Fault node identification method and device, electronic equipment, storage medium and computer program product

Abstract

The invention provides a fault node identification method and device, electronic equipment, a storage medium and a computer program product, and relates to the technical field of the Internet of Things, and the method comprises the steps: constructing an attribute graph corresponding to a plurality of nodes; wherein the attribute graph is used for representing one or more of the following: an identification feature of each node, a connection relationship feature between every two nodes, and an attribute feature of each node in the field of ontology perception and network communication; identifying and determining N first disconnected nodes based on the attribute graph; wherein the first lost node is a node lost from the corresponding initial neighborhood node set; n is an integer greater than 0; and determining a fault node based on the consistency and trend of the attribute feature of each first neighborhood node in the initial neighborhood node set corresponding to each first lost node and the difference between the attribute feature of each first neighborhood node and the attribute feature of each first lost node.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet of Things, and in particular to a fault node identification method and device, an electronic device, a storage medium and a computer program product. BACKGROUND

[0002] In a new generation of intelligent manufacturing and industrial collaborative system, multiple mobile robots or devices need to continuously complete task division and collaborative work in a dynamic and complex environment. The existing industrial device robot fault detection and task rescheduling technical solution includes a platform dominant centralized perception and judgment scheme. This scheme highly depends on the communication stability and computing ability of the central platform. Once the platform fails or the communication link is interrupted, the perception and judgment ability of the whole system will be greatly reduced, and even the lost nodes cannot be found in time. Overall, the existing method mainly relies on platform centralized control in architecture, is single in information source, and lacks neighborhood collaboration and historical trend modeling in judgment mechanism. Therefore, when facing complex scenes such as network fluctuation, node degradation and edge computing, it is difficult to support the high requirements of real-time fault judgment of large-scale, distributed industrial collaborative systems. SUMMARY

[0003] The present application provides a fault node identification method, device, electronic device, storage medium and computer program product.

[0004] The technical solution of the present application is implemented as follows: The present application provides a fault node identification method, device, electronic device, storage medium and computer program product. Constructing an attribute graph corresponding to a plurality of nodes; wherein the attribute graph is used to represent one or more of the following: identification features of each of the nodes, connection relationship features between each two of the nodes, and attribute features of each of the nodes in the field of ontology perception and network communication; Identifying and determining N first lost nodes based on the attribute graph; wherein the first lost node is a node lost with a corresponding initial neighborhood node set; N is an integer greater than 0; Determining a fault node based on the consistency and trend of the attribute features of each first neighborhood node in the initial neighborhood node set corresponding to each first lost node, and the difference between the attribute features of each first neighborhood node and the attribute features of each first lost node.

[0005] In the above solution, the determination of the fault node based on the consistency and trend of the attribute features of each first neighborhood node in the initial neighborhood node set corresponding to each first lost node, and the difference between the attribute features of each first neighborhood node and the attribute features of each first lost node, comprises: screening M second lost nodes from the N first lost nodes based on a consistency analysis result of attribute features of the first neighborhood nodes; screening K third lost nodes from the M second lost nodes based on a trend analysis result of attribute features of second neighborhood nodes of each of the second lost nodes; wherein K is an integer greater than 0 and less than M; screening T fourth lost nodes from the K third lost nodes based on a difference between attribute features of each of the third lost nodes and attribute features of a corresponding each third neighborhood node; wherein T is an integer greater than 0 and less than K; determining a fault score of each fourth neighborhood node for the fourth lost node based on a predetermined attribute feature of a corresponding fourth neighborhood node of each of the fourth lost nodes and an attribute consistency deviation degree of the corresponding fourth neighborhood node of each of the fourth lost nodes; determining the fault node from the T fourth lost nodes based on the fault score of the corresponding fourth neighborhood node of each of the fourth lost nodes.

[0006] In the above scheme, the attribute graph is obtained periodically; the consistency analysis result includes an attribute change amplitude mean value and an attribute change fluctuation rate; the screening of the M second lost nodes from the N first lost nodes based on the consistency analysis result of the attribute features of the first neighborhood nodes includes: determining the attribute change amplitude mean value of the attribute features of the first neighborhood nodes of each of the first lost nodes; determining the attribute change fluctuation rate based on the attribute change amplitude mean value corresponding to each of the first lost nodes and a difference in attribute features of the first neighborhood nodes of each of the first lost nodes within adjacent periods; screening the M second lost nodes from the N first lost nodes based on the attribute change amplitude mean value and the attribute change fluctuation rate corresponding to the sizes respectively.

[0007] In the above scheme, the screening of the M second lost nodes from the N first lost nodes based on the attribute change amplitude mean value and the attribute change fluctuation rate corresponding to the sizes respectively includes any one of the following: screening the M second lost nodes corresponding to the attribute change amplitude mean value less than a first threshold value and the attribute change fluctuation rate greater than a second threshold value from the N first lost nodes; Screening M second lost nodes from N first lost nodes, corresponding to the attribute change amplitude mean less than the first threshold and the attribute change fluctuation rate less than the second threshold.

[0008] In the above scheme, the trend analysis result includes: attribute vector mean and neighborhood trend fluctuation rate; the trend analysis result based on the attribute characteristics of the second neighborhood node of each second lost node screens K third lost nodes from M second lost nodes, including: Determine the attribute vector of the attribute characteristics of the second neighborhood node of each second lost node; wherein the attribute vector is used to represent the attribute change direction per unit time; Based on the attribute vector of each second neighborhood node, determine the attribute vector mean of the corresponding second neighborhood node of each second lost node; Based on the difference between the attribute vector of each second domain node corresponding to each second lost node and the attribute vector mean, determine the neighborhood trend fluctuation rate; Based on the attribute vector mean and the domain trend fluctuation rate corresponding to the size respectively, screen K third lost nodes from M second lost nodes.

[0009] In the above scheme, screening K third lost nodes from M second lost nodes based on the attribute vector mean and the domain trend fluctuation rate corresponding to the size respectively includes any one of the following: Screen K third lost nodes from M second lost nodes, corresponding to the attribute vector mean less than the third threshold and the domain trend fluctuation rate greater than the fourth threshold; Screen K third lost nodes from M second lost nodes, corresponding to the attribute vector mean less than the third threshold and the domain trend fluctuation rate less than the fourth threshold.

[0010] In the above scheme, screening T fourth lost nodes from K third lost nodes based on the difference between the attribute characteristics of each third lost node and the attribute characteristics of each third neighborhood node corresponding to it, including: Based on the difference between the attribute characteristics of each third lost node corresponding to the attribute characteristics of each third neighborhood node corresponding to it, determine the attribute consistency deviation measure corresponding to each third lost node; Based on the sum of the attribute consistency deviation measure and the state information corresponding to the previous period, and the state parameter of whether each third lost node is online, determine the current state information corresponding to each third lost node; Based on the magnitude of the current status information corresponding to each of the third lost nodes, T fourth lost nodes are selected from the K third lost nodes.

[0011] In the above scheme, the step of selecting T fourth lost nodes from K third lost nodes based on the magnitude of the current state information corresponding to each third lost node includes: Among the K third disconnected nodes, T fourth disconnected nodes are selected whose current status information is not less than the fifth threshold.

[0012] In the above scheme, determining the fault score of each fourth neighboring node for the fourth lost-connection node based on the predetermined attribute features of the fourth neighboring nodes corresponding to each fourth lost-connection node and the attribute consistency deviation measure of the fourth neighboring nodes corresponding to each fourth lost-connection node includes: The fault score of each fourth neighboring node for the fourth lost node is determined by weighted summation based on the difference between the heartbeat state and the predetermined value of each fourth neighboring node, the pose drift rate, and the attribute consistency deviation metric of the fourth neighboring node corresponding to each fourth lost node.

[0013] In the above scheme, determining the faulty node among the T fourth lost-connection nodes based on the fault score corresponding to the fourth neighboring node for each fourth lost-connection node includes: Determine the percentage of each fourth disconnected node whose fault score is greater than the sixth threshold; Based on the proportion of information corresponding to each of the fourth lost-connection nodes, the faulty node is determined among the T fourth lost-connection nodes.

[0014] In the above scheme, determining the faulty node among the T fourth lost-connection nodes based on the proportion information corresponding to each of the fourth lost-connection nodes includes: Among the T fourth disconnected nodes, identify the faulty node whose corresponding proportion information is greater than the seventh threshold.

[0015] In the above scheme, the step of identifying and determining N first lost-connection nodes based on the attribute graph includes: Connectivity analysis is performed on the attribute graph to identify and determine multiple initial neighborhood node sets; Based on the attribute graph within the historical time period, N first disconnected nodes that have no connection relationship with any of the nodes in the initial neighborhood node set are determined.

[0016] In the above solution, the attribute features of each node in the fields of ontology perception and network communication include one or more of the following attributes: heartbeat state, path local density, residual power, pose drift rate, communication channel quality, communication delay fluctuation rate, packet loss rate, and throughput.

[0017] The embodiment of the present application further provides a fault node identification device, comprising: An acquisition unit is configured to construct an attribute graph corresponding to a plurality of nodes, wherein the attribute graph is configured to represent one or more of the following: identification features of each node, connection relationship features between two nodes, and attribute features of each node in the fields of ontology perception and network communication. An identification unit is configured to identify N first lost nodes based on the attribute graph, wherein the first lost nodes are nodes lost with a corresponding initial neighborhood node set, and N is an integer greater than 0. The identification unit is further configured to determine a fault node based on consistency and trend of attribute features of each first neighborhood node in the initial neighborhood node set corresponding to each first lost node, and difference between the attribute features of each first neighborhood node and the attribute features of each first lost node.

[0018] The embodiment of the present application further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program capable of running on the processor, and the processor implements steps in the above method when executing the computer program.

[0019] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement steps in the above method.

[0020] The embodiment of the present application further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement steps in the above method.

[0021] In the embodiments of the present application, a plurality of attribute graphs corresponding to nodes are constructed; wherein the attribute graph is used to represent one or more of the following: the identification feature of each node, the connection relationship feature between each two nodes, and the attribute feature of each node in the field of ontology perception and network communication; N first lost nodes are determined based on the attribute graph identification; wherein the first lost node is a node that is lost with the corresponding initial neighborhood node set; N is an integer greater than 0; the fault node is determined based on the consistency and trend of the attribute feature of each first neighborhood node in the initial neighborhood node set corresponding to each first lost node, and the difference between the attribute feature of each first neighborhood node and the attribute feature of each first lost node. In this way, since the attribute graph is used to represent the identification feature of each node, the connection relationship feature between each two nodes, and the attribute feature of each node in the field of ontology perception and network communication, the first lost node is identified based on the multi-source heterogeneous state attribute feature fusion analysis linkage, and the judgment basis is no longer dependent on a single signal such as heartbeat timeout, but based on the structural mutation of ontology perception and network communication to start attribute analysis process for identification, which is more accurate and efficient. And in the scheme of the present application, the composite decision chain of judging the consistency trend of the attribute feature of the neighborhood node and the difference between the attribute feature of the first lost node, compared with the traditional method which can only detect the lost node, this mechanism can pre-position the abnormal evolution trend of the node, and then the fault node can be determined in time. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 Flowchart of the fault node identification method provided by the embodiments of the present application Figure One Figure 2 Effect diagram of the fault node identification method provided by the embodiments of the present application Figure One Figure 3 Flowchart of the fault node identification method provided by the embodiments of the present application Figure Two Figure 4 Flowchart of the fault node identification method provided by the embodiments of the present application Figure Three Figure 5 Structure diagram of the fault node identification device provided by the embodiments of the present application Figure 6 Hardware entity diagram of the electronic device provided by the embodiments of the present application

[0023] It should be noted that the above-mentioned "first", "second" are only used to distinguish different schemes, and do not represent the degree of superiority or priority in the implementation process. DETAILED DESCRIPTION

[0024] ​​​​In order to make the purposes, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be further described in detail below in combination with the drawings and embodiments, and the described embodiments should not be regarded as limitations of the present application. All other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0025] In the following description, "some embodiments" are referred to, which describe a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0026] If similar descriptions of "first / second" appear in the application file, the following description is added. In the following description, the terms "first\second\third" referred to only distinguish similar objects, and do not represent a specific order of the objects. It can be understood that "first\second\third" can be interchanged in a specific order or sequence as allowed, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0027] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0028] The embodiments of the present application provide a fault node identification method, which is applied to a node identification device. The node identification device can include a server, a platform or a terminal with corresponding data processing functions, or can be any one of a plurality of robot nodes. Please refer to Figure 1 The flowchart of the fault node identification method provided by the embodiments of the present application is shown in Figure One The steps shown in Figure 1 will be described: S101, construct an attribute graph corresponding to a plurality of nodes; wherein the attribute graph is used to represent one or more of the following: identification features of each node, connection relationship features between two nodes, and attribute features of each node in the field of ontology perception and network communication.

[0029] In the embodiments of the present application, the node identification device obtains the corresponding identification features, the connection relationship features between two nodes, and the attribute features of each node in the field of ontology perception and the field of network communication from each node. Based on the obtained information, the attribute graph is constructed.

[0030] In the embodiments of the present application, the node identification device can also obtain the corresponding identification features of each node, the connection relationship features between each two nodes, and the attribute features of each node in the field of ontology perception and the field of network communication in each time period. Then, the attribute graph corresponding to each time period is constructed based on the obtained information. The time period can be 1 minute, 2 minutes or 10 minutes. The time period is not limited in the embodiments of the present application.

[0031] The nodes can include a plurality of robot nodes in industrialization, and can also include other device nodes with self-perception and communication functions.

[0032] The attribute features of each node in the field of ontology perception and the field of network communication include one or more attributes of heartbeat state, pose drift rate, communication channel quality, communication delay fluctuation rate, packet loss rate and throughput rate.

[0033] In the embodiments of the present application, the constructed attribute graph can be represented by , wherein is a point set, representing the identification features of all online robot nodes; is an edge set, representing the communicable link between robot nodes, that is, the connection relationship features between each two nodes (wherein connection can be represented by 1, and disconnection can be represented by 0); is the attribute feature of a point, representing the multi-perception attribute feature of each robot node. Specifically, each point contains the following attribute feature vectors:

[0034] , representing the attribute feature set of the t k th robot node.

[0035] : It is a state encoding vector that fuses perception information, wherein each dimension comes from different sources, but can be obtained through the local sensor or communication module of the robot node; : Heartbeat state, whether the heartbeat response of the neighbor node is received in the period is recorded through the communication module (such as Wireless Fidelity (WIFI) / 5G module), 1 represents that the heartbeat packet is successfully received, and 0 represents that the heartbeat packet is not received; : Path local density (surrounding obstacle / robot density), the surrounding obstacles or other nodes are scanned in a unit of time through the laser radar or depth camera system carried by the robot, and the maximum detection density of the region is taken as the normalization standard, which can be represented by formula (1) as follows: Equation (1) wherein, represents the number of scanned obstacles or other nodes in the current unit time. represents the maximum number of scanned obstacles or other nodes in the historical time.

[0036] : remaining power, the percentage of power or voltage level directly provided by the battery management module. This indicator can reflect that low power may cause the node to be passively offline or performance degradation, which is a key attribute for judging "non-communication factors causing abnormality"; the remaining power can be represented by Equation (2).

[0037] Equation (2) wherein, may identify the current percentage of power detected. may represent the minimum value of the percentage of power detected in the historical time. may represent the maximum value of the percentage of power detected in the historical time.

[0038] : pose drift rate, defined as the rate of change of the difference between the node's own estimated pose and the global positioning given by the platform in unit time, which judges whether the node is out of position, and is a core indicator for early detection of "navigation failure" or "visual degradation". The pose drift rate can be calculated by Equation (3) and Equation (4).

[0039] Equation (3) Equation (4) wherein, estimated position by the robot inertial navigation system and odometry, : true value or target reference position provided by the platform or fusion positioning system. represents unit time. represents the intermediate value of the pose offset in unit time. represents the maximum value of the pose drift rate in the historical time.

[0040] : communication channel quality, reflecting the channel condition when the robot receives data, and abnormal decline may indicate channel degradation or communication interruption risk. The 5G communication module used by the robot usually supports querying such network state information through standard AT commands; the communication channel quality can be calculated by Equation (5).

[0041] Equation (5) wherein, represents the current communication channel quality of the query. represents the maximum value of the communication channel quality in the historical time.

[0042] : Communication delay fluctuation rate, reflecting the change range of the round-trip time of periodic communication, mainly used to identify the real-time fluctuation problem existing in the communication link between the robot and the platform. It can be defined as the standard deviation of the round-trip time (RTT) in a unit time window; the communication delay fluctuation rate can be calculated by formula (6) and formula (7).

[0043] Formula (6) Formula (7) Wherein, represents the round-trip time of the lth period. n represents the number of periods. represents the average of the round-trip time of n periods. represents the maximum value of the round-trip time in n periods. represents the median value of the communication delay fluctuation rate.

[0044] : Packet loss rate, which can be obtained through the heartbeat mechanism based on UDP or TCP, and the packet loss rate can be calculated by recording the ratio of the number of heartbeat packets sent in a unit time to the number of successfully received ACKs. If the packet loss rate increases continuously for multiple periods, it may mean that the wireless environment is poor and the platform communication is abnormal. The packet loss rate can be defined as the packet loss rate in a unit time window before t_k; the packet loss rate can be calculated by formula (8).

[0045] Formula (8) Wherein, represents the ratio of the number of heartbeat packets sent in a unit time to the number of successfully received ACKs. represents the maximum ratio of the number of heartbeat packets sent to the number of successfully received ACKs in the historical time.

[0046] : Throughput rate, reflecting the actual data transmission capacity between the robot and the platform. This index can be obtained by the byte rate of sending and receiving packets through the system network interface to achieve statistics, and the transmission rate may indicate that the robot is currently in a weak coverage area, or encounters resource scheduling bottleneck. The throughput rate can be calculated by formula (9).

[0047] Formula (9) Wherein, represents the byte rate of sending and receiving packets currently obtained. represents the average of the received packets and the transmitted byte rate acquired within the historical period.

[0048] In the embodiments of the present application, in order to accurately identify and early warn the abnormal state of the robot node body, a set of attribute parameters complementary to the communication side and the body perception side are fused in the embodiments of the present application. The running health state of the robot node in a complex task environment is comprehensively described. Among them, the heartbeat state is a direct manifestation of the activity of the communication link, which is used to periodically detect the connection status with the neighbor nodes and capture the potential risk of disconnection in the first time; the path local density is perceived by the laser radar or the depth camera to perceive the surrounding obstacles or the density of adjacent robots, which can assist in determining whether the communication anomaly is caused by physical obstruction or dynamic congestion; the remaining power directly reflects the energy health status, which is a key indicator for identifying performance degradation or node disconnection caused by low power; the pose drift rate is judged by the difference rate of the body estimation and the platform global positioning to determine the navigation accuracy, which is suitable for early identification of positioning failure problems such as IMU drift and vision degradation. Further, the expansion of the communication attribute dimension enhances the sensitivity of the system to network side anomalies: the communication channel quality reflects the state of the wireless receiving channel, and the decline may indicate that the robot is in a weak coverage area; the communication delay fluctuation rate measures the instability of the round-trip time, which is used to detect the fluctuation risk of the real-time control link; the packet loss rate reveals the reliability of the communication link, which is an important means to identify high interference, congestion or platform anomalies; the throughput rate describes the data transmission capacity, and a sudden drop may mean channel degradation or system scheduling bottleneck. Overall, this parameter system spans the “body perception-network communication” dual domain, which not only has the ability to describe the fault precursor trend, but also can trace the cause analysis of communication anomalies and device degradation, providing multi-dimensional reliable support for subsequent state judgment, fault voting and platform intervention.

[0049] S102, determine N first disconnection nodes based on the attribute graph; wherein the first disconnection node is a node disconnected with the corresponding initial neighbor node set; N is an integer greater than 0.

[0050] In the embodiments of the present application, the node recognition device can also identify N first disconnection nodes in each time period through traversal method or step iteration method for the attribute graph corresponding to each time period. Among them, the first disconnection node is a node disconnected with the corresponding initial neighbor node set; N is an integer greater than 0.

[0051] In the embodiments of the present application, the node recognition device can identify the attribute graph in each time period t perform connectivity analysis (traversal method, step iteration method) to identify a plurality of initial neighbor node sets. If a node is stripped from the corresponding initial neighbor node set, i.e. no longer belongs to the nodes it consists of, then mark The first lost node is for. The process detects mutations in the communication structure (such as node disconnection, signal blocking), and the marked first lost node will form a candidate fault node pool as a trigger for subsequent attribute analysis.

[0052] The initial neighborhood node set is a set of robot nodes that have a communication connection relationship with each other when the plurality of robot nodes are initialized.

[0053] For example, in combination with Figure 2 The node identification device can determine that the second initial neighborhood node set 102 and the third initial neighborhood node set 103 each include a first lost node after identifying the attribute graph corresponding to the three initial neighborhood node sets.

[0054] S103, based on the consistency and trend of the attribute features of each first neighborhood node in each initial neighborhood node set corresponding to each first lost node, and the difference between the attribute features of each first neighborhood node and the attribute features of each first lost node, determine the fault node.

[0055] In the embodiments of the application, the node identification device can obtain the attribute features of each first neighborhood node in the corresponding initial neighborhood node set corresponding to each lost node in the attribute graph after determining the first lost node corresponding to each initial neighborhood node set. Then, according to the consistency, trend of the attribute features of the first neighborhood node, and the difference between the attribute features of each first neighborhood node and the attribute features of each first lost node, gradient screening is performed on the N first lost nodes to determine a fault node with the largest fault probability.

[0056] In the embodiments of the application, the node identification device can perform gradient screening on the N first lost nodes according to the consistency, trend of the attribute features of the first neighborhood node, and the difference between the attribute features of each first neighborhood node and the attribute features of each first lost node, to determine a part of lost nodes with the largest fault probability, and score the fault probability of each lost node in the part of lost nodes according to the key attribute features of the neighborhood nodes corresponding to the lost node. Based on the scoring of the neighborhood nodes corresponding to each lost node, the fault node is determined.

[0057] In the embodiments of the present application, since the attribute graph is used to represent the identification features of each node, the connection relationship features between each two nodes, and the attribute features of each node in the field of ontology perception and network communication, thus, the first lost node is recognized based on the multi-source heterogeneous state attribute feature fusion analysis linkage, and the judgment basis is no longer dependent on a single signal such as heartbeat timeout, but the attribute analysis process is started based on the structural mutation of ontology perception and network communication to perform recognition, and the accuracy and efficiency of recognition are higher. In the scheme of the present application, the consistency trend of the attribute features of the neighbor nodes and the difference between the attribute features of the first lost node are judged to form a composite decision chain, compared with the traditional method which can only detect the lost node, the mechanism can perceive the abnormal evolution trend of the node in advance, and thus the fault node can be determined in time.

[0058] Please refer to Figure 3 The flowchart of the fault node recognition method provided in the embodiments of the present application is shown in Figure Two , Figure 1 The S103 shown can also be implemented through S201 to S205, which will be described in combination with the steps shown in the figure. Figure 3 S201, based on the consistency analysis result of the attribute features of the first neighbor node, M second lost nodes are selected from N first lost nodes; wherein M is an integer greater than 0 and less than N.

[0059] In the embodiments of the present application, the node recognition device determines the consistency analysis result of the attribute features of the first neighbor node corresponding to each first lost node for each first lost node. Then, the second lost node with dramatic changes in the attribute features of the corresponding part of the first neighbor node or stable changes in the overall attribute features is determined according to the consistency analysis result, and then M second lost nodes are determined.

[0060] The consistency analysis result can be determined by the change amplitude and fluctuation rate between the attribute features of the first neighbor node.

[0061] S202, based on the trend analysis result of the attribute features of the second neighbor node of each second lost node, K third lost nodes are selected from M second lost nodes; wherein K is an integer greater than 0 and less than M.

[0062] In the embodiments of the present application, the node recognition device determines the trend analysis result of the attribute features of the second neighbor node corresponding to each second lost node for each second lost node. Then, the third lost node with dramatic changes in the attribute feature trend of the corresponding part of the second neighbor node or stable changes in the attribute feature trend is determined according to the trend analysis result, and then K third lost nodes are determined.

[0063] ​The trend analysis result can be determined by the average trend vector and the trend volatility of the attribute features of the second neighborhood node.

[0064] In S203, T fourth lost nodes are selected from the K third lost nodes based on the difference between the attribute features of each third lost node and the attribute features of each corresponding third neighborhood node.

[0065] In the embodiments of the present application, the node identification device can determine T fourth lost nodes with a difference between attribute features greater than a corresponding threshold value from the K third lost nodes based on the difference between the attribute features of the third domain node of each third lost node and the attribute features of the third lost node for each third lost node.

[0066] In S204, a failure score of each fourth neighborhood node for the fourth lost node is determined based on the predetermined attribute features of the fourth neighborhood node corresponding to each fourth lost node and the attribute consistency deviation metric of the fourth neighborhood node corresponding to each fourth lost node.

[0067] In the embodiments of the present application, the node identification device can determine a failure score of each fourth neighborhood node for the fourth lost node based on the heartbeat state and the pose drift rate of the fourth neighborhood node corresponding to each fourth lost node and the attribute consistency deviation metric of the fourth neighborhood node corresponding to each fourth lost node.

[0068] In the embodiments of the present application, the node identification device can determine a failure score of each fourth neighborhood node for the fourth lost node based on a weighted sum of the difference between the heartbeat state and a predetermined value, the pose drift rate of the fourth neighborhood node, and the attribute consistency deviation metric of the fourth neighborhood node corresponding to each fourth lost node.

[0069] For example, the node identification device can determine the failure score of the fourth lost node by formula (10).

[0070] Formula (10) Wherein, represents the failure score of the fourth neighborhood node j for the fourth lost node i, , and respectively represent the weight coefficients, represents the heartbeat state of the fourth neighborhood node j, represents the attribute consistency deviation of the fourth neighborhood node, represents the pose drift rate of the fourth neighborhood node.

[0071] S205, determine the fault node in the T fourth lost nodes based on the fault score corresponding to the fourth neighborhood node corresponding to each fourth lost node.

[0072] In the embodiments of the application, the node identification device can determine the fault score corresponding to each fourth lost node. Then, the fault node corresponding to the proportion information greater than the predetermined threshold value and the proportion threshold value can be determined in the T fourth lost nodes.

[0073] In the embodiments of the application, the node identification device can also detect whether there is a fault score greater than the preset maximum score threshold value in the plurality of fault scores corresponding to each fourth lost node, and if there is, determine that the fourth lost node is a fault node.

[0074] In the embodiments of the application, the node identification device can determine the proportion information of the fault score greater than the sixth threshold value corresponding to each fourth lost node; and determine the fault node in the T fourth lost nodes based on the size of the proportion information corresponding to each fourth lost node. The fault node corresponding to the proportion information greater than the seventh threshold value can be determined in the T fourth lost nodes. The seventh threshold value can include 50%, and the seventh threshold value is not limited in the embodiments of the application.

[0075] In the embodiments of the application, the second lost node is screened through the consistency analysis result of the attribute features of the first neighborhood node, the third lost node is screened according to the trend analysis result of the attribute features of the second neighborhood node, the fourth lost node is screened according to the difference between the attribute features of the third neighborhood node and the third lost node, and finally the final fault node is determined through the fault score of each fourth lost node. This compound decision chain of gradient screening can detect the lost nodes compared with the traditional method, the mechanism can pre-position the abnormal evolution trend of the node, and then the fault node can be determined in time.

[0076] In the embodiments of the application, S201 shown can also be implemented through S301 to S303, which will be described in combination with the steps: S301, determine the attribute change amplitude mean of the attribute features of the first neighborhood node of each first lost node.

[0077] In the embodiments of the application, the node identification device can determine the attribute change amplitude mean based on the absolute value of the difference between the attribute features of each first neighborhood node in each attribute period and the attribute features in the previous attribute period. The attribute change amplitude mean reflects the overall strength of the neighborhood attribute change, and judges whether it is a global disturbance.

[0078] Exemplarily, the attribute change amplitude mean value can be calculated by formula (11).

[0079] Formula (11) Wherein, represents the attribute change amplitude mean value. represents the number of first neighborhood nodes. represents the attribute feature of the first neighborhood node in the current time period, represents the attribute feature of the first neighborhood node in the previous time period.

[0080] S302, based on the attribute change amplitude mean value corresponding to each first lost node, and the difference of the attribute feature of the first neighborhood node of each first lost node in the adjacent period, determine the attribute change fluctuation rate.

[0081] In the embodiment of the application, the node recognition device can determine the absolute value of the difference between the attribute feature of each first neighborhood node in each attribute period and the attribute feature in the previous attribute period. Then, based on the square ratio of the difference between the absolute value and the attribute change amplitude mean value and the number of first neighborhood nodes, the attribute change fluctuation rate is determined. The attribute change fluctuation rate reflects the consistency of the neighborhood attribute change.

[0082] Exemplarily, the attribute change fluctuation rate can be calculated by formula (12).

[0083] Formula (12) Wherein, represents the attribute change fluctuation rate.

[0084] S303, based on the attribute change amplitude mean value and the attribute change fluctuation rate corresponding to the size, screening M second lost nodes from N first lost nodes.

[0085] In the embodiment of the application, the node recognition device can screen M second lost nodes from N first lost nodes, which correspond to the attribute change amplitude mean value less than the first threshold value and the attribute change fluctuation rate greater than the second threshold value; or screen M second lost nodes from N first lost nodes, which correspond to the attribute change amplitude mean value less than the first threshold value and the attribute change fluctuation rate less than the second threshold value.

[0086] In the embodiment of the application, after determining the attribute change amplitude mean value and the attribute change fluctuation rate, there are four kinds of judgment trends, if in the third Less than the first threshold value, Greater than the second threshold value, and the fourth Less than the first threshold value and If the value is less than the second threshold, the status of the corresponding second lost node can be further confirmed.

[0087] 1. If Greater than the first threshold If the value exceeds the second threshold, a high mean indicates an overall increase in the domain's attribute values, while a high variance indicates varying responses from neighbors, suggesting asynchronous changes. This could be due to widespread disturbances in the surrounding network or uniform environmental interference affecting the robot's location. In this case, the first disconnected node's loss of connection may be due to a non-entity-related fault caused by environmental disturbances, requiring delayed or low-priority processing to prevent misjudgment.

[0088] 2. If Greater than the first threshold If the value is below the second threshold, a high mean indicates an overall increase or decrease in the attribute value, with changes occurring collectively in the neighborhood. A low variance indicates that the changes in each neighborhood are consistent, reflecting a unified response to a certain instruction or state transition, such as entering a new task segment or encountering a unified instruction. In this case, consistent changes in the neighborhood may still indicate a normal state transition.

[0089] 3. If Less than the first threshold A value greater than the second threshold and a low mean indicate a stable overall state in the neighborhood and a stable system environment. A high variance indicates that individual nodes have experienced drastic changes, disrupting local stability. For example, environmental factors such as electromagnetic interference, strong light, or vibration may cause a drastic disturbance at a single point, leading to sluggishness or failure of the sensing module of the main node, abnormal changes in individual nodes, and stability in other nodes in the neighborhood. If a disconnected node corresponds to this situation, it is highly likely to fail.

[0090] 4. If Less than the first threshold The value is less than the second threshold, and the low mean and variance indicate that the neighborhood attributes are generally stable. However, the first disconnected node is separated from the main connected subgraph, indicating that the main node went offline on its own when there were no anomalies in the neighborhood, which is a high probability of failure.

[0091] In the embodiments of the present application, the attribute variation amplitude mean of the attribute feature of the first neighbor node of each first lost node is determined. The attribute variation fluctuation rate is determined based on the attribute variation amplitude mean corresponding to each first lost node and the difference in attribute feature of the first neighbor node of each first lost node in adjacent periods. The second lost node is selected from the N first lost nodes based on the sizes corresponding to the attribute variation amplitude mean and the attribute variation fluctuation rate. In this way, by the attribute feature consistency of the first neighbor node, that is, whether the attribute variation amplitude mean and the attribute variation fluctuation rate of the first neighbor node of the first lost node also change sharply at the same time, the second lost node that is likely to fail can be accurately identified.

[0092] In the embodiments of the present application, S202 shown can also be implemented through S401 to S404, which will be described in combination with the steps: S401, determining an attribute vector of the attribute feature of the second neighbor node of each second lost node; wherein the attribute vector is used to represent the attribute variation direction per unit time.

[0093] In the embodiments of the present application, the node identification device can obtain the attribute feature of each node in multiple time periods. Then, for each second lost node, the difference between the attribute feature of each second neighbor node of each second lost node in each time period and the attribute feature in the previous time period can be determined. Finally, the attribute vector of the second neighbor node is determined according to the number of multiple time periods. The attribute vector reflects the average trend of the attribute vector of the second neighbor node, and represents the attribute variation direction per unit time.

[0094] For example, the attribute vector can be calculated by formula (13).

[0095] Formula (13) Wherein, indicates the attribute vector of the jth second neighbor node in the tth time period. T represents the number of multiple time periods. indicates the attribute feature of the second neighbor node in the t-k+1th time period. indicates the attribute feature of the second neighbor node in the t-kth time period.

[0096] S402, determining the attribute vector mean of the second neighbor node corresponding to each second lost node based on the attribute vector of each second neighbor node.

[0097] In the embodiments of the present application, after determining the attribute vector of each second neighborhood node, the attribute vector mean of the second lost node can be determined based on the number of the second neighborhood nodes of each second lost node. The attribute vector mean reflects the average direction of the neighbor node attribute trend, and is used to determine whether the neighbors as a whole are in an evolution stage, such as task migration or environmental disturbance.

[0098] For example, the attribute vector mean can be calculated by formula (14).

[0099] Formula (14) Wherein, represents the attribute vector mean. N represents the number of the second neighborhood nodes of each second lost node.

[0100] S403, based on the difference between the attribute vector of each second neighborhood node corresponding to each second lost node and the attribute vector mean, the neighborhood trend fluctuation rate is determined.

[0101] In the embodiments of the present application, the node identification device determines the square of the absolute value of the difference between the attribute vector of each second neighborhood node corresponding to each second lost node and the attribute vector mean. Then, the neighborhood trend fluctuation rate is determined by taking the square root after averaging the number of the second neighborhood nodes. The neighborhood trend fluctuation rate reflects the evolution trend of each node in the neighborhood. The larger the value, the more inconsistent the evolution trend of each node in the neighborhood, and the more likely the heterogeneous change in the region.

[0102] For example, the neighborhood trend fluctuation rate can be calculated by formula (15).

[0103] Formula (15) Wherein, represents the neighborhood trend fluctuation rate.

[0104] S404, based on the size corresponding to the attribute vector mean and the field trend fluctuation rate, K third lost nodes are screened out from M second lost nodes.

[0105] In the embodiments of the present application, the node identification device can screen out K third lost nodes from M second lost nodes, which correspond to the attribute vector mean less than a third threshold and the field trend fluctuation rate greater than a fourth threshold; or screen out K third lost nodes from M second lost nodes, which correspond to the attribute vector mean less than the third threshold and the field trend fluctuation rate less than the fourth threshold.

[0106] In the embodiments of the present application, after the attribute vector mean and the neighborhood trend fluctuation rate are determined, there are four kinds of trend judgments. If it is the third kind of Lower, Higher, and the fourth kind of Lower, Lower, the third lost node is further identified. The third lost node is not completely interrupted in communication, but its state attribute deviates from the neighborhood trend for a long time, which may be a degenerative failure (such as battery performance degradation, sensing system accuracy degradation, control execution ability weakening, etc.).

[0107] 1. If is greater than the third threshold value, is greater than the fourth threshold value, it indicates that the overall state of the neighborhood is in a severe fluctuation, and the fluctuation direction is inconsistent, the change amplitude is large and chaotic, which indicates that the system has a "group disturbance" phenomenon. The group as a whole is disturbed in a large range, such as environmental change (electromagnetic field fluctuation, physical impact, large-scale shielding), or the scheduling system updates the instruction uniformly but the robots respond out of sync. At this time, it is not recommended to immediately judge whether the entity node is faulty, which may only be a large-scale state instability of the system, and delay intervention judgment is needed to avoid misjudgment. Judgment buffer, delay trigger mechanism, wait for subsequent trend stability.

[0108] 2. If is greater than the third threshold value, is less than the fourth threshold value, it indicates that the overall neighborhood has a synchronous change, the direction is consistent but the amplitude is large, which indicates that the system is executing a uniform state switching. For example, multiple robots uniformly accept task instructions to adjust the path, switch the task mode, or have a global scheduling reconfiguration. This is a normal system response, and the node is in a normal system state change, and the occurrence of heartbeat packet loss may only be a transient communication anomaly. Directly filter the node anomaly to exclude false positives and not enter the subsequent fault analysis process.

[0109] 3. If is less than the third threshold value, is greater than the fourth threshold value, it indicates that the overall neighborhood is stable, but some individual nodes change severely, breaking the local stability, which indicates that there is a severe disturbance or isolated anomaly in the local area. For example, a robot sensor anomaly causes abnormal feedback, and a limited environmental change (temperature difference, electromagnetic, collision) in a certain area causes a single point to change severely. If the lost node is in such a local severe fluctuation, it is highly likely to be an abnormal node, and the entity may have identification, communication, power supply, etc. problems. Enter mechanism 3 for integral judgment, and preliminarily mark as "suspected fault" and continuously monitor the state evolution trend.

[0110] 4. If is less than the third threshold value, Less than the fourth threshold value, indicating that the neighborhood attribute is overall smooth, and there is no obvious disturbance, indicating that the system is in a normal running state. In the stable operation of the system, the robot position, power, path density and other attributes fluctuate weakly. If the node is disconnected from the main communication subgraph at this time, it means that the ontology node is independently offline in an environment without disturbance, and the disconnection phenomenon is highly suspicious. Without further verification of external disturbance, directly enter mechanism 3 integral judgment and weighted scoring, strongly suspecting node failure (such as power failure, system crash).

[0111] In the embodiment of the application, the attribute vector of the attribute characteristics of the second neighborhood node of each second lost node is determined. Then, the attribute vector mean and the neighborhood trend fluctuation rate are determined according to the attribute vector. Based on the sizes corresponding to the attribute vector mean and the neighborhood trend fluctuation rate, K third lost nodes are screened out from M second lost nodes. In this way, by the trend of the attribute characteristics of the second neighborhood node of the second lost node, that is, whether the attribute vector mean and the neighborhood trend fluctuation rate of the second neighborhood node of the second lost node also change sharply at the same time, the third lost node that may fail can be accurately identified.

[0112] In the embodiment of the application, S203 shown can also be implemented by S501 to S503, which will be described in combination with the steps: S501, based on the difference between the attribute characteristics corresponding to each third lost node and the attribute characteristics of each corresponding third neighborhood node, determine the attribute consistency deviation measure corresponding to each third lost node.

[0113] In the embodiment of the application, the node recognition device can square the absolute value of the difference between the attribute characteristics corresponding to each third lost node and the attribute characteristics of each corresponding third neighborhood node, and then perform mean value calculation to determine the attribute consistency deviation measure. The attribute consistency deviation measure is the difference degree of the third lost node and the third neighborhood node in the attribute space. If it continues to be large, it means that the node behavior / state attribute continuously deviates from the neighborhood, which is a possible deterioration symptom.

[0114] Exemplarily, the attribute consistency deviation measure can be calculated by formula (16).

[0115] Formula (16) Wherein, The attribute consistency deviation measure is represented by N. N represents the number of third neighborhood nodes of the third lost node. The third lost node is represented by N. The jth third neighborhood node is represented by N.

[0116] S502, determine the current state information corresponding to each third missing node based on the sum of the attribute consistency deviation measure and the state information corresponding to the previous period, and the state parameter of whether each third missing node is online.

[0117] In the embodiments of the present application, the node identification device determines the difference between the sum of the attribute consistency deviation measure and the state information corresponding to the previous period, and the product of the state parameter of whether each third missing node is online and the suppression factor, to determine the current state information.

[0118] For example, the current state information can be calculated by formula (17).

[0119] Formula (17) Wherein, represents the current state information, represents the state information corresponding to the previous period, and when is greater than the fifth threshold value Γ, it is determined that the third missing node is a fourth missing node. represents the suppression factor of online feedback on abnormal integration, represents a flag variable indicating whether the node is online, represents an indicator function, which is 1 if online and 0 otherwise. By integrating over discrete time series, it reflects the cumulative trend of abnormalities over time, which is a dynamic scoring method based on time evolution, used to capture "gradual degradation" or "periodic abnormality not recovered" scenarios, avoiding false positives caused by only looking at instantaneous abnormalities.

[0120] S503, select T fourth missing nodes from K third missing nodes based on the size of the current state information corresponding to each third missing node.

[0121] In the embodiments of the present application, the node identification device selects T fourth missing nodes from K third missing nodes, whose corresponding current state information is not less than the fifth threshold value.

[0122] In the embodiments of the present application, after determining the current state information, there are three kinds of judgment trends. In the third case, that is, when the current state information is not less than the fifth threshold value, it is determined that the third missing node is a fourth missing node.

[0123] 1, if ≈0, the node state integral value is close to zero, indicating that the attribute change difference between the node and the neighborhood is very small, the overall consistency is high, and the system state is stable, for example, the node and the neighbor robot have consistent change trends in path density, power level, signal delay and other indicators, indicating that the system is running well and the individual has not appeared degradation or sudden abnormality. At this time, it can be judged as a normal node and does not need to be further judged.

[0124] 2, if 0 < S < S < (Fifth threshold) indicates that the node state integral has a certain degree of accumulation but does not exceed the warning threshold , which means that the node has a persistent slight deviation from the attributes of the neighborhood for a period of time, for example, the node path density is slightly higher than the neighborhood, the moving speed is slightly slower, the power consumption rate is slightly faster, etc., which may be disturbed by the local environment or slightly abnormal interference of the potential module. Such nodes should enter the integral continuous monitoring mechanism of mechanism 3 to confirm whether their state is gradually restored or further deteriorated.

[0125] 3, if S > S ≥ , which means that the state integral value is significantly high and exceeds the warning threshold, which means that the node has a long-term serious deviation from the attributes of the neighborhood, for example, the node continuously shows characteristics such as severe position drift, abnormal power change, or serious deviation of communication parameters from neighbor nodes due to reasons such as sensing module failure, power system aging, control module jamming, etc., indicating that the node may have a degenerative or non- sudden serious fault, and should be strongly marked as a fault candidate node.

[0126] In the embodiment of the application, based on the difference between the attribute characteristics of each third lost node and the attribute characteristics of each third neighborhood node corresponding thereto, the attribute consistency deviation measure corresponding to each third lost node is determined. Based on the sum of the attribute consistency deviation measure and the state information corresponding to the previous period, and the state parameter of whether each third lost node is online, the current state information corresponding to each third lost node is determined. Based on the size of the current state information corresponding to each third lost node, T fourth lost nodes are selected from K third lost nodes. In this way, through the difference between the attribute characteristics of the third neighborhood node and the attribute characteristics of the third lost node, it can be identified whether the behavior or state attribute of the third lost node continuously deviates from the attribute characteristics of the field node, and further, the fourth lost node that may have a fault can be accurately identified.

[0127] In the embodiment of the application, S102 shown can also be implemented through S601 to S602, which will be described in combination with the steps; S601, performing connectivity analysis on the attribute graph to identify and determine a plurality of initial neighborhood node sets.

[0128] In the embodiment of the application, the node recognition device determines a plurality of initial neighborhood node sets through the traversal method or the step iteration method for the attribute graph.

[0129] S602, determining, based on the attribute graph in the historical time period, N first lost nodes that have no connection relationship with any node in the initial neighborhood node set.

[0130] In the embodiments of the present application, the node identification device can determine the N first lost nodes without connection relationship for any node in the initial neighborhood node set based on the nodes connected with the initial neighborhood node set in the attribute graph in the historical time period.

[0131] If a node is connected with the initial neighborhood node set in the historical attribute graph, but is not connected with the initial neighborhood node set in the current attribute graph, it is determined that the node is a first lost node.

[0132] In the embodiments of the present application, the first lost nodes are identified in the initial neighborhood node set. In this way, the nodes that may fail in the plurality of nodes can be initially judged, and the judgment basis based on the consistency, trend and difference of the attribute characteristics is provided, and then the step-by-step lost node judgment can be realized. Compared with the scheme of obtaining the lost node only by the heartbeat mechanism in the related art, the step-by-step lost node judgment method can improve higher accuracy.

[0133] The embodiments of the present application propose a fault node identification method based on graph structure collaborative evolution trend. First, the scheme constructs a time-varying attribute graph and calculates a connected subgraph in real time, actively perceives the connectivity change of the node, identifies the "local lost node" that may be separated from the main graph, breaks through the limitation of the traditional judgment only relying on the heartbeat timeout, and realizes the ability to identify potential abnormalities from the network topology angle. Second, the scheme introduces a neighborhood attribute collaborative evolution analysis mechanism to avoid misjudgment caused by accidental communication fluctuations, and significantly enhances the robustness of the system. On this basis, a state integral mechanism is proposed to dynamically integrate the attribute deviation trend between the node and the neighborhood, establish time accumulation memory, and effectively perceive non-transient and degenerative faults. Further, in the case of uncertain judgment, a property heterogeneous integral mechanism is used to realize fine judgment of the fuzzy boundary, and the sensitivity and discrimination of the identification are improved. Finally, a decentralized fault confirmation is completed through the multi-node voting consensus mechanism in the local subgraph, a distributed decision logic is constructed for the edge autonomy, and the autonomy, robustness and judgment credibility of the system in complex environments are significantly improved, which has significant innovation and application value.

[0134] Please refer to Figure 4 The flowchart of the fault node identification method provided by the embodiments of the present application is shown Figure Three The steps will be described: S11, adjacency detection and graph perception.

[0135] In the embodiments of the present application, the node identification device constructs an attribute graph by acquiring the identification characteristics, connection relationship characteristics and attribute characteristics sent by each node.

[0136] S12, whether there is a first lost node.

[0137] In the embodiments of the present application, the node identification device identifies the attribute graph in the current time period to determine whether there is a first lost node. If yes, S13 is performed. If no, the identification of the attribute graph in the next judgment period is performed.

[0138] S13, neighborhood collaborative analysis.

[0139] In the embodiments of the present application, the node identification device determines the consistency analysis result of the attribute features of the first neighborhood nodes of the first lost node.

[0140] S14, whether there is a second lost node.

[0141] In the embodiments of the present application, whether there is a second lost node is determined by the size of the consistency analysis result. If yes, S15 is performed. If no, the identification of the attribute graph in the next judgment period is performed.

[0142] S15, neighborhood evolution trend analysis.

[0143] In the embodiments of the present application, the trend analysis result of the attribute features of the second neighborhood nodes of the second lost node is determined.

[0144] S16, whether there is a third lost node.

[0145] In the embodiments of the present application, whether there is a third lost node is determined by the size of the trend analysis result. If yes, S17 is performed. If no, the change consistency of the neighborhood attribute features is determined.

[0146] S17, state integral analysis.

[0147] In the embodiments of the present application, the current state information of the third neighborhood nodes of the third lost node is determined.

[0148] S18, whether there is a fourth lost node.

[0149] In the embodiments of the present application, whether there is a fourth lost node is determined by the size of the current state information of the third lost node. If yes, S19 is performed. If no, it is determined whether the integral value of the current state information is small.

[0150] S19, local consensus voting.

[0151] In the embodiments of the present application, based on the predetermined attribute features of the fourth neighborhood nodes corresponding to each of the fourth lost nodes and the attribute consistency deviation degree of the fourth neighborhood nodes corresponding to each of the fourth lost nodes, the failure score of each of the fourth neighborhood nodes for the fourth lost node is determined S20, failure node.

[0152] In the embodiments of the present application, based on the fault score of the fourth neighborhood node corresponding to each fourth missing node, the fault node is determined in the T fourth missing nodes S21, report the platform.

[0153] Please refer to Figure 5 The structure diagram of the fault node identification device provided in the embodiments of the present application is shown.

[0154] The embodiments of the present application also provide a fault node identification device 800, comprising an acquisition unit 801 and an identification unit 802.

[0155] The acquisition unit 801 is configured to construct an attribute graph corresponding to a plurality of nodes; wherein the attribute graph is used to represent one or more of the following: identification features of each node, connection relationship features between two nodes, and attribute features of each node in the field of ontology perception and network communication; The identification unit 802 is configured to identify and determine N first missing nodes based on the attribute graph; wherein the first missing node is a node that is missing with a corresponding initial neighborhood node set; N is an integer greater than 0; The identification unit 802 is further configured to determine a fault node based on the consistency and trend of the attribute features of each first neighborhood node in the initial neighborhood node set corresponding to each first missing node, and the difference between the attribute features of each first neighborhood node and the attribute features of each first missing node.

[0156] In the embodiments of the present application, the identification unit 802 in the fault node identification device 800 is configured to filter out M second missing nodes from the N first missing nodes based on the consistency analysis result of the attribute features of the first neighborhood node; wherein M is an integer greater than 0 and less than N; Filter out K third missing nodes from the M second missing nodes based on the trend analysis result of the attribute features of the second neighborhood node for each second missing node; wherein K is an integer greater than 0 and less than M; Filter out T fourth missing nodes from the K third missing nodes based on the difference between the attribute features of each third missing node and the attribute features of each corresponding third neighborhood node; wherein T is an integer greater than 0 and less than K; Determine the fault score of each fourth neighborhood node for the fourth missing node based on the predetermined attribute features of the fourth neighborhood node corresponding to each fourth missing node, and the attribute consistency deviation measure of the fourth neighborhood node corresponding to each fourth missing node; determine the fault node from the T fourth lost nodes based on the fault score corresponding to the fourth neighborhood node corresponding to each of the fourth lost nodes.

[0157] In the embodiments of the present application, the identification unit 802 in the fault node identification device 800 is configured to determine the average attribute change range of the attribute feature of the first neighborhood node of each of the first lost nodes. determine the attribute change fluctuation rate based on the average attribute change range corresponding to each of the first lost nodes and the difference in attribute feature of the first neighborhood node of each of the first lost nodes within adjacent periods. select K third lost nodes from the M second lost nodes based on the respective sizes of the average attribute vector and the domain trend fluctuation rate.

[0158] In the embodiments of the present application, the identification unit 802 in the fault node identification device 800 is configured to select, from the N first lost nodes, M second lost nodes corresponding to the average attribute change range less than a first threshold value and the attribute change fluctuation rate greater than a second threshold value; and select, from the N first lost nodes, M second lost nodes corresponding to the average attribute change range less than the first threshold value and the attribute change fluctuation rate less than the second threshold value.

[0159] In the embodiments of the present application, the identification unit 802 in the fault node identification device 800 is configured to determine an attribute vector of the attribute feature of the second neighborhood node of each of the second lost nodes; wherein the attribute vector is used to represent the attribute change direction per unit time. determine the average attribute vector of the second neighborhood node corresponding to each of the second lost nodes based on the attribute vector of each of the second neighborhood nodes. determine the neighborhood trend fluctuation rate based on the difference between the attribute vector of each of the second neighborhood nodes corresponding to each of the second lost nodes and the average attribute vector. select K third lost nodes from the M second lost nodes based on the respective sizes of the average attribute vector and the domain trend fluctuation rate.

[0160] In the embodiments of the present application, the identification unit 802 in the fault node identification device 800 is configured to select, from the M second lost nodes, K third lost nodes corresponding to the average attribute vector less than a third threshold value and the domain trend fluctuation rate greater than a fourth threshold value. select K third lost nodes from the M second lost nodes corresponding to the average attribute vector less than the third threshold value and the domain trend fluctuation rate less than the fourth threshold value.

[0161] In the embodiment of the application, the identification unit 802 in the fault node identification device 800 is configured to determine an attribute consistency deviation metric of each third lost node based on a difference between an attribute feature corresponding to each third lost node and an attribute feature of each third neighbor node corresponding to the third lost node. determine current state information of each third lost node based on a sum of the attribute consistency deviation metric and state information corresponding to a previous period, and a state parameter of whether each third lost node is online. Based on the size of the current state information corresponding to each third lost node, T fourth lost nodes are selected from K third lost nodes.

[0162] In the embodiment of the application, the identification unit 802 in the fault node identification device 800 is configured to select, from K third lost nodes, T fourth lost nodes corresponding to the current state information being greater than a fifth threshold value.

[0163] In the embodiment of the application, the identification unit 802 in the fault node identification device 800 is configured to determine a fault score of each fourth neighbor node for the fourth lost node based on a weighted sum of a difference between a heartbeat state of each fourth neighbor node and a predetermined value, a pose drift rate, and an attribute consistency deviation metric of the fourth neighbor node corresponding to each fourth lost node.

[0164] In the embodiment of the application, the identification unit 802 in the fault node identification device 800 is configured to determine an occupancy information of each fourth lost node corresponding to the fault score being greater than a sixth threshold value. Based on the size of the occupancy information corresponding to each fourth lost node, the fault node is determined from T fourth lost nodes.

[0165] In the embodiment of the application, the identification unit 802 in the fault node identification device 800 is configured to determine, from T fourth lost nodes, the fault node corresponding to the occupancy information being greater than a seventh threshold value.

[0166] In the embodiment of the application, the identification unit 802 in the fault node identification device 800 is configured to perform connectivity analysis on the attribute graph to identify and determine a plurality of initial neighbor node sets. Based on the attribute graph in a historical time period, N first lost nodes having no connection relationship with nodes in any initial neighbor node set are determined.

[0167] In this embodiment of the application, the attribute characteristics of each node in the fields of ontology perception and network communication include one or more of the following attributes: heartbeat state, path local density, remaining power, pose drift rate, communication channel quality, communication latency fluctuation rate, packet loss rate, and throughput.

[0168] It should be noted that, in the embodiments of this application, if the above-described fault node identification method is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, or the part that contributes to the related technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a fault node identification device (which may be a personal computer, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware and software combination.

[0169] Correspondingly, embodiments of this application provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps in the method on one side of the node identification device.

[0170] It should be noted that the descriptions of the storage medium and device embodiments above are similar to the descriptions of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0171] It should be noted that, Figure 6 A schematic diagram of a hardware entity of an electronic device provided in an embodiment of this application, such as... Figure 6 As shown, this application embodiment provides an electronic device 900, including a memory 902 and a processor 901. The memory 902 stores a computer program that can run on the processor 901. When the processor 901 executes the program, it implements the steps in the above-described method, wherein; The processor 901 typically controls the overall operation of the electronic device 900.

[0172] The memory 902 is configured to store instructions and applications executable by the processor 901, and can also cache data (e.g., image data, audio data, voice communication data, and video communication data) to be processed by the processor 901 and modules in the electronic device 900, and can be implemented by a FLASH or a Random Access Memory (RAM).

[0173] Correspondingly, the embodiments of the present application also provide a computer program product, which comprises a computer program executable by the processor 901 of the electronic device 900 to complete the steps in the method of the node identification device 800.

[0174] It should be understood that the "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that the size of the sequence number of the above processes does not mean the execution order in various embodiments of the present application, and the execution order of the processes should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The sequence number of the above embodiments of the present application is only for description, not representing the advantages and disadvantages of the embodiments.

[0175] It should be noted that in this document, the terms "comprising", "containing", or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or device. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0176] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the various components shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0177] The units described above as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units; they can be located in one place or distributed on multiple network units; and part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0178] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be realized in the form of hardware or in the form of hardware plus software functional unit.

[0179] Those skilled in the art can understand that all or part of the steps of the above method embodiments can be completed by program instruction related hardware, and the above program can be stored in a computer readable storage medium, and the program executes the steps including the above method embodiments when executed; and the above storage medium includes mobile storage device, read only memory (ReadOnly Memory, ROM), magnetic disc or optical disc and various storage program codes.

[0180] Alternatively, the integrated unit of the present application, if implemented in the form of a software function module and sold or used as an independent product, can also be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present application. The above storage medium includes mobile storage devices, ROM, magnetic discs or optical discs and various storage program codes.

[0181] The above is only an embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for identifying faulty nodes, characterized in that, include: Construct an attribute graph corresponding to multiple nodes; wherein the attribute graph is used to represent one or more of the following: the identification features of each node, the connection relationship features between each two nodes, and the attribute features of each node in the domain of ontology perception and network communication. Based on the attribute graph, N first lost-connection nodes are identified; wherein, the first lost-connection node is a node that has lost connection with its corresponding initial neighborhood node set; N is an integer greater than 0; Based on the consistency and trend of the attribute characteristics of each first neighboring node in the initial neighboring node set corresponding to each first lost node, and the difference between the attribute characteristics of each first neighboring node and the attribute characteristics of each first lost node, the faulty node is determined.

2. The fault node identification method according to claim 1, characterized in that, The method of determining the faulty node based on the consistency and trend of the attribute characteristics of each first neighboring node in the initial neighboring node set corresponding to each first lost node, and the difference between the attribute characteristics of each first neighboring node and the attribute characteristics of each first lost node, includes: Based on the consistency analysis results of the attribute characteristics of the first neighboring nodes, M second missing nodes are selected from the N first missing nodes; where M is an integer greater than 0 and less than N. Based on the trend analysis results of the attribute characteristics of the second neighboring nodes of each second lost node, K third lost nodes are selected from the M second lost nodes; where K is an integer greater than 0 and less than M. Based on the attribute characteristics of each of the third lost nodes and the difference between the attribute characteristics of each of the corresponding third neighboring nodes, T fourth lost nodes are selected from the K third lost nodes; where T is an integer greater than 0 and less than K. Based on the predetermined attribute features of the fourth neighboring nodes corresponding to each fourth lost node, and the attribute consistency deviation measure of the fourth neighboring nodes corresponding to each fourth lost node, the fault score of each fourth neighboring node for the fourth lost node is determined. Based on the fault score corresponding to the fourth neighboring node corresponding to each fourth lost node, the fault node is determined among the T fourth lost nodes.

3. The fault node identification method according to claim 2, characterized in that, The attribute graph is acquired periodically; the consistency analysis results include: the average magnitude of attribute changes and the volatility of attribute changes; based on the consistency analysis results of the attribute characteristics of the first neighboring nodes, M second missing nodes are selected from the N first missing nodes, including: Determine the average magnitude of attribute changes in the attribute features of the first neighboring nodes of each first lost node; The attribute change volatility is determined based on the average magnitude of the attribute change corresponding to each first lost node and the difference in attribute characteristics of the first neighboring nodes of each first lost node in adjacent periods. Based on the magnitudes of the average change in the attribute and the volatility of the attribute change, M second disconnected nodes are selected from the N first disconnected nodes.

4. The fault node identification method according to claim 3, characterized in that, Based on the magnitudes of the average change in the attribute and the volatility of the attribute change, M second lost-contact nodes are selected from the N first lost-contact nodes, including any one of the following: From the N first lost-connection nodes, select M second lost-connection nodes whose corresponding average change magnitude of the attribute is less than a first threshold and whose attribute change volatility is greater than a second threshold. From the N first lost-connection nodes, select M second lost-connection nodes whose average change in the corresponding attribute is less than the first threshold and whose volatility of the attribute change is less than the second threshold.

5. The fault node identification method according to claim 2, characterized in that, The trend analysis results include: the mean of the attribute vector and the neighborhood trend volatility; based on the trend analysis results of the attribute characteristics of the second neighboring nodes for each second missing node, K third missing nodes are selected from the M second missing nodes, including: Determine the attribute vector of the attribute characteristics of the second neighboring nodes of each second lost node; wherein the attribute vector is used to characterize the direction of attribute change per unit time; Based on the attribute vector of each second neighboring node, determine the mean value of the attribute vector of the second neighboring nodes corresponding to each second lost node; The neighborhood trend volatility is determined based on the difference between the attribute vector of each second neighborhood node corresponding to each second lost node and the mean of the attribute vector. Based on the magnitudes corresponding to the mean of the attribute vector and the volatility of the domain trend, K third disconnected nodes are selected from the M second disconnected nodes.

6. The fault node identification method according to claim 5, characterized in that, Based on the magnitudes corresponding to the mean of the attribute vector and the volatility of the domain trend, K third disconnected nodes are selected from the M second disconnected nodes, including any one of the following: From the M second disconnected nodes, select K third disconnected nodes whose corresponding attribute vector mean is less than the third threshold and whose domain trend volatility is greater than the fourth threshold. From the M second disconnected nodes, select K third disconnected nodes whose corresponding attribute vector mean is less than the third threshold and whose domain trend volatility is less than the fourth threshold.

7. The fault identification method according to claim 2, characterized in that, The step of selecting T fourth missing nodes from K third missing nodes based on the difference between the attribute features of each third missing node and the attribute features of its corresponding third neighbor node includes: Based on the difference between the attribute features corresponding to each third lost node and the attribute features of each corresponding third neighbor node, the attribute consistency deviation measure corresponding to each third lost node is determined. Based on the sum of the attribute consistency deviation metric and the status information corresponding to the previous period, and the status parameter of whether each third disconnected node is online, the current status information corresponding to each third disconnected node is determined. Based on the magnitude of the current status information corresponding to each of the third lost nodes, T fourth lost nodes are selected from the K third lost nodes.

8. The fault node identification method according to claim 7, characterized in that, The step of selecting T fourth lost nodes from K third lost nodes based on the magnitude of the current state information corresponding to each third lost node includes: Among the K third disconnected nodes, T fourth disconnected nodes are selected whose current status information is not less than the fifth threshold.

9. The fault node identification method according to claim 2, characterized in that, The step of determining the fault score of each fourth neighboring node for the fourth lost node based on the predetermined attribute features of the fourth neighboring nodes corresponding to each fourth lost node and the attribute consistency deviation measure of the fourth neighboring nodes corresponding to each fourth lost node includes: The fault score of each fourth neighboring node for the fourth lost node is determined by weighted summation based on the difference between the heartbeat state and the predetermined value of each fourth neighboring node, the pose drift rate, and the attribute consistency deviation metric of the fourth neighboring node corresponding to each fourth lost node.

10. The fault node identification method according to claim 2, characterized in that, The step of determining the faulty node among the T fourth lost-connection nodes based on the fault score corresponding to the fourth neighboring node for each fourth lost-connection node includes: Determine the percentage of each fourth disconnected node whose fault score is greater than the sixth threshold; Based on the proportion of information corresponding to each of the fourth lost-connection nodes, the faulty node is determined among the T fourth lost-connection nodes.

11. The fault node identification method according to claim 10, characterized in that, The step of determining the faulty node among the T fourth lost-connection nodes based on the proportion information corresponding to each of the fourth lost-connection nodes includes: Among the T fourth disconnected nodes, identify the faulty node whose corresponding proportion information is greater than the seventh threshold.

12. The fault node identification method according to any one of claims 1 to 11, characterized in that, The process of identifying N first disconnected nodes based on the attribute graph includes: Connectivity analysis is performed on the attribute graph to identify and determine multiple initial neighborhood node sets; Based on the attribute graph within the historical time period, N first disconnected nodes that have no connection relationship with any of the nodes in the initial neighborhood node set are determined.

13. The fault node identification method according to any one of claims 1 to 11, characterized in that, Each node's attribute characteristics in the fields of ontology perception and network communication include one or more of the following attributes: heartbeat state, path local density, remaining battery power, pose drift rate, communication channel quality, communication delay fluctuation rate, packet loss rate, and throughput rate.

14. A fault node identification device, characterized in that, include: An acquisition unit is used to construct an attribute graph corresponding to multiple nodes; wherein the attribute graph is used to represent one or more of the following: the identification features of each node, the connection relationship features between each two nodes, and the attribute features of each node in the fields of ontology perception and network communication. The identification unit is used to identify and determine N first lost-connection nodes based on the attribute graph; wherein, the first lost-connection node is a node that has lost connection with the corresponding initial neighborhood node set; N is an integer greater than 0; The identification unit is further configured to determine the faulty node based on the consistency and trend of the attribute features of each first neighboring node in the initial neighboring node set corresponding to each first lost node, and the difference between the attribute features of each first neighboring node and the attribute features of each first lost node.

15. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program that can run on the processor, the processor executing the computer program to implement the steps of the method according to any one of claims 1 to 13.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 13.

17. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 13.

Images