CN-122027645-A - IoT (internet of things) patrol network collaborative perception and decision-making system based on edge intelligence

CN122027645ACN 122027645 ACN122027645 ACN 122027645ACN-122027645-A

Abstract

The invention provides an IoT (internet of things) patrol network collaborative awareness and decision-making system based on edge intelligence, and relates to the technical field of the internet of things. The system realizes the fine management of heterogeneous resources by dynamically generating a device operation description body containing executable actions and operation constraints for access devices, generates unified state data by structuring original perception data, generates candidate collaborative behavior schemes based on multi-device state data, synchronously establishes traceable forming path records, performs consistency check and failure tracing on the schemes by utilizing the path records, positions conflict sources by means of inverse dependency analysis, generates adjustment information to update the schemes, realizes self-optimization, and finally quantitatively evaluates the feasible schemes and issues new instructions. The invention improves the adaptability, traceability and overall efficiency of the cooperation of multiple devices in the dynamic patrol environment.

Inventors

SUN QINGQING
LI ZHIQIANG
LIU KUIWEI

Assignees

嘉和众拓科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260114

Claims (8)

1. An edge intelligence based IoT patrol network collaborative awareness and decision-making system, comprising: the equipment access and management unit is configured to receive access requests of all edge equipment in the patrol network and generate corresponding equipment operation description bodies for each access equipment, wherein the equipment operation description bodies comprise equipment identifiers, executable action sets, operation constraint parameters and current cooperative state identifiers; The data acquisition preprocessing unit is configured to receive original perception data from the edge equipment, and perform structural processing on the original perception data based on the equipment operation description body to generate state data associated with corresponding equipment identifiers; The distributed computing unit is configured to receive state data from a plurality of edge devices and generate a collaborative behavior scheme set based on the state data, wherein each collaborative behavior scheme is associated with a forming path record, and the forming path record is used for identifying a state data source participating in generating the collaborative behavior scheme and a processing sequence thereof; The network collaborative management unit is configured to receive the collaborative behavior scheme set, form a path record and a device operation description body, and perform consistency verification on the collaborative behavior scheme set based on the device operation description body, and perform inverse dependence analysis according to the formed path record to determine a source causing verification failure when verification fails; and the collaborative decision-making and executing unit is configured to quantitatively evaluate the collaborative behavior scheme which passes the verification and to execute decision-making of the edge equipment based on the quantitative evaluation.
2. The IoT patrol network collaborative awareness and decision-making system based on edge intelligence according to claim 1, wherein the device access and management unit is configured to receive the access request, directly separate a first field from the access request as a device identifier, separate a candidate action list described in a second field, compare the candidate action list with an effective action range reflected by status data currently processed by the data collection preprocessing unit, bring candidate actions belonging to the effective action range into the executable action set, dynamically calculate an upper limit of a single task data amount and an interval between adjacent tasks in the operation constraint parameter according to a preset proportion according to a total number of devices with which the current collaborative status is not collaborative, assign the current collaborative status identifier as ready, and finally bind the device identifier, the filtered executable action set, the dynamically calculated operation constraint parameter and the current collaborative status identifier of the ready status to complete generation of a device operation descriptor.
3. The IoT patrol network collaborative awareness and decision-making system based on edge intelligence according to claim 1, wherein the data collection preprocessing unit is configured to receive raw awareness data, extract a device identifier in the device operation description, reversely deduce a data category necessary for generating the state data from an executable action set according to a device type indicated by the device identifier, parse and classify the raw awareness data according to the deduced data category, distribute data belonging to different categories to different internal storage queues, read an upper limit of a single task data amount in the operation constraint parameter, and respectively perform data normalization operation for each internal storage queue, wherein when the number of data units in the queue exceeds the upper limit of the single task data amount, remove overage data with earliest generation time, and when the number of the data units in the queue does not reach an upper limit, copy and fill the last data unit in the queue until the number reaches the upper limit, finally, combine the data units in the queues in a regular order according to a data output format fixed with the device type, and combine the data units in the queues and the regular order according to a data output format fixed with the device type, and generate the state data associated with the device identifier.
4. The IoT patrol network collaborative awareness and decision-making system based on edge intelligence according to claim 1, wherein the distributed computing unit is configured to receive status data from a plurality of edge devices, retrieve, for each received status data, a corresponding device operation description according to its associated device identifier to obtain the set of executable actions, group the plurality of status data received within a same time window in units of a time window, traverse the set of executable actions corresponding to all devices for each group of status data, rank and combine actions therein to generate a preliminary action sequence of all probabilities, perform a feasibility assessment on each preliminary action sequence according to specific values contained in each status data, generate an intermediate decision result including a constraint check result, retain a sequence conforming to all device operation constraint parameters as candidate collaborative action schemes, and record, while generating each candidate collaborative action scheme, the device identifier of each status data according to the sequence in which the status data is retrieved and processed, and the specific step number used for the device identifier of each status data according to record, so as to generate the path record scheme.
5. The IoT patrol network collaborative awareness and decision-making system based on edge intelligence according to claim 4, wherein the forming path record includes creating a tree path record structure for each candidate collaborative behavior scheme synchronously, creating a common second-layer node for the first-layer nodes corresponding to the state data according to which the first-layer nodes correspond to generate the scheme, and recording the step numbers, all associated device identifications and intermediate decision results generated based on the first-layer nodes, creating a second-layer node under the corresponding device identification nodes when performing constraint check on one action in the preliminary action sequence based on one state data in the feasibility evaluation process, recording the checked actions, the corresponding step numbers and the intermediate decision results of the second-layer nodes, creating a common second-layer node for the first-layer nodes corresponding to the state data if the check on one step number needs to be logically judged simultaneously, and recording the step numbers, all associated device identifications and the intermediate decision results generated based on the first-layer nodes, and performing a final-layer node structure according to the step numbers and the step numbers of the first-layer nodes and the second-layer nodes after the complete evaluation, forming a tree structure.
6. The IoT (internet of things) patrol network collaborative awareness and decision-making system based on edge intelligence according to claim 1, wherein the network collaborative management unit is configured to perform the consistency check by traversing the collaborative behavior scheme set, extracting an action sequence scheduled to be executed by each device in each collaborative behavior scheme and a corresponding device identifier, comparing each device identifier with an executable action set in a device operation description body of each device identifier, confirming that each action is contained in the set, and simultaneously calculating a total data amount and an execution time interval corresponding to an action sequence that the collaborative behavior scheme requires the device to execute according to an operation constraint parameter in the device operation description body, and comparing the total data amount and the execution time interval with a single task data amount upper limit and an adjacent task interval in the operation constraint parameter respectively.
7. The IoT patrol network collaborative awareness and decision-making system based on edge intelligence according to claim 1, wherein when comparing with the upper limit of the single task data amount and the adjacent task interval in the operation constraint parameter, the network collaborative management unit performs an inverse dependency analysis according to the forming path record to determine a source, and includes searching all nodes containing the device identifier in the corresponding forming path record according to the recorded device identifier which does not pass, identifying the node with the largest step number from the nodes, positioning an evaluation operation to last use the device status data when the distributed computing unit generates the scheme according to the step number, tracing forward along the node hierarchy and sequence relationship recorded in the forming path record, analyzing intermediate judgment results which are relied by the evaluation operation and are generated by participation of other device status data, and determining the intermediate judgment results, all other device identifiers recorded in the forming path record and the source which does not pass, as the source which causes the failure of the verification.
8. The edge-intelligence based IoT defense network collaborative awareness and decision-making system according to claim 1, wherein the collaborative decision-making and execution unit is configured to: Receiving a set of co-behavioural schemes that pass the verification, performing, for each scheme, the following quantitative evaluation: Generating a time sequence coordination score of the scheme based on a planning action sequence of each device and corresponding operation constraint parameters thereof, wherein the time sequence coordination score is specifically calculated by comparing a time interval sequence of adjacent actions in the planning action sequence with an adjacent task interval sequence in the operation constraint parameters and calculating the coincidence proportion of the two sequences at all corresponding positions; Calculating the predicted time consumption by utilizing the total data quantity of the real-time acquisition frequency and the planning action sequence, and calculating the deviation degree of the predicted time consumption and the historical average task time consumption to be used as the efficiency predicted score; And selecting the scheme with the highest comprehensive evaluation value as an execution decision, and issuing corresponding action instructions to each device.

Description

IoT (internet of things) patrol network collaborative perception and decision-making system based on edge intelligence Technical Field The invention relates to the technical field of the Internet of things, in particular to an internet of things (IoT) patrol network collaborative awareness and decision-making system based on edge intelligence. Background In the field of intelligent patrol driven by the current internet of things technology, a distributed sensing device is connected with an executing device through a network, and the distributed sensing device and the executing device are tried to work cooperatively to complete tasks such as monitoring, early warning and intervention, and the like, so that the intelligent patrol has become an important technical development direction. The existing system architecture is usually focused on centralized access and instruction issuing of equipment, and can achieve a certain degree of task coordination. However, in the dynamically changing patrol environment, the system has difficulty in accurately verifying the constraint compliance of the feasibility of the collaborative scheme matched with the current context before the task is executed in the face of the situations of equipment isomerism, task diversity and real-time fluctuation of the resource state. When the cooperative behavior cannot be executed due to constraint conflict or logic contradiction, the existing method can only give the overall failure result, but cannot clearly reveal the root cause of the failure and the specific source thereof, so that the system adjustment and optimization lack clear and efficient basis, and the overall cooperative robustness and adaptation are limited. Disclosure of Invention In order to achieve the purpose, the invention is realized by the following technical scheme that the IoT patrol network collaborative awareness and decision-making system based on edge intelligence comprises: the equipment access and management unit is configured to receive access requests of all edge equipment in the patrol network and generate corresponding equipment operation description bodies for each access equipment, wherein the equipment operation description bodies comprise equipment identifiers, executable action sets, operation constraint parameters and current cooperative state identifiers; The data acquisition preprocessing unit is configured to receive original perception data from the edge equipment, and perform structural processing on the original perception data based on the equipment operation description body to generate state data associated with corresponding equipment identifiers; The distributed computing unit is configured to receive state data from a plurality of edge devices and generate a collaborative behavior scheme set based on the state data, wherein each collaborative behavior scheme is associated with a forming path record, and the forming path record is used for identifying a state data source participating in generating the collaborative behavior scheme and a processing sequence thereof; The network collaborative management unit is configured to receive the collaborative behavior scheme set, form a path record and a device operation description body, and perform consistency verification on the collaborative behavior scheme set based on the device operation description body, and perform inverse dependence analysis according to the formed path record to determine a source causing verification failure when verification fails; and the collaborative decision-making and executing unit is configured to quantitatively evaluate the collaborative behavior scheme which passes the verification and to execute decision-making of the edge equipment based on the quantitative evaluation. The device access and management unit is used for realizing dynamic management of the newly accessed edge device; when an access request is received, a field splitting operation is first performed to extract a predefined first field directly from the request message or data frame as a globally unique device identification, such as a device MAC address or factory serial number, while extracting a second field, which is a structured list describing all candidate actions supported by the device hardware or firmware, such as image acquisition, temperature reading, temperature determining, and/or temperature determining, Mobile control or audible and visual alarm; to generate a feasible set of executable actions, the unit introduces a dynamic screening mechanism; the method comprises the steps of comparing a candidate action list with an effective action range reflected by state data which is processed and cached currently by a data acquisition preprocessing unit in real time, wherein the effective action range represents an action subset which is proved to be effective and supported by other cooperative devices in the context and environment of a current network task, bringing actions which are both