Special-shaped hanging basket safety monitoring system based on internet of things

Abstract

The present application relates to the technical field of industrial monitoring, in particular to a special-shaped hanging basket safety monitoring system based on the Internet of Things. In the present application, the real-time identification of the torsion state is realized by dynamically constructing the angle sequence of the gravity vector trajectory and the displacement path synchronously collected during the vertical operation of the hanging basket, the relative comparison of the diagonal load trend is further carried out based on the corner stress value to judge whether the posture deviates from the dominant gravity center path, the spatial coordinates and load values of the operating personnel and equipment are introduced on the load distribution level to construct a two-dimensional plane mapping diagram, the trend transfer sequence is extracted from the gravity center vector trajectory and embedded into a Markov chain model to analyze the jump path and probability, thereby forming the deviation trend identification process of state transition, and then the duration interval and time length of the deviation path are extracted in the time dimension, whether it exceeds the preset duration threshold is counted, and then the running time period of the hanging basket in the abnormal posture is accurately marked.

Description

Technical Field

[0001] This invention relates to the field of industrial monitoring technology, and in particular to a safety monitoring system for irregularly shaped suspended platforms based on the Internet of Things. Background Technology

[0002] The field of industrial monitoring technology involves real-time observation, data acquisition, and safety management of industrial site operations. Core aspects include sensor deployment, signal transmission, data analysis, and alarm response, with the aim of ensuring the safety of industrial equipment and personnel and improving production efficiency.

[0003] Among them, the traditional irregular suspended platform safety monitoring system refers to the system used in the construction of building curtain walls and irregular structures to monitor parameters such as the operating status, load distribution, and tilt angle of the suspended platform, as well as to monitor and identify anomalies in the operating status of irregular suspended platforms in complex construction scenarios.

[0004] Traditional safety monitoring systems for irregularly shaped suspended platforms rely on independent monitoring of parameters such as the platform's operating status, load distribution, and tilt angle. Lacking a multi-dimensional data fusion mechanism, they cannot generate dynamic trend trajectories in attitude monitoring. This makes it difficult to promptly identify anomalies when slight torsion or minor shifts caused by asymmetrical loads occur. Especially in scenarios with frequent personnel movement or dynamic changes in equipment distribution, rapid shifts in the center of gravity can easily be misjudged or missed by the system. Furthermore, these systems lack effective trend evolution modeling paths and in-depth analysis of load center of gravity changes over multiple time periods, making it difficult to identify attitude hazards caused by persistent shifts. For example, when the load on a corner of the suspended platform gradually increases and lasts for a long time, but the single-point deviation does not exceed the standard, the system cannot identify the potential risk, resulting in a failure to provide effective early warning in the early stages of tilt instability, thus creating potential safety hazards. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing an IoT-based safety monitoring system for irregularly shaped suspended platforms.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: An IoT-based safety monitoring system for irregularly shaped suspended platforms includes:

[0007] Torsional posture recognition module: synchronously collects the gravity vector trajectory and real-time displacement path of the suspended platform at each time period during the vertical movement of the suspended platform through the Internet of Things communication link, identifies the torsional posture of the suspended platform, and obtains the torsional posture recognition result;

[0008] Attitude deviation verification module: acquires the stress values ​​of the four corner points of the suspended platform over a specified time period, calculates the load changes of the corresponding diagonal corners, and performs counter-verification in combination with the torsional attitude recognition results to obtain the attitude deviation judgment result.

[0009] Load center of gravity calculation module: Based on the attitude offset determination result, collect the real-time coordinates of the operator and equipment and the corresponding equivalent load value, and calculate the load center of gravity to obtain the load center of gravity sequence;

[0010] Trend path identification module: Input the load center of gravity sequence into the Markov chain model, identify the state transition path of the load center of gravity in multiple time periods, compare it with the state transition probability threshold, and filter the offset trend sequence;

[0011] Attitude Anomaly Monitoring Module: Calculates the duration of continuous offset segments in the offset trend sequence, compares it with the offset trend duration threshold, determines whether the attitude state of the suspended platform is in an abnormal operating state, and uploads the attitude stability monitoring results to the cloud monitoring platform.

[0012] As a further aspect of the present invention, the torsional attitude recognition result includes the fitting error of the included angle change sequence, the attitude rotation direction recognition type, and the attitude torsional amplitude level. The attitude offset judgment result specifically includes the diagonal load difference characteristics, the load change trend direction, and the verification judgment label. The load centroid sequence includes the centroid plane coordinates, the centroid position offset vector, and the load distribution time sequence label. The offset trend sequence specifically includes the state transition path sequence, the start and end time of the offset duration, and the state transition probability value. The attitude stability monitoring result includes the abnormal attitude judgment state, the duration of continuous offset, and the monitoring time period identifier.

[0013] As a further aspect of the present invention, the torsional posture recognition module includes:

[0014] Gravity vector acquisition submodule: Acquires acceleration vector data from the three-axis accelerometers deployed along the hoist's running path for each time period, transmits it synchronously to the data receiving end via the IoT communication link, calculates the direction angle corresponding to each acceleration vector and arranges them in chronological order to obtain the sequence of gravity direction changes;

[0015] Displacement path analysis submodule: Based on the gravity direction change sequence, obtain the displacement vector information of each time period recorded by the displacement encoder, construct the angle sequence by calculating the angle between the gravity direction vector of the basket and the displacement path vector of the basket, perform a linear fitting operation on the sequence, and identify the angle fitting offset trend;

[0016] The attitude torsion judgment submodule analyzes the slope trend of the angle change over a continuous time period based on the fitting offset trend of the included angle, determines whether there is a continuous offset segment, and generates the torsion attitude recognition result by combining the trend span with the time interval.

[0017] As a further aspect of the present invention, the attitude offset verification module includes:

[0018] Stress data acquisition submodule: acquires stress output data from stress sensors deployed at the four corners of the suspended platform within a specified time period, categorizes the stress value corresponding to each corner point according to four identifiers: left rear, right front, right rear, and left front, and generates a set of corner stress measurement values.

[0019] Load difference calculation submodule: Based on the set of corner stress measurements, construct two load combinations for the corresponding diagonal regions of left rear-right front and right rear-left front, calculate the stress difference sequence corresponding to the time period within each diagonal combination, obtain the relative trend of the difference change between the two combinations, and generate the diagonal load change trend.

[0020] Opposite Consistency Verification Submodule: Based on the diagonal load change trend and the torsional attitude recognition result, compare whether there is an interleaved shift in the trend direction within the same time period, determine whether the attitude deviates from the dominant path of the center of gravity, and output the attitude shift determination result.

[0021] As a further aspect of the present invention, the load centroid calculation module includes:

[0022] Operational load data acquisition submodule: Based on the attitude deviation determination result, it acquires the coordinate information and equivalent load value of the operators and equipment for the corresponding time period recorded in the UWB positioning tag and RFID identification tag configured on the suspended platform, and obtains the operational load coordinate value pair set;

[0023] Load distribution construction submodule: Based on the set of work load coordinate values, establish a spatial distribution mapping of work load in a two-dimensional plane, and map the load values ​​to the plane distribution map according to the coordinate position to obtain the load position mapping map;

[0024] The centroid coordinate extraction submodule calls the load position mapping map, uses coordinate weights and load values ​​as calculation factors to calculate the composite centroid coordinates of the suspended platform in the two-dimensional plane in the current time period, and outputs the centroid positions in the continuous time period as sequence data to generate a load centroid sequence.

[0025] As a further aspect of the present invention, the trend path identification module includes:

[0026] State partitioning construction submodule: Call the load centroid sequence, divide the discrete state space according to the plane coordinate range of the two-dimensional platform, and place all centroid points into the corresponding blocks according to their positions to generate a load centroid state distribution column.

[0027] Transition probability calculation submodule: Based on the load centroid state distribution column, the Markov chain model is applied to calculate the transition frequency between adjacent states, identify the jump probability change between each path node, and obtain the state transition probability set;

[0028] Offset path filtering submodule: Based on the set of state transition probabilities, the jump probability of each state path is filtered, the transition path segments that are continuously higher than the state transition probability threshold are identified and extracted as offset trend data, and an offset trend sequence is generated.

[0029] As a further aspect of the present invention, the attitude anomaly monitoring module includes:

[0030] Trend Continuity Statistics Submodule: Calls the offset trend sequence, sequentially analyzes the start and end ranges of continuous offset segments in the time dimension, calculates the corresponding duration data according to the time span of each segment, and establishes a statistical index for all segments to obtain continuous offset duration groups.

[0031] Abnormal state judgment submodule: Based on the continuous offset duration group, compare each duration segment item by item, identify whether there are continuous segments that are greater than the offset trend duration threshold, mark the state identifier of the corresponding time period, and generate an abnormal posture segment marker set;

[0032] Monitoring result upload submodule: Based on the abnormal posture segment marker set, construct the time-series monitoring result data structure of the hanging basket posture state, encapsulate it into a structured transmission format, and upload it to the cloud monitoring platform through the Internet of Things link to generate posture stability monitoring results.

[0033] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0034] In this invention, the gravity vector trajectory and displacement path collected synchronously during the vertical operation of the suspended platform are dynamically constructed into an angle sequence to achieve real-time identification of the torsional state. Based on the corner stress value, the diagonal load trend is compared to further determine whether the posture deviates from the dominant center of gravity path. The displacement trend and load distribution are effectively combined to achieve dual verification. At the load distribution level, the spatial coordinates of the workers and equipment and the load values ​​are introduced to construct a two-dimensional plane mapping map. The trend transfer sequence is extracted by the center of gravity vector trajectory and embedded into a Markov chain model to analyze its jump path and probability, thus forming a process of identifying the offset trend of state transition. Then, the duration interval and time length of the offset path are extracted from the time dimension, and it is statistically analyzed whether it exceeds the preset duration threshold. This accurately marks the time period when the suspended platform is in an abnormal posture, and the structured results are uploaded to the cloud platform for remote monitoring. The whole process connects multiple analysis links such as posture recognition, load offset verification, trend path modeling and posture anomaly recognition, forming a spatiotemporally linked suspended platform posture stability analysis chain, which greatly improves the identification accuracy, response timeliness and intelligence level of the monitoring system for abnormal states of suspended platforms in complex construction scenarios. Attached Figure Description

[0035] Figure 1This is a system flowchart of the present invention;

[0036] Figure 2 This is a flowchart of the torsional posture recognition module of the present invention;

[0037] Figure 3 This is a flowchart of the attitude offset verification module of the present invention;

[0038] Figure 4 This is a flowchart of the load center of gravity calculation module of the present invention;

[0039] Figure 5 This is a flowchart of the trend path recognition module of the present invention;

[0040] Figure 6 This is a flowchart of the attitude anomaly monitoring module of the present invention. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0042] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0043] Please see Figure 1 The IoT-based safety monitoring system for irregularly shaped suspended platforms includes:

[0044] Torsional posture recognition module: synchronously collects the gravity vector trajectory and real-time displacement path of the suspended platform at each time period during the vertical movement of the suspended platform through the Internet of Things communication link, identifies the torsional posture of the suspended platform, and obtains the torsional posture recognition result;

[0045] Attitude deviation verification module: Obtain the stress values ​​of the four corner points of the suspended platform over a specified time period, calculate the load changes of the corresponding diagonal corners, and perform counter-current verification in combination with the torsional attitude recognition results to obtain the attitude deviation judgment result.

[0046] Load center of gravity calculation module: Based on the attitude offset judgment results, it collects the real-time coordinates of the operators and equipment and the corresponding equivalent load values, calculates the load center of gravity, and obtains the load center of gravity sequence;

[0047] Trend path identification module: Input the load center of gravity sequence into the Markov chain model, identify the state transition path of the load center of gravity in multiple time periods, compare it with the state transition probability threshold, and filter the offset trend sequence.

[0048] Attitude Anomaly Monitoring Module: Calculates the duration of continuous offset segments in the offset trend sequence, compares it with the offset trend duration threshold, determines whether the attitude state of the suspended platform is in an abnormal operating state, and uploads the attitude stability monitoring results to the cloud monitoring platform.

[0049] The torsional attitude recognition results include the fitting error of the angle change sequence, the attitude rotation direction recognition type, and the attitude torsional amplitude level. The attitude offset judgment results specifically include the diagonal load difference characteristics, the load change trend direction, and the verification judgment label. The load center of gravity sequence includes the center of gravity plane coordinates, the center of gravity position offset vector, and the load distribution time sequence label. The offset trend sequence specifically includes the state transition path sequence, the start and end time of the offset duration, and the state transition probability value. The attitude stability monitoring results include the abnormal attitude judgment status, the duration of continuous offset, and the monitoring time period identifier.

[0050] Please see Figure 2 The twisting posture recognition module includes:

[0051] Gravity vector acquisition submodule: Acquires acceleration vector data from the three-axis accelerometers deployed along the hoist's running path for each time period, transmits it synchronously to the data receiving end via the IoT communication link, calculates the direction angle corresponding to each acceleration vector and arranges them in chronological order to obtain the sequence of gravity direction changes;

[0052] Data is collected by deploying triaxial accelerometers along the suspended platform's path. First, each sensor needs to be positioned along... The sensors are fixedly installed on key nodes of the suspended platform (such as connecting beams and load-bearing platforms) along three axes. The sensors record triaxial acceleration data every 0.1 seconds, and the raw data recorded consists of acceleration values ​​in the three axes (e.g., ...). ),in: They represent along axial acceleration components (unit: 1 ≈9.81m / s²).

[0053] These values ​​are synchronously uploaded to a remote server via an IoT communication module (such as a LoRa or NB-IoT module). At the data receiving end, the direction angle of the acceleration vector is calculated using the received triaxial acceleration values. The direction angle is calculated using the spatial vector angle method, that is, the gravity vector is expressed as... And normalized to a unit vector, the formula for calculating the angle between the direction angle and the perpendicular direction is:

[0054] ;

[0055] in: The direction of gravity and the vertical direction during the current time period (i.e., The angle between the axes (in degrees). The magnitude of the triaxial acceleration vector is calculated using the following formula: .

[0056] Taking actual data as an example, if ,but: Therefore: Such calculations are performed at set times based on actual circumstances, corresponding to different time points. (like ), each Sort to get sequence.

[0057] Displacement path analysis submodule: Based on the gravity direction change sequence, obtain the displacement vector information of each time period recorded by the displacement encoder, construct the angle sequence by calculating the angle between the gravity direction vector of the basket and the displacement path vector of the basket, perform linear fitting operation on the sequence, and identify the angle fitting offset trend.

[0058] Based on the sequence of changes in the direction of gravity, the current displacement vector information of the suspended platform is obtained at each sampling time point. The displacement vector is measured by an integrated displacement encoder. For example, if the displacement encoder records 0.05 meters at a certain moment, the displacement vector is represented as follows: This indicates that at that moment the basket was only along Displacement occurred in the axial direction. , , : These respectively indicate that the suspended platform is in , , Displacement components in three directions, in meters (m). Displacement vector, a spatial vector composed of three-axis components, in the form of... , : This represents the gravity direction vector obtained from the triaxial accelerometer at the same time point, in units of . (Standard gravitational acceleration).

[0059] The angle between the displacement vector and the gravity direction vector is calculated using the following formula for spatial vector angles:

[0060] ;

[0061] in: :express and The dot product is calculated using the following formula: , : Gravity vector magnitude , : Magnitude of the displacement vector .

[0062] If the acceleration value at a certain moment is The displacement is ,but:

[0063] ;

[0064] ;

[0065] ;

[0066] .

[0067] The angle calculation described above is repeated at multiple time points to construct a time series. The following set of sampling data is provided (unit: seconds, angle): , .

[0068] The fitting objective is to obtain a linear expression for the relationship between time and the included angle:

[0069] ;

[0070] in: Slope: Indicates the rate at which the angle increases per second, measured in units of... , :exist The initial value of the included angle at time, in units of .

[0071] The least squares method is used for fitting, and the fitting formula is as follows:

[0072] ;

[0073] in: : No. Each sampling time point :and The corresponding included angle measurement value, : Average value at time points The average value of the included angle. : Number of data points, in this example .

[0074] First, calculate the average:

[0075] , ;

[0076] Then calculate the numerator and denominator:

[0077] ;

[0078] ;

[0079] .

[0080] Calculate the intercept : .

[0081] Final Fitting Function for: .

[0082] Assuming that the angle rises on average every second by approximately This indicates that the movement path of the suspended basket gradually deviates from the vertical direction during this period.

[0083] This linear trend is used by the subsequent attitude judgment module to identify whether it is a structural attitude change, an unstable state of the basket, etc. All fitting processes are performed based on actual sampled data, without abstract judgments or empirical adjustments.

[0084] Attitude Twist Judgment Submodule: Based on the fitting offset trend of the included angle, analyze the slope trend of the included angle change in a continuous time period, determine whether there is a continuous offset segment, and combine the trend span to mark the time interval to generate the torsion attitude recognition result.

[0085] Based on the fitted angle-time function Analyze its slope value This value is much higher than the threshold set by engineering experience. This initially indicates the existence of a segment with continuous attitude deviation. To verify its persistence, the module iterates through the continuous slope data within the sliding window; if continuous... The slope values ​​in each window all satisfy... This confirms the offset trend. The minimum slope threshold for torsion determination is set based on historical stable basket test results. Angle changes below this value can be considered natural swaying. The set value is... This corresponds to the boundary of the normal acceleration variation range. The minimum number of consecutive judgment windows is set to 3, corresponding to a stable trend with a continuous time of no less than 3 seconds.

[0086] Let the sampling time range be to Divided into three overlapping windows:

[0087] Assume window 1: Fitting slope Window 2: Fitting slope Window 3: Fitting slope The fitting slopes for all three windows are greater than [a certain value]. This meets the criteria for sustained offset. Further assessment of the trend duration is needed. To meet the minimum span requirement for trend continuation .

[0088] To further identify the magnitude of the offset, the maximum angle within this interval is calculated. Minimum value as follows:

[0089] ;

[0090] ;

[0091] .

[0092] Finally, the identified intervals will be labeled as It outputs the results of the torsional posture segment recognition: the time interval is to The fitted slope is The maximum angle change range is This data is output and stored in the identification data record.

[0093] Please see Figure 3 The attitude offset verification module includes:

[0094] Stress data acquisition submodule: acquires stress output data from stress sensors deployed at the four corners of the suspended platform within a specified time period, categorizes the stress value corresponding to each corner point according to four identifiers: left rear, right front, right rear, and left front, and generates a set of corner stress measurement values.

[0095] Stress sensors are deployed at the four corners of the suspended platform, typically at the left rear, right front, right rear, and left front. These sensors are installed at the corners of the suspended platform's structural components. Each sensor collects stress values ​​at its location every 0.2 seconds, measured in megapascals (MPa). The system labels the stress values ​​collected at each corner point within each time period, for example, left rear as S1, right front as S2, right rear as S3, and left front as S4. Each record contains three pieces of information: collection time, location label, and stress value. The data is stored as a time series and categorized into four sets of stress data based on the four corner locations. For example, within a certain time period, the values ​​might be recorded as S1=22.4, S2=20.9, S3=23.1, and S4=21.3 MPa. This process continues throughout the monitoring cycle, outputting a set of stress time series values ​​from all collected points, thus generating a set of corner stress measurement values.

[0096] Load difference calculation submodule: Based on the set of corner stress measurement values, construct two load combinations for the corresponding diagonal regions of left rear-right front and right rear-left front, calculate the stress difference sequence corresponding to the time period within each diagonal combination, obtain the relative trend of the difference change between the two combinations, and generate the diagonal load change trend.

[0097] First, the stress data sets at the four corner points are categorized into two diagonal combinations based on structural location: the left rear-right front combination and the right rear-left front combination. For each time step, the difference sequence between the stress values ​​at the two corresponding corner points is calculated. For example, if at time t1 the left rear is 22.4 MPa and the right front is 20.9 MPa, the difference for this combination is 1.5 MPa; if the right rear is 23.1 MPa and the left front is 21.3 MPa, the difference for this combination is 1.8 MPa. MPa, the difference between the two combinations at each moment is recorded as two sequences, then the changing trend of the difference between the two diagonal combinations in each time period is statistically analyzed, the direction of increase or decrease between the differences between each pair of adjacent time points is calculated using the simple difference method, and the difference change curve of the two combinations is plotted according to the time axis. Finally, the rising, falling or stable trend of the two change curves is compared. If the difference between the left rear and the right front continues to rise while the difference between the right rear and the left front continues to fall in a certain period of time, then their changing directions are relatively opposite, or if the two change synchronously, it indicates that the trends are consistent. Based on the above comparison results, the trend change is output as the diagonal load change trend.

[0098] Opposite Consistency Verification Submodule: Based on the diagonal load change trend and torsional attitude recognition results, compare whether there is an interleaved shift in the trend direction within the same time period, determine whether the attitude deviates from the dominant path of the center of gravity, and output the attitude shift judgment result.

[0099] The module acquires the diagonal load variation trend and the previously confirmed attitude torsion segment and angle change records from the torsion attitude recognition results. It extracts the variation trends of both within the same time interval. For example, if the torsion recognition segment is from 0 to 4 seconds, the stress difference trend of the corresponding diagonal combination within this time interval is: the left rear-right front combination increases from 1.5 MPa to 2.3 MPa per second, while the right rear-left front combination decreases from 1.8 MPa to 1.2 MPa. This indicates that the two sets of differences show an alternating and opposite variation trend. The module compares these two trend directions to determine whether there is a significant intersection phenomenon within the same time interval. The judgment standard is set as follows: if the two sets of difference change directions show opposite trends for more than 3 consecutive sampling points, it is considered that there is an alternating offset phenomenon. It records whether this trend is consistent with the identified attitude torsion direction. If the offset direction is opposite to the dominant path of the center of gravity, that is, the direction of difference growth is opposite to the direction of torsion angle change, it is determined that the basket attitude has deviated from the dominant path of the center of gravity, and the attitude offset judgment result within this time interval is output.

[0100] Please see Figure 4 The load center of gravity calculation module includes:

[0101] Operational load data acquisition submodule: Based on the attitude offset determination result, it collects the coordinate information and equivalent load value of the operators and equipment in the corresponding time period recorded in the UWB positioning tag and RFID identification tag configured on the suspended platform, and obtains the operational load coordinate value pair set;

[0102] The system collects operational data from UWB positioning tags and RFID identification tags installed on the suspended platform. UWB positioning tags are used to obtain the real-time two-dimensional coordinates of the workers, while RFID identification tags are used to identify the construction equipment on the suspended platform and locate its current position. The system is set to a sampling period of 0.5 seconds, collecting the coordinate-load correspondence once at each time point. The data format is a triplet of horizontal coordinate, vertical coordinate, and equivalent load value, with coordinate units in meters and load units in kilograms. The worker's load value is determined by their weight and movement status. The movement status uses a correction coefficient set by the construction standard movement level classification system. The equipment's load value is determined by a combination of the equipment's rated weight and its operational status correction factor. The rated weight is the standard mass of the equipment in a static state, and the operational status correction factor is determined according to the equipment type and current movement. Different speed settings are set, for example, the static correction coefficient for a regular electric drill is 1.0, low-speed drilling is 1.1, and high-speed impact is 1.2. If a certain equipment has a rated weight of 6kg and is currently in a high-speed impact state, then its equivalent load is 6×1.2=7.2kg. Finally, the load values ​​of the operator and the equipment are summed and combined to the corresponding coordinate points. For example, if a construction worker weighs 70kg and is walking, with a correction coefficient of 1.0, the equivalent load of the electric drill he is carrying is 7.2kg, and the worker's coordinates are (1.2, 0.8), then his total equivalent load is 70+7.2=77.2kg, and the recorded coordinate load pair is (1.2, 0.8, 77.2). The system generates multiple such coordinate load pairs each time it samples, and summarizes the data of all time periods into a set of coordinate load pairs arranged in timestamp order.

[0103] Load distribution construction submodule: Based on the set of work load coordinate values, establish a spatial distribution mapping of work load in a two-dimensional plane, and map the load values ​​to the plane distribution map according to the coordinate position to obtain the load position mapping map;

[0104] A two-dimensional coordinate system is established with the actual length and width dimensions of the suspended platform as its boundaries. All collected work load coordinate values ​​are projected into this coordinate system. The horizontal and vertical coordinates represent the horizontal and vertical positions on the platform, respectively. The system is constructed based on a set grid cell division. For example, if the grid size is set to 0.5 meters, and the suspended platform is 4 meters long and 2 meters wide, then it is divided into 32 grid cells: 8 horizontal cells and 4 vertical cells. Each coordinate value pair is categorized according to its coordinate value falling into a specific grid cell. If a sampling point is located at (1.2, 0.8), corresponding to the 3rd horizontal cell and the 2nd vertical cell, the system will classify it within that grid cell. The load values ​​at each point are accumulated. If multiple objects or devices fall on the same grid within the same time period, their load values ​​are accumulated. For example, if the load values ​​recorded at three consecutive time points in this grid are 77.2 kg, 68.5 kg, and 72.9 kg, then the total load of this grid is 218.6 kg. After all grids have been accumulated, the system stores the center coordinates of each grid cell together with the corresponding total load value to generate a load location dataset that describes the spatial distribution. This dataset is then displayed in a visual two-dimensional matrix, where the horizontal and vertical axes represent spatial locations, and the values ​​in the matrix cells represent the cumulative load values. The whole dataset constitutes a load location mapping map.

[0105] The centroid coordinate extraction submodule calls the load position mapping map, uses coordinate weights and load values ​​as calculation factors to calculate the composite centroid coordinates of the suspended platform in the two-dimensional plane in the current time period, and outputs the centroid positions in the continuous time period as sequence data to generate a load centroid sequence.

[0106] Based on the spatial distribution information of each grid cell in the load location mapping map, the center point coordinates of each grid cell and its corresponding cumulative load are extracted as calculation inputs. A two-dimensional spatial weighted average method is executed within each sampling time period to obtain the current composite centroid coordinates of the suspended platform load. The specific execution process is as follows:

[0107] Suppose that within a certain time period, there are a total of There are 10 valid grid cells participating in the calculation, and each cell has the following three properties: : No. The horizontal coordinates of the center point of each grid cell (unit: meters). : No. The vertical coordinates of the center point of each grid cell (unit: meters). : No. Cumulative load value of each grid cell (unit: kilograms).

[0108] First, calculate the total load value of all grids within this time period, denoted as: .

[0109] Next, the weighted coordinate values ​​of each grid cell in the horizontal and vertical directions are calculated, and the following normalization calculation is performed for each direction: , .

[0110] in: : The coordinates of the center of gravity of the gantry in the lateral direction during the current time period (unit: meters). : The longitudinal load center of gravity coordinates of the suspended platform during the current time period (unit: meters). : No. The weight ratio of each grid cell relative to the total load (unitless).

[0111] Let there be a total of There are 1 effective grid cell, and their specific data are as follows: Grid 1: , , Grid 2: , , Grid 3: , , Grid 4: , , .

[0112] First, calculate the total load: ;

[0113] Calculate the weight ratio (normalized weight) for each grid cell: , , , .

[0114] Calculate the lateral centroid coordinates : ;

[0115] Calculate the longitudinal centroid coordinates : ;

[0116] Finally, the composite centroid coordinates of the two-dimensional loads on the suspended platform during this time period are: .

[0117] The system writes the point as the centroid coordinate corresponding to the current timestamp into the sequence data, and continues to repeat the above calculation process for all subsequent sampling time periods. Finally, it outputs the complete load centroid sequence in chronological order, which can be used for further comparative analysis with attitude change trends or to construct a hoist center of gravity offset trajectory map.

[0118] Please see Figure 5 The trend path identification module includes:

[0119] State partitioning construction submodule: Call the load centroid sequence, divide the discrete state space according to the plane coordinate range of the two-dimensional platform, and place all centroid points into the corresponding blocks according to their positions to generate a load centroid state distribution column.

[0120] Based on the two-dimensional centroid coordinates at each time point in the load centroid sequence, and referring to the physical dimensions of the suspended platform, several equidistant state blocks are divided in the two-dimensional coordinate system to construct a discrete state space. Each state block is defined as a rectangular unit of finite size to accommodate the centroid coordinates falling within that area. For example, if the suspended platform is 4 meters long and 2 meters wide, and is divided into 8 horizontal and 4 vertical unit areas respectively, then the entire platform is divided into 32 rectangular state units. Each block is numbered from S1 to S32, corresponding to its spatial position in the coordinate system. The entire centroid sequence is traversed, and the horizontal and vertical coordinates of each centroid point are compared with the boundary coordinates of the state unit to determine its block. For example, if the centroid coordinates are (0.5714, 0.5356), it falls into the second horizontal column and the second vertical row block, numbered S6, and the state at that time point is recorded as S6. The entire centroid sequence is processed in this way, mapping the centroid coordinates at each time point to a specific state number, and finally forming a load centroid state distribution column arranged by time, while maintaining consistency with the original timestamp, for subsequent state path and jump analysis.

[0121] Transition probability calculation submodule: Based on the load centroid state distribution column, the Markov chain model is applied to calculate the transition frequency between adjacent states, identify the jump probability change between each path node, and obtain the state transition probability set;

[0122] Based on the load centroid state distribution, the time series transitions of the centroid between discrete states are statistically analyzed to construct a Markov chain state transition matrix. The specific execution process includes: first, extracting the state numbers of any two consecutive time points in the state distribution to form a set of state transition pairs. Let the centroid state sequence for a certain time period be S1, S2, S3, S2, S3, S3, S4, S2, S3. Then the extracted state transition pairs are (S1→S2), (S2→S3), (S3→S2), (S2→S3), (S3→S3), (S3→S4), (S4→S2), (S2→S3). A transition frequency matrix is ​​then established according to all possible combinations of state pairs, where each element... The frequency, or transition frequency, represents the number of times a state transitions from state Sa to state Sb. For example, the state S2→S3 appears 3 times in the above sequence, with a frequency of The state S3→S3 occurs only once, with a frequency of The state S1→S2 occurs once, with a frequency of The state S3→S2 occurs once, with a frequency of .

[0123] Next, the frequency of all jumps originating from each state Sa is normalized to obtain the state transition probability. The calculation formula is as follows:

[0124] ;

[0125] in: : Represents the probability (unitless) of transitioning from state Sa to state Sb. : Indicates the number of times (in seconds) a jump occurs from state Sa to state Sb. : Represents the total number of times (in units) the state Sa transitions to all possible states. : Index of discrete state number, representing all possible state path labels.

[0126] Starting from state S2, the transition pairs are (S2→S3), (S2→S3), (S2→S3), (S2→S3), a total of 4 times. Therefore: , ,so .

[0127] Starting from state S3, the transitions include: S3→S2, S3→S3, and S3→S4, each occurring once.

[0128] , , , ,so: , , .

[0129] Similarly, traverse the entire state distribution sequence, recording all valid transition pairs in the state transition matrix, and finally output the result consisting of all... The set of state transition probabilities, composed of jump probabilities, is used to analyze the behavioral tendencies and dynamic distribution paths of the center of gravity in different regions.

[0130] Offset path filtering submodule: Based on the set of state transition probabilities, the jump probability of each state path is filtered, the transition path segments that are continuously higher than the state transition probability threshold are identified and extracted as offset trend data, and an offset trend sequence is generated.

[0131] Read the set of state transition probabilities and perform a search on all state pairs within it. The corresponding jump probability The system performs item-by-item comparison and screening, setting a transition probability threshold as a certain numerical limit to distinguish between high-frequency jumps and random fluctuations. It traverses each actual state path, extracting state path segments with multiple consecutive jump probabilities greater than the threshold as candidate offset segments. For example, if the threshold is set to 0.6, and a path S5→S6, S6→S10, and S10→S11 are found to have transition probabilities of 0.75, 0.82, and 0.68 respectively (all higher than 0.6), then this path segment is determined to be a continuous offset path segment. The system records the starting state, ending state, starting time, ending time, corresponding state number, and jump probability value for each segment. Simultaneously, if a brief jump probability below the threshold occurs in the middle of a continuous segment (e.g., only once below 0.6), the system allows the entire path segment to be retained. If the cumulative number of jump probabilities below the threshold exceeds two, the segment judgment is terminated. Finally, all high-frequency state transition path segments that meet the conditions are combined in chronological order to form an offset trend sequence.

[0132] Please see Figure 6 The attitude anomaly monitoring module includes:

[0133] Trend Continuity Statistics Submodule: Calls the offset trend sequence, sequentially analyzes the start and end ranges of continuous offset segments in the time dimension, calculates the corresponding duration data according to the time span of each segment, and establishes a statistical index for all segments to obtain continuous offset duration groups.

[0134] Each extracted offset path segment is processed sequentially along the time dimension. First, the start and end status numbers, start and end timestamps of each offset path segment are extracted. Calculations are then performed to obtain the time span of the path segment, using a time difference method, i.e., subtracting the start time from the end time to obtain the duration value, in seconds or milliseconds. The timestamp format is uniformly standardized to standard UTC or Unix timestamp format. For example, if an offset path segment starts at 14:00:30 (UTC) and ends at 14:00:55, then the duration of this segment is 25 seconds. The system records the calculation result of each segment as an offset segment duration record entry, including the segment number, start and end status, start and end times, and duration value. Then, all segments are sorted by start time, and a numbered index list is established. The numbering format can be T01, T02, T03, etc., to support subsequent location referencing and querying. Finally, the duration group of all offset trend segments is output.

[0135] Abnormal state judgment submodule: Based on the continuous offset duration group, each duration segment is compared item by item to identify whether there are continuous segments that are greater than the offset trend duration threshold, and the corresponding time period status identifier is marked to generate an abnormal posture segment marker set;

[0136] Based on the continuous offset duration group, the duration of each offset segment is retrieved and judged item by item. The judgment method is based on the comparison and analysis of the set offset trend duration threshold. This threshold is the maximum stable allowable offset time set according to the basket structure and operating conditions. For example, if the threshold is set to 15 seconds, it means that if a segment continuously offsets for more than 15 seconds, it can be considered an abnormal state. The system compares the duration of each segment with 15 seconds in turn. If the duration of a segment T03 is 18 seconds and T07 is 21 seconds, both of which meet the condition of being greater than the threshold, these segments are marked as abnormal. The start and end times, duration, status number and judgment result of the segment are recorded. An abnormal flag bit is set in the marking information, with a value of 1 indicating abnormality and 0 indicating normality. An abnormal status marking table is generated, and all abnormal segment numbers are written into the abnormal attitude segment marking set.

[0137] Monitoring result upload submodule: Based on the abnormal posture segment marker set, construct the time-series monitoring result data structure of the hanging basket posture state, encapsulate it into structured transmission format content, and upload it to the cloud monitoring platform through the Internet of Things link to generate posture stability monitoring results;

[0138] The system acquires a set of abnormal attitude segment markers. Based on the segments' numbers, time ranges, status numbers, and abnormal flag values, it constructs a time-series monitoring structure for the suspended platform's attitude status. The structure is encapsulated in JSON or XML format, with fields including timestamp, state_code, duration, and status_flag. The system arranges each record in chronological order and embeds data source identifiers and device numbers to form a complete data structure. Then, it uses the IoT communication module integrated into the suspended platform to call the network link for packet encapsulation and remote push. The upload address is the specified interface URL of the cloud monitoring platform or an MQTT topic channel. The upload frequency can be set to 10 seconds / upload or push immediately after each segment is completed. After the system completes the packet encapsulation, it sends it to the cloud. The receiving end parses the packet and uses it to generate attitude stability monitoring results.

[0139] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A safety monitoring system for irregularly shaped suspended platforms based on the Internet of Things, characterized in that, The system includes: Torsional posture recognition module: synchronously collects the gravity vector trajectory and real-time displacement path of the suspended platform at each time period during the vertical movement of the suspended platform through the Internet of Things communication link, identifies the torsional posture of the suspended platform, and obtains the torsional posture recognition result; Attitude deviation verification module: acquires the stress values ​​of the four corner points of the suspended platform over a specified time period, calculates the load changes of the corresponding diagonal corners, and performs counter-verification in combination with the torsional attitude recognition results to obtain the attitude deviation judgment result. Load center of gravity calculation module: Based on the attitude offset determination result, collect the real-time coordinates of the operator and equipment and the corresponding equivalent load value, and calculate the load center of gravity to obtain the load center of gravity sequence; Trend path identification module: Input the load center of gravity sequence into the Markov chain model, identify the state transition path of the load center of gravity in multiple time periods, compare it with the state transition probability threshold, and filter the offset trend sequence; Attitude anomaly monitoring module: Calculates the duration of continuous offset segments in the offset trend sequence, compares it with the offset trend duration threshold, determines whether the attitude state of the suspended platform is in an abnormal operating state, and uploads the attitude stability monitoring results to the cloud monitoring platform. The load centroid calculation module includes: Operational load data acquisition submodule: Based on the attitude deviation determination result, it acquires the coordinate information and equivalent load value of the operators and equipment for the corresponding time period recorded in the UWB positioning tag and RFID identification tag configured on the suspended platform, and obtains the operational load coordinate value pair set; Load distribution construction submodule: Based on the set of work load coordinate values, establish a spatial distribution mapping of work load in a two-dimensional plane, and map the load values ​​to the plane distribution map according to the coordinate position to obtain the load position mapping map; Centroid Coordinate Extraction Submodule: Calls the load position mapping map, uses coordinate weights and load values ​​as calculation factors to calculate the composite centroid coordinates of the suspended platform in the two-dimensional plane during the current time period, and outputs the centroid positions in the continuous time period as sequence data to generate a load centroid sequence. The trend path identification module includes: State partitioning construction submodule: Call the load centroid sequence, divide the discrete state space according to the plane coordinate range of the two-dimensional platform, and place all centroid points into the corresponding blocks according to their positions to generate a load centroid state distribution column. Transition probability calculation submodule: Based on the load centroid state distribution column, the Markov chain model is applied to calculate the transition frequency between adjacent states, identify the jump probability change between each path node, and obtain the state transition probability set; Offset path filtering submodule: Based on the set of state transition probabilities, the jump probability of each state path is filtered, the transition path segments that are continuously higher than the state transition probability threshold are identified and extracted as offset trend data, and an offset trend sequence is generated.

2. The IoT-based safety monitoring system for irregularly shaped suspended platforms according to claim 1, characterized in that, The torsional attitude recognition results include the fitting error of the included angle change sequence, the attitude rotation direction recognition type, and the attitude torsional amplitude level. The attitude offset judgment results specifically include the diagonal load difference characteristics, the load change trend direction, and the verification judgment label. The load centroid sequence includes the centroid plane coordinates, the centroid position offset vector, and the load distribution time sequence label. The offset trend sequence specifically includes the state transition path sequence, the start and end time of the offset duration, and the state transition probability value. The attitude stability monitoring results include the abnormal attitude judgment state, the duration of continuous offset, and the monitoring time period identifier.

3. The IoT-based safety monitoring system for irregularly shaped suspended platforms according to claim 1, characterized in that, The torsional posture recognition module includes: Gravity vector acquisition submodule: Acquires acceleration vector data from the three-axis accelerometers deployed along the hoist's running path for each time period, transmits it synchronously to the data receiving end via the IoT communication link, calculates the direction angle corresponding to each acceleration vector and arranges them in chronological order to obtain the sequence of gravity direction changes; Displacement path analysis submodule: Based on the gravity direction change sequence, obtain the displacement vector information of each time period recorded by the displacement encoder, construct the angle sequence by calculating the angle between the gravity direction vector of the basket and the displacement path vector of the basket, perform a linear fitting operation on the sequence, and identify the angle fitting offset trend; The attitude torsion judgment submodule analyzes the slope trend of the angle change over a continuous time period based on the fitting offset trend of the included angle, determines whether there is a continuous offset segment, and generates the torsion attitude recognition result by combining the trend span with the time interval.

4. The IoT-based safety monitoring system for irregularly shaped suspended platforms according to claim 3, characterized in that, The attitude offset verification module includes: Stress data acquisition submodule: acquires stress output data from stress sensors deployed at the four corners of the suspended platform within a specified time period, categorizes the stress value corresponding to each corner point according to four identifiers: left rear, right front, right rear, and left front, and generates a set of corner stress measurement values. Load difference calculation submodule: Based on the set of corner stress measurements, construct two load combinations for the corresponding diagonal regions of left rear-right front and right rear-left front, calculate the stress difference sequence corresponding to the time period within each diagonal combination, obtain the relative trend of the difference change between the two combinations, and generate the diagonal load change trend. Opposite Consistency Verification Submodule: Based on the diagonal load change trend and the torsional attitude recognition result, compare whether there is an interleaved shift in the trend direction within the same time period, determine whether the attitude deviates from the dominant path of the center of gravity, and output the attitude shift determination result.

5. The IoT-based safety monitoring system for irregularly shaped suspended platforms according to claim 1, characterized in that, The attitude anomaly monitoring module includes: Trend Continuity Statistics Submodule: Calls the offset trend sequence, sequentially analyzes the start and end ranges of continuous offset segments in the time dimension, calculates the corresponding duration data according to the time span of each segment, and establishes a statistical index for all segments to obtain continuous offset duration groups. Abnormal state judgment submodule: Based on the continuous offset duration group, compare each duration segment item by item, identify whether there are continuous segments that are greater than the offset trend duration threshold, mark the state identifier of the corresponding time period, and generate an abnormal posture segment marker set; Monitoring result upload submodule: Based on the abnormal posture segment marker set, construct the time-series monitoring result data structure of the hanging basket posture state, encapsulate it into a structured transmission format, and upload it to the cloud monitoring platform through the Internet of Things link to generate posture stability monitoring results.

Images