Train communication quality prediction method and train

Abstract

The invention provides a train communication quality prediction method and a train, which can be applied to the technical field of network monitoring. The train communication quality prediction method comprises the steps that respective operation parameter sets of target network equipment in a train at multiple moments in a first preset time period are acquired, and the operation parameter set at each moment comprises respective operation parameters of multiple ports of the target network equipment at the moments; projecting the plurality of operation parameter groups into a multi-dimensional vector space to obtain respective operation vectors at a plurality of moments, the dimensions of the multi-dimensional vector space being determined according to the number of ports of the target network device; according to the sequence of the multiple moments in the first preset time period, the multiple operation vectors are clustered in sequence, and respective clustering results of the multiple moments are obtained; and fitting the respective clustering centers of the plurality of abnormal clusters to determine a communication quality prediction result of the target network equipment.

Description

Technical Field

[0001] This disclosure relates to the field of network monitoring technology, specifically to a method for predicting train communication quality and a train. Background Technology

[0002] The train communication network is one of the key core systems of high-speed trains. Its functional role is equivalent to the "central system" and "neural network" of an intelligent train, undertaking the control tasks and data communication of the entire train. Various business data, control commands, equipment status data, diagnostic alarm information, video surveillance, and passenger information transmitted within the train must be transmitted through the train communication network. Therefore, the health of the train communication network is directly related to the safety and reliability of train operation.

[0003] In the process of realizing the present invention, it was discovered that the network monitoring and diagnostic system for trains in the related technology is not stable enough and is prone to misjudging and misinterpreting network fluctuations, resulting in a waste of maintenance costs. Summary of the Invention

[0004] In view of the above problems, this disclosure provides a method for predicting train communication quality and a train.

[0005] According to a first aspect of this disclosure, a method for predicting train communication quality is provided, comprising: acquiring operating parameter sets for multiple moments of a target network device in a train within a first preset time period, wherein each moment's operating parameter set includes operating parameters of multiple ports of the target network device at that moment; projecting the multiple operating parameter sets into a multi-dimensional vector space to obtain operating vectors for each moment, wherein the dimension of the multi-dimensional vector space is determined according to the number of ports of the target network device; clustering the multiple operating vectors sequentially according to the order of the multiple moments in the first preset time period to obtain clustering results for each moment, wherein each clustering result includes normal clusters representing the target network device in a normal state and abnormal clusters representing the target network device in an abnormal state; fitting the cluster centers of the multiple abnormal clusters to determine the prediction result of the communication quality of the target network device.

[0006] According to embodiments of this disclosure, fitting the cluster centers of multiple anomalous clusters to determine the communication quality prediction result for a target network device includes: for each anomalous cluster, fitting the cluster centers of the anomalous cluster at multiple time points to obtain parameter fitting results representing the cluster centers of the anomalous clusters; determining the communication quality prediction result for the port to be tested based on the parameter fitting results and the communication quality thresholds of multiple ports; and determining the communication quality prediction result for the target network device based on the communication quality prediction results of multiple ports.

[0007] According to embodiments of this disclosure, based on parameter fitting results and the communication quality thresholds of multiple ports, the communication quality prediction result of the port under test is determined, including: for each port, if the port's operating parameters reach the port's communication quality threshold based on the port's parameter fitting results, the communication quality detection result of the port under test indicates that the port's communication quality will be abnormal, and the abnormal moment when the operating parameters reach the port's communication quality threshold is determined.

[0008] According to embodiments of this disclosure, determining the communication quality prediction result for a target network device based on the communication quality prediction results of multiple ports includes: when the communication quality prediction results of multiple ports indicate that the target network device has ports with normal communication quality, determining that the port corresponding to the abnormal result has a communication abnormality based on the abnormal result indicating abnormal communication quality among the multiple communication quality prediction results; and when the communication quality prediction results of multiple ports indicate that the target network device does not have ports with normal communication quality, determining that the target network device has a fault.

[0009] According to embodiments of this disclosure, fitting the cluster centers of anomaly clusters at multiple time points to obtain parameter fitting results representing the cluster centers of anomaly clusters includes: determining the components of the cluster centers of anomaly clusters at multiple time points in each dimension; arranging the components in each dimension according to the chronological order of the multiple time points to obtain parameter sequences in each dimension; performing linear fitting on the parameter sequences in each dimension to obtain fitting sub-results in each dimension; and determining the parameter fitting results representing the cluster centers of anomaly clusters based on multiple fitting sub-results.

[0010] According to embodiments of this disclosure, there is a correspondence between multiple ports and multiple abnormal clusters, and the correspondence is determined in the following way: for each abnormal cluster, the cluster center of the abnormal cluster is determined; the projection distance between the cluster center of the abnormal cluster and the cluster center of the normal cluster in multiple dimensions of the multidimensional vector space is determined, wherein each dimension of the multidimensional vector space corresponds to a port; based on the multiple projection distances, at least one port corresponding to the abnormal cluster is determined.

[0011] According to embodiments of this disclosure, determining at least one port corresponding to an abnormal cluster based on multiple projection distances includes: determining at least one abnormal projection distance greater than a preset distance threshold from the multiple projection distances; and determining at least one port corresponding to an abnormal cluster based on the dimension of each of the at least one abnormal projection distance and the correspondence between multiple dimensions of the multidimensional vector space and multiple ports.

[0012] According to embodiments of this disclosure, multiple sets of operating parameters are projected into a multidimensional vector space to obtain operating vectors for multiple time moments, including: for each set of operating parameters at each time moment, based on the correspondence between multiple dimensions of the multidimensional vector space and multiple ports, multiple operating parameters in the set of operating parameters are projected onto the dimensions corresponding to the ports that collect the operating parameters, to obtain the components of the operating parameters in the dimensions; and the operating vector at each time moment is determined based on the components of the operating parameters in the multiple dimensions.

[0013] According to embodiments of this disclosure, multiple operating vectors are clustered sequentially according to the order of multiple moments in a first preset time period to obtain clustering results for each moment. This includes: obtaining operating parameter groups for each moment of the target network device within a second preset time period, wherein the last moment of the second preset time period is the same as the first moment of the first preset time period; for each moment in the first preset time period, selecting a preset number of operating parameter groups prior to the moment as historical operating parameter groups for that moment, wherein the preset number is determined based on the number of moments in the second preset time period; and using a preset clustering algorithm to cluster the historical operating parameter groups and operating vectors of the moment to obtain clustering results for that moment.

[0014] Another aspect of this disclosure provides a train equipped with a controller and a target network device, wherein the controller executes the train communication quality prediction method described above to obtain a communication quality prediction result for the target network device.

[0015] According to embodiments of this disclosure, the operating parameters of multiple ports on a target network device at different times are mapped sequentially into a vector space to obtain operating vectors at different times. These operating vectors can accurately represent the operating status of ports in the target network device, providing detailed data support for clustering operations. Then, the obtained operating vectors are clustered to obtain normal clusters reflecting normal operation of the target network device and abnormal clusters reflecting abnormal operation of the target network device, thus distinguishing between normal and abnormal operating vectors. Next, the cluster centers of each abnormal cluster are fitted sequentially to determine the changes in the cluster centers of each abnormal cluster, thereby determining the overall communication quality trend of the port corresponding to that abnormal cluster. This allows for accurate differentiation between occasional fluctuations in operating parameters and actual faults, improving prediction accuracy. Furthermore, since cluster-based anomaly prediction can identify fluctuation anomalies without triggering network device alarms for every occasional fluctuation, unnecessary maintenance costs can be reduced. Attached Figure Description

[0016] The foregoing contents, as well as other objects, features, and advantages of this disclosure, will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0017] Figure 1 An exemplary system architecture for a train communication quality prediction method according to embodiments of this disclosure is illustrated.

[0018] Figure 2 A flowchart illustrating a train communication quality prediction method according to an embodiment of the present disclosure is shown schematically.

[0019] Figure 3 A data flow diagram of a train communication quality prediction method according to an embodiment of the present disclosure is illustrated.

[0020] Figure 4 A schematic block diagram of a train communication quality prediction device according to an embodiment of the present disclosure is shown.

[0021] Figure 5 A block diagram of an electronic device suitable for implementing a train communication quality prediction method according to an embodiment of the present disclosure is shown schematically. Detailed Implementation

[0022] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0023] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0024] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0025] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0026] With the rapid development of train communication networks, they have begun to undertake the control tasks and data communication of the entire vehicle. Various business data, control commands, equipment status data, diagnostic alarm information, video surveillance and passenger information transmitted inside the vehicle must be transmitted through the train communication network.

[0027] However, the common methods for trend analysis and prediction of network communication quality in related technologies are to construct regression analysis models using communication quality parameters at different times. These regression analysis models analyze and predict abnormal trends in network communication quality parameters entirely through linear fitting curves.

[0028] The above analysis method has the following shortcomings: due to network fluctuations and other issues, switches may experience occasional port communication quality degradation even when the port link is normal. Existing methods usually issue an alarm after detecting a degradation in port communication quality, resulting in unnecessary maintenance costs.

[0029] In view of this, this disclosure provides a method for predicting train communication quality, comprising: acquiring operating parameter sets for multiple times of a target network device in a train within a first preset time period, wherein each operating parameter set includes operating parameters of multiple ports of the target network device at each time; projecting the multiple operating parameter sets into a multi-dimensional vector space to obtain operating vectors for each of the multiple times, wherein the dimension of the multi-dimensional vector space is determined according to the number of ports of the target network device; clustering the multiple operating vectors sequentially according to the order of the multiple times in the first preset time period to obtain clustering results for each of the multiple times, wherein each clustering result includes normal clusters representing the target network device in a normal state and abnormal clusters representing the target network device in an abnormal state; fitting the cluster centers of the multiple abnormal clusters to determine the prediction result of the communication quality of the target network device.

[0030] Figure 1 An exemplary system architecture for a train communication quality prediction method according to embodiments of this disclosure is illustrated.

[0031] It is important to note that Figure 1 The examples shown are merely examples of application scenarios that can be applied to the embodiments of this disclosure, in order to help those skilled in the art understand the technical content of this disclosure, but do not mean that the embodiments of this disclosure cannot be used in other devices, systems, environments or scenarios.

[0032] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a train 101, a target network device 102, and a controller 103.

[0033] Train 101 can be any type of train equipped with network devices, such as a high-speed train. The target network device 102 and controller 103 are installed on the train. The target network device 102 can be an Ethernet switch or other device used to provide communication link ports.

[0034] The controller 103 can be used to acquire the operating parameters of the target network device 102 in order to obtain the communication quality prediction results of the target network device. The controller 103 can be various electronic devices with a display screen and support web browsing, including but not limited to tablet computers, laptops, and desktop computers.

[0035] It should be noted that the train communication quality prediction method provided in this embodiment can generally be executed by the controller 103. Correspondingly, the train communication quality prediction device provided in this embodiment can generally be located in the controller 103. The train communication quality prediction method provided in this embodiment can also be executed by the controller 103, or it can be executed by other terminal devices different from the controller 103. Correspondingly, the train communication quality prediction device provided in this embodiment can also be located in the controller 103, or it can be located in other terminal devices different from the controller 103.

[0036] It should be understood that Figure 1 The number of trains, target network devices, and controllers shown is merely illustrative. Any number of trains, target network devices, and controllers can be used depending on implementation needs.

[0037] Figure 2 A flowchart illustrating a train communication quality prediction method according to an embodiment of the present disclosure is shown schematically.

[0038] like Figure 2 As shown, the train communication quality prediction method 200 of this embodiment includes operations S210 to S240.

[0039] In operation S210, the operating parameter groups of the target network device in the train are obtained at multiple times within a first preset time period. Each time period's operating parameter group includes the operating parameters of multiple ports of the target network device at each time period.

[0040] According to embodiments of this disclosure, the target network device in the train may have Each port serves as an independent and single data acquisition point, collecting its operating parameters. The first preset time period includes multiple moments; for each moment, the parameters of each port at that moment can be collected. A set of operating parameters for a network device at a given moment includes the parameters for that moment. Multiple operating parameters for each port.

[0041] The above Each operating parameter can include the number of bytes received at the port, the number of packets lost, etc. When obtaining operating parameters, they can be obtained based on the time step using historical data stored in the target network device.

[0042] In operation S220, multiple sets of operating parameters are projected into a multi-dimensional vector space to obtain operating vectors at multiple time points. The dimensions of the multi-dimensional vector space are determined according to the number of ports of the target network device.

[0043] According to embodiments of this disclosure, the multiple sets of operating parameters obtained above can be converted into a vector form of operating vector. The operating vector includes multiple dimensions, and each dimension corresponds to a port in the target network device. For example, the target network device may have... If there are ports, then the running vector at a certain moment has Each dimension.

[0044] In operation S230, multiple running vectors are clustered sequentially according to the order of multiple moments in the first preset time period to obtain the clustering results for each moment. Each clustering result includes normal clusters representing the target network device in a normal state and abnormal clusters representing the target network device in an abnormal state.

[0045] According to embodiments of this disclosure, in In a multi-dimensional vector space, the multiple running vectors obtained above are clustered sequentially in the vector space based on their corresponding time order, yielding clustering results for each time step. Clustering can be performed using distance-based clustering algorithms, such as k-means clustering.

[0046] Specifically, distance-based clustering algorithms primarily utilize distance functions to measure the similarity between running vectors, and then group similar running vectors together according to certain rules. First, a suitable distance formula, such as Euclidean distance or Manhattan distance, can be defined to calculate the distance between running vectors. Next, a distance-based grouping strategy can be designed, such as setting a distance threshold to group running vectors with a distance less than that threshold into a single cluster. Furthermore, adjustment steps can be introduced, such as merging smaller clusters or optimizing centroids, to make the cluster structure more rational.

[0047] The operating vectors of the target network device during normal communication are relatively concentrated in the multidimensional space, while outliers appear in the operating vectors during abnormal communication. As the clustering algorithm continues to process the operating vectors in chronological order, normal clusters representing the target network device in a normal state and outlier clusters representing the target network device in an abnormal state will emerge in the multidimensional space.

[0048] In operation S240, the cluster centers of each of the multiple anomalous clusters are fitted to determine the communication quality prediction results for the target network device.

[0049] According to embodiments of this disclosure, after obtaining anomalous clusters through clustering, the cluster centers of multiple anomalous clusters can be fitted using linear regression. The fitting process reveals the changing patterns of the cluster centers over time. If the cluster center continuously moves away from the cluster centers of normal clusters, it can be preliminarily determined that communication quality predictions indicate a potential communication anomaly. If the cluster center gradually approaches the cluster centers of normal clusters, it can be determined that the anomalous cluster may be caused by occasional interference factors and does not require maintenance. Because the clustering method is used to obtain the stable state of communication quality parameters, long-term monitoring of port link communication quality parameters is unnecessary. Furthermore, it can perceive the state of communication quality parameters over different time periods, enabling trend analysis and prediction of network communication quality parameters across different time periods.

[0050] According to embodiments of this disclosure, the operating parameters of multiple ports on a target network device at different times are mapped sequentially into a vector space to obtain operating vectors at different times. These operating vectors can accurately represent the operating status of ports in the target network device, providing detailed data support for clustering operations. Then, the obtained operating vectors are clustered to obtain normal clusters reflecting normal operation of the target network device and abnormal clusters reflecting abnormal operation of the target network device, thus distinguishing between normal and abnormal operating vectors. Next, the cluster centers of each abnormal cluster are fitted sequentially to determine the changes in the cluster centers of each abnormal cluster, thereby determining the overall communication quality trend of the port corresponding to that abnormal cluster. This allows for accurate differentiation between occasional fluctuations in operating parameters and actual faults, improving prediction accuracy. Furthermore, since cluster-based anomaly prediction can identify fluctuation anomalies without triggering network device alarms for every occasional fluctuation, unnecessary maintenance costs can be reduced.

[0051] According to embodiments of this disclosure, fitting the cluster centers of multiple anomalous clusters to determine the communication quality prediction result for a target network device includes: for each anomalous cluster, fitting the cluster centers of the anomalous cluster at multiple time points to obtain parameter fitting results representing the cluster centers of the anomalous clusters; determining the communication quality prediction result for the port to be tested based on the parameter fitting results and the communication quality thresholds of multiple ports; and determining the communication quality prediction result for the target network device based on the communication quality prediction results of multiple ports.

[0052] According to embodiments of this disclosure, firstly, the cluster centers of each anomalous cluster at multiple time points are fitted using quadratic or cubic polynomial regression to obtain parameter fitting results for each anomalous cluster, thereby capturing the changing trend of the cluster centers. The parameter fitting results can indicate whether the corresponding anomalous cluster moves away from or closer to the normal cluster over time. Furthermore, since each dimension in the vector space represents a port of a target network device, a communication quality threshold can be set for each port of the target network device. The communication quality threshold can be the distance between the position of the running vector in the vector space when the corresponding port is normal and the position of the running vector in the vector space when the corresponding port is faulty. Then, by comparing the communication quality thresholds for each of the multiple ports with the parameter fitting results obtained above, the communication quality prediction results for the corresponding ports can be obtained. Finally, based on the communication quality prediction results for the multiple ports, the communication quality prediction result for the target network device can be determined.

[0053] According to embodiments of this disclosure, the generation of anomalous clusters is caused by abnormal values ​​in the operating parameters of one or more ports. Therefore, the parameter fitting results of the cluster centers of multiple anomalous clusters can reflect the parameter changes of the ports corresponding to each anomalous cluster. For example, if only the third port of a target network device containing eight ports experiences an anomaly within a sampling period, then the operating vector of the target network device will generate an anomalous cluster due to the anomaly of the third port during this sampling period. Fitting these anomalous clusters will yield the relative changes between the cluster centers of the anomalous clusters caused by the anomaly of the third port and the cluster centers of the normal clusters. Then, the communication quality threshold of the third port can be compared with the relative changes to determine whether the parameters of the third port will reach the communication quality threshold in the future, thereby determining the communication quality prediction result. Similarly, if multiple ports experience anomalies, since the dimensions correspond one-to-one with the ports, the relative changes of multiple ports and the communication quality thresholds of each port can be compared separately and then combined to obtain the communication quality prediction result.

[0054] According to embodiments of this disclosure, the parameter fitting process can use various forms, such as linear fitting, polynomial fitting, exponential fitting, trigonometric function fitting, etc., depending on the data characteristics and trends of the outlier cluster. Linear fitting is suitable for steady changes, polynomial fitting is suitable for non-linear growth or decay, exponential fitting is used for geometric progression changes, and trigonometric function fitting captures periodic fluctuations.

[0055] According to embodiments of this disclosure, for each port, based on the parameter fitting results of the port, if the operating parameters of the port reach the communication quality threshold of the port, the communication quality detection result of the port under test can indicate that the communication quality of the port will be abnormal, and the abnormal moment when the operating parameters reach the communication quality threshold of the port can be determined.

[0056] According to embodiments of this disclosure, the fitted result is compared with the communication quality threshold of the corresponding port. Based on the trend of the fitted change, it can be determined whether the communication quality threshold will be exceeded at some point in the future. Furthermore, the moment when the fitted result exceeds the communication quality threshold can be taken as the port abnormal moment, thus completing the prediction of train communication quality.

[0057] According to embodiments of this disclosure, for example, if the parameter fitting result is obtained based on linear fitting, then the parameter fitting result can be a fitted straight line. This fitted straight line can be compared with the communication quality threshold of the corresponding port. If the fitted straight line reaches the communication quality threshold of the port, it can be determined that the corresponding port may be experiencing an anomaly. Furthermore, the moment when the fitted straight line reaches the communication quality threshold can be used to determine the moment of the anomaly in the corresponding port.

[0058] According to embodiments of this disclosure, when predicting communication quality, a clustering algorithm is first used to identify abnormal cluster centers. Linear regression is then performed on the changes in these abnormal cluster centers at different times to achieve trend analysis and prediction of abnormal network communication quality parameters. Therefore, analysis can be performed using the parameter change trends of abnormal ports, reducing the probability of misjudgments. Furthermore, if the parameter fitting result of a certain port reaches the communication quality threshold for that port, it can be determined that the communication quality of that port will be abnormal. The moment when the operating parameters reach this communication quality threshold is designated as the abnormal moment, and an early warning is issued, providing convenience for maintenance personnel.

[0059] According to embodiments of this disclosure, determining the communication quality prediction result for a target network device based on the communication quality prediction results of multiple ports includes: when the communication quality prediction results of multiple ports indicate that the target network device has ports with normal communication quality, determining that the port corresponding to the abnormal result has a communication abnormality based on the abnormal result indicating abnormal communication quality among the multiple communication quality prediction results; and when the communication quality prediction results of multiple ports indicate that the target network device does not have ports with normal communication quality, determining that the target network device has a fault.

[0060] According to embodiments of this disclosure, when predicting the communication quality of a target network device, various results can be obtained, such as an anomaly in one port of the target network device, an anomaly in multiple ports of the target network device, or an anomaly in all ports of the target network device. Furthermore, if the anomaly result obtained through port detection data is due to a fault in the port itself, or a fault in the communication link connected to the corresponding port, then the anomaly may be due to a fault in the port itself.

[0061] According to embodiments of this disclosure, in the vector space, since each dimension corresponds to a port on the target network device, the location of the running vector can be used to determine which port is functioning correctly. If the running vector indicates that all ports are malfunctioning, since it is generally unlikely that all links connected to the target network device will fail, it can be determined that the target network device has malfunctioned when this occurs.

[0062] According to embodiments of this disclosure, in the above-mentioned communication quality prediction results, for the target network device, if there are normal ports and several ports experience communication anomalies, it indicates that the links corresponding to those ports are experiencing communication anomalies; however, if all ports experience communication anomalies, it indicates that the switch itself is faulty. Therefore, by observing the status of multiple ports, it is possible to quickly determine whether the problem is a local port link anomaly or a fault in the network device itself, such as the switch, thereby quickly locating the root cause of the problem. In complex network environments, this can save time for maintenance personnel, speed up problem resolution, improve maintenance efficiency, and thus provide strong support for network maintenance.

[0063] According to embodiments of this disclosure, fitting the cluster centers of anomaly clusters at multiple time points to obtain parameter fitting results representing the cluster centers of anomaly clusters includes: determining the components of the cluster centers of anomaly clusters at multiple time points in each dimension; arranging the components in each dimension according to the chronological order of the multiple time points to obtain parameter sequences in each dimension; performing linear fitting on the parameter sequences in each dimension to obtain fitting sub-results in each dimension; and determining the parameter fitting results representing the cluster centers of anomaly clusters based on the multiple fitting sub-results.

[0064] According to embodiments of this disclosure, when fitting anomaly clusters at multiple time points, the running vectors within the anomaly clusters are first divided according to dimensions and arranged in the order of time to obtain parameter sequences for each dimension. The parameter sequences for each dimension can reflect the changing trend of the running parameters in the current dimension. Then, linear fitting is performed on the parameter sequences for each dimension to obtain fitting sub-results for each dimension. These fitting sub-results can clearly show the changes in the running parameters of each port on the target network device. Finally, based on these fitting sub-results, the parameter fitting result for the cluster center of the overall anomaly cluster is obtained.

[0065] According to embodiments of this disclosure, fitting a target network device with only a single port anomaly is relatively easy. However, when multiple ports are anomaly-prone, each port needs to be separated, sorted based on the sampling time, and fitted individually. This can reduce the risk of increased error due to multi-dimensional fitting.

[0066] According to embodiments of this disclosure, multiple multidimensional vectors in a multidimensional space are fitted in each dimension to obtain a multidimensional fitting result. This allows for comprehensive analysis of anomaly clusters through meticulous fitting operations, thereby enabling accurate prediction of communication quality anomalies, adapting to complex communication environments, reducing fitting errors, and improving device diagnostic efficiency and reliability.

[0067] According to embodiments of this disclosure, there is a correspondence between multiple ports and multiple abnormal clusters, and the correspondence is determined in the following way: for each abnormal cluster, the cluster center of the abnormal cluster is determined; the projection distance between the cluster center of the abnormal cluster and the cluster center of the normal cluster in multiple dimensions of the multidimensional vector space is determined, wherein each dimension of the multidimensional vector space corresponds to a port; based on the multiple projection distances, at least one port corresponding to the abnormal cluster is determined.

[0068] According to embodiments of this disclosure, after clustering is completed, multiple clusters can be obtained. Generally, the normal cluster contains the most running vectors, while the remaining abnormal clusters, which contain fewer running vectors, are caused by abnormalities in running parameters due to anomalies in one or more ports. Since there is a one-to-one correspondence between dimensions and ports, when a port becomes abnormal at a certain moment, the running vector at that moment will be clustered into the abnormal cluster. When determining which ports cause abnormal clusters, the determination can be based on the projected distance between the cluster centers of the abnormal clusters and the cluster centers of the normal clusters.

[0069] According to embodiments of this disclosure, determining at least one port corresponding to an abnormal cluster based on multiple projection distances includes: determining at least one abnormal projection distance greater than a preset distance threshold from the multiple projection distances; and determining at least one port corresponding to an abnormal cluster based on the correspondence between the dimensions of each of the at least one abnormal projection distance and multiple dimensions of the multidimensional vector space and multiple ports.

[0070] According to embodiments of this disclosure, when calculating the cluster center for each anomalous cluster, the mean method can typically be used, that is, the average value of multiple running vectors within the cluster in each dimension is taken as the coordinates of the cluster center. When determining the cluster center of a normal cluster, the normal cluster is composed of running vectors under normal communication conditions, and its cluster center is also the average value of these data points in each dimension.

[0071] In a multidimensional vector space, each dimension corresponds to a specific operating parameter of a port. For example, dimension 1 might correspond to the number of bytes sent by port 1, and dimension 2 might correspond to the number of bytes sent by port 2, and so on. The projected distances between the cluster centers of the abnormal clusters and the cluster centers of the normal clusters in each dimension of the multidimensional vector space are calculated. These multiple projected distances can be compared with the communication quality thresholds corresponding to each port. If the projected distance of a port in a given dimension is close to or exceeds its corresponding communication quality threshold, then that port is determined to correspond to that abnormal cluster, indicating that the port may have communication anomalies.

[0072] According to embodiments of this disclosure, abnormal ports can be identified by projection distance and dimension, resulting in more accurate location, which can quickly narrow down the scope of troubleshooting, improve the diagnostic efficiency of maintenance personnel, and save time and manpower costs. It is especially suitable for complex and ever-changing communication network environments.

[0073] According to embodiments of this disclosure, multiple sets of operating parameters are projected into a multidimensional vector space to obtain operating vectors for multiple time moments. This includes: for each set of operating parameters at each time moment, based on the correspondence between multiple dimensions of the multidimensional vector space and multiple ports, projecting multiple operating parameters in the set of operating parameters onto the dimensions corresponding to the ports that collect the operating parameters, to obtain the components of the operating parameters in the dimensions; and determining the operating vector for the corresponding time moment based on the components of the operating parameters in the multiple dimensions.

[0074] In the embodiments of this disclosure, when projecting multiple sets of operating parameters onto a multidimensional vector space, firstly, the correspondence between each port and the corresponding dimension in the multidimensional vector space is determined. Then, for each time-varying set of operating parameters, the operating parameters of each port are projected onto the corresponding dimension to obtain the components of the operating parameters in that dimension, and the operating vector is determined based on the multiple components.

[0075] For example, if the target network device has three ports, each port has three parameters: the number of bytes sent, the number of bytes received, and the packet loss rate. For each time moment, the features of the three parameters for each port can be extracted and fused to obtain fused features. The fused features of the three ports are then projected onto the dimension corresponding to that port to obtain the components in that dimension. Finally, the running vector is determined based on the components in the three dimensions.

[0076] In the embodiments of this disclosure, by embedding the operating parameters into a multi-dimensional vector space in the form of vectors, the operating parameters can be effectively organized, providing standardized input for subsequent cluster analysis and other tasks, and preparing the data at the data level in advance for troubleshooting and maintenance.

[0077] In the embodiments of this disclosure, multiple operating vectors are clustered sequentially according to the order of multiple moments in a first preset time period to obtain clustering results for each moment. This includes: obtaining operating parameter groups for each moment of the target network device within a second preset time period, wherein the last moment of the second preset time period is the same as the first moment of the first preset time period; for each moment in the first preset time period, selecting a preset number of operating parameter groups prior to the moment as the historical operating parameter groups for that moment, wherein the preset number is determined based on the number of moments in the second preset time period; and using a preset clustering algorithm to cluster the historical operating parameter groups and the operating vectors of the moment to obtain the clustering results for that moment.

[0078] In the embodiments of this disclosure, since this application adopts a distance-based clustering algorithm, the clustering of this disclosure needs to be carried out multiple times over a period of time with a time step. However, for the first moment in the first preset time period, the data used for clustering only has the running parameter set of the first moment, which makes it difficult to complete the clustering.

[0079] Therefore, a second preset time period can be set before the first preset time period, so that the last moment of the second preset time period is the same as the first moment of the first preset time period. The second preset time period also includes multiple sets of operating parameters. During clustering, for multiple moments in the first preset time period, the operating vector of that moment and multiple sets of operating parameters preceding that moment can be selected for clustering. The number of selected operating parameter sets can be determined based on the number of moments in the second preset time period.

[0080] In the embodiments of this disclosure, when fitting the cluster centers of multiple anomaly clusters, there are two fitting methods. The first method is to fit the multiple running vectors included in the second preset time period and the running vectors from the first preset time period together after clustering is completed at the first moment of the first preset time period. The second method is to use the running vectors from the first moment of the first preset time period as the starting point for fitting, that is, discarding the running vectors included in the second preset time period during fitting. In actual train communication quality prediction, the appropriate fitting method can be selected based on the actual situation.

[0081] In the embodiments of this disclosure, at the first moment of the first preset time period, there is only one running vector, making clustering difficult. Therefore, by acquiring multiple sets of running parameters within the second preset time period, data support can be provided for the clustering operation at the first moment of the first preset time period, enabling successful clustering and subsequent prediction and judgment. Using the second preset time period allows for capturing the changing trend of the running status at the beginning of the first preset time period, thereby achieving accurate prediction of the operating quality of ports on the target network device and providing a reliable basis for preventive maintenance of complex network devices, reducing the risk of later failures and maintenance costs.

[0082] Figure 3 A data flow diagram of a train communication quality prediction method according to an embodiment of the present disclosure is illustrated.

[0083] The following combination Figure 3 The train communication quality prediction method disclosed herein will be further explained.

[0084] In the embodiments of this disclosure, firstly, the target network device in the train is acquired within a first preset time period. The respective operating parameter groups at each time point, as shown in the figure. . Figure 3 China against the Communist Party An example representation of omitting the parameter group at each time step.

[0085] Then, Each set of operating parameters is projected into a multidimensional vector space to obtain... Each time step has its own execution vector. The dimension of each execution vector corresponds one-to-one with the port of the target network device.

[0086] Then, following the sequence of multiple moments within the first preset time period, multiple running vectors are clustered sequentially to obtain... The clustering results at each time point. Specifically, The clustering results at time step are based on The vectors obtained at time points and before are given. The clustering results at time step are based on The running vector at each time step is obtained. The clustering results at time step are based on The vector at each time step is obtained. This process continues for other time steps, without further explanation.

[0087] The multiple clustering results obtained above may include multiple anomalous clusters.

[0088] For the obtained multiple anomaly clusters, the cluster centers of each anomaly cluster at multiple time points are fitted to obtain parameter fitting results representing the cluster centers of the anomaly clusters. The parameter fitting results at multiple time points can include... The abnormal cluster center at any given moment.

[0089] Based on the parameter fitting results of multiple ports and the communication quality thresholds of each port, the communication quality prediction result of the port under test is determined by comparing the projection distance. Furthermore, based on the communication quality prediction results of each port, the communication quality prediction result for the target network device can be determined.

[0090] According to an embodiment of this disclosure, a train is also proposed, on which a controller and a target network device are configured. The controller executes the steps of the train communication quality prediction method described above to obtain the communication quality prediction result of the target network device.

[0091] Based on the above-described train communication quality prediction method, this disclosure also provides a train communication quality prediction device. The following will be combined with... Figure 4 The device is described in detail.

[0092] Figure 4 A schematic block diagram of a train communication quality prediction device according to an embodiment of the present disclosure is shown.

[0093] like Figure 4 As shown, the train communication quality prediction device 400 of this embodiment includes an acquisition module 410, a projection module 420, a clustering module 430, and a prediction module 440.

[0094] The acquisition module 410 is used to acquire the operating parameter sets of the target network device in the train at multiple times within a first preset time period. Each time-based operating parameter set includes the operating parameters of multiple ports of the target network device at that time. In one embodiment, the acquisition module 410 can be used to perform the operation S210 described above, which will not be repeated here.

[0095] The projection module 420 projects multiple sets of operating parameters into a multi-dimensional vector space to obtain operating vectors at multiple time points. The dimensions of the multi-dimensional vector space are determined based on the number of ports of the target network device. In one embodiment, the projection module 420 can be used to perform the operation S220 described above, which will not be repeated here.

[0096] The clustering module 430 is used to cluster multiple running vectors sequentially according to the order of multiple moments in a first preset time period, obtaining clustering results for each moment. Each clustering result includes normal clusters representing the target network device in a normal state and abnormal clusters representing the target network device in an abnormal state. In one embodiment, the clustering module 430 can be used to perform the operation S230 described above, which will not be repeated here.

[0097] The prediction module 440 is used to fit the cluster centers of each of the multiple anomaly clusters to determine the predicted communication quality of the target network device. In one embodiment, the prediction module 440 may be used to perform the operation S230 described above, which will not be repeated here.

[0098] According to embodiments of this disclosure, the prediction module 440 may include a first prediction submodule, a second prediction submodule, and a third prediction submodule.

[0099] The first prediction submodule is used to fit the cluster centers of each anomalous cluster at multiple time points for each anomalous cluster, and obtain the parameter fitting results representing the cluster centers of the anomalous cluster.

[0100] The second prediction submodule is used to determine the communication quality prediction result of the port to be tested based on the parameter fitting results and the communication quality thresholds of multiple ports.

[0101] The third prediction submodule is used to determine the communication quality prediction result for the target network device based on the communication quality prediction results of multiple ports.

[0102] According to embodiments of this disclosure, the second prediction submodule may include the first prediction unit.

[0103] The first prediction unit is used to determine, for each port, that the communication quality detection result of the port under test indicates that the communication quality of the port will be abnormal when the port's operating parameters reach the port's communication quality threshold, based on the port's parameter fitting results, and to determine the abnormal moment when the operating parameters reach the port's communication quality threshold.

[0104] According to embodiments of this disclosure, the third prediction submodule may include a second prediction unit and a third prediction unit.

[0105] The second prediction unit is used to determine, based on the abnormal results indicating abnormal communication quality among the multiple communication quality prediction results, that the port corresponding to the abnormal result has a communication abnormality when the communication quality prediction results of multiple ports indicate that there are ports with normal communication quality in the target network device.

[0106] The third prediction unit is used to determine that the target network device is faulty when the communication quality prediction results of multiple ports indicate that there are no ports with normal communication quality in the target network device.

[0107] According to embodiments of this disclosure, the first prediction submodule may include a fourth prediction unit, a fifth prediction unit, a sixth prediction unit, and a seventh prediction unit.

[0108] The fourth prediction unit is used to determine the components of the cluster centers of the anomalous clusters in each dimension at multiple time points.

[0109] The fifth prediction unit is used to arrange the components in each dimension according to the chronological order of multiple time points to obtain the parameter sequence in each dimension.

[0110] The sixth prediction unit is used to perform linear fitting on the parameter sequences in each dimension to obtain the fitting sub-results in each dimension.

[0111] The seventh prediction unit is used to determine the parameter fitting results representing the cluster centers of anomalous clusters based on multiple fitting results.

[0112] According to embodiments of this disclosure, the prediction module 440 may further include a fourth prediction submodule, a fifth prediction submodule, and a sixth prediction submodule.

[0113] The fourth prediction submodule is used to determine the cluster center of each anomalous cluster.

[0114] The fifth prediction submodule is used to determine the projection distances of the cluster centers of abnormal clusters and normal clusters in multiple dimensions of the multidimensional vector space, where each dimension of the multidimensional vector space corresponds to a port.

[0115] The sixth prediction submodule is used to determine at least one port corresponding to the anomaly cluster based on multiple projection distances.

[0116] According to embodiments of this disclosure, the sixth prediction submodule may include an eighth prediction unit and a ninth prediction unit.

[0117] The eighth prediction unit is used to determine at least one abnormal projection distance that is greater than a preset distance threshold from multiple projection distances.

[0118] The ninth prediction unit is used to determine at least one port corresponding to the anomaly cluster based on the correspondence between the dimensions of each anomaly projection distance and multiple dimensions of the multidimensional vector space and multiple ports.

[0119] According to embodiments of this disclosure, the projection module 420 may include a first projection submodule and a second projection submodule.

[0120] The first projection submodule is used to project multiple operating parameters in the operating parameter group at each time step onto the dimension corresponding to the port that collects the operating parameters, based on the correspondence between multiple dimensions of the multidimensional vector space and multiple ports, to obtain the components of the operating parameters in the dimension.

[0121] The second projection submodule is used to determine the running vector at any given time based on the components of the running parameters in multiple dimensions.

[0122] According to embodiments of this disclosure, clustering module 430 may include a first clustering submodule, a second clustering submodule, and a third clustering submodule.

[0123] The first clustering submodule is used to obtain the operating parameter groups of the target network device at multiple times within a second preset time period, wherein the last time of the second preset time period is the same as the first time of the first preset time period.

[0124] The second clustering submodule is used to select a preset number of running parameter groups before each moment in the first preset time period as the historical running parameter group for that moment, wherein the preset number is determined according to the number of moments in the second preset time period.

[0125] The third clustering submodule is used to cluster the historical running parameter group and the running vector at each time using a preset clustering algorithm to obtain the clustering result at each time.

[0126] According to embodiments of this disclosure, any plurality of modules among the acquisition module 410, projection module 420, clustering module 430, and prediction module 440 may be combined into one module, or any one of these modules may be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of other modules and implemented in one module. According to embodiments of this disclosure, at least one of the acquisition module 410, projection module 420, clustering module 430, and prediction module 440 may be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the acquisition module 410, projection module 420, clustering module 430 and prediction module 440 may be implemented at least partially as a computer program module that can perform corresponding functions when the computer program module is run.

[0127] Figure 5 A block diagram of an electronic device suitable for implementing a train communication quality prediction method according to an embodiment of the present disclosure is shown schematically.

[0128] like Figure 5 As shown, an electronic device 500 according to an embodiment of the present disclosure includes a processor 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage portion 508 into a random access memory (RAM) 503. The processor 501 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 501 may also include onboard memory for caching purposes. The processor 501 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0129] RAM 503 stores various programs and data required for the operation of electronic device 500. Processor 501, ROM 502, and RAM 503 are interconnected via bus 504. Processor 501 performs various operations of the method flow according to embodiments of the present disclosure by executing programs in ROM 502 and / or RAM 503. It should be noted that the programs may also be stored in one or more memories other than ROM 502 and RAM 503. Processor 501 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in said one or more memories.

[0130] According to embodiments of this disclosure, the electronic device 500 may further include an input / output (I / O) interface 505, which is also connected to a bus 504. The electronic device 500 may also include one or more of the following components connected to the I / O interface 505: an input section 506 including a keyboard, mouse, etc.; an output section 507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 510 as needed so that computer programs read from it can be installed into the storage section 508 as needed.

[0131] This disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of this disclosure.

[0132] According to embodiments of this disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this disclosure, the computer-readable storage medium may include ROM 502 and / or RAM 503 and / or one or more memories other than ROM 502 and RAM 503 described above.

[0133] Embodiments of this disclosure also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to cause the computer system to implement the item recommendation method provided in the embodiments of this disclosure.

[0134] When the computer program is executed by the processor 501, it performs the functions defined in the system / apparatus of this disclosure embodiments. According to embodiments of this disclosure, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0135] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via the communication section 509, and / or installed from a removable medium 511. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0136] In such an embodiment, the computer program can be downloaded and installed from a network via communication section 509, and / or installed from removable medium 511. When the computer program is executed by processor 501, it performs the functions defined in the system of this disclosure embodiment. According to embodiments of this disclosure, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0137] According to embodiments of this disclosure, program code for executing the computer programs provided in embodiments of this disclosure can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can execute entirely on a user's computing device, partially on a user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0138] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0139] Those skilled in the art will understand that the features described in the various embodiments of this disclosure can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. In particular, the features described in the various embodiments of this disclosure can be combined and / or combined in various ways without departing from the spirit and teachings of this disclosure. All such combinations and / or combinations fall within the scope of this disclosure.

[0140] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A train communication quality prediction method characterized by, The method comprises: obtaining a plurality of running parameter groups of a target network device in a train at respective time points in a first preset time period, wherein each running parameter group at a time point comprises running parameters of a plurality of ports of the target network device at the time point; projecting the plurality of running parameter groups into a multi-dimensional vector space respectively to obtain a plurality of running vectors at respective time points, wherein the dimension of the multi-dimensional vector space is determined according to the number of ports of the target network device; sequentially clustering the plurality of running vectors in the order of the plurality of time points in the first preset time period to obtain a plurality of clustering results at respective time points, wherein each clustering result in the plurality of clustering results comprises a normal cluster representing that the target network device is in a normal state and an abnormal cluster representing that the target network device is in an abnormal state; and fitting clustering centers of the plurality of abnormal clusters respectively to determine a communication quality prediction result of the target network device.

2. The method of claim 1, wherein, The fitting of the clustering centers of the plurality of abnormal clusters respectively to determine the communication quality prediction result of the target network device comprises: fitting the clustering centers of the abnormal cluster at respective time points to obtain a parameter fitting result representing the clustering center of the abnormal cluster for each abnormal cluster; determining a communication quality prediction result of the to-be-tested port based on the parameter fitting result and a communication quality threshold of the plurality of ports respectively; and determining the communication quality prediction result of the target network device based on the communication quality prediction result of the plurality of ports respectively.

3. The method of claim 2, wherein, The determination of the communication quality prediction result of the to-be-tested port based on the parameter fitting result and the communication quality threshold of the plurality of ports respectively comprises: for each port, determining that the communication quality detection result of the to-be-tested port represents that the communication quality of the port will be abnormal and determining an abnormal time point at which the running parameter reaches the communication quality threshold of the port based on the parameter fitting result of the port and determining that the running parameter of the to-be-tested port reaches the communication quality threshold of the port.

4. The method of claim 3, wherein, The determination of the communication quality prediction result of the target network device based on the communication quality prediction result of the plurality of ports respectively comprises: in a case where the communication quality prediction result of the plurality of ports respectively represents that the target network device has a port with normal communication quality, determining that a port corresponding to an abnormal result representing abnormal communication quality in the plurality of communication quality prediction results has abnormal communication quality; and in a case where the communication quality prediction result of the plurality of ports respectively represents that the target network device does not have a port with normal communication quality, determining that the target network device has a fault.

5. The method of claim 2, wherein, The fitting of the clustering centers of the abnormal cluster at respective time points to obtain a parameter fitting result representing the clustering center of the abnormal cluster comprises: determining components of the clustering centers of the abnormal cluster at respective time points in each dimension; arranging the components in each dimension in the order of the plurality of time points respectively to obtain a parameter sequence in each dimension; and fitting the parameter sequence in each dimension to obtain the parameter fitting result representing the clustering center of the abnormal cluster. linearly fitting the parameter sequence on each dimension to obtain a fitting sub-result on each dimension; and determining a parameter fitting result representing a cluster center of the abnormal cluster based on the fitting sub-results.

6. The method of claim 2, wherein, The correspondence between the plurality of ports and the plurality of abnormal clusters is determined by: determining, for each abnormal cluster, a cluster center of the abnormal cluster; determining, respectively, a projection distance between the cluster center of the abnormal cluster and the cluster center of the normal cluster on each dimension of the multi-dimensional vector space, wherein each dimension of the multi-dimensional vector space corresponds to one port; determining at least one port corresponding to the abnormal cluster based on the projection distances.

7. The method of claim 6, wherein, The determining at least one port corresponding to the abnormal cluster based on the projection distances comprises: determining, from the projection distances, at least one abnormal projection distance greater than a preset distance threshold; and determining at least one port corresponding to the abnormal cluster based on the dimension where each abnormal projection distance is located and the correspondence between the plurality of dimensions of the multi-dimensional vector space and the plurality of ports.

8. The method of claim 6, wherein, The projecting the plurality of running parameter groups into the multi-dimensional vector space to obtain a running vector of each time comprises: projecting, for each running parameter group of each time, the running parameters in the running parameter group into a dimension corresponding to a port from which the running parameters are collected based on the correspondence between the plurality of dimensions of the multi-dimensional vector space and the plurality of ports to obtain a component of the running parameters on the dimension; and determining the running vector of the time based on the components of the running parameters on the plurality of dimensions.

9. The method of claim 1, wherein, The clustering the plurality of running vectors in the order of the plurality of times in the first preset time period to obtain a clustering result of each time comprises: obtaining a plurality of running parameter groups of the target network device at a plurality of times in a second preset time period, wherein the last time of the second preset time period is the same as the first time of the first preset time period; selecting, for each time in the first preset time period, a preset number of running parameter groups before the time as a historical running parameter group of the time, wherein the preset number is determined according to the number of times in the second preset time period; and clustering the historical running parameter group of the time and the running vector of the time using a preset clustering algorithm to obtain a clustering result of the time.

10. A train loaded with a controller and target network device, characterized by, The controller executes the steps of the method of any one of claims 1-9 to obtain a communication quality prediction result of the target network device.

Images