Distributed operation and maintenance intelligent management method and system based on cloud computing and Internet of Things

Abstract

The invention provides a distributed operation and maintenance intelligent management method and system based on cloud computing and the Internet of Things, and relates to the technical field of operation and maintenance of the Internet of Things, firstly, an inference fault knowledge chain of potential fault activities is accurately acquired and screened, concerned target fault activities are determined, and the pertinence of fault recognition is improved. The target Internet of Things interaction event flow is coded and expressed, the fault logic diagram coding vector and the fault penetration node vector are generated and further stacked to form the fault logic flow transfer vector, fault knowledge and event flow characteristics are effectively fused, and the depth and accuracy of fault analysis are enhanced. And finally, generating a target stack vector by combining the first interaction path vector of the target Internet of Things interaction event flow with the fault logic flow vector, and predicting a fault root cause description tag of the target fault activity accordingly, thereby significantly improving the speed and accuracy of fault positioning, and improving the accuracy of fault positioning. And data support is provided for operation and maintenance management in cloud computing and Internet of Things environments.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things operation and maintenance technology, and in particular to a distributed operation and maintenance intelligent management method and system based on cloud computing and the Internet of Things. Background Art

[0002] With the rapid development of cloud computing and IoT technologies, distributed systems have become the mainstream architecture for information processing and service provision. In these complex distributed environments, the interactive event flows between IoT devices are becoming increasingly large and complex, resulting in unprecedented challenges for operation and maintenance management. Traditional operation and maintenance methods often rely on manual experience and simple monitoring tools, which are difficult to effectively deal with the fault location and management problems in large-scale, highly dynamic IoT interactive event flows.

[0003] Especially in cloud computing environments, the interactive event flows between IoT devices involve numerous potential fault activities, which are often interrelated and mutually influential, forming a complex fault chain. In order to accurately locate and resolve these faults, operation and maintenance personnel need to have a deep understanding of the logical relationships and propagation paths of the fault activities, which is often time-consuming, labor-intensive and error-prone in actual operations. Summary of the invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things, the method comprising: Obtaining the inference fault knowledge chain corresponding to each potential fault activity in the target IoT interaction event stream in the cloud computing environment, determining a concerned fault knowledge chain from a plurality of the inference fault knowledge chains, wherein the concerned fault knowledge chain is used to guide the location of the concerned target fault activity among the plurality of the potential fault activities; The target IoT interaction event stream is respectively encoded according to the concerned fault knowledge chain and each of the inference fault knowledge chains, a plurality of fault logic diagram encoding vectors are generated, and the fault penetration node vectors corresponding to the concerned fault knowledge chain and each of the inference fault knowledge chains are extracted; Stacking each of the fault logic diagram encoding vectors with the corresponding fault penetration node vectors to generate a plurality of fault logic flow vectors; Extracting a first interaction path vector of the target Internet of Things interaction event flow, stacking the first interaction path vector and a plurality of the fault logic flow vectors to generate a target stacking vector; Predictions are made based on the target stacked vector to generate a target fault root cause description label of the target fault activity.

[0005] In a possible implementation manner of the first aspect, stacking the first interaction path vector and the plurality of fault logic flow vectors to generate a target stacking vector includes: Determine the knowledge chain length of each of the inference fault knowledge chains respectively, and stack the fault logic flow vectors corresponding to each of the inference fault knowledge chains in descending order of the knowledge chain lengths to generate a first stacking vector; The first interaction path vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain are stacked to generate a target stacking vector.

[0006] In a possible implementation manner of the first aspect, the target fault root cause description label is generated by a first deep learning network, and stacking the first interaction path vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain to generate a target stacking vector includes: Generate guidance information for guiding the first deep learning network to generate a fault root cause description label; A guidance knowledge vector of the guidance information is extracted, and the first interaction path vector, the guidance knowledge vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain are stacked to generate a target stacking vector.

[0007] In a possible implementation of the first aspect, the target IoT interaction event stream is respectively encoded according to the concerned fault knowledge chain and each of the reasoned fault knowledge chains to generate multiple fault logic diagram encoding vectors, including: Performing multi-stage encoding representation on the target Internet of Things interaction event stream to generate multi-stage encoding features of the target Internet of Things interaction event stream; According to the concerned fault knowledge chain and each of the inference fault knowledge chains, respectively, the multi-stage coding features are subjected to knowledge chain coding mapping to generate a plurality of multi-stage mapping coding features; The multi-stage mapping coding features are respectively coded and integrated to generate a plurality of fault logic diagram coding vectors.

[0008] In a possible implementation of the first aspect, acquiring the inference fault knowledge chain corresponding to each potential fault activity in the target IoT interaction event stream in the cloud computing environment, and determining a focus fault knowledge chain from a plurality of the inference fault knowledge chains, includes: Acquire a target Internet of Things interaction event stream and guidance identification information of the target Internet of Things interaction event stream, wherein the target Internet of Things interaction event stream includes multiple potential fault activities, and the guidance identification information is used to guide the location of a target fault activity of concern in the multiple potential fault activities; Fault knowledge chains are extracted from the target IoT interaction event stream to generate inference fault knowledge chains corresponding to each of the potential fault activities, and a focus fault knowledge chain is determined in each of the inference fault knowledge chains according to the guidance identification information.

[0009] In a possible implementation of the first aspect, the step of extracting a fault knowledge chain from the target IoT interaction event stream to generate a reasoning fault knowledge chain corresponding to each potential fault activity includes: Traversing the target IoT interaction event stream, adding each event in the target IoT interaction event stream as a node to an event association graph, and determining the association relationship between events according to the time sequence, event source, and event target attributes between events, and optimizing the event association graph after adding corresponding edges to the event association graph; A graph traversal algorithm is used to traverse the nodes in the optimized event association graph, and potential fault activities are identified according to the association relationship between events and the preset fault identification rules. The identified fault activities are represented in the form of a node set and added to the potential fault activity list; For each potential fault activity in the potential fault activity list, extract corresponding key features, wherein the key features include event type distribution, event time interval distribution, event source and target distribution, etc.; quantizing the key features into numerical data, combining them into a fault feature vector, and then normalizing the fault feature vector; For each fault feature vector of a potential fault activity, a similarity calculation algorithm is used to find the most similar fault feature in a pre-built fault knowledge base, and corresponding fault knowledge data is obtained based on the found fault features, and a reasoning fault knowledge chain is constructed based on the association relationship between fault types and the logical relationship between fault causes in the fault knowledge data; For example, in a possible implementation of the first aspect, the steps of respectively encoding the target IoT interaction event stream according to the concerned fault knowledge chain and each of the inference fault knowledge chains to generate a plurality of fault logic diagram encoding vectors, and extracting the fault penetration node vectors corresponding to the concerned fault knowledge chain and each of the inference fault knowledge chains include: Uniquely identify each event in the target IoT interaction event stream, and construct an initial empty matrix as an event stream-knowledge chain association matrix; Traverse each event in the target IoT interaction event stream, and for each event, analyze whether the event appears in the concerned fault knowledge chain or any reasoning fault knowledge chain. If so, mark the corresponding position in the event stream-knowledge chain association matrix as 1, otherwise mark it as 0; The target IoT interactive event stream is divided into multiple stages by using a time window partitioning or event type clustering method. For each stage, the key features of the events in the stage are extracted, and the key features are encoded as numerical data to form the multi-stage coding features of the stage. The multi-stage coding features of all stages are sequentially combined to obtain the overall multi-stage coding features of the target IoT interactive event stream. The key features include event type, event frequency, and time interval between events. For each reasoning fault knowledge chain, determining the knowledge chain node associated with the event in the target IoT interaction event stream according to the event stream-knowledge chain association matrix; For each associated knowledge chain node, knowledge chain coding mapping is performed according to the position and function of the knowledge chain node in the knowledge chain and the multi-stage coding features of the corresponding event. The mapping process uses a neural network mapping method to convert the multi-stage coding features of the corresponding event into mapping coding features corresponding to the knowledge chain node. For each reasoning fault knowledge chain, the mapping coding features of all associated knowledge chain nodes are combined in sequence to form the multi-stage mapping coding features corresponding to the reasoning fault knowledge chain. Inputting each multi-stage mapping coding feature into a recurrent neural network, so that the recurrent neural network generates a fault logic diagram coding vector by learning and extracting the association relationship and timing information between each multi-stage mapping coding feature; Analyze the fault propagation path and fault penetration node of each reasoning fault knowledge chain, wherein the fault penetration node refers to a key node that can affect or trigger other faults when the fault propagates in the reasoning fault knowledge chain; According to the position and function of the fault permeation node in the reasoning fault knowledge chain and the concerned fault knowledge chain, and the association relationship with other fault permeation nodes, the key features of the fault permeation node are extracted, and the key features are encoded into numerical data to form a fault permeation node vector, wherein the key features include node type, node degree, and path length between nodes; For each reasoning fault knowledge chain, the corresponding fault logic diagram encoding vector and the fault penetration node vector are stacked to obtain the fault logic flow vector corresponding to the reasoning fault knowledge chain. The stacking process uses an attention mechanism or a gating mechanism to dynamically adjust the weights of the fault logic diagram encoding vector and the fault penetration node vector.

[0010] In a possible implementation manner of the first aspect, the target fault root cause description label is generated by a first deep learning network, and the training step of the first deep learning network includes: Acquire a template IoT interaction event stream and a second fault knowledge chain corresponding to a template fault activity in the template IoT interaction event stream, extract the fault knowledge chain from the template IoT interaction event stream, and generate a first fault knowledge chain corresponding to each fault interaction activity in the template IoT interaction event stream, wherein the template fault activity is one of the multiple fault interaction activities; Extracting a second interaction path vector of the template IoT interaction event flow, encoding the template IoT interaction event flow according to the second fault knowledge chain and each of the first fault knowledge chains, generating a plurality of template encoding features, and extracting a template fault penetration node vector corresponding to the second fault knowledge chain and each of the first fault knowledge chains; The second interaction path vector, the template encoding feature, and the template fault penetration node vector are stacked and loaded into the first deep learning network for prediction to generate a fault root cause description prediction result, wherein the fault root cause description prediction result is used to determine a fault root cause description label of the template fault activity; Obtain a first fault root cause description label of a template fault knowledge point associated with the template IoT interaction event flow, determine a training error based on the fault root cause description prediction result and the first fault root cause description label, optimize the first deep learning network based on the training error, and use the template fault knowledge point to guide the positioning of the template fault activity.

[0011] In a possible implementation of the first aspect, the template IoT interaction event flow, the second fault knowledge chain, and the first fault root cause description label are all acquired from a sample database, and before acquiring the template IoT interaction event flow and the second fault knowledge chain corresponding to the template fault activity in the template IoT interaction event flow, the method further includes: Acquire multiple basic Internet of Things interaction event streams and fault attention knowledge data corresponding to each of the basic Internet of Things interaction event streams, and determine key interaction point data corresponding to each of the basic Internet of Things interaction event streams based on each of the basic Internet of Things interaction event streams and the corresponding fault attention knowledge data, wherein the fault attention knowledge data is used to guide and locate key fault activities in the corresponding basic Internet of Things interaction event streams; Acquire multiple reference fault knowledge points, and determine, according to each of the key interaction point data, the associated fault knowledge points corresponding to each of the basic Internet of Things interaction event flows in the multiple reference fault knowledge points; According to each of the basic Internet of Things interaction event flows and the corresponding fault concern knowledge data, respectively determine the fault knowledge chain label data of each of the basic Internet of Things interaction event flows, and the fault knowledge chain label data is used to guide the positioning of the corresponding key fault activities; Each of the basic Internet of Things interaction event streams, the corresponding fault knowledge chain label data, and the corresponding associated fault knowledge points are integrated and added to the sample database; the template Internet of Things interaction event stream is randomly extracted from a plurality of the basic Internet of Things interaction event streams; the second fault knowledge chain is the fault knowledge chain label data corresponding to the template Internet of Things interaction event stream; and the template fault knowledge point is the associated fault knowledge point associated with the template Internet of Things interaction event stream.

[0012] In a possible implementation of the first aspect, determining the fault knowledge chain label data of each of the basic Internet of Things interaction event flows according to each of the basic Internet of Things interaction event flows and the corresponding fault attention knowledge data includes: Loading each of the fault concern knowledge data into the second deep learning network for prediction, and generating fault summary description data corresponding to each of the basic Internet of Things interaction event flows; Perform fault activity location according to each of the basic Internet of Things interaction event flows and the corresponding fault summary description data, and generate a basic fault logic path corresponding to each of the basic Internet of Things interaction event flows, wherein the basic fault logic path is used to guide the location of the corresponding key fault activity; Each of the basic Internet of Things interaction event flows and the corresponding basic fault logic paths are loaded into a first fault knowledge chain extraction network to extract the fault knowledge chain, and generate fault knowledge chain label data corresponding to each of the basic Internet of Things interaction event flows.

[0013] On the other hand, an embodiment of the present invention also provides a distributed operation and maintenance intelligent management system based on cloud computing and the Internet of Things, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0014] Based on the above aspects, the embodiments of the present application realize efficient fault management and positioning of the interactive event stream of the Internet of Things in the cloud computing environment. First, the inference fault knowledge chain of potential fault activities is accurately obtained and screened, the target fault activities of concern are determined, and the pertinence of fault identification is improved. By encoding and representing the target Internet of Things interactive event stream, and generating a fault logic graph encoding vector and a fault penetration node vector, and further stacking to form a fault logic flow vector, the fault knowledge and event flow characteristics are effectively integrated, and the depth and accuracy of fault analysis are enhanced. Finally, by combining the first interactive path vector and the fault logic flow vector of the target Internet of Things interactive event stream, a target stacking vector is generated, and the fault root cause description label of the target fault activity is predicted based on this, which significantly improves the speed and accuracy of fault location, and provides data support for operation and maintenance management in cloud computing and Internet of Things environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the execution flow of the distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things provided by an embodiment of the present invention.

[0016] Figure 2 It is a schematic diagram of the hardware architecture of a distributed operation and maintenance intelligent management system based on cloud computing and the Internet of Things provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 The present invention is a flowchart of a distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things provided by an embodiment of the present invention. The distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things is introduced in detail below.

[0018] Step S110, obtaining the inference fault knowledge chain corresponding to each potential fault activity in the target IoT interaction event flow in the cloud computing environment, determining a focus fault knowledge chain from multiple inference fault knowledge chains, and the focus fault knowledge chain is used to guide the positioning of the target fault activity of concern in multiple potential fault activities.

[0019] In this embodiment, the intelligent operation and maintenance scenario of radio monitoring facilities based on distributed mobile terminals is taken as an example. The radio monitoring facilities include numerous IoT devices, which interact with each other to realize the monitoring function. For example, the sensor device transmits the collected radio signal data to the data processing unit, and the data processing unit interacts with the storage device to store the data. The cloud computing environment carries the event stream generated by the interaction of these IoT devices.

[0020] The server obtains the target IoT interaction event stream in the cloud computing environment. This target IoT interaction event stream includes the acquisition process of the reasoning fault knowledge chain corresponding to multiple potential fault activities. Assuming that there is a sensor node in the radio monitoring facility, it will normally send the monitored radio frequency band data to the data aggregation node at regular intervals. If the data transmission of this sensor node is not received for a period of time, this may be a potential fault activity.

[0021] For this potential fault activity, the server builds a reasoning fault knowledge chain by analyzing historical data, equipment operation manuals, known fault modes, etc. This knowledge chain may include a series of possible fault factors and their associations, from equipment hardware failure (such as sensor component damage) to software failure (such as data transmission protocol error) to network level failure (such as network connection interruption between sensor and aggregation node).

[0022] Similarly, for other potential fault activities, such as the data processing unit processing data too slowly, the storage device failing to write data, etc., the server also builds corresponding reasoning fault knowledge chains.

[0023] Then, the concerned fault knowledge chain is determined among the numerous reasoning fault knowledge chains. For example, the current operation and maintenance personnel report that the accuracy of the monitoring data in a specific area has seriously decreased. The server determines the concerned fault knowledge chain related to this problem in each reasoning fault knowledge chain based on this guidance identification information (corresponding to the problem of decreased monitoring data accuracy in this specific area). If the accuracy of the monitoring data has decreased, the concerned fault knowledge chain may focus on the accuracy of the sensor data collection, the data integrity during the data transmission process, and the data calibration in the data processing unit. Fault factors and their correlations, this concerned fault knowledge chain will guide the positioning of the target fault activities that are concerned among multiple potential fault activities, that is, those fault activities that are directly or indirectly related to the decrease in the accuracy of the monitoring data.

[0024] Step S120, respectively perform encoding representation of the target IoT interaction event stream according to the concerned fault knowledge chain and each of the inference fault knowledge chains, generate multiple fault logic diagram encoding vectors, and extract the fault penetration node vectors corresponding to the concerned fault knowledge chain and each of the inference fault knowledge chains.

[0025] In this embodiment, each event in the target IoT interaction event stream, such as an event in which a sensor collects a radio signal strength of a certain value, or an event in which a data processing unit spends a specific amount of time processing a batch of data, needs to be encoded and represented.

[0026] The server first performs multi-stage encoding representation on the target IoT interactive event stream. For example, the entire interactive event stream is divided into multiple time stages according to the chronological order of events, and the key features of the events are encoded in each stage. For sensor acquisition events, key features may include acquisition time, signal strength, signal frequency, etc.; for data processing events, key features may include processing data volume, processing time, processing algorithm, etc. After these key features are encoded as numerical data, the multi-stage encoding features of the target IoT interactive event stream are generated.

[0027] Next, the multi-stage coding features are mapped to knowledge chains based on the concerned fault knowledge chain and each reasoned fault knowledge chain. Taking the concerned fault knowledge chain as an example, if the concerned fault knowledge chain emphasizes the accuracy of sensor data, then the multi-stage coding features of the sensor acquisition event will be mapped according to the fault knowledge chain rules related to the accuracy of sensor data. For example, if the knowledge chain indicates that abnormal fluctuations in signal strength are related to sensor aging, then in the coding mapping process, the feature of signal strength will have a specific mapping code in the dimension related to sensor aging.

[0028] Then, the multi-stage mapping coding features are encoded and integrated to generate multiple fault logic diagram coding vectors. For example, the multi-stage mapping coding features related to sensor acquisition, data processing, data storage and other links are integrated according to certain logic and algorithms to obtain a coding vector that can comprehensively reflect the fault logic relationship.

[0029] At the same time, the server also needs to extract the fault penetration node vectors corresponding to the concerned fault knowledge chain and each inference fault knowledge chain. In radio monitoring facilities, fault penetration nodes may be those nodes that will affect multiple subsequent links once a fault occurs. For example, if the data verification module in the data processing unit fails, it may cause erroneous data to flow into the storage device, thereby affecting the accuracy of the entire monitoring data. For this data verification module, its key features include module type (data verification), connection relationship with other modules (receiving data processed from sensors and outputting data to storage devices), etc. Encoding these key features as numerical data forms a fault penetration node vector.

[0030] Step S130 , stacking each of the fault logic diagram coding vectors with the corresponding fault penetration node vectors to generate a plurality of fault logic flow vectors.

[0031] Taking the sensor acquisition link as an example, it is assumed that the previously generated sensor acquisition-related fault logic diagram encoding vector contains the fault logic information of various aspects of sensor acquisition, and the corresponding fault penetration node vector contains the key information of the fault penetration node such as the sensor element.

[0032] The server stacks these two vectors through a specific algorithm (such as weighted summation, etc.). This stacking process is like integrating the fault logic information in the sensor acquisition link with the key node information that may cause the fault to propagate, forming a fault logic flow vector that more comprehensively reflects the fault flow of the sensor acquisition link. Similarly, for other links such as data processing and data storage, the fault logic diagram encoding vector and the fault penetration node vector are stacked to generate the fault logic flow vector of each link.

[0033] Step S140: extracting a first interaction path vector of the target IoT interaction event flow, stacking the first interaction path vector and a plurality of the fault logic flow vectors, and generating a target stacking vector.

[0034] In this embodiment, the first interaction path vector reflects the order and path relationship of the interaction between IoT devices. For example, the sensor transmits data to the data processing unit, and the data processing unit transmits the processed data to the storage device. This transmission order and the device relationship involved constitute the first interaction path vector.

[0035] Then, the server stacks the first interaction path vector and multiple fault logic flow vectors to generate a target stack vector. For example, the fault logic flow vectors corresponding to each inference fault knowledge chain are stacked in descending order of the knowledge chain length of each inference fault knowledge chain to generate a first stack vector. If the knowledge chain length corresponding to the fault logic flow vector related to the data processing unit is longer in the previous fault logic flow vector, it may be stacked first.

[0036] Next, the first interaction path vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain are stacked. This process is like integrating the overall path information of device interaction, the fault flow information of each link, and the fault flow information of the focused fault link to form a target stacking vector.

[0037] Step S150 , performing prediction based on the target stacking vector to generate a target fault root cause description label of the target fault activity.

[0038] For example, if the target fault activity is the decrease in the accuracy of monitoring data, the target fault root cause description tag may include a detailed fault analysis starting from the sensor collection link, which may describe the external interference (such as signal interference from other nearby radio devices) when the sensor collects data, how this interference causes deviations in the data collected by the sensor, and when the deviated data passes through the data processing unit, because a certain algorithm in the data processing unit does not effectively correct this deviation (it may be the limitation of the algorithm itself or the unreasonable setting of the algorithm parameters), resulting in large errors in the processed data, and then these erroneous data are stored in the storage device, which ultimately leads to a decrease in the accuracy of the monitoring data. At the same time, this target fault root cause description tag may also include the impact of data packet loss during network transmission on data accuracy, and the impact of time synchronization problems in the entire IoT interactive system on data accuracy. Complex logical relationships and detailed descriptions of fault factors.

[0039] Therefore, in the intelligent operation and maintenance scenario of radio monitoring facilities based on distributed mobile terminals, it is possible to accurately analyze the fault activities in the target IoT interaction event stream, find the root cause of the target fault activity and generate a detailed target fault root cause description label.

[0040] Based on the above steps, the embodiment of the present application realizes efficient fault management and positioning of the interactive event stream of the Internet of Things in the cloud computing environment. First, the inference fault knowledge chain of potential fault activities is accurately obtained and screened, the target fault activities of concern are determined, and the pertinence of fault identification is improved. By encoding and representing the target Internet of Things interactive event stream, and generating a fault logic graph encoding vector and a fault penetration node vector, and further stacking to form a fault logic flow vector, the fault knowledge and event flow characteristics are effectively integrated, and the depth and accuracy of fault analysis are enhanced. Finally, by combining the first interactive path vector and the fault logic flow vector of the target Internet of Things interactive event stream, a target stacking vector is generated, and the fault root cause description label of the target fault activity is predicted based on this, which significantly improves the speed and accuracy of fault location, and provides data support for operation and maintenance management in cloud computing and Internet of Things environments.

[0041] In a possible implementation, step S140 includes: Step S141 , respectively determining the knowledge chain length of each of the inference fault knowledge chains, and stacking the fault logic flow vectors corresponding to each of the inference fault knowledge chains in descending order of the knowledge chain lengths to generate a first stacking vector.

[0042] Step S142: stacking the first interaction path vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain to generate a target stacking vector.

[0043] In a possible implementation manner, the target fault root cause description label is generated by a first deep learning network, and step S142 includes: Step S1421, generating guidance information for guiding the first deep learning network to generate a fault root cause description label.

[0044] Step S1422: extract the guidance knowledge vector of the guidance information, stack the first interaction path vector, the guidance knowledge vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain to generate a target stacking vector.

[0045] In this embodiment, the knowledge chain length of each reasoning fault knowledge chain is first determined. For example, taking the interaction between the aforementioned sensor, data processing unit and storage device as an example, different fault reasoning knowledge chains may exist in each link. For example, for the sensor link, its reasoning fault knowledge chain may start from the sensor hardware itself (such as component aging, damage, etc.), extend to its connection with peripheral devices (such as loose connection lines, interface failures, etc.), and then to possible external interference (such as electromagnetic interference, etc.), etc., which constitutes a longer knowledge chain. For the data processing unit, its reasoning fault knowledge chain may be more concentrated on internal algorithms (such as algorithm errors, improper parameter settings, etc.), software operating environment (such as operating system failures affecting data processing), etc., and the length of the knowledge chain may be relatively short.

[0046] Assuming that the inference fault knowledge chain of the sensor link is the longest, the fault logic flow vector related to the sensor will be stacked first. This fault logic flow vector contains all the logical information of the sensor link from data collection to possible faults and propagation. When stacking in descending order of knowledge chain length, the fault logic flow vectors of the data processing unit, storage device and other links are stacked in sequence. This process is like integrating the fault logic flow information of each link in the order from the most complex and comprehensive fault logic relationship to the relatively simple one to form the first stacking vector.

[0047] The first interaction path vector reflects the interaction sequence between IoT devices in the radio monitoring facility, such as the complete interaction process information of the sensor collecting data and transmitting it to the data processing unit, and the data processing unit then transmitting the processed data to the storage device. This interaction path vector is stacked with the first stacking vector generated previously and the fault logic flow vector corresponding to the concerned fault knowledge chain. If the concerned fault knowledge chain revolves around data accuracy, then its corresponding fault logic flow vector contains the fault logic relationships related to data accuracy that may affect data accuracy in each link from sensor collection to final storage. Through such stacking, the target stacking vector integrates multiple aspects of information such as the device interaction path, the fault logic flow of each link, and the fault logic relationship of focus.

[0048] When the target fault root cause description label is generated by the first deep learning network, in the process of stacking to generate the target stacking vector, the server must first generate guidance information for guiding the first deep learning network to generate the fault root cause description label. In the intelligent operation and maintenance scenario of radio monitoring facilities, the guidance information may come from the operation and maintenance personnel's preliminary description of the fault phenomenon, specific monitoring index requirements, or key performance parameters of equipment operation. For example, the operation and maintenance personnel reported that the radio signal data of a specific frequency band monitored during a certain period of time fluctuated abnormally, and this information is part of the guidance information. At the same time, the monitoring facilities have specific index requirements for data accuracy, data integrity, etc., and these requirements also constitute guidance information.

[0049] Among them, the server then extracts the guidance knowledge vector of the guidance information. This guidance knowledge vector is obtained by quantifying and encoding various contents in the guidance information. For example, for the information of abnormal signal data fluctuations reported by the operation and maintenance personnel, the amplitude range of the fluctuations, the frequency of the fluctuations, etc. can be quantified and encoded. For the indicator requirements of the monitoring facilities, such as the data accuracy requirement of more than 99%, this 99% and other related indicator values ​​can be encoded. Then the first interactive path vector, the guidance knowledge vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain are stacked to generate a target stacking vector. In this process, the addition of the guidance knowledge vector is like injecting specific target-oriented information into the entire stacking process. The first interactive path vector provides the basic framework for device interaction, the first stacking vector contains the fault logic flow of each link, the fault logic flow vector corresponding to the concerned fault knowledge chain focuses on the fault logic relationship of focus, and the guidance knowledge vector guides the deep learning network to generate fault root cause description labels in the direction that meets the operation and maintenance requirements and fault phenomenon characteristics. Through such stacking, the target stacking vector integrates all necessary information so that the subsequent first deep learning network can accurately generate a complex target fault root cause description label based on this target stacking vector. This label will describe in detail the root cause of the fault in the entire process from sensor acquisition to data storage and the logical relationship between various factors.

[0050] In a possible implementation, step S120 may include: The target Internet of Things interaction event stream is represented by multi-stage coding to generate multi-stage coding features of the target Internet of Things interaction event stream.

[0051] According to the concerned fault knowledge chain and each of the inference fault knowledge chains, knowledge chain encoding mapping is performed on the multi-stage encoding features to generate a plurality of multi-stage mapping encoding features.

[0052] The multi-stage mapping coding features are respectively coded and integrated to generate a plurality of fault logic diagram coding vectors.

[0053] In this embodiment, the IoT interactive event stream covers the interaction process between many devices, such as the interaction between sensors and data processing units, data processing units and storage devices, etc. The server divides this interactive event stream into multiple stages according to the time sequence or specific task flow. For example, taking the sensor collecting radio signals and transmitting them to the data processing unit as an example, the initial calibration stage after the sensor is started can be set as the first stage, the stage of normal data collection and transmission can be set as the second stage, and the adjustment stage when encountering external interference can be set as the third stage, etc.

[0054] For events in each stage, the server extracts its key features for encoding. In the initial calibration stage of the sensor, key features may include the sensor's startup time, initial self-test parameters (such as voltage, current, etc.), the version of the calibration algorithm, etc. These features are quantified into numerical data to form the encoding representation of this stage. In the normal acquisition and transmission stage, the key features include the strength, frequency, acquisition time interval, data transmission rate, etc. of the collected radio signal. For the adjustment stage when encountering external interference, the key features may be the time when the interference is detected, the type of interference (such as the frequency band of electromagnetic interference, etc.), the adjustment strategy adopted by the sensor (such as the parameters corresponding to the operation of adjusting gain, etc.). Combining these encodings of each stage generates the multi-stage encoding features of the target IoT interaction event stream.

[0055] Next, the server performs knowledge chain encoding mapping on the multi-stage encoding features according to the concerned fault knowledge chain and each reasoning fault knowledge chain, and generates multiple multi-stage mapping encoding features. In the intelligent operation and maintenance scenario of radio monitoring facilities, if the concerned fault knowledge chain focuses on data accuracy, each element in the multi-stage encoding feature must be mapped according to this concerned fault knowledge chain. Taking the sensor data collection stage as an example, in the multi-stage encoding feature, if the concerned fault knowledge chain indicates that the signal strength abnormality may be related to sensor aging, then in the knowledge chain encoding mapping process, the signal strength value will be mapped to the encoding dimension related to sensor aging. If the signal strength is within the normal range, it may be mapped to a value indicating a normal aging state. If the signal strength is too low and exceeds the normal fluctuation range, it may be mapped to a value indicating that sensor aging seriously affects the accuracy of data collection.

[0056] The same is true for each reasoning fault knowledge chain. For example, there is a reasoning fault knowledge chain about the impact of network transmission on data accuracy. When performing knowledge chain encoding mapping, if the transmission rate is lower than the normal level, the data transmission rate element in the multi-stage coding feature may be mapped to a coding dimension that indicates that network congestion or network equipment failure affects data transmission and thus affects data accuracy according to this reasoning fault knowledge chain. In this way, each element in each multi-stage coding feature is mapped according to the concerned fault knowledge chain and each reasoning fault knowledge chain, thereby generating multiple multi-stage mapping coding features.

[0057] Finally, the server encodes and integrates each multi-stage mapping coding feature to generate multiple fault logic diagram coding vectors. In radio monitoring facilities, taking the operation and maintenance of sensors, data processing units and storage devices as an example, for the multi-stage mapping coding features related to sensors, the server integrates the codes mapped at different stages according to specific logic and algorithms. For example, the multi-stage mapping coding features related to sensors such as the initial calibration stage of the sensor, the normal acquisition and transmission stage, and the adjustment stage when encountering external interference are weighted summed or combined according to specific logical rules. In this process, different weights may be assigned according to the importance of the impact of different stages on the overall fault logic. For example, the impact weight of the normal acquisition and transmission stage on the overall fault logic may be higher because this stage lasts for a long time and the accuracy of data collection is directly related to the operation effect of the entire monitoring facility.

[0058] Similarly, similar encoding integration operations are performed for the multi-stage mapping encoding features related to the data processing unit and the storage device. Through such an integration process, the server generates multiple fault logic diagram encoding vectors, which can comprehensively reflect the fault logic relationship of each link (such as sensors, data processing units, storage devices, etc.) in the entire IoT interactive event flow based on the attention to the fault knowledge chain and the reasoning fault knowledge chain, providing an important data basis for subsequent fault analysis and root cause location.

[0059] In a possible implementation, step S110 includes: Step S111, obtaining a target IoT interaction event stream and guidance identification information of the target IoT interaction event stream, wherein the target IoT interaction event stream includes multiple potential fault activities, and the guidance identification information is used to guide the location of a target fault activity of concern among the multiple potential fault activities.

[0060] Step S112, extracting fault knowledge chains from the target IoT interaction event stream, generating inference fault knowledge chains corresponding to each of the potential fault activities, and determining a focus fault knowledge chain in each of the inference fault knowledge chains according to the guidance identification information.

[0061] In a possible implementation, step S112 includes: Step S1121, traverse the target IoT interaction event stream, add each event in the target IoT interaction event stream as a node to the event association graph, and determine the association relationship between events according to the time sequence, event source and event target attributes between events, and after adding corresponding edges in the event association graph, optimize the event association graph.

[0062] Step S1122, using a graph traversal algorithm to traverse the nodes in the optimized event association graph, identifying potential fault activities based on the association relationship between events and preset fault identification rules, representing the identified fault activities in the form of a node set, and adding them to the potential fault activity list.

[0063] Step S1123: For each potential fault activity in the potential fault activity list, extract corresponding key features, wherein the key features include event type distribution, event time interval distribution, event source and target distribution, etc.

[0064] Step S1124, quantizing the key features into numerical data, combining them into a fault feature vector, and then normalizing the fault feature vector.

[0065] Step S1125, for the fault feature vector of each potential fault activity, a similarity calculation algorithm is used to search for the most similar fault feature in the pre-built fault knowledge base, and based on the found fault feature, the corresponding fault knowledge data is obtained, and based on the association relationship between the fault types in the fault knowledge data and the logical relationship between the fault causes, an inference fault knowledge chain is constructed.

[0066] In this embodiment, the target IoT interaction event stream includes interaction events between many devices, such as events such as a sensor device sending radio signal data to a data processing unit, and a data processing unit storing the data in a storage device after processing the data. The guidance identification information may come from the needs of the operation and maintenance personnel or specific monitoring indicators of the monitoring system. For example, the operation and maintenance personnel found that the accuracy of the radio monitoring data has decreased recently. This information about the decrease in data accuracy is the guidance identification information, which will guide the server to locate the target fault activity related to the decrease in data accuracy among many potential fault activities.

[0067] Next, in the process of fault knowledge chain extraction, the target IoT interaction event stream can be traversed, and each event in it can be added as a node to the event association graph. Taking the event that the sensor sends data to the data processing unit as an example, the event that the sensor sends data is a node, and the event that the data processing unit receives data is also a node. Then, the association relationship between the events is determined according to the time sequence between the events, the event source and the event target attributes, and the corresponding edges are added to the event association graph. If the sensor sends data first and the data processing unit receives data later, then an edge from the sensor sending data node to the data processing unit receiving data node is established according to this time sequence. In addition, since the sensor is the event source and the data processing unit is the event target, this also clarifies the direction of the edge. Afterwards, the server will optimize the event association graph, which may remove some redundant edges or merge some nodes with similar attributes.

[0068] After the optimized event association graph is formed, the server uses a graph traversal algorithm to traverse the nodes in it. Potential fault activities are identified based on the association between events and the preset fault identification rules. For example, according to the preset rules, if the data sent by the sensor exceeds the normal data range for multiple consecutive times, this may be a potential fault activity. All identified fault activities are represented in the form of a node set and added to the list of potential fault activities.

[0069] For each potential fault activity in the potential fault activity list, the server extracts the corresponding key features. In the radio monitoring facility scenario, taking the potential fault activity of abnormal sensor data transmission as an example, in terms of event type distribution, it may be that there is an error in the data transmission type (for example, data of other frequency bands were sent when data of a specific frequency band should have been sent); in terms of event time interval distribution, it may be that the time interval for sending data does not match the normal setting (for example, data is normally sent every 1 second, but now it is sent every 5 seconds); in terms of event source and target distribution, the source is the sensor and the target is the data processing unit. If the data always fails to reach the target accurately, this is also one of the key features.

[0070] Then, the server quantifies these key features into numerical data, combines them into a fault feature vector, and normalizes the fault feature vector. For example, the data transmission type error is converted into a numerical value according to the pre-set coding rules, such as the type error is coded as 1 and the normal is 0; for the time interval, the difference with the normal time interval is calculated and quantized into a numerical value; after these numerical values ​​are combined into a fault feature vector, the numerical values ​​in the vector are mapped to a specific interval, such as the [0,1] interval, through the normalization algorithm.

[0071] Finally, for each fault feature vector of a potential fault activity, the server uses a similarity calculation algorithm to search for the most similar fault feature in the pre-built fault knowledge base. In the fault knowledge base of the radio monitoring facility, feature data of various known faults are stored. For example, for the fault feature vector of abnormal sensor data transmission, the server will search for similar fault features in the knowledge base. If a similar fault feature is found, its corresponding fault knowledge data may contain the fault type when this fault occurred before (such as sensor hardware failure or software configuration error) and the logical relationship between the fault causes (such as hardware failure causing a change in data transmission frequency or software configuration error causing data transmission type error). Based on the association between these fault types and the logical relationship between the fault causes, the server constructs an inference fault knowledge chain. This inference fault knowledge chain may start from the failure of a certain component of the sensor hardware and deduce a series of logical relationships such as how it affects the data transmission type or time interval of the sensor, and then affects the accuracy of the data received by the data processing unit.

[0072] After constructing the inference fault knowledge chains corresponding to each potential fault activity, the server determines the concerned fault knowledge chain in these inference fault knowledge chains based on the previously obtained guidance identification information. Since the guidance identification information is about the decline in data accuracy, the inference fault knowledge chain related to data accuracy will be determined as the concerned fault knowledge chain in each inference fault knowledge chain. For example, those inference fault knowledge chains that contain logical relationships such as inaccurate sensor data collection affecting the accuracy of subsequent data processing and algorithm errors in the data processing unit causing a decline in data accuracy will be the focus of attention. This concerned fault knowledge chain will play an important role in guiding the location of target fault activities in the subsequent fault analysis process.

[0073] For example, in a possible implementation, step S120 may further include: Step S121, uniquely identify each event in the target IoT interaction event stream, and construct an initial empty matrix as an event stream-knowledge chain association matrix.

[0074] In this embodiment, the IoT interactive event stream contains many events, such as events of sensors collecting radio signals, events of data processing units receiving sensor data, and events of storage devices storing processed data. The server assigns a unique identifier to each such event, just like attaching a specific label to each event so that subsequent processing can accurately identify it. Then, an initially empty matrix is ​​created. This matrix is ​​the event stream-knowledge chain association matrix, which will be used to record the association between events and the concerned fault knowledge chain and each reasoning fault knowledge chain.

[0075] Step S122, traverse each event in the target IoT interaction event stream, and for each event, analyze whether the event appears in the concerned fault knowledge chain or any reasoning fault knowledge chain. If so, mark the corresponding position in the event stream-knowledge chain association matrix as 1, otherwise mark it as 0.

[0076] For each event, analyze whether the event appears in the concerned fault knowledge chain or any reasoning fault knowledge chain. In radio monitoring facilities, taking the sensor data collection event as an example, if the sensor data collection event is included in the concerned fault knowledge chain (assuming that the concerned fault knowledge chain is related to data accuracy) or a reasoning fault knowledge chain (such as the reasoning fault knowledge chain about the impact of sensor hardware failure on data collection), then the corresponding position in the event flow-knowledge chain association matrix is ​​marked as 1, indicating that there is an association; if it is not in any related knowledge chain, it is marked as 0. For example, if the data verification event in the data processing unit is related to the current concerned fault knowledge chain or a reasoning fault knowledge chain, the corresponding position in the matrix is ​​marked as 1, and if it is not related, it is marked as 0.

[0077] Step S123, using time window division or event type clustering method, divide the target IoT interaction event flow into multiple stages, for each stage, extract the key features of the events in the stage, and encode the key features into numerical data to form the multi-stage coding features of the stage, then combine the multi-stage coding features of all stages in sequence to obtain the overall multi-stage coding features of the target IoT interaction event flow, wherein the key features include event type, event frequency, and time interval between events.

[0078] For example, in terms of time window division, during the operation of radio monitoring facilities, the entire interactive event stream can be divided according to fixed time intervals, such as one time window per hour. For each stage, the key features of the events in the stage are extracted, and the key features are encoded as numerical data to form the multi-stage coding features of the stage. Then, the multi-stage coding features of all stages are combined in sequence to obtain the overall multi-stage coding features of the target IoT interactive event stream. For example, in a time window, the key features of the sensor data collection event include event type (collecting radio signals), event frequency (such as collecting once every 10 minutes), time interval between events (each collection interval is 10 minutes), etc. These features are converted into numerical data according to pre-set coding rules. For example, the event type can be represented by a specific number, the event frequency can be encoded according to the actual frequency value or the difference with the standard frequency, and the time interval between events is also quantified and encoded in a similar manner. This operation is performed for each stage, and then the coding features of each stage are combined in sequence to obtain the overall multi-stage coding features.

[0079] Step S124: for each reasoning fault knowledge chain, determine the knowledge chain node associated with the event in the target IoT interaction event flow according to the event flow-knowledge chain association matrix.

[0080] Assume that there is a reasoning fault knowledge chain about the impact of sensor faults on data transmission. According to the event stream-knowledge chain association matrix, if sensor startup events, sensor data sending events, etc. are associated with this reasoning fault knowledge chain, then the nodes corresponding to these events are the associated knowledge chain nodes in this reasoning fault knowledge chain.

[0081] Step S125, for each associated knowledge chain node, knowledge chain coding mapping is performed according to the position and function of the knowledge chain node in the knowledge chain, and the multi-stage coding features of the corresponding event. The mapping process adopts a neural network mapping method to convert the multi-stage coding features of the corresponding event into mapping coding features corresponding to the knowledge chain node. And, for each reasoning fault knowledge chain, the mapping coding features of all associated knowledge chain nodes are combined in sequence to form the multi-stage mapping coding features corresponding to the reasoning fault knowledge chain.

[0082] For example, in the inference fault knowledge chain where sensor faults affect data transmission, the sensor start node may be at the starting position in the knowledge chain, which has a certain impact on the subsequent fault propagation. If the multi-stage coding features of the sensor start event include the start time, self-test parameters at start-up, etc., the neural network mapping will convert these multi-stage coding features into mapping coding features suitable for the representation of this node in the knowledge chain according to the meaning of the sensor start node in the knowledge chain. For each inference fault knowledge chain, the mapping coding features of all associated knowledge chain nodes are combined in sequence to form the multi-stage mapping coding features corresponding to the inference fault knowledge chain.

[0083] Step S126, input each multi-stage mapping coding feature into a recurrent neural network, so that the recurrent neural network generates a fault logic diagram coding vector by learning and extracting the association relationship and timing information between each multi-stage mapping coding feature.

[0084] For the reasoning fault knowledge chain related to the interactive event flow involving multiple links such as sensors, data processing units, and storage devices, the recurrent neural network will analyze the relationship between the multi-stage mapping coding features of events related to different stages and different devices. For example, there may be a temporal sequence relationship and a logical causal relationship between the multi-stage mapping coding features of sensor data collection and the multi-stage mapping coding features of data processing units. The recurrent neural network can learn these relationships and integrate them into the fault logic diagram coding vector. This fault logic diagram coding vector can comprehensively reflect the fault logic relationship in the entire reasoning fault knowledge chain.

[0085] Step S127 , analyzing the fault propagation path and fault penetration nodes of each inference fault knowledge chain, wherein the fault penetration node refers to a key node that can affect or trigger other faults when the fault propagates in the inference fault knowledge chain.

[0086] Take the fault propagation in the data processing unit as an example. If an algorithm error in the data processing unit is the source of the fault, this error may affect the subsequent data verification, data storage and other links along the data processing process. This is the fault propagation path. The fault penetration node refers to the key node that can affect or trigger other faults when the fault propagates in the reasoning fault knowledge chain. For example, the data verification module in the data processing unit is a fault penetration node because once it fails, it may cause erroneous data to flow into the storage device, which in turn causes more faults.

[0087] Step S128, based on the position, function and association relationship of the fault permeation node with other fault permeation nodes in the inference fault knowledge chain and the focus fault knowledge chain, extract the key features of the fault permeation node, and encode the key features into numerical data to form a fault permeation node vector, wherein the key features include node type, node degree and path length between nodes.

[0088] Take the fault penetration node of the data verification module in the data processing unit as an example. The node type is data verification module, which is a specific functional type. The node degree indicates the number of edges connected to this node, reflecting its connection relationship with other nodes. For example, if it is connected to both the data processing module and the storage module, then the node degree is 2. The path length between nodes, such as the shortest path length from it to the storage module, etc. Converting these key features into numerical data according to specific encoding rules forms a fault penetration node vector.

[0089] Step S129: For each inference fault knowledge chain, the corresponding fault logic diagram encoding vector and the fault penetration node vector are stacked to obtain the fault logic flow vector corresponding to the inference fault knowledge chain. The stacking process uses an attention mechanism or a gating mechanism to dynamically adjust the weights of the fault logic diagram encoding vector and the fault penetration node vector.

[0090] Finally, taking the sensor-related reasoning fault knowledge chain as an example, the fault logic diagram encoding vector contains the fault logic relationship of the sensor from data acquisition to data transmission, and the fault permeation node vector contains the key features of the key components inside the sensor (such as the sensor element, which is a fault permeation node). When stacking, the attention mechanism or gating mechanism will dynamically assign weights according to the actual situation. If the current focus is on the impact of the sensor element failure on the entire fault logic flow, the weight of the fault permeation node vector may be relatively high; if more attention is paid to the overall fault logic relationship, the weight of the fault logic diagram encoding vector may dominate. Through such stacking, the obtained fault logic flow vector can more comprehensively reflect the fault flow in the reasoning fault knowledge chain, and provide richer information for subsequent fault analysis and processing.

[0091] In a possible implementation manner, the target fault root cause description label is generated by a first deep learning network, and the training step of the first deep learning network includes: Step S101, obtaining a template IoT interaction event flow and a second fault knowledge chain corresponding to a template fault activity in the template IoT interaction event flow, extracting the fault knowledge chain from the template IoT interaction event flow, and generating a first fault knowledge chain corresponding to each fault interaction activity in the template IoT interaction event flow, wherein the template fault activity is one fault activity among the multiple fault interaction activities.

[0092] Step S102: extract the second interaction path vector of the template IoT interaction event flow, respectively perform encoding representation of the template IoT interaction event flow according to the second fault knowledge chain and each of the first fault knowledge chains, generate multiple template encoding features, and extract the template fault penetration node vector corresponding to the second fault knowledge chain and each of the first fault knowledge chains.

[0093] Step S103, stacking the second interaction path vector, the template encoding feature, and the template fault penetration node vector and loading them into the first deep learning network for prediction to generate a fault root cause description prediction result, which is used to determine the fault root cause description label of the template fault activity.

[0094] Step S104, obtaining a first fault root cause description label of a template fault knowledge point associated with the template IoT interaction event flow, determining a training error based on the fault root cause description prediction result and the first fault root cause description label, optimizing the first deep learning network based on the training error, and the template fault knowledge point is used to guide the positioning of the template fault activity.

[0095] In a possible implementation manner, the template IoT interaction event flow, the second fault knowledge chain, and the first fault root cause description tag are all acquired from a sample database. Before step S101, the method further includes: Step A110, obtaining multiple basic Internet of Things interaction event flows and fault attention knowledge data corresponding to each of the basic Internet of Things interaction event flows, and determining key interaction point data corresponding to each of the basic Internet of Things interaction event flows based on each of the basic Internet of Things interaction event flows and the corresponding fault attention knowledge data, wherein the fault attention knowledge data is used to guide the positioning of key fault activities in the corresponding basic Internet of Things interaction event flows.

[0096] Step A120, obtaining a plurality of reference fault knowledge points, and determining, according to each of the key interaction point data, the associated fault knowledge points corresponding to each of the basic IoT interaction event flows from the plurality of reference fault knowledge points.

[0097] Step A130, according to each of the basic Internet of Things interaction event flows and the corresponding fault concern knowledge data, respectively determine the fault knowledge chain label data of each of the basic Internet of Things interaction event flows, and the fault knowledge chain label data is used to guide the positioning of the corresponding key fault activities.

[0098] Step A140, integrate and add each of the basic IoT interaction event flows, the corresponding fault knowledge chain label data, and the corresponding associated fault knowledge points into the sample database, the template IoT interaction event flow is randomly extracted from multiple basic IoT interaction event flows, the second fault knowledge chain is the fault knowledge chain label data corresponding to the template IoT interaction event flow, and the template fault knowledge point is the associated fault knowledge point associated with the template IoT interaction event flow.

[0099] In a possible implementation, step A130 includes: Step A131, respectively load each of the fault concern knowledge data into the second deep learning network for prediction, and generate fault summary description data corresponding to each of the basic IoT interaction event flows.

[0100] Step A132, locating fault activities according to each of the basic IoT interaction event flows and the corresponding fault summary description data, generating basic fault logic paths corresponding to each of the basic IoT interaction event flows, and the basic fault logic paths are used to guide the location of the corresponding key fault activities.

[0101] Step A133, respectively load each of the basic Internet of Things interaction event flows and the corresponding basic fault logic paths into the first fault knowledge chain extraction network to extract the fault knowledge chain, and generate fault knowledge chain label data corresponding to each of the basic Internet of Things interaction event flows.

[0102] In this embodiment, first, in the training step of the first deep learning network, the server needs to obtain the template IoT interaction event stream and the second fault knowledge chain corresponding to the template fault activity in the template IoT interaction event stream, and extract the fault knowledge chain from the template IoT interaction event stream to generate the first fault knowledge chain corresponding to each fault interaction activity in the template IoT interaction event stream. In the radio monitoring facility, the template IoT interaction event stream can be regarded as a representative set of interaction events selected from the IoT interaction event streams in many actual operation and maintenance scenarios. For example, this template IoT interaction event stream may include a series of interaction events between the sensor device and the data processing unit and the storage device within a specific time period. The template fault activity, for example, is that the data collected by the sensor at a certain moment has abnormal fluctuations. The second fault knowledge chain corresponding to this template fault activity may include a series of fault factors and their associations from the sensor hardware level (such as component aging, external interference, etc.) to the software level (such as data acquisition algorithm errors, etc.), and then to the interaction level with other devices (such as problems with the transmission protocol of the data processing unit, etc.).

[0103] The server extracts the fault knowledge chain from this template IoT interaction event stream and generates the first fault knowledge chain corresponding to each fault interaction activity. For example, in addition to the template fault activity of abnormal fluctuation of sensor data mentioned above, there may also be a fault interaction activity of data processing unit processing data delay. For the fault interaction activity of data processing unit processing data delay, the server constructs the corresponding first fault knowledge chain by analyzing each event in the event stream, such as the time when the data processing unit receives the data, the time for processing the data, the execution of the processing algorithm, etc. This knowledge chain may involve fault factors such as insufficient hardware performance of the data processing unit, high complexity of the software algorithm, and resource competition when interacting with other devices, as well as their interrelationships.

[0104] Next, the server extracts the second interaction path vector of the template IoT interaction event stream. In the radio monitoring facility scenario, this second interaction path vector reflects the interaction order and path relationship between the devices in the template IoT interaction event stream. For example, the sensor collects data and transmits it to the data processing unit, which stores the data in the storage device after processing. The transmission order and connection relationship between the devices constitute the second interaction path vector. At the same time, the server performs encoding representation of the template IoT interaction event stream based on the second fault knowledge chain and each first fault knowledge chain, generates multiple template encoding features, and extracts the template fault penetration node vector corresponding to the second fault knowledge chain and each first fault knowledge chain.

[0105] For the encoding representation, taking the sensor data collection event as an example, according to the second fault knowledge chain or a first fault knowledge chain (such as the fault knowledge chain related to sensor hardware), the server will extract the key features of the sensor data collection event, such as the collection time, collection frequency, and the collected data intensity, and encode these features into numerical data according to specific encoding rules. After encoding multiple such events, they are combined to form the template encoding features. For the extraction of the template fault penetration node vector, for example, in the sensor hardware fault knowledge chain, the sensor element may be a fault penetration node. The server extracts its key features according to the position of this node in the knowledge chain (such as being at the starting position, it is more likely to be the source of the fault), the role (such as directly affecting the accuracy of the collected data), and the association with other fault penetration nodes (such as the signal conversion module of the sensor, etc.). For example, it may have an association relationship with the sensor's signal conversion module, etc.), such as node type (sensor element), node degree (the number of connections with other related nodes, assuming that it is connected to the signal conversion module and the power module, the node degree is 2), and the path length between nodes (the number of data processing links passed by the shortest path to the storage device, etc.), and after encoding these key features into numerical data, the template fault penetration node vector is formed.

[0106] Then, the server stacks the second interaction path vector, the template coding features, and the template fault penetration node vector and loads them into the first deep learning network for prediction, generating a fault root cause description prediction result. In the scenario of intelligent operation and maintenance of radio monitoring facilities, the first deep learning network is a network model designed to analyze the root cause of the fault. This stacked vector contains the device interaction path in the template IoT interaction event stream, the coding features of each fault interaction activity, and the information of the fault penetration node. After inputting it into the first deep learning network, the first deep learning network performs analysis and prediction based on its internal algorithm and model structure. For example, if the abnormal fluctuation of sensor data is a template fault activity, the network may predict that the collected data is inaccurate due to the aging of the sensor element. During the data transmission process, certain settings of the transmission protocol amplify this inaccuracy, which ultimately leads to abnormal fluctuations in the data. This prediction result is the fault root cause description prediction result, which is used to determine the fault root cause description label of the template fault activity.

[0107] Afterwards, the server obtains the first fault root cause description label of the template fault knowledge point associated with the template IoT interactive event flow, determines the training error based on the fault root cause description prediction result and the first fault root cause description label, and optimizes the first deep learning network based on the training error. In the radio monitoring facility scenario, the template fault knowledge point is a knowledge set closely related to the template fault activity, which is used to guide the location of the template fault activity. For example, the template fault knowledge point clearly points out that the abnormal fluctuation of sensor data may be related to the combined effect of sensor hardware aging and software algorithm, which is the first fault root cause description label. The server compares the prediction result with this label. If the prediction result only mentions the aging of sensor hardware and does not involve the influence of the software algorithm, then there is a certain deviation, which is the training error. Based on this training error, the server adjusts the parameters of the first deep learning network, such as the weight and bias of the neural network, to improve the prediction accuracy of the network.

[0108] Before obtaining the template IoT interaction event stream and the second fault knowledge chain corresponding to the template fault activity in the template IoT interaction event stream, the server needs to perform a series of preparatory work to build a sample database. First, the server obtains multiple basic IoT interaction event streams and the fault attention knowledge data corresponding to each basic IoT interaction event stream. In radio monitoring facilities, the basic IoT interaction event stream is a collection of broader IoT device interaction events. For example, the interaction event stream between sensors and data processing units and storage devices in different areas. Fault attention knowledge data is used to guide the positioning of key fault activities in the corresponding basic IoT interaction event stream. For example, the sensor data transmission in a certain area is unstable, and this information about the unstable transmission is the fault attention knowledge data.

[0109] Based on each basic IoT interaction event stream and the corresponding fault concern knowledge data, the server determines the key interaction point data corresponding to each basic IoT interaction event stream. For example, in a basic IoT interaction event stream where sensor data transmission is unstable, the key interaction point data may be the relevant data at the transmission interface between the sensor and the data processing unit, such as fluctuations in the transmission rate, packet loss of the transmission data, etc. Then, the server obtains multiple reference fault knowledge points, and determines the associated fault knowledge points corresponding to each basic IoT interaction event stream in multiple reference fault knowledge points based on each key interaction point data. The reference fault knowledge point is a collection of various fault types and their related knowledge. According to the key interaction point data, such as transmission rate fluctuations and packet loss, the associated fault knowledge points that match them are found in the reference fault knowledge points, which may be knowledge about transmission protocol errors or network congestion.

[0110] Then, the server determines the fault knowledge chain label data of each basic IoT interaction event stream based on each basic IoT interaction event stream and the corresponding fault concern knowledge data. Specifically, the server loads each fault concern knowledge data into the second deep learning network for prediction, and generates fault summary description data corresponding to each basic IoT interaction event stream. In the radio monitoring facility scenario, taking the unstable sensor data transmission as an example, after the fault concern knowledge data (transmission instability information) is loaded into the second deep learning network, the network may predict that the unstable data transmission is caused by problems in the network transmission link. This result is the fault summary description data.

[0111] Next, the server locates the fault activity according to each basic IoT interaction event stream and the corresponding fault summary description data, and generates the basic fault logic path corresponding to each basic IoT interaction event stream. For example, according to the fault summary description data of unstable sensor data transmission and problems in the network transmission link, the basic fault logic path may start from the sensor sending data, and after being processed by the transmission protocol, due to network congestion or transmission protocol errors during the network transmission process, the data becomes unstable when it reaches the data processing unit.

[0112] Finally, the server loads each basic IoT interaction event stream and the corresponding basic fault logic path into the first fault knowledge chain extraction network for fault knowledge chain extraction, and generates fault knowledge chain label data corresponding to each basic IoT interaction event stream. This fault knowledge chain label data can guide the location of the corresponding key fault activities. For example, in the case of unstable sensor data transmission, the fault knowledge chain label data contains multiple factors such as sensor hardware, transmission protocol, network environment, and their interrelationships. These relationships can clearly point out which factors are the key factors that lead to the key fault activity of unstable data transmission, thereby providing important data support for the subsequent construction of the sample database and the training of the first deep learning network.

[0113] In this way, the first deep learning network can be effectively trained to accurately analyze the root cause of the fault and generate an accurate root cause description label.

[0114] Figure 2 The hardware structure of the distributed operation and maintenance intelligent management system 100 based on cloud computing and the Internet of Things for implementing the above-mentioned distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things provided by an embodiment of the present invention is shown as follows: Figure 2 As shown, the distributed operation and maintenance intelligent management system 100 based on cloud computing and the Internet of Things may include a processor 110 , a machine-readable storage medium 120 , a bus 130 , and a communication unit 140 .

[0115] The machine-readable storage medium 120 may store data and / or instructions. In some embodiments, the machine-readable storage medium 120 may store data obtained from an external terminal. In some embodiments, the machine-readable storage medium 120 may store data and / or instructions that the distributed operation and maintenance intelligent management system 100 based on cloud computing and the Internet of Things uses to execute or use to complete the exemplary method described in the present invention.

[0116] During the specific implementation process, one or more processors 110 execute computer executable instructions stored in the machine-readable storage medium 120, so that the processor 110 can execute the distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things as described in the above method embodiment. The processor 110, the machine-readable storage medium 120 and the communication unit 140 are connected through the bus 130, and the processor 110 can be used to control the sending and receiving actions of the communication unit 140.

[0117] The specific implementation process of the processor 110 can refer to the various method embodiments executed by the above-mentioned distributed operation and maintenance intelligent management system 100 based on cloud computing and the Internet of Things. The implementation principles and technical effects are similar, and this embodiment will not be repeated here.

[0118] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer executable instructions are preset. When a processor executes the computer executable instructions, the above-mentioned distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things is implemented.

[0119] It should be noted that in order to simplify the description of the present invention and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, various features are sometimes combined into one embodiment, drawing or description thereof.

Claims

1. A distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things, characterized in that: The method comprises: Obtaining the inference fault knowledge chain corresponding to each potential fault activity in the target IoT interaction event stream in the cloud computing environment, determining a concerned fault knowledge chain from a plurality of the inference fault knowledge chains, wherein the concerned fault knowledge chain is used to guide the location of the concerned target fault activity among the plurality of the potential fault activities; The target IoT interaction event stream is respectively encoded according to the concerned fault knowledge chain and each of the inference fault knowledge chains, a plurality of fault logic diagram encoding vectors are generated, and the fault penetration node vectors corresponding to the concerned fault knowledge chain and each of the inference fault knowledge chains are extracted; Stacking each of the fault logic diagram encoding vectors with the corresponding fault penetration node vectors to generate a plurality of fault logic flow vectors; Extracting a first interaction path vector of the target Internet of Things interaction event flow, stacking the first interaction path vector and a plurality of the fault logic flow vectors to generate a target stacking vector; Predictions are made based on the target stacked vector to generate a target fault root cause description label of the target fault activity.

2. The distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things according to claim 1 is characterized in that: The stacking of the first interaction path vector and the plurality of fault logic flow vectors to generate a target stacking vector includes: Determine the knowledge chain length of each of the inference fault knowledge chains respectively, and stack the fault logic flow vectors corresponding to each of the inference fault knowledge chains in descending order of the knowledge chain lengths to generate a first stacking vector; The first interaction path vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain are stacked to generate a target stacking vector.

3. The distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things according to claim 2 is characterized in that: The target fault root cause description label is generated by a first deep learning network, and the first interaction path vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain are stacked to generate a target stacking vector, including: Generate guidance information for guiding the first deep learning network to generate a fault root cause description label; A guidance knowledge vector of the guidance information is extracted, and the first interaction path vector, the guidance knowledge vector, the first stacking vector, and the fault logic flow vector corresponding to the concerned fault knowledge chain are stacked to generate a target stacking vector.

4. The distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things according to claim 1 is characterized in that: The target IoT interaction event stream is respectively encoded according to the concerned fault knowledge chain and each of the inference fault knowledge chains to generate a plurality of fault logic diagram encoding vectors, including: Performing multi-stage encoding representation on the target Internet of Things interaction event stream to generate multi-stage encoding features of the target Internet of Things interaction event stream; According to the concerned fault knowledge chain and each of the inference fault knowledge chains, respectively, the multi-stage coding features are subjected to knowledge chain coding mapping to generate a plurality of multi-stage mapping coding features; The multi-stage mapping coding features are respectively coded and integrated to generate a plurality of fault logic diagram coding vectors.

5. The distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things according to claim 1 is characterized in that: The acquiring of the inference fault knowledge chain corresponding to each potential fault activity in the target IoT interaction event stream in the cloud computing environment, and determining a focus fault knowledge chain from a plurality of the inference fault knowledge chains, includes: Acquire a target Internet of Things interaction event stream and guidance identification information of the target Internet of Things interaction event stream, wherein the target Internet of Things interaction event stream includes multiple potential fault activities, and the guidance identification information is used to guide the location of a target fault activity of concern in the multiple potential fault activities; Fault knowledge chains are extracted from the target IoT interaction event stream to generate inference fault knowledge chains corresponding to each of the potential fault activities, and a focus fault knowledge chain is determined in each of the inference fault knowledge chains according to the guidance identification information.

6. The distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things according to claim 5 is characterized in that: The step of extracting the fault knowledge chain from the target IoT interaction event stream and generating the inference fault knowledge chain corresponding to each potential fault activity includes: Traversing the target IoT interaction event stream, adding each event in the target IoT interaction event stream as a node to an event association graph, and determining the association relationship between events according to the time sequence, event source, and event target attributes between events, and optimizing the event association graph after adding corresponding edges to the event association graph; A graph traversal algorithm is used to traverse the nodes in the optimized event association graph, and potential fault activities are identified according to the association relationship between events and the preset fault identification rules. The identified fault activities are represented in the form of a node set and added to the potential fault activity list; For each potential fault activity in the potential fault activity list, extract corresponding key features, wherein the key features include event type distribution, event time interval distribution, event source and target distribution, etc.; quantizing the key features into numerical data, combining them into a fault feature vector, and then normalizing the fault feature vector; For the fault feature vector of each potential fault activity, a similarity calculation algorithm is used to find the most similar fault feature in the pre-built fault knowledge base, and based on the found fault features, the corresponding fault knowledge data is obtained, and based on the association relationship between the fault types in the fault knowledge data and the logical relationship between the fault causes, an inference fault knowledge chain is constructed.

7. The distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things according to claim 1 is characterized in that: The target fault root cause description label is generated by a first deep learning network, and the training steps of the first deep learning network include: Acquire a template IoT interaction event stream and a second fault knowledge chain corresponding to a template fault activity in the template IoT interaction event stream, extract the fault knowledge chain from the template IoT interaction event stream, and generate a first fault knowledge chain corresponding to each fault interaction activity in the template IoT interaction event stream, wherein the template fault activity is one of the multiple fault interaction activities; Extracting a second interaction path vector of the template IoT interaction event flow, encoding the template IoT interaction event flow according to the second fault knowledge chain and each of the first fault knowledge chains, generating a plurality of template encoding features, and extracting a template fault penetration node vector corresponding to the second fault knowledge chain and each of the first fault knowledge chains; The second interaction path vector, the template encoding feature, and the template fault penetration node vector are stacked and loaded into the first deep learning network for prediction to generate a fault root cause description prediction result, wherein the fault root cause description prediction result is used to determine a fault root cause description label of the template fault activity; Obtain a first fault root cause description label of a template fault knowledge point associated with the template IoT interaction event flow, determine a training error based on the fault root cause description prediction result and the first fault root cause description label, optimize the first deep learning network based on the training error, and use the template fault knowledge point to guide the positioning of the template fault activity.

8. The distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things according to claim 7 is characterized in that: The template IoT interaction event flow, the second fault knowledge chain, and the first fault root cause description tag are all obtained from a sample database. Before obtaining the template IoT interaction event flow and the second fault knowledge chain corresponding to the template fault activity in the template IoT interaction event flow, the method further includes: Acquire multiple basic Internet of Things interaction event streams and fault attention knowledge data corresponding to each of the basic Internet of Things interaction event streams, and determine key interaction point data corresponding to each of the basic Internet of Things interaction event streams based on each of the basic Internet of Things interaction event streams and the corresponding fault attention knowledge data, wherein the fault attention knowledge data is used to guide and locate key fault activities in the corresponding basic Internet of Things interaction event streams; Acquire multiple reference fault knowledge points, and determine, according to each of the key interaction point data, the associated fault knowledge points corresponding to each of the basic Internet of Things interaction event flows in the multiple reference fault knowledge points; According to each of the basic Internet of Things interaction event flows and the corresponding fault concern knowledge data, respectively determine the fault knowledge chain label data of each of the basic Internet of Things interaction event flows, and the fault knowledge chain label data is used to guide the positioning of the corresponding key fault activities; Each of the basic Internet of Things interaction event streams, the corresponding fault knowledge chain label data, and the corresponding associated fault knowledge points are integrated and added to the sample database; the template Internet of Things interaction event stream is randomly extracted from a plurality of the basic Internet of Things interaction event streams; the second fault knowledge chain is the fault knowledge chain label data corresponding to the template Internet of Things interaction event stream; and the template fault knowledge point is the associated fault knowledge point associated with the template Internet of Things interaction event stream.

9. The distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things according to claim 8 is characterized in that: Determining the fault knowledge chain label data of each of the basic Internet of Things interaction event flows according to each of the basic Internet of Things interaction event flows and the corresponding fault concern knowledge data respectively includes: Loading each of the fault concern knowledge data into the second deep learning network for prediction, and generating fault summary description data corresponding to each of the basic Internet of Things interaction event flows; Perform fault activity location according to each of the basic Internet of Things interaction event flows and the corresponding fault summary description data, and generate a basic fault logic path corresponding to each of the basic Internet of Things interaction event flows, wherein the basic fault logic path is used to guide the location of the corresponding key fault activity; Each of the basic Internet of Things interaction event flows and the corresponding basic fault logic paths are loaded into a first fault knowledge chain extraction network to extract the fault knowledge chain, and generate fault knowledge chain label data corresponding to each of the basic Internet of Things interaction event flows.

10. A distributed operation and maintenance intelligent management system based on cloud computing and the Internet of Things, characterized in that: The distributed operation and maintenance intelligent management system based on cloud computing and the Internet of Things includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the distributed operation and maintenance intelligent management method based on cloud computing and the Internet of Things as described in any one of claims 1 to 9.

Images