An LED lamp strip monitoring system based on LoRa bee colony dual power supply AI diagnosis

Abstract

The application relates to the technical field of LED lighting intelligent monitoring, non-contact electrical detection, low-power wireless communication, industrial safety control and artificial intelligence fault diagnosis, and discloses an LED light bar monitoring system based on LoRa bee colony dual-power supply AI diagnosis, which comprises a data acquisition module, a bee colony Mesh ad hoc network transmission module, an AI intelligent diagnosis module and a relay fault safety control module. Through lossless sensing and hardware-level configurable safety strategy technology, the LED light bar full-life-cycle monitoring demand can be covered, the monitoring result is more accurate in reflecting the real operation condition of the LED light bar, and a more scientific basis is provided for LED light bar operation and maintenance and safety management.

Description

Technical Field

[0001] This invention relates to the fields of intelligent monitoring of LED lighting, non-contact electrical detection, low-power wireless communication, industrial safety control, and artificial intelligence fault diagnosis. More specifically, it discloses an LED light strip monitoring system based on LoRa colony dual power supply AI diagnosis. Background Technology

[0002] LED light strips are linear lighting devices composed of multiple LED beads connected in series or parallel. Due to their flexible structure, high brightness, and low energy consumption, they are widely used in highway and railway tunnel lighting, municipal street lights, airport navigation lights, industrial equipment status indicators, and outdoor landscape lighting. In actual operation, light strips are often deployed on a large scale in a multi-parallel configuration and are often in complex environments such as no mains power, strong electromagnetic interference, and high humidity and oil contamination. Relying solely on manual inspection is not only inefficient but also fails to identify potential faults such as damaged LED beads, aging circuits, and short circuits or overcurrents in a timely manner.

[0003] The prior art patent document with authorization announcement number CN119658981A discloses "A method and system for monitoring the operation of an LED silicone light strip extruder". Within a preset period, the extrusion state data of the extruder is acquired, and extrusion state behavior values ​​of the extruder within the preset period are obtained based on the extrusion state data. The extruder state is divided according to the extrusion state behavior values ​​to obtain extruder state signals. Based on the extruder operating state difference signals, the extruder is modularly divided to obtain the divided operating area units, and the monitoring anomaly value of each operating area unit is acquired. Based on the monitoring anomaly values ​​of the operating area units, the operating area units are linked to obtain monitoring reminder values ​​for the operating area units, and the targeted monitoring of the extruder operating area units is completed based on the monitoring reminder values.

[0004] The patent document with authorization announcement number CN107590499B discloses "A video-based method and system for monitoring the status of LED indicator lights on a device", which includes: acquiring image data of the back panel where the device's LED indicator lights are located; separating the possible LED indicator light areas from the device's back panel based on the statistical distribution differences in color space features between the LED indicator lights and the device's back panel; calculating the shape features of the separated areas, removing areas whose shape features do not conform to the shape features of the LED indicator lights from the separation results, and using areas whose shape features conform to the shape features of the LED indicator lights as the LED indicator lights to be detected; detecting the working status of the separated LED indicator lights, comparing the detected information with the information on normal device operation, and determining whether the device is working properly.

[0005] While existing technologies can achieve automatic monitoring of LED indicator lights through video imaging, improving the efficiency of equipment monitoring and the accuracy of indicator light extraction in computer rooms, and can accurately identify the operating status of LED silicone light strip extruders, and achieve intelligent monitoring of their maintenance locations through modular division, existing technologies have not formed a complete solution for the full-scenario monitoring needs of multi-parallel LED light strips. They generally suffer from problems such as the need for wire breaking for sampling, high electrical safety risks, reliance on central gateways for wireless networking, susceptibility to single-point failures and lack of network self-healing capabilities, and reliance on fixed current thresholds for fault diagnosis, which cannot accurately identify the number of faulty LEDs and the aging trend of the light strips, resulting in high false alarm rates and a lack of predictive maintenance capabilities. The power supply mode is singular, making it difficult to balance equipment endurance and real-time monitoring, and there are no hardware-level configurable security policies. They have poor adaptability to all scenarios and cannot meet the compliance requirements of highly regulated scenarios such as civil aviation, resulting in low overall operation and maintenance efficiency. Summary of the Invention

[0006] The main technical problem solved by this invention is to provide an LED light strip monitoring system based on LoRa colony dual power supply AI diagnostics, which can solve the problems mentioned in the background art.

[0007] To solve the above-mentioned technical problems, according to one aspect of the present invention, more specifically, an LED light strip monitoring system based on LoRa cellular dual power supply AI diagnostics, comprising: a data acquisition module, a cellular Mesh self-organizing network transmission module, an AI intelligent diagnostic module, and a relay fault safety control module; The data acquisition module performs non-contact current sampling on the LED light strip circuit to obtain circuit current data; The cellular mesh self-organizing network transmission module is based on LoRa to build a decentralized cellular mesh self-organizing network, realizing multi-hop transmission of current data and network self-healing. The AI ​​intelligent diagnostic module extracts current timing features to complete fault diagnosis and lifespan prediction of LED light strips; The relay fault safety control module performs fault safety isolation and authorized reset operations on the LED light strip based on the fault judgment results of the AI ​​intelligent diagnostic module.

[0008] Furthermore, the data acquisition module includes: a Hall sampling module, a zero-point self-calibration module, and a temperature drift adaptive compensation module; Hall sampling module: Employs a clamp-on Hall sensor to perform non-contact current acquisition; Zero-point self-calibration module: Performs zero-point self-calibration of current sampling to eliminate sensor factory deviation; Temperature drift adaptive compensation module: performs dynamic temperature drift compensation on current data based on ambient temperature.

[0009] Furthermore, the swarm mesh self-organizing network transmission module includes: a decentralized networking module, a route self-reconstruction module, and a local caching and retransmission module; Decentralized networking module: Constructs a decentralized networking architecture with all nodes peering; The self-reconstruction routing module identifies faulty nodes based on node heartbeat packets and completes autonomous network routing reconstruction. Local caching and retransmission module: Used to store data locally when communication is interrupted, and retransmit it in batches according to timestamps after communication is restored.

[0010] Furthermore, the AI ​​intelligent diagnostic module includes: a time-series feature extraction module, a diagnostic module, and an aging and lifespan prediction module; Timing feature extraction module: extracts current amplitude, steady-state decay rate, fluctuation frequency, transient changes, and step current characteristics of LED damage. Diagnostic module: Based on AI models, it completes fault identification, counts the number of faulty LEDs, and determines the fault level; Aging and Lifespan Prediction Module: Fits aging curves, analyzes aging trends, and predicts remaining lifespan.

[0011] Furthermore, the relay fault safety control module includes: a fault level matching module, a safety isolation module, and a manual authorized reset module; Fault level matching module: classifies ordinary / high-risk faults and matches control actions, and adapts the matching logic to hardware-level configurable safety policies; Safety isolation module: Used for circuit power-off isolation in case of high-risk faults; the relay adopts a fault pass-through hardware design. Manually authorized reset module: Performs remote authorized reset of relays through the operation and maintenance platform.

[0012] Furthermore, it also includes a dual-redundant intelligent power supply module and an integrated waterproof protection and easy installation module.

[0013] Furthermore, the dual-redundant intelligent power supply module includes: a power supply switching module, a low-power dynamic adaptation module, and a capacitor freewheeling current protection module; Power supply switching module: Based on light intensity, external DC power input status, and remaining lithium battery power, it intelligently and dynamically switches between three modes: solar power supply, external DC power supply, and lithium battery backup power supply, with external DC power supply being the primary mode. Low-power dynamic adaptation module: dynamically adjusts the current sampling frequency and data reporting cycle according to the remaining lithium battery power; Capacitor freewheeling protection module: Provides freewheeling power during power supply mode switching through a large-capacity electrolytic capacitor.

[0014] Furthermore, the integrated waterproof protection and convenient installation module includes: a waterproof protection module and a snap-on installation module; Waterproof protection module: It adopts an integrated waterproof and flame-retardant shell and is equipped with a sealing structure with a corresponding protection level; Clip-on installation module: Equipped with a clip-on installation base to prevent detachment, the base is compatible with conventional installation carriers such as light poles and tunnel walls.

[0015] An LED strip monitoring terminal based on LoRa cellular dual-power supply and AI diagnostics includes: the terminal is an integrated modular hardware, integrating a main control MCU unit, a data acquisition module, a dual-redundant intelligent power supply module, a cellular mesh self-organizing network transmission module, an AI intelligent diagnostic module, a relay fault safety control module, and an integrated waterproof protection and convenient installation module; each functional module is electrically connected to the main control MCU unit through a standardized SPI interface, and the main control MCU unit coordinates the working logic linkage and data interaction of each module.

[0016] A method for monitoring LED light strips based on LoRa colony dual-power AI diagnostics includes the following steps: S1. The data acquisition module uses a clamp-on Hall sensor to perform non-contact current acquisition of the LED light strip circuit. First, it automatically completes the zero-point self-calibration of the current sampling, then collects the sensor's working environment temperature in real time and performs temperature drift adaptive compensation on the current acquisition data to obtain the calibrated LED light strip circuit current data. S2. The bee colony mesh self-organizing network transmission module constructs a decentralized self-organizing network architecture with all nodes peer-to-peer. It transmits the calibrated current data obtained in step S1 to the outside through a multi-hop relay method. When communication is interrupted, the current data is stored locally. After communication is restored, the stored data is retransmitted in batches according to the timestamp. The faulty node is determined by the node heartbeat packet and the network routing is autonomously reconstructed based on the greedy algorithm. The S3 AI intelligent diagnostic module extracts current amplitude, steady-state decay rate, fluctuation frequency, transient change characteristics, and current step characteristics after lamp bead damage from the transmitted current data. Based on a one-dimensional CNN + random forest fusion AI model, it completes lamp strip fault identification, counts the number of faulty lamp beads, and determines the level of ordinary and high-risk faults. The determination process adopts a multi-level anti-false alarm mechanism with consistent results from multiple consecutive acquisitions. At the same time, it fits the aging curve based on the lamp strip current decay trend, analyzes the lamp strip aging trend, and predicts the remaining service life. S4. The relay fault safety control module matches the fault level judgment result of the AI ​​intelligent diagnosis module with the control action. For high-risk faults, the relay is triggered to perform power-off isolation of the LED light strip circuit. The relay hardware adopts a fault pass-through design. After the fault is handled, the remote reset operation of the relay is performed by manual authorization through the operation and maintenance platform.

[0017] The beneficial effects of this invention, an LED strip monitoring system based on LoRa cellular dual-power AI diagnostics, are as follows: Through non-destructive sensing and hardware-level configurable security strategies, it can cover the monitoring needs of the entire LED strip lifecycle, making the monitoring results more accurately reflect the actual operating status of the LED strip and providing a more scientific basis for LED strip operation and maintenance and safety management. Furthermore, through cellular decentralized mesh self-organizing network and data caching and retransmission technology, it achieves efficient integration of the monitoring terminal and management platform, breaking down data transmission barriers and enabling real-time data sharing and interaction between nodes. This improves the efficiency and scientific nature of LED strip resource management, enhances the accuracy and adaptability of the diagnostic model, and provides more reliable and compliant safety control measures through hardware-level configurable security strategies and AI intelligent diagnostic technology. It can clearly adapt to the safety requirements of different scenarios. Based on real-time current data and advanced diagnostic model decision support, it can quickly provide scientific handling solutions in emergency faults, greatly improving the decision-making level of LED strip management and the ability to respond to sudden faults. Attached Figure Description

[0018] The present invention will now be described in further detail with reference to the accompanying drawings and specific implementation methods.

[0019] Figure 1 This is a schematic diagram of the system module architecture; Figure 2 This is a schematic diagram of the monitoring terminal structure; Figure 3 A schematic diagram of the logic flow for switching between dual power supply modes; Figure 4 This is a schematic diagram of the closed-loop process for AI diagnosis and fault safety control. Detailed Implementation

[0020] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the present application can be combined with each other.

[0021] According to one aspect of the invention, such as Figures 1-4 As shown, an LED strip monitoring system based on LoRa cellular dual-power AI diagnostics is provided, including: a data acquisition module for non-contact current sampling of the LED strip circuit to obtain circuit current data. This module includes: Hall sampling module: Employs a clamp-on Hall sensor to perform non-contact current acquisition; First, open the jaws of the clamp-on Hall sensor along the opening and closing buckle, and directly snap it onto the external conductor of the LED light strip power supply bus. After snapping it in, close the buckle to complete the mechanical locking and fixation. The sensor has no electrical connection with the power supply bus, achieving electrical isolation from the LED light strip circuit. No modification operations such as powering off, cutting wires, or rewiring the original circuit are required. Then, based on the Hall effect electromagnetic induction principle, the sensor converts the changes in the current and magnetic field of the power supply bus into a linear analog electrical signal. The analog electrical signal is then amplified, filtered, and converted into a digital current signal by the built-in signal conditioning circuit of the sensor, thus providing a precise basic data source for subsequent data processing.

[0022] Zero-point self-calibration module: Performs zero-point self-calibration of current sampling to eliminate sensor factory deviation; After the terminal is powered on and initialized, the main control MCU unit immediately sends a calibration trigger command to the zero-point self-calibration module. At this time, the Hall sampling module remains in an unloaded working state, the clamp sensor does not collect any effective current signal, and the module enters a continuous unloaded current acquisition stage, continuously acquiring unloaded electrical signals at a high frequency of 100Hz. Then, the acquired no-load electrical signal is subjected to mean filtering to remove abnormal extreme data that deviate from the mean, and a stable no-load current reference value is calculated. This reference value is stored in the Flash non-volatile storage area of ​​the main control MCU unit as a permanent zero-point calibration reference. All subsequent sampled current data are corrected for zero-point offset with reference to this reference value, thus completely eliminating the systematic measurement error caused by hardware deviation and initial zero-point drift of the sensor at the factory.

[0023] Temperature drift adaptive compensation module: dynamically compensates for temperature drift in current data based on ambient temperature; Specifically, the main control MCU unit uses a high-precision NTC temperature sensor built into the temperature drift adaptive compensation module to collect the ambient temperature of the core working area of ​​the Hall sensor in real time. The collected temperature range fully covers the working temperature range of the Hall sensor (e.g., -40℃ to 85℃). The original temperature drift characteristic curve of the Hall sensor and the corresponding linear compensation formula are pre-written into the module's storage unit. The module automatically retrieves the corresponding temperature drift compensation coefficient from the temperature drift characteristic curve based on the real-time collected ambient temperature. Simultaneously, the compensation coefficient is substituted into the linear compensation formula to dynamically correct the temperature drift value of the digital current signal output in real time by the Hall sampling module. Each current sampling performs a temperature drift compensation operation synchronously, ensuring that the collected current data can maintain stable high accuracy under different ambient temperature conditions, completely eliminating the measurement deviation caused by changes in ambient temperature, and ensuring the accuracy of the data source for subsequent AI intelligent diagnosis.

[0024] The cellular mesh self-organizing network transmission module, based on LoRa, constructs a decentralized cellular mesh self-organizing network to achieve multi-hop transmission of current data and network self-healing. This module includes: Decentralized networking module: Constructs a decentralized networking architecture with all nodes peering; First, after the monitoring terminal is powered on and completes system initialization, it automatically wakes up the LoRa radio frequency communication unit, configures it to network mode, and scans the surrounding monitoring terminal nodes of the same type at a frequency of 1Hz in the preset LoRa working frequency band (433MHz). It sends encrypted identity authentication requests to the scanned nodes, and only the legitimate nodes that pass the authentication can join the network system. After scanning and authenticating all legitimate nodes, a decentralized Mesh networking architecture with all nodes being equal is constructed. Each node is assigned a unique physical identifier ID, and all nodes are guaranteed to have three core functions: data collection, multi-hop relay, and autonomous route maintenance. No central gateway needs to be configured during the networking process, and all nodes participate equally in data transmission and network maintenance. After the network architecture is finally built, each node interacts with its working status and routing information with its neighboring nodes in real time, automatically generates and stores a dynamic routing table locally. The routing table contains key information such as neighboring node identifiers, signal strength, and transmission delay. When a new node joins or a non-faulty node leaves, all nodes in the network update the routing table synchronously, realizing dynamic adaptive networking and adapting to different deployment scales of node addition and removal scenarios.

[0025] The self-reconstruction routing module identifies faulty nodes based on node heartbeat packets and completes autonomous network routing reconstruction. The determination of faulty nodes is achieved through heartbeat packet interaction. After the network is formed, all nodes send encrypted heartbeat packets to their neighboring nodes at a fixed time interval of 5 seconds. The heartbeat packet contains the node identifier, current working status and communication status information. If a node's neighboring nodes fail to receive its heartbeat packet multiple times in a row, the node is determined to be a faulty node. The neighboring nodes immediately broadcast routing update information to the entire network, triggering the network routing self-reconstruction process. The route reconstruction based on the greedy algorithm is as follows: after receiving the route update broadcast, all nodes in the network immediately retrieve their local dynamic routing tables. Using the shortest data transmission distance, the best signal strength, and the minimum transmission delay as the greedy selection criteria, the optimal combination of relay nodes is selected from the available legal nodes. The multi-hop relay path for data transmission is replanned, and faulty nodes are automatically removed from the transmission path to ensure uninterrupted data transmission and achieve rapid self-healing of the network.

[0026] Local caching and retransmission module: used to store data locally when communication is interrupted, and retransmit it in batches according to timestamps after communication is restored; First, the main control MCU unit detects the network communication status in real time through the communication feedback signal of the LoRa radio frequency communication unit. When it detects no communication feedback or data transmission failure for more than a preset number of consecutive times (e.g., 3 times), it immediately determines that the network communication is interrupted. At the same time, it sends a data caching instruction to the data acquisition module. The acquired current timing data will be encrypted and packaged according to the millisecond-level precision timestamp and stored in the terminal's built-in large-capacity Flash non-volatile cache unit. The cached data contains complete information such as sampling time, calibrated current value, temperature drift compensation coefficient, and node identifier, and is continuously stored in ascending order of timestamp without overwriting historical data. Immediately after the LoRa RF communication unit receives the network synchronization signal again and restores normal communication capability, it immediately sends a communication recovery signal to the main control MCU unit. The main control MCU unit then issues a data retransmission command, retrieves all unuploaded cached data from the local Flash cache unit in ascending order of timestamp, and transmits it to the target node in batch packet mode through multi-hop relay of the reconstructed Mesh network. After each packet of data transmission is completed, an integrity check is performed. Only after the check passes can the next packet be transmitted. After all cached data retransmission is completed and confirmed to be correct, the main control MCU unit will automatically delete the successfully uploaded data from the local Flash cache, release the cache space, and ensure the continuous storage of subsequent data.

[0027] The AI-powered intelligent diagnostic module extracts current timing characteristics to perform fault diagnosis and lifespan prediction for LED light strips. This module includes: Timing feature extraction module: extracts current amplitude, steady-state decay rate, fluctuation frequency, transient changes, and step current characteristics of LED damage. First, the calibrated current time series data output by the bee colony Mesh self-organizing network transmission module is received. Then, the data is preprocessed by moving average filtering to remove abnormal outliers caused by electromagnetic interference and signal fluctuations, while retaining the true change pattern of the current data in the time dimension, thus ensuring the accuracy of the data source for feature extraction. Then, the preprocessed current timing data is segmented according to a preset fixed time window, with 10 seconds as a basic time window unit. Within each time window, the maximum, minimum and average values ​​of the current amplitude, the decay rate of the current in the steady state, the inherent fluctuation frequency of the current signal, the peak value and duration of the current transient change when the circuit is abnormal, and the current step change value and duration after the lamp bead is damaged are calculated and extracted to completely extract five types of core current timing features. Finally, the five extracted core features are normalized and standardized to eliminate the dimensional differences between different feature dimensions, map all feature values ​​to a unified numerical range, and form a standardized time-series feature dataset. At the same time, each feature dataset is bound to a corresponding terminal node identifier and data collection timestamp to provide standardized input for subsequent diagnostic model inference.

[0028] Diagnostic module: Based on AI models, it completes fault identification, counts the number of faulty LEDs, and determines the fault level; The diagnostic reasoning process integrating the AI ​​model is as follows: First, a standardized time-series feature dataset is input into a one-dimensional CNN network. Through convolutional layers and pooling layers, deep feature extraction is performed on the time-series data to capture the fine-grained fault features hidden in the current time-series data. Then, the deep feature vector output by the one-dimensional CNN network is input into a trained random forest classifier. The random forest classifier completes the accurate identification of fault types and the quantitative statistics of the number of faulty LEDs. At the same time, according to the scope and severity of the fault, the fault is divided into two categories: ordinary faults and high-risk faults. Ordinary faults include single / a few damaged LEDs and slight aging of LED strips. High-risk faults include multiple LEDs damaged in a concentrated manner, short circuits, and overcurrent in the circuit. The multi-level false alarm prevention mechanism is implemented as follows: the fault diagnosis result output by the model in a single instance is temporarily set as the preliminary judgment result. The main control MCU unit continuously collects and extracts the current timing features within multiple consecutive time windows, and inputs them sequentially into the fused AI model for diagnostic reasoning. Only when the preliminary judgment results of multiple diagnoses are completely consistent will the result be recognized as the final valid fault diagnosis result. If any judgment result is inconsistent, it is judged as a false judgment caused by interference signals, and the fault alarm is canceled to ensure the authenticity and reliability of the fault diagnosis result.

[0029] Aging and Lifespan Prediction Module: Fits aging curves, analyzes aging trends, and predicts remaining lifespan; Specifically, the process involves first establishing a basic aging characteristic database for LED light strips, storing standard current decay benchmark data for different models and specifications of LED light strips, and simultaneously collecting real-time current time-series data for the entire lifecycle of each monitoring terminal's corresponding LED light strip from its new state to the operational stage. The steady-state current decay rate characteristics of different operational stages are extracted, and an aging curve specific to each light strip is constructed using a non-linear fitting method with operating time as the horizontal axis and steady-state current decay rate as the vertical axis. As the operating time of the light strip increases, new current decay data is continuously added to dynamically correct and optimize the aging curve, ensuring a high degree of match between the curve and the actual aging state of the light strip. Then, the real-time extracted steady-state current decay rate of the light strip is compared with the aging curve to determine the current aging stage of the light strip. Combined with the overall decay trend of the aging curve, the remaining operating time from the current state to the failure threshold is calculated, achieving accurate prediction of the remaining service life. It should be noted that this module supports the construction of personalized aging curves for LED strips of different brands, power, and number of LEDs. For the parameters of the LED strip bound to each monitoring terminal, it automatically matches the corresponding basic aging characteristic data. At the same time, it synchronizes the aging stage and remaining service life prediction results to the operation and maintenance platform at a preset cycle, providing accurate data support for operation and maintenance personnel to formulate predictive maintenance plans. Moreover, the prediction results will be dynamically updated according to the actual operating status of the LED strip, ensuring the timeliness and accuracy of the prediction.

[0030] The relay fail-safe control module, based on the fault determination results from the AI ​​intelligent diagnostic module, enables fault-safe isolation and authorized reset operations for the LED light strip. This module includes: Fault level matching module: classifies ordinary / high-risk faults and matches control actions, and adapts the matching logic to hardware-level configurable safety policies; Common faults include damage to a single or a small number of LED beads, slight aging of LED strips, and faults that do not affect the overall lighting performance or circuit safety. These types of faults only require the main control MCU unit to trigger a text warning on the operation and maintenance platform and a lightweight prompt on the terminal. No circuit isolation action is required. The warning information will be archived by binding the fault node identifier, fault type, and collection timestamp. High-risk faults include those caused by the concentrated damage of multiple LEDs, short circuits, overcurrent in the circuit, and LED strips aging to a critical state, which may lead to circuit burnout, electric shock, and other safety accidents. These types of faults require the main control MCU unit to immediately trigger a circuit power-off isolation action. The matching logic between the fault level and the control action will dynamically adapt to the two working modes of the hardware-configurable safety strategy. The hardware-level configuration is achieved through a physical isolation mechanism connected in series with the relay drive circuit inside the terminal. This physical isolation mechanism can be selected from any one or more combinations of DIP switches, jumper caps, zero-ohm resistance solder joints, physical pluggable connectors, and fuses, and is equipped with an anti-tampering sealing structure. Once the mode is locked, it cannot be modified in a non-destructive way. Unsealing will leave a trace. In the disconnected state, the main control MCU unit cannot output drive signals to the relay at all, achieving forced isolation at the physical level. This ensures that there is no possibility of automatic disconnection in the pure monitoring mode, and the on / off state can be directly identified visually without the need for power-on detection, adapting to the safety management requirements of different scenarios such as civil aviation, tunnels, and municipal administration.

[0031] Safety isolation module: Used for circuit power-off isolation in case of high-risk faults; the relay adopts a fault pass-through hardware design. First, after the main control MCU receives the final valid high-risk fault diagnosis result issued by the AI ​​intelligent diagnosis module, it immediately sends a high-level isolation trigger command to the drive circuit of the relay fault safety control module. At the same time, the command is encrypted and locked to prevent repeated triggering or false triggering caused by electromagnetic interference. Then, after receiving a valid trigger command, the relay drive circuit disconnects the relay contacts connected in series in the main power supply circuit of the LED light strip, realizing rapid power-off isolation of the fault circuit of the LED light strip. The relay adopts a fault-pass hardware design, which is normally closed to ensure the normal power supply of the light strip. It always remains normally closed when there is no electrical signal input. Even if the main control MCU or drive circuit has a hardware failure, the relay will not disconnect on its own, avoiding false isolation without command that affects the normal operation of the light strip. Finally, after the relay completes the isolation action, the module's built-in status detection circuit will collect the actual on / off status of the relay and the power supply status of the light strip circuit in real time, and feed back the electrical signal of isolation completion to the main control MCU. The main control MCU will then encrypt and upload information such as fault isolation status, isolation trigger time, and corresponding monitoring node identifier to the operation and maintenance platform, and trigger a dual audible and visual alarm for high-risk faults on the operation and maintenance platform and the field terminal, while continuously monitoring the status of the fault circuit in real time.

[0032] Manual authorized reset module: Performs remote authorized reset of relays via the operation and maintenance platform; Specifically, after the operation and maintenance platform receives the fault handling completion confirmation information uploaded by the on-site operation and maintenance personnel, it initiates a dedicated manual authorization process for remote relay reset. First, the operation and maintenance personnel who initiate the reset operation undergo three levels of identity authentication, namely, platform account password verification, job operation permission verification, and mobile phone SMS dynamic verification code secondary verification. Only legitimate operation and maintenance personnel who pass all authentications can enter the dedicated reset operation interface. At the same time, the operation and maintenance platform will automatically retrieve the historical fault information, isolation trigger time, and fault handling records of the fault node for automated secondary verification. Only after confirming that the fault has been effectively handled and the circuit reset power supply conditions are met will the permission to issue the reset command be granted. If the verification fails, the reset operation function will be directly blocked. After the maintenance personnel confirm and issue the reset command on the platform, the platform generates an encrypted reset command data packet containing the unique physical identifier of the faulty node, the reset execution command, the operator's identity information, and the command generation timestamp. This data packet is then transmitted via a cellular mesh self-organizing network transmission module using encrypted multi-hop relay to the main control MCU unit of the corresponding monitoring terminal. Upon receiving the reset command data packet, the main control MCU first performs integrity verification and decryption verification on the data packet. After successful verification, it sends a reset trigger low-level command to the relay drive circuit. Upon receiving the command, the drive circuit de-energizes the electromagnetic coil of the electromagnetic relay, causing the relay contacts to close again and restoring the normal power supply to the LED light strip circuit. The built-in status detection circuit of the safety isolation module collects the on / off status of the relay after reset and the power supply recovery status of the light strip circuit in real time. It feeds back the electrical signal of the reset completion to the main control MCU. The main control MCU then encrypts and uploads data such as the reset completion status, reset execution time, and operator information to the operation and maintenance platform for recording and archiving. After the reset, the main control MCU will automatically increase the current sampling frequency of the light strip circuit to the highest level and continuously monitor the circuit operation status for a certain period of time. If no fault characteristics are detected, the normal sampling frequency will be restored. If high-risk fault characteristics are detected again, the power-off isolation will be triggered again and a secondary fault alarm will be sent to the operation and maintenance platform to realize closed-loop monitoring after fault reset.

[0033] A dual-redundant intelligent power supply module provides power to all modules of the system, enabling intelligent dynamic switching of power supply modes, dynamic adaptation to low power consumption, and capacitor freewheeling protection during power supply switching. This module includes: Power supply switching module: Based on light intensity, external DC power input status, and remaining lithium battery power, it intelligently and dynamically switches between three modes: solar power supply, external DC power supply, and lithium battery backup power supply, with external DC power supply being the primary mode. First, the module's built-in voltage detection chip collects the input voltage value of the external DC power supply in real time, the light sensor collects the ambient light intensity value, and the lithium battery power detection chip accurately collects the remaining power percentage and cell voltage of the lithium battery. All collected data is transmitted to the main control MCU unit in real time for data parsing and status determination. Then, the main control MCU unit selects the power supply mode according to the preset judgment threshold. If the input voltage of the external DC power supply is within the rated working range, it is directly judged to be in the external DC power supply mode, and the external DC power supply powers the entire system module and simultaneously float-charges the lithium battery. If the input of the external DC power supply is abnormal, it is judged whether the light intensity reaches the solar power supply threshold. If it does, it switches to the solar power supply mode, and the solar panel converts light energy into electrical energy to power the system and simultaneously charges the lithium battery. If it does not reach the threshold, it directly switches to the lithium battery backup power supply mode. Finally, the main control MCU unit stores the real-time power supply mode status and the working parameters of each power supply unit in the local cache and uploads them to the operation and maintenance platform at the same time. It also continuously monitors the status changes of each power supply unit. When a change in the trigger condition is detected, it immediately performs a seamless switch of the power supply mode. The switching command is sent to the power supply switching execution circuit through the standardized SPI interface to ensure the timeliness and intelligence of the mode switching.

[0034] Low-power dynamic adaptation module: dynamically adjusts the current sampling frequency and data reporting cycle according to the remaining lithium battery power; The current sampling frequency is divided into three levels of adaptation standards according to the remaining lithium battery power. For example, when the remaining lithium battery power is ≥50%, the default high-frequency sampling of 1 second / time is maintained; when the remaining lithium battery power is ≤20% and <50%, the sampling frequency is adjusted to 5 seconds / time; and when the remaining lithium battery power is <20%, the sampling frequency is switched to 10 seconds / time. All sampling frequency adjustment commands are sent to the data acquisition module in real time by the main control MCU unit to ensure that the sampling frequency matches the power supply capacity. The data reporting cycle is simultaneously adjusted in three levels according to the remaining lithium battery power. For example, when the remaining lithium battery power is ≥50%, the default real-time reporting frequency of 5 seconds / report is maintained; when the remaining lithium battery power is ≤20% and <50%, the interval reporting is adjusted to 20 seconds / report; and when the remaining lithium battery power is <20%, the frequency reporting is switched to 30 seconds / report. The reporting cycle adjustment command is issued by the main control MCU unit to the cellular Mesh self-organizing network transmission module. At the same time, when the remaining lithium battery power recovers to more than 50% and stabilizes for 5 minutes, it automatically switches back to the default sampling and reporting parameters, realizing a dynamic balance between battery life and real-time monitoring.

[0035] Capacitor freewheeling protection module: Provides freewheeling power during power supply mode switching through a large-capacity electrolytic capacitor; Specifically, a large-capacity electrolytic capacitor adapted to the system's operating voltage is connected in parallel at the common power output terminal of the dual-redundant intelligent power supply module. This capacitor is an industrial-grade electrolytic capacitor with high capacitance and low leakage current, and its capacity meets the millisecond-level continuous power supply requirements of each module in the system. During the normal power supply phase of the system, the capacitor is always in a fully charged floating charge state, continuously storing electrical energy for backup. When the main control MCU unit issues a power supply mode switching command, and the power supply circuit of the original power supply unit is disconnected while the power supply circuit of the new power supply unit has not yet been fully connected, the electrolytic capacitor immediately releases the stored electrical energy to provide a continuous and stable working power supply for the core modules of the system, such as the main control MCU, data acquisition, and wireless transmission. The follow current time completely covers the entire gap process of power supply switching, ensuring that there is no power interruption, no data loss, and no equipment disconnection during the switching process. Moreover, the capacitor quickly resumes charging after the new power supply unit is connected, preparing for the next switch, adapting to seamless switching scenarios between all power supply modes.

[0036] This integrated waterproof and easy-to-install module provides the monitoring terminal with comprehensive adaptability to complex outdoor environments. It integrates waterproof and flame-retardant protection with drill-free, quick-installation and disassembly capabilities. Its modular design, integrated with the terminal, adapts to various harsh operating conditions such as high humidity, oil contamination, dust, and strong vibration, while also meeting the rapid installation requirements of various carriers including light poles and tunnel walls. The module includes: Waterproof protection module: It adopts an integrated waterproof and flame-retardant shell and is equipped with a sealing structure with a corresponding protection level; Specifically, the terminal shell is made of flame-retardant ABS+PC composite engineering plastic, precision-molded into a single piece without any seams. The overall protection level of the shell is designed to be IP67, meeting the requirements for outdoor use, including immersion in 1 meter of water for 30 minutes without water ingress and complete dust protection in dusty environments. Customized EPDM rubber sealing gaskets are used at all interfaces and mechanical openings. These gaskets undergo special treatment to resist high and low temperatures, aging, and oil stains, and precisely fit into the pre-drilled slots on the shell for a seamless seal. The terminal's LoRa RF antenna and power supply interface both adopt a waterproof aviation plug design, with a built-in double-layer silicone sealing ring. The screw-on connection structure achieves secondary sealing at the interface. The clamp-shaped opening and closing port of the Hall sampling module is equipped with an elastic waterproof silicone sleeve. After the buckle is engaged, the silicone sleeve tightly wraps the opening and closing gap to prevent moisture and dust from entering. In addition, all circuit boards and electronic component solder joints inside the terminal are professionally coated with conformal coating. The conformal coating forms a dense protective film, achieving triple protection against moisture, mildew, and salt spray. Moisture-absorbing cotton is also installed inside the shell to further absorb trace amounts of internal moisture, ensuring that the circuit system of each internal functional module can operate stably without short circuits, open circuits, or other faults in complex environments such as wide temperature ranges, high humidity tunnels, and outdoor coastal salt spray.

[0037] Clip-on installation module: Equipped with an anti-drop clip-on installation base, which is compatible with conventional installation carriers such as light poles and tunnel walls; Specifically, the mounting base is made of high-strength aluminum alloy through die casting, and undergoes anodizing treatment to improve its corrosion resistance, wear resistance, and vibration resistance. The base is designed as a split-type combination structure of a fixed bracket and a movable buckle. The fixed bracket can be pre-fixed to mounting carriers such as light poles, tunnel walls, and angle steel brackets using expansion screws or strong industrial adhesive. The surface of the bracket has anti-slip serrations and positioning grooves to effectively prevent the base from sliding or shifting. The movable buckle and the monitoring terminal shell are integrated injection-molded structures. Anti-slip rubber pads are pasted on the inside of the buckle, and positioning protrusions that match the positioning grooves of the fixed bracket are provided. During terminal installation, simply align the movable buckle with the positioning groove of the fixed bracket and snap it down. The buckle's built-in spring-loaded self-locking tongue will... The automatic spring-loaded locking slot ensures a rigid fixation between the terminal and the base, resisting external forces such as strong outdoor winds and equipment vibrations, effectively preventing the terminal from falling off. The movable buckle is equipped with a dedicated unlocking wrench hole, allowing the terminal to be disassembled only by prying open the self-locking tongue with the included dedicated unlocking wrench. This prevents non-professionals from disassembling the terminal, ensuring its safety after installation. Meanwhile, the base is equipped with multiple sizes of replaceable buckle bushings, which can be adapted to different types of installation carriers such as round light poles, flat tunnel walls, and irregularly shaped metal brackets. No custom-made base is required, enabling universal installation in all scenarios. Furthermore, the entire installation and disassembly process requires no drilling or wiring and can be completed by a single person, greatly improving the convenience of on-site construction and subsequent maintenance.

[0038] An LED strip monitoring terminal based on LoRa cellular dual-power supply and AI diagnostics includes: the terminal is an integrated modular hardware, integrating a main control MCU unit, a data acquisition module, a dual-redundant intelligent power supply module, a cellular mesh self-organizing network transmission module, an AI intelligent diagnostic module, a relay fault safety control module, and an integrated waterproof protection and convenient installation module; each functional module is electrically connected to the main control MCU unit through a standardized SPI interface, and the main control MCU unit coordinates the working logic linkage and data interaction of each module.

[0039] A method for monitoring LED light strips based on LoRa colony dual-power AI diagnostics includes: S1. The data acquisition module uses a clamp-on Hall sensor to perform non-contact current acquisition of the LED light strip circuit. First, it automatically completes the zero-point self-calibration of the current sampling, then collects the sensor's working environment temperature in real time and performs temperature drift adaptive compensation on the current acquisition data to obtain the calibrated LED light strip circuit current data. S2. The bee colony mesh self-organizing network transmission module constructs a decentralized self-organizing network architecture with all nodes peer-to-peer. It transmits the calibrated current data obtained in step S1 to the outside through a multi-hop relay method. When communication is interrupted, the current data is stored locally. After communication is restored, the stored data is retransmitted in batches according to the timestamp. The faulty node is determined by the node heartbeat packet and the network routing is autonomously reconstructed based on the greedy algorithm. The S3 AI intelligent diagnostic module extracts current amplitude, steady-state decay rate, fluctuation frequency, transient change characteristics, and current step characteristics after lamp bead damage from the transmitted current data. Based on a one-dimensional CNN + random forest fusion AI model, it completes lamp strip fault identification, counts the number of faulty lamp beads, and determines the level of ordinary and high-risk faults. The determination process adopts a multi-level anti-false alarm mechanism with consistent results from multiple consecutive acquisitions. At the same time, it fits the aging curve based on the lamp strip current decay trend, analyzes the lamp strip aging trend, and predicts the remaining service life. S4. The relay fault safety control module matches the fault level judgment result of the AI ​​intelligent diagnosis module with the control action. For high-risk faults, the relay is triggered to perform power-off isolation of the LED light strip circuit. The relay hardware adopts a fault pass-through design. After the fault is handled, the remote reset operation of the relay is performed by manual authorization through the operation and maintenance platform.

[0040] To illustrate the applicable scenarios of this invention, the following scenarios are provided as examples: Highway / railway tunnel lighting scenes: In response to the characteristics of long distances, multiple obstructions, lack of light, and difficult wiring in highway / railway tunnels, the monitoring terminal of this invention is deployed according to a general core implementation method. It is installed on the tunnel wall through a snap-on base, with an overall protection level of not less than IP67. It adopts an external DC priority power supply mode, utilizing the tunnel's original DC power supply system, while a lithium battery is used as a backup. The terminal automatically builds a cellular mesh network, achieving full coverage communication in long-distance tunnels through multi-hop relays, eliminating the need for a central gateway and solving the problems of difficult wiring and communication obstruction in tunnels; the AI ​​diagnostic module identifies tunnel light strip faults in real time, and automatically cuts off the power supply to the circuit for short circuit and overcurrent faults to prevent electrical fires in the tunnel. Fault information is pushed to tunnel maintenance personnel via mobile devices. Combined with the terminal's unique physical ID, the fault can be accurately located and responded to quickly, replacing manual inspections and optimizing the tunnel lighting maintenance system.

[0041] Municipal street light scene: In response to the characteristics of municipal streetlights being outdoors without mains power, widely distributed, and difficult to detect single-point faults, the monitoring terminal of this invention is deployed according to a general core implementation method. It is installed on the streetlight pole through a snap-on base, with an overall protection level of not less than IP65. It adopts a solar main power supply mode with a lithium battery as backup, eliminating the need for an external power source and achieving wiring-free deployment. The terminal automatically forms a cellular mesh network to cover the entire municipal area, eliminating the need for a central gateway and solving the problems of scattered distribution and high wiring costs of municipal streetlights. The AI ​​diagnostic module identifies street light faults and aging trends in real time, and performs predictive maintenance based on current decay curves to prevent street lights from going out at night and affecting traffic, while reducing the frequency of manual inspections.

[0042] General application scenarios for airport non-flying area auxiliary lighting: For airport non-flying area ancillary lighting scenarios, including terminal building indoor and outdoor lighting, cargo area lighting, airport perimeter lighting, parking lot lighting, business jet building ancillary lighting and other non-civil aviation strongly regulated areas, such scenarios are characterized by wide distribution, difficulty in wiring without mains power, and great difficulty in inspection coverage. The monitoring terminal is attached to the power supply bus of the corresponding lighting circuit through the opening and closing Hall effect card, which can be deployed without power interruption or wire breaking. The terminal adopts a dual-redundant intelligent power supply module with solar + DC dual power supply mode, which is adapted to outdoor lighting conditions and indoor no-light scenarios, ensuring that the equipment operates 24 / 7 without interruption. The terminal automatically builds a cellular mesh network without a central gateway, avoiding single point of failure from affecting the entire network operation and solving the problems of high cabling costs and difficult communication coverage in large areas of airports. The AI ​​diagnostic module can accurately identify lighting circuit faults in real time. For short circuit and overcurrent faults, it can quickly cut off the power supply to the corresponding circuit to prevent the fault from escalating and causing electrical fires and other safety accidents. After the fault is resolved, maintenance personnel can remotely reset the relay on the maintenance center platform without on-site operation, shortening the fault recovery time and optimizing the manual inspection process.

[0043] The applicable scenarios for monitoring core navigational aid lights in the flight area of ​​civil airports (runways, taxiways, parking ramps, takeoff and landing strips, and other areas with high civil aviation regulatory oversight): Each set of core navigation light LED strips in the flight area is equipped with one intelligent monitoring terminal as described in this embodiment. The terminal adopts a clamp-on Hall current sampling module, which is directly connected to the power supply circuit bus of the navigation light LED strip. There is no need to disconnect the power, cut the wire, or lay the wiring. It is fixed to the light pole or equipment bracket by a snap-on base that prevents it from falling off, which meets the relevant safety requirements for construction without interrupting operation at transport airports. The overall protection level of the terminal is not lower than IP67, which is suitable for the harsh operating environment of the flight area, such as de-icing fluid spray, high-pressure cleaning, and high humidity oil pollution. The system adopts a "wireless terminal + fixed redundant primary and backup gateway" architecture, disables the decentralized self-networking function of the terminals, and installs the gateway in an unobstructed area of ​​the airport lighting station or tower. All terminals are uniformly allocated a 433MHz civil aviation approved frequency band, ≤14dBm transmission power and fixed communication route. All data is connected to the airport navigation lighting monitoring network through the gateway, realizing full-process management, controllability, traceability and shutdown, meeting all requirements of relevant civil airport electromagnetic environment testing specifications, and eliminating the risk of harmful interference from aviation radio. The terminal eliminates the solar power unit and adopts a dual redundant power supply architecture of airport isolated DC auxiliary power supply + industrial-grade lithium battery backup. It is connected to the airport's original isolated auxiliary power supply system, which meets the mandatory specifications of no glare, no protruding foreign objects, and high wind resistance in the flight area. When there is no external power supply, it automatically switches to lithium battery backup power supply. During the switching process, capacitor current is used to ensure no data loss and no equipment disconnection. The physical isolation mechanism of the relay drive circuit is permanently disconnected and tamper-proof lead seals are applied, locking the equipment in pure monitoring mode; the AI ​​diagnostic module collects real-time current timing data of the navigation light circuit, completes fault identification, counts the number of faulty LEDs, analyzes aging trends and predicts remaining lifespan, and only outputs alarm signals to the airport operation and maintenance platform without performing automatic disconnection actions; for high-risk faults such as short circuits and overcurrents, only emergency alarms are pushed, fully guaranteeing the airport's exclusive operational control over the navigation light system, which meets the requirements for civil aviation special equipment certification; Terminal nodes are bound to a unique physical ID and the installation location of navigation lights to achieve precise fault location; data is cached locally when the network is interrupted and automatically retransmitted after communication is restored; all data is retained locally for no less than one year, which complies with the relevant requirements of civil aviation safety audit and accident tracing.

[0044] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention are also within the protection scope of the present invention.

Claims

1. An LED strip monitoring system based on LoRa colony dual-power AI diagnostics, characterized in that, include: Data acquisition module, bee colony Mesh self-organizing network transmission module, AI intelligent diagnosis module, relay fault safety control module; The data acquisition module performs non-contact current sampling on the LED light strip circuit to obtain circuit current data; The cellular mesh self-organizing network transmission module is based on LoRa to build a decentralized cellular mesh self-organizing network, realizing multi-hop transmission of current data and network self-healing. The AI ​​intelligent diagnostic module extracts current timing features to complete fault diagnosis and lifespan prediction of LED light strips; The relay fault safety control module performs fault safety isolation and authorized reset operations on the LED light strip based on the fault judgment results of the AI ​​intelligent diagnostic module.

2. The LED strip monitoring system based on LoRa colony dual-power AI diagnostics according to claim 1, characterized in that: The data acquisition module includes: a Hall sampling module, a zero-point self-calibration module, and a temperature drift adaptive compensation module; Hall sampling module: Employs a clamp-on Hall sensor to perform non-contact current acquisition; Zero-point self-calibration module: Performs zero-point self-calibration of current sampling to eliminate sensor factory deviation; Temperature drift adaptive compensation module: performs dynamic temperature drift compensation on current data based on ambient temperature.

3. The LED strip monitoring system based on LoRa colony dual-power AI diagnostics according to claim 1, characterized in that: The bee colony Mesh self-organizing network transmission module includes: a decentralized networking module, a route self-reconstruction module, and a local caching and retransmission module; Decentralized networking module: Constructs a decentralized networking architecture with all nodes peering; The self-reconstruction routing module identifies faulty nodes based on node heartbeat packets and completes autonomous network routing reconstruction. Local caching and retransmission module: Used to store data locally when communication is interrupted, and retransmit it in batches according to timestamps after communication is restored.

4. The LED strip monitoring system based on LoRa colony dual-power AI diagnostics according to claim 1, characterized in that: The AI ​​intelligent diagnostic module includes: a time-series feature extraction module, a diagnostic module, and an aging and lifespan prediction module; Timing feature extraction module: extracts current amplitude, steady-state decay rate, fluctuation frequency, transient changes, and step current characteristics of LED damage. Diagnostic module: Based on AI models, it completes fault identification, counts the number of faulty LEDs, and determines the fault level; Aging and Lifespan Prediction Module: Fits aging curves, analyzes aging trends, and predicts remaining lifespan.

5. The LED strip monitoring system based on LoRa colony dual-power AI diagnostics according to claim 1, characterized in that: The relay fault safety control module includes: a fault level matching module, a safety isolation module, and a manual authorized reset module; Fault level matching module: classifies ordinary / high-risk faults and matches control actions, and adapts the matching logic to hardware-level configurable safety policies; Safety isolation module: Used for circuit power-off isolation in case of high-risk faults; the relay adopts a fault pass-through hardware design. Manually authorized reset module: Performs remote authorized reset of relays through the operation and maintenance platform.

6. The LED strip monitoring system based on LoRa colony dual power supply AI diagnostics according to claim 1 further includes a dual redundant intelligent power supply module and an integrated waterproof protection and convenient installation module.

7. The LED strip monitoring system based on LoRa colony dual-power AI diagnostics according to claim 6, characterized in that, The dual-redundant intelligent power supply module includes: a power supply switching module, a low-power dynamic adaptation module, and a capacitor freewheeling current protection module. Power supply switching module: Based on light intensity, external DC power input status, and remaining lithium battery power, it intelligently and dynamically switches between three modes: solar power supply, external DC power supply, and lithium battery backup power supply, with external DC power supply being the primary mode. Low-power dynamic adaptation module: dynamically adjusts the current sampling frequency and data reporting cycle according to the remaining lithium battery power; Capacitor freewheeling protection module: Provides freewheeling power during power supply mode switching through a large-capacity electrolytic capacitor.

8. The LED strip monitoring system based on LoRa colony dual-power AI diagnostics according to claim 6, characterized in that, The integrated waterproof protection and convenient installation module includes: a waterproof protection module and a snap-on installation module; Waterproof protection module: It adopts an integrated waterproof and flame-retardant shell and is equipped with a sealing structure with a corresponding protection level; Clip-on installation module: Equipped with a clip-on installation base to prevent detachment, the base is compatible with conventional installation carriers such as light poles and tunnel walls.

9. A monitoring terminal for LED light strips based on LoRa colony dual-power AI diagnostics, characterized in that, include: The terminal is an integrated modular hardware, which integrates a main control MCU unit, a data acquisition module, a dual redundant intelligent power supply module, a cellular Mesh self-organizing network transmission module, an AI intelligent diagnostic module, a relay fault safety control module, and an integrated waterproof protection and convenient installation module. Each functional module is electrically connected to the main control MCU unit through a standardized SPI interface, and the main control MCU unit coordinates the working logic linkage and data interaction of each module.

10. A method for monitoring LED light strips based on LoRa colony dual-power AI diagnostics, characterized in that, Includes the following steps: S1. The data acquisition module uses a clamp-on Hall sensor to perform non-contact current acquisition of the LED light strip circuit. First, it automatically completes the zero-point self-calibration of the current sampling, then collects the sensor's working environment temperature in real time and performs temperature drift adaptive compensation on the current acquisition data to obtain the calibrated LED light strip circuit current data. S2. The bee colony mesh self-organizing network transmission module constructs a decentralized self-organizing network architecture with all nodes peer-to-peer. It transmits the calibrated current data obtained in step S1 to the outside through a multi-hop relay method. When communication is interrupted, the current data is stored locally. After communication is restored, the stored data is retransmitted in batches according to the timestamp. The faulty node is determined by the node heartbeat packet and the network routing is autonomously reconstructed based on the greedy algorithm. The S3 AI intelligent diagnostic module extracts current amplitude, steady-state decay rate, fluctuation frequency, transient change characteristics, and current step characteristics after lamp bead damage from the transmitted current data. Based on a one-dimensional CNN + random forest fusion AI model, it completes lamp strip fault identification, counts the number of faulty lamp beads, and determines the level of ordinary and high-risk faults. The determination process adopts a multi-level anti-false alarm mechanism with consistent results from multiple consecutive acquisitions. At the same time, it fits the aging curve based on the lamp strip current decay trend, analyzes the lamp strip aging trend, and predicts the remaining service life. S4. The relay fault safety control module matches the fault level judgment result of the AI ​​intelligent diagnosis module with the control action. For high-risk faults, the relay is triggered to perform power-off isolation of the LED light strip circuit. The relay hardware adopts a fault pass-through design. After the fault is handled, the remote reset operation of the relay is performed by manual authorization through the operation and maintenance platform.

Images