An intelligent monitoring and identifying system for illegal construction of power transmission line

Abstract

The present application relates to the technical field of power transmission line monitoring, and discloses a kind of power transmission line construction site illegal intelligent monitoring identification system.The space modeling module of the system constructs three-dimensional dynamic model, environmental perception module collects multimodal environmental characteristic data, and action capture module obtains the action flow of construction personnel and equipment;Specification database module stores digital safety regulations, illegal modeling module combines three-dimensional dynamic model, environmental data and regulations to construct illegal behavior identification model matching current construction stage;Illegal analysis module outputs illegal probability vector accordingly, and spatial offset determination module establishes dynamic safety boundary and generates spatial offset parameters;Decision module generates multi-level response instructions by comprehensively generating the above data, and alarm and braking equipment are driven by execution control module;Feedback correction module updates model parameters through alarm record and false data, to realize dynamic optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission line monitoring, and in particular to an intelligent monitoring and identification system for violations at a power transmission line construction site. Background Art

[0002] Transmission line construction often takes place outdoors, involving high-altitude work, large equipment operation, and complex terrain. Frequent on-site personnel turnover and fluctuating equipment conditions present numerous challenges for safety management. Traditional construction site monitoring relies on manual inspections and fixed-point camera recordings. The former, limited by labor costs and inspection frequency, struggles to achieve full-time coverage. The latter, however, has blind spots and can only capture a single visual signal, failing to integrate environmental parameters like temperature, humidity, and wind speed, resulting in a fragmented perception of site conditions.

[0003] Safety margins are not fixed during construction. As foundation pouring, tower assembly, and conductor installation progress, equipment placement and operating ranges dynamically adjust. Static safety zones often differ from actual needs. For example, the safety distances during conductor deployment differ significantly from those during tower assembly, making boundary determinations prone to discrepancies when the same standards are applied.

[0004] Existing systems often rely on pre-set rule bases for comparison. These bases often contain general terms that are difficult to match with the detailed requirements of specific construction scenarios. The same action may have different compliance requirements at different stages, such as when a crane arm rotates in an open area versus when it rotates near a live object. Fixed rules can easily lead to misjudgments. Furthermore, when the system generates false alarms, it lacks an effective correction mechanism and is unable to optimize the judgment logic based on actual alarm records. Consequently, its adaptability gradually decreases with long-term use.

[0005] Insufficient multi-source data processing capabilities are also a prominent issue. Construction site video streams, equipment movement trajectories, and environmental sensor data are often stored and analyzed independently, lacking a collaborative processing mechanism. This makes it difficult to verify violations from multiple dimensions. For example, judging whether a person has entered a dangerous area based solely on video while ignoring the impact of wind speed on the person's balance can lead to inaccurate judgments. These factors combine to limit traditional monitoring systems in terms of real-time performance, accuracy, and adaptability, making them unable to meet the safety and control requirements of complex construction scenarios. Summary of the Invention

[0006] The purpose of the present invention is to provide an intelligent monitoring and identification system for violations at a transmission line construction site to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above-mentioned object, the present invention provides an intelligent monitoring and identification system for violations at a transmission line construction site, the system comprising:

[0008] The space modeling module is configured to construct a three-dimensional dynamic model of the construction site;

[0009] The environment perception module is configured to collect multi-modal environmental feature data of the construction site in real time;

[0010] The behavior capturing module is configured to synchronously acquire the action flow of the construction personnel and equipment through multi-angle visual sensors;

[0011] The specification database module stores the digital provisions of the power transmission line construction safety regulations;

[0012] The violation modeling module receives the three-dimensional dynamic model, multi-modal environmental feature data and digital provisions, and constructs a violation behavior identification model matched with the current construction stage;

[0013] The violation analysis module receives the action flow of the construction personnel and equipment in real time, and outputs a violation probability vector in combination with the violation behavior identification model;

[0014] The space offset determination module establishes a dynamic safety boundary according to the three-dimensional dynamic model, and generates a space offset parameter when the action flow of the construction personnel and equipment reaches the safety boundary;

[0015] The decision module generates multi-level response instructions by comprehensively combining the violation probability vector and the space offset parameter;

[0016] The execution control module parses the multi-level response instructions and drives the on-site alarm device and braking equipment;

[0017] The feedback correction module continuously receives the alarm device triggering record and the false positive data of the violation analysis module, and updates the violation behavior identification model parameters.

[0018] Preferably, the environment perception module includes an array of meteorological sensors, an electromagnetic field strength detector and a ground settlement monitor;

[0019] The multi-modal environmental feature data specifically includes real-time wind speed gradient, equipment electromagnetic radiation intensity value and ground bearing deformation variable;

[0020] The environment perception module transmits the wind speed gradient to the space modeling module for correcting the wind load parameter of the three-dimensional dynamic model;

[0021] The electromagnetic radiation intensity value is transmitted to the decision module for evaluating the equipment operation risk level;

[0022] The ground bearing deformation variable is transmitted to the space offset determination module for adjusting the dynamic safety boundary threshold value.

[0023] Preferably, the behavior capturing module deploys an array of wide-angle infrared cameras and millimeter wave radars;

[0024] The construction personnel and equipment motion streams include the displacement sequence of skeletal joints, the angle change value of the heavy machinery hydraulic arm, and the equipment grounding status identifier;

[0025] The skeletal joint point displacement sequence is input into the violation analysis module for posture compliance detection;

[0026] The hydraulic arm angle change value is transmitted to the spatial offset determination module to calculate the mechanical operation safety margin;

[0027] The equipment grounding status identifier input decision module activates the anti-electric shock protection protocol.

[0028] Preferably, the specification database module is associated with a construction progress node index table;

[0029] Digital terms are stored according to the foundation excavation stage, tower assembly stage, and conductor installation stage;

[0030] When the spatial modeling module detects the construction phase switching, it sends a phase switching instruction to the specification database module;

[0031] The specification database module responds to the stage switching instruction and outputs the protective equipment wearing standards, equipment safety distance thresholds, and high-altitude work protection measures requirements for the corresponding stage to the violation modeling module.

[0032] Preferably, the violation modeling module performs the following operations:

[0033] Integrate the protective equipment wearing standards during foundation excavation with the geological structure data of the 3D dynamic model to generate a landslide risk prediction model;

[0034] Combine the equipment safety spacing threshold and wind speed gradient during the tower assembly phase to output the wind-resistant overturning safety factor;

[0035] Correlate the protective measures required for high-altitude work during the conductor installation phase with the electromagnetic radiation intensity value to construct an arc damage prevention assessment matrix;

[0036] The landslide risk prediction model, wind-resistant overturning safety factor, and arc damage prevention assessment matrix are integrated into a violation behavior identification model and transmitted to the violation analysis module.

[0037] Preferably, the operation of the violation analysis module includes:

[0038] Compare the skeletal joint displacement sequence with the landslide risk prediction model to calculate the violation probability value of not wearing safety equipment;

[0039] Analyze the deviation of the hydraulic arm angle change value under the wind-resistant overturning safety factor;

[0040] Detect the matching degree between the grounding status identifier of the equipment and the arc damage prevention assessment matrix;

[0041] Based on the violation probability value, deviation degree, and matching degree, a violation probability vector containing a timestamp is generated and transmitted to the decision module.

[0042] Preferably, the spatial offset determination module performs the following operations:

[0043] Receive the ground bearing deformation and dynamically adjust the safety boundary coordinates of the tower base operation area;

[0044] Recalculate the safe operating radius of the lifting equipment based on the wind speed gradient;

[0045] When the skeletal joint point displacement sequence enters the adjusted safety boundary coordinate range, the person's out-of-bounds offset is generated;

[0046] When the hydraulic arm angle change exceeds the safe operating radius, the device posture abnormality parameters are output;

[0047] The personnel crossing-boundary offset and the equipment posture abnormality parameters are combined into spatial offset parameters and transmitted to the decision module.

[0048] Preferably, the processing logic of the decision module includes:

[0049] When the probability of not wearing safety equipment in the violation probability vector exceeds the set threshold and the person's crossing-border offset is positive, a level 1 alarm instruction is generated;

[0050] When the abnormal posture parameters of the equipment conflict with the wind-resistant overturning safety factor, a secondary braking command is output;

[0051] If the equipment grounding status identifier is abnormal and the arc damage prevention assessment matrix triggers an alarm, the third-level power-off command is initiated;

[0052] The first-level alarm instruction, second-level braking instruction, and third-level power-off instruction containing the alarm level, target device number, and geographic coordinates are encapsulated as multi-level response instructions and transmitted to the execution control module.

[0053] Preferably, the execution control module performs the following operations:

[0054] Parse the first-level alarm command to drive the sound and light alarm corresponding to the target device number;

[0055] Responding to the secondary braking command, controlling the hydraulic arm locking mechanism to intervene;

[0056] Execute the third-level power-off command to cut off the power supply circuit at the specified geographic coordinates;

[0057] At the same time, the triggering time of the sound and light alarm, the recorded value of the hydraulic arm locking angle, and the power-off operation timestamp are fed back to the feedback correction module.

[0058] Preferably, the operation of the feedback correction module includes:

[0059] Compare the triggering time of the sound and light alarm with the timestamp deviation of the violation probability vector;

[0060] Analyze the correlation between the recorded values ​​of the hydraulic arm locking angle and the abnormal equipment posture parameters;

[0061] Verify the response speed of the arc damage prevention assessment matrix based on the power-off operation timestamp;

[0062] The safety margin coefficient in the violation behavior identification model is updated based on timestamp deviation, correlation analysis results, and response speed verification data.

[0063] Compared with the prior art, the present invention has the following beneficial effects:

[0064] The system uses a spatial modeling module to construct a dynamic three-dimensional model of the construction site, mapping changes in site topography, equipment layout, and personnel positions in real time. This allows monitoring to transcend the limitations of a two-dimensional perspective and present the construction scene in a three-dimensional perspective. The environmental perception module collects multimodal environmental feature data, encompassing visual, audio, meteorological, and other information, enabling a comprehensive capture of the on-site environment and avoiding information loss associated with a single data type.

[0065] The behavior capture module leverages multi-angle visual sensors to simultaneously capture the movement of construction personnel and equipment, breaking through the limitations of fixed viewpoints and ensuring the integrity and consistency of recorded actions. This encompasses everything from the physical movements of workers working at height to the operational trajectory of large machinery. The standardized database module stores digitized safety regulations, providing a unified standard for determining violations. These regulations are stored in digital form, facilitating rapid comparison with real-time data.

[0066] The violation modeling module receives the 3D dynamic model, multimodal environmental feature data, and digital terms to construct a violation behavior recognition model tailored to the current construction phase. This model is tailored to the specific requirements of different phases, such as foundation construction, tower assembly, and conductor installation, eliminating the inapplicability of general models in specific scenarios. The violation analysis module receives the action stream in real time and, combined with the recognition model, outputs a violation probability vector. Rather than simply providing a "yes" or "no" judgment, it presents the likelihood of different violation types in vector form, providing richer reference information for subsequent decision-making.

[0067] The spatial offset determination module establishes a dynamic safety boundary based on a three-dimensional dynamic model. This boundary adjusts in real time with construction progress, equipment movement, and environmental changes. The spatial offset parameters generated when personnel or equipment reach the boundary accurately reflect the extent and location of the violation, making boundary control more targeted. The decision module combines the violation probability vector and spatial offset parameters to generate multi-level response instructions, avoiding over- or under-response caused by a single parameter. The hierarchical setting of instructions can adapt to scenarios of varying urgency, such as localized alarms for minor violations and equipment braking for serious violations.

[0068] The execution control module parses instructions and activates on-site alarms and braking equipment, achieving a rapid transition from judgment to response, ensuring the timeliness of alarms and braking actions. The feedback correction module continuously receives alarm trigger records and false alarm data, updating the parameters of the violation identification model. This allows the model to gradually adapt to actual site conditions over long-term use, reducing judgment errors caused by environmental differences and different operating habits, and improving the system's adaptability to different construction scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 This is a working principle diagram of the intelligent monitoring and identification system for violations at transmission line construction sites according to the present invention;

[0070] Figure 2 Flowchart of the working of the environment perception module;

[0071] Figure 3 Flowchart for standardizing the work of database modules;

[0072] Figure 4 Flowchart of the violation analysis module. DETAILED DESCRIPTION

[0073] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0074] See also Figure 1 The present invention provides a transmission line construction site violation intelligent monitoring and identification system, the system comprising:

[0075] The spatial modeling module constructs a three-dimensional dynamic model of the construction site and updates the model's spatial coordinates through real-time terrain scanning and equipment positioning. The environmental perception module simultaneously collects meteorological, electromagnetic, and geological data to form a multimodal environmental feature data stream. The behavior capture module uses a wide-angle infrared camera array and millimeter-wave radar to capture the displacement sequence of construction workers' skeletal joints, the angle change of the equipment's hydraulic arm, and the grounding status identifier. The specification database module stores digital safety regulations indexed by construction phase (foundation excavation / tower assembly / conductor installation). The violation modeling module receives the three-dimensional model, environmental data, and corresponding phase regulations to generate a violation behavior recognition model for the current construction phase. The violation analysis module inputs the skeletal displacement sequence and hydraulic arm angle values ​​into the violation behavior recognition model and outputs a timestamp-matched violation probability vector. The spatial offset determination module establishes a dynamic safety boundary based on the three-dimensional model and generates spatial offset parameters when the action flow reaches the boundary. The decision module associates the violation probability vector with the spatial offset parameters to generate multi-level response instructions. The execution control module parses the instructions to drive the sound and light alarm, hydraulic arm locking mechanism, or power disconnect device. The feedback correction module collects alarm trigger records and false alarm data to iteratively optimize the violation behavior recognition model parameters.

[0076] Example 1: See Figure 2 The environmental perception module is equipped with a meteorological sensor array, an electromagnetic field intensity detector, and a foundation settlement monitor to achieve multimodal environmental feature acquisition. The meteorological sensor array captures atmospheric parameters at a sampling frequency of 10Hz. A vertical profile is constructed using 16 distributed air pressure sensors. Real-time wind speed gradient profile data is calculated using differential air pressure values, and the three-dimensional wind speed vector is decomposed into horizontal and vertical components. The electromagnetic field intensity detector deploys an 8-channel circular probe array within a 5-meter radius around the construction equipment. Each channel captures the alternating electromagnetic field at a sampling rate of 50kHz. The electromagnetic radiation intensity spectrum is generated through a fast Fourier transform, and the radiation intensity characteristic values ​​from the power frequency to the megahertz frequency band are extracted. The foundation settlement monitor uses a combination of a laser ranging unit and a dual-axis inclination sensor. The laser unit monitors the displacement of surface benchmarks with a resolution of 0.1mm, and the inclination sensor measures the inclination angle of the geological structure with an accuracy of 0.01°. The data from the two are integrated to calculate the ground bearing deformation. Wind speed gradient profile data is transmitted in real time to the dynamic model correction unit of the spatial modeling module. The model recalculates the wind load distribution coefficient of the tower structure based on the horizontal wind shear rate. The electromagnetic radiation intensity characteristic value is input into the risk assessment unit of the decision-making module, and compared with the preset equipment electromagnetic exposure threshold table to generate an operation risk level index (scaled from 0 to 100). The ground bearing deformation variable is synchronously transmitted to the boundary adaptation engine of the spatial offset judgment module, triggering the dynamic adjustment algorithm of the safety boundary threshold.

[0077] The behavior capture module integrates eight wide-angle infrared cameras and four millimeter-wave radars to synchronously capture spatial motion streams. The wide-angle infrared cameras capture images at a resolution of 1280×1024 and a frame rate of 30fps. Using a deep neural network model, they extract the 3D coordinates of 17 skeletal joints of construction workers, generating a time series of joint displacements. This data includes the rate of change of the 3D coordinates and the acceleration vector. The millimeter-wave radar uses 77GHz frequency-modulated continuous wave technology to scan the heavy machinery operation area with a 0.5° azimuth resolution. Point cloud clustering analysis is used to extract real-time angular changes at the six articulated joints of the hydraulic arm. The angle sampling interval is 33ms, achieving an angular resolution accuracy of ±0.5°. Equipment grounding status identifiers are obtained through a radio frequency identification system. 125kHz passive RFID tags are installed at the grounding terminals of construction equipment. A reader array is deployed within a three-meter radius, polling the tag status every 200ms and outputting a Boolean identifier (0 for disconnected, 1 for valid grounding). The skeletal joint displacement sequence is input into the posture compliance detection unit of the violation analysis module for spatiotemporal alignment with a preset safe working posture template. The hydraulic arm angle change is transmitted to the mechanical dynamics analysis unit of the spatial offset determination module, where the mechanical operation safety margin index is calculated in real time based on a multibody dynamics model. The equipment grounding status identifier is connected to the protocol trigger unit of the decision module. When the identifier detects a 0, the three-level anti-electric shock protection protocol is immediately activated, forcing the equipment to shut down. Multimodal sensing data operates collaboratively to form a closed loop. When the laser ranging unit detects ground bearing deformation exceeding 3mm, the millimeter-wave radar automatically increases the hydraulic arm monitoring frequency to 60Hz. When the electromagnetic radiation intensity exceeds the 4kV / m threshold, the infrared camera activates high-sensitivity mode to enhance the recognition of protective equipment worn by the operator. A time synchronization mechanism is established during the data collection process. The GPS timing module provides microsecond-level time synchronization for all sensors, ensuring a consistent time base for spatiotemporal data such as wind speed gradients, skeletal displacement sequences, and hydraulic arm angles. The sampling period of the environment perception module and the frame rate of the behavior capture module are kept at a coupling ratio of 1:30 through the system clock controller, ensuring that each round of wind speed gradient update corresponds to thirty frames of action flow analysis.

[0078] Example 2: See Figure 3The specification database module stores structured digital clauses of construction safety regulations, enabling dynamic access through a construction progress node index table. The construction progress node index table establishes a three-dimensional mapping relationship: the spatial modeling module collects the tower base concrete pouring volume completion rate, tower assembly height percentage, and conductor deployment length progress values ​​in real time, and generates stage marking parameters after normalization. When the concrete pouring completion rate exceeds 95% and the tower assembly height reaches 15% of the design value, the index table marks it as the tower assembly stage; when the conductor deployment progress exceeds 20%, it switches to the conductor installation stage. The index table is updated every five seconds and transmitted to the scheduling unit of the specification database. The digital terms are classified and stored in physical partitions according to the three stages of foundation excavation, tower assembly, and conductor installation: the foundation excavation partition stores the slope safety distance threshold (2.5 times the pit depth), the impact strength standard of the protective net (50kN), and the safety helmet wearing status detection specification; the tower assembly partition defines the calculation formula for the safe operating radius of the lifting equipment, the wind speed-lifting mass limit curve table, and the connection specification of the high-altitude work fall arrester; the conductor installation partition sets the minimum safety distance matrix for live work, the conductivity detection standard for shielding clothing, and the temporary grounding device layout specification.

[0079] When the laser scanning unit in the spatial modeling module detects that the tower assembly height has reached 89.5% of the design value, its stage identifier sends a timestamped switch request to the specification database. Upon receiving the request, the specification database searches the index table to confirm that the current progress parameters meet the conditions for activating the conductor erection stage (height ≥ 88% and conductor deployment > 15%). It then dispatches a conductor erection zone data package. This data package includes a set of protective equipment clauses (hardhats, shielding suits, and fall arresters), a set of equipment operation clauses (grounding specifications for hydraulic equipment, electromagnetic protection requirements for tensioners), and a set of environmental response clauses (wind speed-conductor sway safety margin table). This set of clauses is transmitted via an encrypted channel to the data fusion engine in the violation modeling module.

[0080] The violation modeling module integrates multi-source heterogeneous data. During the foundation excavation phase, it receives geological parameters from the spatial modeling module (soil shear strength 1.2 MPa, groundwater level -3.2 m), combines them with slope safety distance thresholds from the regulatory database, and generates a landslide risk prediction model through geological stability simulation. This model inputs the construction area coordinates and outputs a five-level risk index (ranging from 0.1 to 0.9). When the risk index exceeds 0.6, the equipment detection mechanism is triggered. During the tower assembly phase, wind speed gradient data (horizontal component 8.3 m / s, vertical component 1.2 m / s) from the environmental perception module is integrated with the wind speed-hoisting limit curve from the regulatory database. This data is then substituted into the boom overturning moment equilibrium equation to output a dynamic wind-resistant overturning safety factor, setting a critical threshold of 1.28. During the conductor installation phase, the electromagnetic protection standards of the specification database are matched with the electromagnetic radiation intensity spectrum of the environmental perception module (peak value in the 125kHz-10MHz frequency band) to construct a three-dimensional arc damage prevention assessment matrix. The matrix dimensions include voltage level (0-220kV), operating distance (1-15m), and protection level (level 1-4) indicator layers.

[0081] The data fusion process utilizes a timestamp alignment mechanism: the geological parameter sampling time, wind speed gradient timestamp, and electromagnetic radiation detection time are all tied to the version of the clause transmitted in the specification database. When fusing data from the foundation excavation phase, the geological parameter time deviation must be controlled within 500ms. Wind speed gradient data from the tower assembly phase utilizes a sliding window mean filter, with the window width matching the response time of the lifting equipment. Electromagnetic radiation data from the conductor installation phase undergoes frequency domain alignment, increasing the frequency resolution of the radiation spectrum to 0.5kHz to accommodate the frequency band accuracy of the assessment matrix. The resulting three types of model data: a landslide risk prediction model (including geological coordinate mapping), a wind-induced overturning safety factor (including a dynamic threshold update algorithm), and an arc damage assessment matrix (including a frequency-to-space domain conversion module). After model compression, these data are integrated into a violation identification model. This model is transmitted to the violation analysis module's buffer pool via Gigabit Ethernet, synchronized with the construction phase scans, with a minimum interval of three seconds. When the model is updated, the boundary condition parameters of the previous stage are automatically inherited. For example, when switching from foundation excavation to tower assembly, the slope risk parameter is converted into the tower foundation settlement warning threshold.

[0082] Double verification is implemented during construction phase transitions: When the specification database outputs new phase terms, the violation modeling module requests environmental parameter recalibration from the spatial modeling module. The environmental perception module simultaneously activates high-frequency sampling (wind speed gradient acquisition frequency increased to 20Hz, and foundation settlement monitoring accuracy improved to 0.05mm) to ensure that model parameters deviate from on-site conditions by less than 3%. The system maintains a ten-second transition buffer during which the old and new phase models run in parallel. Final transition is completed only when the consistency of three consecutive analysis results exceeds 95%. Phase marker parameters are written to a log file and linked to all generated model versions, supporting full-cycle traceability and review.

[0083] Example 3: See Figure 4 The violation analysis module receives in real time the skeletal joint displacement sequence, hydraulic arm angle change, and equipment grounding status identifier transmitted by the behavior capture module. The skeletal joint displacement sequence is a three-dimensional coordinate time series in the format [timestamp, joint ID, x, y, z]. The lumbar vertebrae (ID: L3) and head joint (ID: H1) serve as core monitoring points. The module invokes the collapse risk prediction model within the violation behavior recognition model to extract the geological risk factor for the construction area. When the head joint's Z-axis height remains below 1.5 meters and the helmet identification indicator at point H1 is 0, the violation probability is calculated based on the regional geological risk factor. This calculation uses a probabilistic superposition algorithm: the geological risk factor is normalized to a 0-1 weighted range, and the absence of a helmet is used as an independent event trigger probability with an initial value of 0.6. The product of these two factors generates the final violation probability. The hydraulic arm angle change processing flow is as follows: The six joint angle data streams collected by the millimeter-wave radar are received [timestamp, joint ID, angle]. The hydraulic main arm joint (ID: Arm01) is selected as the analysis target. The critical value of the wind-resistance overturning safety factor is used as the reference angle θ, and the real-time angle data φ is processed by the deviation calculation algorithm. The deviation is calculated using the normalized differential equation:

[0084]

[0085] in is the deviation output value, is the real-time hydraulic arm elevation angle, is the safe elevation angle allowed by the current working conditions, The wind speed impact factor (range: 0.8-1.2, provided in real time by the environmental perception module) is used. The equipment grounding status identifier processing unit receives a Boolean parameter (0 / 1) and an electromagnetic radiation intensity value. When the identifier is 0, the minimum protection requirement for the corresponding voltage level in the arc damage prevention assessment matrix is ​​retrieved. The matching degree is calculated as follows: subtract the real-time electromagnetic radiation intensity value from the protection level standard parameter, and divide the result by the standard parameter to obtain a percentage matching degree. A negative value indicates a risk state.

[0086] Output data structured packaging: Generates a three-element violation probability vector [equipment missing probability, hydraulic deviation, ground fit] and appends a millisecond timestamp (UTC time format). When the equipment missing probability is greater than 0.6, the hydraulic deviation is greater than 0.25, or the ground fit is less than 0, the vector emergency flag protocol (EMERG_FLAG=1) is initiated. The spatial offset determination module operation process runs in parallel with the violation analysis module. Receives the ground bearing deformation from the foundation settlement monitor. (Unit: mm), adjust the safety margin according to the deformation level: When the tower base operation area is safe, the radius of the boundary is reduced. Meters; receive wind speed gradient data (unit ), safe operating radius of lifting equipment The update formula is: ,in is the base radius (12 meters), is the attenuation coefficient (0.5), is the wind speed threshold ( Real-time matching of skeletal joint displacement sequence: When the 3D coordinates of the lumbar joint (ID: L3) fall within the compressed safety boundary, the displacement vector modulus is calculated as the person's out-of-bounds offset. ; If the spatial projection point coordinates corresponding to the angle of the hydraulic arm hinge point (ID: Arm01) exceed Boundary, records the minimum Euclidean distance from the projection point to the safety boundary As the device posture abnormality parameter. Generate a spatial offset parameter group [D, E] and append a timestamp, where A positive value indicates the intrusion depth (in meters). The distance that the device exceeds the limit (in meters).

[0087] Data processing timing control mechanism: The analysis cycle for skeletal joint displacement sequences is fixed at 33ms (corresponding to a 30fps frame rate), and the spatial offset parameter generation cycle is synchronized with the environmental perception module (100ms). The module has a built-in time calibration buffer, which implements time window matching between the violation probability vector and the spatial offset parameter. Using the spatial offset parameter timestamp as a reference, the violation probability vector is captured 50ms back and correlated. When hydraulic arm angle changes are simultaneously input into both modules, a direct data pipeline is established: the raw hydraulic arm angle data is diverted to the violation analysis module for deviation calculation and to the spatial offset determination module for safety radius verification. Output data transmission utilizes a differential protocol: the violation probability vector is sent to the decision module via a high-speed data bus with a transmission latency of less than 10ms; the spatial offset parameter is transmitted via a fiber optic channel with a latency of less than 5ms. The system creates an independent analysis thread for each piece of heavy equipment, sharing environmental parameters such as wind speed gradient and geological risk factor between threads via a memory-mapped file. Critical state handling mechanism: When the person's out-of-bounds offset D increases for three consecutive cycles or the device's abnormal posture parameter E exceeds 1.5 meters, the parameter accelerated transmission mode is triggered, increasing the data transmission frequency to three times the standard value. All output data packets are appended with a data quality check code generated from the raw data feature values ​​and used for authenticity verification by the decision module. The module operation log records snapshots of key parameters, including the safety margin adjustment range, the ω factor value in the deviation calculation, the electromagnetic reference value in the matching calculation, and other metadata supporting traceability analysis.

[0088] Example 4: The decision module receives the violation probability vector transmitted by the violation analysis module and the spatial offset parameters generated by the spatial offset determination module. The violation probability vector data structure contains three elements: the probability of missing head safety equipment P1 (such as 0.87), the hydraulic arm angle deviation α (such as 0.31), and the ground state matching degree M (such as 0.1); the vector is attached with a millisecond-level timestamp (such as 20250728T150301.456Z). The spatial offset parameters include the personnel crossing the boundary offset D (such as +2.1m) and the equipment posture abnormality parameter E (such as 1.7m). The timestamp is aligned with the violation probability vector. The module has a built-in three-level response rule engine, and the association conditions stored in the rule base are shown in Table 1:

[0089] Table 1: Multi-level response rule engine trigger conditions and instruction correspondence table

[0090]

[0091] The specific judgment process is as follows: When the system detects that the probability vector of violation in a tower base operation area (timestamp 20250728T150301.456Z) contains (Not wearing a helmet), 、 , while associating the spatial offset parameters (People intruding into the slope area), The rule engine first matches the first level alarm condition: value and positive , triggering the ALERT_1 instruction. The instruction binds the RFID tag ID of the person (OPERATOR_048) and its Beidou positioning coordinates (39.126°N, 116.873°E). At the same time, the secondary braking condition is detected: Not exceeding threshold , but the real-time position of the hydraulic main arm Exceeded Limit. The engine queries the real-time value of the wind-resistance overturning safety factor ( ) and the device posture abnormality parameters Conflict (coefficient Critical value and ), the second-level braking condition is met to generate the BRAKE_2 instruction, specifying the device number EXCAVATOR_07 and the safety elevation angle 58°. No third-level power-off condition is triggered during this process ( ).

[0092] The execution control module parses the multi-level response instruction set. For the ALERT_1 instruction: call the sound and light alarm network within 500 meters of the target coordinates, drive the device number SIREN_12 to start the third-level alarm mode (120dB pulse buzzer frequency 2Hz, red rotating warning light flashing frequency 1Hz), and the duration is initially set to 15 seconds. For the BRAKE_2 instruction: control the hydraulic arm locking mechanism of EXCAVATOR_07, send the brake command code (0x5FA2) to the device controller via the CAN bus, limit the hydraulic cylinder stroke and force the elevation angle to , the brake intervention time is recorded as 20250728T150302.112Z. Module synchronous startup status monitoring: the sound and light alarm trigger signal strength is transmitted back in real time (current value 118dB), the hydraulic arm locking angle is sampled every second (stable value after locking) ).

[0093] When the system detects new data in the wire installation area (timestamp 20250728T150305.789Z): violation probability vector , spatial offset parameters The rules engine detects ground matching And the real-time value of electromagnetic radiation intensity Threshold, immediately trigger POWEROFF_3 instruction. Instruction binds power supply circuit number CIRCUIT_NO_33 and fault point coordinates (39.131 °N, 116.877 °E). After the execution control module responds: send the opening instruction (OP_CODE=0x91) to the intelligent circuit breaker, the circuit breaking time stamp is recorded as 20250728T150306.003Z, and the response delay is 214 ms. The circuit breaker status feedback contact opens successfully (STATUS=0x00).

[0094] Multi-instruction concurrent processing mechanism: when multiple groups of violations occur at the same time window (within 500 ms), the module establishes an instruction priority queue. Process in the order of three-level power-off > two-level brake > one-level alarm. For example, at a certain time, a wire erection area power-off instruction (POWEROFF_3) and an adjacent area brake instruction (BRAKE_2) are received, and the power-off operation is performed first and then the brake is started. All execution records are packaged into log units in real time, and the data fields include:

[0095] Alarm instruction: device number, trigger start and end time, maximum sound intensity

[0096] Brake instruction: device number, target angle, actual locking angle

[0097] Power-off instruction: circuit number, instruction sending time, circuit breaker response time

[0098] After the log unit is compressed, it is fed back to the feedback correction module through an independent channel. Abnormal processing flow: if the hydraulic arm locking mechanism does not reach the target angle within 300 ms, the execution control module automatically upgrades the response level, supplements the device shutdown instruction (STOP_CMD) and marks it as an urgent event (URGENT_FLAG=1). The system retains a retry mechanism for instructions that have not been successfully executed, such as automatically switching to a backup device (SIREN_15) when the audible and visual alarm is offline. Time synchronization accuracy control: all instruction timestamps use GPS time source, and the actuator clock error is less than 1 ms.

[0099] Example 5: The feedback correction module continuously receives the operation record data stream uploaded by the execution control module and the original analysis results of the violation analysis module. The operation record data packet contains three types of structured information: the audible-light alarm trigger record carries the device number, sound intensity peak, warning light state, and duration field; the hydraulic arm locking operation record contains the machine number, target safe elevation angle, actual locking angle value, and locking intervention timestamp sequence; the power-off operation record stores the loop number, power-off instruction issuance time, and circuit breaker actual action timestamp. The original analysis results are extracted from the violation analysis module log pool and include the violation probability vector generation timestamp, landslide risk prediction model call parameters, and hydraulic arm deviation degree calculation process snapshot. The built-in time trajectory analyzer in the module performs the following operations: align the audible-light alarm opening time (e.g., 20250728T150302.000Z) with the violation probability vector marking time (20250728T150301.456Z), and calculate the absolute time difference as 544 milliseconds; mark the record items with a time difference exceeding the preset 300-millisecond threshold, and generate a time calibration parameter vector [device number, time offset].

[0100] Hydraulic arm operation data analysis process: extract the actual locking angle (57.8°) and target safe elevation angle (58.0°) of the number EXCAVATOR_07 in the execution record, calculate the absolute angle difference 0.2°. Retrieve the same period spatial offset judgment module log to obtain the device pose abnormal parameter original value E = 1.7 meters (corresponding to the theoretical correction requirement 58°). Start the residual risk verification mechanism: input the actual locking angle 57.8° into the anti-wind overturning safety coefficient calculation model, combined with the real-time wind speed gradient data of the environment perception module (horizontal wind speed 8.3 m / s), obtain the current overturning coefficient calculation result 1.13 (lower than the critical value 1.25). The residual risk value (1.25-1.13=0.12) is output as the correlation index. Power-off response verification operation: compare the power-off instruction issuance time (20250728T150306.003Z) with the anti-arc injury assessment matrix alarm trigger time (20250728T150305.789Z), obtain a response interval of 214 milliseconds; at the same time, retrieve the electromagnetic radiation intensity historical record of the environment perception module at the corresponding coordinate point (9.3 kV / m), and calculate the time deviation rate 7% according to the voltage level protection standard (requirement ≤200ms response) specified by the matrix.

[0101] Based on the above analysis results, model parameter iteration is performed: when the offset in the time calibration parameter vector exceeds 300 milliseconds for three consecutive times, the timestamp compensation algorithm is triggered, and the clock synchronization parameters of the violation analysis module are updated. The residual risk value is statistically processed to generate a safety margin coefficient adjustment amount: if the residual risk value > 0.1 appears five times in a row, the anti-wind overturn safety coefficient threshold (originally 1.25 adjusted to 1.28) is increased; if the arc response time deviation rate > 5% accumulates for ten times, the response determination threshold of the anti-arc injury evaluation matrix is tightened from 200ms to 180ms. The parameter update logic includes a version control mechanism: each adjustment generates an incremental parameter package, including modified fields, new parameter values, and effective time window. For example, for the adjustment of the collapse risk prediction model: during the period of frequent sound-light alarm time delay events (more than twenty delays per day), the risk probability threshold is lowered from 0.7 to 0.68, and the equipment detection range is expanded to include back protection equipment identification.

[0102] The model update operation implements double-channel verification: the update parameter package is first loaded into the test environment, and historical three-month operation data sets (about 120 million records) are injected to perform simulation analysis. Compare the alarm trigger difference rate between the original model and the new parameter model: when the difference rate is less than three percent, allow the production environment to update; if the difference rate exceeds five percent, trigger the manual review process. When officially deployed, a gray release strategy is adopted: on the first day, update five percent of the edge computing nodes, and simultaneously monitor the false positive rate changes of these nodes; after three days, the false positive rate fluctuation range is less than one point five percent, and the full amount is pushed to all terminals. All update records generate version snapshots, including operation time, environment state fingerprints (memory usage, CPU load average), update effect indicators (residual risk change trend, response time standard deviation). The snapshot is stored through the blockchain to generate an unalterable log, supporting full life cycle parameter tracing through device number + time range.

[0103] The system retains parameter rollback capability: when key indicators are abnormal during the running of the new parameter model (such as a 200% increase in daily secondary braking instructions), automatically revert to the previous stable version. The rollback operation generates a rollback report, detailing four types of information: parameter update version number, abnormal event trigger time, rollback execution time, and indicator recovery status. Environment adaptive parameters are continuously updated through background learning: hydraulic arm lock angle deviation data is generated every eight hours to generate a statistical distribution graph, and when the deviation value distribution center continuously shifts to the positive direction, the control compensation coefficient of the locking mechanism is adjusted; the grounding state response time is calculated every hour, and this value is included in the time margin adjustment factor of the anti-electric shock protection protocol. The module maintains a dynamic weight knowledge base, storing the history of various parameter adjustments and their impact assessments.

[0104] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0105] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent monitoring and identification system for violations at transmission line construction sites, characterized in that: include: The spatial modeling module is used to build a three-dimensional dynamic model of the construction site; The environmental perception module is used to collect multimodal environmental feature data of the construction site in real time; The behavior capture module is used to synchronously capture the motion flow of construction personnel and equipment through multi-angle visual sensors; The specification database module stores the digital clauses of the transmission line construction safety regulations; The violation modeling module receives the 3D dynamic model, multimodal environmental feature data, and digital terms to build a violation behavior recognition model that matches the current construction stage; The violation analysis module receives the construction personnel and equipment action stream in real time, inputs the skeleton displacement sequence and hydraulic arm angle value into the violation behavior recognition model, and outputs the violation probability vector with a timestamp; The spatial offset determination module establishes a dynamic safety boundary based on the 3D dynamic model and generates spatial offset parameters when the movement flow of construction personnel and equipment reaches the safety boundary; The decision module integrates the violation probability vector and the spatial offset parameter to generate a multi-level response instruction; The execution control module parses the multi-level response instructions and drives the on-site alarm device and braking equipment; The feedback correction module continuously receives the triggering records of the alarm device and the false alarm data of the violation analysis module, and updates the parameters of the violation behavior recognition model.

2. The intelligent monitoring and identification system for violations at a power transmission line construction site according to claim 1, characterized in that: The environmental perception module includes a meteorological sensor array, an electromagnetic field intensity detector and a foundation settlement monitor; Multimodal environmental characteristic data specifically includes real-time wind speed gradient, equipment electromagnetic radiation intensity value, and ground bearing deformation; The environmental perception module transmits the wind speed gradient to the spatial modeling module for correcting the wind load parameters of the three-dimensional dynamic model; Transmit the electromagnetic radiation intensity value to the decision module for evaluating the equipment operation risk level; The ground bearing deformation is transmitted to the spatial offset determination module for adjusting the dynamic safety boundary threshold.

3. The intelligent monitoring and identification system for violations at a power transmission line construction site according to claim 2, characterized in that: The behavior capture module deploys a wide-angle infrared camera array and millimeter-wave radar; The construction personnel and equipment motion streams include the displacement sequence of skeletal joints, the angle change value of the heavy machinery hydraulic arm, and the equipment grounding status identifier; The skeletal joint point displacement sequence is input into the violation analysis module for posture compliance detection; The hydraulic arm angle change value is transmitted to the spatial offset determination module to calculate the mechanical operation safety margin; The equipment grounding status identifier input decision module activates the anti-electric shock protection protocol.

4. The intelligent monitoring and identification system for violations at a power transmission line construction site according to claim 3, characterized in that: The specification database module is associated with the construction progress node index table; Digital terms are stored according to the foundation excavation stage, tower assembly stage, and conductor installation stage; When the spatial modeling module detects the construction phase switching, it sends a phase switching instruction to the specification database module; The specification database module responds to the stage switching instruction and outputs the protective equipment wearing standards, equipment safety distance thresholds, and high-altitude work protection measures requirements for the corresponding stage to the violation modeling module.

5. The intelligent monitoring and identification system for violations at a power transmission line construction site according to claim 4, characterized in that: The Violation Modeling module performs the following operations: Integrate the protective equipment wearing standards during foundation excavation with the geological structure data of the 3D dynamic model to generate a landslide risk prediction model; Combine the equipment safety spacing threshold and wind speed gradient during the tower assembly phase to output the wind-resistant overturning safety factor; Correlate the protective measures required for high-altitude work during the conductor installation phase with the electromagnetic radiation intensity value to construct an arc damage prevention assessment matrix; The landslide risk prediction model, wind-resistant overturning safety factor, and arc damage prevention assessment matrix are integrated into a violation behavior identification model and transmitted to the violation analysis module.

6. The intelligent monitoring and identification system for violations at a power transmission line construction site according to claim 5, characterized in that: The operations of the violation analysis module include: Compare the skeletal joint displacement sequence with the landslide risk prediction model to calculate the violation probability value of not wearing safety equipment; Analyze the deviation of the hydraulic arm angle change value under the wind-resistant overturning safety factor; Detect the matching degree between the grounding status identifier of the equipment and the arc damage prevention assessment matrix; Based on the violation probability value, deviation degree, and matching degree, a violation probability vector containing a timestamp is generated and transmitted to the decision module.

7. The intelligent monitoring and identification system for violations at a power transmission line construction site according to claim 6, characterized in that: The spatial offset determination module performs the following operations: Receive the ground bearing deformation and dynamically adjust the safety boundary coordinates of the tower base operation area; Recalculate the safe operating radius of the lifting equipment based on the wind speed gradient; When the skeletal joint point displacement sequence enters the adjusted safety boundary coordinate range, the person's out-of-bounds offset is generated; When the hydraulic arm angle change exceeds the safe operating radius, the device posture abnormality parameters are output; The personnel crossing-boundary offset and the equipment posture abnormality parameters are combined into spatial offset parameters and transmitted to the decision module.

8. The intelligent monitoring and identification system for violations at a power transmission line construction site according to claim 7, characterized in that: The processing logic of the decision module includes: When the probability of not wearing safety equipment in the violation probability vector exceeds the set threshold and the person's crossing-border offset is positive, a level 1 alarm instruction is generated; When the abnormal posture parameters of the equipment conflict with the wind-resistant overturning safety factor, a secondary braking command is output; If the equipment grounding status identifier is abnormal and the arc damage prevention assessment matrix triggers an alarm, the third-level power-off command is initiated; The first-level alarm instruction, second-level braking instruction, and third-level power-off instruction containing the alarm level, target device number, and geographic coordinates are encapsulated as multi-level response instructions and transmitted to the execution control module.

9. The intelligent monitoring and identification system for violations at a power transmission line construction site according to claim 8, characterized in that: The execution control module performs the following operations: Parse the first-level alarm command to drive the sound and light alarm corresponding to the target device number; Responding to the secondary braking command, controlling the hydraulic arm locking mechanism to intervene; Execute the third-level power-off command to cut off the power supply circuit at the specified geographic coordinates; At the same time, the triggering time of the sound and light alarm, the recorded value of the hydraulic arm locking angle, and the power-off operation timestamp are fed back to the feedback correction module.

10. The intelligent monitoring and identification system for violations at a power transmission line construction site according to claim 9, characterized in that: The operations of the feedback correction module include: Compare the triggering time of the sound and light alarm with the timestamp deviation of the violation probability vector; Analyze the correlation between the recorded values ​​of the hydraulic arm locking angle and the abnormal equipment posture parameters; Verify the response speed of the arc damage prevention assessment matrix based on the power-off operation timestamp; The safety margin coefficient in the violation behavior identification model is updated based on timestamp deviation, correlation analysis results, and response speed verification data.

Images