Barrier gate remote monitoring system and method based on Internet of Things

Abstract

The invention relates to the technical field of intelligent control, in particular to a barrier gate remote monitoring system and method based on the Internet of Things, and the method comprises the steps: building a dual-channel communication link through a 4G module and a Bluetooth / WiFi module, achieving the synchronous collection and uploading of the current of a barrier gate motor, a position sensor and environment data, deploying an LSTM model at an edge calculation layer, and carrying out the remote monitoring of a barrier gate. Time sequence prediction and abnormal deviation quantification are carried out on the operation state, video acquisition is triggered, vehicle retention and rod body deformation are identified in combination with a lightweight YOLOv5 model, abnormal event grading judgment is realized, spare part inventory is associated according to event types, and a maintenance scheme is dynamically generated. An intelligent work order including a thermodynamic diagram, a video frame and a prediction curve is constructed by using a three-dimensional visualization template, evidence storage processing is completed through a block chain module, and finally optimal personnel scheduling and order dispatching are executed according to a maintenance scheme. According to the method, multi-source data fusion modeling, visual linkage anomaly recognition and full-link credible operation and maintenance are realized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and in particular to a barrier gate remote monitoring system and method based on the Internet of Things. Background Art

[0002] With the continuous development of smart transportation and urban management systems, barrier gates, as a crucial component of traffic access control, have been widely used in scenarios such as parking management, campus access control, and urban road management. Modern barrier gate systems typically incorporate a variety of sensors, control circuits, and communication modules to implement functions such as vehicle detection, traffic control, and operational monitoring. To enhance operational reliability and remote maintenance capabilities, an increasing number of barrier gates are being connected to IoT platforms. Leveraging wireless communication and edge computing, these devices enable preliminary data collection and status analysis, providing the foundation for unmanned operation and intelligent maintenance.

[0003] However, existing barrier gate monitoring systems still face several key technical bottlenecks. For one thing, most systems rely solely on a single-path data upload mechanism (such as a 4G link), which is susceptible to communication interruptions in complex environments, making it difficult to ensure data integrity and video continuity. Furthermore, existing anomaly detection methods generally rely on single-variable threshold judgments and lack the ability to perform time series modeling and joint analysis on multi-source data (such as current, angle, temperature and humidity), resulting in delayed anomaly identification and high false alarm rates. Furthermore, systems often lack visual representation of faults and reliable evidence storage mechanisms, preventing them from forming a closed-loop response chain covering "perception-analysis-determination-action-recording," hindering their widespread application in critical security and high-frequency traffic scenarios. Summary of the Invention

[0004] Based on the above objectives, the present invention provides a barrier gate remote monitoring system and method based on the Internet of Things.

[0005] A remote monitoring method for a gate based on the Internet of Things includes the following steps: S1, dual-mode communication link establishment: A parallel communication channel is established through the 4G module and the Bluetooth / WiFi module to collect the gate motor current, position sensor data and environmental temperature and humidity parameters in real time to form a status monitoring data set; S2, edge computing layer anomaly detection: The state monitoring data set is input into the LSTM prediction model deployed in the embedded function box to generate a predicted value of the gate operation state. If the deviation between the measured value and the predicted value exceeds a first threshold, it is marked as a level 1 abnormal event; S3, hierarchical response processing: If a level 1 abnormal event is detected, the local camera is activated through the Bluetooth / WiFi module to capture multi-angle video, and the lightweight YOLOv5 model of the edge computing unit is simultaneously called for real-time analysis. If a vehicle is stranded or a pole is deformed, it is upgraded to a level 2 abnormal event; S4, intelligent generation of maintenance work orders: Automatically associate spare parts inventory data according to the level of abnormal events, generate a three-dimensional visual work order including fault location map, recommended maintenance plan and optimal personnel scheduling, and upload it to the cloud operation and maintenance platform through the blockchain evidence module.

[0006] Optionally, the S1 includes: S11, initialization configuration: start the 4G module and Bluetooth / WiFi module at the same time, configure the heartbeat packet sending interval of the 4G module to 10 seconds, the transmission buffer size of the Bluetooth / WiFi module to 512 bytes, and verify the connection status of the two modules with the cloud operation and maintenance platform through the handshake protocol.

[0007] S12, dynamic establishment of dual-mode communication link: Based on the initialized 4G module and Bluetooth / WiFi module, two parallel communication channels are established and channel priority strategies are set.

[0008] S13, synchronous acquisition of multi-source sensors: The analog-to-digital converter connected to the Bluetooth / WiFi module synchronously acquires raw data, including the three-phase current value of the gate motor, the pulse signal of the position sensor of the gate pole, and the temperature and humidity values ​​output by the ambient temperature and humidity sensor.

[0009] S14, data preprocessing and verification: preprocess and verify the original data to generate a preprocessed data set; S15, generating a state monitoring data set: uploading the pre-processed data set to the embedded database of the edge computing layer in parallel through the dual channels of the 4G module and the Bluetooth / WiFi module to form a state monitoring data set.

[0010] Optionally, the S2 includes: S21, preprocessing of the state monitoring data set: reading the state monitoring data set from the embedded database, and generating a standardized time series data set by normalizing the current characteristics and performing time series sliding window segmentation; S22, dynamic loading of LSTM prediction model: loading the pre-trained LSTM prediction model through the hardware accelerator of the embedded function box, performing model structure initialization and digital signature verification to implement model deployment; S23, multivariate time series prediction execution: input the standardized time series data set into the LSTM prediction model, and generate a prediction result set through forward reasoning and inverse transformation processing.

[0011] Optionally, the S2 further includes: S24, abnormal deviation quantification calculation: extract the measured value from the condition monitoring data set, and calculate the real-time deviation between the measured value and the predicted result. If the deviation between the measured value and the predicted result exceeds a first threshold, it is marked as a level 1 abnormal event; S25, abnormal event feature association: For the data segment marked as a first-level abnormal event, an abnormal event report is generated through abnormal type identification and environmental disturbance calculation.

[0012] Optionally, the S3 includes: S31, abnormal event triggers video acquisition: When the first-level abnormal event is detected, a start instruction is sent to the local camera array through the Bluetooth / WiFi module, multi-angle image acquisition is performed to generate abnormality-related video streams, and the video streams are stored in the edge storage pool.

[0013] S32, lightweight YOLOv5 model loading: loading a preset lightweight YOLOv5 model through the edge computing unit, and implementing model deployment through model structure tailoring, quantization configuration, and hardware binding; S33, real-time video analysis execution: read the video stream data from the edge storage pool, input it into the lightweight YOLOv5 model for frame-by-frame recognition and state inference, and generate a set of detection results.

[0014] Optionally, the S3 further includes: S34, Level 2 abnormal event upgrade determination: performing cross-frame correlation analysis on the detection result set, determining upgrade conditions based on event feature combinations, and generating a Level 2 abnormal event report; S35, pre-triggering work order generation: fusing the secondary abnormal event report with the primary abnormal event report, and generating a three-dimensional visualization template through video-sensor data alignment.

[0015] Optionally, the S4 includes: S41, intelligent association of spare parts inventory: By accessing the spare parts inventory database of the cloud operation and maintenance platform, performing exception type analysis and inventory query, and generating a spare parts matching result set; S42, dynamic maintenance plan generation: by analyzing the fault location map and combining the spare parts matching result set, performing maintenance strategy matching and man-hour estimation, and generating a recommended maintenance plan document; S43, 3D visualization work order construction: by calling the 3D visualization template, executing data fusion and interactive control construction, a 3D visualization work order is generated.

[0016] Optionally, the S4 further includes: S44, blockchain evidence processing: performing hash calculation and contract writing through the blockchain evidence module to generate an evidence record; S45, optimal personnel scheduling execution: performing maintenance personnel screening and route optimization through the recommended maintenance plan document, and generating a scheduling instruction package.

[0017] A remote monitoring system for a gate based on the Internet of Things is used to implement the above-mentioned remote monitoring method for a gate based on the Internet of Things, and includes the following modules: 4G module and Bluetooth / WiFi module: used to establish a TCP long connection with the cloud operation and maintenance platform and a UDP short connection with the embedded function box, respectively, to achieve parallel transmission of video streams and sensor data, and support signal strength perception and channel switching of the communication link; Edge computing unit: This includes an embedded function box, an NPU acceleration core, and a local camera array. It is used to perform status data acquisition and preprocessing, LSTM prediction analysis, lightweight YOLOv5 target recognition and video status judgment, and generate abnormal event reports and visual tagging results. Condition Monitoring and Model Analysis Module: This module is used to standardize, model and predict gate current, location, and environmental data, quantify deviations, and classify anomalies, supporting intelligent judgment of primary and secondary abnormal events. Visual work order generation module: used to build 3D work order templates including video screenshots, stress hot spots, and comparison of predicted and measured data, supporting interactive display and parameter controllable backtracking; Blockchain evidence storage module: This module is used to perform SHA-256 hash operations on 3D visual work orders and write the hash value, timestamp, and event ID into a Merkle Patricia Tree structure through smart contracts, thus achieving tamper-proof evidence storage on the chain. Maintenance decision-making and personnel scheduling module: used to match faulty parts, generate maintenance plans, evaluate inventory status, perform optimal personnel path scheduling, and generate scheduling instruction packages containing navigation paths and personnel information.

[0018] Beneficial effects of the present invention: The present invention achieves parallel and stable transmission of sensor data and high-definition video streams by building a dual-mode communication link of 4G and Bluetooth / WiFi. It combines the LSTM prediction model to perform multivariate time series modeling of motor current, angle and environmental data, which can predict operational anomalies in advance. It further integrates denormalization processing and dynamic deviation quantization strategy to effectively identify potential overloads, motor imbalances or abnormal mechanical response of the rod, and realize millisecond-level early warning response to abnormal events.

[0019] This invention uses an anomaly-triggered video acquisition mechanism combined with a lightweight YOLOv5 model to accurately identify vehicle holdups and pole deformation. It also employs multi-frame confidence logic and a curvature calculation strategy to implement a secondary upgrade mechanism for determining abnormal vehicle-pole coupling states. The system can perform cross-modal cross-validation based on detected current anomalies and image content, effectively avoiding false positives and missed detections, and improving the stability and adaptive processing capabilities of barrier gate equipment under high-frequency use and harsh environments.

[0020] The present invention automatically generates structured three-dimensional work orders based on prediction data, visual results and fault maps, and embeds interactive controls to support visual analysis of the fault evolution process; further hashes and uploads the work order information through the blockchain evidence storage module to ensure that the work order data is authentic and traceable; combines the spare parts inventory status with the personnel GIS scheduling optimization algorithm to generate the optimal dispatch instructions, realizing the closed-loop management of "discovery-analysis-dispatching-evidence storage", and significantly improving the intelligence, efficiency and safety compliance of gate equipment maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 Schematic diagram of a method flow in an embodiment of the present invention; Figure 2 Schematic diagram of the system flow of an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. It is also noted that, to provide a more detailed description, the following embodiments are best and preferred embodiments, and those skilled in the art may employ alternative methods for implementing certain known technologies. Furthermore, the accompanying drawings are intended only to provide a more detailed description of the embodiments and are not intended to limit the present invention.

[0024] like Figure 1 As shown, a remote monitoring method for a gate based on the Internet of Things includes the following steps: S1, dual-mode communication link establishment: A parallel communication channel is established through the 4G module and the Bluetooth / WiFi module to collect the gate motor current, position sensor data and environmental temperature and humidity parameters in real time to form a status monitoring data set; S2, edge computing layer anomaly detection: The status monitoring dataset is input into the LSTM prediction model deployed in the embedded function box to generate a predicted value of the gate operation status. If the deviation between the measured value and the predicted value exceeds the first threshold, it is marked as a level 1 anomaly event; S3, hierarchical response processing: If a level 1 abnormal event is detected, the local camera is activated through the Bluetooth / WiFi module to capture multi-angle video, and the lightweight YOLOv5 model of the edge computing unit is simultaneously called for real-time analysis. If a vehicle is stranded or a pole is deformed, it is upgraded to a level 2 abnormal event; S4, intelligent generation of maintenance work orders: Automatically associate spare parts inventory data according to the level of abnormal events, generate a three-dimensional visual work order including fault location map, recommended maintenance plan and optimal personnel scheduling, and upload it to the cloud operation and maintenance platform through the blockchain evidence module.

[0025] S1 includes: S11, initialization configuration: Start the 4G module and Bluetooth / WiFi module at the same time, configure the 4G module's heartbeat packet sending interval to 10 seconds, the Bluetooth / WiFi module's transmission buffer size to 512 bytes, and verify the connection status of the two modules with the cloud operation and maintenance platform through the handshake protocol. Specifically, it includes: Start the 4G module and configure the heartbeat mechanism to periodically send heartbeat packets to maintain the TCP connection with the cloud. Start the Bluetooth / WiFi module and configure local communication parameters, including buffer size and data packet format; Verify the connection validity and availability of the encrypted channel between the two communication modules based on TLS or a custom handshake protocol.

[0026] S12, dynamic establishment of dual-mode communication link: Based on the initialized 4G module and Bluetooth / WiFi module, two parallel communication channels are established and channel priority strategies are set, including: Build a TCP long connection channel between the 4G module and the cloud operation and maintenance platform for high-bandwidth transmission of video streaming data; Build a UDP short connection channel between the Bluetooth / WiFi module and the embedded function box for low-latency transmission of sensor data; Set the channel priority strategy. When the signal strength of the Bluetooth / WiFi module is lower than -70dBm, the sensor data transmission will be automatically switched to the 4G module to ensure communication stability.

[0027] S13, synchronous acquisition of multi-source sensors: The analog-to-digital converter connected to the Bluetooth / WiFi module synchronously collects raw data with a sampling period of 100ms. The raw data includes the three-phase current value of the gate motor, the pulse signal of the gate pole position sensor, and the temperature and humidity values ​​output by the ambient temperature and humidity sensor. Specifically, it includes: Collect the three-phase current value of the gate motor , the sampling range is 0-5A, and the accuracy is ±0.5%; Collect the pulse signal of the position sensor of the gate pole and record it with a resolution of 0.1 degree; Collect the temperature value T and humidity value H output by the ambient temperature and humidity sensor; Add a timestamp to the collected raw data and write it to the temporary buffer in the format of "device ID + timestamp + data value".

[0028] S14, data preprocessing and verification: Preprocess and verify the raw data in the temporary buffer to generate a preprocessed data set, specifically including: Current data verification: If there is any phase current >4.8 A or three-phase unbalance satisfies: ,in , then the overload mark is triggered; Position data conversion: convert the pulse signal into the gate opening and closing angle θ (ranging from 0° to 90°) and calculate the angular velocity ; Environmental data filtering: The temperature and humidity values ​​are processed by sliding average and smoothed using a 5-point moving window. The filtered values ​​are expressed as: , ; The processing results are combined into a preprocessing data set, the fields include {ID, t, , , ,θ,ω,T,H,overload mark}; S15, generating a state monitoring data set: uploading the pre-processed data set to the embedded database of the edge computing layer through the dual channels of the 4G module and the Bluetooth / WiFi module in parallel to form a state monitoring data set, specifically including: Data fusion: Aligns the same device data from two communication channels based on timestamps to eliminate redundant and duplicate data packets; Data encapsulation: encapsulate the fused data into a standard data structure in JSON format; Dataset storage: The encapsulated data is written into the embedded database of the edge computing layer, and a data integrity check code is generated through the hash chain mechanism to form a status monitoring data set.

[0029] S2 includes: S21, condition monitoring data set preprocessing: read the condition monitoring data set from the embedded database, and generate a standardized time series data set by normalizing the current characteristics and segmenting the time series sliding window. Specifically, it includes: Data standardization: three-phase current values Perform Z-score standardization and the calculation formula is: ,in, is the historical current mean, is the historical standard deviation; Time series slicing: Divide the continuous data into time series segments of 50 sampling points in length using a sliding window method, with an overlap rate of 30% between adjacent data segments; Generate a standardized time series dataset with the following fields: {Device ID, window start time, , , ,θ,ω,T,H}; S22, dynamic loading of LSTM prediction model: Load the pre-trained LSTM prediction model through the hardware accelerator of the embedded function box, perform model structure initialization and digital signature verification to implement model deployment, specifically including: Input layer: receives time series input of dimension [50×7], and the 7 features are , , ,θ,ω,T,H; Hidden layer: configured with 64 LSTM units and a Dropout ratio of 0.2 to prevent overfitting; Output layer: uses a fully connected structure to output a set of predicted values ​​for the next five time steps; ; Real-time verification of the digital signature of the loaded model to ensure that the model has not been maliciously tampered with; S23, multivariate time series forecast execution: The standardized time series dataset is input into the LSTM forecast model, and the forecast result set is generated through forward reasoning and inverse transformation processing, specifically including: Forward reasoning: Model reasoning is performed with a running cycle of 100ms, and the predicted value is output in real time: ; Denormalization: Perform an inverse Z-score transformation on the predicted current value. The calculation formula is: ; Generate prediction result set, the fields are: {device ID, current timestamp, , , , , }.

[0030] S2 also includes: S24, Abnormal Deviation Quantification Calculation: Extract the measured value from the condition monitoring data set and calculate the real-time deviation between the measured value and the predicted result. If the deviation between the measured value and the predicted result exceeds a first threshold, it is marked as a level 1 abnormal event. Specifically, it includes: Current deviation calculation: ; Calculation of mechanical deviation: ; Abnormal judgment conditions: When satisfied and When , it is marked as a level 1 abnormal event; S25, Abnormal Event Feature Correlation: For data segments marked as Level 1 abnormal events, an abnormal event report is generated through abnormal type identification and environmental disturbance calculation, specifically including: Exception type identification: Associate the overload flag field in the corresponding data segment. If the overload flag exists, append the exception type code "OL"; Environmental factor analysis: Extract the temperature T and humidity H data within 10 seconds before the abnormality occurs and calculate the coefficient of variation respectively and , calculated as: , ; Generate an abnormal event report with the following fields: {Event ID, Device ID, Trigger Time, , , exception type code, , }.

[0031] S3 includes: S31, abnormal event triggers video acquisition: When a level 1 abnormal event is detected, a start command is sent to the local camera array via the Bluetooth / WiFi module to perform multi-angle image acquisition to generate abnormality-related video streams and store them in the edge storage pool. Specifically, it includes: Activate the three wide-angle cameras installed on the top, bottom, and side of the gate pillars, and set the acquisition parameters to 1920×1080 resolution and 25fps frame rate; Control the camera pan / tilt to scan along the preset trajectory, covering the gate pole pitch angle range of 45-135 degrees, and generate continuous video stream data; The collected video stream is encapsulated into H.264 encoding format, and the event ID and timestamp in the first-level abnormal event report are attached and stored in the edge storage pool.

[0032] S32, lightweight YOLOv5 model loading: Loads a preset lightweight YOLOv5 model through the edge computing unit. Model deployment is achieved through model structure tailoring, quantization configuration, and hardware binding. Specifically, it includes: Model structure verification: Confirm that the loaded model is a channel pruning optimized model, the convolutional layers are compressed from 24 to 16, and the model file size is compressed from 89MB to 34MB; Enable INT8 quantization inference mode, set the model input image size to 640×640 pixels, and the output layer target confidence threshold to 0.65; Bind the model inference task to the NPU acceleration core of the embedded function box and allocate a dedicated memory buffer with a capacity of 256MB; S33, real-time video analysis execution: reads video stream data from the edge storage pool, inputs it into a lightweight YOLOv5 model for frame-by-frame recognition and state inference, and generates a set of detection results, specifically including: Object detection processing: Detecting the vehicle outline (labeled "vehicle") and the barrier structure (labeled "barrier") in the video frame; The status determination logic is as follows: For the target with the "vehicle" tag, calculate its dwell time. If it is greater than 30 seconds and is located in the barrier drop zone, mark it as "vehicle stranded"; For the "barrier" label target, extract the rod skeleton and calculate the curvature: ; If the bending degree is greater than 2%, it is marked as "rod deformation"; Output detection result set, the fields include: {event ID, frame number, vehicle detention flag, pole deformation flag, curvature value}.

[0033] S3 also includes: S34, Level 2 abnormal event upgrade determination: Perform cross-frame correlation analysis on the detection result set, determine the upgrade conditions based on the event feature combination, and generate a Level 2 abnormal event report, specifically including: If the vehicle detention mark appears more than or equal to 8 times in 10 consecutive frames, and the three-phase current deviation value in the corresponding first-level abnormal event meets >1.5A, it will be upgraded to a Level 2 abnormal event; If the "rod deformation" flag is true in any frame and the curvature value corresponding to the frame is greater than 3%, a secondary abnormal event is directly triggered; Output secondary abnormal event report, the fields include: {event ID, upgrade time, trigger type (vehicle / rod), keyframe number set, maximum curvature}; S35, pre-triggering work order generation: Data fusion of the second-level abnormal event report and the first-level abnormal event report is performed, and a 3D visualization template is generated through video-sensor data alignment. Specifically, the following steps are performed: Spatial alignment: Correlate video keyframes with gate angle and angular velocity parameters in the state monitoring dataset based on timestamps; Feature association: superimpose the maximum curvature value on the stress distribution heat map layer of the fault location map; Work order pre-generation: Create a three-dimensional visualization template including video screenshots, pole curvature curves, and current abnormality sections, and store it in the pre-generated work order pool for subsequent step S4.

[0034] S4 includes: S41, intelligent association of spare parts inventory: By accessing the spare parts inventory database of the cloud operation and maintenance platform, performing exception type analysis and inventory query, and generating a spare parts matching result set, specifically including: By parsing the trigger type field in the secondary abnormal event report, match the fault component code: When the trigger type is "vehicle", the associated infrared sensor has the component code IRS-2023; When the trigger type is "rod body", the associated torque limiter has the component code TLQ-5A; Query the inventory status of the matching parts. If the inventory level is less than the safety threshold (IRS-2023: 5 units, TLQ-5A: ​​3 units), a replenishment warning flag is triggered. Generate spare parts matching result set, the fields include: {event ID, component code, inventory quantity, replenishment warning flag}; S42, Dynamic Generation of Maintenance Plans: By analyzing the fault location map and combining it with the spare parts matching result set, maintenance strategy matching and man-hour estimation are performed to generate a recommended maintenance plan document, specifically including: Extract the coordinates of the area where the stress value is greater than 50MPa in the fault location map and construct the hot zone boundary polygon; Select maintenance strategy based on hot zone area: When the area is less than 0.2 m2, a local reinforcement solution (solution code LBQ-01) is used; When the area is greater than or equal to 0.2 m2, the overall replacement plan (plan code ZTH-02) is adopted; Query the historical maintenance record database to obtain the average working hours of the plan (LBQ-01: 1.5h, ZTH-02: 4h); Generate a recommended maintenance plan document, the fields include: {Event ID, solution code, hot zone coordinate set, estimated working hours}; S43, 3D visualization work order construction: By calling the 3D visualization template, performing data fusion and interactive control construction, a 3D visualization work order is generated, specifically including: Data fusion processing: superimpose the predicted current curve and the measured curve for display; Map the thermal layer in the fault location map to the 3D model of the gate; Embed keyframe image thumbnails (from the detection result set); Add interactive controls: Set a click hotspot in the bending area of ​​the rod to display the corresponding bending value; Add a slider in the current anomaly range to support timeline backtracking of historical data; Generate a 3D visualization work order, the fields include: {Work order ID, 3D model data, interactive control set, timeline data}.

[0035] The S4 also includes: S44, blockchain evidence processing: The blockchain evidence module performs hash calculation and contract writing to generate evidence records, specifically including: Perform SHA-256 hash operation on the 3D visualization work order content to generate a 128-bit hash value; Execute the work order notarization smart contract of the cloud operation and maintenance platform and write the following fields: Work order ID, evidence timestamp (accurate to milliseconds), hash value, and associated secondary abnormal event report ID; Write the evidence data into the improved Merkle Patricia Tree structure and set the leaf node expiration time to 365 days; S45, optimal personnel scheduling execution: Based on the recommended maintenance plan document, maintenance personnel screening and route optimization are performed, and a scheduling instruction package is generated, specifically including: Screen the maintenance personnel database for candidates who meet the following conditions: The skills matrix includes the corresponding certification qualifications for the program codes; The distance between the current geographical location and the GIS of the gate site is less than 10km; Calculate the scheduling priority score of each candidate. The calculation formula is: ; The candidate with the highest score is selected as the dispatch object and a dispatch instruction packet is generated. The fields include: {work order ID, personnel ID, navigation path coordinate string, estimated arrival time}.

[0036] like Figure 2 As shown, a gate remote monitoring system based on the Internet of Things is used to implement the above-mentioned gate remote monitoring method based on the Internet of Things, including the following modules: 4G module and Bluetooth / WiFi module: used to establish a TCP long connection with the cloud operation and maintenance platform and a UDP short connection with the embedded function box, respectively, to achieve parallel transmission of video streams and sensor data, and support signal strength perception and channel switching of the communication link; Edge computing unit: This includes an embedded function box, an NPU acceleration core, and a local camera array. It is used to perform status data acquisition and preprocessing, LSTM prediction analysis, lightweight YOLOv5 target recognition and video status judgment, and generate abnormal event reports and visual tagging results. Condition Monitoring and Model Analysis Module: This module is used to standardize, model and predict gate current, location, and environmental data, quantify deviations, and classify anomalies, supporting intelligent judgment of primary and secondary abnormal events. Visual work order generation module: used to build 3D work order templates including video screenshots, stress hot spots, and comparison of predicted and measured data, supporting interactive display and parameter controllable backtracking; Blockchain evidence storage module: This module is used to perform SHA-256 hash operations on 3D visual work orders and write the hash value, timestamp, and event ID into a Merkle Patricia Tree structure through smart contracts, thus achieving tamper-proof evidence storage on the chain. Maintenance decision-making and personnel scheduling module: used to match faulty parts, generate maintenance plans, evaluate inventory status, perform optimal personnel path scheduling, and generate scheduling instruction packages containing navigation paths and personnel information.

[0037] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0038] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A remote monitoring method for a gate based on the Internet of Things, characterized in that: The following steps are involved: S1, dual-mode communication link establishment: A parallel communication channel is established through the 4G module and the Bluetooth / WiFi module to collect the gate motor current, position sensor data and environmental temperature and humidity parameters in real time to form a status monitoring data set; S2, edge computing layer anomaly detection: The state monitoring data set is input into the LSTM prediction model deployed in the embedded function box to generate a predicted value of the gate operation state. If the deviation between the measured value and the predicted value exceeds a first threshold, it is marked as a level 1 abnormal event; S3, hierarchical response processing: If a level 1 abnormal event is detected, the local camera is activated through the Bluetooth / WiFi module to capture multi-angle video, and the lightweight YOLOv5 model of the edge computing unit is simultaneously called for real-time analysis. If a vehicle is stranded or a pole is deformed, it is upgraded to a level 2 abnormal event; S4, intelligent generation of maintenance work orders: Automatically associate spare parts inventory data according to the level of abnormal events, generate a three-dimensional visual work order including fault location map, recommended maintenance plan and optimal personnel scheduling, and upload it to the cloud operation and maintenance platform through the blockchain evidence module.

2. The method for remotely monitoring a gate based on the Internet of Things according to claim 1, characterized in that: Said S1 comprises: S11, initialization configuration: start the 4G module and Bluetooth / WiFi module at the same time, configure the heartbeat packet sending interval of the 4G module to 10 seconds, the transmission buffer size of the Bluetooth / WiFi module to 512 bytes, and verify the connection status of the two modules with the cloud operation and maintenance platform through the handshake protocol; S12, dynamic establishment of dual-mode communication link: Based on the initialized 4G module and Bluetooth / WiFi module, two parallel communication channels are established and channel priority strategy is set; S13, synchronous acquisition of multi-source sensors: The analog-to-digital converter connected to the Bluetooth / WiFi module synchronously acquires raw data, including the three-phase current value of the gate motor, the pulse signal of the gate pole position sensor, and the temperature and humidity values ​​output by the ambient temperature and humidity sensor; S14, data preprocessing and verification: preprocess and verify the original data to generate a preprocessed data set; S15, generating a state monitoring data set: uploading the pre-processed data set to the embedded database of the edge computing layer in parallel through the dual channels of the 4G module and the Bluetooth / WiFi module to form a state monitoring data set.

3. The method for remotely monitoring a gate based on the Internet of Things according to claim 2, characterized in that: The S2 includes: S21, preprocessing of the state monitoring data set: reading the state monitoring data set from the embedded database, and generating a standardized time series data set by normalizing the current characteristics and performing time series sliding window segmentation; S22, dynamic loading of LSTM prediction model: loading the pre-trained LSTM prediction model through the hardware accelerator of the embedded function box, performing model structure initialization and digital signature verification to implement model deployment; S23, multivariate time series prediction execution: input the standardized time series data set into the LSTM prediction model, and generate a prediction result set through forward reasoning and inverse transformation processing.

4. A remote monitoring method for a gate based on the Internet of Things according to claim 3, characterized in that: Said S2 further comprises: S24, abnormal deviation quantification calculation: extract the measured value from the condition monitoring data set, and calculate the real-time deviation between the measured value and the predicted result. If the deviation between the measured value and the predicted result exceeds a first threshold, it is marked as a level 1 abnormal event; S25, abnormal event feature association: For the data segment marked as a first-level abnormal event, an abnormal event report is generated through abnormal type identification and environmental disturbance calculation.

5. The method for remote monitoring of a gate based on the Internet of Things according to claim 4 is characterized in that: The S3 includes: S31, abnormal event triggers video acquisition: when the first-level abnormal event is detected, a start instruction is sent to the local camera array via the Bluetooth / WiFi module to perform multi-angle image acquisition to generate abnormality-related video streams, and store them in the edge storage pool; S32, lightweight YOLOv5 model loading: loading a preset lightweight YOLOv5 model through the edge computing unit, and implementing model deployment through model structure tailoring, quantization configuration, and hardware binding; S33, real-time video analysis execution: read the video stream data from the edge storage pool, input it into the lightweight YOLOv5 model for frame-by-frame recognition and state inference, and generate a set of detection results.

6. The method for remotely monitoring a gate based on the Internet of Things according to claim 5 is characterized in that: Said S3 further comprises: S34, Level 2 abnormal event upgrade determination: performing cross-frame correlation analysis on the detection result set, determining upgrade conditions based on event feature combinations, and generating a Level 2 abnormal event report; S35, pre-triggering work order generation: fusing the secondary abnormal event report with the primary abnormal event report, and generating a three-dimensional visualization template through video-sensor data alignment.

7. The method for remotely monitoring a gate based on the Internet of Things according to claim 6 is characterized in that: The S4 includes: S41, intelligent association of spare parts inventory: By accessing the spare parts inventory database of the cloud operation and maintenance platform, performing exception type analysis and inventory query, and generating a spare parts matching result set; S42, dynamic maintenance plan generation: by analyzing the fault location map and combining the spare parts matching result set, performing maintenance strategy matching and man-hour estimation, and generating a recommended maintenance plan document; S43, 3D visualization work order construction: by calling the 3D visualization template, executing data fusion and interactive control construction, a 3D visualization work order is generated.

8. The method for remotely monitoring a gate based on the Internet of Things according to claim 7, characterized in that: Said S4 further comprises: S44, blockchain evidence processing: performing hash calculation and contract writing through the blockchain evidence module to generate an evidence record; S45, optimal personnel scheduling execution: performing maintenance personnel screening and route optimization through the recommended maintenance plan document, and generating a scheduling instruction package.

9. A gate remote monitoring system based on the Internet of Things, used to implement a gate remote monitoring method based on the Internet of Things as claimed in any one of claims 1 to 8, characterized in that: Includes the following modules: 4G module and Bluetooth / WiFi module: used to establish a TCP long connection with the cloud operation and maintenance platform and a UDP short connection with the embedded function box, respectively, to achieve parallel transmission of video streams and sensor data, and support signal strength perception and channel switching of the communication link; Edge computing unit: This includes an embedded function box, an NPU acceleration core, and a local camera array. It is used to perform status data acquisition and preprocessing, LSTM prediction analysis, lightweight YOLOv5 target recognition and video status judgment, and generate abnormal event reports and visual tagging results. Condition Monitoring and Model Analysis Module: This module is used to standardize, model and predict gate current, location, and environmental data, quantify deviations, and classify anomalies, supporting intelligent judgment of primary and secondary abnormal events. Visual work order generation module: used to build 3D work order templates including video screenshots, stress hot spots, and comparison of predicted and measured data, supporting interactive display and parameter controllable backtracking; Blockchain evidence storage module: This module is used to perform SHA-256 hash operations on 3D visual work orders and write the hash value, timestamp, and event ID into a Merkle Patricia Tree structure through smart contracts, thus achieving tamper-proof evidence storage on the chain. Maintenance decision-making and personnel scheduling module: used to match faulty parts, generate maintenance plans, evaluate inventory status, perform optimal personnel path scheduling, and generate scheduling instruction packages containing navigation paths and personnel information.

Images