Stage suspender safety detection method and device based on feature recognition
By employing multi-source sensor fusion technology and differentiated detection logic, the problems of low accuracy and non-real-time detection of stage hoisting rods have been solved, enabling high-precision, real-time safety status assessment and improving the automation and safety of the detection process.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- XIAN HONGYUAN VIDEO EQUIP CO LTD
- Filing Date
- 2026-01-23
- Publication Date
- 2026-05-08
AI Technical Summary
Existing technologies for stage rigging inspection suffer from low accuracy, lack of real-time monitoring, and high subjectivity. They struggle to identify subtle deformations and hidden cracks, and the results rely on human experience, making them prone to missed or false detections.
By employing multi-source sensor fusion technology, displacement, strain, distance, and acceleration data of the boom are collected in real time. Combined with weighted fusion and threshold judgment, differentiated detection logic is designed for static and moving states to achieve high-precision and real-time safety status assessment.
It achieves real-time, high-precision, and reliable detection of booms, accurately identifying potential hazards under both static and dynamic conditions, reducing the need for manual intervention, and improving the automation and safety of the detection process.
Smart Images
Figure CN121997133A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of multi-source data processing and recognition technology, and in particular to a method for safety inspection of stage rigging using feature recognition technology based on multi-source data. Background Technology
[0002] Stage rigging is a core load-bearing component in theaters, convention centers, and other venues, used to suspend curtains, lighting, sound equipment, and other performing arts gear. Its structural integrity and installation stability directly impact performance safety. Currently, the industry commonly uses manual visual inspection, which suffers from three major problems: 1. Low inspection accuracy, failing to identify subtle deformations, hidden cracks, and other potential hazards; 2. Non-real-time inspection, only possible during performance breaks or when equipment is not in operation, making it difficult to capture safety risks under dynamic conditions; 3. High subjectivity, with inspection results relying on personnel experience, easily leading to missed or incorrect detections. Summary of the Invention
[0003] This application provides a stage rigging safety detection method and apparatus based on feature recognition to solve the above-mentioned problems in the prior art.
[0004] On the one hand, embodiments of this application provide a stage rigging safety detection method based on feature recognition, including: Collect real-time status data of the boom, including real-time displacement data, real-time strain data, real-time distance data, and real-time acceleration data; Determine whether the boom is stationary or moving based on real-time acceleration data; If the boom is stationary, calculate the difference between the real-time displacement data and the displacement calibration value as the displacement difference; calculate the difference between the real-time strain data and the strain calibration value as the strain difference; calculate the difference between the real-time distance data and the distance calibration value as the distance difference. Based on the relationship between the displacement difference, strain difference, and distance difference and the static threshold of the corresponding dimension, determine the static dimension normal number of the boom in each dimension. The real-time displacement data, real-time strain data, and real-time distance data are weighted and fused to obtain fused data. The difference between the fused data and the fused calibration value is calculated as the fusion difference. The static fusion safety status is determined based on the relationship between the fusion difference and the static fusion threshold. The safety status of the boom in a static state is determined by comprehensively considering the static dimension normals and the static fusion safety status. If the boom is in a moving state, the dynamic dimension normal number of the boom in each dimension is determined according to the relationship between the displacement difference, strain difference, distance difference and the dynamic threshold of the corresponding dimension. The dynamic fusion safety status is determined based on the relationship between the fusion difference and the dynamic fusion threshold; the dynamic fusion threshold is determined based on real-time acceleration data, and the dynamic threshold is determined based on the dynamic fusion threshold; the safety status of the boom under the moving state is judged comprehensively based on the dynamic dimension normal number and the dynamic fusion safety status.
[0005] On the other hand, embodiments of this application also provide a stage rigging safety detection device based on feature recognition, comprising: Sensors are used to collect real-time status data of the boom, including real-time displacement data, real-time strain data, real-time distance data, and real-time acceleration data. The data processing equipment is used to determine whether the boom is stationary or moving based on real-time acceleration data. If the boom is stationary, it calculates the difference between the real-time displacement data and the displacement calibration value as the displacement difference, the difference between the real-time strain data and the strain calibration value as the strain difference, and the difference between the real-time distance data and the distance calibration value as the distance difference. Based on the relationship between the displacement difference, strain difference, and distance difference and the static threshold of the corresponding dimension, it determines the static dimension normals of the boom in each dimension. The real-time displacement data, real-time strain data, and real-time distance data are then weighted and fused to obtain fused data. The difference between the fused data and the fused calibration value is calculated as the fusion difference. The static fusion safety status is determined based on the relationship between the fusion difference and the static fusion threshold. The safety status of the boom in a static state is comprehensively judged based on the static dimension normals and the static fusion safety status. If the boom is in a moving state, the dynamic dimension normals of the boom in each dimension are determined based on the relationship between the displacement difference, strain difference, and distance difference and the corresponding dynamic thresholds. The dynamic fusion safety status is determined based on the relationship between the fusion difference and the dynamic fusion threshold. The dynamic fusion threshold is determined based on real-time acceleration data. The safety status of the boom in a moving state is comprehensively judged based on the dynamic dimension normals and the dynamic fusion safety status.
[0006] The stage rigging safety detection method and device based on feature recognition disclosed in this application have the following advantages: 1. Real-time performance: The sensor sampling frequency is up to 200Hz, and the data processing delay is ≤10ms, which can capture the safety status of the boom under static and dynamic working conditions in real time.
[0007] 2. High precision: It adopts multi-source sensor fusion technology, combined with multi-dimensional comparison and fusion verification, and the detection accuracy reaches the millimeter level (displacement) and micro-strain level (strain), which is far higher than manual visual inspection.
[0008] 3. High adaptability: Different detection logic is designed for two working conditions: static and mobile. The threshold is dynamically adjusted in the mobile state to adapt to different working scenarios of the boom.
[0009] 4. High reliability: It has mechanisms such as sensor redundancy, data anomaly processing, and equipment fault alarm to avoid detection failure caused by a single fault.
[0010] 5. Ease of use: High degree of automation, no manual intervention required, intuitive alarm information, traceable maintenance records, and reduced operating costs. Attached Figure Description
[0011] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0012] Figure 1 This is a flowchart illustrating a stage rigging safety detection method based on feature recognition, provided in an embodiment of this application. Detailed Implementation
[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0014] Figure 1 This is a flowchart illustrating a stage rigging safety detection method based on feature recognition, provided in an embodiment of this application. The embodiment of this application provides a stage rigging safety detection method based on feature recognition, comprising: S100 collects real-time status data of the boom, including real-time displacement data, real-time strain data, real-time distance data, and real-time acceleration data.
[0015] For example, the real-time displacement data, real-time strain data, real-time distance data, and real-time acceleration data are acquired by an industrial high-definition camera, a fiber optic strain sensor, a laser rangefinder, and a triaxial accelerometer, respectively.
[0016] Specifically, one camera is installed on each of the pillars on both sides of the stage, with the lens horizontally aligned with the entire length of the rigging rod. The installation height is level with the midpoint of the rigging rod, and the installation position avoids the areas where the curtain and lighting fixtures are suspended to ensure unobstructed views. The two cameras have complementary shooting angles and cover all key stress sections of the rigging rod. Two strain sensors are attached to each of the three key stress points of the boom: ① the midpoint of the boom, i.e., the position of maximum deflection; ② the connection points between the boom and the truss at both ends, i.e., the stress concentration points; ③ 500mm below the suspension point, i.e., the load transfer point. Before attaching, the surface of the boom needs to be sanded and cleaned to ensure that the sensor is tightly attached to the boom and to avoid vibration causing it to fall off.
[0017] The accelerometer is fixed to the center of gravity of the boom with bolts, and the sensor's sensitive axis is aligned with the length of the boom to ensure that the true acceleration of the boom's movement is captured.
[0018] The laser sensors are installed on the truss at the top of the stage, with one sensor installed on each rigging rod. The lens is vertically downward and aligned with the midpoint of the rigging rod. The installation position must be firmly fixed to avoid truss vibration affecting the measurement accuracy.
[0019] S110 determines whether the boom is stationary or moving based on real-time acceleration data.
[0020] For example, a preset movement state determination threshold is first loaded. a 0 = 0.1 m / s 2 When the absolute value of the combined acceleration is | a |≥ a At time 0, the boom is determined to be in a moving state, while when | a ∣< a At time 0, the boom is determined to be in a stationary state.
[0021] In the embodiments of this application, the accelerometer acquires three-dimensional acceleration data at a frequency of 200Hz, i.e. a x , a y , a z Calculate the magnitude of the resultant acceleration:
[0022] If 5 consecutive sampling periods T s The resultant acceleration within the range all satisfy a < a If 0, it is determined to be in a static state. If the combined acceleration of any one sampling period satisfies a ≥ a 0 indicates a moving state.
[0023] Furthermore, when the boom changes from stationary to moving or vice versa, the system will clear the detection data cache of the previous state and restart the detection process of the corresponding state to avoid cross-state data interference.
[0024] S120, if the boom is in a static state, calculate the difference between the real-time displacement data and the displacement calibration value as the displacement difference, calculate the difference between the real-time strain data and the strain calibration value as the strain difference, calculate the difference between the real-time distance data and the distance calibration value as the distance difference, and determine the static dimension normal number of the boom in each dimension according to the relationship between the displacement difference, strain difference, distance difference and the static threshold of the corresponding dimension.
[0025] For example, the system sends acquisition commands to the camera, strain sensor, and laser sensor, while the accelerometer continuously acquires data for status monitoring. Each sensor acquires data at a sampling period. T s Data was collected at 0.01s, resulting in a sequence of real-time status data: Real-time displacement data sequence { x ( t Real-time strain data sequence { y ( t Real-time distance data sequence { z ( t )}, t Indicates the time.
[0026] Denoising real-time status data:
[0027]
[0028]
[0029] in m =5, x raw ( k ), y raw ( k )and z raw ( k These are the raw real-time data for displacement, strain, and distance dimensions, respectively.
[0030] Align sensor data based on timestamps to ensure that they are synchronized at the same time. t corresponding x ( t ), y ( t ), z ( t A one-to-one correspondence is established. If data from a certain sensor is missing, i.e., no data is received for three consecutive sampling periods, the following redundancy mechanism is activated: If displacement data is missing: calculate the compensated displacement using laser distance data. x′ = z 0 z ( t ),in z 0 represents the distance calibration value; If strain data is missing: calculate the compensated strain using displacement data and hanger structural parameters. ,in, E The elastic modulus of the hanger material. I Let be the moment of inertia of the section of the boom. L This refers to the total length of the boom; If distance data is missing: calculate the compensated distance using displacement data. z′ = z 0 x ( t ).
[0031] Furthermore, the methods for generating displacement calibration values, strain calibration values, and distance calibration values include: The control boom is kept stationary, and the original calibration data sequence of the boom is continuously collected. The original calibration data sequence includes camera displacement sequence, strain sensor sequence and laser sensor sequence. The original calibration data sequence is denoised. The average value of the original calibration data sequence after denoising is taken in each dimension and used as the calibration value for each dimension.
[0032] Specifically, first ensure the stage environment is free from vibration and strong light interference, and that the room temperature is controlled at 20±5℃ to avoid thermal deformation caused by temperature changes. Then, start all sensors and run them continuously for 30 minutes. Once the sensors are stable, begin collecting the original calibration data sequence.
[0033] With sampling period T s =0.01s, that is, data is continuously collected at 100Hz for 10 minutes to obtain the original calibration data sequence of each sensor: camera displacement sequence { x 0,raw ( t )}、Strain sensor sequence{ y 0,raw ( t )}、Laser sensor sequence{ z 0,raw ( t )}.
[0034] The denoising process uses a moving average filtering algorithm to denoise the original calibration data, as shown in the following formula:
[0035] in, m=5 indicates that the sliding window size is set to 5 sampling points to balance the denoising effect and the real-time performance of the data. x 0( t ), y 0( t )and z 0( t () indicates the data after noise reduction.
[0036] The average value of the denoised calibration data sequence is taken as the calibration value for each dimension:
[0037]
[0038]
[0039] in, N Indicates the total number of sampling points. x 0、 y 0 and z 0 represents the calibration values for displacement, strain, and distance dimensions, respectively.
[0040] Extracting the maximum value of the denoised data , , This is used for standardized processing during subsequent multi-source data fusion.
[0041] Furthermore, methods for generating static thresholds include: Calculate the standard deviation of the original calibration data sequence in each dimension; The three-standard-deviation method is used to take three times the standard deviation of each dimension as the static threshold of the corresponding dimension.
[0042] Specifically, the static state threshold is determined based on the statistical characteristics of the calibration data, using the "3-standard deviation method" with a 99.73% confidence interval to eliminate the influence of random errors.
[0043] Calculate the standard deviation of the original calibration data series for each dimension. s 1 (displacement) s 2 (Strain) s 3 (distance), threshold is: Δ L 11 =3 s 1, Δ L 12 =3 s 2, Δ L 13 =3 s 3 Where, Δ L 11Δ L 12 and Δ L 13 These are the static thresholds for displacement, strain, and distance dimensions, respectively.
[0044] For the displacement dimension, first calculate the displacement difference Δ x ( t = | x ( t ) x 0 |, if Δ x ( t )<Δ L 11 If the displacement dimension is normal, then the displacement dimension is considered normal; if Δ x ( t )≥Δ L 11 If so, the displacement dimension is determined to be abnormal; For the strain dimension, first calculate the strain difference Δ y ( t = | y ( t ) y 0 |, if Δ y ( t )<Δ L 12 If the strain dimension is normal, then the strain dimension is considered normal; if Δ y ( t )≥Δ L 12 If so, the strain dimension is determined to be abnormal; For the distance dimension, first calculate the distance difference Δ z ( t = | z ( t ) z 0 |, if Δ z ( t )<Δ L 13 If the distance dimension is normal, then the distance dimension is considered normal; if Δ z ( t )≥Δ L 13 If the distance dimension is not specified, then it is considered an anomaly.
[0045] After determining the state of each dimension, the number of dimensions belonging to the normal state is counted to obtain the static dimension normality number. Clearly, the maximum static dimension normality number is 3, and the minimum is 0.
[0046] S130: Weighted fusion of real-time displacement data, real-time strain data and real-time distance data is performed to obtain fused data. The difference between the fused data and the fused calibration value is calculated as the fusion difference. Based on the relationship between the fusion difference and the static fusion threshold, the static fusion safety state is determined.
[0047] For example, a method for computing fused data includes: Extract the maximum value from the original calibration data sequence for each dimension; Calculate the ratio of the calibration value to the corresponding maximum value for each dimension; The ratios of each dimension are weighted and summed according to their respective weights to obtain the fused data.
[0048] Specifically, in order to improve detection accuracy and avoid misjudgment by a single sensor, this application standardizes and weights the preprocessed real-time data.
[0049] Data standardization: Converting real-time data from various dimensions into dimensionless data to eliminate unit differences.
[0050] in, x norm ( t ), y norm ( t )and z norm ( t ) represent the standardized displacement, strain, and distance dimensions, respectively.
[0051] Weighted fusion calculation: F ( t )= w 1 x norm ( t )+ w 2 y norm ( t )+ w 3 z norm ( t ) in, F ( t To integrate data, w 1. w 2 and w 3 represents the fusion weights for displacement, strain, and distance dimensions, respectively.
[0052] Furthermore, the methods for generating the static fusion threshold include: Calculate the standard deviation of the fused data; The three-times-standard-deviation method is used to take three times the standard deviation of the fused data as the static fusion threshold.
[0053] Specifically, first calculate the fused data. F ( t Then calculate its standard deviation. s F Then the static fusion threshold Δ L 0=3 s F .
[0054] The fusion calibration value adopts the weighted fusion method, and the formula is as follows:
[0055] The weights satisfy w 1+ w 2+ w 3=1, in the embodiments of this application, w 1 = 0.3 w 2 = 0.3, w 3 = 0.4.
[0056] For fused data, first calculate the fusion difference Δ F ( t = | F ( t ) F 0 | If Δ F ( t )<Δ L If 0, the fusion dimension is considered normal; if Δ F ( t )≥Δ L If the value is 0, the fusion dimension is considered abnormal.
[0057] S140, the safety status of the boom in a static state is judged by comprehensively considering the static dimension normal numbers and the static fusion safety status.
[0058] For example, when the boom is stationary, in each sampling period, the boom's state is defined as normal when the distance difference, strain difference, or distance threshold is less than the static threshold of the corresponding dimension, and the number of dimensions belonging to the normal state is counted as the static dimension normal number; the boom's state is defined as normal when the fusion difference is less than the static fusion threshold; if the static dimension normal number is equal to 3 and the static fusion safety state is normal, then the boom's safety state in the stationary state is normal; if the static dimension normal number is equal to 2 and the static fusion safety state is normal, or if the static dimension normal number is equal to 3 and the static fusion safety state is abnormal, then the boom's safety state in the stationary state is a warning state; if the static dimension normal number is less than or equal to 1, or if the static dimension normal number is less than or equal to 2 and the static fusion safety state is abnormal, then the boom's safety state in the stationary state is a dangerous state.
[0059] Specifically, for each sampling period T s Statistical analysis of static dimensions and normal numbers C ( t ) and static fusion security state SF ( t The value is 0 for normal and 1 for abnormal. The determination result for a single sampling period is as follows: normal: C ( t )=3 and SF ( t )=0; Warning: C ( t )≥2 and SF ( t )=0, or C ( t )=3 and SF ( t )=1; Danger: C ( t )≤1 or SF ( t )=1 and C ( t )≤2.
[0060] S150, if the boom is in a moving state, determine the dynamic dimension normal number of the boom in each dimension according to the relationship between the displacement difference, strain difference, and distance difference and the dynamic threshold of the corresponding dimension. The dynamic threshold is determined according to the dynamic fusion threshold.
[0061] For example, the dynamic threshold is calculated according to the following formula:
[0062]
[0063]
[0064] in, , and These are the dynamic thresholds in the displacement, strain, and distance dimensions, respectively. , and These are the static thresholds in the displacement, strain, and distance dimensions, respectively, Δ L ( t () represents the dynamic fusion threshold.
[0065] Specifically, the dynamic threshold and dynamic fusion threshold of this application also have threshold constraints, that is, setting upper limits for the dynamic threshold and dynamic fusion threshold to prevent excessive acceleration from causing the threshold to increase indefinitely. These upper limits are set to... , , i The dimension label is used; if the calculation result exceeds the upper limit, the upper limit value is used.
[0066] For the displacement dimension, first calculate the displacement difference Δ x ( t = | x ( t ) x 0 |, if If the displacement dimension is normal, then the displacement dimension is considered normal; if If so, the displacement dimension is determined to be abnormal; For the strain dimension, first calculate the strain difference Δ y ( t = | y ( t ) y 0 |, if If the strain dimension is judged to be normal; if If so, the strain dimension is determined to be abnormal; For the distance dimension, first calculate the distance difference Δ z ( t = | z ( t ) z 0 |, if If the distance dimension is considered normal; if If the distance dimension is not specified, then it is considered an anomaly.
[0067] S160, determine the dynamic fusion safety status based on the relationship between the fusion difference and the dynamic fusion threshold; the dynamic fusion threshold is determined based on real-time acceleration data; and the safety status of the boom under moving conditions is comprehensively judged based on the dynamic dimension normal number and the dynamic fusion safety status.
[0068] For example, a method for generating a dynamic fusion threshold includes: Smooth the real-time acceleration data to obtain smoothed acceleration; The mapping coefficients are calculated based on multiple different accelerations observed in the dynamic experiment.
[0069] in, k Let Δ be the mapping coefficient. L’ For each acceleration a’ The corresponding allowable fusion deviation, Δ L 0 represents the static fusion threshold; The dynamic fusion threshold is determined based on the static fusion threshold, mapping coefficient, and smoothing acceleration.
[0070] Where, Δ L ( t () represents the dynamic fusion threshold. a smoothed ( t () represents smooth acceleration.
[0071] Specifically, embodiments of this application determine the mapping coefficient between acceleration and dynamic fusion threshold through dynamic experiments. k The specific process is as follows: control the boom to accelerate at different speeds. a’ =0.2,0.4,...,2.0m / s 2 Move the device and collect normal data under various accelerations, i.e., deformation data when there are no safety hazards. For each acceleration... a’ Calculate the corresponding allowable fusion deviation Δ L’ Obtained through linear fitting k The value is taken as the average of multiple fitting results as the final value.
[0072] Furthermore, the accelerometer acquires three-dimensional acceleration data at a frequency of 200Hz to calculate the real-time composite acceleration. a ( t Then, a moving average filter is used to smooth the acceleration data to remove high-frequency noise.
[0073] After obtaining the smoothed acceleration, its validity needs to be verified. asmoothed ( t >10m / s 2 This indicates that the sensor's measurement limit has been exceeded, which is determined to be an abnormal acceleration data, triggering an acceleration data abnormality alarm, and replacing it with the average acceleration value from the previous cycle.
[0074] After obtaining the dynamic fusion threshold, based on the fusion difference Δ F ( t Perform state judgment on the fusion dimension, if Δ F ( t )<Δ L ( t If Δ is true, then the fusion dimension is judged as normal; if Δ is false, then the fusion dimension is judged F ( t )≥Δ L ( t If the fusion dimension is not found to be abnormal, then the fusion dimension is considered abnormal.
[0075] Furthermore, when the boom is in motion, in each sampling period, the boom's state is defined as normal when the distance difference, strain difference, or distance threshold is less than the dynamic threshold of the corresponding dimension, and the number of dimensions belonging to the normal state is counted as the dynamic dimension normal number; the boom's state is defined as normal when the fusion difference is less than the dynamic fusion threshold; if the dynamic dimension normal number is equal to 3 and the dynamic fusion safety state is normal, then the boom's safety state in motion is normal; if the dynamic dimension normal number is equal to 2 and the dynamic fusion safety state is normal, or if the dynamic dimension normal number is equal to 3 and the dynamic fusion safety state is abnormal, then the boom's safety state in motion is a warning state; if the dynamic dimension normal number is less than or equal to 1, or if the dynamic dimension normal number is less than or equal to 2 and the dynamic fusion safety state is abnormal, then the boom's safety state in motion is a dangerous state.
[0076] Specifically, consistent with the static state, based on the dynamic dimension positive constants C’ ( t and dynamic fusion security status SF' ( t The result of a single sampling period is determined as normal / warning / danger.
[0077] Furthermore, after each testing cycle, the number of times the boom is in normal, warning, and dangerous states is counted, the safety judgment confidence level is calculated, and the comprehensive judgment result of the boom in static or moving states is determined based on the safety judgment confidence level and the proportion of dangerous states in the total number of times. The testing cycle includes multiple sampling cycles.
[0078] Specifically, in a stationary or moving state, during the detection cycle M Within 1000 sampling points, the number of normal occurrences was counted.N normal Number of warnings N warn Number of dangerous times N danger Calculate the confidence level of the security decision:
[0079] Based on confidence level and frequency percentage, the comprehensive judgment result of the static state is given: Normal state: i ≥ i 1 and N danger =0, where the first confidence threshold i 1 = 0.9; Warning status: i 2≤ i < i 1 and N danger <0.05 M The second confidence threshold i 2 = 0.7; Dangerous situation: i <θ2 or N danger ≥0.05 M .
[0080] Furthermore, when the boom is in motion, the real-time position of the boom is calculated based on the boom's initial position, initial velocity, and real-time acceleration data. The real-time position is then corrected using real-time distance data to obtain the corrected position of the boom. During the boom's movement, real-time status data of the corrected position is collected.
[0081] Specifically, a quadratic integration method is used, based on the initial position. s 0. Initial velocity v 0 and real-time acceleration to calculate real-time position s calc ( t ):
[0082] Where the initial position s 0 and initial velocity v 0 represents the measurement value at the start of the detection cycle.
[0083] Then, the correction value Δ is calculated using the real-time distance data from the laser sensor. s ( t )= z 0 z ( t The corrected position is: s ( t )= s calc ( t )+Δ s ( t ) This correction compensates for the cumulative error of acceleration integration, ensuring a positional accuracy of ≤ ±0.5 mm.
[0084] The system receives the judgment results of each detection cycle in real time under stationary / moving conditions. If any detection cycle indicates a warning but no danger, a safety warning is output, triggering an audible and visual warning, such as a flashing green light and intermittent buzzer alarm, and recording the warning time, location, and deviation data. If any detection cycle indicates danger, a danger status is output, triggering a three-level alarm according to the following strategy: Level 1 alarm: The red light on the audible and visual alarm is constantly on, and the buzzer sounds continuously. Level 2 alarm: Sends a remote notification to the stage management personnel's mobile phone, which includes the dangerous location and deviation data; Level 3 alarm: Triggers an emergency stop signal, controls the boom to stop moving and locks it in the current position to prevent the risk from escalating.
[0085] Safety warnings require manual re-inspection to confirm the absence of hidden dangers before being manually deactivated. Dangerous conditions require maintenance personnel to repair the boom and recalibrate it before the system can perform a self-check to deactivate the warning.
[0086] When data anomalies occur, the following processing procedure is adopted: Data loss handling: If data from a certain sensor is missing, that is, no data is received for 3 consecutive sampling cycles, the system will automatically activate the redundancy mechanism and replace it with data from other sensors. At the same time, the missing data will be recorded and the user will be reminded to check for maintenance. Data mutation handling: If the change in a certain dimension of data exceeds 50% of the maximum value of the calibrated data within one sampling period, it is determined to be a data mutation. At this time, the data point is removed, the average value of the data before and after is used as a replacement, and a data mutation alarm is triggered. Threshold anomaly handling: If the calculated result of the dynamic fusion threshold exceeds the upper limit, for example Δ L ( t )>2Δ L 0. The system automatically sets the threshold to the upper limit and records any abnormal acceleration conditions, reminding the user to check whether the boom movement is normal.
[0087] When equipment failure occurs, the following procedures should be followed: Sensor Failure: If a sensor fails during self-testing or detection, the system will automatically block the sensor data, activate the redundancy mechanism, and trigger a sensor fault alarm, displaying the fault location and type. Communication failure: If the communication link is interrupted, the system will attempt to reconnect. If the connection fails, a communication failure alarm will be triggered, and the collected data will be saved to prevent data loss. Storage failure: If there is insufficient storage space or read / write failure, the system will automatically delete the oldest historical data. If the problem persists, a storage failure alarm will be triggered, data storage will be stopped, but the detection process will continue.
[0088] This application also provides a stage rigging safety detection device based on feature recognition, the device comprising: Sensors are used to collect real-time status data of the boom, including real-time displacement data, real-time strain data, real-time distance data, and real-time acceleration data. The data processing equipment is used to determine whether the boom is stationary or moving based on real-time acceleration data. If the boom is stationary, it calculates the difference between the real-time displacement data and the displacement calibration value as the displacement difference, the difference between the real-time strain data and the strain calibration value as the strain difference, and the difference between the real-time distance data and the distance calibration value as the distance difference. Based on the relationship between the displacement difference, strain difference, and distance difference and the static threshold of the corresponding dimension, it determines the static dimension normals of the boom in each dimension. The real-time displacement data, real-time strain data, and real-time distance data are then weighted and fused to obtain fused data. The difference between the fused data and the fused calibration value is calculated as the fusion difference. The static fusion safety status is determined based on the relationship between the fusion difference and the static fusion threshold. The safety status of the boom in a static state is comprehensively judged based on the static dimension normals and the static fusion safety status. If the boom is in a moving state, the dynamic dimension normals of the boom in each dimension are determined based on the relationship between the displacement difference, strain difference, and distance difference and the corresponding dynamic thresholds. The dynamic fusion safety status is determined based on the relationship between the fusion difference and the dynamic fusion threshold. The dynamic fusion threshold is determined based on real-time acceleration data. The safety status of the boom in a moving state is comprehensively judged based on the dynamic dimension normals and the dynamic fusion safety status.
[0089] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.
[0090] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A stage rigging safety detection method based on feature recognition, characterized in that, include: Collect real-time status data of the boom, including real-time displacement data, real-time strain data, real-time distance data, and real-time acceleration data; The real-time acceleration data is used to determine whether the boom is stationary or moving. If the boom is stationary, calculate the difference between the real-time displacement data and the displacement calibration value as the displacement difference; calculate the difference between the real-time strain data and the strain calibration value as the strain difference; calculate the difference between the real-time distance data and the distance calibration value as the distance difference; and determine the static dimension normal number of the boom in each dimension based on the relationship between the displacement difference, the strain difference, the distance difference, and the static threshold of the corresponding dimension. The real-time displacement data, the real-time strain data, and the real-time distance data are weighted and fused to obtain fused data. The difference between the fused data and the fused calibration value is calculated as the fused difference. Based on the relationship between the fused difference and the static fused threshold, the static fused safety state is determined. The safety status of the boom in a static state is determined by comprehensively considering the static dimension normal numbers and the static fusion safety status. If the boom is in a moving state, the dynamic dimension normal number of the boom in each dimension is determined according to the relationship between the displacement difference, the strain difference, the distance difference and the dynamic threshold of the corresponding dimension. The dynamic fusion safety status is determined based on the relationship between the fusion difference and the dynamic fusion threshold; the dynamic fusion threshold is determined based on the real-time acceleration data, and the dynamic threshold is determined based on the dynamic fusion threshold; the safety status of the boom under the moving state is comprehensively judged based on the dynamic dimension normal number and the dynamic fusion safety status.
2. The stage rigging safety detection method based on feature recognition according to claim 1, characterized in that, The methods for generating the displacement calibration value, the strain calibration value, and the distance calibration value include: The boom is kept stationary while the original calibration data sequence of the boom is continuously collected. The original calibration data sequence includes camera displacement sequence, strain sensor sequence and laser sensor sequence. The original calibration data sequence is then denoised. The average value of the original calibration data sequence after denoising is taken in each dimension and used as the calibration value for each dimension.
3. The stage rigging safety detection method based on feature recognition according to claim 2, characterized in that, The method for calculating the fused data includes: Extract the maximum value from the original calibration data sequence for each dimension; Calculate the ratio of the calibration value to the corresponding maximum value for each dimension; The fused data is obtained by weighting and summing the ratios of each dimension according to their respective weights.
4. The stage rigging safety detection method based on feature recognition according to claim 2, characterized in that, The method for generating the static threshold includes: Calculate the standard deviation of the original calibration data sequence in each dimension; The three-standard-deviation method is used to take three times the standard deviation of each dimension as the static threshold of the corresponding dimension; The method for generating the static fusion threshold includes: Calculate the standard deviation of the fused data; The static fusion threshold is determined by using the 3x standard deviation method, where 3 times the standard deviation of the fused data is taken as the threshold.
5. The stage rigging safety detection method based on feature recognition according to claim 1, characterized in that, The method for generating the dynamic fusion threshold includes: The real-time acceleration data is smoothed to obtain smoothed acceleration; The mapping coefficients are calculated based on multiple different accelerations observed in the dynamic experiment. in, k Let Δ be the mapping coefficient. L i For each acceleration a i The corresponding allowable fusion deviation, Δ L 0 represents the static fusion threshold; The dynamic fusion threshold is determined based on the static fusion threshold, the mapping coefficient, and the smoothing acceleration. Where, Δ L ( t ) is the dynamic fusion threshold. a smoothed ( t ) represents the smooth acceleration.
6. The stage rigging safety detection method based on feature recognition according to claim 5, characterized in that, The dynamic threshold is calculated according to the following formula: in, , and These are the dynamic thresholds in the displacement, strain, and distance dimensions, respectively. , and These are the static thresholds in the displacement dimension, strain dimension, and distance dimension, respectively.
7. The stage rigging safety detection method based on feature recognition according to claim 1, characterized in that, When the boom is stationary, in each sampling period, the boom's state is defined as normal when the distance difference, strain difference, or distance threshold is less than the static threshold of the corresponding dimension, and the number of dimensions belonging to the normal state is counted as the static dimension normal number; the boom's state is defined as normal when the fusion difference is less than the static fusion threshold; if the static dimension normal number is equal to 3 and the static fusion safety state is normal, then the boom's safety state in the stationary state is normal; if the static dimension normal number is equal to 2 and the static fusion safety state is normal, or if the static dimension normal number is equal to 3 and the static fusion safety state is abnormal, then the boom's safety state in the stationary state is a warning state; if the static dimension normal number is less than or equal to 1, or if the static dimension normal number is less than or equal to 2 and the static fusion safety state is abnormal, then the boom's safety state in the stationary state is a dangerous state. When the boom is in motion, in each sampling period, the boom's state is determined to be normal when the distance difference, strain difference, or distance threshold is less than the dynamic threshold of the corresponding dimension, and the number of dimensions belonging to the normal state is counted as the dynamic dimension normal number; the boom's state is determined to be normal when the fusion difference is less than the dynamic fusion threshold; if the dynamic dimension normal number is equal to 3 and the dynamic fusion safety state is normal, then the boom's safety state in motion is normal; if the dynamic dimension normal number is equal to 2 and the dynamic fusion safety state is normal, or if the dynamic dimension normal number is equal to 3 and the dynamic fusion safety state is abnormal, then the boom's safety state in motion is a warning state; if the dynamic dimension normal number is less than or equal to 1, or if the dynamic dimension normal number is less than or equal to 2 and the dynamic fusion safety state is abnormal, then the boom's safety state in motion is a dangerous state.
8. The stage rigging safety detection method based on feature recognition according to claim 7, characterized in that, After each detection cycle, the number of times the boom is in normal, warning, and dangerous states is counted, the safety judgment confidence level is calculated, and the comprehensive judgment result of the boom in static or moving states is determined based on the safety judgment confidence level and the proportion of the number of dangerous states in the total number of times. The detection cycle includes multiple sampling cycles.
9. The stage rigging safety detection method based on feature recognition according to claim 1, characterized in that, When the boom is in a moving state, the real-time position of the boom is calculated based on the boom's initial position, initial velocity, and the real-time acceleration data. The real-time position is then corrected using the real-time distance data to obtain the corrected position of the boom. During the movement of the boom, real-time status data of the corrected position is collected.
10. An apparatus for applying the feature recognition-based stage rigging safety detection method according to any one of claims 1-9, characterized in that, include: Sensors are used to collect real-time status data of the boom, including real-time displacement data, real-time strain data, real-time distance data, and real-time acceleration data. A data processing device is used to determine whether the boom is stationary or moving based on the real-time acceleration data. If the boom is stationary, calculate the difference between the real-time displacement data and the displacement calibration value as the displacement difference; calculate the difference between the real-time strain data and the strain calibration value as the strain difference; calculate the difference between the real-time distance data and the distance calibration value as the distance difference. Based on the magnitude of the displacement difference, the strain difference, and the distance difference, and the static threshold of the corresponding dimension, determine the static dimension normality of the boom in each dimension. Then, weightedly fuse the real-time displacement data, real-time strain data, and real-time distance data to obtain fused data. Calculate the difference between the fused data and the fused calibration value as the fusion difference. Based on the magnitude of the fused difference and the static fusion threshold... The relationship is used to determine the static fusion safety state; the safety state of the boom in the static state is comprehensively judged based on the static dimension normality and the static fusion safety state; if the boom is in a moving state, the dynamic dimension normality of the boom in each dimension is determined based on the magnitude relationship between the displacement difference, the strain difference, the distance difference, and the dynamic threshold of the corresponding dimension; the dynamic fusion safety state is determined based on the magnitude relationship between the fusion difference and the dynamic fusion threshold; the dynamic fusion threshold is determined based on the real-time acceleration data, and the dynamic threshold is determined based on the dynamic fusion threshold; the safety state of the boom in the moving state is comprehensively judged based on the dynamic dimension normality and the dynamic fusion safety state.