CN114268640A - Intelligent routing system of industrial Internet of things with cloud edge cooperation - Google Patents
Intelligent routing system of industrial Internet of things with cloud edge cooperation Download PDFInfo
- Publication number
- CN114268640A CN114268640A CN202010971570.0A CN202010971570A CN114268640A CN 114268640 A CN114268640 A CN 114268640A CN 202010971570 A CN202010971570 A CN 202010971570A CN 114268640 A CN114268640 A CN 114268640A
- Authority
- CN
- China
- Prior art keywords
- data
- service
- edge
- equipment
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Debugging And Monitoring (AREA)
Abstract
An industry thing networking intelligent routing system that cloud limit is cooperative includes: the system comprises an edge data route, an equipment template configuration module, an edge service management module, a flow configuration module, a message distribution module, a flow execution unit, a service registration center and a monitoring analysis module. According to the invention, tasks such as data cleaning, data conversion and data compression are delivered to the edge computing node which is geographically close to the industrial production equipment through equipment service operation instance management, and then the data preprocessed by the edge computing node is transmitted to the cloud service, and complete data monitoring, processing, analysis and other functions are realized on the cloud service based on a preconfigured business process, so that the cloud-edge cooperative business process execution and industrial equipment state monitoring process is realized.
Description
Technical Field
The invention relates to the technology in the field of Internet of things, in particular to an intelligent routing system of an industrial Internet of things with cloud-edge coordination.
Background
In the industrial 4.0 era, the technology of the internet of things plays an important role. The development of industrial intelligence is not independent of the upgrading of equipment, and particularly important point is that the industrial internet of things equipment is combined with a cloud platform by using the internet of things technology, and data obtained by the internet of things equipment in industrial production is properly processed by the powerful computing capacity of the cloud platform. However, the internet of things technology still faces the following problems in the industrial production practice: firstly, a large number of heterogeneous devices exist in industrial production, the devices can be cross-workshop or even cross-region, and how to effectively carry out uniform data acquisition and management on the heterogeneous devices is one of the problems which are urgently needed to be solved in the application process of the internet of things technology; secondly, for the process tasks related to a plurality of heterogeneous devices, how the devices interact and cooperate to complete the business process is also an important problem; in addition, how to perform efficient and accurate data analysis on a large amount of data generated in an industrial production process is also a considerable problem.
The existing solutions include that a server directly establishes long connection with an equipment terminal for communication, or data acquired by equipment is analyzed and decided by developing application and deploying the data to a cloud platform, and the equipment is controlled in a closed loop mode by an industrial control computer, but the technical solutions may have the problems of network communication delay and the like when facing a large number of pieces of equipment in multiple regions, are limited to data acquisition and provision, and are not easy to implement and use.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a cloud-edge cooperative industrial internet of things intelligent routing system, tasks such as data cleaning, data conversion and data compression are delivered to edge computing nodes which are geographically close to industrial production equipment through equipment service operation instance management, the data preprocessed by the edge computing nodes are transmitted to a cloud service, and complete functions such as data monitoring, processing and analysis are realized on the basis of a preset service flow in the cloud service, so that a cloud-edge cooperative service flow execution and industrial equipment state monitoring process is realized.
The invention is realized by the following technical scheme:
the invention relates to an intelligent routing system of an industrial Internet of things with cloud edge coordination, which comprises: an edge data route used for collecting and preprocessing data generated by industrial internet of things equipment at an edge end, an equipment template configuration module used for managing an equipment data template configured by a user and defining a data cleaning strategy and a data conversion rule script related to equipment data processing, an edge service management module used for managing the state of edge end equipment service running in a specific area, a process configuration module, a message distribution module used for monitoring and distributing the data obtained by the edge data route, a process execution unit used for performing data fusion, service call and sequence flow control on new messages from the message distribution module according to the requirement of a business process and performing backpressure control on the data flow, and a process execution unit used for maintaining the state of each equipment service managed by the edge service management module in each current edge computing node and meta information related to the service at a cloud end, the system comprises a service registration center supporting a service call request of a process execution unit and a monitoring analysis module used for performing structured storage on a log from the process execution unit and performing equipment health analysis by combining an equipment service health value from the service registration center, wherein: the service operation examples corresponding to each device send the collected business data to a data collection interface provided by an edge data route, and then the business data is pushed to a cloud message queue after being uniformly preprocessed by the edge data route, a process configuration module verifies a business process modeling and labeling (BPMN) model defined by a user, checks whether the business process model is correct in syntax and analyzes the model, a message distribution module monitors the data of the cloud message queue, and the distribution of the message is completed based on the message distribution rule previously determined by the flow configuration module, the service registration center interacts with the relatively more stable edge service management module through a heartbeat mechanism, during the interaction process, the edge service management module sends the average health value of each equipment service in the time period to the service registry.
The preprocessing process is performed according to the device data template, and the heterogeneous industrial device is adapted through the device data template, and the method specifically includes the following steps: data cleaning, data conversion and data compression, wherein: the data cleaning process is carried out according to a data cleaning strategy configured in the equipment data template; the data conversion process is completed according to a user-defined data conversion rule script in the device data template, and then a DEFLATE data compression algorithm is adopted to perform lossless compression on the data and send the data to the cloud, so that the occupation of network bandwidth from the edge computing node to the cloud service is reduced.
The data cleaning strategy comprises two preset strategies, namely a static sliding window strategy and an adaptive sliding window strategy, wherein the two preset strategies comprise: the static sliding window strategy can configure windows of time granularity or data tuple number granularity, and the influence of noise and missing values is reduced by performing mean smoothing on the acquired data stream based on the windows with fixed sizes; the self-adaptive sliding window strategy adopts a statistical smoothing algorithm and utilizes a probability statistical method to self-adaptively adjust the size of the sliding window. Aiming at different types of equipment, different data acquisition scene requirements are adapted by adjusting a data cleaning strategy of the equipment.
The data conversion rule script defines how data conversion needs to be carried out on data acquired by the edge data routing aiming at the type of equipment. In consideration of the heterogeneity of the devices, the adoption of preset conversion rules may not be enough to meet diversified data conversion requirements, so that the part supports the definition of the data conversion rules by means of custom rule scripts. The device template configuration module stores the data in a distributed manner, so that the device template configuration modules at the edge ends of a plurality of regions can share the template data of devices of the same type.
When the device service is started, the service description is preferably sent to the module for registration by calling the service registration interface of the module, and then the edge service management module interacts with the device service of the edge end through a heartbeat mechanism to maintain the current running state of each device service.
The registration process allows the device service to register a health check interface for the device service, and then in the process of interaction through the heartbeat mechanism, the edge service management module accesses the health check interface of each device service to acquire and record the current running device state data of each device service.
The device state data, i.e. the health state of the device service, includes: UP, BUSY and DOWN are used to indicate the available, BUSY and invalid states of the current device service. When the next time the edge service management module interacts with the service registration center of the cloud, the edge service management module calculates the average health value of each service according to the health state of the service in the time period and sends the average health value to the service registration center.
For subsequent use, the average health value remains within the interval [0,1 ].
Said average health value
Wherein: n is the health state of the specified service obtained by the module in the period from the last time the module interacted with the cloud service registryTotal number of times of (n)UPIs the number of times the health status of the service is UP within the period, nBUSYThe number of times that the service health state is BUSY in the period is determined, and k is a value in the interval (0, 1)]The constant of (2) is an upper limit of the load rate of the device, i.e., an upper limit of the ratio of the number of times the device status is busy.
The calculation formula limits the contribution degree of the busy state of the equipment service to the health value, and the average health value is reduced when the equipment service load rate is too high, so that the calculation formula is more suitable for being used in the subsequent backpressure control and anomaly detection processes. Compared with a mode that the cloud service registration center directly interacts with the device service of the edge end, the mode that the average health value of the computing device service is sent to the cloud can greatly reduce the network communication overhead, so that the cloud service registration center can manage the states of more device services at the same time, and the system has higher expandability.
The analyzing specifically comprises: and (3) regarding the BPMN process model as a graph data structure, traversing each task node in the process model based on breadth-first search, and analyzing each node. For the analysis operation of each task node, firstly, the description of the text form is required to be obtained, then the semantics corresponding to the text is analyzed by combining the domain knowledge graph, and the service interface matched with the task semantics is searched according to the semantic description, so that the data processing rule of the task node in the operation process is determined. In addition, additional configuration, such as configuration of backpressure control policy, etc., which may be performed by a node using BPMN extension, needs to be parsed according to BPMN syntax. According to the analysis process, the process configuration module determines the message distribution rule and the data processing rule required for completing the business process, so as to configure the message distribution module and generate a corresponding process execution unit to support the subsequent responsive business process execution process.
The distribution refers to that: the message distribution module stores the corresponding relation between each message source and the process execution unit by using the hash table, so that a set of the corresponding process execution unit can be quickly found for a new piece of data in the message distribution process, the message is sent to each processing unit in the set, the message is deleted from the memory after the message confirmation of each processing unit is received, and the distribution of the message is completed once.
The data fusion is as follows: and fusing the data from the message distribution module aiming at the task node which needs to process the input of the data sources. The process supports two ways of fusing according to time limit or according to specified fields. The fusion mode according to the time limit specifically means that the flow execution unit stores the received first piece of data into a cache, and fuses the data if messages from all other data sources are received within the limited time, otherwise, the data is used as abnormal data, service calling is not performed, only the related data is sent to the monitoring analysis module, and meanwhile, the related data is removed from the cache. The length of the time limit is a uniform default value, but can also be manually adjusted in the operation stage according to the requirements of users. The fusion mode is suitable for services with relatively low use frequency, such as production of finished product batch records and the like.
For a service used at a high frequency, due to network delay between an edge end and a cloud end, a data tuple sequence may be disordered according to time-limit fusion, and data fusion needs to be performed in a mode of fusion according to a specified field, specifically: when a user configures an equipment data template, a field is designated in a data conversion rule script as a fusion basis, then the data tuple is labeled by an edge data route after data preprocessing is finished, and the label content comprises a field name designated by the user and a corresponding value; when the flow executing unit receives the data tuples with the labels, the field values and the corresponding data tuples are stored in the cache, and when all the messages with the same labels reach the processing unit, the data tuples are fused. And then, the flow execution unit takes the data obtained by fusion as input, calls a service interface matched with the task of the flow execution unit, and performs sequence flow control according to the obtained result.
The sequential flow control specifically means: according to a sequence flow control function in a data processing rule, calculating a task node to be executed next in the flow according to a service calling result, packaging the service calling result and sending the service calling result to a message queue corresponding to the task node, triggering the execution of the next task node by a message distribution module, and sending the context and the processing result of the data processing process to a monitoring analysis module by a flow execution unit for data analysis.
In addition, in the above process, the flow execution unit needs to perform backpressure control on the data flow to prevent data backlog caused by an excessively fast data generation rate.
The service call specifically includes: and according to the definition of the service resources and the interfaces matched with the current task node in the data processing rules, the flow execution unit takes the data obtained by data fusion as input data and initiates a service call request to the service registration center. The service registration center finds out a proper service resource and initiates a calling request to a specific interface thereof according to the definition and input data of the required service resource and interface and the locally maintained service meta-information and service state, and returns the returned result to the flow execution unit. When the service registry cannot find the proper service resource, the service registry returns a call failure to the flow execution unit.
The back pressure control specifically comprises the following steps: the flow execution unit receives the number n of the data tuples in the time period delta tinAnd the number n of data tuples sent to the monitoring analysis moduleoutRespectively counting to represent the message receiving rate and the message processing rate in the time period, wherein: the length of the time period Δ t is determined by the latest average health value h of the service of the equipment involved in the processing unit. The process execution unit will periodically pull the latest average health value of the relevant service from the service registry according to the service interface which the unit needs to access, and when the unit has a plurality of candidate equipment service instances, the health values corresponding to the service instances need to be averaged to obtain
For calculating Δ t.
Said period of time
Wherein: t is t0Is a constant representing the lowest time interval.
The calculation mode enables the flow execution unit to be more sensitive to the device service with poor operation state (namely, to have a lower statistical time interval), so that the flow execution unit can sense the abnormal device service with too high load or failure more quickly to adjust the backpressure control strategy.
Preferably, when the message receiving rate is higher than the message processing rate, the process execution unit starts a backpressure control policy, which specifically includes: buffering, discarding and sampling, wherein: when the backpressure control is carried out through buffering, the flow execution unit locally maintains a buffer queue and firstly caches the message locally; when the local cache queue is full, the processing of the new message can be degraded to a discarding strategy, that is, the processing such as data fusion and service call is not performed, and the new message is only sent to the monitoring analysis module as failure data to perform the recording and analysis of the abnormal log. The backpressure control through sampling requires that extra information is provided for a task in a business process model to specify the sampling frequency of data when the task generates backpressure in the data processing process, and the receiving and processing of messages are controlled according to the sampling frequency, so that the backlog of the data is reduced. After the backpressure control strategy is started to operate for a period of time, when the message processing rate is higher than the message receiving rate, the flow execution unit starts to recover to work, and the normal work of the system is prevented from being influenced by the excessive backpressure control. The method of recovery is related to the backpressure control strategy employed. When a buffer strategy is adopted for backpressure control and is degraded to a discarding strategy, when no accumulated data exists in a local buffer queue, the backpressure control mode of the buffer strategy can be recovered, and if the message processing rate is higher than the message receiving rate in the next time period, the backpressure control strategy can be closed to recover normal work; when the sampling strategy is adopted to carry out backpressure control, the sampling frequency is multiplied until the sampling frequency exceeds 100% after the sampling frequency is increased next time, and if the message processing rate is still higher than the message receiving rate in the next time period, the backpressure control strategy can be closed, and normal work is recovered.
The service registration center maintains the latest average health value of each equipment service, the latest average health value is inquired by the process execution unit to support backpressure control of the data flow, and the average health value of the equipment service is sent to the monitoring analysis module to support the abnormal detection process of the data.
The log comprises: the data processing process uses the service instance information, the service interface information, the input of the interface call, and the return result of the interface call.
The anomaly detection, namely the monitoring analysis module adopts an anomaly detection algorithm based on a hierarchical time sequence memory (HTM) network to carry out anomaly detection on the data, specifically comprising the following steps: the monitoring analysis module finds the latest health value of the instance in the local cache according to the service instance information used in the data processing process, encodes the service instance information, the health value and the interface information by using a scalar encoder, encodes the input and return results of interface calling by using a cooperative encoder, and transmits a plurality of encoded vectors into a multi-encoder to form a vector containing log information of a complete flow execution unit; then, the vector is processed by a space pool to obtain a Sparse Distributed Representation (SDR) and used as the input of the HTM network, the next input SDR in the current data stream is continuously predicted by the learning algorithm of the HTM, and when a new SDR (recorded as d') is received, an original abnormal score is calculated according to the predicted SDR (recorded as d) in the previous round
Wherein: d 'represents the scalar product of the SDR of the previous round of prediction with the new SDR array, | d' | represents the number of bits in the new SDR array representing all values of 1.
In order to tolerate the noise of the equipment data, the monitoring analysis module does not directly set a threshold value for the original abnormal score to detect the abnormality, but calculates the probability of the current abnormal occurrence according to the distribution of the abnormal score, and sets a threshold value for the abnormal probability to detect the abnormality, specifically: the values of the most recent W anomaly scores are recorded,then, the mean value μ and the standard deviation σ of these values are calculated, the distribution of these values is regarded as a gaussian distribution, and the anomaly probability, that is, the anomaly probability is calculated by the right-tail function (Q-function) of the gaussian distribution
Wherein: stAnd (4) representing the abnormal score obtained by the t-th calculation.
The probability of the abnormity is calculated by using the data distribution, and the data abnormity is judged by setting a threshold value on the calculated probability, so that the influence of noise can be effectively reduced, and the robustness of a data analysis algorithm is improved.
Preferably, the monitoring analysis module stores the input log and the abnormal detection result into log historical data, and generates a periodic data analysis report at regular time for service personnel to analyze; and carrying out early warning notification on the relevant users when the detection finds the abnormity.
Technical effects
The invention integrally solves the problem that the prior art can not carry out low-delay data acquisition and unified management on a large number of heterogeneous devices distributed in multiple regions in the environment of industrial Internet of things; in the prior art, for process tasks related to a plurality of heterogeneous devices, interaction among the devices cannot be coordinated, so that the tasks are completed correctly and efficiently; and the inability to quickly and accurately analyze large amounts of data generated during industrial production to detect equipment anomalies that may occur.
Compared with the prior art, the method and the device have the advantages that the utilization of edge computing resources is enhanced, and the pressure of network bandwidth from the internet of things device to the cloud end can be relieved; meanwhile, the edge end provides template-based configuration support for the heterogeneous equipment, so that the access of various heterogeneous equipment in an industrial production environment can be effectively supported, and the method has good expandability; in addition, the cloud end can effectively support the automatic execution of the business process by the cooperation of a plurality of different functional units and the combination of a streaming data processing method, and can carry out comprehensive analysis according to the data processing log, thereby providing an effective data analysis and early warning feedback scheme for the data generated by the internet of things equipment.
Drawings
FIG. 1 is a process framework diagram of the present invention;
fig. 2 is a system configuration diagram according to an embodiment of the present invention.
Detailed Description
As shown in fig. 1, the present embodiment relates to a cloud-edge collaborative intelligent routing system for an industrial internet of things, which takes data collected by an industrial production device control service as input, combines with a business process model, and implements data monitoring, distribution and processing of process tasks in industrial production by using a cloud-edge collaborative method, and records a data processing result and a state as a log, and performs data analysis on the log by using an anomaly detection algorithm, and performs an early warning notification on a relevant user when an anomaly is found in a data analysis result, and records information such as a data analysis report for upper-level enterprise application query, and the system includes: the device comprises an equipment template configuration module, a process configuration module, an edge data route, a message distribution module, a process execution unit, a service registration center, an edge service management module and a monitoring analysis module.
The device template configuration module receives the device template configuration file as input, analyzes the device template configuration file to obtain a data cleaning strategy and a data conversion rule script corresponding to the device, and stores the data into a distributed data storage for subsequent query.
The flow configuration module comprises: a model verification unit and a model parsing unit, wherein: the model verification unit verifies the input business process model according to the grammar rule of the process model, and outputs the grammar in the process model to the model analysis unit after confirming that the grammar in the process model is correct; the model analysis unit carries out topological structure analysis, semantic analysis and service matching on the business process model in sequence, determines a message distribution rule and a data processing rule required by the completion of the business process, and configures the message distribution module and the process execution unit according to the rules.
The edge data routing comprises: data washing unit, data conversion unit and data compression unit, wherein: the data cleaning unit inquires a corresponding equipment data template according to the type of input corresponding equipment, and then cleans the input service data according to a data cleaning strategy defined in the equipment data template and outputs the cleaned service data to the data conversion unit; the data conversion unit defines a data conversion rule script in the cleaned data execution equipment data template and outputs the converted data to the data compression unit; and the data compression unit compresses the converted data and then sends the compressed data to a message queue at the cloud end.
The message distribution module comprises: data listening unit and message distribution unit, wherein: the data monitoring unit monitors data in the message queue, takes the new data from the edge data route as input when acquiring the new data, and triggers the distribution of the message by calling an interface of the message distribution unit. The message distribution unit calculates an index of the message according to the message queue corresponding to the message and the device identifier corresponding to the data contained in the message, searches a set of corresponding flow execution units in a local hash table based on the index, and sends the message to each flow execution unit in the set to complete the distribution of the message.
The flow executing unit comprises: the system comprises a data flow backpressure control unit, a data fusion unit, a service calling unit and a sequential flow control unit, wherein: the data flow backpressure control unit counts the message receiving rate and the message processing rate of the flow execution unit, and starts a backpressure control strategy to carry out backpressure control on the data flow when the message receiving rate is higher than the processing rate; the data fusion unit fuses a plurality of pieces of data from the message distribution module according to needs and outputs the data to the service calling unit; the service calling unit takes the fused data as input, initiates a service calling request to a service registration center, and outputs a result returned by the request to the sequence flow control unit; and the sequence flow control unit takes the service calling result as input, executes a sequence flow control function to determine the next task node to be executed after the flow is completed, packages the service calling result and sends the service calling result to a message queue to trigger the flow to be further executed, and simultaneously sends the relevant data of the data processing process to the monitoring analysis module for data analysis.
The service registration center manages the service meta information and the service state from the edge service management module, and stores the related data into the memory database for the subsequent inquiry or service calling process.
The edge service management module manages the equipment service running at the edge end, receives the description file of the service and analyzes the description file to obtain the metadata of the service, and meanwhile, regularly receives the state data of the service through a heartbeat mechanism, calculates the average health value of the service according to the data and regularly sends the average health value to the service registration center.
The monitoring analysis module comprises: log recording unit, anomaly detection unit, early warning feedback unit and level sequence memory network, wherein: the log recording unit is used for performing structured storage on the log from the flow execution unit and outputting the log to the abnormality detection unit; the abnormality detection unit encodes and processes the space pool according to service metadata corresponding to the log, service calling input, service calling result and other data to obtain SDR, predicts the SDR as input by using a hierarchical time sequence memory network, detects abnormality of new data based on the prediction result, and outputs abnormality related information to the early warning feedback unit when abnormality is detected; the early warning feedback unit carries out early warning notification on related users through a preset contact way in the personal information of the users according to the abnormal related information from the abnormal detection unit, records the abnormal related information, and periodically aggregates the abnormal detection results to form a staged data analysis report for the users to inquire and use.
As shown in fig. 2, the present embodiment relates to a routing method of the above system, and specifically includes: a configuration phase and an operation phase.
In the configuration stage, a user can write an equipment template configuration file in a YAML format and send the equipment template configuration file to an equipment template configuration module at an edge end through RESTAPI. And equipment template configuration modules deployed in different regions deploy MongoDB by adopting a copy set cluster architecture to realize distributed storage of data. The user can also construct a BPMN business process model and send the BPMN business process model to a process configuration module at the cloud end through RESTAPI. The flow configuration module is used for assisting in analyzing the BPMN flow model by inquiring a previously constructed domain knowledge graph stored in a Neo4j graph database, and configuring a flow-related message distribution module and a flow execution unit according to an analysis result.
In the operation stage, real-time data of an internet of things device set in a production workshop can be provided for the OPC Server, and ECN micro-services corresponding to the deployment devices can be developed to interact with the OPC Server to realize the reading and writing of data and instructions. The device ECN micro-service sends a self service description file (WADL or WSDL) to an edge service management module during service registration, and sends self state data of the device and service data acquired by the device to the edge service management module and an edge data route respectively through REST API in JSON format during operation. And the edge service management module calculates the health state of the equipment according to the equipment state data and synchronizes to the service registration center of the cloud. The service registration center updates the state to a Redis memory database for use by a subsequent process execution unit and a monitoring analysis module. And the edge data router preprocesses the service data according to the equipment template and then pushes the data to a Kafka message queue at the cloud. And the message distribution module at the cloud monitors the Kafka message queue and distributes the Kafka message queue to the corresponding flow execution unit when a new message is monitored. When processing a new message, the flow execution unit makes a service call request required by the data processing process through the service registration center, performs backpressure control on the data flow by combining with the health state data of the equipment, sends the data related to the data processing to the monitoring analysis module when the data processing is finished, and constructs a new message to be pushed to the message distribution module when needed, so as to trigger the execution of the next data processing task in the service flow. The monitoring analysis module forms a flow execution log according to data generated in the data processing process and records the flow execution log into the MongoDB database, analyzes the flow execution log by combining with the equipment health state data, carries out early warning notification on a user when abnormality occurs in detection, and stores the analysis result and the early warning notification record into the MongoDB database. And then, an enterprise application such as ERP, MES and the like at the upper layer can obtain an execution log or an early warning notification record of the business process in the JSON format through REST API query provided by cloud service, and can query a data analysis report in the JSON or HTML format, wherein the report in the JSON format can be used for system background recording and processing, and the report in the HTML format can be used for visually displaying a user in an application scene such as an email and the like to assist the user in understanding and reviewing a data analysis result.
The present example is compared with the prior art as shown in the following table.
TABLE 1 comparison of technical characteristics
The method has the advantages of reducing the network bandwidth occupation of data transmission between the industrial equipment and the cloud service, reducing the data uniform acquisition and processing difficulty of heterogeneous equipment and improving the flexibility of the execution process of the multi-equipment collaborative business process. According to the invention, through configuration of the device data template, cleaning, conversion and compression of data acquired by heterogeneous industrial devices at the edge end are realized, an optional data cleaning strategy and a self-defined data conversion rule are provided, and the network overhead of data transmission between the industrial devices and cloud service is reduced by fully utilizing the computing capacity of the edge end; meanwhile, the edge terminals store the device data templates in a distributed manner, so that the distributed and deployed edge terminals can efficiently multiplex the templates of the same device, and the operation and maintenance cost is reduced; the edge nodes are used for assisting the cloud end to manage the equipment service information and the state, so that the pressure of centralized management on each equipment service in multiple regions is relieved; the analysis of a service flow topological structure is realized through the description of the service flow based on the BPMN flow model grammar and the expansion thereof, and meanwhile, the intelligent service flow semantic analysis is realized by the aid of the assistance of the related domain knowledge map of the industrial equipment, so that the flexibility of the multi-equipment collaborative production flow is improved; aiming at the phenomenon of unbalanced data production and consumption rate in the data processing process, a proper backpressure control method is provided to prevent data backlog, so that the robustness of system operation is improved; aiming at the logs generated in the data processing process, an anomaly detection algorithm based on an HTM network is adopted for rapid anomaly detection and analysis, and when the anomalies are detected, the feedback can be timely carried out on the users in a notification mode, so that the faults can be rapidly found, and the loss possibly brought by the faults in the production process can be reduced.
The above components/modules/algorithms/operations, which are original to the present invention, have never been disclosed, and their operation is different from any prior document, include: the method comprises a service state management process finished by the edge end and the cloud end in a cooperative mode, a flow execution process of multi-device service cooperative interaction and an HTM (hyper text transport memory) network-based anomaly detection algorithm aiming at the device service.
The technical details of the service state management process cooperatively completed by the edge end and the cloud end are as follows: the edge service management module interacts with the equipment service of the edge end through a heartbeat mechanism and maintains the current running state of the service, and the average health value of the service is calculated periodically according to the state and is sent to the service registration center, so that the service registration center of the cloud end can sense the change of the equipment service state of the edge end, and excessive network communication overhead from the edge end to the cloud end is not required.
The technical details of the flow execution process of the multi-device service collaborative interaction are specifically as follows: according to the message distribution rule and the data processing rule obtained by analyzing the business process model by the process configuration module, the message distribution module monitors and distributes a message queue corresponding to the data related to the equipment service, and the process execution unit performs data fusion, service call, sequence flow control and data flow backpressure control on the distributed data, so that a user can conveniently define a process task needing to be completed by the cooperation of multiple equipment services through the process model, and meanwhile, the system can quickly respond to new data related to the process.
The technical details of the HTM network-based anomaly detection algorithm for device services are specifically as follows: and the monitoring analysis module encodes and performs space pool processing on the logs from the process execution unit, and calculates the abnormal probability according to the distribution of the prediction result based on the abnormal data through HTM network prediction, thereby obtaining the abnormal detection result. The algorithm relates to a log from a flow execution unit, and simultaneously, static information such as instance information and interface information of equipment service, and dynamic data such as health value of service, input and return of interface calling are taken into consideration range, so that the result of abnormality detection is more reliable.
Compared with the prior art, the performance index of the device/method is improved as follows: by fully utilizing the computing resources of the edge end, the overhead of network communication between the cloud end and the edge end is reduced, and the efficiency of data acquisition and management of the service running at the edge end under the same standard is improved; in the process of flow execution, a proper back pressure control method is adopted to process the data backlog problem, so that the robustness of system operation is improved; the method adopts an anomaly detection algorithm based on the HTM network to rapidly analyze the logs generated in the process of flow execution, so that the system can keep good usability in a large data volume scene,
the foregoing embodiments may be modified in many different ways by those skilled in the art without departing from the spirit and scope of the invention, which is defined by the appended claims and all changes that come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.
Claims (12)
1. The utility model provides a collaborative industry thing networking intelligent routing system in cloud limit which characterized in that includes: an edge data route used for collecting and preprocessing data generated by industrial internet of things equipment at an edge end, an equipment template configuration module used for managing an equipment data template configured by a user and defining a data cleaning strategy and a data conversion rule script related to equipment data processing, an edge service management module used for managing the state of edge end equipment service running in a specific area, a process configuration module, a message distribution module used for monitoring and distributing the data obtained by the edge data route, a process execution unit used for performing data fusion, service call and sequence flow control on new messages from the message distribution module according to the requirement of a business process and performing backpressure control on the data flow, and a process execution unit used for maintaining the state of each equipment service managed by the edge service management module in each current edge computing node and meta information related to the service at a cloud end, the system comprises a service registration center supporting a service call request of a process execution unit and a monitoring analysis module used for performing structured storage on a log from the process execution unit and performing equipment health analysis by combining an equipment service health value from the service registration center, wherein: the method comprises the steps that collected business data are sent to a data collection interface provided by an edge data route by a service operation instance corresponding to each device, then the business data are subjected to uniform preprocessing by the edge data route and then are pushed to a cloud message queue, a process configuration module verifies a BPMN model defined by a user, whether the business process model is correct in syntax is checked, the model is analyzed, a message distribution module conducts data monitoring on the cloud message queue, message distribution is completed based on a message distribution rule previously determined by the process configuration module, a service registration center interacts with a relatively more stable edge service management module through a heartbeat mechanism, and the edge service management module sends an average health value of each device service in the time period to the service registration center in the interaction process.
2. The cloud-edge-coordinated intelligent routing system for the industrial internet of things as claimed in claim 1, wherein the preprocessing is performed according to an equipment data template, and the adaptation of heterogeneous industrial equipment is performed through the equipment data template, specifically comprising: data cleaning, data conversion and data compression, wherein: the data cleaning process is carried out according to a data cleaning strategy configured in the equipment data template; the data conversion process is completed according to a user-defined data conversion rule script in the device data template, and then a DEFLATE data compression algorithm is adopted to perform lossless compression on data and send the data to the cloud, so that the network bandwidth occupation from the edge computing node to the cloud service is reduced;
the data cleaning strategy comprises two preset strategies, namely a static sliding window strategy and an adaptive sliding window strategy, wherein the two preset strategies comprise: the static sliding window strategy can configure windows of time granularity or data tuple number granularity, and the influence of noise and missing values is reduced by performing mean smoothing on the acquired data stream based on the windows with fixed sizes; the self-adaptive sliding window strategy adopts a statistical smoothing algorithm, the size of the sliding window is self-adaptively adjusted by utilizing a probability statistical method, and different data acquisition scene requirements are adapted by adjusting the data cleaning strategy of the equipment aiming at different types of equipment;
the data conversion rule script defines how data conversion is required to be carried out on data acquired by the edge data router aiming at the type of equipment, and in consideration of the heterogeneity of the equipment, the adoption of a preset conversion rule may not be enough to meet diversified data conversion requirements, so that the part supports the definition of the data conversion rule by a user-defined rule script mode, and the equipment template configuration module carries out distributed storage on the data, so that the equipment template configuration modules of a plurality of regional edge terminals can share the template data of the same type of equipment.
3. The intelligent routing system of the industrial internet of things with cloud edge coordination as claimed in claim 1, wherein when the device service is started, the service registration interface of the module is preferably called to send the service description to the module for registration, and then the edge service management module interacts with the device service of the edge end through a heartbeat mechanism to maintain the current running state of each device service; the registration process allows the equipment service to register a health check interface for the equipment service, and then in the process of interaction through a heartbeat mechanism, the edge service management module accesses the health check interface of each equipment service to acquire and record the equipment state data of the current operation of each equipment service;
the device state data, i.e. the health state of the device service, includes: the system comprises an UP module, a BUSY module and a DOWN module, wherein the UP, the BUSY and the DOWN are respectively used for representing three states of available service, BUSY service and failure of the current equipment, and when the edge service management module interacts with a service registration center of a cloud, the edge service management module can calculate the average health value of each service according to the health state of the service in the time period and send the average health value to the service registration center;
said average health value
Wherein: n is the total number of times of obtaining the health state of the specified service by the module in the period from the last time of the interaction between the module and the cloud service registry to the present time, and n is the total number of times of obtaining the health state of the specified service by the moduleUPIs the number of times the health status of the service is UP within the period, nBUSYK is the upper limit of the load rate of the device, that is, the upper limit of the ratio of the number of times the state of the device is BUSY.
4. The cloud-edge collaborative industrial internet of things intelligent routing system according to claim 1, wherein the parsing specifically comprises: regarding a BPMN process model as a graph data structure, traversing each task node in the process model based on breadth-first search, analyzing each node, aiming at the analysis operation of each task node, firstly acquiring the description of the text form, then analyzing the semantics corresponding to the text by combining a domain knowledge graph, searching a service interface matched with the semantics of the task according to the semantic description, thereby determining the data processing rule of the task node in the operation process, in addition, according to the additional configuration which is possibly performed by using BPMN extension by the BPMN syntax analysis node, such as the configuration of a backpressure control strategy, and the like, according to the analysis process, a process configuration module determines the message distribution rule and the data processing rule required by completing the service process, so as to configure a message distribution module and generate a corresponding process execution unit, to support subsequent responsive business process execution.
5. The cloud-edge collaborative industrial internet of things intelligent routing system according to claim 1, wherein the distribution is that: the message distribution module stores the corresponding relation between each message source and the process execution unit by using the hash table, so that a set of the corresponding process execution unit can be quickly found for a new piece of data in the message distribution process, the message is sent to each processing unit in the set, the message is deleted from the memory after the message confirmation of each processing unit is received, and the distribution of the message is completed once.
6. The cloud-edge collaborative industrial internet of things intelligent routing system according to claim 1, wherein the data fusion is that: the process supports two modes of time limit fusion or specified field fusion, specifically, a process execution unit stores received first data into a cache according to the time limit fusion mode, if messages from all other data sources are received within a limited time, the data are fused, otherwise, the data are used as abnormal data, service calling is not carried out, only relevant data are sent to a monitoring analysis module, and meanwhile, the data are removed from the cache, the length of the time limit adopts a uniform default value, but manual adjustment can also be carried out in an operation stage according to user requirements, and the fusion mode is suitable for services with relatively low use frequency, such as finished product batch record production and the like.
7. The cloud-edge collaborative industrial internet of things intelligent routing system according to claim 1, wherein the sequence flow control specifically refers to: according to a sequence flow control function in a data processing rule, calculating a task node to be executed next in the flow according to a service calling result, packaging the service calling result and sending the service calling result to a message queue corresponding to the task node, triggering the execution of the next task node by a message distribution module, and sending the context and the processing result of the data processing process to a monitoring analysis module by a flow execution unit for data analysis.
8. The cloud-edge collaborative industrial internet of things intelligent routing system according to claim 1, wherein the service invocation specifically refers to: according to the definition of the service resource and the interface matched with the current task node in the data processing rule, the flow execution unit takes the data obtained through data fusion as input data, a service call request is sent to the service registration center, the service registration center searches the appropriate service resource and sends a call request to the specific interface according to the definition of the required service resource and the interface and the input data in combination with locally maintained service meta-information and service state, the return result is obtained and then returned to the flow execution unit, and when the service registration center cannot search the appropriate service resource, the flow execution unit returns call failure to the flow execution unit.
9. The cloud-edge collaborative industrial internet of things intelligent routing system according to claim 1, wherein the backpressure control specifically refers to: the flow execution unit receives the number n of the data tuples in the time period delta tinAnd the number n of data tuples sent to the monitoring analysis moduleoutRespectively counting to represent the message receiving rate and the message processing rate in the time period, wherein: the length of the time period delta t is determined by the latest average health value h of the equipment service related to the processing unit, the flow execution unit will periodically pull the latest average health value of the relevant service from the service registration center according to the service interface which the unit needs to access, and when the unit has a plurality of candidate equipment service instances, the health values corresponding to the service instances need to be averaged again to obtain
For calculating time periods
Wherein: t is t0Is a constant representing the lowest time interval;
the service registration center maintains the latest average health value of each equipment service, the latest average health value is inquired by the process execution unit to support backpressure control of the data flow, and the average health value of the equipment service is sent to the monitoring analysis module to support the abnormal detection process of the data.
10. The cloud-edge-coordinated intelligent routing system of the industrial internet of things as claimed in claim 9, wherein when the message receiving rate is higher than the message processing rate, the process execution unit starts a backpressure control strategy, specifically comprising: buffering, discarding and sampling, wherein: when the backpressure control is carried out through buffering, the flow execution unit locally maintains a buffer queue and firstly caches the message locally; when a local cache queue is full, the processing of a new message can be degraded to a discarding strategy, namely, the processing such as data fusion, service call and the like is not carried out, but only the failure data is sent to a monitoring analysis module for recording and analyzing an abnormal log, extra information is provided for a task in a business flow model for carrying out backpressure control through sampling to specify the sampling frequency of the data when the task generates backpressure in the data processing process, and the receiving and processing of the message are controlled according to the sampling frequency, so that the backlog of the data is reduced, after the backpressure control strategy is started to operate for a period of time, a flow execution unit starts to recover the work when the message processing rate is higher than the message receiving rate, the normal work of the system is prevented from being influenced by excessive backpressure control, a method for recovering the work is related to the adopted backpressure control strategy, and when the backpressure control is carried out and the backpressure control is degraded to the discarding strategy by adopting a buffer strategy, when no data is accumulated in the local cache queue, the back pressure control mode of the buffer strategy can be recovered, and if the message processing rate is higher than the message receiving rate in the next time period, the back pressure control strategy can be closed to recover normal work; when a sampling strategy is adopted for backpressure control, the sampling frequency is multiplied until the sampling frequency exceeds 100% after the sampling frequency is increased next time, and if the message processing rate is still higher than the message receiving rate in the next time period, the backpressure control strategy can be closed, and normal work is recovered;
the log comprises: the data processing process uses the service instance information, the service interface information, the input of the interface call, and the return result of the interface call.
11. The cloud-edge-coordinated intelligent routing system of the industrial internet of things as claimed in claim 9, wherein the anomaly detection module, namely the monitoring and analyzing module, performs anomaly detection on the data by using an anomaly detection algorithm based on a hierarchical time-series memory (HTM) network, specifically: the monitoring analysis module finds the latest health value of the instance in the local cache according to the service instance information used in the data processing process, encodes the service instance information, the health value and the interface information by using a scalar encoder, encodes the input and return results of interface calling by using a cooperative encoder, and transmits a plurality of encoded vectors into a multi-encoder to form a vector containing log information of a complete flow execution unit; then, the vector is processed by a space pool to obtain a Sparse Distributed Representation (SDR) and used as the input of the HTM network, the next input SDR in the current data stream is continuously predicted by the learning algorithm of the HTM, and when a new SDR (recorded as d') is received, an original abnormal score is calculated according to the predicted SDR (recorded as d) in the previous round
Wherein: d 'represents the scalar product of the SDR of the previous round of prediction with the new SDR array, | d' | represents the number of bits in the new SDR array representing all values of 1.
12. The cloud-edge-coordinated intelligent routing system of the industrial internet of things as claimed in claim 9, wherein the monitoring analysis module does not directly perform anomaly detection by setting a threshold value for an original anomaly score, but calculates the probability of current anomaly occurrence according to the distribution of the anomaly score, and performs anomaly detection by setting a threshold value for the anomaly probability, specifically: recording the values of the most recent W anomaly scores, then calculating the mean value mu and the standard deviation sigma of the values, regarding the distribution of the values as a Gaussian distribution, and calculating the average value mu and the standard deviation sigma of the values by means of the Gaussian distributionThe right-tail function (Q function) of the cloth to calculate the anomaly probability, i.e.
Wherein: stRepresenting the abnormal score obtained by the t-th calculation; the monitoring analysis module stores the input log and the abnormal detection result into log historical data, and generates a staged data analysis report at regular time for service personnel to analyze; and carrying out early warning notification on the relevant users when the detection finds the abnormity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010971570.0A CN114268640B (en) | 2020-09-16 | 2020-09-16 | Intelligent routing system of industrial Internet of things with cloud edge cooperation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010971570.0A CN114268640B (en) | 2020-09-16 | 2020-09-16 | Intelligent routing system of industrial Internet of things with cloud edge cooperation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114268640A true CN114268640A (en) | 2022-04-01 |
| CN114268640B CN114268640B (en) | 2022-09-09 |
Family
ID=80824180
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010971570.0A Active CN114268640B (en) | 2020-09-16 | 2020-09-16 | Intelligent routing system of industrial Internet of things with cloud edge cooperation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114268640B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114666555A (en) * | 2022-05-23 | 2022-06-24 | 创意信息技术股份有限公司 | Edge gateway front-end system |
| CN114979251A (en) * | 2022-04-26 | 2022-08-30 | 东莞市海数云电子科技有限公司 | Cross-system cooperative service heterogeneous data exchange tool |
| CN115484265A (en) * | 2022-09-19 | 2022-12-16 | 合肥合锻智能制造股份有限公司 | Multi-source heterogeneous data management system based on cloud edge collaboration |
| CN115643159A (en) * | 2022-11-18 | 2023-01-24 | 奇点创新(江苏)智能科技有限公司 | Equipment abnormity early warning method and system based on edge calculation |
| CN116612371A (en) * | 2023-07-18 | 2023-08-18 | 四川弘和数智集团有限公司 | Neural network reasoning method, device, equipment and medium based on Bian Yun decoupling |
| CN116708528A (en) * | 2023-08-02 | 2023-09-05 | 成都卓拙科技有限公司 | Method and device for monitoring TCP long connection between nodes |
| CN116737523A (en) * | 2023-08-10 | 2023-09-12 | 中国电子投资控股有限公司 | Observability data acquisition method based on edge calculation |
| CN117235559A (en) * | 2023-11-14 | 2023-12-15 | 四川万物纵横科技股份有限公司 | Internet of things data acquisition method and system based on edge calculation |
| WO2024000498A1 (en) * | 2022-06-30 | 2024-01-04 | 西门子股份公司 | Industrial control system, edge device, cloud server, apparatus, and method |
| CN118113787A (en) * | 2024-04-30 | 2024-05-31 | 中增智慧城市建设发展有限公司 | Smart city Internet of things data processing method and system based on edge calculation |
| CN118378202A (en) * | 2024-06-26 | 2024-07-23 | 苏州共营互联网科技有限公司 | Data processing method for ERP system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102880475A (en) * | 2012-10-23 | 2013-01-16 | 上海普元信息技术股份有限公司 | Real-time event handling system and method based on cloud computing in computer software system |
| US20170285710A1 (en) * | 2016-04-01 | 2017-10-05 | Intel Corporation | Processor power monitoring and control with dynamic load balancing |
| US20190018815A1 (en) * | 2017-07-01 | 2019-01-17 | Kermin Fleming | Processors, methods, and systems with a configurable spatial accelerator |
| CN109308290A (en) * | 2018-08-06 | 2019-02-05 | 国网安徽省电力有限公司信息通信分公司 | A kind of efficient data cleaning conversion method based on CIM |
| CN109885566A (en) * | 2019-02-25 | 2019-06-14 | 南京世界村云数据产业集团有限公司 | A kind of acquisition of data and edge calculations system |
| CN110035117A (en) * | 2019-03-15 | 2019-07-19 | 启迪云计算有限公司 | One kind is based on configurable monitoring script monitoring system and monitoring method |
| CN111587423A (en) * | 2017-11-13 | 2020-08-25 | 维卡艾欧有限公司 | Hierarchical data policy for distributed storage systems |
-
2020
- 2020-09-16 CN CN202010971570.0A patent/CN114268640B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102880475A (en) * | 2012-10-23 | 2013-01-16 | 上海普元信息技术股份有限公司 | Real-time event handling system and method based on cloud computing in computer software system |
| US20170285710A1 (en) * | 2016-04-01 | 2017-10-05 | Intel Corporation | Processor power monitoring and control with dynamic load balancing |
| US20190018815A1 (en) * | 2017-07-01 | 2019-01-17 | Kermin Fleming | Processors, methods, and systems with a configurable spatial accelerator |
| CN111587423A (en) * | 2017-11-13 | 2020-08-25 | 维卡艾欧有限公司 | Hierarchical data policy for distributed storage systems |
| CN109308290A (en) * | 2018-08-06 | 2019-02-05 | 国网安徽省电力有限公司信息通信分公司 | A kind of efficient data cleaning conversion method based on CIM |
| CN109885566A (en) * | 2019-02-25 | 2019-06-14 | 南京世界村云数据产业集团有限公司 | A kind of acquisition of data and edge calculations system |
| CN110035117A (en) * | 2019-03-15 | 2019-07-19 | 启迪云计算有限公司 | One kind is based on configurable monitoring script monitoring system and monitoring method |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114979251B (en) * | 2022-04-26 | 2023-08-11 | 广东海术云电子科技有限公司 | Cross-system collaborative service heterogeneous data exchange system |
| CN114979251A (en) * | 2022-04-26 | 2022-08-30 | 东莞市海数云电子科技有限公司 | Cross-system cooperative service heterogeneous data exchange tool |
| CN114666555A (en) * | 2022-05-23 | 2022-06-24 | 创意信息技术股份有限公司 | Edge gateway front-end system |
| WO2024000498A1 (en) * | 2022-06-30 | 2024-01-04 | 西门子股份公司 | Industrial control system, edge device, cloud server, apparatus, and method |
| CN115484265A (en) * | 2022-09-19 | 2022-12-16 | 合肥合锻智能制造股份有限公司 | Multi-source heterogeneous data management system based on cloud edge collaboration |
| CN115643159A (en) * | 2022-11-18 | 2023-01-24 | 奇点创新(江苏)智能科技有限公司 | Equipment abnormity early warning method and system based on edge calculation |
| CN116612371A (en) * | 2023-07-18 | 2023-08-18 | 四川弘和数智集团有限公司 | Neural network reasoning method, device, equipment and medium based on Bian Yun decoupling |
| CN116612371B (en) * | 2023-07-18 | 2023-11-03 | 四川弘和数智集团有限公司 | Neural network reasoning method, device, equipment and medium based on Bian Yun decoupling |
| CN116708528A (en) * | 2023-08-02 | 2023-09-05 | 成都卓拙科技有限公司 | Method and device for monitoring TCP long connection between nodes |
| CN116708528B (en) * | 2023-08-02 | 2023-10-03 | 成都卓拙科技有限公司 | Method and device for monitoring TCP long connection between nodes |
| CN116737523B (en) * | 2023-08-10 | 2023-12-01 | 中国电子投资控股有限公司 | Observability data acquisition method based on edge calculation |
| CN116737523A (en) * | 2023-08-10 | 2023-09-12 | 中国电子投资控股有限公司 | Observability data acquisition method based on edge calculation |
| CN117235559A (en) * | 2023-11-14 | 2023-12-15 | 四川万物纵横科技股份有限公司 | Internet of things data acquisition method and system based on edge calculation |
| CN117235559B (en) * | 2023-11-14 | 2024-02-06 | 四川万物纵横科技股份有限公司 | Internet of things data acquisition method and system based on edge calculation |
| CN118113787A (en) * | 2024-04-30 | 2024-05-31 | 中增智慧城市建设发展有限公司 | Smart city Internet of things data processing method and system based on edge calculation |
| CN118113787B (en) * | 2024-04-30 | 2024-06-28 | 中增智慧城市建设发展有限公司 | Smart city Internet of things data processing method and system based on edge calculation |
| CN118378202A (en) * | 2024-06-26 | 2024-07-23 | 苏州共营互联网科技有限公司 | Data processing method for ERP system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114268640B (en) | 2022-09-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114268640B (en) | Intelligent routing system of industrial Internet of things with cloud edge cooperation | |
| US11169506B2 (en) | Predictive data capture with adaptive control | |
| US10685283B2 (en) | Demand classification based pipeline system for time-series data forecasting | |
| US7720841B2 (en) | Model-based self-optimizing distributed information management | |
| US10084722B2 (en) | Modification of computing resource behavior based on aggregated monitoring information | |
| US20210042578A1 (en) | Feature engineering orchestration method and apparatus | |
| US20200134421A1 (en) | Assurance of policy based alerting | |
| CN103761309A (en) | Operation data processing method and system | |
| CN111309550A (en) | Data acquisition method, system, equipment and storage medium of application program | |
| US20210065083A1 (en) | Method for changing device business and business change system | |
| Renart et al. | Distributed operator placement for IoT data analytics across edge and cloud resources | |
| CN112000350B (en) | Dynamic rule updating method, device and storage medium | |
| US9842140B2 (en) | Dynamic input streams handling in DSMS | |
| US20230283530A1 (en) | Service assurance monitoring based on telemetry | |
| CN116166505A (en) | Monitoring platform, method, storage medium and equipment for dual-state IT architecture in financial industry | |
| US20230376372A1 (en) | Multi-modality root cause localization for cloud computing systems | |
| CN116566831A (en) | Mobile network resource management method and system based on cloud computing | |
| Ali et al. | Probabilistic normed load monitoring in large scale distributed systems using mobile agents | |
| CN113114480A (en) | Data reporting method and related equipment | |
| Motlagh et al. | Quality of monitoring for cellular networks | |
| KR101878291B1 (en) | Big data management system and management method thereof | |
| Zehnder et al. | Using virtual events for edge-based data stream reduction in distributed publish/subscribe systems | |
| WO2023093257A1 (en) | Modeling method for self-driving service and related device | |
| CN118316981B (en) | Cloud edge cooperative system for data processing of Internet of things equipment | |
| Mu | Heterogeneous IoT Information Distributed Collection System Based on Embedded Technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |