Cloud-driven intelligent management method and system

Abstract

The invention discloses a cloud drive intelligent management method and system, and relates to the technical field of drive management, and the method comprises the steps: collecting equipment operation state data, and carrying out the preprocessing of the equipment operation state data; based on the preprocessed equipment operation state data, extracting compatibility characteristics of the equipment, and predicting the matching degree of the equipment and a driver according to the compatibility characteristics of the equipment; dynamically selecting an optimal drive version based on the matching degree of the equipment and the drive; the equipment state is monitored in real time and compared with the equipment normal state baseline, and the driving fault probability is dynamically recognized according to the comparison result; and performing dynamic early warning according to the identification result of the driving fault probability. According to the method, the compatibility characteristics are extracted based on the preprocessed equipment operation state data, and the matching degree of the equipment and the drive is predicted, so that the complex relationship between the equipment and the drive is accurately quantified, and the compatibility and the stability are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of drive management, and in particular to a cloud drive intelligent management method and system. Background Art

[0002] In recent years, with the rapid development of information technology and the wide application of intelligent devices, the importance of audio and video conferencing systems in enterprises, educational institutions and various organizations has become increasingly prominent. As a key component among them, the cloud drive intelligent management method is playing an increasingly important role. Traditional drive management methods mainly rely on local storage and physical media distribution, such as optical discs or USB flash drives, etc. This method is not only inefficient, but also difficult to meet the rapidly changing meeting requirements. In addition, the drive file formats and versions of different devices may vary, resulting in compatibility issues and affecting the normal operation of the devices. With the continuous development and popularization of cloud computing technology, the application of cloud services in drive management is becoming more and more extensive, providing advantages such as elastic expansion, high availability and centralized management, and significantly improving the efficiency and quality of drive management.

[0003] However, although the existing technologies have made certain progress, there are still some deficiencies. First, in terms of the management and update of drive files, traditional methods often require developers to separately adapt and produce device drivers for each project, which not only consumes a large amount of time and energy, but also increases the risk of errors. Second, when on-site implementers install and configure devices, they usually need to carry multiple drive media and perform complex operations on-site. Once problems such as drive incompatibility or version errors are encountered, a large amount of time is required to troubleshoot and solve them, seriously affecting the implementation progress and quality of the project. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a cloud drive intelligent management method to solve the problems of low efficiency in drive file management and update and the project implementation progress delay and quality problems caused by complex on-site installation and configuration.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a cloud drive intelligent management method, which includes collecting device operation status data and preprocessing the device operation status data; based on the preprocessed device operation status data, extracting the compatibility characteristics of the device, and according to the compatibility characteristics of the device, predicting the matching degree between the device and the drive; based on the matching degree between the device and the drive, dynamically selecting the best drive version; real-time monitoring the device status and comparing it with the normal device status baseline, and dynamically identifying the drive failure probability according to the comparison result; and performing dynamic early warning according to the identification result of the drive failure probability.

[0008] As a preferred solution of the cloud-driven intelligent management method described in the present invention, wherein: the device operation status data includes device hardware configuration data, real-time performance indicators, log information, error information, driver interaction status, and environmental parameters;

[0009] The preprocessing of the device operation status data is specifically carried out according to the following steps:

[0010] The device operation status data is cleaned by using the interpolation method and the mean filling method;

[0011] The device operation status data is normalized by using the standardization process;

[0012] The random noise in the device operation status data is removed by using the filtering algorithm.

[0013] As a preferred solution of the cloud-driven intelligent management method described in the present invention, wherein: the compatibility characteristics of the device include hardware configuration characteristics, driver interaction characteristics, error log characteristics, performance fluctuation characteristics, and environmental adaptation characteristics, and the extraction process is specifically as follows:

[0014] The hardware configuration characteristics are extracted by analyzing the device model, hardware version, and driver version;

[0015] The driver interaction characteristics are extracted by analyzing the call success rate, average interface response time, and I / O error rate of the device;

[0016] The error log characteristics are extracted by counting the number of driver loading failures, error type distribution, and error frequency;

[0017] The performance fluctuation characteristics are extracted by analyzing the fluctuation of the performance indicators after the device loads the driver through time series analysis;

[0018] The environmental adaptation characteristics are extracted by analyzing the interference degree of environmental parameters on the device behavior.

[0019] As a preferred solution of the cloud-driven intelligent management method described in the present invention, wherein: the prediction of the matching degree between the device and the driver according to the compatibility characteristics of the device is specifically carried out according to the following steps:

[0020] Based on the compatibility characteristics of the device, through the linear combination and nonlinear processing of the hardware configuration characteristics and the driver interaction characteristics, and combining the periodic adjustment effect of the environmental adaptation characteristics to form the numerator, and at the same time placing the cumulative influence of the negative characteristics of the error log characteristics and the performance fluctuation characteristics in the denominator to suppress the compatibility score, finally realizing the balanced fusion of positive and negative characteristics and environmental adjustment, and generating a comprehensive feature vector, and the expression is:

[0021]

[0022] Among them, α is the weight coefficient of hardware configuration feature and driver interaction feature, β is the weight coefficient of error log feature and performance fluctuation feature, γ is the amplitude of periodic adjustment of compatibility score by environment adaptation feature, C i is the hardware configuration feature of the ith device, D i is the driver interaction feature of the i-th device, E i is the error log feature of the ith device, P i is the performance fluctuation characteristic of the i-th device, A i is the environmental adaptation characteristic of the i-th device, F i is the comprehensive feature vector of the i-th device;

[0023] Define target characteristics and equipment and drive operating status standards based on the historical operating data of equipment and drives and the average of historical performance indicators;

[0024] According to the comprehensive feature vector F i The deviation from the target feature is used to calculate the similarity of the compatibility features of the device using a nonlinear attenuation function to obtain the initial matching degree between the device and the driver;

[0025] For the comprehensive feature vector F i Perform logarithmic transformation and combine it with periodic analysis to explore the periodic compatibility characteristics between equipment and drivers;

[0026] Based on the periodic compatibility characteristics and combined with the historical interaction records between the device and the driver, the positive contribution of long-term stability to the matching degree is extracted, the initial matching degree is optimized, and finally the matching degree prediction result of the device and the driver is generated. The expression is:

[0027]

[0028] Among them, U is the reference value of the compatibility characteristic, λ is the nonlinear attenuation factor, ω is the parameter of the periodic compatibility characteristic between the device and the driver, and R i It is the historical interaction record between the device and the driver. i is the error value between the equipment and the drive during operation and the operating state standard, and β is the measure of the error value E i Parameters that affect the matching degree, M i It is the matching degree between the device and the driver.

[0029] As a preferred solution of the cloud drive intelligent management method of the present invention, the following specific steps are taken to dynamically select the best driver version based on the matching degree between the device and the driver, and establish a behavioral baseline for the normal working state of the device:

[0030] Define the matching reference benchmark based on the mean and fluctuation range of historical operation data and performance indicators;

[0031] Sort all the matching degrees M of the drivers i in descending order, and select the driver version with the highest matching degree from the sorting result as the best matching version for the current device;

[0032] Continuously monitor the matching degree of the current best driver version. If the best driver version of the current device is lower than the matching degree reference benchmark, then the matching degrees M of all drivers i are re-sorted, and the driver version with the highest matching degree is selected again as the best driver version for the current device.

[0033] As a preferred solution of the cloud driver intelligent management method described in the present invention, wherein: the specific steps of the real-time monitoring of the device status and comparison with the device normal status baseline are as follows:

[0034] Based on the historical data of the device running under the best driver version, analyze the mean and fluctuation range of the device historical running status data, and establish the device normal status baseline;

[0035] Real-time monitor the running status data of the device when running under the current best driver version;

[0036] If the device running data is within the normal status baseline range, the device running status data is marked as normal;

[0037] If the device running data exceeds the normal status baseline range, the device running status data is marked as abnormal;

[0038] The dynamic identification of the driver failure probability according to the comparison result includes constructing a driver identification model, and the specific steps are as follows,

[0039] The input layer extracts the time series data of the driver of the device under the current best driver version;

[0040] The LSTM layer captures the long-term dependence relationship of the time series data through the LSTM network and extracts the dynamic running status features of the driver;

[0041] The Attention layer dynamically assigns weights to the dynamic running status features of the driver at different time points through the Attention mechanism;

[0042] The fully connected layer extracts the high-dimensional running status features based on the weights of the dynamic running status features of the driver;

[0043] The output layer finally outputs the failure probability of the driver;

[0044] Integrate the input layer, the LSTM layer, the Attention layer, the fully connected layer and the output layer, and finally form a failure identification model;

[0045] When the operating state of the device is marked as abnormal, the failure probability of the drive is predicted through the failure identification model, and the expression is:

[0046]

[0047] Among them, P is the failure probability of the current drive, r t is the weight of the dynamic operating state characteristics of the drive at time point t, h t is the dynamic operating state characteristic at time point t, is the mean value of the dynamic operating state characteristic h t of, q is the deviation control parameter of the dynamic operating state characteristic h t from the mean value of the dynamic operating state characteristic of, h max is the maximum value of the dynamic operating state characteristic, z t is the high-dimensional operating state characteristic output by the fully connected layer, ∈ is the regularization factor, and T represents the total number of time steps of the drive operating state data.

[0048] As a preferred solution of the cloud drive intelligent management method described in the present invention, wherein: the dynamic early warning is carried out according to the identification result of the drive failure probability, and the specific steps are as follows:

[0049] Based on the failure probability of the historical drive, define the low-risk threshold P1 and the high-risk threshold P2;

[0050] When P < P1, it is considered that there is no failure in the current drive;

[0051] When P1 ≤ P < P2, it is considered that there is a failure risk in the current drive, and the failure detection early warning is triggered;

[0052] When P ≥ P2, it is considered that there is a failure in the current drive, and the emergency failure early warning is triggered.

[0053] In a second aspect, the present invention provides a cloud drive intelligent management system, including a data acquisition module, a matching degree prediction module, a drive version selection module, a failure identification module, and a dynamic early warning module;

[0054] The data acquisition module is used to collect the device operating state data and preprocess the device operating state data;

[0055] The matching degree prediction module is used to extract the compatibility characteristics of the device based on the preprocessed device operating state data, and predict the matching degree between the device and the drive according to the compatibility characteristics of the device;

[0056] The drive version selection module is used to dynamically select the best drive version based on the matching degree between the device and the drive;

[0057] A fault identification module, which is used to monitor the device status in real time, compare it with the baseline of the normal device status, and dynamically identify the driving fault probability according to the comparison result;

[0058] A dynamic warning module, which is used to perform dynamic warning according to the identification result of the driving fault probability.

[0059] Thirdly, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the cloud-driven intelligent management method described in the first aspect of the present invention is implemented.

[0060] Fourthly, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the cloud-driven intelligent management method described in the first aspect of the present invention is implemented.

[0061] The beneficial effects of the present invention are as follows: by extracting compatibility features based on the preprocessed device operation status data and predicting the matching degree between the device and the driver, the precise quantification of the complex relationship between the device and the driver is realized, the compatibility and stability are significantly improved, seamless connection between new and old devices is ensured, and the fault risk caused by incompatibility is reduced; at the same time, by monitoring the device status in real time and comparing it with the normal status baseline, the driving fault probability is dynamically identified, the present invention can timely discover potential risks and trigger warnings before problems occur, improve reliability and efficiency, reduce maintenance costs and downtime, and ensure the continuity and stability of critical services. Description of the Drawings

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0063] Figure 1 It is a flowchart of the cloud-driven intelligent management method in Embodiment 1.

[0064] Figure 2 It is a schematic diagram of the cloud-driven intelligent management system in Embodiment 1. Detailed Embodiments

[0065] In order to make the above objects, features and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the drawings in the specification.

[0066] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways than those specifically described herein. Those skilled in the art can make similar extensions without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0067] Secondly, as used herein, an "embodiment" or "embodiments" refers to specific features, structures, or characteristics that may be included in at least one implementation of the present invention. The phrase "in an embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments.

[0068] Embodiment 1, referring to Figure 1 and Figure 2 , is the first embodiment of the present invention. This embodiment provides a cloud-driven intelligent management method, including the following steps:

[0069] S1. Collect device operation status data and preprocess the device operation status data.

[0070] The device operation status data includes device hardware configuration data, real-time performance metrics, log information, error information, driver interaction status, and environmental parameters.

[0071] It should be noted that the device hardware configuration data is obtained by reading information such as the device model, hardware version, and driver version; the real-time performance metrics are collected by monitoring dynamic performances such as CPU usage rate, memory occupancy, and network connection status; the log information is extracted from the log files generated by the device, covering content such as event type, severity level, and description; the error information comes from various exceptions and warnings recorded during the device operation; the driver interaction status is determined by analyzing the device call success rate, interface response time, and I / O operation conditions; and the environmental parameters are measured by built-in or external sensors for physical conditions such as temperature and humidity to comprehensively reflect the working environment where the device is located.

[0072] The preprocessing of the device operation status data is specifically as follows:

[0073] Use the interpolation method and the mean filling method to clean the device operation status data.

[0074] The specific process is as follows: For missing data points, first identify and remove the obviously abnormal or unreasonable data, and then use the relationship between adjacent data points to fill in the missing values in the numerical data through the interpolation method; for categorical data, perform mean filling according to the context logic or the most likely category selection to ensure the integrity and consistency of the data set.

[0075] Normalize the device operation status data through standardization processing.

[0076] The specific process is as follows: Convert all numerical data to the same scale range. By performing a linear transformation on the original data, make the mean of each feature become zero and the standard deviation become one. This process ensures the comparability of data with different dimensions in subsequent analysis, improving the consistency and accuracy of data processing.

[0077] Adopt a filtering algorithm to remove random noise in the device operation status data.

[0078] The specific process is as follows: Apply digital filtering technology to process time series data, identify and weaken or eliminate high-frequency random fluctuations, and retain the main trends and features of the data. The data after smoothing can more accurately reflect the true operation status of the device, reducing misjudgments caused by noise.

[0079] S2. Based on the preprocessed device operation status data, extract the compatibility features of the device, and predict the matching degree between the device and the driver according to the compatibility features of the device.

[0080] Extract the hardware configuration features by analyzing the device model, hardware version, and driver version.

[0081] Furthermore, first, access the information library of the device to obtain detailed device model and hardware version information; then, check the installed driver files to confirm their version numbers. Then, organize all the collected information in a predefined format to ensure that the hardware configuration information of each device is clearly and accurately reflected in the record.

[0082] Extract the driver interaction features by analyzing the device call success rate, average interface response time, and I / O error rate.

[0083] Furthermore, start the log parsing tool to extract the log entries of the device calling the driver, and mark the result (success or failure) of each call. For each call, record the interface response time, and conduct statistical analysis on these times to determine the average response time and the distribution of response times. At the same time, collect the error logs in all input / output operations, count the occurrence times of each type of error, and form a comprehensive record of driver interaction behavior.

[0084] Extract the error log features by counting the number of driver loading failures, error type distribution, and error frequency.

[0085] Further, read the log file of the device, identify and extract the log entries containing driver loading information. For each log, determine whether it involves a loading failure and count the total number of failures. Classify different error types according to the content of the error message and count the occurrence frequency of each type of error. Finally, organize all the statistical results into tabular data for subsequent analysis.

[0086] Through time series analysis, extract the fluctuation of performance metrics after the device loads the driver and extract the performance fluctuation characteristics.

[0087] Further, regularly collect key performance metric data of the device, such as CPU utilization, memory occupancy, etc., to form a time series data set. Apply time series analysis methods to identify the change patterns and abnormal fluctuation points of the performance metrics. By calculating the differences between adjacent time points, evaluate the stability of the performance metrics and record the time points and amplitudes of significant fluctuations. Finally, generate a detailed performance fluctuation report for evaluating the impact of driver loading on device performance.

[0088] By analyzing the degree of interference of environmental parameters on device behavior, extract environmental adaptability characteristics.

[0089] Further, deploy sensors to monitor environmental conditions such as temperature and humidity of the device's location, regularly collect this environmental data and synchronously record it with the device's behavior data. Compare the operating states of the device under different environmental conditions, identify the impact of environmental changes on device performance and stability. By establishing an association model between environmental conditions and device behavior, quantify the degree of interference of environmental factors on device behavior, and thus extract environmental adaptability characteristics to assist the device in maintaining optimal performance in different environments.

[0090] Based on the compatibility characteristics of the device, through the linear combination and non-linear processing of hardware configuration characteristics and driver interaction characteristics, and combining the periodic adjustment effect of environmental adaptation characteristics to form the numerator, while placing the cumulative impact of negative characteristic error log characteristics and performance fluctuation characteristics in the denominator to suppress the compatibility score, finally achieve the balanced integration of positive and negative characteristics and environmental adjustment, and generate a comprehensive feature vector. The expression is:

[0091]

[0092] Among them, α is the weight coefficient of hardware configuration characteristics and driver interaction characteristics, β is the weight coefficient of error log characteristics and performance fluctuation characteristics, γ is the amplitude of the periodic adjustment of environmental adaptation characteristics on the compatibility score, C i is the hardware configuration characteristic of the i-th device, D i is the driver interaction characteristic of the i-th device, E i is the error log characteristic of the i-th device, P iis the performance fluctuation characteristic of the i-th device, A i is the environmental adaptation characteristic of the i-th device, F i is the comprehensive feature vector of the ith device.

[0093] Further, the hardware configuration features and driver interaction features of each device are read, the two are added and multiplied by the corresponding weight coefficient, and then the sine function is applied for nonlinear transformation to form part of the numerator. Next, the environmental adaptation feature is introduced to further enrich the numerator content through the periodic adjustment effect. The specific method is to multiply the environmental adaptation feature by the periodic adjustment amplitude and then apply the cosine function to process it. The result is added to the above calculation result to form a complete numerator expression. For the denominator, the error log features and performance fluctuation features of each device are collected, and the cumulative impact of these negative features is quantified and placed in the denominator to suppress the undesirable compatibility score. The specific process is: the absolute value of the error log feature and the square value of the performance fluctuation feature are counted, the two are added and then multiplied by the corresponding weight coefficient, and finally 1 is added to ensure that the denominator is not zero. This process effectively balances the impact of positive and negative features, so that the comprehensive feature vector can accurately reflect the compatibility level between the device and the driver. Finally, through the construction of the above numerator and denominator, the comprehensive feature vector of each device is generated.

[0094] Based on the historical operation data of equipment and drivers and the average of historical performance indicators, define target characteristics and equipment and driver operation status standards.

[0095] According to the comprehensive feature vector F i The deviation from the target feature is calculated using a nonlinear attenuation function to calculate the similarity of the device's compatibility features and obtain the initial matching degree between the device and the driver.

[0096] The specific process is as follows: First, read the comprehensive feature vector F of each device i , and determine the predefined target features. Then, calculate the comprehensive feature vector F i The deviation value between the device and the target feature. Next, a nonlinear attenuation function is applied to process the deviation value, which maps the deviation value to a numerical range between 0 and 1, where values ​​close to 1 indicate high similarity and values ​​close to 0 indicate large differences. In this way, the similarity between the device and the target feature is quantified, thereby obtaining a preliminary matching degree between the device and the driver.

[0097] For the comprehensive feature vector F i Logarithmic transformation is performed and combined with periodic analysis to explore the periodic compatibility characteristics between devices and drivers.

[0098] The specific process is as follows: First, the comprehensive feature vector F iPerform a logarithmic transformation operation to enhance the impact of small changes in the data and make the feature changes in different time periods more obvious. Then, apply time series analysis methods to identify the trends and periodic patterns of the comprehensive feature vector over time. By analyzing these periodic patterns, discover the compatibility change rules between the device and the drive in different time periods. Finally, combine the results of the periodic analysis to extract features that can reflect the long-term stability between the device and the drive, which are used to optimize the initial matching degree to ensure that the final generated matching degree result more accurately reflects the actual compatibility between the device and the drive.

[0099] Based on the periodic compatibility characteristics, combined with the historical interaction records between the device and the drive, extract the positive contribution of long-term stability to the matching degree, optimize the initial matching degree, and finally generate the matching degree prediction result between the device and the drive, with the expression:

[0100]

[0101] where U is the reference value of the compatibility feature, λ is the non-linear attenuation factor, ω is the parameter of the periodic compatibility characteristic between the device and the drive, R i is the historical interaction record between the device and the drive, E i is the error value from the operating state standard during the operation of the device and the drive, and β is the parameter measuring the influence degree of the error value E i on the matching degree, and M i is the matching degree between the device and the drive.

[0102] Furthermore, first, read the comprehensive feature vector F i , and determine the predefined reference value U of the compatibility feature. Calculate the absolute deviation value between F i and U, input this deviation value into the non-linear attenuation function, and through exponential attenuation processing, obtain a value reflecting the influence of the deviation. Then, apply the cosine function to process the result of adding 1 to the logarithmically transformed comprehensive feature vector F i to capture the influence of the periodic compatibility characteristic. Next, perform hyperbolic sine function processing on the historical interaction record between the device and the drive to quantify the positive contribution of the historical interaction to the matching degree. At the same time, obtain the error value of the device and the drive relative to the operating state standard during the operation process, and calculate the power of its absolute value as part of the denominator to measure the influence of the error on the matching degree. Finally, combine the results of the above parts and generate the final matching degree prediction result M i .

[0103] S3. Dynamically select the best drive version based on the matching degree between the device and the drive.

[0104] Define the matching degree reference benchmark based on the mean and fluctuation range of the historical operation data and performance metrics.

[0105] Sort all the matching degrees M of the drivers i in descending order, and select the driver version with the highest matching degree from the sorting result as the best matching version for the current device.

[0106] Continuously monitor the matching degree of the current best driver version. If the best driver version of the current device is lower than the matching degree reference benchmark, then the matching degrees M of all the drivers i are re-sorted, and the driver version with the highest matching degree is selected again as the best driver version for the current device.

[0107] The specific process is as follows: First, collect the operation status data of the device during previous operations. Then, perform statistical analysis on these historical data, calculate the average value of each performance metric, and determine its fluctuation range, that is, the interval between the maximum value and the minimum value. Then, based on the calculated average value and fluctuation range, set a reasonable reference interval to measure whether the future operation status is normal. For each performance metric, use the historical average value of this metric as the core part of the reference benchmark, and at the same time consider its fluctuation range to ensure that the reference benchmark can accurately reflect the performance level of the device during normal operation. Finally, integrate the reference benchmarks of all performance metrics to form a comprehensive matching degree reference benchmark for subsequent matching degree evaluation and comparison.

[0108] S4. Real-time monitor the device status and compare it with the device normal status baseline, and dynamically identify the driver failure probability according to the comparison result.

[0109] Based on the historical data of the device during operation under the best driver version, analyze the average value and fluctuation range of the device historical operation status data, and establish a device normal status baseline.

[0110] The specific process is as follows: First, access the database storing the device operation records, and filter out all the historical data of the device during operation under the best driver version. Then, perform a detailed analysis on these historical data, calculate the average value of each performance metric during the entire operation cycle as the central reference point of this metric. At the same time, determine the maximum value and the minimum value of each performance metric to define its fluctuation range, ensuring that all reasonable changes during normal operation can be captured. For example, for the CPU utilization metric, collect the CPU utilization data of the device during all time periods of operation under the best driver version, calculate the average utilization of these data as the average value of this metric. Then, find the highest value and the lowest value of the CPU utilization during this period to form a fluctuation range. This process is repeatedly applied to all relevant performance metrics, such as memory occupancy, network response time, etc. Finally, integrate the average values and fluctuation ranges of all performance metrics to construct a baseline that comprehensively reflects the normal operation status of the device.

[0111] Real-time monitoring of the operating status data of the device when it is running under the current best driver version.

[0112] If the device operation data is within the normal state baseline range, the device operation status data is marked as normal.

[0113] If the device operation data exceeds the normal state baseline range, the device operation status data is marked as abnormal.

[0114] Dynamically identify the drive failure probability according to the comparison result, including building a drive identification model. The specific steps are as follows.

[0115] The input layer extracts the time series data of the drive when the device is under the current best driver version.

[0116] Furthermore, read the historical operation data of the device under the best driver version from the database storing the device operation records. These data include key performance indicators during the device operation, such as CPU utilization rate, memory occupancy rate, etc., which are arranged in chronological order to form a time series. For example, for a specific device, collect its CPU utilization rate data every minute in the past week, ensuring that the data covers different load conditions and operation scenarios, providing a detailed basis for subsequent analysis.

[0117] The LSTM layer captures the long-term dependencies of the time series data through the LSTM network and extracts the dynamic operation status features of the drive.

[0118] Furthermore, input the device operation time series data provided by the input layer into the LSTM network. The LSTM network processes each data point in the time series one by one, using its internal memory units to capture patterns and trends over a long time span. For example, when analyzing the time series of CPU utilization rate, LSTM can identify regular high-load situations that occur during certain specific time periods (such as 9 am every day) and extract them as important dynamic operation status features.

[0119] The Attention layer dynamically assigns weights to the dynamic operation status features of the drive at different time points through the Attention mechanism.

[0120] Furthermore, based on the output of the LSTM layer, introduce the Attention mechanism to evaluate the importance of each time point. The Attention mechanism automatically calculates and assigns different weights according to the relevance of each data point in the time series to the current analysis task. For example, if the CPU utilization rate abnormally increases during a certain period of time and this change is crucial for subsequent fault prediction, then the Attention mechanism will assign a higher weight to this period of time.

[0121] The fully connected layer extracts high-dimensional operating state features based on the weights of the drive's dynamic operating state characteristics.

[0122] Furthermore, it receives the dynamically weighted operating state features from the Attention layer and performs complex non-linear transformations through multiple layers of neurons. Each neuron is connected to all the features of the previous layer, enabling the fully connected layer to comprehensively consider the features and their weights at all time points, thereby extracting higher-dimensional abstract features. For example, for the CPU utilization rate and memory occupancy rate weighted by the Attention mechanism, the fully connected layer can identify the complex interaction patterns between the two, generate higher-level operating state features, and provide richer information for fault prediction.

[0123] The output layer finally outputs the fault probability of the drive.

[0124] Integrating the input layer, LSTM layer, Attention layer, fully connected layer, and output layer finally constructs a fault identification model.

[0125] When the device operating state is marked as abnormal, the fault probability of the drive is predicted through the fault identification model, and the expression is:

[0126]

[0127] where P is the current fault probability of the drive, r t is the weight of the drive's dynamic operating state characteristics at time point t, h t is the dynamic operating state characteristics at time point t, is the mean value of the dynamic operating state characteristics h t q is the deviation control parameter of the dynamic operating state characteristics h t from the mean value of the dynamic operating state characteristics h max is the maximum value of the dynamic operating state characteristics, z t is the high-dimensional operating state characteristics output by the fully connected layer, ∈ is the regularization factor, and T represents the total number of time steps of the drive operating state data.

[0128] The specific process is as follows: First, extract the dynamic operating state characteristics h t at each time point t from the time series data and calculate the deviation of these characteristics from their mean value . For each time point, apply the weight r t to weight the dynamic operating state characteristics, then adjust the degree of deviation according to the deviation control mechanism, and then map the adjusted deviation to a specific range through a sine function to reflect the importance of the characteristics at that time point. Next, combine the high-dimensional operating state characteristics z t output by the fully connected layer, calculate a comprehensive score to measure the correlation and changes between different time points. In this process, the characteristic differences between the current time point and the previous time point, as well as the characteristic changes between the current time point and the previous two time points, are considered to ensure a comprehensive capture of the change trend of the device operation status. Finally, the results after the above processing for all time points are summarized and adjusted by a normalization factor to ensure that the final output failure probability P is within a reasonable numerical range.

[0129] S5. Conduct dynamic early warning based on the identification result of the drive failure probability.

[0130] Define a low-risk threshold P1 and a high-risk threshold P2 based on the historical drive failure probability.

[0131] When P < P1, it is considered that there is no failure in the current drive.

[0132] For example, if the calculated failure probability P is 0.1 and the low-risk threshold P1 is set to 0.3, it is determined that the current drive is operating normally and no action needs to be taken.

[0133] When P1 ≤ P < P2, it is considered that there is a failure risk in the current drive, and a failure detection early warning is triggered.

[0134] For example, if the failure probability P is calculated as 0.4, which is between the low-risk threshold P1 (0.3) and the high-risk threshold P2 (0.7), a failure detection early warning is triggered to prompt the operator to pay attention to potential problems and conduct further inspections.

[0135] When P ≥ P2, it is considered that there is a failure in the current drive, and an emergency failure early warning is triggered.

[0136] For example, if the failure probability P reaches or exceeds 0.8, exceeding the high-risk threshold P2 (0.7), an emergency failure early warning is immediately triggered, instructing the operator to quickly take measures to prevent possible equipment failures or service interruptions.

[0137] This embodiment also provides a cloud drive intelligent management system, including: a data acquisition module, a matching degree prediction module, a drive version selection module, a failure identification module, and a dynamic early warning module;

[0138] The data acquisition module is used to collect device operation status data and preprocess the device operation status data;

[0139] The matching degree prediction module is used to extract the compatibility characteristics of the device based on the preprocessed device operation status data and predict the matching degree between the device and the drive according to the compatibility characteristics of the device;

[0140] The drive version selection module is used to dynamically select the best drive version based on the matching degree between the device and the drive;

[0141] A fault identification module, which is used to monitor the device status in real time and compare it with the baseline of the normal device status, and dynamically identify the drive fault probability according to the comparison result;

[0142] A dynamic warning module, which is used to perform dynamic warning according to the identification result of the drive fault probability.

[0143] This embodiment also provides a computer device, which is applicable to the case of the cloud drive intelligent management method, and includes: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to implement the cloud drive intelligent management method proposed in the above embodiment.

[0144] The computer device may be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a button, a trackball or a touchpad provided on the outer shell of the computer device, or an external keyboard, a touchpad or a mouse, etc.

[0145] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the cloud drive intelligent management method proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM for short), Electrically Erasable Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read Only Memory (EPROM for short), Programmable Red-Only Memory (PROM for short), Read-Only Memory (ROM for short), magnetic memory, flash memory, a magnetic disk or an optical disc.

[0146] In summary, the present invention realizes the precise quantification of the complex relationship between the device and the driver by extracting compatibility features based on the preprocessed device operation status data and predicting the matching degree between the device and the driver, significantly improving compatibility and stability, ensuring seamless connection between new and old devices and reducing the risk of failures caused by incompatibility. At the same time, by monitoring the device status in real time and comparing it with the normal status baseline to dynamically identify the probability of driver failure, the present invention can detect potential risks in a timely manner before problems occur and trigger warnings, improving reliability and efficiency, reducing maintenance costs and downtime, and ensuring the continuity and stability of critical services.

[0147] Embodiment 2, referring to Table 1, is the second embodiment of the present invention. To further verify the technical solution of the present invention, experimental simulation data of the cloud drive intelligent management method is given.

[0148] To verify the effectiveness and superiority of the cloud drive intelligent management method, a typical enterprise-level data center was selected as the experimental environment. The data center is equipped with multiple models of servers that run different operating systems and use various versions of driver programs. To ensure the comprehensiveness and accuracy of the experimental data, 5 servers with different configurations were selected from the data center as experimental objects, named Server A, Server B, Server C, Server D, and Server E respectively.

[0149] First, before the experiment started, a detailed hardware configuration check was performed on each server, including device model, hardware version, and driver version. Monitoring software was deployed to collect real-time performance metrics (such as CPU usage, memory occupancy, and network connection status), and a logging tool was set up to capture event information. Technicians manually recorded all error messages and analyzed the interface response time and I / O operation conditions to determine the driver interaction status. In addition, environmental sensors were installed to monitor temperature and humidity to comprehensively reflect the working environment of the servers.

[0150] Second, the collected data was subjected to preprocessing steps for cleaning, normalization, and denoising. The interpolation method and the mean filling method were used to fill in the missing values to ensure data integrity and consistency; the numerical data was converted to the same scale range through standardization processing to improve the accuracy of data processing; a filtering algorithm was applied to remove random noise to ensure the quality of subsequent analysis.

[0151] Next, after the preprocessing was completed, it entered the feature extraction stage, and a comprehensive feature vector was constructed based on hardware configuration, driver interaction, error logs, performance fluctuations, and environmental adaptability features. For example, when analyzing Server A, it was found that the CPU utilization fluctuated greatly during a specific period, and the association between this fluctuation and external conditions was further confirmed by combining environmental parameters such as the rising temperature.

[0152] Finally, calculate the matching degree between each server and the current driver version according to the comprehensive feature vector, introduce a non-linear attenuation function to evaluate the similarity, and capture the long-term stability characteristics through periodic analysis. Dynamically select the best driver version and establish a normal state baseline for real-time monitoring. When it is detected that the operating state exceeds the baseline range, start the fault identification model to predict the fault probability. For example, Server C triggered a fault detection warning due to a memory occupancy rate significantly higher than the baseline.

[0153] As shown in Table 1 below:

[0154] Table 1 Driver Fault Analysis Table

[0155]

[0156] Through the data analysis of the above table, it can be clearly seen that the present invention significantly improves the stability and compatibility of the server under different driver versions through the intelligent management method. For example, under the best driver version, the CPU usage rate and memory occupancy rate of Server A remain at a relatively stable level, and the driver fault probability is only 0.12, far lower than the high-risk threshold P2 (assumed to be 0.7), which indicates that even under a relatively high load, the device can maintain a good operating state and reduce the performance degradation caused by driver incompatibility.

[0157] In addition, the present invention greatly improves the accuracy of fault prediction by introducing advanced machine learning models such as the LSTM network and the Attention mechanism. For example, Server C had a memory occupancy rate significantly higher than the baseline during a certain period, triggering a fault detection warning (P = 0.45), indicating that there may be potential problems that need further inspection. In contrast, traditional methods often can only handle the faults after they actually occur, resulting in an increase in maintenance costs and an increase in the risk of service interruption.

[0158] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A cloud-driven intelligent management method, characterized in that: include, Collect equipment operation status data and pre-process the equipment operation status data; Based on the pre-processed device operation status data, the compatibility characteristics of the device are extracted, and the matching degree between the device and the driver is predicted according to the compatibility characteristics of the device; Dynamically select the best driver version based on the matching degree between the device and the driver; Monitor the device status in real time and compare it with the device normal status baseline, and dynamically identify the probability of drive failure based on the comparison results; Dynamic early warning is carried out based on the identification results of the drive failure probability.

2. The cloud-driven intelligent management method according to claim 1, characterized in that: The device operation status data includes device hardware configuration data, real-time performance indicators, log information, error information, driver interaction status, and environmental parameters; The specific steps of preprocessing the equipment operation status data are as follows: Use interpolation and mean filling methods to clean the equipment operation status data; Use standardization processing to normalize the equipment operation status data; A filtering algorithm is used to remove random noise from the equipment operation status data.

3. The cloud-driven intelligent management method according to claim 2, characterized in that: The compatibility features of the device include hardware configuration features, driver interaction features, error log features, performance fluctuation features, and environmental adaptability features. The extraction process is as follows: Extract hardware configuration features by analyzing device model, hardware version, and driver version; By analyzing the device's call success rate, interface response time average, and I / O error rate, driver interaction features are extracted; Extract error log features by counting the number of driver loading failures, error type distribution, and error frequency; Through time series analysis, the fluctuation of performance indicators after the device is loaded and driven is extracted, and the performance fluctuation characteristics are extracted; By analyzing the degree of interference of environmental parameters on device behavior, environmental adaptability characteristics are extracted.

4. The cloud-driven intelligent management method according to claim 3, characterized in that: The specific steps of predicting the matching degree between the device and the driver according to the compatibility characteristics of the device are as follows: Based on the compatibility characteristics of the device, the numerator is formed by linear combination and nonlinear processing of hardware configuration characteristics and driver interaction characteristics, combined with the periodic regulation of environmental adaptation characteristics. At the same time, the cumulative impact of negative feature error log characteristics and performance fluctuation characteristics is placed in the denominator to suppress the compatibility score. Finally, a balanced fusion of positive and negative characteristics and environmental regulation is achieved to generate a comprehensive feature vector, which is expressed as: Among them, α is the weight coefficient of hardware configuration feature and driver interaction feature, β is the weight coefficient of error log feature and performance fluctuation feature, γ is the amplitude of periodic adjustment of compatibility score by environment adaptation feature, C i is the hardware configuration feature of the ith device, D i is the driver interaction feature of the i-th device, E i is the error log feature of the ith device, P i is the performance fluctuation characteristic of the i-th device, A i is the environmental adaptation characteristic of the i-th device, F i is the comprehensive feature vector of the i-th device; Define target characteristics and equipment and drive operating status standards based on the historical operating data of equipment and drives and the average of historical performance indicators; According to the comprehensive feature vector F i The deviation from the target feature is used to calculate the similarity of the compatibility features of the device using a nonlinear attenuation function to obtain the initial matching degree between the device and the driver; For the comprehensive feature vector F i Perform logarithmic transformation and combine it with periodic analysis to explore the periodic compatibility characteristics between equipment and drivers; Based on the periodic compatibility characteristics and combined with the historical interaction records between the device and the driver, the positive contribution of long-term stability to the matching degree is extracted, the initial matching degree is optimized, and finally the matching degree prediction result of the device and the driver is generated. The expression is: Among them, U is the reference value of the compatibility characteristic, λ is the nonlinear attenuation factor, ω is the parameter of the periodic compatibility characteristic between the device and the driver, and R i It is the historical interaction record between the device and the driver. i is the error value between the equipment and the drive during operation and the operating state standard, and β is the measure of the error value E i Parameters that affect the matching degree, M i It is the matching degree between the device and the driver.

5. The cloud-driven intelligent management method according to claim 4, characterized in that: Based on the matching degree between the device and the driver, the best driver version is dynamically selected, and a behavior baseline of the normal working state of the device is established. The specific steps are as follows: Define the matching reference benchmark based on the mean and fluctuation range of historical operation data and performance indicators; Set the matching degree of all drivers to M i Sort from high to low, and select the driver version with the highest matching degree from the sorting results as the best matching version for the current device; Continuously monitor the matching degree of the current best driver version. If the best driver version of the current device is lower than the matching degree reference benchmark, the matching degree of all drivers M i Re-sort and select the driver version with the highest matching degree again as the best driver version for the current device.

6. The cloud-driven intelligent management method according to claim 5, characterized in that: The specific steps of real-time monitoring of device status and comparing it with the baseline of normal status of the device are as follows: Based on the historical data of the equipment running under the optimal driver version, analyze the mean and fluctuation range of the equipment's historical operating status data to establish a baseline for the normal status of the equipment; Real-time monitoring of the operating status data of the device when it is running under the current optimal driver version; If the equipment operation data is within the normal status baseline range, the equipment operation status data is marked as normal; If the equipment operation data exceeds the normal status baseline range, the equipment operation status data is marked as abnormal; The method of dynamically identifying the probability of a drive failure according to the comparison result includes constructing a drive identification model. The specific steps are as follows: The input layer extracts the time series data of the device driven under the current best driver version; The LSTM layer captures the long-term dependencies of time series data through the LSTM network and extracts the dynamic operating status characteristics of the driver; The Attention layer dynamically assigns weights to the dynamic running state features driven at different time points through the Attention mechanism; The fully connected layer extracts high-dimensional operating state features based on the weights of the dynamic operating state features of the driver; The failure probability of the final output driver of the output layer; The input layer, LSTM layer, Attention layer, fully connected layer, and output layer are integrated to form a fault identification model. When the equipment operation status is marked as abnormal, the fault probability of the drive is predicted by the fault identification model, and the expression is: Where P is the failure probability of the current drive, r t is the weight of the dynamic operating state characteristics of the driver at time point t, h t is the dynamic running state characteristic at time point t, is the dynamic running state characteristic h t The mean of q is the dynamic running state characteristic h t The mean of the dynamic running state characteristics The deviation control parameter, h max is the maximum value of the dynamic running state characteristic, z t is the high-dimensional running state feature output by the fully connected layer, ∈ is the regularization factor, and T represents the total time steps of the driving running state data.

7. The cloud-driven intelligent management method according to claim 6, characterized in that: The specific steps of performing dynamic early warning according to the identification result of the drive failure probability are as follows: Based on the history-driven failure probability, define the low-risk threshold P1 and the high-risk threshold P2; When P<P1, it is considered that there is no fault in the current drive; When P1≤P<P2, it is considered that the current drive has a fault risk, triggering a fault detection warning; When P≥P2, it is considered that the current drive has a fault, triggering an emergency fault warning.

8. A cloud-driven intelligent management system, based on the cloud-driven intelligent management method according to any one of claims 1 to 7, characterized in that: Including data acquisition module, matching prediction module, driver version selection module, fault identification module and dynamic warning module; A data acquisition module is used to collect equipment operation status data and pre-process the equipment operation status data; A matching degree prediction module is used to extract the compatibility characteristics of the device based on the pre-processed device operation status data, and predict the matching degree between the device and the driver according to the compatibility characteristics of the device; The driver version selection module is used to dynamically select the best driver version based on the matching degree between the device and the driver; Fault identification module, which is used to monitor the equipment status in real time and compare it with the equipment normal status baseline, and dynamically identify the probability of drive failure based on the comparison results; The dynamic warning module is used to issue dynamic warnings based on the identification results of the drive failure probability.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the cloud-driven intelligent management method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the cloud-driven intelligent management method described in any one of claims 1 to 7 are implemented.

Images