Host system configuration checking method for multi-source data analysis

Abstract

The invention discloses a host system configuration checking method for multi-source data analysis, and the method comprises the following steps: 1, dynamically collecting host multi-source environment parameters through a multi-source sensing array, analyzing the host multi-source environment parameters, judging whether the host environment meets the checking environment requirements or not, and when the host environment parameters meet the environment requirements, executing the step 2; performing the next step; and 2, inserting the configuration checking equipment into a corresponding interface of the host, performing image acquisition of the configuration checking equipment at the moment, performing access state verification on an image, performing mechanical-electrical dual stability verification after the access state verification is passed, and performing the next step when the mechanical-electrical dual stability verification is seriously passed. Through multi-dimensional environment evaluation, flexible access state verification, mechanical-electrical dual stability verification and targeted system configuration verification, all-directional and precise verification of host system configuration is realized, and the comprehensiveness and effectiveness of verification are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of verification methods, in particular to a host system configuration verification method based on multi-source data analysis. BACKGROUND

[0002] The compliance and stability of host system configuration are the core to ensure its safe and efficient operation, which needs to be covered by a comprehensive verification mechanism covering environment adaptation, device access, system core configuration and other key links.

[0003] In the existing host system configuration verification process, there are problems such as single verification dimension and insufficient verification mechanism: environmental assessment often relies on single parameter judgment, which is difficult to dynamically capture potential risks in complex environments; verification of device access state lacks flexible alternative solutions; connection stability verification does not take into account the mechanical and electrical dual characteristics, which may miss connection hazards; system configuration verification does not take into account the differentiated characteristics of different types of storage devices, making it difficult to accurately identify storage abnormalities. Therefore, a host system configuration verification method based on multi-source data analysis is proposed. SUMMARY

[0004] In view of the defects in the prior art, the present application provides a host system configuration verification method based on multi-source data analysis, comprising the following steps:

[0005] Step one: dynamically collect host multi-source environmental parameters through a multi-source sensor array, analyze the host multi-source environmental parameters, and determine whether the host environment meets the verification environmental requirements. When it meets the environmental requirements, proceed to the next step.

[0006] Step two: insert the configuration verification device into the corresponding interface of the host, at this time, image collection of the configuration verification device is performed, the access state verification is performed on the image, and after the access state verification is passed, the mechanical-electrical dual stability verification is performed, and after the mechanical-electrical dual stability verification is passed, the next step is performed.

[0007] Step three: perform host system configuration verification, and the system configuration verification content includes operating system verification, account content verification, storage system verification and hardware information verification. When any one of them is abnormal, a prompt information is generated.

[0008] Step four: send the prompt information generated in the verification process of step three to the corresponding receiving terminal.

[0009] Further, the specific process of analyzing the host environmental parameters to determine whether the host environment meets the verification environmental requirements is as follows:

[0010] The multi-source sensor array is used to collect environmental temperature information, environmental vibration information and environmental electromagnetic information.

[0011] The environmental temperature information includes a real-time temperature and a temperature gradient vector, the environmental vibration information includes a real-time vibration force and a vibration wavelet packet energy entropy, and the environmental electromagnetic information includes a real-time electromagnetic intensity and an electromagnetic power spectrum density;

[0012] When any one of the real-time vibration force, the real-time electromagnetic intensity and the real-time electromagnetic intensity exceeds a corresponding item threshold value, it indicates that the host environment does not meet the verification environment requirement.

[0013] The vibration wavelet packet energy entropy, the temperature gradient vector and the electromagnetic power spectrum density are input into an environmental risk prediction model, and a dynamic risk coefficient Kenv is output, and when the Kenv is within a preset safety threshold value, it indicates that the verification environment requirement is met.

[0014] Further, the specific construction process of the environmental risk prediction model is as follows:

[0015]

[0016] wherein, σ is a Sigmoid function, is a temperature gradient vector, E vib is a vibration wavelet packet energy entropy, P EM is an electromagnetic power spectrum density, TCN is a time convolution network, wi is a weight coefficient of a multi-source parameter, i is an index variable of a summation operation, used to traverse each sample or parameter item participating in calculation, taking a value from 1 to n, n is a total sample quantity or a total number of parameter items participating in model calculation, that is, an upper limit of a summation operation range.

[0017] Further, the specific process of the step two of verifying the access state of the image is as follows:

[0018] Baseline point preset: preset two baseline points a1 and a2 on the configuration verification device, and set two baseline points b1 and b2 on the host, wherein a1 and b1 are located on the same side, and a2 and b2 are located on the other side.

[0019] Real-time line acquisition: extract the image of the configuration verification device, extract the baseline points a1 and a2 from the image, and extract the baseline points b1 and b2 from the host; then perform a line connection operation, connect a1 with b1 and b2 respectively to obtain real-time lines L1 and L2; connect a2 with b1 and b2 respectively to obtain real-time lines L3 and L4.

[0020] Baseline line setting: preset baseline lines F1, F2, F3 and F4, wherein F1 corresponds to the real-time line L1, F2 corresponds to the real-time line L2, F3 corresponds to the real-time line L3, and F4 corresponds to the real-time line L4.

[0021] Verification judgment: calculate the length difference of the baseline and the corresponding real-time line, that is, the length difference of F1 and L1, the length difference of F2 and L2, the length difference of F3 and L3, and the length difference of F4 and L4. If any of the above length differences exceeds the preset range, it is determined that the access state verification fails. If all four length differences are within the preset range, it is determined that the access state verification is passed.

[0022] Further, when the reference points a1 and a2 cannot be collected in the image of the configuration verification device, the following process verification is performed:

[0023] The physical size parameters (such as length, width) and structural features (such as interface position, screw hole distribution) of the configuration verification device are pre-stored;

[0024] When the reference points a1 and a2 cannot be collected, the edge midpoint of the device or the pre-device identification point, such as the device surface two-dimensional code or the groove, is automatically identified as the temporary reference points a1' (on the same side as b1) and a2' (on the same side as b2);

[0025] The temporary real-time lines L1'~L4' are generated according to the original logic, and compared with the baseline F1~F4. Since the accuracy of the substitute points is slightly lower, the allowable error range is expanded by 15%;

[0026] If the substitute points cannot be identified, the complete contour of the configuration verification device is extracted from the image of the configuration verification device, and the shortest distances d1, d2 from the contour edge to the reference points b1, b2 are calculated;

[0027] Compare with the preset standard distances d10, d20. If |d1-d10|≤0.5cm and |d2-d20|≤0.5cm, and the center line of the contour and the host computer have an angle within the preset range (such as ±3°), it is determined that the verification is passed;

[0028] When the standard points b1 and b2 cannot be collected, the position parameters of the host side interface (such as USB port, HDMI port) are pre-stored as substitute features of the reference points b1 and b2;

[0029] When the reference points b1 and b2 cannot be collected, the center position of the host interface is identified as the temporary reference points b1' (on the same side as a1) and b2' (on the same side as a2);

[0030] Then generate real-time lines L1'~L4' and compare with baseline F1~F4.

[0031] Further, the specific process of the mechanical-electrical dual stability verification is as follows:

[0032] Mechanical stability verification, specifically:

[0033] Pressure sensor matrix configuration: a distributed flexible pressure sensor array is adopted, containing Q×K piezoresistive sensor units (Q≥3, K≥3), which are uniformly distributed on the interface contact surface, and the resolution of each sensor unit is 0.1N and the sampling frequency is 100Hz;

[0034] Contact area calculation: the output signal of the distributed flexible pressure sensor array is processed by threshold segmentation method to obtain the effective contact area, specifically:

[0035]

[0036] wherein Sik is the binary contact state (Sik=1 when the pressure value is greater than or equal to 5N, otherwise Sik=0), and ΔA is the area of a single sensor unit;

[0037] Stability coefficient calculation: the mechanical stability coefficient is calculated based on the contact area and the number of insertions, specifically:

[0038] When η is greater than or equal to the corresponding threshold value, it is determined to pass the verification;

[0039] A0 is the initial design contact area, which is obtained by calibration with a standard test piece; N is the number of recorded insertions, which is stored in the host non-volatile memory; Nmax is the maximum allowable number of insertions determined by the interface material fatigue life test;

[0040] The specific process of electrical stability verification is as follows:

[0041] Test signal injection: a composite signal with a frequency set F={f1, f2,..., fk} is sent to the interface through a test signal generator, wherein f1=100kHz (low frequency component), f2=1MHz (medium frequency component), and f3=10MHz (high frequency component), and the amplitude of each frequency signal is 1Vpp;

[0042] Signal acquisition and analysis: the response signal is collected at the interface receiving end through a spectrum analyzer, and the attenuation rate of each frequency component is calculated, the specific process is as follows:

[0043]

[0044] wherein A in (f) is the amplitude of the injected signal, A out (f) is the amplitude of the received signal;

[0045] When the attenuation rate ρf of all frequency components is less than or equal to the corresponding threshold value, it is determined that the electrical connection of the interface is reliable; if there is a frequency component with ρf greater than the corresponding threshold value, further principal component analysis is performed to locate the fault frequency band, and a targeted maintenance suggestion is generated;

[0046] When the mechanical stability verification and the electrical stability verification are verified to pass at the same time, it is determined that the mechanical-electrical dual stability verification passes.

[0047] Further, the content verified by the operating system includes verifying the host identification and the asset information.

[0048] The specific verification process of the host identification and the asset information includes verifying whether the host name, the IP address, and the MAC address are consistent with the asset list, and whether there is an unregistered host. When there is an inconsistent or unregistered host, prompt information is generated.

[0049] Whether the business, the person in charge, or the purpose is clear, and whether there is an unauthorized change of purpose (such as a production environment host being used for testing). When there is an unclear business, person in charge, or purpose, or an unauthorized change of purpose, prompt information is generated.

[0050] Whether the system installation time and version information meet the standard requirements. When the system installation time or version information does not meet the standard requirements, prompt information is generated.

[0051] The specific process of the account content verification includes whether there are long-term inactive accounts, redundant accounts (such as accounts of employees who have left the company not being deleted), and shared accounts. When there are long-term inactive accounts or redundant accounts (such as accounts of employees who have left the company not being deleted) or shared accounts, prompt information is generated.

[0052] The content of the storage system verification includes whether the disk space usage is too high (such as more than 85% which may affect system operation) and whether there is abnormal occupation (such as large log files, unknown large files). When there is a high disk space usage or abnormal occupation, prompt information is generated.

[0053] Whether the mounted external storage (such as USB, NFS shared) has passed security authentication, and whether the read and write permissions are limited. When the mounted external storage has not passed security authentication or the read and write permissions are limited, prompt information is generated.

[0054] Abnormal analysis is performed on disk read and write to determine whether there is an abnormality. When there is an abnormality in disk read and write, prompt information is generated.

[0055] Further, in the process of abnormal analysis of disk read and write, a differentiated detection strategy is adopted according to the disk type. The specific process is as follows:

[0056] When the detected disk is an SSD, the read and write speed sequence V = {v1, v2,..., vn} and the temperature sequence T = {t1, t1,..., tn} are synchronously collected with a sampling period τ, and the dynamic performance attenuation coefficient δ is calculated. SSD

[0057] ​When the detected disk is HDD, the read-write speed sequence VHDD = {v1', v2',..., vm'} is synchronously collected with the same sampling period τ, and the noise spectrum NHDD(f) = {n1(f), n2(f),..., nm(f)} is collected through the pre-set microphone array, and the mechanical health index η is calculated HDD ;

[0058] If δ SSD > Γth or η HDD < Hth, it is determined that the disk read-write is abnormal and a prompt information is generated.

[0059] Wherein, the threshold value Γth is the SSD abnormality determination threshold value, and the threshold value Hth is the HDD abnormality determination threshold value, which are determined by a logistic regression model trained by historical fault data.

[0060] Further, the acquisition process of the dynamic performance attenuation coefficient δ SSD is as follows:

[0061] The calculation is as follows:

[0062] First, a temperature-speed correlation matrix is constructed, specifically:

[0063]

[0064] k ≥ 10, ΔT i = t i - t i-1 , ΔV i = |v i - v base |;

[0065] Wherein, k is the number of samples collected continuously in the current sampling period (k ≥ 10 is required to ensure calculation stability), and the last k continuous samples are cut from the synchronously collected temperature sequence TSSD and read-write speed sequence V; ti is the temperature value (unit: ℃) at the i-th sampling time in TSSD, which is collected in real time by the built-in temperature sensor of the SSD; vi is the read-write speed value (unit: MB / s) at the i-th sampling time in VSSD, which is read in real time through the disk controller interface; vbase is the SSD reference speed, which is measured by the manufacturer before leaving the factory under the standard environment of 25℃ (humidity 50% ± 5%, no vibration) for 1 hour of continuous read-write test, stored in the calibration parameter area of the disk firmware, and can be read through the API provided by the manufacturer;

[0066] The temperature change vector M T = [ΔT1, ΔT2,..., ΔT k ] T and the speed attenuation vector M V = [ΔV1, ΔV2,..., ΔVk ] T ;

[0067] Then the attenuation coefficient is calculated, the specific process is:

[0068]

[0069] Wherein, is the gradient of the temperature change vector, which is obtained by performing a first-order difference operation on MT, that is ||·||2 is the L2 norm, which is calculated according to the square root of the sum of the squares of the vector elements; cov(M T , M V ) is the covariance of two vectors, which is calculated by using the statistical standard formula, that is and are the mean values of M T and M V , respectively; var(M V ) is the variance of the velocity attenuation vector, which is calculated by using the statistical standard formula ∈=10 -5 , is a preset constant, which is set by the system in advance (fixed in the monitoring program), to avoid the calculation anomaly of the denominator being zero; t max is the highest temperature in the current sampling period, which is obtained by traversing the temperature values t1, t2...tk of the k samples in the TSSD to obtain the maximum value; β(t max )=1+0.05(t max -70) is a temperature penalty term, which is calculated in real time based on t max (when t max ≤70℃, β(t max )≥1);

[0070] Further, the mechanical health index ηHDD is calculated as follows:

[0071] Perform MFCC transformation on each nj(f) (j=1, 2,..., m) in the collected noise spectrum N HDD (f)={n1(f), n2(f),..., nm(f)}: use a 40-channel mel filter bank, the frequency range is 500Hz-15kHz, and the first 12-dimensional coefficients are extracted

[0072] Wherein, nj(f) is the noise spectrum (unit: dB / Hz) at the jth sampling time in NHDD(f), which is obtained by converting the original audio signal collected by the microphone array (sampling rate 44.1kHz, 16-bit quantization) deployed on the surface of the HDD shell through fast Fourier transform (FFT);

[0073] 40-channel Mel filter bank parameters (center frequency, bandwidth) are preset by the system (based on HDD mechanical noise characteristics optimization, solidified in the audio processing module), and the frequency range of 500Hz-15kHz is the main distribution interval of HDD mechanical failure noise;

[0074] The first 12-dimensional feature coefficients after MFCC transformation are extracted through the standard MFCC interface of the open-source audio processing library (such as Librosa);

[0075] The mean value of m sampling points is taken to construct a feature vector Among them

[0076] Among them: m is the total number of samples in the current sampling period, which is consistent with the length of V HDD , that is, m is equal to the number of sampling times in the sampling period τ;

[0077] The mean value of the i-th MFCC coefficient is calculated by taking the mean value of m sampling points The arithmetic mean value is calculated;

[0078] The health index is calculated:

[0079] C0 is the baseline feature vector of the HDD, which is established by taking the mean value of 100 repeated samplings of the same type of HDD by the manufacturer before leaving the factory in a standard silent environment (background noise ≤ 30dB) and stored in the baseline database of the monitoring system;

[0080] ||C-C0||2 is the L2 norm of the two vectors, which is calculated by taking the square root of the sum of the square differences of the corresponding elements;

[0081] γ=0.1 is the speed fluctuation weight coefficient, which is determined by historical data statistics (based on the speed fluctuation characteristics before failure of 1000 same type HDDs, solidified in the monitoring algorithm);

[0082] The absolute value of the read-write speed change rate is calculated by performing first-order difference calculation on V HDD , and then taking the absolute value, that is, Unit: MB / (s·s), where v′ j is the read-write speed at the j-th sampling time in V HDD , and τ is the sampling period.

[0083] The beneficial effects of the present application are:

[0084] By collecting multi-dimensional parameters such as environmental temperature, vibration, and electromagnetic through a multi-source sensing array, not only the basic environmental safety is judged through real-time value threshold, but also dynamic risk coefficients are output by an environmental risk prediction model combined with a time convolution network, realizing dynamic and comprehensive evaluation of environmental safety and avoiding the limitation of single parameter judgment; the access state of the device is verified by comparing the preset reference point connection, and when the reference point cannot be collected, a temporary reference point can be automatically identified or a substitute solution such as contour analysis is used for verification, improving the adaptability and accuracy of the access state verification and ensuring the compliance of the device access; mechanical and electrical dual stability verification is adopted, the stability coefficient is calculated by combining the contact area and the number of plug-in times on the mechanical level, and the interface reliability is analyzed by the multi-frequency signal attenuation rate on the electrical level, so that the mechanical stability and electrical transmission reliability of the interface connection are ensured, and the verification error caused by connection problems is reduced; multi-dimensional verification of the operating system, account, and storage system is covered, and differentiated detection strategies are adopted for SSD and HDD in the storage system to realize accurate identification of abnormalities of different types of disks and improve the pertinence and effectiveness of the verification; through multi-source data fusion, multi-link verification, and differentiated strategies, potential risks of the host system can be more comprehensively found, omissions are reduced, the accuracy and reliability of the configuration verification are improved, and the security protection capability of the host system is enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0085] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the specific embodiments or the prior art description will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual scale.

[0086] Figure 1 The overall flowchart of the present application. DETAILED DESCRIPTION

[0087] The embodiments of the technical solutions of the present application will be described in detail below with reference to the drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present application, and therefore only serve as examples, but cannot limit the protection scope of the present application.

[0088] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in the present application should be understood as the usual meanings understood by the skilled in the art to which the present application belongs.

[0089] As shown in Figure 1 A multi-source data analysis host system configuration verification method, comprising the following steps:

[0090] Step one: dynamically collect host multi-source environmental parameters through a multi-source sensing array, analyze the host multi-source environmental parameters, determine whether the host environment meets the verification environment requirements, and when it meets the environment requirements, proceed to the next step;

[0091] Step two: insert the configuration verification device into the corresponding interface of the host, at this time, image acquisition of the configuration verification device is performed, access state verification is performed on the image, and when the access state verification is passed, mechanical-electrical dual stability verification is performed, and when the mechanical-electrical dual stability verification is passed, the next step is performed;

[0092] Step three: perform host system configuration verification, and the system configuration verification content includes operating system verification, account content verification, storage system verification and hardware information verification, and when any one is abnormal, a prompt information is generated;

[0093] Step four: send the prompt information generated in the step three verification process to the corresponding receiving terminal.

[0094] The specific process of analyzing the host environmental parameters and determining whether the host environment meets the verification environment requirements is as follows:

[0095] The multi-source sensing array is used to collect environmental temperature information, environmental vibration information and environmental electromagnetic information;

[0096] The environmental temperature information includes real-time temperature and temperature gradient vector, the environmental vibration information includes real-time vibration force and vibration wavelet packet energy entropy, and the environmental electromagnetic information includes real-time electromagnetic intensity and electromagnetic power spectral density;

[0097] When any one of the real-time vibration force, the real-time electromagnetic intensity and the real-time electromagnetic intensity exceeds the corresponding item threshold value, it means that the host environment does not meet the verification environment requirements;

[0098] The vibration wavelet packet energy entropy, the temperature gradient vector and the electromagnetic power spectral density are input into an environmental risk prediction model, and a dynamic risk coefficient Kenv is output, and when Kenv is within a preset safety threshold, it means that it meets the verification environment requirements;

[0099] The three core parameters of environmental temperature, vibration and electromagnetic are collected by a multi-source sensing array, covering the key environmental factors affecting the stable operation of the host, avoiding the limitations of single environmental parameter monitoring, realizing multi-dimensional and stereoscopic evaluation of the host environment, and adopting a dual judgment mechanism of basic threshold screening and dynamic risk prediction: on the one hand, the threshold of real-time vibration force and real-time electromagnetic intensity and other parameters can directly and quickly identify the environment that does not meet the requirements, ensuring basic safety; on the other hand, the vibration wavelet packet energy entropy and temperature gradient vector and other parameters with more dynamic characteristics are input into the environmental risk prediction model, and the dynamic risk coefficient Kenv is output to realize fine and dynamic evaluation of the potential risk of the environment, improve the accuracy of environmental judgment, and ensure that only in the environment that meets the requirements can the subsequent configuration verification be carried out, avoiding the distortion of verification data or damage to equipment caused by environmental interference, and ensuring the effectiveness and safety of the entire verification process.

[0100] The specific construction process of the environmental risk prediction model is as follows:

[0101]

[0102] wherein σ is a Sigmoid function, is a temperature gradient vector, E vib is a vibration wavelet packet energy entropy, P EM is an electromagnetic power spectral density, TCN is a time convolution network, wi is a weight coefficient of the multi-source parameters, and i is an index variable of summation operation, used to traverse each sample or parameter item participating in calculation, taking values from 1 to n, n is the total number of samples participating in model calculation or the total number of parameter items, i.e. the upper limit of the range of summation operation;

[0103] The model integrates three types of core dynamic environmental parameters, i.e. temperature gradient vector, vibration wavelet packet energy entropy and electromagnetic power spectral density, rather than single static data. By processing the time sequence characteristics of these parameters through the time convolution network, the dynamic change law of environmental parameters over time can be captured, avoiding the limitations of single moment data, making the risk assessment more in line with the dynamic characteristics of the actual environment, and mapping the comprehensive analysis results of multi-source parameters to the dynamic risk coefficient Kenv in the 0-1 interval through the Sigmoid function, realizing the quantitative expression of environmental risk. Compared with qualitative description, the quantitative Kenv is more convenient for setting a clear safety threshold, so that the operator can intuitively judge whether the environment meets the verification requirements, reducing subjective judgment errors; the weight coefficient of multi-source parameters is introduced, which can adjust the influence weight of each parameter according to the actual application scene. For example, in the electromagnetically sensitive machine room, the weight of electromagnetic power spectral density (PEM) can be increased, so that the model focuses more on key risk factors and improves the pertinence;

[0104] Suppose in a data center host environment verification:

[0105] Temperature gradient vector The temperature shows a rapid rise from 25℃ to 30℃ in a short time, and the gradient changes significantly;

[0106] Vibration wavelet packet energy entropy (E vib ) shows that the display device vibrates slightly, but the fluctuation amplitude is small;

[0107] Electromagnetic power spectral density (P EM ) shows that there are intermittent electromagnetic signals around, but the intensity does not exceed the basic threshold.

[0108] If only a single parameter is used for judgment, it may be misjudged that the environmental risk is too high due to the significant change in temperature gradient. However, through the model set in this case:

[0109] TCN processes the time series of the above parameters and identifies that the change in temperature gradient is a short-term fluctuation (not a sustained increase), and the influence of vibration and electromagnetic signals is weak;

[0110] The weight coefficient (wi) is adjusted according to the data center scenario, and the temperature weight is slightly higher, but it does not amplify short-term fluctuations excessively;

[0111] Finally, the dynamic risk coefficient Kenv=0.3 is output through the Sigmoid function (the preset safety threshold is 0-0.5), and it is determined that the environment meets the requirements.

[0112] The specific process of the image access state verification in step two is as follows:

[0113] Baseline point preset: preset two baseline points a1 and a2 on the configuration verification device, and set two baseline points b1 and b2 on the host, wherein a1 and b1 are located on the same side, and a2 and b2 are located on the other side.

[0114] Real-time line acquisition: extract the image of the configuration verification device, extract the baseline points a1 and a2 from the image, and extract the baseline points b1 and b2 from the host; then perform a line operation to connect a1 with b1 and b2 respectively, to obtain real-time lines L1 and L2; connect a2 with b1 and b2 respectively to obtain real-time lines L3 and L4.

[0115] Baseline line setting: preset baseline lines F1, F2, F3, and F4, wherein F1 corresponds to real-time line L1, F2 corresponds to real-time line L2, F3 corresponds to real-time line L3, and F4 corresponds to real-time line L4.

[0116] Verification judgment: Calculate the length difference between the reference line and the corresponding real-time line, i.e. the length difference between F1 and L1, the length difference between F2 and L2, the length difference between F3 and L3, and the length difference between F4 and L4. If any of the above length differences exceeds the preset range, the access state verification fails. If all four length differences are within the preset range, the access state verification is successful.

[0117] By presetting the reference points of the device and the host, four corresponding lines are constructed. The length difference is used as the basis for judgment. Multi-dimensional comparison avoids the one-sidedness of single reference point verification, and can more accurately determine whether the configuration verification device is correctly and stably connected to the host interface. This reduces the distortion of verification data caused by access deviation. Based on the length difference of the reference point line extracted from the image, the access state is verified through specific numerical values rather than subjective observation. The standard is clear and has strong operability, reduces human judgment errors, and improves the consistency of verification results.

[0118] Suppose in the configuration verification of a host USB interface:

[0119] The reference points a1 (left edge midpoint) and a2 (right edge midpoint) on the preset configuration verification device, and the reference points b1 (2 cm left of the USB interface) and b2 (2 cm right of the USB interface) on the host.

[0120] The preset reference lines F1 (standard distance between a1 and b1) is 5 cm, F2 (between a1 and b2) is 9 cm, F3 (between a2 and b1) is 9 cm, and F4 (between a2 and b2) is 5 cm. The preset length difference allowed range is ±0.2 cm.

[0121] After the device is inserted, the image is captured, the real-time reference points are extracted and connected, and L1=5.1 cm, L2=8.9 cm, L3=9.1 cm, and L4=4.9 cm are obtained.

[0122] Calculate the length difference: F1 and L1 differ by 0.1 cm, F2 and L2 differ by 0.1 cm, F3 and L3 differ by 0.1 cm, and F4 and L4 differ by 0.1 cm, all within the allowed range. The access state verification is successful, ensuring that the device is correctly connected and providing a reliable basis for subsequent verification.

[0123] If L1=5.3 cm (0.3 cm difference from F1, exceeding the range) in a certain verification, the access state verification fails directly, avoiding unstable electrical connection or data transmission errors caused by the device being inserted at an angle.

[0124] When the reference points a1 and a2 cannot be captured in the image of the configuration verification device, the following process is verified:

[0125] The physical dimensions (such as length and width) and structural features (such as interface location and screw hole distribution) of the configuration verification equipment are pre-stored;

[0126] When reference points a1 and a2 cannot be collected, the device edge midpoint or a preset backup marker point (such as a QR code or groove on the device surface) is automatically identified as a temporary reference point a1' (on the same side as b1) and a2' (on the same side as b2).

[0127] Temporary real-time lines L1' to L4' are generated according to the original logic and compared with the baseline lines F1 to F4. The allowable error range is increased by 15% (because the accuracy of the replacement points is slightly lower).

[0128] If a substitute point cannot be identified, extract the complete outline of the configuration verification equipment from the image of the configuration verification equipment, and calculate the shortest distances d1 and d2 from the edge of the outline to the reference points b1 and b2.

[0129] If the distances are compared with the preset standard distances d10 and d20, and |d1-d10|≤0.5cm and |d2-d20|≤0.5cm, and the angle between the center line of the contour and the host is within the preset range (e.g., ±3°), then the verification is considered successful.

[0130] When standard points b1 and b2 cannot be acquired, the position parameters of the host-side interface (such as USB port, HDMI port) are pre-stored as alternative features for reference points b1 and b2.

[0131] When reference points b1 and b2 cannot be acquired, the center position of the host interface is identified as temporary reference points b1' (on the same side as a1) and b2' (on the same side as a2).

[0132] After generating real-time lines L1' to L4', they are compared with baseline lines F1 to F4.

[0133] The above process addresses scenarios where reference points (a1, a2 or b1, b2) cannot be collected. By using pre-set alternative solutions (identifying temporary reference points, contour analysis, interface feature substitution, etc.), it avoids verification interruptions due to missing reference points and ensures that access status verification can still be completed in complex environments (such as equipment surface wear and reference point obstruction), thus enhancing the anti-interference capability of the verification process.

[0134] The alternative solution has clear operational logic and quantitative standards (such as temporary reference point generation rules, contour distance difference ≤ 0.5cm, included angle ± 3°, etc.), avoiding subjective judgment without standards, and ensuring that even when reference points are missing, the verification results can still reflect the actual access status of the device and maintain the rigor of the verification.

[0135] For example, configure the verification device to be inserted into the host's USB port:

[0136] If the reference points a1 and a2 on the surface of the verification equipment cannot be acquired from the image due to wear and tear from long-term use, then real-time lines L1 to L4 cannot be generated:

[0137] The system calls an alternative solution: The physical dimensions (length 10cm, width 3cm) and structural features (two symmetrical screw holes on the interface edge) of the verification device are pre-stored, and the midpoints of the upper and lower edges of the device are automatically identified as temporary reference points a1' (on the same side as b1) and a2' (on the same side as b2);

[0138] Temporary real-time lines L1' to L4' are generated according to the original logic and compared with the baseline lines F1 to F4 (the allowable error range is increased by 15% compared with the original range, the original range is ±0.2cm, and the increased range is ±0.23cm).

[0139] The calculated length differences between L1' and F1 are 0.18cm, L2' and F2 are 0.21cm, L3' and F3 are 0.20cm, and L4' and F4 are 0.19cm, all within the expanded allowable range. Therefore, the access status verification is deemed successful.

[0140] If the temporary reference point still cannot be identified, the system extracts the complete outline of the verification device, calculates the shortest distances from the outline edge to the host reference points b1 and b2, d1 = 5.2cm and d2 = 5.3cm, and compares them with the preset standard distances d10 = 5.0cm and d20 = 5.0cm. If |d1-d10| = 0.2cm ≤ 0.5cm and |d2-d20| = 0.3cm ≤ 0.5cm, and the angle between the outline centerline and the host is +2° (within the range of ±3°), the verification is still considered successful, ensuring that the access status verification is uninterrupted and the results are reliable.

[0141] The specific process of the mechanical-electric dual stability verification is as follows:

[0142] Mechanical stability verification, specifically:

[0143] Pressure sensor matrix configuration: A distributed flexible pressure sensor array is adopted, containing Q×K piezoresistive sensor units (Q≥3, K≥3), which are evenly distributed on the interface contact surface. Each sensor unit has a resolution of 0.1N and a sampling frequency of 100Hz.

[0144] Contact area calculation: The effective contact area is obtained by processing the output signal of the distributed flexible pressure sensor array using a threshold segmentation method, specifically as follows:

[0145]

[0146] Where Sik represents the binary contact state (Sik = 1 when the pressure value is ≥ 5N, otherwise Sik = 0), and ΔA represents the area of ​​a single sensor unit;

[0147] Stability coefficient calculation: mechanical stability coefficient is calculated based on contact area and plug-in times, specifically:

[0148] When η is greater than or equal to the corresponding threshold, it is determined to pass the verification;

[0149] A0 is the initial design contact area, which is obtained by standard test piece calibration; N is the recorded plug-in times, stored in the host non-volatile memory; Nmax is the maximum allowed plug-in times determined by interface material fatigue life test;

[0150] The specific process of electrical stability verification is:

[0151] Test signal injection: a composite signal of frequency set F = {f1, f2,..., fk} is sent to the interface through a test signal generator, where f1 = 100 kHz (low frequency component), f2 = 1 MHz (medium frequency component), f3 = 10 MHz (high frequency component), and the amplitude of each frequency signal is 1 Vpp;

[0152] Signal acquisition and analysis: the response signal is collected at the interface receiving end through a spectrum analyzer, and the attenuation rate of each frequency component is calculated, the specific process is:

[0153]

[0154] Where A in (f) is the amplitude of the injected signal, A out (f) is the amplitude of the received signal;

[0155] When the attenuation rate of all frequency components is less than or equal to the corresponding threshold, it is determined that the electrical connection of the interface is reliable; if there is a frequency component with attenuation rate greater than the corresponding threshold, further principal component analysis is used to locate the fault frequency band, and a targeted maintenance suggestion is generated;

[0156] When the mechanical stability verification and the electrical stability verification pass at the same time, it is determined that the mechanical-electrical dual stability verification passes;

[0157] Through mechanical-electrical dual verification, the stability of the physical connection of the interface is evaluated from the mechanical level, and the reliability of signal transmission is evaluated from the electrical level, avoiding the limitations of single-dimensional verification, realizing all-round control of the connection state of the interface, and the mechanical stability coefficient is combined with the effective contact area and the number of plug-in times to ensure the physical stability of the current connection and consider the influence of equipment aging on the connection, improve the dynamic and predictive nature of verification, through the attenuation rate analysis of low-frequency, medium-frequency and high-frequency composite signals, the electrical connection quality under different frequencies can be comprehensively reflected; if there is an anomaly, the fault frequency band can be located through principal component analysis, and targeted maintenance suggestions are generated to improve problem solving efficiency, only when the mechanical and electrical verification passes at the same time, it is determined that the connection is stable, the strict determination logic reduces the misjudgment caused by single-dimensional standard but actual hidden danger (such as mechanical loosening leading to unstable electrical signal), and ensures the actual reliability of the interface connection;

[0158] In the scene of configuring the verification device to insert the host USB interface, after completing the access state verification, the re-stability verification is performed:

[0159] Mechanical stability verification:

[0160] The interface contact surface is deployed with a 3×3 (Q=3, K=3) distributed flexible pressure sensor array (resolution 0.1N, sampling frequency 100Hz), the signal is processed by threshold segmentation method (Sik=1 when pressure≥5N), and the effective contact area is calculated (ΔA is the area of a single sensor 0.2cm 2 ), Ac=8.5cm 2 ;

[0161] The initial design contact area A0=10cm 2 is known, the number of plug-in times of this interface N=200 times, the maximum allowable plug-in times Nmax=1000 times, the stability coefficient η=(8.5 / 10)×(1-200 / 1000)=0.85×0.8=0.68 is calculated by substituting the formula, if the preset threshold is 0.6, the mechanical verification passes.

[0162] Electrical stability verification:

[0163] The composite signal of frequency set F={100kHz, 1MHz, 10MHz} (amplitude is 1Vpp) is injected into the interface, the response signal is collected at the receiving end, and the attenuation rate of each frequency is calculated:

[0164] 100kHz: Ain=1Vpp, Aout=0.95Vpp, ρf=(1-0.95) / 1×100%=5%;

[0165] 1MHz: Ain=1Vpp, Aout=0.92Vpp, ρf=8%;

[0166] 10MHz: Ain = 1Vpp, Aout = 0.88Vpp, pf = 12%;

[0167] If the preset frequency threshold is 15%, the three frequency attenuation rates are all less than or equal to the threshold, and the electrical verification is passed.

[0168] Final determination: both mechanical and electrical verifications are passed, it is determined that the mechanical-electrical dual stability verification of the USB interface is passed, the physical connection is stable and the signal transmission is reliable after the device is connected, and a stable hardware foundation is provided for subsequent configuration checking.

[0169] If the 10MHz signal attenuation rate pf = 20% (exceeding the threshold of 15%) in a certain verification, the system locates the high-frequency fault through principal component analysis, generates a maintenance suggestion of "checking whether the high-frequency transmission contact inside the interface is oxidized", and improves the problem troubleshooting efficiency.

[0170] The content of the operating system verification includes verifying the host identification and asset information;

[0171] The specific verification process of the host identification and asset information is to verify whether the host name, IP address and MAC address are consistent with the asset list, whether there is an unregistered host, and generate a prompt information when there is an inconsistent or unregistered host;

[0172] Whether the business, responsible person or purpose is clear, whether there is unauthorized change of purpose, such as a production environment host being used for testing, and whether there is an unclear business, responsible person or purpose or unauthorized change of purpose, and generate a prompt information;

[0173] Whether the system installation time and version information meet the standard requirements, and generate a prompt information when the system installation time or version information does not meet the standard requirements;

[0174] The specific process of the account content verification is whether there is a long-term inactive account, a redundant account (such as an account of a former employee not deleted) and a shared account, and a prompt information is generated when there is a long-term inactive account or a redundant account (such as an account of a former employee not deleted) or a shared account;

[0175] The content of the storage system verification includes whether the disk space usage rate is too high (such as more than 85% which may affect system operation) and whether there is abnormal occupation (such as too large log files, unknown large files), and a prompt information is generated when there is a high disk space usage rate or abnormal occupation;

[0176] Whether the mounted external storage (such as U disk, NFS shared) is security certified, and whether the read-write permission is limited, and a prompt information is generated when the mounted external storage is not security certified or the read-write permission is limited;

[0177] Anomaly analysis is performed on the disk read and write, and it is judged whether there is an anomaly. When the disk read and write is abnormal, prompt information is generated.

[0178] In the process of analyzing the anomaly of the disk read and write, a differentiated detection strategy is adopted according to the disk type, and the specific process is as follows:

[0179] When the detected disk is SSD, the read and write speed sequence V={v1, v2,..., vn} and the temperature sequence T={t1, t1,..., tn} are synchronously collected with a sampling period τ, and the dynamic performance attenuation coefficient δ is calculated SSD ;

[0180] When the detected disk is HDD, the read and write speed sequence V HDD ={v1', v2',..., vm'} is synchronously collected with the same sampling period τ, and the noise spectrum NHDD(f)={n1(f), n2(f),..., nm(f)} is collected through a pre-set microphone array, and the mechanical health index η is calculated HDD ;

[0181] If δ SSD >Γth or η HDD <Hth, it is determined that the disk read and write is abnormal and prompt information is generated;

[0182] Wherein, the threshold value Γth is the SSD abnormality determination threshold value, and Hth is the HDD abnormality determination threshold value, which are determined by a logistic regression model trained by historical fault data;

[0183] It covers three key areas of operating system, account, storage system, including basic attributes such as host identification and asset information, and core risk points such as account security, storage resource usage and external storage access, avoiding the omission of single-dimensional verification, realizing the all-round control of host system configuration, discovering unregistered hosts or inconsistent devices in time through verifying the consistency of host name, IP address, MAC address and asset list, preventing unauthorized device access; verifying information such as business, responsible person, purpose and system version can ensure that the host usage conforms to the specification, and protect system compliance and traceability, screening risk accounts such as long-term non-login, redundant and shared accounts can reduce the risk of illegal use of accounts, avoid unauthorized access or data leakage due to account management oversight, improve the security of system identity authentication and permission management, and early warning of insufficient storage space through monitoring disk space usage and abnormal occupation can avoid affecting system operation; verifying the security authentication and read-write permission of external storage can prevent unauthorized external devices (such as U disk) from introducing malicious programs or leaking data; analyzing disk read-write can discover potential faults of storage media in time, ensuring the integrity and availability of data storage, and generating prompt information for all kinds of verification anomalies (such as inconsistent information, unauthorized changes, risk accounts, storage anomalies, etc.) can help managers quickly locate problems and take measures, improve risk disposal efficiency and reduce the probability of security incidents.

[0184] The dynamic performance attenuation coefficient δ SSD The acquisition process is as follows:

[0185] The following steps are calculated:

[0186] First, build a temperature-speed correlation matrix, specifically:

[0187]

[0188] k≥10, ΔT i =t i -t i-1 , ΔV i =|v i -v base |;

[0189] Wherein, k is the number of consecutive samples collected in the current sampling period (k≥10 is required to ensure calculation stability), the latest k consecutive samples are intercepted from the synchronously collected temperature sequence TSSD and read-write speed sequence V; ti is the temperature value (unit: ℃) at the i-th sampling time in T SSD , which is collected in real time by the built-in temperature sensor of SSD; vi is the V SSDThe read-write speed value (unit: MB / s) at the i-th sampling time, read in real time through the disk controller interface; vbase is the SSD reference speed, measured by the manufacturer before shipment in a standard environment of 25℃ (humidity 50%±5%, no vibration) for 1 hour of continuous read-write test, stored in the calibration parameter area of the disk firmware, and can be read through the API provided by the manufacturer;

[0190] Extract the temperature change vector M of the matrix T = [ΔT1, ΔT2,..., ΔT k ] T And the speed attenuation vector M V = [ΔV1, ΔV2,..., ΔV k ] T ;

[0191] Then calculate the attenuation coefficient, the specific process is:

[0192]

[0193] Wherein, is the gradient of the temperature change vector, obtained by performing a first-order difference operation on MT, that is,

[0194] ||·||2 is the L2 norm, calculated as the square root of the sum of the squares of the vector elements;

[0195] cov(M T , M V ) is the covariance of the two vectors, calculated using the statistical standard formula, that is,

[0196] And are the mean values of M T and M V , respectively;

[0197] var(M V ) is the variance of the speed attenuation vector, calculated using the statistical standard formula

[0198]

[0199] ∈=10 -5 , is a preset constant, set by the system in advance (fixed in the monitoring program), to avoid calculation anomalies when the denominator is zero;

[0200] t max is the highest temperature in the current sampling period, obtained by traversing the temperature values t1, t2...tk of the k samples in TSSD and taking the maximum value;

[0201] β(t max) = 1 + 0.05(t max -70) is a temperature penalty term, based on t max Real-time calculation (when t max ≤70℃, β(t max )≥1);

[0202] The above process not only considers speed attenuation alone, but also quantifies the correlation between temperature change and speed attenuation (such as whether high temperature exacerbates speed decline) through covariance, avoiding the one-sidedness of evaluation by a single parameter (such as only looking at speed), and better fitting the temperature-sensitive characteristics of SSD (temperature rise easily leads to performance degradation).

[0203] Based on the time series data of the last k samples (k≥10), rather than a single time value, the dynamic degradation trend of SSD performance can be reflected, the interference of accidental fluctuations on the result can be reduced, and the evaluation is more stable.

[0204] By β(t max ), the impact of high temperature on performance is strengthened (such as t max =80℃, β=1+0.05×10=1.5, amplifying the attenuation coefficient under high temperature), so that the evaluation result is more consistent with the actual performance of SSD in different temperature environments.

[0205] Introducing statistical methods such as L2 norm, covariance, and variance to quantify features, and avoiding calculation anomalies with zero denominator through ∈, to ensure the scientificity and stability of coefficient calculation.

[0206] Suppose a certain SSD in the host is monitored:

[0207] Sampling period τ=5s, k=10 (satisfying k≥10);

[0208] Temperature sequence T(℃): [30, 32, 33, 35, 36, 38, 39, 40, 41, 42], calculate ΔT i (℃): [2, 1, 2, 1, 2, 1, 1, 1, 1];

[0209] Read and write speed sequence V(MB / s): [500, 490, 485, 480, 475, 470, 465, 460, 455, 450], reference speed v base =500MB / s, calculate ΔV i (MB / s): [0, 10, 15, 20, 25, 30, 35, 40, 45, 50];

[0210] Extract MT=[2, 1, 2, 1, 2, 1, 1, 1, 1] T , MV=[0, 10, 15, 20, 25, 30, 35, 40, 45, 50] T ;

[0211] calculate: The L2 norm is approximately 2.45; cov(MT, MV) = 12.3; var(MV) = 264.4; t max =42℃ (β=1, because t max ≤70℃);

[0212] Substitute into the formula: δ SSD = (2.45 × 12.3) / (264.4 + 10) -5 )×1≈29.935 / 264.4≈0.113.

[0213] If the anomaly threshold Γth of the SSD is 0.2, then δ SSD =0.113 < Γth, indicating that the current performance degradation is normal; if δ is found to be normal in subsequent monitoring... SSD If the value rises to 0.25 (>Γth), it is determined that the performance degradation is abnormal and maintenance should be requested.

[0214] Furthermore, the calculation process for the mechanical health index ηHDD is as follows:

[0215] For each nj(f) (j = 1, 2, ..., m) in the acquired noise spectrum NHDD(f) = {n1(f), n2(f), ..., nm(f)}, perform MFCC transformation: use a 40-channel Mel filter bank with a frequency range of 500Hz-15kHz, and extract the first 12 dimensions of coefficients.

[0216] Where nj(f) is N HDD The noise spectrum (unit: dB / Hz) at the j-th sampling time in (f) is obtained by fast Fourier transform (FFT) after the original audio signal is acquired by a microphone array (sampling rate 44.1kHz, 16-bit quantization) deployed on the surface of the HDD housing.

[0217] The parameters (center frequency, bandwidth) of the 40-channel Mel filter bank are preset by the system (optimized based on the mechanical noise characteristics of HDD and embedded in the audio processing module), with a frequency range of 500Hz-15kHz, which is the main distribution range of HDD mechanical fault noise.

[0218] The first 12 feature coefficients after MFCC transformation are extracted using the standard MFCC interface of an open-source audio processing library (such as Librosa).

[0219] Construct a feature vector by taking the mean of m sampling points. in

[0220] Where: m is the total number of samples in the current sampling period, which is consistent with the length of VHDD, that is, m is equal to the number of samplings in the sampling period τ;

[0221] The mean of the i-th dimension MFCC coefficients is obtained by sampling m points. m) Calculate using the arithmetic mean;

[0222] Calculate your health index:

[0223] C0 is the HDD baseline feature vector, which is established by the manufacturer by performing 100 repeated samplings on the same model of HDD under a standard quiet environment (background noise ≤30dB) before leaving the factory and taking the average value, and is stored in the baseline database of the monitoring system.

[0224] ||C-C0||2 is the L2 norm of two vectors, calculated by taking the square root of the sum of the squares of the differences between their corresponding elements;

[0225] γ = 0.1 is the speed fluctuation weighting coefficient, which is determined by historical data statistics (based on the pre-fault speed fluctuation characteristics of 1000 HDDs of the same model, and is embedded in the monitoring algorithm);

[0226] The absolute value of the rate of change of read / write speed is given by V. HDD Perform a first-order difference calculation and then take the absolute value to obtain the result. Unit: MB / (s·s), where v′ j For V HDD The read / write speed at the j-th sampling moment (read via the disk controller interface), where τ is the sampling period;

[0227] Combination of mechanical noise and read-write speed dimensions: HDD failures are often accompanied by mechanical component wear (such as motors, heads), which manifest as noise abnormalities (such as abnormal sounds) and speed fluctuations. The combination of the two is more comprehensive than a single indicator (such as only looking at speed), and can capture potential failures earlier. Extract the first 12-dimensional coefficients of the noise spectrum through MFCC transformation: MFCC is good at capturing audio features related to human auditory perception, effectively filtering irrelevant noise and focusing on the characteristic frequencies of HDD mechanical failures (500Hz-15kHz, the document specifies the fault noise interval), more accurate than original spectrum analysis; Take the factory reference feature vector C0 as the reference, quantify the difference between the current noise feature and the reference through L2 norm, convert the abstract health status into a calculable value, avoid subjective judgment, and make the health index have a clear physical meaning (the smaller the difference, the higher the health degree), HDD performance degradation often manifests as unstable speed (increasing fluctuations), this index can reflect the trend of dynamic performance changes, avoiding the limitations of evaluating only through static speed values (such as average speed) (such as average speed is normal but short-term fluctuations are severe), balance the influence of noise feature difference and speed fluctuation through γ = 0.1: the weight coefficient calibrated by historical data makes the model more consistent with the actual failure rules of HDD (such as noise abnormalities have a greater impact on health, but severe speed fluctuations also need to be considered), improving the scientific nature of the evaluation.

[0228] For example, monitoring a certain type of HDD in a host:

[0229] Sampling period τ = 8s, collect m = 15 samples, i.e. 15 noise spectra and read-write speed data;

[0230] For noise spectrum N HDD (f) Perform MFCC transformation to extract the first 12-dimensional coefficients, and calculate the mean to get feature vector C;

[0231] The reference feature vector C0 (same type factory reference) is known, and ||C-C0||2 = 0.6 is calculated (reflecting the difference between the current noise and the reference);

[0232] Read-write speed sequence V HDD The first-order difference calculation gives the mean of the absolute value of the speed change rate

[0233] Substitute the formula: η HDD = 1 / (1+0.6)-0.1×0.3≈0.625-0.03=0.595.

[0234] If the abnormal threshold H th of this type of HDD is 0.4, then η HDD = 0.595 > H th, the current mechanical health state is normal; if in the subsequent monitoring, ||C-C0||2 increases to 1.2 and the speed change rate increases to 0.8 due to head wear, then HDD =1 / (1+1.2)-0.1x0.8≈0.455-0.08=0.375<Hth, it is determined to be abnormal and a prompt message is generated.

[0235] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and they should be covered in the scope of the claims and the description of the present application.

Claims

1. A method for verifying the configuration of a host system for multi-source data analysis, characterized in that, Includes the following steps: Step 1: Dynamically collect multi-source environmental parameters of the host through a multi-source sensor array, analyze the multi-source environmental parameters of the host, and determine whether the host environment meets the verification environment requirements. If it meets the environmental requirements, proceed to the next step. Step 2: Insert the configuration verification device into the corresponding interface of the host. At this time, the configuration verification device will capture images and verify the access status by the images. After the access status verification is successful, mechanical-electrical dual stability verification will be performed. After the mechanical-electrical dual stability verification is successful, proceed to the next step. Step 3: Perform host system configuration verification. The system configuration verification includes operating system verification, account content verification, storage system verification, and hardware information verification. If any item is abnormal, a prompt message will be generated. Step 4: Send the prompt information generated during the verification process in Step 3 to the corresponding receiving terminal.

2. The host system configuration verification method for multi-source data analysis according to claim 1, characterized in that: The specific process of analyzing host environment parameters to determine whether the host environment meets the verification environment requirements is as follows: Multi-source sensor arrays are used to collect environmental temperature information, environmental vibration information, and environmental electromagnetic information. Ambient temperature information includes real-time temperature and temperature gradient vector; ambient vibration information includes real-time vibration force and vibration wavelet packet energy entropy; ambient electromagnetic information includes real-time electromagnetic intensity and electromagnetic power spectral density. When any one of the real-time vibration force, real-time electromagnetic intensity, and real-time electromagnetic intensity exceeds the threshold of the corresponding item, it indicates that the host environment does not meet the verification environment requirements. The vibration wavelet packet energy entropy, temperature gradient vector, and electromagnetic power spectral density are input into the environmental risk prediction model, and the dynamic risk coefficient Kenv is output. When Kenv is within the preset safety threshold, it means that the environmental requirements for verification are met.

3. The host system configuration verification method for multi-source data analysis according to claim 2, characterized in that: The specific construction process of the environmental risk prediction model is as follows: Where σ is the Sigmoid function, Let E be the temperature gradient vector. vib Let P be the energy entropy of the vibrational wavelet packet. EM is the electromagnetic power spectral density, TCN is the temporal convolutional network, wi is the weight coefficient of the multi-source parameters, and i is the index variable for the summation operation, used to traverse each sample or parameter item involved in the calculation, with a value from 1 to n, where n is the total number of samples or parameter items involved in the model calculation, i.e., the upper limit of the range of the summation operation.

4. The host system configuration verification method for multi-source data analysis according to claim 1, characterized in that: The specific process of verifying the access status of the image in step two is as follows: Preset reference points: Preset two reference points a1 and a2 on the configuration verification device, and set two reference points b1 and b2 on the host, where a1 and b1 are on the same side, and a2 and b2 are on the other side; Real-time line acquisition: Extract the image of the configuration verification device, extract reference points a1 and a2 from the image, and extract reference points b1 and b2 from the host; then perform a connection operation to connect a1 with b1 and b2 respectively to obtain real-time lines L1 and L2. Connect a2 to b1 and b2 respectively to obtain real-time lines L3 and L4; Baseline settings: Preset baselines F1, F2, F3, and F4, where F1 corresponds to real-time line L1, F2 corresponds to real-time line L2, F3 corresponds to real-time line L3, and F4 corresponds to real-time line L4. Verification and judgment: Calculate the length difference between the baseline and the corresponding real-time line, namely the length difference between F1 and L1, the length difference between F2 and L2, the length difference between F3 and L3, and the length difference between F4 and L4. If any of the above length differences exceeds the preset range, the access status verification is judged to have failed; if all four length differences are within the preset range, the access status verification is judged to have passed.

5. The host system configuration verification method for multi-source data analysis according to claim 4, characterized in that: When reference points a1 and a2 cannot be captured in the images from the configured verification equipment, the following verification process is performed: The physical dimensions and structural features of the configuration verification equipment are pre-stored; When reference points a1 and a2 cannot be collected, the device edge midpoint or a preset backup marker point is automatically identified as temporary reference points a1' and a2'. Temporary real-time lines L1' to L4' are generated according to the original logic and compared with the baseline lines F1 to F4. If a substitute point cannot be identified, extract the complete outline of the configuration verification equipment from the image of the configuration verification equipment, and calculate the shortest distances d1 and d2 from the edge of the outline to the reference points b1 and b2. If the distances are compared with the preset standard distances d10 and d20, and |d1-d10|≤0.5cm and |d2-d20|≤0.5cm, and the angle between the center line of the contour and the host is within the preset range, then the verification is considered successful. When standard points b1 and b2 cannot be collected, the position parameters of the host-side interface are pre-stored as substitute features for reference points b1 and b2. When reference points b1 and b2 cannot be acquired, the center position of the host interface is identified as temporary reference points b1' and b2'. After generating real-time lines L1' to L4', they are compared with baseline lines F1 to F4.

6. The host system configuration verification method for multi-source data analysis according to claim 4, characterized in that: The specific process of the mechanical-electric dual stability verification is as follows: Mechanical stability verification, specifically: Pressure sensor matrix configuration: A distributed flexible pressure sensor array is adopted, containing Q×K piezoresistive sensor units, which are evenly distributed on the interface contact surface; Contact area calculation: The effective contact area is obtained by processing the output signal of the distributed flexible pressure sensor array using a threshold segmentation method, specifically as follows: Where Sik represents the binary contact state, and ΔA represents the area of ​​a single sensor unit; Stability coefficient calculation: The mechanical stability coefficient is calculated based on the contact area and the number of insertions and removals, specifically as follows: The verification is considered successful when η is greater than or equal to the corresponding threshold. A0 is the initial design contact area, obtained through calibration using standard test pieces; N is the recorded number of insertions and removals, stored in the host non-volatile memory; Nmax is the maximum permissible number of insertions and removals determined by fatigue life testing of the interface material; The specific process for electrical stability verification is as follows: Test signal injection: A composite signal with a frequency set F = {f1, f2, ..., fk} is sent to the interface through a test signal generator; Signal Acquisition and Analysis: The response signal is acquired at the interface receiver using a spectrum analyzer, and the attenuation rate of each frequency component is calculated. The specific process is as follows: Where A in (f) represents the amplitude of the injected signal, A out (f) represents the amplitude of the received signal; When the attenuation rate ρf of all frequency components is less than or equal to the corresponding threshold, the electrical connection of the interface is determined to be reliable; if there are frequency components with ρf greater than the corresponding threshold, the faulty frequency segment is further located through principal component analysis, and targeted maintenance suggestions are generated. When both mechanical stability verification and electrical stability verification are passed simultaneously, the mechanical-electrical dual stability verification is deemed to have passed.

7. The host system configuration verification method for multi-source data analysis according to claim 1, characterized in that: The operating system verification includes verifying host identifiers and asset information; The specific verification process for host identification and asset information is as follows: verify whether the host name, IP address, and MAC address are consistent with the asset list, and whether there are any unregistered hosts. If there are inconsistencies or unregistered hosts, a prompt message will be generated. Whether the business, responsible person, or purpose is clearly defined, and whether there are any unauthorized changes to the purpose; if the business, responsible person, or purpose is unclear or there are unauthorized changes to the purpose, a prompt message will be generated. If the system installation time and version information do not meet the standard requirements, a prompt message will be generated. The specific process of account content verification is as follows: whether there are accounts that have not been logged in for a long time, redundant accounts or shared accounts; if there are accounts that have not been logged in for a long time, or redundant accounts or shared accounts, a prompt message is generated. The storage system checks whether the disk space usage is abnormal and whether there is abnormal usage. When there is abnormal disk space usage or abnormal usage, a prompt message is generated. Whether the mounted external storage has undergone security authentication and whether read and write permissions have been restricted. If the mounted external storage has not undergone security authentication or read and write permissions have been restricted, a prompt message will be generated. Perform anomaly analysis on disk read / write operations to determine if any anomalies exist. If anomalies are found during disk read / write operations, a prompt message will be generated.

8. The host system configuration verification method for multi-source data analysis according to claim 7, characterized in that: During the anomaly analysis of disk read / write operations, a differentiated detection strategy is adopted based on the disk type. The specific process is as follows: When the disk is detected to be an SSD, the read / write speed sequence V = {v1, v2, ..., vn} and the temperature sequence T = {t1, t1, ..., tn} are synchronously collected at a sampling period τ, and the dynamic performance degradation coefficient δ is calculated. SSD ; When the disk is detected as an HDD, the read / write speed sequence VHDD = {v1′, v2′, ..., vm′} is synchronously collected with the same sampling period τ, and the noise spectrum NHDD(f) = {n1(f), n2(f), ..., nm(f)} is collected through a pre-set microphone array to calculate the mechanical health index η. HDD ; If δ SSD >Γth or η HDD If <Hth, then a disk read / write error is detected and a prompt message is generated; Among them, the threshold Γth is the SSD anomaly detection threshold, and Hth is the HDD anomaly detection threshold, both of which are determined by a logistic regression model trained on historical fault data.

Images