Power internet of things agent system for dynamic resource management

Abstract

The application discloses a power internet of things agent system for dynamic resource management, which is connected between an internet of things terminal and an internet of things management platform, and comprises a device atomic layer, a device arrangement layer and a network communication layer; the device atomic layer, the device arrangement layer and the network communication layer are sequentially connected in series; the device atomic layer is connected with the internet of things terminal and realizes data acquisition of the internet of things terminal through protocol analysis and data preprocessing; the device arrangement layer realizes connection with the device atomic layer and the network communication layer through a software defined network interface, and realizes matching of resources and tasks through construction of a virtual resource pool; the network communication layer is connected with the internet of things management platform through an encryption channel, and realizes hierarchical transmission of data of the internet of things terminal through a master-standby communication link and a queue buffer mode. The application not only realizes agent data transmission between the internet of things terminal and the internet of things management platform, but also has higher reliability and better real-time performance.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of electrical automation, and particularly relates to a power internet of things agent system for dynamic resource management. BACKGROUND

[0002] With the development of economy and technology and the improvement of people's living standards, electric energy has become an essential secondary energy in people's production and life, bringing endless convenience to people's production and life. Therefore, ensuring the stable and reliable supply of electric energy has become one of the most important tasks of the power system.

[0003] At present, the power system has developed rapidly towards the direction of internet of things and intelligence. The power internet of things has gradually become the core support of the intelligence of the power system. The goal of the power internet of things is to realize real-time sensing, remote control and data analysis of device status through the interconnection of a large number of terminal devices (such as smart meters, sensors, relay protection devices, etc.). However, the large-scale deployment of the power internet of things faces the following technical bottlenecks: (1) device heterogeneity: electric terminal devices involve multiple communication protocols (such as RS-485, Modbus-TCP, DL / T645-2007, etc.) and interface types, resulting in high system integration complexity; (2) resource limitation: edge device computing power is limited, making it difficult to cope with high-concurrency data processing requirements, and traditional centralized resource scheduling mode cannot adapt to dynamic business load; (3) unreliable communication: power control type business (such as relay protection) has strict requirements on delay and reliability (millisecond-level response), but existing communication links are easily disturbed by the environment, and lack intelligent redundancy mechanism; (4) contradiction between data quality and real-time performance: raw data on the edge side often contains noise, and traditional cloud cleaning scheme has high delay, and local processing algorithm lacks dynamic adaptability.

[0004] At present, the power system adopts transmission schemes including single-link redundancy backup scheme and priority queue scheduling scheme to solve the above problems. The single-link redundancy backup scheme realizes master-slave link switching through double network cards or double SIM cards; the priority queue scheduling scheme divides the transmission queue based on the type of business (such as QoS classification in CoAP protocol). However, the single-link redundancy backup scheme realizes link switching through traditional heartbeat detection mechanism (such as TCP retransmission), and the switching time is usually in seconds, which is difficult to meet the millisecond-level switching requirement of power control type business. In the priority queue scheduling scheme, the priority and bandwidth allocation are fixed, and cannot be dynamically adjusted according to the network state. SUMMARY

[0005] The present application aims to provide a power internet of things agent system for dynamic resource management with high reliability and good real-time performance.

[0006] The power IoT agent system for dynamic resource management provided by this invention connects IoT terminals and an IoT management platform, and includes a device atomic layer, a device orchestration layer, and a network communication layer. These three layers are connected in series. The device atomic layer connects to the IoT terminals and, through protocol parsing and data preprocessing, enables data acquisition from the IoT terminals. The device orchestration layer connects to the device atomic layer and the network communication layer via a software-defined network interface, and achieves resource-task matching through the construction of a virtual resource pool. The network communication layer connects to the IoT management platform through an encrypted channel and, through primary and backup communication links and a queue buffer mode, enables hierarchical transmission of IoT terminal data.

[0007] The device orchestration layer includes a virtual resource pool module and a dynamic scheduling module connected in series. The virtual resource pool module constructs a digital twin model of IoT devices based on graph database technology, then calculates the performance consumption value of each IoT device based on the constructed digital twin model, and sets resource profile tags for each IoT device based on the performance consumption value. The dynamic scheduling module sets the dynamic scheduling strategy of IoT devices based on the data information uploaded by the virtual resource pool module, and generates acquisition and transmission strategies based on the Nash equilibrium scheme.

[0008] The device atomic layer includes a protocol parsing module and an edge preprocessing module connected in series. The protocol parsing module is used to realize the power system protocol parsing and data structure conversion of various IoT terminals. The edge preprocessing module is used to perform standardization processing and noise warning on the received data.

[0009] The protocol parsing module's processing procedure includes the following steps:

[0010] Data information from various IoT terminals is collected through protocols supported by the power system; the protocols supported by the power system include RS-485 protocol, Modbus-TCP protocol and DL / T645-2007 protocol.

[0011] The received data from each IoT terminal is converted into a JSON structure conforming to the IEC 61850 standard; the JSON structure includes device ID, timestamp, measurement value, and extended field checksum.

[0012] The edge preprocessing module's processing steps include the following:

[0013] Data standardization: The acquired data is standardized using a sliding window algorithm as follows:

[0014]

[0015] In the formula Standardized data information; is the data information before standardization; is the data mean value in the sliding window; is the data standard deviation in the sliding window; the window length is set according to the corresponding IoT device of the obtained data; the sliding step size is set as , is a set proportion coefficient, and w is the length of the sliding window;

[0016] Noise processing: the value of the noise level L is defined as ; when L is less than a first noise threshold, it is marked as no noise; when L is greater than or equal to the first noise threshold and less than a second noise threshold, it is marked as mild noise, at which time the data is retained and an abnormal identifier is added in the data; when L is greater than or equal to the second noise threshold, it is marked as severe noise, at which time the data is discarded and an alarm is reported; wherein the first noise threshold is less than the second noise threshold.

[0017] The processing process of the virtual resource pool module includes the following steps:

[0018] The Neo4j graph database is used to construct the digital twin model of the IoT device; the device attribute in the digital twin model of the i-th IoT device includes the real-time state vector , wherein is the CPU usage rate interacting with the i-th IoT device, is the memory usage rate interacting with the i-th IoT device, is the network delay interacting with the i-th IoT device;

[0019] The real-time state vector of the i-th IoT device is normalized to a set scale, represented as , wherein is the normalized real-time state vector of the i-th IoT device, is the normalized , is the normalized , is the normalized ;

[0020] The performance consumption value of the i-th IoT device is calculated as , wherein is a set first weight value, is a set second weight value, is a set third weight value;

[0021] For all IoT devices, the mean value and the variance of the performance consumption value are calculated ;

[0022] Set the resource image tag of the connected device:

[0023] If , the i-th connected device is determined as a high-performance consumption device;

[0024] If , the i-th connected device is determined as a central energy consumption device;

[0025] If , the i-th connected device is determined as a low-performance consumption device.

[0026] The processing procedure of the dynamic scheduling module includes the following steps:

[0027] A. According to the data information of the virtual resource pool module, a scheduling scheme is generated:

[0028] If the service type is a control instruction, the priority is set to the highest level, and the transmission strategy is a preemptive real-time queue;

[0029] If the service type is real-time data uploading, the priority is set to high, and the transmission strategy is an elastic real-time queue;

[0030] If the service type is periodic data uploading, the priority is set to medium, and the transmission strategy is a first buffer queue;

[0031] If the service type is batch data transmission, the priority is set to low, and the transmission strategy is a second buffer queue;

[0032] The preemptive real-time queue is defined as a resource that can preempt a low-priority queue and is always sent first. The elastic real-time queue is defined as a resource that does not preempt a low-priority queue and is sent first when the bandwidth is higher than a set threshold, and is downgraded to a buffer queue when the bandwidth is lower than or equal to the set threshold. The first buffer queue is defined as having the highest priority in the buffer queue, but the priority is lower than the downgraded elastic real-time queue, and the bandwidth allocation ratio is greater than a set threshold. The second buffer queue is defined as allowing data transmission according to a set delay value and packet loss rate, and the bandwidth allocation ratio is lower than a set threshold.

[0033] B. According to the data information of the virtual resource pool module, a collection strategy and a transmission strategy are generated:

[0034] The resource competition process is set as a non-cooperative game process, and Nash equilibrium is used to realize resource allocation, so as to ensure that the CPU usage rate and memory occupancy rate of each connected device do not exceed a set threshold, and the difference between the performance consumption values of each connected device is within a set range, avoiding a single connected device monopolizing resources and ensuring fairness.

[0035] The network communication layer comprises a dual-pass routing module and a dual-queue buffer module; the dual-pass routing module is used for switching the sending communication channel according to the set primary and secondary channel switching mechanism; and the dual-queue buffer module is used for carrying out data communication according to the set dual-queue mode according to the queue content.

[0036] The processing procedure of the dual-pass routing module comprises the following steps:

[0037] The configured dual channel comprises a fiber link and a 5G wireless link;

[0038] The normal communication process adopts the fiber link; when the UDP probe detects that consecutive probe packets are lost or the time delay exceeds ms, the link switching is started, and the communication link is switched to the 5G wireless link.

[0039] The processing procedure of the dual-queue buffer module comprises the following steps:

[0040] The configured dual queue comprises a real-time queue and a cache queue;

[0041] For the real-time queue: the priority scheduling mode is adopted, and the data transmission is carried out according to the generated scheduling scheme;

[0042] For the cache queue: when it is monitored that the bandwidth utilization is less than the set value, the batch transmission is triggered, and the packet survival time is set to seconds.

[0043] The power internet-of-things agent system for dynamic resource management provided by the application is connected between the internet-of-things terminal and the internet-of-things management platform, adopts a hierarchical connection mode, takes business interaction as the target, and through data acquisition, resource-task matching and main and backup communication links and queue buffer modes, not only realizes the agent data transmission between the internet-of-things terminal and the internet-of-things management platform, but also has higher reliability and better real-time performance. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 It is a functional module schematic diagram of the system of the application. DETAILED DESCRIPTION

[0045] As Figure 1The functional module schematic diagram of the system is shown: the power internet of things agent system for dynamic resource management disclosed by the application is connected between the internet of things terminal and the internet of things management platform, and comprises a device atomic layer, a device arrangement layer and a network communication layer; the device atomic layer, the device arrangement layer and the network communication layer are sequentially connected in series; the device atomic layer is used for connecting the internet of things terminal and realizing data acquisition of the internet of things terminal through protocol analysis and data preprocessing; the device arrangement layer realizes connection with the device atomic layer and the network communication layer through a software defined network interface, and realizes matching of resources and tasks through construction of a virtual resource pool; the network communication layer is connected with the internet of things management platform through an encrypted channel, and realizes hierarchical transmission of data of the internet of things terminal through a master-standby communication link and a queue buffer mode;

[0046] The device arrangement layer comprises a virtual resource pool module and a dynamic scheduling module connected in series; the virtual resource pool module constructs a digital twin model of the internet of things device based on a graph database technology, then calculates performance consumption values of the internet of things devices according to the constructed digital twin model, and sets resource portrait tags of the internet of things devices according to the performance consumption values; the dynamic scheduling module sets a dynamic scheduling strategy of the internet of things device according to data information uploaded by the virtual resource pool module, and generates a collection strategy and a transmission strategy based on a Nash equilibrium scheme.

[0047] In specific implementation, the device atomic layer comprises a protocol analysis module and an edge preprocessing module connected in series; the protocol analysis module is used for realizing power system protocol analysis and data structure conversion of various internet of things terminals; and the edge preprocessing module is used for standardizing the received data and warning of noise.

[0048] The processing process of the protocol analysis module comprises the following steps:

[0049] Data information of each internet of things terminal is collected through a protocol mode supported by a power system; the protocol mode supported by the power system comprises an RS-485 protocol, a Modbus-TCP protocol and a DL / T645-2007 protocol; through the protocol analysis process, plug-and-play access of multi-protocol power equipment can be realized.

[0050] The received data information of each internet of things terminal is converted into a JSON structure body conforming to an IEC 61850 standard; the JSON structure body comprises a device ID, a timestamp, a measurement value and an extended field check code.

[0051] The processing process of the edge preprocessing module comprises the following steps:

[0052] Data standardization: the acquired data information is subjected to sliding window standardization processing by using the following formula:

[0053]

[0054] wherein is the standardized data information; is the non-standardized data information; is the data mean value in the sliding window; is the data standard deviation in the sliding window; the window length is set according to the corresponding IoT device of the obtained data; the sliding step is the value of , is the set proportion coefficient (preferably the value is 0.1), and w is the length of the sliding window (preferably the value is );

[0055] Noise processing: the value of the noise level L is defined as ; when L is less than the first noise threshold, it is marked as no noise; when L is greater than or equal to the first noise threshold and less than the second noise threshold, it is marked as light noise, at which time the data is retained and an abnormal identifier (such as adding an abnormal identifier 0xEE) is added in the data; when is greater than or equal to the second noise threshold, it is marked as heavy noise, at which time the data is discarded and an alarm is reported; wherein the first noise threshold is less than the second noise threshold, and the first noise threshold is preferably 3 and the second noise threshold is preferably 5.

[0056] In specific implementation, the processing process of the virtual resource pool module includes the following steps:

[0057] A Neo4j graph database is used to construct a digital twin model of the IoT device; the device attributes in the digital twin model of the i-th IoT device include a real-time state vector , wherein is the CPU usage rate interacting with the i-th IoT device, is the memory usage rate interacting with the i-th IoT device, is the network delay interacting with the i-th IoT device;

[0058] The real-time state vector of the i-th IoT device is normalized to a set scale (preferably normalized to the range of 0~1), represented as , wherein is the normalized real-time state vector of the i-th IoT device, is the normalized , is the normalized , is the normalized ;

[0059] Calculate the performance consumption value of the ith IoT device For , is a set first weight value, is a set second weight value, is a set third weight value; in specific implementation, , and can be set according to specific circumstances, for example, if CPU usage is emphasized, the value of can be increased;

[0060] For all IoT devices, the mean value and the variance of the performance consumption value are calculated;

[0061] The resource profile label of the set IoT device is:

[0062] If , the ith IoT device is determined to be a high-performance consumption device;

[0063] If , the ith IoT device is determined to be a center performance consumption device;

[0064] If , the ith IoT device is determined to be a low-performance consumption device.

[0065] In specific implementation, the processing process of the dynamic scheduling module includes the following steps:

[0066] A. According to the data information of the virtual resource pool module, a scheduling scheme is generated:

[0067] If the service type is a control instruction (such as relay protection data), the priority is set to the highest level, and the transmission strategy is preemptive real-time queue;

[0068] If the service type is real-time data uploading (such as fault recording data), the priority is set to high level, and the transmission strategy is elastic real-time queue;

[0069] If the service type is periodic data uploading (such as device state monitoring data), the priority is set to intermediate level, and the transmission strategy is the first buffer queue;

[0070] If the service type is batch data transmission (such as historical data return data), the priority is set to low level, and the transmission strategy is the second buffer queue;

[0071] The pre-emptive real-time queue is defined as being able to pre-empt the resources of a low-priority queue and always being sent in priority; the elastic real-time queue is defined as not pre-empting the resources of a low-priority queue and being sent in priority when the bandwidth is higher than a set threshold and being degraded to a cache queue when the bandwidth is lower than or equal to the set threshold; the first cache queue is defined as having the highest priority in the cache queue but a priority lower than the degraded elastic real-time queue and having a bandwidth allocation ratio greater than a set threshold (preferably 20%); and the second cache queue is defined as allowing data transmission according to a set delay value and packet loss rate (generally allowing a higher delay and packet loss rate) and having a bandwidth allocation ratio lower than a set threshold (preferably 10%);

[0072] B. According to the data information of the virtual resource pool module, generate collection strategy and transmission strategy:

[0073] The resource competition process is set as a non-cooperative game process, and Nash equilibrium is used to realize resource allocation, so as to ensure that the CPU usage rate and memory occupancy rate of each Internet of Things device do not exceed the set threshold, and the difference between the performance consumption values of each Internet of Things device is within the set range, avoiding that a single Internet of Things device monopolizes resources and ensuring fairness.

[0074] In specific implementation, the network communication layer includes a dual-channel routing module and a dual-queue buffer module; the dual-channel routing module is used to switch the sending communication channel according to a set primary-secondary channel switching mechanism; and the dual-queue buffer module is used to perform data communication according to the queue content according to a set dual-queue mode.

[0075] The processing process of the dual-channel routing module includes the following steps:

[0076] The configured dual channel includes a fiber link and a 5G wireless link;

[0077] The normal communication process uses the fiber link; when the UDP probe detects that consecutive probe packets (preferably 3 probe packets) are lost or the delay exceeds ms (preferably 100 ms), the link switching is started, and the communication link is switched to the 5G wireless link.

[0078] The processing process of the dual-queue buffer module includes the following steps:

[0079] The configured dual queue includes a real-time queue and a cache queue;

[0080] For the real-time queue: a priority scheduling mode is used, and data transmission is performed according to the generated scheduling scheme;

[0081] For the cache queue: when the bandwidth utilization is less than a set value (preferably 70%), trigger batch transmission, and set the packet survival time to seconds (preferably 300 seconds).

[0082] In the scheme, plug-and-play access of multi-protocol power equipment is realized, mainstream power protocols (such as RS-485, Modbus-TCP, DL / T645-2007) and standardized data conversion (IEC 61850 JSON structure) are supported, and system compatibility is improved; through the establishment of a virtual resource pool and dynamic resource arrangement for terminal equipment, the paradigm of power edge agents changes from "device-oriented" to "service-oriented", not only solving the pain points of tight coupling and low efficiency of traditional systems, but also providing support for flexible services of smart grids; through the virtual resource pool and dynamic scheduling engine, the device resource state (CPU, memory, latency) is monitored in real time, and the resource- task matching closed-loop control is realized based on the resource portrait label and the non-cooperative game model (Nash equilibrium), which improves the resource utilization and system dynamic adaptability; through the multi-factor decision algorithm and priority mapping mechanism, four levels of business priority and corresponding transmission strategies (such as preemptive queue) are defined, combined with device resource portrait and communication layer state, the collection frequency and transmission strategy are dynamically adjusted to ensure the resource exclusivity of high-priority tasks and the overall fairness of the system; through the dual-channel intelligent routing (fiber + 5G redundant link) and millisecond-level switching mechanism (UDP probe detects packet loss or latency > 100ms triggers), combined with dual-queue hierarchical transmission (real-time queue preemptive scheduling + cache queue dynamic triggering), the low latency and high reliability of high-priority services (such as relay protection) are guaranteed; through the sliding window standardization, dynamic threshold adjustment and noise classification processing (light noise identification reservation, heavy noise discard + alarm), the data cleaning efficiency and quality are considered, and the dynamic data fluctuations caused by device state switching are adapted.

[0083] The following further illustrates the method of the application in combination with an embodiment: a typical existing scheme A (based on static queue combined with traditional centralized scheduling architecture) is selected as a comparison object, and under the same data set conditions (30 terminal devices of a certain 110kV substation, 450 data collected per minute), tests are carried out in three typical application scenarios. The performance of the existing mainstream scheme is compared in terms of average data transmission delay, high-priority service transmission success rate, resource scheduling response time and data processing accuracy, and the results are shown in Table 1 as follows:

[0084] Table 1 Performance evaluation result table

[0085]

[0086] As can be seen from the above table, the method of the application effectively overcomes the problems of resource preemption chaos and single-device resource exclusive in the traditional method through virtual resource pool modeling and non-cooperative game scheduling strategy; significantly enhances the low latency and high reliability of high-priority services through intelligent switching of dual-path routing and the UDP probe mechanism; and optimizes the data quality through a hierarchical identification mechanism in noise processing, avoiding false positives and data pollution.

Claims

1. A power internet of things agent system for dynamic resource management, characterized in that The connection is between the Internet of Things terminal and the Internet of Things management platform, including a device atomic layer, a device arrangement layer and a network communication layer; the device atomic layer, the device arrangement layer and the network communication layer are sequentially connected; the device atomic layer is used for connecting the Internet of Things terminal, and through protocol analysis and data preprocessing, the data acquisition of the Internet of Things terminal is realized; the device arrangement layer realizes the connection with the device atomic layer and the network communication layer through the software defined network interface, and realizes the matching of resources and tasks through the construction of a virtual resource pool; the network communication layer is connected with the Internet of Things management platform through an encrypted channel, and realizes the hierarchical transmission of the Internet of Things terminal data through the master and backup communication links and the queue buffer mode; The device arrangement layer includes a virtual resource pool module and a dynamic scheduling module which are connected in sequence; the virtual resource pool module constructs a digital twin model of the Internet of Things device based on a graph database technology, then calculates the performance consumption value of each Internet of Things device according to the constructed digital twin model, and sets the resource portrait label of each Internet of Things device according to the performance consumption value; the dynamic scheduling module sets the dynamic scheduling strategy of the Internet of Things device according to the data information uploaded by the virtual resource pool module, and generates the collection strategy and the transmission strategy based on the Nash equilibrium scheme; In specific implementation, the processing process of the virtual resource pool module includes the following steps: adopting a Neo4j graph database to construct a digital twin model of the i-th IoT device; the device attributes in the digital twin model of the i-th IoT device include a real-time state vector wherein is a CPU usage rate for interacting with the i-th IoT device, is a memory usage rate for interacting with the i-th IoT device, is a network latency for interacting with the i-th IoT device; real-time state vector of the i-th IoT device normalized to a set scale, denoted as wherein is the normalized real-time state vector of the i-th IoT device, is the normalized , is the normalized , is the normalized ; calculating a performance consumption value of the ith iot device for wherein is a set first weight value, is a set second weight value, is a set third weight value; For all the connected devices, the mean value of the performance consumption value is calculated and the variance ; The set resource portrait label of the Internet of Things device: If then determine the ith IoT device as a high-performance consuming device; If then determine that the ith IoT device is a central consumable device; If then determine the ith IoT device as a low-performance consuming device; The processing process of the dynamic scheduling module includes the following steps: A. According to the data information of the virtual resource pool module, a scheduling scheme is generated: If the business type is a control instruction, the priority is set to the highest level, and the transmission strategy is a preemptive real-time queue; If the business type is real-time data uploading, the priority is set to high level, and the transmission strategy is an elastic real-time queue; If the business type is periodic data uploading, the priority is set to medium level, and the transmission strategy is a first buffer queue; If the business type is batch data transmission, the priority is set to low level, and the transmission strategy is a second buffer queue; The preemptive real-time queue is defined as a resource that can preempt a low-priority queue and is always sent first; the elastic real-time queue is defined as a resource that does not preempt a low-priority queue and is sent first when the bandwidth is higher than a set threshold, and is downgraded to a cache queue when the bandwidth is lower than or equal to the set threshold; the first buffer queue is defined as a cache queue with the highest priority but lower than the downgraded elastic real-time queue, and the bandwidth allocation ratio is greater than a set threshold; the second buffer queue is defined as allowing data transmission according to a set delay value and packet loss rate, and the bandwidth allocation ratio is lower than a set threshold; B. According to the data information of the virtual resource pool module, a collection strategy and a transmission strategy are generated: The resource competition process is set as a non-cooperative game process, and Nash equilibrium is adopted to realize resource allocation, so as to ensure that the CPU usage rate and the memory occupation rate of each Internet of Things device do not exceed a set threshold, and the difference between the performance consumption values of each Internet of Things device is within a set range, avoiding that a single Internet of Things device monopolizes resources and ensuring fairness. 2.The power internet-of-things agent system for dynamic resource management of claim 1, wherein The device atom layer comprises a protocol analysis module and an edge preprocessing module connected in series; the protocol analysis module is used for realizing power system protocol analysis and data structure conversion of various types of Internet of Things terminals; and the edge preprocessing module is used for standardizing and warning noise of the received data. 3.The power internet-of-things agent system for dynamic resource management of claim 2, wherein The processing procedure of the protocol analysis module comprises the following steps: Data information of various types of Internet of Things terminals is collected through a protocol mode supported by a power system; the protocol mode supported by the power system comprises an RS-485 protocol, a Modbus-TCP protocol and a DL / T645-2007 protocol; The data information of various types of Internet of Things terminals is converted into a JSON structure body conforming to an IEC 61850 standard; the JSON structure body comprises a device ID, a time stamp, a measurement value and an extended field check code. 4.The power internet of things agent system for dynamic resource management of claim 2, wherein The processing procedure of the edge preprocessing module comprises the following steps: Data standardization: the acquired data information is processed by sliding window standardization as follows: ; wherein is the standardized data information; is the data information before standardization; is the data mean in the sliding window; is the data standard deviation in the sliding window; the sliding step is , is a set proportion coefficient, and w is the length of the sliding window; Noise processing: the value of the noise level L is defined as ; When L is less than a first noise threshold, it is marked as no noise; when L is greater than or equal to the first noise threshold and less than a second noise threshold, it is marked as mild noise, at this time, the data is retained and an abnormal identifier is added in the data; when L is greater than or equal to the second noise threshold, it is marked as severe noise, at this time, the data is discarded and an alarm is given; wherein the first noise threshold is less than the second noise threshold. 5.The power internet-of-things agent system for dynamic resource management of claim 1, wherein The network communication layer comprises a dual-channel routing module and a dual-queue buffer module; the dual-channel routing module is used for switching a sending communication channel according to a set main-vice channel switching mechanism; and the dual-queue buffer module is used for performing data communication according to queue content according to a set dual-queue mode. 6.The power internet-of-things agent system for dynamic resource management of claim 5, wherein The processing procedure of the dual-channel routing module comprises the following steps: The configured dual channel comprises a fiber link and a 5G wireless link; Normal communication process, using optical fiber link; when the UDP probe detects that more than consecutive probe packets are lost or the delay exceeds ms, start link switching and switch the communication link to a 5G wireless link. 7.The power internet-of-things agent system for dynamic resource management of claim 5, wherein The processing procedure of the dual-queue buffer module comprises the following steps: The configured dual queue comprises a real-time queue and a cache queue; For the real-time queue: a priority scheduling mode is adopted, and data transmission is performed according to a generated scheduling scheme; For the cache queue: when the bandwidth utilization is monitored to be less than a set value, trigger batch transmission, and set the packet lifetime to seconds.

Images