A distributed network switch networking method
By monitoring and optimizing the operating status of low-quality switches, the problem of low-quality switches in the networking area affecting networking effect is solved, and the effect of improving communication task completion and networking stability is achieved.
Patent Information
- Application Number
- CN202410249141.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-05
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-03-05
AI Technical Summary
In the networking area, the existence of low-quality switches will affect the overall networking effect, resulting in poor data transmission speed and low completion rate, affecting the quality of network communications.
By monitoring the switch operating status, low-quality switches are selected, optimized, replaced and maintained according to the distribution density and completion thresholds, and used training models to predict and optimize the switch operating status and maintenance solutions to improve maintenance efficiency.
It improves the working state optimization and maintenance efficiency of low-quality switches, ensures the completion of communication tasks in the networking area, predicts and judges the networking effect, and ensures the stability and efficiency of the networking effect.
Smart Images

Figure CN118233333B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of switch networking, and in particular to a distributed network switch networking method. Background Art
[0002] A switch is a device that completes the information exchange function in a communication system. It regenerates information and forwards it to a designated port after internal processing. It has automatic addressing capabilities and switching functions. In computer network systems, switches are introduced to address the weaknesses of the shared working mode. All ports of the switch are connected to this back bus. When the control circuit receives a data packet, the processing port will look up the address comparison table in the memory to determine which port the NIC (network card) of the destination MAC (hardware address of the network card) is connected to, and quickly transmit the data packet to the destination port through the internal switching matrix. If the destination MAC does not exist, the switch will broadcast to all ports. After the receiving port responds, the switch will "learn" the new address and add it to the internal address table.
[0003] In the application publication number CN110943939A, a network card unit, a distributed network switch architecture and a networking method are disclosed. The network card unit includes a network card device, and the network card device includes a processor, a memory, a power management module and a communication interface; the communication interface includes a PCIE interface and an Ethernet port, wherein the PCIE interface is used to connect to a server, and the Ethernet port includes port one, port two, port three and port four; the processor is respectively connected to the memory, the power management module and the communication interface; the four Ethernet ports on the network card device are used for the network card devices to interconnect and form a switching network. According to the structural form of the network card unit, the present invention realizes a distributed network switch architecture and a networking method.
[0004] In combination with the above applications and prior art, when a higher network communication quality is required in the networking area, several network switches need to be networked to ensure the execution effect of the communication task. However, if some network switches are in poor operating condition, when they are used for networking and used as nodes, these low-quality switches have poor data transmission speeds and low data transmission completion rates, which may interfere with the overall networking effect and affect the effect of switch networking.
[0005] To this end, the present invention provides a distributed network switch networking method. Summary of the invention
[0006] 1. Technical problems solved:
[0007] In view of the deficiencies of the prior art, the present invention provides a distributed network switch networking method, which generates the completion degree of the communication task from the task execution data. If the completion degree obtained exceeds the completion degree threshold, the low-quality switches are replaced in sequence according to the replacement priority. If the completion degree obtained after the replacement is lower than the completion degree threshold, the spare switches are enabled in the networking area, the low-quality switches are fault-detected and matched with corresponding maintenance plans, the maintenance plans are analyzed, the maintenance degree of the low-quality switches is constructed based on the analysis results, and the maintenance plans are screened based on the maintenance degree. The low-quality switches are maintained, the fault characteristics are given through fault detection, and the maintenance plans are quickly given for the switches to be maintained, which can improve the maintenance efficiency, thereby solving the technical problems raised in the background technology.
[0008] (II) Technical solution:
[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: A distributed network switch networking method, including monitoring the operating status of switches in the networking area, and constructing the operating status coefficient of the switch based on the monitoring data. ,If the distribution density of the selected low-quality switches exceeds the density threshold, an optimization instruction is issued;
[0010] The optimized switch performs the communication task in the networking area, and the task execution data generates the completion degree of the communication task. , if the completion degree obtained When the completion threshold is exceeded, a replacement instruction is issued;
[0011] Get replacement priority , according to the replacement priority Replace the low-quality switches one by one. If the completion degree after replacement is If the completion rate is lower than the threshold, the standby switch is activated in the networking area and a maintenance instruction is issued;
[0012] Detect faults of low-quality switches and match corresponding maintenance plans, analyze the maintenance plans, and build the maintenance level of low-quality switches based on the analysis results. , according to the degree of maintenance The maintenance plans are screened and the screened maintenance plans are executed.
[0013] Furthermore, a distribution map of switches is constructed based on the distribution status of switches in the networking area, the working status of switches is monitored at each monitoring node, and the acquired monitoring data is aggregated to construct a working status data set; the operating status coefficient of the switch is generated from the working status set of the switch. , if the operating state coefficient If the working status threshold is not exceeded, it will be marked as a low-quality switch.
[0014] Furthermore, the location of each low-quality switch is marked on the electronic map, and the corresponding location information is obtained to calculate the distribution density of the low-quality switches. If the distribution density If the density threshold is exceeded, an optimization instruction needs to be issued.
[0015] Furthermore, the trained optimization model is obtained to improve the operating state of the low-quality switch as the optimization goal, the current operating parameters of the low-quality switch are optimized, and the optimized operating parameters are obtained; the optimized operating parameters are executed for the low-quality switch, and the communication task execution data in the networking area is collected at each observation node, and the task execution data set is constructed after aggregation.
[0016] Further, after receiving the replacement instruction, the location information of several low-quality switches is obtained, and the proximity center of the low-quality switches in the networking area is calculated. , and obtain the operating status coefficient of each optimized low-quality switch .
[0017] Furthermore, the replacement priority of each low-quality switch is calculated , according to the replacement priority obtained Sort the low-quality switches from high to low to obtain the replacement order; replace the low-quality switches in sequence according to the replacement order until the completion rate in the networking area is Exceeded completion threshold.
[0018] Furthermore, the switch communication model is obtained by training with sample data; the communication task used for testing is used as input, and the trained switch communication model is used to predict the communication status of the switches in the networking area, and the degree of completion is predicted. ; If the prediction gets the completion degree If the completion rate is lower than the threshold, the standby switches in the networking area are enabled one by one until the completion rate reaches the threshold. The completion threshold is exceeded and maintenance instructions are issued for low-quality switches.
[0019] Furthermore, after receiving the maintenance instruction, the low-quality switch is subjected to fault detection to obtain the corresponding fault characteristics; a number of maintenance plans corresponding to the fault characteristics are obtained, the maintenance plans are summarized, and a maintenance plan library is pre-built; and the corresponding maintenance plans are matched for the low-quality switch based on the correspondence between the fault characteristics and the maintenance plans.
[0020] Furthermore, the switch test model is trained and obtained, and the maintenance plan is used as input. The trained switch test model is used to predict the maintenance process of the switch, and after several maintenances, the operating status coefficient after maintenance is obtained. , and then build a maintenance data set after aggregation.
[0021] Furthermore, the maintenance degree of low-quality switches is constructed from the maintenance data set. , to obtain the maintenance degree The highest maintenance plan is used as the target plan, and the next highest is used as the backup plan; the target plan is executed to maintain the low-quality switches.
[0022] (III) Beneficial effects:
[0023] The present invention provides a distributed network switch networking method, which has the following beneficial effects:
[0024] 1. Optimize the working status of low-quality switches, and build the completion degree of current communication tasks based on the communication tasks within the networking area. ; Based on the completion of acquisition Evaluate and judge the ability of the optimized low-quality switches to complete communication tasks. If the switches in the networking area still cannot complete the current communication tasks after optimization, further processing of the low-quality switches is required to ensure the networking effect when the switches are networked.
[0025] 2. Build a communication model of switches in the networking area to predict and judge the networking effect of the switches. If it is still difficult to achieve the expected ability to perform communication tasks, pre-install backup switches in the networking area to improve the current networking status. If the expected effect is achieved, the current networking is completed. Otherwise, it is necessary to maintain some low-quality switches that affect the networking status, thereby ensuring the networking effect of the switches.
[0026] 3. If enabling the backup switch still cannot improve the networking effect, the low-quality switch needs to be maintained. By giving the fault characteristics through fault detection and pre-maintaining the solution library, a maintenance solution can be quickly given for the switch to be maintained, which can improve maintenance efficiency during maintenance.
[0027] 4. Use the trained switch test model to test each maintenance plan and build the corresponding operation status coefficient based on the test data , after several tests, build the maintenance level of low-quality switches , according to the degree of maintenance The maintenance plans are screened, and the target plans and backup plans are determined. When maintenance of low-quality switches is required, the best maintenance plan can be given. When it is used to maintain low-quality switches, the maintenance effect and efficiency can be improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 The figure is a flow chart of the distributed network switch networking method of the present invention. DETAILED DESCRIPTION
[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0030] See also Figure 1 The present invention provides a distributed network switch networking method, comprising:
[0031] Step 1: Monitor the operating status of the switches in the networking area and construct the switch operating status coefficient based on the monitoring data. , and use it to filter out low-quality switches from the networking area. If the distribution density exceeds the density threshold, an optimization instruction is issued;
[0032] The step 1 includes the following contents:
[0033] Step 101: construct a distribution map of switches according to the distribution status of switches in the networking area, and mark the position of each switch on the distribution map; set a monitoring period and include a number of equally spaced monitoring nodes in the monitoring period, monitor the working status of the switch at each monitoring node, and aggregate the acquired monitoring data to construct a working status data set;
[0034] Step 102: Generate the switch operation state coefficient from the switch operation state set , the method is as follows: collect the port throughput and transmission speed Perform linear normalization and map the corresponding data values to the interval According to the following formula:
[0035] ;
[0036] The meaning of the parameters are: , is the number of monitoring nodes, weight coefficient: , ,and , whose specific value is set by the user. is the mean port throughput, is the mean transmission speed; is the qualified standard value of port throughput; is the qualified standard value of transmission speed; as a supplementary explanation, is the value of port throughput on monitoring node i, is the value of transmission speed at monitoring node i;
[0037] Based on historical data and the expected use and management of the switch, the working status threshold of the switch is pre-set; if the operating status coefficient If it does not exceed the working status threshold, it means that the working status of the current switch is poor and it may not play the expected role when used. Therefore, it is marked as a low-quality switch;
[0038] Step 103: After filtering out the low-quality switches, mark the locations of the low-quality switches on the electronic map, obtain the corresponding location information, and calculate the distribution density of the low-quality switches. ;
[0039] ;
[0040] Where n is the number of low-quality switches in the networking area, is the distance from low-quality switch i to low-quality switch j, and r is the maximum diameter of the networking area;
[0041] Based on the expected operation and management of low-quality switches, the density threshold is pre-set. If the density threshold is exceeded, an optimization instruction needs to be issued;
[0042] When using, combine the contents in steps 101 to 103:
[0043] When the switch is in the networking area and in operation, the operation status coefficient is constructed by monitoring its operation status data. , so that the operating state coefficient The operating status of the switch is evaluated, so as to screen out low-quality switches from a number of switches. When there are many low-quality switches, the low-quality switches can be optimized in a targeted manner, so as to improve the working status of the switch.
[0044] In combination with the above applications and prior art, when a higher network communication quality is required in the networking area, several network switches need to be networked. If some network switches are in poor operating condition, they will be used for networking. When they are used as nodes, there may be certain interference with the networking effect, affecting the networking effect.
[0045] Step 2: Optimize the operation status of the low-quality switch, combine it with the remaining switches, perform the communication task in the networking area, and generate the completion degree of the communication task from the task execution data. If the obtained completion degree exceeds the completion degree threshold, issue a replacement instruction to the outside;
[0046] The step 2 includes the following contents:
[0047] Step 201: Use the ant colony algorithm to build an initial model, and use sample data to train the initial model to obtain a trained optimization model; take improving the operating state of the low-quality switch as the optimization goal, optimize the current operating parameters of the low-quality switch, and obtain the optimized operating parameters;
[0048] Step 202: Optimize the operating parameters for the low-quality switches and set an observation period including a number of equally spaced observation nodes. After setting the communication task, collect the communication task execution data in the networking area at each observation node, and then aggregate and construct a task execution data set.
[0049] Step 203: Generate the degree of completion of the communication task from the execution data set , as follows: After assigning the communication tasks, obtain the transmission completion rate on each observation node And transmission accuracy , after linear normalization of both, the corresponding data values are mapped to the interval According to the following formula:
[0050] ;
[0051] Among them, the weight coefficient is: , ,and , is the number of observation nodes, is the mean value of the transmission completion rate, is the qualified standard value of the transmission completion rate, is the transmission completion rate on the i-th observation node; is the mean value of transmission accuracy, is the qualified standard value of transmission accuracy, is the transmission accuracy rate on the i-th observation node;
[0052] Based on historical data and expectations for the execution of communication tasks, a completion threshold is pre-set. If the completion threshold is exceeded, it means that after optimizing the low-quality switches in the networking area, optimization alone is not enough to perform the communication tasks in the networking area. Therefore, some switches need to be replaced and replacement instructions need to be issued.
[0053] When using, combine the contents in steps 201 to 203:
[0054] After obtaining the various operating parameters and operating indicators of the switch, use the trained optimization model to optimize it, obtain the corresponding optimization plan, and after executing the optimization plan, optimize the working state of the low-quality switch, and build the completion degree of the current communication task based on the communication tasks in the networking area. ;
[0055] Based on the degree of completion obtained Evaluate and judge the ability of the optimized low-quality switches to complete communication tasks. If the switches in the networking area still cannot complete the current communication tasks after optimization, then further processing of the low-quality switches is required, thereby ensuring the networking effect when the switches are networked.
[0056] Step 3: Calculate the replacement priority based on the location and operating status of the low-quality switch , according to the replacement priority Replace the low-quality switches one by one. If the completion degree after replacement is If the completion rate is lower than the threshold, the standby switch is activated in the networking area and a maintenance instruction is issued;
[0057] The step three includes the following contents:
[0058] Step 301: After receiving the replacement instruction, obtain the location information of several low-quality switches and calculate the proximity center of the low-quality switches in the networking area. , and obtain the operating status coefficient of each optimized low-quality switch ; Calculate the replacement priority of each low-quality switch , as follows: close to the centrality And operating status coefficient Do linear normalization and map the corresponding data values to the interval According to the following formula:
[0059] ;
[0060] Weight coefficient: , , i is the serial number of the low-quality switch, is the number of low-quality switches, ; is the mean of several closeness centralities, is the mean of several operating status coefficients.
[0061] Step 302: Based on the obtained replacement priority Sort low-quality switches from high to low to obtain a replacement order; replace low-quality switches in sequence according to the replacement order, and test the ability of switches in the networking area to perform communication tasks to obtain the corresponding completion degree Until the completion of the networking area Exceeding completion threshold;
[0062] When in use, the replacement priority is calculated based on the location and operating status of the low-quality switch. When a low-quality switch needs to be replaced, the most important parts can be replaced first, reducing the number of subsequent replacements, improving replacement efficiency, and completing switch networking in a shorter time.
[0063] Step 303: collect data such as specification performance, location data, and communication tasks of switches in the networking area; construct an initial model using a convolutional neural network, extract part of the collected data as sample data; train and test the initial model using the sample data, and obtain a trained switch communication model;
[0064] Taking the communication task for testing as input, the trained switch communication model is used to predict the communication status of switches in the networking area and predict the acquisition completion degree. ; If the prediction gets the completion degree If the completion rate is lower than the threshold, the standby switches in the networking area are enabled one by one until the completion rate reaches the threshold. Exceeding the completion threshold and issuing maintenance instructions for low-quality switches;
[0065] When using, combine the contents in steps 301 to 303:
[0066] After replacing the low-quality switches in the networking area, build a switch communication model in the networking area. After completing the initial switch networking, predict and judge the networking effect of the switches. If it is still difficult to make the ability to perform communication tasks meet expectations, then pre-install backup switches in the networking area to improve the current networking status. If the expected effect is achieved, it means that the current networking is completed. Otherwise, it is necessary to maintain some low-quality switches that affect the networking status, thereby ensuring the networking effect of the switches.
[0067] Step 4: Detect faults on low-quality switches and match corresponding maintenance plans. Use the trained switch test model to analyze the maintenance plans and construct the maintenance level of low-quality switches based on the analysis results. , according to the degree of maintenance Screen the maintenance plans and execute the screened maintenance plans;
[0068] The step 4 includes the following contents:
[0069] Step 401: After receiving the maintenance instruction, perform fault detection on the low-quality switch and obtain corresponding detection data; after setting the abnormal standard, identify the obtained detection data and obtain the corresponding fault characteristics; obtain several maintenance plans corresponding to the fault characteristics through deep search or offline collection, summarize the maintenance plans, and pre-build a maintenance plan library;
[0070] After obtaining the fault features, the trained matching model is used to match the corresponding maintenance plan for the low-quality switch based on the correspondence between the fault features and the maintenance plan.
[0071] When in use, after receiving the maintenance instruction, if enabling the standby switch still cannot improve the networking effect, the low-quality switch needs to be maintained. By giving the fault characteristics through fault detection and pre-maintaining the solution library, a maintenance solution can be quickly given for the switch to be maintained, which can improve the maintenance efficiency during maintenance.
[0072] Step 402: collect the specification performance data and operation status data of the switch, build an initial model with a convolutional neural network, train and test the initial model with the collected data, and obtain a trained switch test model; use the maintenance plan as input, use the trained switch test model to predict the maintenance process of the switch, and after several maintenances, obtain the operation status coefficient after maintenance , build and maintain data sets after aggregation;
[0073] Step 403: Construct the maintenance level of the low-quality switch from the maintenance data set , as follows:
[0074] ;
[0075] in, , To maintain the number of predictions, the weight coefficient is: , and ; is the operating state coefficient of the i-th test, is its mean value; is the qualified standard value of the operating status coefficient;
[0076] To obtain maintenance The highest maintenance plan is used as the target plan, and the next highest is used as the backup plan. The target plan is executed to maintain the low-quality switches.
[0077] When used, combine the contents in steps 401 to 403:
[0078] After several maintenance plans are given, the trained switch test model is used to test each maintenance plan, and the corresponding operating status coefficient is constructed based on the test data. , and then, after several tests, the maintenance level of low-quality switches is built , according to the degree of maintenance The maintenance plans are screened and the target plans and backup plans are determined. Thus, when maintenance of low-quality switches is required, the best maintenance plan can be given. When it is used for maintenance of low-quality switches, the maintenance effect and efficiency can be improved.
[0079] Among them, the weight coefficient can be obtained by referring to the hierarchical analysis method. The hierarchical analysis method is an analysis method that combines qualitative and quantitative methods. It can decompose complex problems into multiple levels. By comparing the importance of factors at each level, it can help decision makers make decisions on complex problems and determine the final decision-making plan. In this process, the hierarchical analysis method can be used to determine the weight coefficients of these indicators.
[0080] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0081] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
[0082] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.
Claims
1. A distributed network switch networking method, characterized in that: include, Monitor the operating status of switches in the networking area, and construct the switch operating status coefficient Yxs based on the monitoring data. If the distribution density of the low-quality switches screened out exceeds the density threshold, issue an optimization instruction; The optimized switch executes the communication task in the networking area, and generates the completion degree Wp of the communication task from the task execution data. If the obtained completion degree Wp exceeds the completion degree threshold, a replacement instruction is issued; Obtain the replacement priority Yd, and replace the low-quality switches in turn according to the replacement priority Yd. If the completion degree Wp obtained after the replacement is lower than the completion degree threshold, enable the spare switch in the networking area and issue a maintenance instruction; Perform fault detection on low-quality switches and match corresponding maintenance plans, analyze the maintenance plans, construct the maintenance degree Pd of the low-quality switches based on the analysis results, screen the maintenance plans based on the maintenance degree Pd, and execute the screened maintenance plans; The operation status coefficient Yxs is calculated as follows: linear normalization is performed on the collected port throughput Dv and transmission speed Cv, and the corresponding data values are mapped to the interval [0,1] according to the following formula: The meaning of the parameters is: i = 1, 2, ..., n, n is the number of monitoring nodes, weight coefficient: 0 ≤ F1 ≤ 1, 0 ≤ F2 ≤ 1, and F2 + F1 = 1, the specific value is adjusted by the user, the Dv avg is the mean of the port throughput, Cv avg is the mean transmission speed; is the qualified standard value of port throughput; It is the qualified standard value of transmission speed; Dv i is the value of port throughput at monitoring node i, Cv i is the value of transmission speed at monitoring node i; The degree of completion Wp is calculated as follows: After assigning the communication tasks, the transmission completion rate Ru and the transmission accuracy Rt on each observation node are obtained. After linear normalization of the two, the corresponding data values are mapped to the interval [0, 1] according to the following formula: Among them, the weight coefficient is: 0≤β≤1, 0≤α≤1, and α+β=1, i=2,…,k, k is the number of observation nodes, Ru avg is the mean value of the transmission completion rate, is the qualified standard value of the transmission completion rate, Ru i is the transmission completion rate on the i-th observation node; Rt avg is the mean value of transmission accuracy, is the qualified standard value of transmission accuracy, Rt i is the transmission accuracy rate on the i-th observation node; The replacement priority Yd is calculated as follows: linearly normalize the proximity center Cu and the operating state coefficient Yxs, and map the corresponding data values to the interval [0,1] according to the following formula: Weight coefficient: 0≤ζ≤1, 0≤χ≤1, i is the serial number of the low-quality switch, p is the number of low-quality switches, i=1, 2, …, p; Cu avg is the mean of several closeness centers, Yxs avg is the mean of several operating state coefficients; The maintenance degree Pd is calculated as follows: Where i = 1, 2, ..., n, n is the number of maintenance predictions, weight coefficient: 0 ≤ F1 ≤ 1, 0 ≤ F2 ≤ 1 and F2 + F1 = 1; Yxs i is the operating state coefficient of the i-th test, the Yxs avg is its mean value; It is the qualified standard value of the operating status coefficient.
2. A distributed network switch networking method according to claim 1, characterized in that: A distribution map of switches is constructed based on the distribution status of switches in the networking area. The working status of switches is monitored at each monitoring node, and the acquired monitoring data is aggregated to construct a working status data set. The operating status coefficient Yxs of the switch is generated from the working status set of the switch. If the operating status coefficient Yxs does not exceed the working status threshold, it is marked as a low-quality switch.
3. A distributed network switch networking method according to claim 2, characterized in that: The location of each low-quality switch is marked on the electronic map, and the corresponding location information is obtained, and the distribution density Mj of the low-quality switches is calculated; if the distribution density Mj exceeds the density threshold, an optimization instruction needs to be issued.
4. A distributed network switch networking method according to claim 1, characterized in that: Obtain the trained optimization model, take improving the operating status of the low-quality switch as the optimization goal, optimize the current operating parameters of the low-quality switch, and obtain the optimized operating parameters; execute the optimized operating parameters for the low-quality switch, collect communication task execution data in the networking area at each observation node, and construct a task execution data set after aggregation.
5. A distributed network switch networking method according to claim 1, characterized in that: After receiving the replacement instruction, the location information of several low-quality switches is obtained, the proximity center Cu of the low-quality switches in the networking area is calculated, and the operating state coefficient Yxs of each optimized low-quality switch is obtained.
6. A distributed network switch networking method according to claim 5, characterized in that: Calculate the replacement priority Yd of each low-quality switch, sort the low-quality switches from high to low according to the obtained replacement priority Yd, and obtain the replacement order; replace the low-quality switches in sequence according to the replacement order until the completion degree Wp in the networking area exceeds the completion degree threshold.
7. A distributed network switch networking method according to claim 6, characterized in that: The switch communication model is obtained by training with sample data. The communication task used for testing is used as input, and the trained switch communication model is used to predict the communication status of the switches in the networking area, and the completion degree Wp is predicted and obtained. If the predicted completion degree Wp is lower than the completion degree threshold, the spare switches are enabled one by one in the networking area until the completion degree Wp exceeds the completion degree threshold, and maintenance instructions are issued to the low-quality switches.
8. A distributed network switch networking method according to claim 1, characterized in that: After receiving the maintenance instruction, the low-quality switch is fault-detected to obtain the corresponding fault characteristics; Obtain several maintenance plans corresponding to the fault characteristics, summarize the maintenance plans, and pre-build a maintenance plan library; match the corresponding maintenance plan for the low-quality switch based on the correspondence between the fault characteristics and the maintenance plan.
9. A distributed network switch networking method according to claim 8, characterized in that: The switch test model is trained and obtained, and the maintenance plan is used as input. The trained switch test model is used to predict the maintenance process of the switch. After several maintenances, the post-maintenance operating status coefficient Yxs is obtained and summarized to construct a maintenance data set.
10. A distributed network switch networking method according to claim 9, characterized in that: The maintenance degree Pd of the low-quality switch is constructed from the maintenance data set to obtain the maintenance plan with the highest maintenance degree Pd as the target plan, and the next one as the backup plan; the target plan is executed to maintain the low-quality switch.
