Server hardware fault early warning method and system based on intelligent optimization algorithm

Abstract

The application relates to a server hardware fault early warning method and system based on an intelligent optimization algorithm, and the method comprises the following steps: collecting running data information of a server; performing running fault detection on the server according to the running data information to obtain running fault detection information; collecting a hardware image of the server; identifying hardware position defects by using a computer vision recognition algorithm to obtain physical fault detection information; judging whether the server has a fault risk according to the running fault detection information and the physical fault detection information, and if yes, outputting fault early warning information; and if not, outputting running normal information; the application can improve the judgment accuracy by predicting and judging the server fault by using operation data and hardware positions.

Description

Technical Field

[0001] This invention relates to the field of server monitoring technology, and specifically to a server hardware fault early warning method and system based on intelligent optimization algorithms. Background Technology

[0002] The core methods for monitoring server operation status mainly involve installing professional monitoring software (such as Prometheus, Zabbix, Nagios, etc.) to collect basic indicators such as CPU, memory, disk I / O, and network traffic in real time; configuring data storage and visualization (such as Grafana to display time-series data, and Prometheus to store historical data); deploying a log analysis system (such as ELKStack) to centrally manage logs and trigger alarms through keywords; monitoring CPU utilization (monitoring processing capacity and peak load), monitoring memory usage (monitoring utilization and swap space status), monitoring disk performance (monitoring disk I / O throughput and storage space), monitoring network traffic (monitoring bandwidth usage and number of connections), and monitoring service health (monitoring process status and response time); and then issuing alarms or automatically repairing based on the monitoring data. For example, setting tiered threshold alarms (such as an emergency alarm for CPU >90%, and a warning for >70%), configuring multi-channel notifications (email / SMS / IM) to ensure timely response; and writing remediation scripts to achieve automatic recovery from common faults.

[0003] As can be seen, existing technologies for server monitoring all rely on information acquisition, that is, using the operational information of each piece of server hardware to monitor the operation of each hardware component in order to determine whether the server hardware is faulty. However, server hardware failures can include not only performance failures but also physical failures, such as loose or detached hardware, or dust accumulation, all of which can affect the server. These potential problems can all have a certain impact on the use of the server, but acquiring hardware operational information cannot determine whether the server is at risk of failure, thus posing operational security risks to the server. Summary of the Invention

[0004] To address the technical problem in existing technologies where physical hardware failures in servers cannot be identified, thus posing operational security risks, this invention provides a server hardware failure early warning method and system based on intelligent optimization algorithms.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: A server hardware fault early warning method based on intelligent optimization algorithms includes the following steps: Collect server operation data; Based on the operational data information, the server is subjected to operational fault detection to obtain operational fault detection information; Acquire hardware images of the server; Computer vision recognition algorithms are used to identify hardware location defects and obtain physical fault detection information; Based on the operational fault detection information and the physical fault detection information, determine whether the server has a fault risk. If so, output fault warning information; otherwise, output normal operation information.

[0006] The beneficial effects of this invention are: by collecting hardware operation data and hardware location information, it can determine whether the server has a risk of failure. The operation data can reflect the hardware performance failure of the server, and the hardware location information can reflect whether the hardware has a risk of falling off or shifting. If there is a risk of falling off or shifting, it cannot be determined by the operation data. Moreover, once it occurs, falling off or shifting will directly cause the server to fail to operate. Therefore, using computational data and hardware location to predict and judge server failures can improve the accuracy of the judgment.

[0007] Based on the above technical solution, the present invention can be further improved as follows.

[0008] Furthermore, the server's operational data is collected, including the following steps: The CPU of the server was tested for floating-point operations and integer operations to obtain a CPU performance score. A memory testing tool was used to perform a full scan of the memory, and the memory test results were obtained. Simulate read and write operations on the disk to obtain disk throughput; Perform multi-threaded network transmission detection to detect the luminous power of the network transmission module; The operational data information includes the computing performance score, the memory detection result, the disk throughput, and the luminous power.

[0009] Furthermore, based on the operational data information, operational fault detection is performed on the server to obtain operational fault detection information, including the following steps: The computational performance score is compared with a preset computational performance score. If the computational performance score is less than the preset computational performance score, a computational fault message is output; if the computational performance score is greater than or equal to the preset computational performance score, no message is output. If the memory detection result includes memory error information, then a memory status exception message will be output; if the memory detection result does not include the aforementioned memory error information, then no information will be output. The disk throughput is compared with a preset throughput threshold. If the disk throughput is less than the preset throughput threshold, disk status abnormality information is output; if the disk throughput is greater than or equal to the preset throughput threshold, no information is output. The luminous power is compared with a preset luminous efficiency threshold. If the luminous power is less than the preset luminous efficiency threshold, a network communication error message is output. If the luminous power is greater than or equal to the preset luminous efficiency threshold, no message is output. The operational fault detection information includes computational fault information, memory status abnormality information, disk status abnormality information, and network communication abnormality information.

[0010] Furthermore, computer vision recognition algorithms are used to identify hardware location defects and obtain physical fault detection information, including the following steps: The hardware image is preprocessed to obtain a preprocessed image; The position and size of each hardware component of the server in the preprocessed image are identified using computer vision recognition algorithms. Determine whether a physical fault has occurred based on the location and size of each piece of hardware. If so, output the physical fault detection information; otherwise, output no information.

[0011] Further, the hardware image is preprocessed to obtain a preprocessed image; this includes the following steps: The hardware image is then processed to remove the background. The hardware image with the background removed is then processed into grayscale to obtain the preprocessed image.

[0012] Furthermore, the location and dimensions of each piece of hardware from the server in the preprocessed image are identified using computer vision recognition algorithms, including the following steps: The computer vision recognition algorithm is used to perform window binarization on each hardware component in the preprocessed image to obtain the binarized window of each hardware component. The positional dimensions of each hardware component are calculated based on the distance between the binarized windows of each component.

[0013] Furthermore, the positional dimensions of each hardware component are calculated based on the distance between their binarized windows, including the following steps: Use the binarized window of one of the hardware components as the reference window; Construct a rectangular coordinate system using the sides and corners of the reference window; Anchor points are determined on all binarized windows except the reference window; The coordinates of the anchor points on each of the binarized windows in the Cartesian coordinate system are calculated to obtain the position and size of each hardware component.

[0014] Furthermore, the binarized window of one of the hardware components is used as the reference window. The specific steps are as follows: randomly select the binarized window of one of the hardware components as the reference window.

[0015] Furthermore, determining whether a physical fault has occurred based on the location and dimensions of each piece of hardware includes the following steps: Based on the absolute value of the coordinates of the anchor points on each of the binarized windows in the rectangular coordinate system; The absolute values ​​of the coordinates of the anchor points on each of the binarized windows are compared with a preset relative position threshold. If the absolute value of the coordinates of any anchor point on the binarized window is less than or equal to the relative position threshold, it is determined that the position size of the hardware has not experienced a physical fault; if the absolute value of the coordinates of any anchor point on the binarized window is greater than the relative position threshold, it is determined that the position size of the hardware has experienced a physical fault.

[0016] To address the aforementioned technical problems, this invention also provides a server hardware fault early warning system based on intelligent optimization algorithms, the specific technical content of which is as follows: A server hardware fault early warning system based on intelligent optimization algorithms includes: The data acquisition module is used to collect the server's operational data and to acquire hardware images of the server. The fault detection module is used to perform operational fault detection on the server based on the operational data information and obtain operational fault detection information; and to use computer vision recognition algorithms to identify hardware location defects and obtain physical fault detection information. The fault diagnosis module is used to determine whether the server has a fault risk based on the operational fault detection information and the physical fault detection information. If so, it outputs a fault warning message; if not, it outputs a normal operation message. Attached Figure Description

[0017] Figure 1 This is a flowchart of the server hardware fault early warning method based on intelligent optimization algorithm in an embodiment of the present invention; Figure 2 This is a schematic diagram of the server hardware fault early warning system based on intelligent optimization algorithm in an embodiment of the present invention. Detailed Implementation

[0018] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0019] like Figure 1 As shown, this embodiment provides a server hardware fault early warning method based on intelligent optimization algorithms, including the following steps: S1. Collect server operation data information; Collect server operational data, including the following steps: S101. Perform floating-point operation tests and integer operation tests on the CPU of the server to obtain the CPU's computing performance score; S102. Use a memory testing tool to perform a complete scan of the memory and obtain the memory test results; S103. Perform simulated read and write operations on the disk to obtain the disk throughput; S104. Perform multi-threaded network transmission detection to detect the luminous power of the network transmission module; The operational data information includes the computing performance score, the memory detection result, the disk throughput, and the luminous power.

[0020] S2. Perform operational fault detection on the server based on the operational data information to obtain operational fault detection information; Based on the operational data information, the server is subjected to operational fault detection to obtain operational fault detection information, including the following steps: S201. Compare the computational performance score with a preset computational performance score. If the computational performance score is less than the preset computational performance score, output computational fault information; if the computational performance score is greater than or equal to the preset computational performance score, no information is output. S202. If the memory detection result includes memory error information, then output memory status abnormality information; if the memory detection result does not include the memory error information, then no information is output. S203. Compare the disk throughput with a preset throughput threshold. If the disk throughput is less than the preset throughput threshold, output disk status abnormal information; if the disk throughput is greater than or equal to the preset throughput threshold, no information is output. S204. Compare the luminous power with a preset luminous efficiency threshold. If the luminous power is less than the preset luminous efficiency threshold, output network communication abnormality information; if the luminous power is greater than or equal to the preset luminous efficiency threshold, no information is output. S205, The operational fault detection information includes the computational fault information, the memory status abnormal information, the disk status abnormal information, and the network communication abnormal information.

[0021] S3. Acquire the hardware image of the server; Specifically, a high-definition camera is installed inside the server to capture a diagram of the server's internal hardware. The high-definition camera is located inside the server to capture a diagram of the connection relationships between the various hardware components. Since the connection relationships of most servers can be observed on the CPU installation side, the high-definition camera can be placed on the CPU installation side to capture a diagram of the positional relationships of the various hardware components on the CPU installation side.

[0022] S4. Use computer vision recognition algorithms to identify hardware location defects and obtain physical fault detection information; Using computer vision recognition algorithms to identify hardware location defects and obtain physical fault detection information includes the following steps: S401. Preprocess the hardware image to obtain a preprocessed image; The hardware image is preprocessed to obtain a preprocessed image; this includes the following steps: The hardware image is then processed to remove the background. The hardware image with the background removed is then processed into grayscale to obtain the preprocessed image.

[0023] S402. Use a computer vision recognition algorithm to identify the location and dimensions of each piece of hardware in the preprocessed image of the server; the various pieces of hardware in the server are core components such as the processor (CPU) component, memory (RAM), storage device, motherboard, network interface card (NIC), power supply unit (PSU), cooling system, and RAID controller. Since the processor (CPU) component is usually covered by a heatsink, the acquisition of the processor (CPU) component image is to acquire the image of the processor cooling component.

[0024] Identifying the position and dimensions of various hardware components of the server in the preprocessed image using computer vision recognition algorithms includes the following steps: The computer vision recognition algorithm is used to perform window binarization on each hardware component in the preprocessed image to obtain the binarized window of each hardware component. The positional dimensions of each hardware component are calculated based on the distance between the binarized windows of each component.

[0025] The positional dimensions of each hardware component are calculated based on the distance between their binarized windows, including the following steps: The binarized window of one of the hardware components is used as the reference window. The specific steps are as follows: randomly select the binarized window of one of the hardware components as the reference window.

[0026] Construct a rectangular coordinate system using the sides and corners of the reference window; Anchor points are determined on all binarized windows except the reference window; The coordinates of the anchor points on each of the binarized windows in the Cartesian coordinate system are calculated to obtain the position and size of each hardware component.

[0027] When a binary window corresponding to a processor (CPU) component is randomly selected as the reference window, a corner and two sides of the processor (CPU) component can be selected as a Cartesian coordinate system. The vertex of the corner is used as the origin of the coordinate system, and the two sides are used as the two coordinate axes. If the two sides are not perpendicular, any side can be selected, and the line perpendicular to the selected side on an imaginary fish can be used as the coordinate axis. Anchor points are set on the binary windows of core components such as memory (RAM), storage devices, motherboards, network interface cards (NICs), power supply units (PSUs), cooling systems, and RAID controllers. The anchor points are preferably set to the vertices of the corners.

[0028] Preferably, considering that some hardware has an irregular shape, a white anchor buckle is fixed on each piece of hardware, and the anchor buckle is square.

[0029] The rectangular coordinate system is constructed using the sides and corners of the reference window. Specifically, the rectangular coordinate system is constructed using the sides and corners of the binary square image of the anchor buckle on the hardware corresponding to the reference window.

[0030] Anchor points are determined on all binarized windows except the reference window, specifically as follows: The anchor point is determined by setting a preset angle of the binary square image of the anchor buckle on all binary windows except the reference window. The preset angle is any right angle of the anchor buckle.

[0031] As can be seen, by setting white anchor buckles, the anchor buckles can be easily distinguished during window binarization. Furthermore, the anchor buckles are fixed to the hardware and their positions are not easily moved. Since the connection between hardware is usually made by a card slot plug-in structure, the hardware is prone to displacement. Therefore, the fixed anchor buckles can better reflect the positional relationship between the various hardware components.

[0032] S403. Determine whether a physical fault has occurred based on the position and size of each piece of hardware. If so, output the physical fault detection information; otherwise, no information is output.

[0033] Determining whether a physical fault has occurred based on the location and dimensions of each piece of hardware includes the following steps: Based on the absolute value of the coordinates of the anchor points on each of the binarized windows in the rectangular coordinate system; The absolute values ​​of the coordinates of the anchor points on each of the binarized windows are compared with a preset relative position threshold. If the absolute value of the coordinates of any anchor point on the binarized window is less than or equal to the relative position threshold, it is determined that the position size of the hardware has not experienced a physical fault; if the absolute value of the coordinates of any anchor point on the binarized window is greater than the relative position threshold, it is determined that the position size of the hardware has experienced a physical fault.

[0034] S5. Based on the operational fault detection information and the physical fault detection information, determine whether the server has a fault risk. If so, output fault warning information; otherwise, output normal operation information.

[0035] Based on the operational fault detection information and the physical fault detection information, the specific steps for determining whether the server has a fault risk are as follows: If the operational fault detection information includes any one of the following: computational fault information, memory status abnormality information, disk status abnormality information, network communication abnormality information, or physical fault detection information, then it is determined whether the server has a fault risk; if the operational fault detection information does not include the computational fault information and / or the memory status abnormality information and / or the disk status abnormality information and / or the network communication abnormality information and / or the physical fault detection information, then it is determined that the server does not have a fault risk.

[0036] It should be noted that the computer vision recognition algorithms in this invention include, but are not limited to, the following algorithms: Convolutional Neural Networks (CNNs): These extract image features through convolutional layers and are widely used in image classification, object detection, and other fields. Classic models include AlexNet and ResNet (which addresses the degradation problem of deep networks).

[0037] Support Vector Machine (SVM): It achieves classification by constructing a hyperplane in a high-dimensional space. It is suitable for small sample data and is often used for face recognition and medical image analysis.

[0038] Feature extraction algorithms, including SIFT, SURF, ORB, etc., are used to extract key points in images and achieve matching.

[0039] Optical flow method: calculates the motion vector field of pixels in an image sequence. Commonly used algorithms include Lucas-Kanade and Horn-Schunck.

[0040] Object detection algorithms: YOLO series, which treats object detection as a regression problem, offering high real-time performance; the latest versions (such as YOLOv8) achieve a balance between accuracy and speed. Faster R-CNN, an object detection framework based on region proposal networks. SSD, a single-stage object detection algorithm that predicts bounding boxes using multi-scale feature maps.

[0041] Image generation algorithm: Generative Adversarial Networks (GANs) generate realistic images through generators and discriminators, and are applied to style transfer and image inpainting.

[0042] Variational autoencoders (VAEs) are models that generate continuous data distributions.

[0043] This invention assesses server malfunction risk by collecting hardware operation data and hardware location information. Operation data reflects hardware performance failures, while location information indicates potential risks such as detachment or displacement. Risks of detachment or displacement cannot be determined from operation data alone, and such occurrences directly render the server inoperable. Therefore, using computational data and hardware location information to predict and assess server malfunctions improves accuracy. During hardware location detection, a computer vision algorithm is employed to identify the coordinates of the hardware's anchor points in the coordinate system. This determines the positional relationship of each hardware component relative to a reference hardware component (the component corresponding to the reference window). When hardware displacement or loosening occurs, its relative position changes. Significant changes in position indicate a hardware malfunction risk, enabling multi-angle hardware malfunction warnings and improving the accuracy of hardware risk warnings.

[0044] In this invention, since each piece of hardware is fixedly equipped with a white anchor buckle, the dust accumulation status of the server hardware is determined by identifying the average grayscale value of the binarized square image of the anchor buckle. The specific steps for determining the dust accumulation status of the server hardware by identifying the average grayscale value of the binarized square image of the anchor buckle are as follows: Obtain the mean grayscale value of the binarized square image of the anchor buckle; If the average grayscale value of the binarized square image of the anchor buckle is less than the preset grayscale threshold, the dust accumulation status of the server hardware is determined to be high dust accumulation risk, and high dust accumulation risk information is output. If the average grayscale value of the binarized square image of the anchor buckle is greater than or equal to the preset grayscale threshold, the dust accumulation status of the server hardware is determined to be low dust accumulation risk, and no dust accumulation risk information is output; where the preset grayscale threshold is 255.

[0045] By identifying the grayscale mean of the binarized square image of the anchor buckle, the dust accumulation status of the server hardware can be determined. This allows for monitoring of the dust accumulation on the server hardware and prevents excessive dust accumulation from causing server malfunctions.

[0046] like Figure 2 As shown, in some other embodiments, a server hardware fault early warning system based on intelligent optimization algorithms is also provided, including: The data acquisition module is used to collect the server's operational data and to acquire hardware images of the server. The fault detection module is used to perform operational fault detection on the server based on the operational data information and obtain operational fault detection information; and to use computer vision recognition algorithms to identify hardware location defects and obtain physical fault detection information. The fault diagnosis module is used to determine whether the server has a fault risk based on the operational fault detection information and the physical fault detection information. If so, it outputs a fault warning message; otherwise, it outputs a normal operation message. The data acquisition module, fault detection module, and fault diagnosis module can be program modules or computers or computing devices with corresponding functions.

[0047] In some other embodiments, a storage medium is also provided, which stores a computer program or computer instructions. When the computer program or computer instructions are executed by the processor of a computer, the steps of the above-described server hardware fault early warning method based on intelligent optimization algorithms are implemented.

[0048] The storage medium can be an internal storage unit of any data processing device described in any of the foregoing embodiments, such as a hard disk or memory. The storage medium can also be an external storage device of any data processing device, such as a plug-in hard disk, smart memory card, SD card, flash memory card, etc., mounted on the device. Furthermore, the storage medium can include both internal storage units and external storage devices of any data processing device. The computer-readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0049] In other embodiments, a computer is also provided, including a memory and one or more processors, wherein executable code is stored in the memory, and when the one or more processors execute the executable code, the steps of the server hardware fault early warning method based on intelligent optimization algorithm in Embodiment 1 are implemented.

[0050] The memory can be an internal storage unit of any data processing device described in any of the foregoing embodiments, such as a hard disk or RAM. The memory can also be an external storage device of any data processing device, such as a plug-in hard disk, smart memory card, SD card, flash memory card, etc., mounted on the device. Furthermore, the memory can include both internal storage units and external storage devices of any data processing device. The memory is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0051] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the concept and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A server hardware fault early warning method based on intelligent optimization algorithms, characterized in that, Includes the following steps: Collect server operation data; Based on the operational data information, the server is subjected to operational fault detection to obtain operational fault detection information; Acquire hardware images of the server; Computer vision recognition algorithms are used to identify hardware location defects and obtain physical fault detection information; Based on the operational fault detection information and the physical fault detection information, determine whether the server has a fault risk. If so, output a fault warning message; otherwise, output a normal operation message. Collect server operational data, including the following steps: The CPU of the server was tested for floating-point operations and integer operations to obtain a CPU performance score. A memory testing tool was used to perform a full scan of the memory, and the memory test results were obtained. Simulate read and write operations on the disk to obtain disk throughput; Perform multi-threaded network transmission detection to detect the luminous power of the network transmission module; The operational data information includes the computing performance score, the memory detection result, the disk throughput, and the luminous power; Based on the operational data information, the server is subjected to operational fault detection to obtain operational fault detection information, including the following steps: The computational performance score is compared with a preset computational performance score. If the computational performance score is less than the preset computational performance score, a computational fault message is output; if the computational performance score is greater than or equal to the preset computational performance score, no message is output. If the memory test results include memory error information, then output memory status abnormal information; If the memory error information is not included in the memory test results, no information will be output. The disk throughput is compared with a preset throughput threshold. If the disk throughput is less than the preset throughput threshold, disk status abnormality information is output. If the disk throughput is greater than or equal to the preset throughput threshold, no information is output; The luminous power is compared with a preset luminous efficiency threshold. If the luminous power is less than the preset luminous efficiency threshold, a network communication error message is output. If the luminous power is greater than or equal to the preset luminous efficiency threshold, no message is output. The operational fault detection information includes the computational fault information, the memory status abnormal information, the disk status abnormal information, and the network communication abnormal information; Using computer vision recognition algorithms to identify hardware location defects and obtain physical fault detection information includes the following steps: The hardware image is preprocessed to obtain a preprocessed image; The position and size of each hardware component of the server in the preprocessed image are identified using computer vision recognition algorithms. Determine whether a physical fault has occurred based on the location and size of each piece of hardware. If so, output the physical fault detection information; otherwise, no information is output. The positional dimensions of each hardware component are calculated based on the distance between their binarized windows, including the following steps: Use the binarized window of one of the hardware components as the reference window; Construct a rectangular coordinate system using the sides and corners of the reference window; Anchor points are determined on all binarized windows except the reference window; Calculate the coordinates of the anchor points on each of the binarized windows in the Cartesian coordinate system to obtain the position and size of each hardware component; The binarized window of one of the hardware components is used as the reference window. The specific steps are as follows: randomly select a binarized window of one of the hardware components as the reference window. Determining whether a physical fault has occurred based on the location and dimensions of each piece of hardware includes the following steps: Based on the absolute value of the coordinates of the anchor points on each of the binarized windows in the rectangular coordinate system; The absolute values ​​of the coordinates of the anchor points on each of the binarized windows are compared with a preset relative position threshold. If the absolute value of the coordinates of any anchor point on the binarized window is less than or equal to the relative position threshold, it is determined that the position size of the hardware has not experienced a physical fault; if the absolute value of the coordinates of any anchor point on the binarized window is greater than the relative position threshold, it is determined that the position size of the hardware has experienced a physical fault.

2. The server hardware fault early warning method based on intelligent optimization algorithm according to claim 1, characterized in that, The hardware image is preprocessed to obtain a preprocessed image; this includes the following steps: The hardware image is then processed to remove the background. The hardware image with the background removed is then processed into grayscale to obtain the preprocessed image.

3. A server hardware fault early warning system based on intelligent optimization algorithms, characterized in that, include: The data acquisition module collects operational data from the server. Acquire hardware images of the server; The fault detection module is used to perform operational fault detection on the server based on the operational data information and obtain operational fault detection information; Computer vision recognition algorithms are used to identify hardware location defects and obtain physical fault detection information; The fault diagnosis module is used to determine whether the server has a fault risk based on the operational fault detection information and the physical fault detection information. If so, it outputs a fault warning message; if not, it outputs a normal operation message. Collect server operational data, including the following steps: The CPU of the server was tested for floating-point operations and integer operations to obtain a CPU performance score. A memory testing tool was used to perform a full scan of the memory, and the memory test results were obtained. Simulate read and write operations on the disk to obtain disk throughput; Perform multi-threaded network transmission detection to detect the luminous power of the network transmission module; The operational data information includes the computing performance score, the memory detection result, the disk throughput, and the luminous power; Based on the operational data information, the server is subjected to operational fault detection to obtain operational fault detection information, including the following steps: The computational performance score is compared with a preset computational performance score. If the computational performance score is less than the preset computational performance score, a computational fault message is output; if the computational performance score is greater than or equal to the preset computational performance score, no message is output. If the memory test results include memory error information, then output memory status abnormal information; If the memory error information is not included in the memory test results, no information will be output. The disk throughput is compared with a preset throughput threshold. If the disk throughput is less than the preset throughput threshold, disk status abnormality information is output. If the disk throughput is greater than or equal to the preset throughput threshold, no information is output; The luminous power is compared with a preset luminous efficiency threshold. If the luminous power is less than the preset luminous efficiency threshold, a network communication error message is output. If the luminous power is greater than or equal to the preset luminous efficiency threshold, no message is output. The operational fault detection information includes the computational fault information, the memory status abnormal information, the disk status abnormal information, and the network communication abnormal information; Using computer vision recognition algorithms to identify hardware location defects and obtain physical fault detection information includes the following steps: The hardware image is preprocessed to obtain a preprocessed image; The position and size of each hardware component of the server in the preprocessed image are identified using computer vision recognition algorithms. Determine whether a physical fault has occurred based on the location and size of each piece of hardware. If so, output the physical fault detection information; otherwise, no information is output. The positional dimensions of each hardware component are calculated based on the distance between their binarized windows, including the following steps: Use the binarized window of one of the hardware components as the reference window; Construct a rectangular coordinate system using the sides and corners of the reference window; Anchor points are determined on all binarized windows except the reference window; Calculate the coordinates of the anchor points on each of the binarized windows in the Cartesian coordinate system to obtain the position and size of each hardware component; The binarized window of one of the hardware components is used as the reference window. The specific steps are as follows: randomly select a binarized window of one of the hardware components as the reference window. Determining whether a physical fault has occurred based on the location and dimensions of each piece of hardware includes the following steps: Based on the absolute value of the coordinates of the anchor points on each of the binarized windows in the rectangular coordinate system; The absolute values ​​of the coordinates of the anchor points on each of the binarized windows are compared with a preset relative position threshold. If the absolute value of the coordinates of any anchor point on the binarized window is less than or equal to the relative position threshold, it is determined that the position size of the hardware has not experienced a physical fault; if the absolute value of the coordinates of any anchor point on the binarized window is greater than the relative position threshold, it is determined that the position size of the hardware has experienced a physical fault.

Images