaluation index of each computer component in the set time window; The weight coefficients of different computer components and the component performance evaluation index of each computer component within a set time window are multiplied and then summed to obtain the component health evaluation index of the computer within the set time window before reaching the time point.

10. The computer structure health monitoring system based on PINN according to claim 6, characterized in that: The obtaining of the chassis status evaluation index by using the PINN model according to the use condition and the health evaluation data packet specifically includes: According to the chassis vibration data of each component of the current computer in the health assessment data packet, the chassis vibration data of the computer within a set time window is obtained, and according to the use conditions, an allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition is set; Compare the chassis vibration data of the computer within the set time window with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each use condition, and obtain the cumulative time length that the chassis vibration data of the computer within the set time window is higher than the allowable amplitude threshold value as the high vibration time length; According to the chassis vibration data of each component of the current computer in the health assessment data packet, extract the highest chassis vibration data of the computer within a set time window, and calculate the ratio with the allowable amplitude threshold of the chassis vibration amplitude corresponding to each corresponding use condition, so as to obtain the vibration performance value of the computer within the set time window; According to the intervals of each group of durations corresponding to the high vibration duration, the additional coefficient corresponding to the interval of each group of durations is obtained; The additional coefficient corresponding to the interval of each group of duration is multiplied by the vibration performance value of the computer within the set time window to obtain the chassis status evaluation index of the computer within the set time window before reaching the time point.

Images