Power distribution cabinet cluster internet of things networking and data acquisition method based on wireless ad hoc network

Abstract

The application relates to a power distribution cabinet cluster Internet of Things networking and data acquisition method based on a wireless ad hoc network, and belongs to the technical field of the power Internet of Things.The method comprises the following steps: after each power distribution cabinet node is powered on, a master node is automatically elected through wireless link quality evaluation, and the remaining nodes automatically become slave nodes; each slave node elects a parent node and determines a network level based on wireless link quality, so as to form a cluster network topology; the master node and each slave node respectively pre-process original acquisition data of each node to generate standardized data packets; each slave node monitors the communication link state between the slave node and the master node; when the link is normal, the standardized data packets are uploaded to the master node; when the link is interrupted, each slave node automatically reads the cache data and uploads the cache data to the master node after the link is recovered; and the master node receives the standardized data packets uploaded by each slave node and forwards the standardized data packets to a remote terminal. The application has the effect of effectively scheduling power distribution cabinet cluster data acquisition.

Description

Technical Field

[0001] This application relates to the field of power Internet of Things (IoT) technology, and in particular to a method for IoT networking and data acquisition of distribution cabinet clusters based on wireless self-organizing networks. Background Technology

[0002] As the terminal equipment for power distribution and control, the real-time monitoring of the operating status of distribution cabinets is crucial for ensuring power supply reliability and equipment safety. With the development of Internet of Things (IoT) technology, more and more distribution cabinets are being equipped with wireless communication modules. These modules use wireless networking to aggregate the operating data of each distribution cabinet and upload it to a monitoring platform, enabling remote centralized monitoring and management of distribution cabinet clusters.

[0003] Currently, wireless data acquisition in power distribution cabinet clusters typically employs an architecture where each distribution cabinet has a wireless node, each node connects to a central aggregation node via a wireless protocol, and the aggregation node uploads data uniformly. However, this architecture has the following shortcomings in practical applications: First, the networking method relies on manual configuration and lacks adaptability. In existing solutions, the topology between wireless nodes is usually pre-set by maintenance personnel or uses a fixed networking structure. When the power distribution cabinet cluster needs to add or remove nodes or a node goes offline due to a fault, network parameters often need to be manually reconfigured. This makes it impossible to achieve automatic adjustment and self-healing of the topology, resulting in low efficiency in cluster deployment and maintenance and difficulty in adapting to the dynamic changes in the power distribution cabinet operating environment.

[0004] Secondly, data transmission lacks orderly scheduling, making it difficult to prioritize critical data. In power distribution cabinet cluster scenarios, numerous nodes need to upload collected data to the same aggregation node. Existing solutions often use random contention to occupy wireless channels. When multiple nodes transmit data simultaneously, channel conflicts are highly likely to occur, leading to data retransmissions and increased transmission delays. It is particularly noteworthy that critical data such as fault alarms and abnormal states are not prioritized compared to regular operational data during transmission. With equal competition for channel resources, congestion can cause critical data to be delayed or even lost, severely impacting the timely reporting and handling of faults.

[0005] In summary, existing wireless acquisition schemes for power distribution cabinet clusters have significant shortcomings in terms of adaptive adjustment of network topology and orderly scheduling of data transmission, which restricts the effective improvement of the overall perception capability of power distribution cabinet clusters. Summary of the Invention

[0006] To achieve effective scheduling of data acquisition for power distribution cabinet clusters, this application provides a method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless ad hoc networks.

[0007] Firstly, this application provides a method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless ad hoc networks, applicable to each power distribution cabinet node in a power distribution cabinet cluster, including: After each power distribution cabinet node is powered on, a master node is automatically elected through wireless link quality assessment, and the remaining nodes automatically become slave nodes. Each slave node elects a parent node and determines the network level based on wireless link quality, forming a cluster network topology with the master node as the root node and multiple redundant communication paths. The master node and each slave node preprocess their respective raw collected data locally to generate standardized data packets; Each slave node monitors the communication link status between itself and the master node. When the link is normal, the master node publishes a synchronization timing reference to each slave node. Each slave node uploads the standardized data packet to the master node within the allocated transmission time slot according to the preset data priority. When the link is interrupted, each slave node caches the standardized data packet locally. After the link is restored, it automatically reads the cached data and uploads it to the master node. The master node receives the standardized data packets uploaded by each slave node, and together with the standardized data packets pre-processed by the master node itself, they are aggregated, processed, and then forwarded to the remote terminal.

[0008] By adopting the above technical solution, after each distribution cabinet node is powered on, a master node and slave nodes are automatically elected through wireless link quality assessment. Slave nodes elect a parent node and determine the hierarchy based on link quality, eliminating the need for manual topology configuration. The network can automatically reorganize itself when nodes are added, deleted, or their locations change. Each node preprocesses the raw collected data locally, reducing the data cleaning burden on the master node and remote terminals. Slave nodes monitor the communication link status with the master node. When the link is normal, data is uploaded in an orderly manner according to time slots, avoiding channel conflicts and delays in critical data. When the link is interrupted, data is cached locally and retransmitted after recovery to avoid data loss. The master node aggregates the data uploaded by each slave node with its own data and forwards it to the remote terminal, enabling the remote terminal to obtain complete data.

[0009] Optionally, the method further includes: Each slave node continuously monitors the quality of its wireless link with its current parent node. When the wireless link quality is detected to be lower than a preset threshold, it automatically switches to a backup parent node to reconstruct the cluster network topology.

[0010] By adopting the above technical solution, the node continuously monitors the quality of the wireless link with the current parent node. When the quality is lower than the threshold, it automatically switches to the backup parent node and reconstructs the topology, avoiding communication interruption caused by the degradation of the parent node link, and enabling the network to automatically recover when the parent node fails or the signal weakens.

[0011] Optionally, the method further includes: The master node records statistical information related to the network communication behavior of each slave node during operation, and generates a multi-dimensional profile to characterize the network communication quality characteristics of each slave node based on the statistical information. The master node uses the multi-dimensional profile as the basis for each slave node to elect a parent node based on link quality and to upload data within the allocated transmission time slots.

[0012] By adopting the above technical solution, the master node records the network communication behavior statistics of each slave node and generates a multi-dimensional profile. This profile is used as the basis for each slave node to elect a parent node and allocate transmission time slots. This allows the selection of the parent node and the allocation of time slots to take into account the long-term communication behavior characteristics of the node, avoiding unreasonable decisions caused by relying solely on instantaneous link quality.

[0013] Optionally, the method further includes: The master node records the wireless link quality degradation patterns of each slave node before the historical topology reconstruction occurs, and constructs a historical degradation feature library; During the continuous monitoring of the wireless link quality between itself and its current parent node, each slave node records the trend information of the wireless link quality changing over time and attaches the trend information when uploading data. The master node compares the trend information with the historical degradation feature database and combines it with the multi-dimensional profiles of each slave node to comprehensively determine whether each slave node has a risk of failure. When a risk of failure is determined, a switching command is sent to the corresponding target slave node in advance, so that the switch to the backup parent node occurs before the wireless link fails.

[0014] By adopting the above technical solution, the master node records the link quality degradation patterns before historical topology reconstruction and constructs a historical degradation feature library. The slave nodes record trend information during the link quality monitoring process and attach it to the uploaded data. The master node compares the trend information with the historical degradation feature library and combines it with multi-dimensional profiles to comprehensively judge the risk of link failure. When a risk is determined, a switching instruction is sent to the target slave node in advance, so that it switches to the backup parent node before the link failure occurs. This avoids data transmission interruption caused by switching only after the link failure and reduces packet loss during the switching process.

[0015] Optionally, the method further includes: The master node dynamically adjusts the frequency at which each slave node uploads data within its allocated transmission time slot based on the multi-dimensional profile of each slave node; wherein, slave nodes with relatively low communication quality as represented by the multi-dimensional profile are assigned a relatively higher upload frequency.

[0016] By adopting the above technical solution, the master node dynamically adjusts the upload frequency of each slave node based on the multi-dimensional profile of each slave node. Slave nodes with lower communication quality are assigned a higher upload frequency, enabling them to complete data upload before the link deteriorates further, reducing the probability of data loss due to link breakage, and improving the integrity of data collection from all nodes in the network.

[0017] Optionally, the method further includes: When a topology reconfiguration occurs, the master node acquires network communication behavior data of the target slave node before and after the break occurs, and determines the type of break cause based on the network communication behavior data. The master node corrects the dimensions associated with the breakage cause in the multidimensional profile of the target slave node based on the determined breakage cause category, and uses the corrected multidimensional profile as the basis for the target slave node to subsequently elect a parent node based on wireless link quality and upload data in the allocated transmission time slot.

[0018] By adopting the above technical solution, when topology reconstruction occurs, the master node obtains network communication behavior data before and after the target slave node breaks down and determines the type of cause of the breakdown. Based on the determined type, the corresponding dimension in the multi-dimensional profile is corrected, and the corrected profile is used for the subsequent parent node election and time slot allocation of the node. This allows the experience of the breakdown event to be incorporated into subsequent decisions, forming a closed loop of continuous optimization.

[0019] Optionally, the method further includes: When the cause of the breakage belongs to a preset category, the master node obtains the data of other slave nodes around the target slave node in the same time period, and calculates the data of the target slave node during the breakage based on the historical data association between the target slave node and other slave nodes around the target slave node before the breakage. The master node then sends the data along with the data of the master node itself and other slave nodes to the remote terminal. Once the target slave node is able to upload data again, the master node compares the actual data uploaded by the target slave node with the estimated data, and corrects the historical data association based on the comparison results.

[0020] By adopting the above technical solution, when the cause of the fracture falls within a preset category, the master node acquires data from other slave nodes surrounding the target slave node within the same time period. Based on the historical data correlation between the target slave node and other surrounding slave nodes before the fracture, it infers the data during the fracture period and forwards the inferred data along with the data from the master node itself and other slave nodes to the remote terminal. This ensures that even if the parent node's hardware failure causes the cache retransmission mechanism to fail, the remote terminal can still obtain the inferred data during the node's fracture period. After the target slave node is able to upload data again, the master node compares its actual data with the inferred data and corrects the historical data correlation based on the comparison results, gradually improving the accuracy of subsequent inferences as fracture events accumulate.

[0021] Secondly, this application provides a power distribution cabinet cluster IoT networking and data acquisition system based on a wireless self-organizing network, including: The networking module is used to automatically elect a master node through wireless link quality assessment after each power distribution cabinet node is powered on. The remaining nodes automatically become slave nodes. Each slave node elects a parent node and determines the network level based on the wireless link quality, forming a cluster network topology with the master node as the root node and multiple redundant communication paths. A preprocessing module is set in the master node and each of the slave nodes, and is used to preprocess their respective raw collected data locally to generate standardized data packets; A communication scheduling module, located in each slave node, is used to monitor the communication link status between each slave node and the master node. When the link is normal, the master node publishes a synchronization timing reference to each slave node, and each slave node uploads the standardized data packets to the master node within its allocated transmission time slot according to a preset data priority. When the link is interrupted, each slave node caches the standardized data packets locally, and automatically reads the cached data and uploads it to the master node after the link is restored. The aggregation and forwarding module, located in the master node, is used to receive standardized data packets uploaded by each slave node, aggregate and process them together with the standardized data packets preprocessed by the master node itself, and then forward them to the remote terminal.

[0022] Thirdly, this application provides a power distribution cabinet cluster IoT networking and data acquisition device based on a wireless self-organizing network, including a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any of the methods in the first aspect.

[0023] Fourthly, this application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described in any of the first aspects.

[0024] In summary, this application includes at least one of the following beneficial technical effects: In this application, after each distribution cabinet node is powered on, a master node and slave nodes are automatically elected through wireless link quality assessment. Slave nodes elect a parent node and determine the hierarchy based on link quality, eliminating the need for manual topology configuration. The network can automatically reorganize itself when nodes are added, deleted, or their locations change. Each node preprocesses the raw collected data locally, reducing the data cleaning burden on the master node and remote terminals. Slave nodes monitor the communication link status with the master node. When the link is normal, data is uploaded in an orderly manner according to time slots, avoiding channel conflicts and delays in critical data. When the link is interrupted, data is cached locally and retransmitted after recovery to avoid data loss. The master node aggregates the data uploaded by each slave node with its own data and forwards it to the remote terminal, enabling the remote terminal to obtain complete data.

[0025] Furthermore, the node continuously monitors the quality of the wireless link with the current parent node. When the quality falls below a threshold, it automatically switches to a backup parent node and reconstructs the topology to avoid communication interruptions caused by parent node link degradation, enabling the network to automatically recover when the parent node fails or the signal weakens. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart illustrating the IoT networking and data acquisition method for power distribution cabinet clusters based on wireless self-organizing networks disclosed in this application.

[0028] Figure 2 This is a structural block diagram of the distribution cabinet cluster IoT networking and data acquisition system based on a wireless self-organizing network disclosed in the embodiments of this application.

[0029] Explanation of reference numerals in the attached diagram: 201, Networking module; 202, Preprocessing module; 203, Communication scheduling module; 204, Aggregation and forwarding module. Detailed Implementation

[0030] The following is in conjunction with the appendix Figure 1-2 This application will be described in further detail.

[0031] This application discloses a method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless ad hoc networks (hereinafter referred to as the network management method), the execution subject of which is a power distribution cabinet cluster IoT networking and data acquisition system based on wireless ad hoc networks (hereinafter referred to as the network management system). The following will be combined with... Figure 1This section details the execution process of the network management system for network management methods.

[0032] The network management method is applied to each distribution cabinet node in a distribution cabinet cluster, and includes the following steps: S101 After each distribution cabinet node is powered on, a master node is automatically elected through wireless link quality assessment, and the remaining nodes automatically become slave nodes. Each slave node elects a parent node and determines the network hierarchy based on wireless link quality, forming a cluster network topology with the master node as the root node and multiple redundant communication paths.

[0033] S102, the master node and each slave node preprocess their respective raw collected data locally to generate standardized data packets.

[0034] S103, each slave node monitors the communication link status between itself and the master node; when the link is normal, the master node publishes a synchronization timing reference to each slave node, and each slave node uploads standardized data packets to the master node within its allocated transmission time slot according to the preset data priority; when the link is interrupted, each slave node caches the standardized data packets locally, and automatically reads the cached data and uploads it to the master node after the link is restored. S104, the master node receives the standardized data packets uploaded by each slave node, and after processing them together with the standardized data packets pre-processed by the master node itself, forwards them to the remote terminal.

[0035] S105, each slave node continuously monitors the quality of the wireless link between itself and the current parent node. When the wireless link quality is detected to be lower than a preset threshold, it automatically switches to the backup parent node to reconstruct the cluster network topology.

[0036] In implementation, in step S101, the network management system monitors the power-on status of each distribution cabinet node in real time. Specifically, each distribution cabinet node has a built-in power management module and voltage detection circuit. The voltage detection circuit is connected to the power supply port of the microcontroller unit of each distribution cabinet node via a signal line. When the distribution cabinet is powered on, the power management module outputs a stable DC operating voltage (e.g., DC 3.3V or DC 5V). The voltage detection circuit continuously detects the voltage value of the power supply port. When the voltage value reaches the operating voltage threshold of the microcontroller unit (e.g., 3.0V) and remains stable for more than a preset stabilization time (e.g., 500 milliseconds), the microcontroller unit determines that the node has completed power-on. After each distribution cabinet node completes power-on, its built-in wireless communication module (e.g., a WiFi module supporting the IEEE 802.11 protocol or a ZigBee module supporting the IEEE 802.15.4 protocol) automatically starts and begins scanning for surrounding wireless signals.

[0037] After all distribution cabinet nodes have been powered on, the network management system initiates an automatic election process. Each distribution cabinet node's wireless communication module periodically sends probe request frames on its operating channel. These frames contain the node's unique identifier (e.g., MAC address or pre-configured device number). For any two distribution cabinet nodes capable of receiving each other's probe request frames, each independently calculates its own comprehensive wireless link quality score with the other node from its own perspective.

[0038] Specifically, after receiving a probe request frame from another node, any distribution cabinet node acquires multiple dimensions of wireless link quality indicators based on the received probe request frame, including but not limited to received signal strength indication, signal-to-noise ratio (SNR), packet reception rate, and transmission delay. The received signal strength indication is directly measured by the baseband chip of the wireless communication module; the SNR is calculated by the baseband chip based on the ratio of signal power to noise power; the packet reception rate is obtained by the ratio of the number of successfully received packets per unit time to the total number of packets that should be received; and the transmission delay is obtained by calculating the difference between the sending timestamp and the receiving timestamp of the probe request frame.

[0039] Each node performs a weighted summation of the aforementioned multiple dimensions of indicators according to preset weighting coefficients (e.g., Received Signal Strength Index (RSI) weighted at 0.4, Signal-to-Noise Ratio (SNR) weighted at 0.3, Data Packet Receiver Ratio (DRFR) weighted at 0.2, and Transmission Delay weighted at 0.1) to obtain a comprehensive wireless link quality score between itself and the peer node that sent the probe request frame. Each distribution cabinet node maintains its own score table, which records the comprehensive wireless link quality score between itself and every other distribution cabinet node in the network.

[0040] After the detection phase, each distribution cabinet node calculates the average of its overall wireless link quality score with all other nodes in the network, based on its own maintained scoring table. This average score serves as the node's representative value for the overall network link quality, which is then broadcast to all other nodes via its wireless communication module. Upon receiving the representative values ​​from other nodes, each node compares them with its own overall network link quality representative value. The node with the highest representative value is elected as the master node, and the remaining nodes automatically become slave nodes. When multiple nodes have the same highest representative value, a unique node identifier (e.g., MAC address) is used as an auxiliary comparison criterion; the node with the smaller identifier is elected as the master node. After the master node election, the master node broadcasts a master node announcement frame to all slave nodes within its coverage area. This frame contains the master node's unique identifier and hierarchy information (root node hierarchy is 0) to notify all slave nodes that the network's master node has been determined.

[0041] After the master node is determined, the network management system enters the parent node election and network hierarchy determination phase. The master node, acting as the root node, periodically broadcasts beacon frames containing its unique identifier and current network hierarchy information. Upon receiving the beacon frames broadcast by the master node, each slave node measures the overall wireless link quality score with the master node (this score is calculated in the same way as in the previous steps, based on a weighted sum of received signal strength indication, signal-to-noise ratio, packet reception rate, and transmission delay), and compares this score with a preset direct access threshold. If the score reaches or exceeds the direct access threshold, the slave node directly elects the master node as its parent node, and its network hierarchy is determined to be level 1. For a slave node whose overall wireless link quality score with the master node does not reach the direct access threshold, the slave node continues to listen to the secondary beacon frames broadcast by other slave nodes with determined network layers. The secondary beacon frames contain the unique identifier of the slave node that sent the frame and its current network layer information. Based on the received signal strength indication of each secondary beacon frame, the slave node calculates the overall wireless link quality score with each neighboring slave node (the calculation method is the same as described above), and selects the neighboring slave node with the highest overall wireless link quality score as its parent node. Its network layer is determined as the network layer of the parent node plus 1.

[0042] If a slave node fails to elect a parent node within a preset time window (e.g., 30 seconds), it proactively sends a network entry request frame to the master node. This frame includes the number of detectable neighboring nodes recorded by the slave node during the probing phase, along with the unique identifier of each neighboring node. Upon receiving this request frame, the master node, based on the detectable neighboring node information, selects the node with the lowest network level and whose number of child nodes does not exceed a preset limit as the recommended parent node. The master node then sends the recommendation result to the slave node in the form of a parent node assignment frame. The slave node determines its parent node and network level according to the assignment result.

[0043] After determining its parent node and network layer, each slave node sends an acknowledgment frame to its parent node. Upon receiving the acknowledgment frame, the parent node records the slave node in its child node list. During the parent node election process, in addition to electing a primary parent node, each slave node also records and maintains a candidate parent node list. This candidate parent node list is generated as follows: the slave node records its adjacent nodes ranked second to fourth in the comprehensive radio link quality score during the parent node election process as candidate parent nodes, and sorts them from highest to lowest according to the comprehensive radio link quality score. This list is used for the parent node switching scenario in step S105.

[0044] The network management system generates a cluster network topology with the master node as the root node and multiple redundant communication paths based on the parent node relationships and network hierarchy information of each node. Specifically, the master node constructs a complete network topology diagram by collecting parent node confirmation information reported by each slave node. This network topology diagram includes a unique identifier for each slave node, a unique identifier for its parent node, its own network hierarchy, and a list of child nodes. After completing the construction of the network topology diagram, the master node sends the complete cluster network topology information (including the network hierarchy of each slave node, parent node identifier, and child node list) to each slave node through a downlink broadcast channel. Each slave node stores the topology information in its local non-volatile memory upon receiving it, thus concluding the network construction process in step S101.

[0045] For S102, after the master node and all slave nodes complete the network formation, the network management system controls each node to start the data acquisition and preprocessing process. Each distribution cabinet node (including the master node and slave nodes) is equipped with a data acquisition sensor group. This sensor group is connected to each monitoring point in the distribution cabinet via signal lines to collect raw data during the operation of the distribution cabinet. This raw data includes, but is not limited to, three-phase current values, three-phase voltage values, internal temperature of the distribution cabinet, ambient humidity, circuit breaker opening and closing status, and switch input status. Among them, the three-phase current value is acquired through current transformer and analog-to-digital converter, the three-phase voltage value is acquired through voltage transformer and analog-to-digital converter, the temperature value is acquired through thermistor or digital temperature sensor, the humidity value is acquired through capacitive humidity sensor, the circuit breaker opening and closing status and the switch input status are obtained by judging the high and low level signals of the digital input port (e.g., optocoupler isolation input port), the sampling frequency of the analog-to-digital converter is, for example, 1000 samples per second, and the output is a 12-bit or 16-bit digital quantized value, and the level sampling frequency of the digital input port is, for example, once every 10 milliseconds.

[0046] The master node and each slave node preprocess the raw collected data locally to generate standardized data packets. This preprocessing includes three sub-steps: filtering and denoising, time base alignment, and outlier removal.

[0047] For filtering and noise reduction, taking analog signal acquisition channels (such as current, voltage, and temperature acquisition channels) as an example, the microcontroller unit of each acquisition channel allocates a circular buffer in its local memory. This circular buffer is used to store N consecutively acquired sample values ​​(N is an odd number between 5 and 9, preferably 7). Whenever a new sample value enters the circular buffer, the microcontroller unit sorts all N sample values ​​in the circular buffer according to their numerical values, removes the maximum and minimum values, calculates the arithmetic mean of the remaining N-2 sample values, and uses this arithmetic mean as the filtered output value at that moment. For digital signal acquisition channels (such as circuit breaker opening / closing status and fault alarm flags), the microcontroller unit uses a three-step continuous confirmation method: after each reading of the status value of the digital input port, the current reading value is compared with the status values ​​of the previous two consecutive readings. Only when the three readings are completely consistent is the status value recognized as a valid status value and output; if there are inconsistencies among the three readings, the valid status value of the previous output remains unchanged, and the abnormal status value of the current reading is recorded in the error log.

[0048] For time reference alignment, the master node maintains a global time reference internally (e.g., calibrated via a built-in real-time clock chip or periodically obtained from a remote terminal using a network time protocol). Each slave node, during each data interaction with the master node (e.g., upon receiving a beacon frame broadcast by the master node), extracts the global timestamp embedded by the master node from the beacon frame. Each slave node's microcontroller unit compares this global timestamp with the current time of its local real-time clock, calculates the time deviation, and stores this deviation in a local variable. When a slave node generates a timestamp for each piece of acquired data during subsequent preprocessing, the microcontroller unit adds the aforementioned time deviation to the local real-time clock's timing value to obtain a unified timestamp aligned with the global time reference.

[0049] For outlier removal, the microcontroller has preset reasonable value ranges for each acquired physical quantity in its local memory (e.g., temperature range of -20℃ to +80℃, voltage range of ±20% of rated voltage, and current range of 0 to 120% of rated current). After each sampled value is filtered and output, the microcontroller determines whether the filtered output value falls within the corresponding preset reasonable value range. If it does, the data is retained as valid data; otherwise, the data is discarded and the event is recorded in the anomaly log.

[0050] After completing the filtering, noise reduction, time base alignment, and outlier removal processes, the master node and each slave node package each valid data along with its corresponding unified timestamp and unique node identifier according to a preset data encapsulation format to generate a standardized data packet. The format of this standardized data packet is, for example, as follows: a data packet header (containing a data packet type identifier, data packet length, source node identifier, and destination node identifier), a data body (containing one or more data fields, each containing a data type encoding, data length, timestamp, and a specific value), and a data packet trailer (containing a checksum, such as a CRC16 cyclic redundancy check code, used by the receiving end to verify the integrity of the data packet). After the standardized data packet is generated, each node temporarily stores it in its local transmission buffer, awaiting scheduling and uploading in step S103.

[0051] For S103, the network management system controls each slave node to execute the data scheduling and uploading process. Each slave node's microcontroller unit continuously monitors the communication link status between the slave node and the master node through its wireless communication module. Specifically, each slave node sends a heartbeat request frame to the master node at a preset heartbeat cycle (e.g., once every 5 seconds). This heartbeat request frame contains the slave node's unique identifier and current timestamp. Upon receiving the heartbeat request frame, the master node immediately replies with a heartbeat response frame, which contains the master node's unique identifier and the receiving timestamp of the received heartbeat request frame. The slave node's microcontroller unit records the sending timestamp when sending the heartbeat request frame and the receiving timestamp when receiving the heartbeat response frame, using the difference between the receiving and sending timestamps as the round-trip time. Simultaneously, the slave node measures the received signal strength indication value of the heartbeat response frame. The microcontroller unit of the slave node evaluates the current communication link status by combining the round-trip time and the received signal strength indicator: when the round-trip time is less than a preset time threshold (e.g., 100 milliseconds) and the received signal strength indicator is greater than a preset strength threshold (e.g., -70 dBm), the communication link is considered normal; when the round-trip time is greater than or equal to the preset time threshold, or the received signal strength indicator is less than or equal to the preset strength threshold, and this status continues for more than a preset number of consecutive timeouts (e.g., 3 consecutive heartbeat timeouts), the communication link is considered interrupted. The microcontroller unit of the slave node stores the result of each link status determination in its local status register, which is used to trigger different upload branches.

[0052] When the communication link status is determined to be normal, the network management system executes the normal upload branch. The master node broadcasts a synchronization timing reference frame to all slave nodes according to a preset superframe period (e.g., one superframe every 1 second). This synchronization timing reference frame contains a global timestamp of the superframe start time, the total length of the superframe, and a time slot allocation table for each slave node within the superframe. This time slot allocation table is pre-calculated and generated by the master node based on the number of slave nodes currently connected to the network and the data priority of each slave node. Specifically, the master node assigns each slave node a unique starting transmission offset (e.g., a microsecond-level offset relative to the superframe start time) and a fixed transmission window duration (e.g., 10 milliseconds) in the time slot allocation table. After receiving the synchronization timing reference frame, each slave node extracts the superframe start time and its corresponding starting transmission offset from the frame, and sets a timer locally. This timer starts counting from the superframe start time, and triggers a transmission interruption when the count reaches the node's starting transmission offset.

[0053] During the interrupt service routine, the microcontroller unit of the slave node retrieves the standardized data packets to be uploaded from its local transmit buffer and assigns a priority label to each packet. Specifically, the microcontroller unit identifies the data priority based on the content type of the data body within the standardized data packet. If the data body contains fault alarm flags or circuit breaker opening / closing status change events, the data packet is marked as urgent (highest priority). If the data body contains core operating parameters such as three-phase current values ​​and three-phase voltage values, the data packet is marked as important (medium priority). If the data body only contains environmental or statistical data such as temperature and humidity, the data packet is marked as normal (lowest priority). When multiple standardized data packets of different priorities are waiting to be transmitted on the same slave node, the microcontroller unit transmits them sequentially in the order of urgent priority over important priority, and important priority over normal priority. Data packets within the same priority are transmitted in the order of their generation time.

[0054] During transmission, the microcontroller unit of the slave node performs an idle channel assessment before starting transmission. Specifically, the physical layer of the slave node's wireless communication module has a built-in power detection circuit and a frame synchronization detection circuit. The power detection circuit generates a received signal strength indicator value by integrating the power of the radio frequency signal on the current working channel. The frame synchronization detection circuit performs preamble synchronization detection on the received radio frequency signal to determine whether there are IEEE 802.11 or IEEE 802.15.4 standard data frames being transmitted in the current channel. The microcontroller unit continuously reads the received signal strength indicator value output by the power detection circuit within a preset assessment time window (e.g., 128 microseconds) and simultaneously reads the frame synchronization flag bit of the frame synchronization detection circuit. If, within the assessment time window, the received signal strength indicator value is continuously lower than the preset idle channel assessment threshold (e.g., -80dBm) and the frame synchronization flag bit is "no frame synchronization detected", the microcontroller unit determines that the current channel is idle. If the received signal strength indicator value reaches or exceeds the idle channel assessment threshold at any point within the assessment time window, or the frame synchronization flag bit is "frame synchronization detected", the microcontroller unit determines that the current channel is busy. When the channel is determined to be idle, the microcontroller immediately initiates the physical layer transmission procedure for the data frame. When the channel is determined to be busy, the microcontroller performs a random backoff, for example, taking a random value within the range of 0 to 5 milliseconds as the backoff time. After the backoff ends, the above idle channel assessment procedure is re-executed. If the channel is determined to be busy in three consecutive idle channel assessments, the microcontroller discards the current data frame and records the frame loss event in the local anomaly log.

[0055] After transmission is complete, the slave node waits for an acknowledgment frame from the master node. If no acknowledgment frame is received within the preset acknowledgment waiting time (e.g., 20 milliseconds), the transmission is considered a failure, and the data frame is returned to the transmission buffer to await retransmission in the next transmission slot. After completing the above transmission process, the slave node transmits each data field in the standardized data packet to the master node via the wireless communication module in the form of electromagnetic waves.

[0056] When the communication link status is determined to be interrupted, the network management system executes an abnormal upload branch. After detecting the communication link interruption, the microcontroller unit of the slave node immediately switches the working mode flag in its local status register from "real-time upload mode" to "local cache mode".

[0057] In local caching mode, each standardized data packet generated by the microcontroller unit of the slave node in step S102 is no longer attempted to be uploaded to the master node. Instead, it is written sequentially to a cache queue in local non-volatile storage media (such as a Flash memory chip with an SPI interface or an SD card) according to the order of its generation time. This cache queue adopts a circular queue data structure, with a preset queue capacity to store a preset number of data records (e.g., 10,000 standardized data packets). Each standardized data packet is accompanied by a write timestamp when written. When the cache queue is full and the communication link has not yet been restored, newly generated standardized data packets overwrite the oldest unuploaded data packet in the queue according to the first-in-first-out principle. In local caching mode, the microcontroller unit of the slave node continues to send heartbeat request frames to the master node according to a preset heartbeat cycle (e.g., once every 5 seconds) and continuously monitors the reception of heartbeat response frames to detect the restoration of the communication link in real time. The microcontroller unit of the slave node also records the start time of the communication link interruption, the duration of the interruption, and the number of standardized data packets cached during the interruption, and stores this information in a local status log.

[0058] When the microcontroller unit of the slave node determines that the communication link has returned to normal based on the received heartbeat response frame, it immediately switches the working mode flag in its local status register from "local cache mode" back to "real-time upload mode," and then triggers the retransmission process. In the retransmission process, the microcontroller unit first reads the cached standardized data packets from the local non-volatile storage medium's cache queue in a first-in, first-out order, reading a preset batch number of data packets each time (e.g., 50 packets per batch). These data packets are then uploaded to the master node in batches via the wireless communication module, with the start and end sequence numbers of each batch appended to the end of the data packets. After successfully receiving and verifying each standardized data packet, the master node returns an acknowledgment frame to the slave node for each packet. Upon receiving an acknowledgment frame for a data packet, the slave node marks that data packet as "acknowledged" from the cache queue and will not read that data packet again in subsequent retransmission processes. If the slave node fails to receive an acknowledgment frame within a preset acknowledgment waiting time after sending a data packet, the data packet is retained in the cache queue and retransmitted after the other data packets in the current batch have been sent. The retransmission process and the real-time upload process are executed in parallel. That is, while the slave node is retransmitting historical cached data, the newly collected and preprocessed standardized data packets are still uploaded in real time according to the time slot allocation rules of the normal upload branch. Both share the same transmission queue of the wireless communication module, but the transmission priority of the retransmitted data packets is lower than that of the real-time upload data packets. The retransmission process continues to execute until all data packets marked "unacknowledged" in the cache queue have received acknowledgment frames and the cache queue is empty. At this point, the retransmission process ends, and the slave node fully resumes the normal real-time upload mode.

[0059] For S104, the network management system controls the master node to execute the data aggregation and forwarding process. The master node's wireless communication module continuously receives standardized data packets (including real-time uploaded data packets and retransmitted historical buffered data packets) uploaded by each slave node through step S103. After receiving each standardized data packet, the master node first performs an integrity check on the data packet, extracts the CRC16 checksum at the end of the data packet, and compares it with the CRC16 checksum recalculated by the master node based on the data packet header and data body content. If the two match, the data packet is considered complete; if they do not match, the data packet is discarded and a negative acknowledgment frame is returned to the slave node that sent the data packet, triggering the slave node to retransmit. After the integrity check passes, the master node's microcontroller unit temporarily stores the received standardized data packets in the master node's local receive buffer and sorts and organizes the data from different slave nodes according to the source node identifier and timestamp field in the data packet header, in chronological order. At the same time, the master node also completes the preprocessing of its own raw collected data locally through step S102 to generate standardized data packets belonging to the master node itself.

[0060] After receiving, verifying the integrity of, and sorting all data uploaded by the slave nodes, the master node aggregates its own standardized data packets with those of each slave node. This aggregation process includes: data format standardization: converting data packets from different slave nodes into a preset standard data exchange format (e.g., adding aggregation timestamps, total data count, and a list of data source nodes to the outer layer of the data packets); data redundancy removal: identifying and removing duplicate or redundant data fields from different slave nodes; and data compression and packaging: compressing multiple standardized data packets using a preset compression algorithm (e.g., LZ77 compression or Huffman coding) to reduce data size.

[0061] After aggregation processing is complete, the master node forwards the aggregated data packets to the remote terminal through its uplink communication interface. This uplink communication interface can be, for example, a wired Ethernet interface (connected to an industrial Ethernet switch via an RJ45 Ethernet port and twisted-pair cable, thus accessing the remote monitoring network) or a cellular mobile communication interface (e.g., accessing the internet via a 4G or 5G cellular module, establishing a TCP / IP connection with a remote terminal server deployed in the cloud). During forwarding, the master node and the remote terminal interact using a preset uplink communication protocol (e.g., MQTT or Modbus TCP). The master node, acting as a client, actively initiates a connection to the remote terminal and pushes the aggregated data packets. After successful reception, the remote terminal returns a reception acknowledgment frame. Upon receiving the acknowledgment frame, the master node clears the forwarded data from its local receive buffer to free up storage space.

[0062] For S105, the network management system continuously executes the topology maintenance and reconstruction process during the normal operation of each slave node. This process is executed in parallel with the data scheduling and uploading process in step S103 and the data aggregation and forwarding process in step S104, and is independent of the execution status of the above processes. The microcontroller unit of each slave node continuously monitors the quality of the wireless link between itself and the current parent node through its wireless communication module. The monitoring mechanism is as follows: each slave node measures the received signal strength indication value of the most recent data frame (including beacon frame, heartbeat response frame, or data acknowledgment frame) received from the current parent node in each preset monitoring period (e.g., once every 100 milliseconds), and at the same time calculates the data frame transmission packet loss rate (i.e., the percentage of data frames that the parent node failed to receive out of the total number of data frames sent) between itself and the parent node in the past preset time window (e.g., the most recent 10 seconds), and calculates the moving average of the round-trip time of the most recent (e.g., the most recent 10) heartbeat request frames. The microcontroller unit of the slave node performs a weighted sum of the monitoring values ​​of the above three dimensions according to preset weighting coefficients (e.g., the weight of the received signal strength indicator is 0.5, the weight of the packet loss rate is 0.3, and the weight of the round-trip time is 0.2) to obtain the comprehensive wireless link quality score at the current moment. The microcontroller unit of the slave node compares the comprehensive wireless link quality score with a preset handover threshold, which is a value lower than the direct access threshold in step S101, for example, the direct access threshold is 75 points (out of 100 points), and the handover threshold is 45 points.

[0063] When a slave node detects that the overall wireless link quality score of its current parent node is lower than a preset handover threshold, the slave node's microcontroller unit initiates the parent node handover process. First, the slave node queries its locally stored list of candidate parent nodes. This list consists of neighboring nodes ranked second to fourth in overall wireless link quality score, which were recorded and maintained during the parent node election process in step S101. The slave node's microcontroller unit then measures the instantaneous overall wireless link quality score with each candidate parent node in descending order of their scores, selecting the candidate parent node with the highest score as the backup parent node. Before the handover, the slave node sends a handover warning frame to its current parent node, notifying it that the connection is about to be lost. Upon receiving the handover warning frame, the current parent node removes the slave node from its list of child nodes. Subsequently, the slave node sends a network access request frame to the selected backup parent node. Upon receiving the network access request frame, the backup parent node measures the overall radio link quality score between itself and the slave node. If the score reaches or exceeds a preset admission threshold (e.g., 50 points), it returns a network access confirmation frame to the slave node and adds the slave node to its own child node list. Simultaneously, the slave node updates its parent node to the backup parent node and updates its own network layer to the backup parent node's network layer plus 1. If the overall radio link quality score carried in the network access confirmation frame returned by the backup parent node is lower than the admission threshold, the slave node continues to try the next candidate parent node in the candidate parent node list until a successful handover or all attempts in the candidate parent node list fail. If all candidate parent nodes in the candidate parent node list fail to meet the admission threshold requirements, the slave node sends an emergency network access request to the master node, and the master node reassigns a parent node to the slave node according to the parent node assignment procedure in step S101.

[0064] After a slave node completes a parent node switchover, it sends a topology update notification frame to its new parent node. Upon receiving this notification frame, the new parent node confirms the existence of the slave node in its local child node list and reports the updated child node list to the master node level by level via the uplink. After receiving the topology update information reported by nodes at each level, the master node updates its maintained cluster network topology diagram and incrementally broadcasts the updated topology information to all slave nodes. Each slave node, upon receiving the incremental topology information, synchronously updates its local topology structure storage. When the network topology changes, the master node recalculates and generates a new time slot allocation table based on the number of currently connected slave nodes and the changed network hierarchy. This new time slot allocation table is then broadcast to all slave nodes in the synchronization timing reference frame of the next superframe period. Each slave node adjusts its transmission time slots according to the new time slot allocation table to adapt to the reconstructed network topology.

[0065] Thus, the network management system, through the cyclical execution of steps S101 to S105, has realized a complete methodology for the automatic networking, data acquisition and preprocessing, orderly scheduling and uploading, aggregation and forwarding, and autonomous maintenance and reconstruction of the topology of the power distribution cabinet cluster.

[0066] Optionally, the network management method may also include the following steps: S301, the master node records statistical information related to the network communication behavior of each slave node during operation, and generates a multi-dimensional profile to characterize the network communication quality characteristics of each slave node based on the statistical information; S302, the master node uses the multi-dimensional profile as the basis for each slave node to elect a parent node based on link quality and to upload data within the allocated transmission time slots.

[0067] S401, the master node records the wireless link quality degradation patterns of each slave node before the historical topology reconstruction occurs, and constructs a historical degradation feature library; S402, each slave node, while continuously monitoring the wireless link quality between itself and its current parent node, records the trend information of the wireless link quality changing over time and attaches the trend information when uploading data. S403, the master node compares the trend information with the historical degradation feature database and combines it with the multi-dimensional profiles of each slave node to comprehensively determine whether each slave node has a risk of failure. When a risk of failure is determined, a switching command is sent to the corresponding target slave node in advance, so that the switch to the backup parent node occurs before the wireless link fails.

[0068] S501, the master node dynamically adjusts the frequency of data upload by each slave node within the allocated transmission time slot based on the multi-dimensional profile of each slave node; among them, slave nodes with relatively low communication quality represented by the multi-dimensional profile are assigned a relatively higher upload frequency.

[0069] In implementation, within S301, the network management system initiates a network communication behavior statistics and profile generation process on the master node side. This process continues to run after the cluster network topology is completed and all slave nodes are successfully connected. The master node receives and records statistical information related to network communication behavior generated by each slave node during operation through its wireless communication module and wired network interface. Specifically, the master node allocates an independent statistical information storage area for each connected slave node. This storage area is located in the master node's local memory (e.g., DDR memory or Flash memory chip) and is indexed by the slave node's unique identifier.

[0070] Within each preset statistical period (e.g., every 10 minutes), the master node extracts and records the following statistical information from each communication interaction with the slave node: the average and variance of the comprehensive wireless link quality score between the slave node and the parent node within the statistical period (this score is obtained by weighted summation of received signal strength indication, signal-to-noise ratio, data packet reception rate, and transmission delay, and the calculation method is the same as in step S101), the continuous online duration and historical disconnection count and average disconnection duration of the slave node within the statistical period (the disconnection judgment criterion is that the number of consecutive center hop timeouts in step S103 exceeds a preset threshold), the response delay of the slave node to the synchronization timing reference frame published by the master node (i.e., the time difference between the slave node receiving the synchronization timing reference frame and starting to send standardized data packets), the average daily data upload volume of the slave node within the statistical period (in bytes), and the number of neighboring nodes that the slave node can detect (obtained by counting the number of nodes in the neighbor node list reported by the slave node). The above statistical information is compiled and persistently stored in the master node's local non-volatile memory at the end of each statistical period by the microcontroller unit of the master node.

[0071] Based on the aforementioned statistical information, the master node generates a multi-dimensional profile for each slave node, characterizing its network communication quality. This multi-dimensional profile consists of feature coefficients across multiple dimensions, each corresponding to a network communication quality characteristic. Specifically, the master node's microcontroller calculates a stability coefficient based on the variance of the slave node's comprehensive wireless link quality score: a smaller variance results in a higher stability coefficient. For example, the inverse of the variance is normalized to a range of 0 to 1 to obtain the stability coefficient, which characterizes the fluctuation of the slave node's link quality. A reliability coefficient is calculated based on the ratio of the slave node's continuous online duration to its historical disconnection count: a larger ratio results in a higher reliability coefficient. For example, the ratio of continuous online duration to total runtime is multiplied by (1 minus the ratio of disconnection counts to total runtime), normalized to a range of 0 to 1 to obtain the reliability coefficient, which characterizes the slave node's online stability over time. A response coefficient is calculated based on the slave node's average response delay to the synchronization timing reference frame: a shorter response delay results in a higher response coefficient. For example, the average response delay is calculated by multiplying the continuous online duration to the total runtime by (1 minus the ratio of disconnection counts to total runtime), normalizing to a range of 0 to 1 to obtain the reliability coefficient, which characterizes the slave node's online stability over time. The difference between the maximum allowable response latency (e.g., 50 milliseconds) and the actual average response latency of the slave node is normalized to the range of 0 to 1 to obtain the response coefficient, which characterizes the slave node's responsiveness to scheduling commands. The load coefficient is calculated based on the slave node's average daily uploaded data volume: the larger the average daily uploaded data volume, the higher the load coefficient. Specifically, the ratio of the slave node's average daily uploaded data volume to a preset load capacity limit is normalized to the range of 0 to 1 to obtain the load coefficient, which characterizes the slave node's data throughput capacity. The connectivity coefficient is calculated based on the number of detectable neighboring nodes of the slave node: the more neighboring nodes, the higher the connectivity coefficient. Specifically, the ratio of the slave node's number of neighboring nodes to a preset maximum neighbor count limit is normalized to the range of 0 to 1 to obtain the connectivity coefficient, which characterizes the slave node's connectivity coverage in the network.

[0072] The five coefficients mentioned above together constitute the multidimensional profile of the slave node. The master node associates the multidimensional profile of each slave node with its unique identifier and stores it in the local profile database. As the running time progresses, the master node recalculates the multidimensional profile of each slave node at the end of each statistical period and updates the corresponding record in the profile database using an exponentially weighted moving average. This ensures that the multidimensional profile can smoothly reflect the long-term changing trend of the network communication quality characteristics of each slave node, rather than being dominated by short-term fluctuations.

[0073] For S302, the network management system uses the multi-dimensional profile generated in S301 as the basis for decision-making when each slave node performs subsequent network actions. Specifically, during the parent node election process in S101, when a slave node needs to select the optimal parent node from multiple candidate parent nodes, the slave node queries the master node for the stability coefficient and reliability coefficient in the multi-dimensional profile of each candidate parent node. It prioritizes the selection of candidate parent nodes whose stability coefficient and reliability coefficient are both higher than a preset threshold (e.g., stability coefficient greater than 0.7 and reliability coefficient greater than 0.7) as the final parent node, rather than selecting solely based on the instantaneous wireless link quality score, thereby avoiding the selection of a historically poor-performing parent node due to instantaneous signal fluctuations. During the transmission time slot allocation process in S103, when generating the time slot allocation table, the master node queries the response coefficient and load coefficient in the multi-dimensional profile of each slave node. It allocates a higher starting transmission offset to slave nodes with lower response coefficients or higher load coefficients to reduce the accumulated delay caused by waiting for the transmission window.

[0074] For S401, the network management system initiates the process of building a historical degradation feature database on the master node side. This process is coupled with the topology reconfiguration event in S105. Specifically, whenever a slave node in S105 triggers a parent node switch due to its wireless link quality falling below a preset threshold, the master node records the wireless link quality degradation pattern of that slave node before the current topology reconfiguration. Specifically, after receiving a switchover warning frame from a slave node or a topology update notification reported by the slave node through its new parent node, the master node determines that the slave node has just completed a topology reconfiguration. The master node then retrieves all historical records of the slave node's overall wireless link quality score within a preset time window (e.g., 60 seconds before the break) from its local historical data storage area. This historical record includes the overall wireless link quality score value reported by the slave node to the master node with its current parent node in each monitoring cycle (e.g., every 100 milliseconds), along with the corresponding reporting timestamp. The master node's microcontroller unit arranges these historical records in chronological order, forming a complete sequence of wireless link quality degradation trajectories.

[0075] The master node extracts degradation pattern features from the degradation trajectory sequence, including the degradation start time, degradation rate, and critical break value. The degradation start time is determined as follows: the master node starts from the end of the degradation trajectory sequence (i.e., the break occurs) and backtracks point by point, calculating the difference between every two adjacent sampling points in the sequence (the score value of the later time minus the score value of the previous time). When a predetermined number of consecutive negative differences (e.g., 5 consecutive sampling points) occur (i.e., the score value continues to decrease), the time corresponding to the first sampling point where the consecutive negative differences begin is marked as the degradation start time, thereby quantifying and defining the starting boundary of the "continuous downward trend".

[0076] The degradation rate is calculated as follows: the total decrease in the overall wireless link quality score (the score at the start of degradation minus the critical value of the break) during the time period from the start of degradation to the break occurs is divided by the duration of that time period to obtain the average decrease in the score per unit time.

[0077] The threshold for signal breakage is determined as follows: when a slave node triggers a parent node switchover, the last reported comprehensive wireless link quality score between the slave node and the current parent node is used; this score is the score at the moment the breakage occurs. The master node also incorporates the short-term deviation of the comprehensive wireless link quality score from the start of degradation to the moment of the breakage into the degradation pattern characteristics. Specifically, the master node calculates the difference between the actual score at each sampling point in the degradation trajectory sequence and the predicted score, which decreases linearly according to the degradation rate. The maximum absolute value of this difference is taken as the fluctuation amplitude feature, used to characterize the degree of oscillation of signal quality around the overall downward trend during the degradation process. The larger the value of the fluctuation amplitude feature, the more severe the short-term oscillations during the degradation process; the smaller the value, the smoother the degradation process.

[0078] The master node packages the aforementioned degradation start time, degradation rate, fracture threshold, and fluctuation amplitude characteristics, along with the slave node's unique identifier and the parent node identifier at the time of fracture, into a historical degradation feature record. This record is then stored in the historical degradation feature library in the master node's local non-volatile memory. As the cluster's runtime increases and topology reconfiguration events accumulate, the historical degradation feature records stored in this library become increasingly rich, covering various degradation patterns. For example, a rapid degradation rate with small fluctuation amplitude corresponds to sudden degradation, a slow degradation rate with small fluctuation amplitude corresponds to slow and smooth degradation, and a moderate degradation rate with large fluctuation amplitude corresponds to oscillating degradation. This provides diverse historical reference samples for similarity comparison and trend prediction in step S403.

[0079] For S402, the network management system initiates the link quality trend information recording and reporting process on each slave node side. During normal operation, each slave node's control unit continuously measures the overall wireless link quality score between itself and its current parent node in each monitoring cycle (e.g., every 100 milliseconds) according to the monitoring mechanism defined in S105, and maintains a trend information recording queue in local memory. This queue is used to store the historical values ​​of the overall wireless link quality score for the most recent preset number of monitoring cycles (e.g., the most recent 100 monitoring cycles, corresponding to a 10-second time window).

[0080] After each new measurement is completed, the microcontroller unit of the slave node appends the new score value to the tail of the trend information recording queue and removes the oldest historical value from the head of the queue, keeping the queue length constant. In the normal upload branch of step S103, when the slave node uploads a standardized data packet within its allocated transmission time slot, the microcontroller unit of the slave node adds a trend information appendix field to the data body of the data packet. The content of this appendix field is all the historical values ​​of the comprehensive wireless link quality score stored in the current trend information recording queue, arranged in chronological order.

[0081] The microcontroller sends standardized data packets carrying the aforementioned trend information appendix to the master node within the allocated time slot, enabling the master node to obtain the latest link quality trend information from each slave node in each transmission cycle. It should be noted that the addition of this trend information appendix does not affect the normal transmission of the original data body content in the standardized data packet. That is, the valid data fields after filtering, time base alignment, and outlier removal remain unchanged; the trend information is simply inserted as an appendix at the end of the data body. When parsing the standardized data packet, the receiving end (master node) parses the original data fields and the trend information appendix separately.

[0082] For S403, the network management system performs risk identification and prediction decisions for network outages at the master node. After receiving standardized data packets from each slave node in each round, the master node's microcontroller extracts the historical sequence of the overall wireless link quality score from the trend information appendix field of the data packet. The master node first compares this sequence with the degradation trajectory sequences of each historical degradation feature record in the historical degradation feature database constructed in S401.

[0083] Specifically, the master node uses a dynamic time warping algorithm to calculate the similarity distance between the current trend information sequence and the degradation trajectory sequence in each historical degradation feature record. This algorithm non-linearly warps the time axis of the two sequences to achieve optimal alignment in the time dimension between two sequences that may have different lengths, and calculates the cumulative Euclidean distance between corresponding points after alignment as the similarity distance. The smaller this distance, the more similar the degradation patterns of the two sequences are. The master node selects the minimum distance from all similarity distances. If this minimum distance is less than a preset similarity threshold, it is determined that the link quality degradation mode of the current slave node is highly similar to the degradation mode before the historical break, and the slave node is at risk of breakage. If the minimum distance is greater than or equal to the similarity threshold, the master node further inputs the sequence into the local trend extrapolation prediction model. The model uses a linear regression algorithm, with the measurement time corresponding to each time point in the sequence as the independent variable and the comprehensive wireless link quality score as the dependent variable, to fit a least squares regression line, and calculates the predicted score value for the future preset time (e.g., the next 5 seconds) based on the slope of the regression line. If the predicted score value is lower than the switching threshold defined in step S105, the slave node is also determined to have a risk of failure.

[0084] Based on this, the master node combines the results of the above comparison or prediction with the multi-dimensional profile of the slave node to comprehensively determine whether the slave node has a risk of failure. Specifically, the master node queries the stability coefficient in the multi-dimensional profile of the slave node. If the stability coefficient of the slave node is high (e.g., greater than 0.8), it indicates that the link quality of the slave node has been relatively stable in the past. At this time, the master node assigns a low confidence weight to the comparison results of trend information and historical degradation feature database (i.e., a single trend match is not enough to determine the existence of risk), and only issues a switching command when the comparison results and prediction results both point to a risk of failure.

[0085] If the stability coefficient of the slave node is low (e.g., less than 0.5), it indicates that the link quality of the slave node has fluctuated greatly in the past. In this case, the master node assigns a high confidence weight to the comparison results of trend information and historical degradation feature database. When either the comparison result or the prediction result points to the risk of failure, a switching command can be issued, thereby achieving more sensitive risk prediction on low stability nodes.

[0086] When the master node determines that the slave node is at risk of failure, it sends a handover instruction to the target slave node through the downlink broadcast channel before the next transmission time slot allocated to the slave node arrives. The handover instruction contains the identifier of the target backup parent node recommended by the master node for the slave node (the master node recommends the target backup parent node based on the adjacent nodes with higher connectivity coefficients in the multi-dimensional profile of the slave node).

[0087] Upon receiving the handover instruction, the target slave node proactively executes the parent node handover process before the actual break in the wireless link with the current parent node occurs. The specific handover execution method is the same as the parent node handover process described in step S105, including sending a handover warning frame to the current parent node, sending a network access request frame to the target backup parent node, receiving a network access confirmation frame, and updating the local parent node record. Because the handover action is performed before the actual link break occurs, the target slave node will not experience data loss or transmission interruption during the handover process, achieving an upgrade from "passive response-based reconstruction" to "proactive predictive handover."

[0088] For S501, the network management system executes a dynamic time slot allocation optimization process on the master node side. Before generating a new time slot allocation table in each superframe cycle (e.g., one superframe every 1 second), the microcontroller unit of the master node reads the multi-dimensional profile of each slave node generated and continuously updated in step S301. The master node extracts the stability coefficient and reliability coefficient of each slave node from the multi-dimensional profile of each slave node, and comprehensively evaluates the overall communication quality level of each slave node based on the above two coefficients: the overall communication quality level is determined by the weighted average of the stability coefficient and the reliability coefficient (e.g., the stability coefficient has a weight of 0.6 and the reliability coefficient has a weight of 0.4), and the higher the weighted average, the higher the overall communication quality level.

[0089] When generating the time slot allocation table, the master node dynamically adjusts the transmission time slot allocation strategy for each slave node based on its overall communication quality level. For slave nodes with relatively low overall communication quality levels (e.g., ranking in the bottom 50% of all slave nodes), the master node allocates a relatively higher upload frequency. For example, the normal superframe period is shortened from 1 second to a dedicated short period for that slave node (e.g., 500 milliseconds), giving these slave nodes more upload opportunities per unit of time. Specifically, the master node increases the number of initial transmission offsets allocated to these slave nodes in the time slot allocation table, meaning it allocates multiple independent transmission time slots to them within the same superframe period, instead of allocating only one transmission time slot as to slave nodes with higher overall communication quality levels. By tilting time slot resources towards slave nodes with relatively low communication quality, the network management system enables these nodes to upload collected data to the master node at a higher frequency before the link quality deteriorates further, thereby reducing the risk of data loss due to sudden link failures. When the multi-dimensional profile of a low-communication-quality slave node is updated and improved in subsequent statistical periods, the master node correspondingly reduces the number of over-allocated time slots and restores the normal upload frequency, so that the data collection integrity of all slave nodes in the network tends to be balanced.

[0090] Optionally, the network management method may also include the following steps: S601, when a topology reconfiguration occurs, the master node obtains network communication behavior data of the target slave node before and after the break occurs, and determines the type of break cause based on the network communication behavior data.

[0091] S602, the master node corrects the dimensions related to the cause of the break in the multi-dimensional profile of the target slave node according to the determined cause of the break, and uses the corrected multi-dimensional profile as the basis for the target slave node to elect a parent node based on the wireless link quality and upload data in the allocated transmission time slot.

[0092] S701, when the cause of the breakage belongs to a preset category, the master node obtains the data of other slave nodes around the target slave node in the same time period, and calculates the data of the target slave node during the breakage period based on the historical data relationship between the target slave node and other slave nodes around the target slave node before the breakage, and sends the pushed data together with the data of the master node itself and other slave nodes when forwarding to the remote terminal; S702, after the target slave node is able to upload data again, the master node compares the real data uploaded by the target slave node with the estimated data, and corrects the historical data correlation based on the comparison results.

[0093] In implementation, for S601, the network management system executes a fault determination process on the master node side. This process is triggered by the slave node triggering a parent node switch and completing topology reconstruction in step S105. That is, whenever a slave node switches to a backup parent node due to the wireless link quality being lower than a preset threshold, and the master node receives the topology update notification reported by the slave node through the new parent node, it determines that a topology reconstruction event has been completed, and then starts the fault determination process.

[0094] The microcontroller unit of the master node first retrieves network communication behavior data of the target slave node (i.e., the slave node that has experienced a failure and completed a switchover) from the local historical data storage area within a preset time window before and after the failure. This time window includes a first preset duration before the failure (e.g., 120 seconds before the failure) and a second preset duration after the failure (e.g., 30 seconds after the failure) to cover the entire process from the start of the failure to the completion of the switchover.

[0095] The retrieved data includes, but is not limited to: a comprehensive wireless link quality score record for each target slave node within the time window and its corresponding reporting timestamp (this record is periodically reported to the master node by the target slave node in step S402 in the form of a trend information appended field); the load history data of the target slave node's current parent node (i.e., the parent node before the break) within the time window, including the number of child nodes, the total amount of data received, and the average response delay of sending acknowledgment frames in each statistical period (extracted from the running status information reported by the parent node to the master node); and the auxiliary listening records of the target slave node's neighbor nodes before and after the break, i.e., whether the neighbor nodes can still detect the target slave node's wireless signal within the preset time window before and after the break (obtained through the historical log of the target slave node's received signal strength indication value recorded by the neighbor nodes when continuously monitoring the surrounding wireless signals in step S105).

[0096] The microcontroller unit of the master node inputs the aforementioned network communication behavior data into a preset fracture cause classifier to determine the fracture cause category. This fracture cause classifier performs pattern matching sequentially according to the following determination logic: First, the determination of physical environment interference. The master node checks the changes in the overall wireless link quality score of other slave nodes around the target slave node during the same time period before and after the break. If the master node finds that during the same time period when the break occurs, the overall wireless link quality score of at least two neighboring nodes of the target slave node also shows a significant synchronous decrease (i.e., the decrease exceeds a preset synchronous decrease threshold, for example, the decrease exceeds 30% of their respective normal levels), and the difference between the start time of the decrease of the above-mentioned nodes and the start time of the decrease of the target slave node is within a preset time tolerance range (e.g., ±3 seconds), then the master node determines that the cause of the break is physical environment interference. Typical scenarios of this type include, but are not limited to, strong electromagnetic interference sources in the area where the power distribution cabinet is located, signal attenuation caused by the removal of temporary metal obstructions, or sudden interference from surrounding wireless devices on the same frequency band.

[0097] Second, the determination of parent node overload type. If the determination condition of physical environment interference type is not met, the master node further checks the parent node's load history data. The master node calculates the average response latency of the parent node within a preset statistical window before the failure (e.g., 60 seconds before the failure) and compares it with the baseline response latency of the parent node under normal operation (this baseline value is provided by the historical average response latency statistically calculated by the parent node in step S301). If the average response latency of the parent node before the failure exceeds a preset multiple of the baseline response latency (e.g., more than twice the baseline value), and the number of child nodes of the parent node before the failure reaches or exceeds its preset maximum child node capacity (e.g., the maximum capacity is 8 child nodes), then the master node determines that the failure is due to parent node overload. A typical scenario of this type is that the parent node's processing capacity is saturated due to too many connected child nodes or too heavy data forwarding tasks, making it unable to respond to the child nodes' data sending requests in a timely manner.

[0098] Third, the determination of parent node hardware failure. If neither of the above two types is met, the master node further checks whether the target slave node's parent node is completely offline after the failure occurs. Specifically, the master node queries whether the parent node still reports any wireless signals (including heartbeat response frames, beacon frames, secondary beacon frames, or data acknowledgment frames) to the master node within a preset offline detection window (e.g., 60 seconds after the failure). If the master node does not receive any wireless signals from the parent node within this offline detection window, and all child nodes under the parent node fail and switch to other parent nodes at similar times (e.g., within ±5 seconds), the master node determines that the failure is due to parent node hardware failure. Typical scenarios for this type are unexpected power outages in the distribution cabinet where the parent node is located, hardware damage to the parent node's wireless communication module, or deadlock of the parent node's microcontroller unit.

[0099] Fourth, the determination of self-occlusion type. If none of the above three types are met, the master node further examines the degradation curve shape of the target slave node's overall wireless link quality score before the break occurred. If the degradation curve shows a slow and continuous downward trend (i.e., the absolute value of the degradation rate is less than the preset slow degradation threshold, for example, the score decrease is less than 0.5 points / second), while the wireless link quality of other nodes (including the parent node and other neighboring nodes) remains stable during the same period, the master node determines that the break is caused by self-occlusion. A typical scenario for this type is that the distribution cabinet door is closed or there are temporary obstructions around the cabinet, causing the slave node's wireless signal to be attenuated.

[0100] The master node performs pattern matching sequentially according to the above judgment logic. When the first matching type is found, subsequent matching stops, and this type is taken as the determination result for the fracture cause category. If none of the above four types meet the matching conditions, the master node marks the fracture cause category as "unknown type" and records the event in the local anomaly log for subsequent manual analysis by operations and maintenance personnel. The master node associates and stores the determined fracture cause category with the unique identifier of the target slave node, the fracture occurrence timestamp, and the parent node identifier at the time of the fracture, for use in step S602.

[0101] For S602, the network management system performs targeted corrections to the multi-dimensional profile of the target slave node based on the breakage cause category determined in S601, and uses the corrected profile as the basis for subsequent network behavior decisions. The microcontroller unit of the master node first reads the current multi-dimensional profile of the target slave node stored in the local profile database. Subsequently, the master node determines the profile dimensions that need to be corrected and the direction of correction based on the breakage cause category determined in step S601: If the cause of the break is classified as physical environment interference, it indicates that the wireless signal environment at the physical location of the target slave node is subject to occasional but strong external interference. This interference is unrelated to the target slave node's own hardware, but it will intermittently affect the link quality stability of the node. Therefore, the master node lowers the stability coefficient in the multi-dimensional profile of the target slave node. Specifically, the reduction is, for example, by multiplying the current stability coefficient by a preset attenuation factor (e.g., 0.8), to reflect the degradation of the stability characteristics of this type of node due to external environmental interference. At the same time, the node is marked as a "vulnerable node to environmental interference", so that when the master node performs parent node election and topology reconstruction in the future, it will be given higher early warning sensitivity when judging the break risk of this node (i.e., when judging the break risk by combining the multi-dimensional profile as described in weight 4, a higher confidence weight is given to this type of node), and when there are switchable candidate parent nodes around the node, more backup parent nodes will be allocated to it first. After the stability coefficient is lowered, the master node adds an environmental interference flag field to the node. This field records the occurrence time and duration of the most recent interference event for subsequent periodic checks. If the node does not experience a similar failure in the following period of operation, the master node will gradually restore the stability coefficient to the normal level.

[0102] If the cause of the failure is classified as parent node overload, it indicates that the failure is not caused by the target slave node itself, but by the insufficient processing capacity of its parent node. Therefore, the master node does not modify the profile of the target slave node, but instead adjusts the load coefficient in the multi-dimensional profile of the target slave node's parent node (i.e., the overloaded parent node that caused the failure) upwards. Specifically, the adjustment is made by multiplying the current load coefficient by a preset increment factor (e.g., 1.2), and the reliability coefficient of the parent node is reduced. This lowers the priority of the parent node when it is elected as the parent node by other slave nodes in the future, thereby reducing the risk of it becoming overloaded again in the future.

[0103] If the cause of the failure is classified as a parent node hardware failure, it indicates that the parent node has failed entirely, and its reliability as a parent node is fundamentally problematic. The master node significantly lowers the reliability coefficient in the multi-dimensional profile of the failed parent node. Specifically, the reduction is, for example, by multiplying the current reliability coefficient by a preset severe decay factor (e.g., 0.5), and prohibits the parent node from being elected as the master node or serving as the parent node of any slave node for a preset penalty period (e.g., 72 hours). If the failed parent node subsequently comes back online (i.e., is able to upload data to the master node again) and does not experience a similar failure within a preset observation period (e.g., 7 consecutive days), the master node gradually restores its reliability coefficient upwards according to a preset recovery rate (e.g., increasing by 0.05 per statistical period) until it reaches a normal level.

[0104] If the cause of the disconnection is classified as node self-obstruction, it indicates that the disconnection is caused by the wireless signal at the physical location of the target slave node being obstructed, leading to continuous signal degradation. Therefore, the master node adjusts the connectivity coefficient in the multi-dimensional profile of the target slave node by multiplying the current connectivity coefficient by a preset attenuation factor (e.g., 0.85) and increasing the number of backup parent nodes for that node (e.g., from the preset 2 backup parent nodes to 4). This provides the node with more candidate parent nodes to choose from during future parent node switching, thereby reducing the probability of data transmission interruption due to single link obstruction.

[0105] If the cause of the break is unknown, the master node will not automatically correct it, but will record the event and trigger a system alarm, waiting for maintenance personnel to conduct manual investigation.

[0106] After the master node completes the above-mentioned portrait correction, it persists the corrected multidimensional portrait in the local portrait database, overwriting the original portrait record of the node. From the time the correction is completed, the target slave node (or its parent node, depending on the correction object) will use the corrected multidimensional portrait as the decision basis when performing parent node election in S101, transmission time slot allocation in S103, and topology reconstruction in step S105. That is, in the parent node election, the adjacent node with higher stability coefficient and reliability coefficient is given priority as the parent node; in the transmission time slot allocation, the master node dynamically adjusts the time slot allocation strategy according to the corrected portrait (the specific execution method is the same as in step S501); in the topology reconstruction, the master node configures a corresponding number of backup parent nodes for the node according to the corrected connectivity coefficient.

[0107] Thus, S601 and S602 form a complete adaptive closed loop of "fracture diagnosis → profile correction → feedback to subsequent decision-making", enabling the network management system to extract experience from each fracture event and continuously optimize subsequent network decision-making behavior.

[0108] For S701, after determining the type of breakage in S601, the network management system further determines whether the breakage cause category belongs to a preset category that would prevent the slave node from performing local caching operations. This preset category includes, but is not limited to, the parent node hardware failure type defined in step S601. In this type of breakage scenario, the parent node of the target slave node is completely offline, but the target slave node itself has not lost power and can still perform data acquisition and preprocessing normally. However, because the parent node is completely offline, although the "local cache retransmission" mechanism in step S103 can work normally (i.e., the target slave node can cache standardized data packets in local non-volatile storage media), when the parent node recovers and re-uploads, the data locally cached by the parent node during the breakage is missing. This is because the parent node's original acquired data could not be uploaded to the master node during the breakage, and the parent node, as the main body of the breakage, cannot perform local caching operations in a completely offline state. Therefore, when the breakage cause category belongs to the parent node hardware failure type, a data replenishment calculation process needs to be initiated.

[0109] When the breakage cause category determined by S601 belongs to a preset category, the microcontroller unit of the master node initiates the complement calculation process. The master node first obtains data from other slave nodes around the target slave node (in this scenario, the parent node that experienced the hardware failure, hereinafter referred to as the faulty node). Specifically, the master node queries the connectivity coefficient in the multidimensional profile of the faulty node from the local profile database to determine the list of other slave nodes that are physically adjacent to the faulty node (i.e., can detect each other's wireless signals), and retrieves the valid data fields (including three-phase current values, three-phase voltage values, temperature values, etc.) from the standardized data packets of the aforementioned adjacent slave nodes during the same time period of the faulty node's breakage from the local data storage area.

[0110] The master node simultaneously retrieves historical data sequences from its local historical data storage area, covering the last continuous online period before the fault node's failure. These historical data sequences span a pre-defined time window before the failure (e.g., 60 seconds prior to failure), and include the values ​​of various physical quantities uploaded by the fault node within that time window. The master node's microcontroller unit uses the data sequences of adjacent slave nodes during the same time period during the failure as input features and the historical data sequences of the fault node within the pre-defined time window before the failure as output labels. A linear regression algorithm is used to establish a data association model between the fault node and its adjacent slave nodes. Specifically, for each acquired physical quantity (e.g., phase A current), a multiple linear regression equation is established between the value of that physical quantity at the fault node and the values ​​of the same physical quantity at each adjacent slave node. The regression coefficients are determined by least squares fitting. The physical meaning of this data association model is that, under normal operating conditions, there is a fixed linear proportional relationship between the value of a certain acquired physical quantity at the fault node and the values ​​of the same physical quantity at its adjacent nodes. This relationship is determined by the coupling characteristics of the loads of each distribution cabinet in the distribution cabinet cluster.

[0111] After the data association model is established, the microcontroller unit of the master node substitutes the actual data sequences of each adjacent slave node during the fault node's interruption period into the data association model to calculate the estimated values ​​of each physical quantity collected by the fault node at each moment during the interruption period. These estimated values ​​are then arranged in chronological order to generate the estimated data sequence of the fault node during the interruption period. This estimated data sequence is not simply calculated independently for a single moment, but rather calculated sequentially for each data acquisition moment during the interruption period (e.g., every 100 milliseconds is one acquisition moment), thus forming a complete data sequence for the interruption period. The master node packages the above estimated data sequence together with the actual data from itself and other slave nodes during the same time period in the aggregation processing stage of step S104, and sends it together when forwarding it to the remote terminal. An estimation flag and confidence level identifier are appended to each data field of the estimated data sequence. The inference marker is used to inform the remote terminal that the data is inferred data rather than actual collected data. The confidence level is calculated as the normalized product of the goodness of fit R² of the data association model (i.e., the coefficient of determination of the regression model, with a value ranging from 0 to 1, and a higher value indicating a higher model fitting accuracy) and the number of neighboring slave nodes involved in the modeling. This is used to quantify the credibility of the inferred data. For example, if there are 3 neighboring slave nodes around the faulty node involved in the modeling and the goodness of fit R² is 0.95, then the confidence level is 0.95; if there is only 1 neighboring slave node involved in the modeling and the goodness of fit R² is 0.7, then the confidence level is 0.7. The remote terminal can decide whether to use the inferred data for fault analysis or operational decision-making based on the confidence level.

[0112] In S702, after the faulty node is able to upload data again (i.e., after the faulty node restores power or the wireless communication module returns to normal and reconnects to the network, and uploads standardized data packets normally through step S103), the network management system executes the self-correction process of data association.

[0113] The microcontroller unit of the master node extracts the actual collected data at each moment during the interruption from the received real standardized data packets uploaded after the faulty node recovers (i.e., although the faulty node cannot upload data during the interruption, its microcontroller unit continuously collects and preprocesses data during the interruption, and caches the preprocessed standardized data packets in local non-volatile storage media; after recovery, the faulty node uploads the standardized data packets cached during the interruption to the master node in batches during the retransmission process, and the specific retransmission mechanism is the same as the retransmission process in step S103). The master node compares the actual collected data of the faulty node at each moment during the interruption with the estimated data sequence generated in step S701, calculates the estimation error (the absolute difference between the actual value and the estimated value) at each collection moment, and calculates the average estimation error and the maximum estimation error during the entire interruption period.

[0114] The master node modifies the historical data association established in step S701 based on the comparison results. Specifically, if the average inference error is less than a preset first error threshold (e.g., the average inference error is less than 5% of the true value), it indicates that the current data association model has high accuracy. The master node maintains the model parameters unchanged and only adds the comparison results of the true data and inference data during the current break as new training samples to the sample set when establishing the data association model again, in order to improve the model's generalization ability. If the average inference error is greater than or equal to the first error threshold but less than a preset second error threshold (e.g., the average inference error is less than 15% of the true value), it indicates that the current model has biases but the overall trend is usable. The master node includes the comparison results of the true data and inference data during the current break in the training sample set, re-executes linear regression fitting, updates the regression coefficients, and generates the modified data association model.

[0115] If the average extrapolation error is greater than or equal to the second error threshold, it indicates that the current data association model is severely inaccurate. The master node will mark this comparison result as an abnormal event, trigger a system alarm to notify the operations and maintenance personnel for manual investigation, and suspend the use of the data association model for subsequent extrapolations until the cause is identified. For correction scenarios where the error is between the first and second thresholds, the master node will store the corrected data association model in the local model library, replacing the original historical data association relationship, as the basis for extrapolating the complement in the next similar break scenario.

[0116] Through periodic comparisons of real and estimated data and iterative model corrections, the estimation accuracy of the data association model gradually improves with the passage of time and the accumulation of broken samples, realizing the self-evolution of data association. The real data generated by each broken event not only fills the data gap during the network outage, but also corrects and optimizes the mathematical model used for estimation.

[0117] This application also discloses a power distribution cabinet cluster IoT networking and data acquisition system based on a wireless self-organizing network, including: The networking module 201 is used to automatically elect a master node through wireless link quality assessment after each power distribution cabinet node is powered on, and the remaining nodes automatically become slave nodes. Each slave node elects a parent node and determines the network level based on the wireless link quality, forming a cluster network topology with the master node as the root node and multiple redundant communication paths. The preprocessing module 202 is set in the master node and each of the slave nodes, and is used to preprocess their respective raw collected data locally to generate standardized data packets; A communication scheduling module 203, located in each of the slave nodes, is used to monitor the communication link status between each slave node and the master node. When the link is normal, the master node publishes a synchronization timing reference to each slave node, and each slave node uploads the standardized data packets to the master node within the allocated transmission time slot according to a preset data priority. When the link is interrupted, each slave node caches the standardized data packets locally, and automatically reads the cached data and uploads it to the master node after the link is restored. The aggregation and forwarding module 204 is located in the master node and is used to receive the standardized data packets uploaded by each slave node, aggregate and process them together with the standardized data packets preprocessed by the master node itself, and then forward them to the remote terminal.

[0118] This application also discloses a power distribution cabinet cluster IoT networking and data acquisition device based on a wireless ad hoc network. The power distribution cabinet cluster IoT networking and data acquisition device based on a wireless ad hoc network includes a memory and a processor. The memory stores a computer program that can be loaded by the processor and executed as described above for the power distribution cabinet cluster IoT networking and data acquisition method based on a wireless ad hoc network.

[0119] This application also discloses a computer-readable storage medium that stores a computer program that can be loaded by a processor and executed as described above regarding the IoT networking and data acquisition method for power distribution cabinet clusters based on a wireless ad hoc network. The computer-readable storage medium includes, for example, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0120] It should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0121] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit the scope of protection of the application. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

Claims

1. A method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless ad hoc networks, applied to each power distribution cabinet node in a power distribution cabinet cluster, characterized in that, include: After each power distribution cabinet node is powered on, a master node is automatically elected through wireless link quality assessment, and the remaining nodes automatically become slave nodes. Each slave node elects a parent node and determines the network level based on wireless link quality, forming a cluster network topology with the master node as the root node and multiple redundant communication paths. The master node and each slave node preprocess their respective raw collected data locally to generate standardized data packets; Each slave node monitors the communication link status between itself and the master node. When the link is normal, the master node publishes a synchronization timing reference to each slave node. Each slave node uploads the standardized data packet to the master node within the allocated transmission time slot according to the preset data priority. When the link is interrupted, each slave node caches the standardized data packet locally. After the link is restored, it automatically reads the cached data and uploads it to the master node. The master node receives the standardized data packets uploaded by each slave node, and together with the standardized data packets pre-processed by the master node itself, they are aggregated, processed, and then forwarded to the remote terminal.

2. The method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless self-organizing networks according to claim 1, characterized in that, The method further includes: Each slave node continuously monitors the quality of its wireless link with its current parent node. When the wireless link quality is detected to be lower than a preset threshold, it automatically switches to a backup parent node to reconstruct the cluster network topology.

3. The method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless self-organizing networks according to claim 1, characterized in that, The method further includes: The master node records statistical information related to the network communication behavior of each slave node during operation, and generates a multi-dimensional profile to characterize the network communication quality characteristics of each slave node based on the statistical information. The master node uses the multi-dimensional profile as the basis for each slave node to elect a parent node based on link quality and to upload data within the allocated transmission time slots.

4. The method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless self-organizing networks according to claim 3, characterized in that, The method further includes: The master node records the wireless link quality degradation patterns of each slave node before the historical topology reconstruction occurs, and constructs a historical degradation feature library; During the continuous monitoring of the wireless link quality between itself and its current parent node, each slave node records the trend information of the wireless link quality changing over time and attaches the trend information when uploading data. The master node compares the trend information with the historical degradation feature database and combines it with the multi-dimensional profiles of each slave node to comprehensively determine whether each slave node has a risk of failure. When a risk of failure is determined, a switching command is sent to the corresponding target slave node in advance, so that the switch to the backup parent node occurs before the wireless link fails.

5. The method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless self-organizing networks according to claim 3, characterized in that, The method further includes: The master node dynamically adjusts the frequency at which each slave node uploads data within its allocated transmission time slot based on the multi-dimensional profile of each slave node; wherein, slave nodes with relatively low communication quality as represented by the multi-dimensional profile are assigned a relatively higher upload frequency.

6. The method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless self-organizing networks according to claim 4, characterized in that, The method further includes: When a topology reconfiguration occurs, the master node acquires network communication behavior data of the target slave node before and after the break occurs, and determines the type of break cause based on the network communication behavior data. The master node corrects the dimensions associated with the breakage cause in the multidimensional profile of the target slave node based on the determined breakage cause category, and uses the corrected multidimensional profile as the basis for the target slave node to subsequently elect a parent node based on wireless link quality and upload data in the allocated transmission time slot.

7. The method for IoT networking and data acquisition of power distribution cabinet clusters based on wireless self-organizing networks according to claim 6, characterized in that, The method further includes: When the cause of the breakage belongs to a preset category, the master node obtains the data of other slave nodes around the target slave node in the same time period, and calculates the data of the target slave node during the breakage based on the historical data association between the target slave node and other slave nodes around the target slave node before the breakage. The master node then sends the data along with the data of the master node itself and other slave nodes to the remote terminal. Once the target slave node is able to upload data again, the master node compares the actual data uploaded by the target slave node with the estimated data, and corrects the historical data association based on the comparison results.

8. A power distribution cabinet cluster IoT networking and data acquisition system based on a wireless self-organizing network, characterized in that, include: The networking module (201) is used to automatically elect a master node through wireless link quality assessment after each power distribution cabinet node is powered on, and the remaining nodes automatically become slave nodes. Each slave node elects a parent node and determines the network level based on the wireless link quality, forming a cluster network topology with the master node as the root node and multiple redundant communication paths. The preprocessing module (202) is set in the master node and each of the slave nodes, and is used to preprocess their respective raw collected data locally to generate standardized data packets; A communication scheduling module (203) is installed in each of the slave nodes to monitor the communication link status between each slave node and the master node. When the link is normal, the master node publishes a synchronization timing reference to each slave node. Each slave node uploads the standardized data packet to the master node within the allocated transmission time slot according to the preset data priority. When the link is interrupted, each slave node caches the standardized data packet locally. After the link is restored, it automatically reads the cached data and uploads it to the master node. The aggregation and forwarding module (204) is located in the master node and is used to receive the standardized data packets uploaded by each slave node, aggregate and process them together with the standardized data packets generated by the master node itself after preprocessing, and then forward them to the remote terminal.

9. A distribution cabinet cluster IoT networking and data acquisition device based on a wireless self-organizing network, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer program is stored that can be loaded by a processor and executed as described in any one of claims 1 to 7.

Images