Network card interface connection method and device

Abstract

The invention discloses a network card interface connection method and device, and relates to the field of network card interface connection, and the method comprises the steps: collecting a level signal of a PCIE interface and a network use condition of network equipment; adjusting the priority of the PCIE interface by using an intelligent algorithm based on the network use condition, and generating a dynamic priority list; a standby PCIE interface is configured for a key task, a level signal of the PCIE interface is monitored to judge whether the PCIE interface fails or not, if the PCIE interface fails, the standby interface is rapidly switched, the state of the PCIE interface is re-evaluated after maintenance and repair by a dispatcher, the PCIE interface is switched back to the original PCIE interface, and a flow prediction model is constructed through a machine learning algorithm. And inputting the historical network traffic and the level signal of the PCIE interface into the traffic prediction model for training, outputting a traffic prediction value, judging a traffic demand state according to the traffic prediction value, and setting a PCIE interface state based on the traffic demand state. According to the invention, comprehensive monitoring of the current state is realized by collecting the level signal of the PCIE interface and the network use condition of the network equipment.

Description

Technical Field

[0001] The present invention relates to the field of network card interface connection, and particularly to a network card interface connection method and device. Background Art

[0002] With the rapid development of information technology, the importance of network communication in modern society has become increasingly prominent. Especially in data centers and cloud computing environments, an efficient network card interface connection method is crucial for ensuring the reliability and low latency of data transmission. Traditional network card interface connection methods mainly rely on static configuration, that is, resources are allocated according to preset rules during initialization and remain unchanged throughout the operation cycle. Although this static configuration method is simple and easy to implement, it seems inadequate in the face of dynamic network loads. In recent years, with the development of virtualization technology and software-defined network (SDN), more and more research has been dedicated to improving the flexibility and intelligence level of network card interfaces. For example, some research optimizes network traffic allocation through intelligent algorithms, and others attempt to use machine learning technology to predict future network loads and thus adjust resource configuration in advance. However, most of these methods are limited to single-dimensional optimization and fail to fully consider the overall performance improvement under the synergistic effect of multiple factors.

[0003] Although the existing technologies have made significant progress in some aspects, there are still many deficiencies. First, traditional PCIE interface management methods usually rely on static configuration and are difficult to dynamically adjust according to real-time network usage. This static configuration method often fails to respond in a timely manner in the face of network traffic fluctuations, resulting in low resource utilization and even bottleneck phenomena. Second, the existing technologies are still insufficient in predicting network traffic demands, often relying on simple statistical analysis of historical data and lacking the support of deep learning models, making it difficult to accurately capture complex and changing network behavior patterns. The existence of these problems not only limits the effective utilization of network resources but also affects the user experience to a certain extent. Summary of the Invention

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

[0005] Therefore, the present invention provides a network card interface connection method to solve the problems of insufficient prediction of network traffic demands, lack of support from deep learning models, and difficulty in accurately capturing complex and changing network behavior patterns.

[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 method for connecting a network card interface, which includes collecting the level signals of the PCIE interface and the network usage of the network device, adjusting the PCIE interface priority using an intelligent algorithm based on the network usage to generate a dynamic priority list, configuring a backup PCIE interface for critical tasks, monitoring the level signals of the PCIE interface, when the level signals of the PCIE interface are continuously high or continuously low, determining that the PCIE interface has a fault, quickly switching to the backup interface, dispatching personnel for repair, re-evaluating the status of the PCIE interface after repair, and switching back to the original PCIE interface when the stability of the level signal is normal. A traffic prediction model is constructed through a machine learning algorithm, and the historical network traffic and the level signals of the PCIE interface are used to train the traffic prediction model to output a traffic prediction value. According to the traffic prediction value, the traffic demand status is judged, and based on the traffic demand status, the status of the PCIE interface is set.

[0008] As a preferred solution of the network card interface connection method of the present invention, wherein: the collecting of the level signals of the PCIE interface and the network usage of the network device is specifically as follows:

[0009] Start the server, scan all available PCIE interfaces, and collect the level signals of the PCIE interfaces and the network usage of the network device;

[0010] Use a high-precision voltage sensor to monitor the level signals of each PCIE interface to obtain the stability of the level signals, and use software tools to collect the network usage, where the network usage refers to network traffic, CPU load, and memory usage rate;

[0011] Use a network splitter TAP to collect network traffic in real time, and use the monitoring tool Windows Task Manager to obtain the CPU load and memory usage rate;

[0012] As a preferred solution of the network card interface connection method of the present invention, wherein: the adjusting of the PCIE interface priority using an intelligent algorithm based on the network usage to generate a dynamic priority list is specifically as follows:

[0013] Use an intelligent algorithm with a linear regression model as the core to calculate the CPU load and memory usage rate of each PCIE interface,

[0014] Adjust the PCIE interface priority according to the calculated CPU load and memory usage rate and calculate the priority score P of each PCIE interface;

[0015] Sort based on the priority score P in descending order;

[0016] Set a priority threshold Q. When the priority score P > Q, it means that the PCIE interface has a high priority when processing tasks;

[0017] When the priority score P ≤ Q, it indicates that the PCIE interface has a low priority when processing tasks;

[0018] Use the built-in function sorted in Python to sort all PCIE interfaces in descending order to generate a dynamic priority list;

[0019] The dynamic priority list refers to the priority score of each PCIE interface and the corresponding task assignment. The interface with a higher priority will be assigned to critical tasks, and the interface with a lower priority will be assigned to non-critical tasks and used as a backup interface.

[0020] As a preferred solution of the network card interface connection method described in the present invention, wherein: a backup PCIE interface is configured for critical tasks, and the level signal of the PCIE interface is monitored. When the level signal of the PCIE interface is continuously high or continuously low, it is determined that the PCIE interface has a fault, and it is quickly switched to the backup interface, and personnel are dispatched for repair. After repair, the status of the PCIE interface is re-evaluated. When the stability of the level signal is normal, it is switched back to the original PCIE interface. Specifically,

[0021] Configure multiple backup PCIE interfaces for critical tasks;

[0022] The critical tasks refer to high-quality video stream transmission and data synchronization operations;

[0023] Use a high-precision voltage sensor to monitor the level signal of each PCIE interface, calculate the voltage of the level signal, and monitor the voltage change of the PCIE interface;

[0024] Set the high-level threshold as V h and the low-level threshold as V l and the high-level time threshold as T h and the low-level time threshold as T l ;

[0025] Compare the calculated voltage of the level signal with the high-level threshold V h and the low-level threshold V l to determine the fault;

[0026] When the voltage of the level signal is higher than V h and the duration exceeds the high-level time threshold T h , it is determined that the PCIE interface has a fault;

[0027] When the voltage of the level signal is lower than V l and the duration exceeds the low-level time threshold T l , it is determined that the PCIE interface has a fault;

[0028] When the voltage of the level signal is neither continuously high nor continuously low, it is determined that the PCIE interface is normal;

[0029] When the fault condition is met, the switching device is immediately triggered to quickly transfer the network communication task to the standby PCIE interface;

[0030] After the PCIE interface is switched to the standby interface, a maintenance work order for the physical location of the faulty interface and the abnormal waveform of the level signal is generated through the enterprise asset management EAM, and the maintenance work order is pushed to the handheld terminal of the maintenance personnel through the positioning of the Internet of Things device. At the same time, the physical slot of the faulty interface is locked;

[0031] Dispatch maintenance personnel to the site for maintenance. After the maintenance personnel arrive at the site, perform the repair work, replace the damaged components and adjust the configuration;

[0032] After the repair is completed, inject a gradient pressure load into the repaired PCIE interface through a programmable flow generator, pressurize in stages, synchronously collect the voltage of the level signal, and re-evaluate the status of the PCIE interface;

[0033] The status of the PCIE interface refers to faulty and normal;

[0034] Recalculate the voltage of the level signal. When V l ≤ level signal voltage ≤ V h , the stability of the level signal is normal, and it can be switched back to the original PCIE interface. When the level signal voltage > V h and the level signal voltage < V l , it is non-standard, and the repair continues.

[0035] As a preferred solution of the network card interface connection method described in the present invention, wherein: the traffic prediction model constructed based on the machine learning algorithm is specifically

[0036] Use a high-precision voltage sensor to collect the level signal of the PCIE interface, use a network splitter TAP to collect the original network traffic data, and apply Kalman filtering to eliminate electromagnetic interference EMI from the collected level signal;

[0037] Use moving average to extract the time series features in the level signal, use Pandas to divide the original network traffic data into time windows to generate statistical features;

[0038] Use an autoencoder to fuse the time series features and statistical features to generate a feature dataset;

[0039] Use the LSTM model as the basic framework of the network traffic prediction model, and define the basic framework of the network traffic prediction model as the input layer, LSTM layer, fully connected layer, and output layer.

[0040] As a preferred solution of the network card interface connection method described in the present invention, wherein: training the traffic prediction model with the historical network traffic and the level signals of the PCIE interface, and outputting a traffic prediction value, specifically,

[0041] Dividing the feature data set into a training set and a validation set;

[0042] The input layer inputs the training set for training;

[0043] The LSTM layer receives the training set from the input layer, captures the hidden state vectors in the feature data set through the gating mechanism, and passes them to the fully connected layer;

[0044] The fully connected layer receives the hidden state vectors, performs a linear transformation on the hidden state vectors, and outputs a prediction vector;

[0045] The output layer outputs the prediction vector as the final predicted traffic value;

[0046] Using the mean square error MSE and combining with an adaptive adjustment factor, calculate the difference between the predicted traffic value and the actual traffic value;

[0047] Calculate the gradient through the backpropagation algorithm, adjust the weights and biases of the traffic prediction model, and update the traffic prediction model parameters using the Adam optimizer;

[0048] Use the validation set to calculate the mean absolute error, and adjust the hyperparameters of the traffic prediction model according to the mean absolute error index.

[0049] As a preferred solution of the network card interface connection method described in the present invention, wherein: judging the traffic demand state according to the traffic prediction value, and adjusting the state of the priority PCIE interface based on the traffic demand state, specifically,

[0050] Set the peak threshold as R according to the traffic prediction value 1 , and the trough threshold as R 2 , compare the traffic prediction value with the peak threshold and the trough threshold to judge the traffic demand state;

[0051] When the predicted traffic value > R 1 , it is determined as the traffic peak period, when the predicted traffic value < R 2 , it is determined as the traffic trough period, and when R 2 ≤ the predicted traffic value ≤ R 1 , it is regarded as the normal traffic period;

[0052] The traffic demand state refers to the traffic peak period, the traffic trough period, and the traffic normal period;

[0053] Adjust the state of the priority PCIE interface based on the traffic demand state;

[0054] The state of the priority PCIE interface is a connected state and a disconnected state;

[0055] Read the dynamic priority list to identify the currently available high-priority PCIE interfaces and low-priority PCIE interfaces;

[0056] When the PCIE interface is in a peak traffic period, set the high-priority PCIE interface to the connected state;

[0057] When the PCIE interface is in a low traffic period, set the low-priority PCIE interface to the disconnected state;

[0058] When the PCIE interface is in a normal traffic period, keep the existing interface state unchanged.

[0059] In a second aspect, the present invention provides a network card interface connection device, including a data collection module, a priority adjustment module, a fault detection module, a prediction module, and a dynamic adjustment module.

[0060] The data collection module is used to collect the level signal of the PCIE interface and the network usage of the network device;

[0061] The priority adjustment module is used to adjust the PCIE interface priority using an intelligent algorithm based on the network usage to generate a dynamic priority list;

[0062] The fault detection module is used to configure a standby PCIE interface for critical tasks, monitor the level signal of the PCIE interface. When the level signal of the PCIE interface is continuously high or continuously low, it is determined that the PCIE interface has a fault, quickly switch to the standby interface, dispatch personnel for repair, re-evaluate the state of the PCIE interface after repair, and switch back to the original PCIE interface when the stability of the level signal is normal;

[0063] The prediction module is used to construct a traffic prediction model through a machine learning algorithm, train the traffic prediction model with historical network traffic and the level signal of the PCIE interface, and output a traffic prediction value;

[0064] The dynamic adjustment module is used to judge the traffic demand state according to the traffic prediction value, and set the PCIE interface state based on the traffic demand state.

[0065] In a third aspect, the present invention provides a computer device, including a memory and a processor. The memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the network card interface connection method described in the first aspect of the present invention is implemented.

[0066] Fourthly, the present invention provides a computer-readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by a processor, any step of the network card interface connection method described in the first aspect of the present invention is implemented.

[0067] The beneficial effects of the present invention are as follows: by collecting the level signals of the PCIE interface and the network usage of the network device, the present invention realizes the comprehensive monitoring of the current state, which not only provides basic data for subsequent intelligent algorithm adjustment, but also ensures the real-time and accuracy of the data. By dynamically adjusting the priority, the present invention realizes the intelligent scheduling of the PCIE interface, optimizes the resource allocation, and improves the ability to cope with sudden high loads. By real-time monitoring the level signals, it quickly switches to the standby interface to ensure the continuity and low latency of critical tasks. By precise repair and evaluation mechanisms, it ensures long-term stable operation and reduces downtime. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings 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.

[0069] Figure 1 It is a flowchart of the network card interface connection method in Embodiment 1.

[0070] Figure 2 It is a schematic diagram of the dynamic priority list in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0071] In order to make the above-mentioned objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention with reference to the accompanying drawings of the specification.

[0072] Many specific details are set forth in the following description in order to provide a thorough understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0073] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude other embodiments.

[0074] Embodiment 1, refer to Figure 1 and Figure 2, which is the first embodiment of the present invention. This embodiment provides a method for connecting a network card interface, including the following steps:

[0075] S1. Collect the level signals of the PCIE interface and the network usage of the network device.

[0076] Start the server, scan all available PCIE interfaces, and collect the level signals of the PCIE interface and the network usage of the network device;

[0077] Use a high-precision voltage sensor (such as a high-precision voltage sensor with an accuracy of 0.01V) to monitor the level signals of each PCIE interface. During monitoring, collect the level signals once per second and control the voltage resolution at 0.01V to obtain the stability of the level signals. Use software tools to collect the network usage, where the network usage refers to network traffic, CPU load, and memory usage rate;

[0078] Use a network splitter TAP to collect network traffic in real time, and use the monitoring tool Windows Task Manager to obtain the CPU load and memory usage rate.

[0079] S2. Based on the network usage, use an intelligent algorithm to adjust the PCIE interface priority and generate a dynamic priority list.

[0080] Use an intelligent algorithm with a linear regression model as the core to calculate the CPU load and memory usage rate of each PCIE interface. The expression is:

[0081] C i =α 1 X 1i +α 2 X 2i +…+α n X ni +b;

[0082] M i =β 1 Y 1i +β 2 Y 2i +…+β m Y mi +c;

[0083] Where, C i represents the CPU load of the i-th PCIE interface, X ni represents the n-th characteristic variable of the i-th PCIE interface, α n represents the weight corresponding to the CPU load, b represents the bias term of the CPU load, M i represents the memory usage rate of the i-th PCIE interface, Y mi represents the m-th characteristic variable of the i-th PCIE interface, βm w represents the weight corresponding to the memory usage rate, c represents the bias term of the memory usage rate, i represents the index variable of the PCIE interface, n represents the characteristic variable of the CPU load, and m represents the characteristic variable of the memory usage rate;

[0084] Adjust the PCIE interface priority according to the calculated CPU load and memory usage rate, and calculate the priority score P of each PCIE interface. The expression is:

[0085]

[0086] Among them, P represents the priority score of the PCIE interface, C i represents the current CPU occupancy rate of the i-th PCIE interface, B i represents the transmission capacity of the i-th PCIE interface, L i represents the data transmission delay of the i-th PCIE interface, log represents the logarithmic function, and i represents the index variable of the PCIE interface;

[0087] Sort in descending order based on the priority score P;

[0088] Set the priority threshold Q, set the high priority threshold to 0.75, and the low priority threshold to 0.25. The specific values can be adjusted according to the actual situation;

[0089] When the priority score P > Q, it means that the PCIE interface has a high priority when processing tasks;

[0090] When the priority score P ≤ Q, it means that the PCIE interface has a low priority when processing tasks;

[0091] Use the built-in function sorted in Python to sort all PCIE interfaces in descending order to generate a dynamic priority list;

[0092] The dynamic priority list refers to the priority score of each PCIE interface and the corresponding task allocation. The interface with a high priority will be assigned to critical tasks, and the interface with a low priority will be assigned to non-critical tasks and used as a backup interface.

[0093] S3. Configure a backup PCIE interface for critical tasks, monitor the level signal of the PCIE interface. When the level signal of the PCIE interface is continuously high or continuously low, it is determined that the PCIE interface has a fault, quickly switch to the backup interface, dispatch personnel for repair, re-evaluate the status of the PCIE interface after repair, and switch back to the original PCIE interface when the stability of the level signal is normal.

[0094] Configure multiple backup PCIE interfaces for critical tasks;

[0095] The critical tasks refer to high-quality video stream transmission (single or multiple video streams transmitted simultaneously) and data synchronization operations (single or multiple data sets synchronized simultaneously);

[0096] Use a high-precision voltage sensor to monitor the level signal of each PCIE interface, calculate the voltage of the level signal, and monitor the voltage change of the PCIE interface;

[0097] Set the high-level threshold to V h , and the low-level threshold to V l , the high-level time threshold to T h , and the low-level time threshold to T l , set the high-level threshold V h to 3.3V, the low-level threshold V l to 0.8V, the high-level time threshold T h to 5s, and the low-level time threshold T l to 5s. The specific values can be adjusted according to the actual situation;

[0098] Compare the calculated level signal voltage with the high-level threshold V h and the low-level threshold V l to judge the fault. The expression is:

[0099]

[0100] Among them, V(t) represents the level signal voltage at time t, V h represents the high-level threshold, V l represents the low-level threshold, N represents the total number of samples within the time window, t h represents the time threshold for continuous high level, T h represents the set high-level time threshold, t l represents the time threshold for continuous low level, T l represents the set low-level time threshold, F represents the fault voltage, t represents time, h represents the index variable of high level, and l represents the index variable of low level;

[0101] When the level signal voltage is higher than V h , and the duration (such as 10 seconds) exceeds the high-level time threshold T h , then it is determined that the PCIE interface is faulty;

[0102] When the level signal voltage is lower than V l , and the duration (such as 10 seconds) exceeds the low-level time threshold T l , then it is determined that the PCIE interface is faulty;

[0103] When the voltage of the level signal is not continuously high and not continuously low, it is determined that the PCIE interface is normal;

[0104] When the fault condition is met, the switching device is immediately triggered to quickly transfer the network communication task to the standby PCIE interface;

[0105] After the PCIE interface is switched to the standby interface, a maintenance work order for the physical location of the faulty interface and the abnormal waveform of the level signal is generated through the enterprise asset management EAM, and the maintenance work order is pushed to the handheld terminal of the maintenance personnel through the positioning of the Internet of Things device. At the same time, the physical slot of the faulty interface is locked;

[0106] Dispatch maintenance personnel to the site for repair. After the maintenance personnel arrive at the site, perform the repair work, replace the damaged components and adjust the configuration;

[0107] After the repair is completed, inject a gradient pressure load into the repaired PCIE interface through a programmable flow generator, pressurize in stages, synchronously collect the voltage of the level signal, and re-evaluate the status of the PCIE interface;

[0108] The status of the PCIE interface refers to faulty and normal;

[0109] Recalculate the voltage of the level signal. When V l ≤ level signal voltage ≤ V h , the stability of the level signal is normal, and it can be switched back to the original PCIE interface. When the level signal voltage > V h and the level signal voltage < V l , it is not standard, and the repair continues.

[0110] S4. Construct a traffic prediction model through a machine learning algorithm, train the traffic prediction model with historical network traffic and the level signal of the PCIE interface, and output the traffic prediction value.

[0111] Use a high-precision voltage sensor to collect the level signal of the PCIE interface, use a network splitter TAP to collect the original network traffic data, and apply Kalman filtering to eliminate electromagnetic interference EMI from the collected level signal;

[0112] Use moving average to extract the timing features in the level signal, use Pandas to divide the original network traffic data into time windows to generate statistical features;

[0113] Use an autoencoder Autoencoder to fuse the timing features and statistical features to generate a feature dataset;

[0114] Use the LSTM model as the basic framework of the network traffic prediction model, and define the basic framework of the network traffic prediction model as the input layer, LSTM layer, fully connected layer and output layer;

[0115] Divide the feature dataset into a training set and a validation set; (The division ratio of the feature dataset is 80%:20%)

[0116] The training set continuously adjusts the model parameters by using the mean squared error (MSE) as the loss function, enabling the training set to learn from the input features how to accurately predict the output network traffic. The validation set is used to evaluate the performance of the model on unseen data, helping to adjust the model hyperparameters and prevent overfitting;

[0117] The input layer inputs the training set for training;

[0118] The LSTM layer receives the training set from the input layer, captures the hidden state vectors in the feature dataset through the gating mechanism, and passes them to the fully connected layer;

[0119] The LSTM layer processes the time series data in the feature dataset through the gating mechanisms of its internal input gate, forget gate, and output gate, thereby effectively capturing the long-term dependencies in the time series data. At each time step, the LSTM cell updates its internal cell state and hidden state. Finally, the hidden state vector at the last time step is passed to the fully connected layer for further processing;

[0120] The fully connected layer receives the hidden state vectors, performs a linear transformation on the hidden state vectors, and outputs the prediction vectors;

[0121] The output layer outputs the prediction vectors as the final predicted traffic values;

[0122] Use the mean squared error (MSE) and combine it with an adaptive adjustment factor to calculate the difference between the predicted traffic value and the actual traffic value. The expression is:

[0123]

[0124] L represents the difference between the predicted traffic value and the actual traffic value, N j represents the total number of feature datasets, w represents the adaptive adjustment factor, y j represents the actual network traffic value of the jth feature dataset, f(x j ) represents the traffic prediction value of the jth feature dataset, x j represents the input feature dataset, (y j -f(x j )) 2 represents the squared error between the actual traffic value and the traffic prediction value, and j represents the index variable of the feature dataset;

[0125] Calculate the gradient through the backpropagation algorithm, adjust the weights and biases of the traffic prediction model, and use the Adam optimizer to update the traffic prediction model parameters;

[0126] Use the validation set to calculate the mean absolute error, and adjust the hyperparameters of the traffic prediction model according to the mean absolute error index.

[0127] S5. According to the traffic prediction value, judge the traffic demand status, and set the PCIE interface status based on the traffic demand status.

[0128] Set the peak threshold as R according to the traffic prediction value 1 , and the low threshold as R 2 , and the expression is:

[0129]

[0130] Among them, R 1 represents the peak threshold, R 2 represents the low threshold, N θ represents the total number of actual network traffic values, y represents the actual network traffic value, and σ y represents the standard deviation of the actual network traffic value, k 1 represents the adjustment coefficient of the peak threshold, k 2 represents the adjustment coefficient of the low threshold, and θ represents the index variable of the actual network traffic value;

[0131] The value range of the adjustment coefficient of the peak threshold is [1.5, 3.0];

[0132] The value range of the adjustment coefficient of the low threshold is [1.5, 3.0];

[0133] The specific value is adjusted according to the actual situation;

[0134] Compare the traffic prediction value with the peak threshold and the low threshold to judge the traffic demand status. The formula is:

[0135]

[0136] Among them, S represents the traffic demand, represents the traffic prediction value;

[0137] The traffic demand status refers to the traffic peak period, the traffic low period, and the traffic normal period;

[0138] When the predicted traffic value > R 1 , it is determined as the traffic peak period. When the predicted traffic value < R 2 , it is determined as the traffic low period. When R 2 ≤ the predicted traffic value ≤ R 1 , it is regarded as the normal traffic period;

[0139] Adjust the status of the priority PCIe interface based on the traffic demand status. The status of the priority PCIe interface is the connected status and the disconnected status;

[0140] Read the dynamic priority list to identify the currently available high-priority PCIe interfaces and low-priority PCIe interfaces;

[0141] When the PCIe interface is in the peak traffic period, set the high-priority PCIe interface to the connected status;

[0142] When the PCIe interface is in the off-peak traffic period, set the low-priority PCIe interface to the disconnected status;

[0143] When the PCIe interface is in the normal traffic period, keep the existing interface status unchanged.

[0144] This embodiment also provides a network card interface connection device, including: a data collection module, a priority adjustment module, a fault detection module, a prediction module, and a dynamic adjustment module,

[0145] The data collection module is used to collect the level signal of the PCIe interface and the network usage of the network device;

[0146] The priority adjustment module is used to adjust the PCIe interface priority using an intelligent algorithm based on the network usage to generate a dynamic priority list;

[0147] The fault detection module is used to configure a standby PCIe interface for critical tasks, monitor the level signal of the PCIe interface. When the level signal of the PCIe interface is continuously high or continuously low, it is determined that the PCIe interface has a fault, and it quickly switches to the standby interface, dispatches personnel for repair, re-evaluates the status of the PCIe interface after repair, and switches back to the original PCIe interface when the stability of the level signal is normal;

[0148] The prediction module is used to construct a traffic prediction model through a machine learning algorithm, train the traffic prediction model with historical network traffic and the level signal of the PCIe interface, and output a traffic prediction value;

[0149] The dynamic adjustment module is used to judge the traffic demand status according to the traffic prediction value, and set the PCIe interface status based on the traffic demand status.

[0150] This embodiment also provides a computer device, applicable to the case of the network card interface connection method, including: 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 network card interface connection method proposed in the above embodiment.

[0151] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via 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 computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be achieved through WIFI, carrier network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0152] 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 method for connecting a network card interface as 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, magnetic disk or optical disk.

[0153] In summary, the present invention realizes comprehensive monitoring of the current state by collecting the level signals of the PCIE interface and the network usage of network devices, not only provides basic data for subsequent intelligent algorithm adjustment, but also ensures the real-time and accuracy of data. By dynamically adjusting the priority, it realizes intelligent scheduling of the PCIE interface, optimizes resource allocation, and improves the ability to handle sudden high loads. By real-time monitoring of level signals, it quickly switches to the standby interface to ensure the continuity and low latency of critical tasks. Through an accurate repair and evaluation mechanism, it ensures long-term stable operation and reduces downtime.

[0154] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than 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 within the scope of the claims of the present invention.

Claims

1. A network card interface connection method, characterized in that: include, Collect the level signal of the PCIE interface and the network usage of the network device; Adjust PCIE interface priority using intelligent algorithms based on network usage and generate a dynamic priority list; Configure a spare PCIE interface for critical tasks and monitor the level signal of the PCIE interface. When the level signal of the PCIE interface is continuously high or low, it is determined that the PCIE interface is faulty, and the interface is quickly switched to the spare interface. Personnel are dispatched for repair. After the repair, the PCIE interface status is re-evaluated. When the stability of the level signal is normal, switch back to the original PCIE interface. A traffic prediction model is built through machine learning algorithms, and the traffic prediction model is trained with historical network traffic and PCIE interface level signals to output traffic prediction values. According to the traffic prediction value, the traffic demand state is judged, and based on the traffic demand state, the PCIE interface state is set.

2. The network card interface connection method according to claim 1, wherein: The collecting of the level signal of the PCIE interface and the network usage of the network device is specifically, Start the server, scan all available PCIE interfaces, and collect the level signals of the PCIE interfaces and the network usage of network devices; Use high-precision voltage sensors to monitor the level signal of each PCIE interface to obtain the stability of the level signal, and use software tools to collect network usage information, which refers to network traffic, CPU load, and memory usage; Use the network tap TAP to collect network traffic in real time, and use the monitoring tool Windows Task Manager to obtain CPU load and memory usage.

3. The network card interface connection method according to claim 2, wherein: The PCIE interface priority is adjusted by using an intelligent algorithm based on network usage to generate a dynamic priority list, specifically, An intelligent algorithm based on the linear regression model is used to calculate the CPU load and memory usage of each PCIE interface. Adjust the PCIE interface priority according to the calculated CPU load and memory usage and calculate the priority score P of each PCIE interface; Sort by priority score P from high to low; Set the priority threshold Q. When the priority score P>Q, it means that the PCIE interface has a high priority when processing tasks. When the priority score P≤Q, it means that the PCIE interface has a low priority when processing tasks; Use the built-in function sorted in Python to sort all PCIE interfaces from high to low to generate a dynamic priority list; The dynamic priority list refers to the priority score of each PCIE interface and the corresponding task allocation. The interface with high priority will be allocated to the critical task, and the interface with low priority will be allocated to the non-critical task and used as a backup interface.

4. The network card interface connection method according to claim 3, characterized in that: The method configures a spare PCIE interface for a critical task, monitors the level signal of the PCIE interface, determines that the PCIE interface fails when the level signal of the PCIE interface is continuously high or low, quickly switches to the spare interface, dispatches personnel for repair, re-evaluates the PCIE interface status after repair, and switches back to the original PCIE interface when the stability of the level signal is normal. Specifically, Configure multiple spare PCIE interfaces for critical tasks; The key tasks mentioned are high-quality video streaming and data synchronization operations; Use a high-precision voltage sensor to monitor the level signal of each PCIE interface, calculate the level signal voltage, and monitor the voltage change of the PCIE interface; Set the high level threshold to V h , the low level threshold is V l , the high level time threshold is T h , the low level time threshold is T l , The calculated level signal voltage is compared with the high level threshold V h and low level threshold V l Make comparisons and determine faults; When the level signal voltage is higher than V h , and the duration exceeds the high level time threshold T h When the PCIE interface is faulty, When the level signal voltage is lower than V l , and the duration exceeds the low level time threshold T l When the PCIE interface is faulty, When the level signal voltage is not continuously high or low, it is determined that the PCIE interface is normal; When the fault condition is met, the switching device is immediately triggered to quickly transfer the network communication task to the standby PCIE interface; After the PCIE interface switches to the backup interface, a maintenance work order with the physical location of the faulty interface and the abnormal waveform of the level signal is generated through the enterprise asset management EAM, and the maintenance work order is pushed to the maintenance personnel's handheld terminal through the IoT device positioning, and the physical slot of the faulty interface is locked at the same time; Dispatch maintenance personnel to the site for repair. After arriving at the site, the maintenance personnel will perform repair work, replace damaged components and adjust the configuration; After the repair is completed, a gradient pressure load is injected into the repaired PCIE interface through a programmable flow generator, pressurized in stages, and the voltage of the level signal is synchronously collected to re-evaluate the PCIE interface status; The PCIE interface status refers to fault and normal; Recalculate the level signal voltage, when V l ≤Level signal voltage≤V h When the level signal is stable, you can switch back to the original PCIE interface. h and level signal voltage <V l If it is not standard, continue repairing.

5. The network card interface connection method according to claim 4, characterized in that: The traffic prediction model is constructed by machine learning algorithm, specifically, Use high-precision voltage sensors to collect the level signals of the PCIE interface, use the network splitter TAP to collect the original network traffic data, and apply Kalman filtering to eliminate electromagnetic interference EMI from the collected level signals; Use moving average to extract time series features from level signals, use Pandas to divide the original network traffic data into time windows, and generate statistical features; Use the autoencoder to fuse the time series features and statistical features to generate a feature dataset; The LSTM model is used as the basic framework of the network traffic prediction model, and the basic framework of the network traffic prediction model is defined as the input layer, LSTM layer, fully connected layer and output layer.

6. The network card interface connection method according to claim 5, characterized in that: The flow prediction model is trained by using the historical network flow and the level signal of the PCIE interface to output the flow prediction value, specifically, Divide the feature dataset into training set and validation set; The input layer inputs the training set for training; The LSTM layer receives the training set from the input layer, captures the hidden state vector in the feature data set through the gating mechanism, and passes it to the fully connected layer; The fully connected layer receives the hidden state vector, performs a linear transformation on the hidden state vector, and outputs a prediction vector; The output layer outputs the prediction vector as the final predicted flow value; The mean square error (MSE) is used in combination with the adaptive adjustment factor to calculate the difference between the predicted flow value and the actual flow value. The gradient is calculated through the back-propagation algorithm, the weight and bias of the traffic prediction model are adjusted, and the parameters of the traffic prediction model are updated using the Adam optimizer; Use the validation set to calculate the mean absolute error, and adjust the traffic prediction model hyperparameters based on the mean absolute error indicator.

7. The network card interface connection method according to claim 6, characterized in that: The method of determining the traffic demand state according to the traffic prediction value and setting the PCIE interface state based on the traffic demand state is specifically as follows: According to the predicted traffic value, set the peak threshold as R1 and the trough threshold as R2. Compare the predicted traffic value with the peak threshold and the trough threshold to determine the traffic demand status; When the predicted traffic value > R1, it is determined as the traffic peak period. When the predicted traffic value < R2, it is determined as the traffic trough period. When R2 ≤ predicted traffic value ≤ R1, it is regarded as the normal traffic period; The traffic demand status refers to the traffic peak period, the traffic trough period, and the traffic normal period; Based on the traffic demand status, adjust the status of the priority PCIE interface; The status of the priority PCIE interface is the connected state and the disconnected state; Read the dynamic priority list to identify the currently available high-priority PCIE interfaces and low-priority PCIE interfaces; When the PCIE interface is in the traffic peak period, set the high-priority PCIE interface to the connected state; When the PCIE interface is in the traffic trough period, set the low-priority PCIE interface to the disconnected state; When the PCIE interface is in the normal traffic period, keep the existing interface status unchanged.

8. A network card interface connection device, based on the network card interface connection method according to any one of claims 1 to 7, characterized in that: It includes a data collection module, a priority adjustment module, a fault detection module, a prediction module, and a dynamic adjustment module. The data collection module is used to collect the level signals of the PCIE interface and the network usage of the network device; The priority adjustment module is used to adjust the PCIE interface priority using an intelligent algorithm based on the network usage and generate a dynamic priority list; The fault detection module is used to configure a standby PCIE interface for critical tasks, monitor the level signals of the PCIE interface. When the level signals of the PCIE interface are continuously high or continuously low, it is determined that the PCIE interface has a fault, quickly switch to the standby interface, dispatch personnel for repair, re-evaluate the status of the PCIE interface after repair, and switch back to the original PCIE interface when the stability of the level signal is normal; The prediction module is used to construct a traffic prediction model through a machine learning algorithm, train the traffic prediction model with historical network traffic and the level signals of the PCIE interface, and output the predicted traffic value; The dynamic adjustment module is used to determine the traffic demand status according to the predicted traffic value and set the status of the PCIE interface based on the traffic demand status.

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, it implements the steps of the network card interface connection method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the network card interface connection method according to any one of claims 1 to 7.

Images