A tower crane jacking multi-parameter coupling prediction early warning and intelligent decision method

Abstract

This invention discloses a multi-parameter coupling prediction, early warning, and intelligent decision-making method for tower crane jacking, comprising: collecting multiple parameters covering dimensions of state safety, mechanical balance, and environmental adaptation through a multi-modal intelligent monitoring module integrating multiple sensors; preprocessing the data; constructing a three-level risk quantification model containing root nodes, intermediate coupling nodes, and leaf nodes based on multiple parameters using an improved probabilistic graphical network model combined with a dynamic balancing self-calibration algorithm; dynamically allocating the weights of each parameter according to different jacking processes, calculating and outputting the graded coupling risk level; analyzing the collaborative change trend of multiple parameters based on the risk level and its temporal changes through a time-series prediction model, predicting the risk level changes in future periods, and triggering early warning information in advance based on the prediction results; outputting multi-dimensional early warnings and operational guidance based on the graded coupling risk level, and automatically interlocking control or executing emergency shutdown for subsequent processes according to the risk level.

Description

Technical Field

[0001] The present invention relates to the technical field of tower crane jacking safety control, and particularly relates to a multi-parameter coupling prediction, warning and intelligent decision-making method for tower crane jacking. Background Art

[0002] The information provided in this part is only background information related to the present disclosure, and it is not necessarily prior art.

[0003] The jacking operation of a tower crane is a key high-risk link in construction. According to the statistical data of the Ministry of Housing and Urban-Rural Development of China, 89% of the jacking instability accidents are caused by the combined action of multiple parameters and the risk conduction across processes. The influencing factors include the inclination of the sleeve frame, non-standard hanging boot operation, operation against the wind, unfastened bolts, and the driver's fatigue operation, etc. The accidents caused by single factors account for only 11%. There are many prominent technical pain points in the existing tower crane jacking monitoring systems and related technologies:

[0004] Limited adaptability: Most of the existing multi-degree-of-freedom installation schemes are complex physical structures, which need to be customized according to different tower crane models. The installation is cumbersome, and the compensation effect for the manufacturing errors of old tower cranes is limited, affecting the multi-parameter acquisition accuracy and the reliability of coupling calculation, and the universality is insufficient; Lack of process linkage: Only the parameters of a single process are monitored, and the coupling relationship between "multi-parameters of the previous process - multi-parameters of the current process" is not established, resulting in incomplete risk quantification; Shortcomings of sensors and transmission: The sensors lack a fault self-diagnosis and standby switching mechanism, and no multi-parameter cross-verification logic is formed, which is prone to misjudgment due to single-point failures; Dependence on traditional wired transmission or basic wireless transmission, without cache and standby channels, unable to meet the multi-parameter real-time coupling transmission requirements for UDP wireless transmission; Limited trimming accuracy: The trimming recognition is only based on a single parameter or a fixed algorithm, and the multi-parameter coupling calculation logic such as "amplitude, inclination angle, slewing angle" is not established, nor is it dynamically calibrated in combination with working condition parameters such as the service life of the tower crane and the number of attachment layers. The trimming accuracy is greatly affected by the environment and equipment status; Single decision-making guidance: The operation guidance only targets single-risk scenarios, and does not generate differentiated solutions based on the multi-parameter coupling risk level, and does not consider the coupling risks between personnel status parameters such as the driver's fatigue and illegal operation and equipment parameters, and cannot accurately handle the complex risks caused by the combination of multiple parameters.

[0005] Although the prior art has achieved multi-sensor monitoring and basic wireless transmission, it has not broken through the core technical bottleneck of multi-parameter coupling - the associated calculation logic between parameters has not been established, the parameter weights have not been dynamically allocated according to the process, the process stage division and associated risk key points have not been clarified, and the coupling analysis between the driver's state parameters and equipment parameters has not been incorporated, resulting in low risk quantification accuracy and delayed warning, and unable to fundamentally solve the safety control and universality pain points of the jacking operation. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this solution provides a multi-parameter coupled prediction, early warning, and intelligent decision-making system for tower crane jacking. This system addresses the technical pain points of existing tower crane jacking monitoring systems, including limited adaptability, lack of process linkage, sensor malfunctions leading to misjudgments, insufficient transmission stability, limited balancing accuracy, and failure to consider operator status coupling risks. The core of this solution is establishing a full-process multi-parameter coupling logic, clearly defining core parameters, and dividing the core stages of the process. This achieves the technical goals of accurate risk quantification, proactive early warning, coordinated guidance and control, reliable and stable system, and easy and universally applicable installation. Based on the core invention concept of "multi-parameter coupling," the system follows a sequential process of "data acquisition → processing and transmission → risk quantification → time-series prediction → decision linkage → cloud optimization," plus some real-time linkage logic. The specific steps are as follows:

[0007] Step 1: Multimodal data acquisition. Through a multimodal intelligent monitoring module integrating multiple sensors, multiple core coupled parameters covering the dimensions of state safety, mechanical balance and environmental adaptation are collected simultaneously.

[0008] Step 2: Data preprocessing and transmission. The data collected in Step 1 is preprocessed, including noise reduction, filtering and artificial intelligence feature verification. The validity of the data is ensured through a multi-parameter cross-validation mechanism. The processed data is then transmitted in real time via wireless transmission.

[0009] Step 3: Cross-process coupling risk quantification. Based on the multi-class parameters preprocessed in Step 2, a three-level risk quantification model containing root nodes, intermediate coupling nodes, and leaf nodes is constructed by combining an improved probabilistic graphical network model with a dynamic balancing self-calibration algorithm. The weights of each parameter are dynamically allocated according to different lifting processes, and the hierarchical coupling risk level is calculated and output.

[0010] Step 4: Phased time series prediction. Based on the risk level and its time series changes output in Step 3, the time series prediction model is used to analyze the trend of multi-parameter coordinated changes, predict the risk level changes in future periods, and trigger early warning information in advance based on the prediction results. The warning in this step is: if something may happen in the future, inform in advance, only remind and do not intervene.

[0011] Step 5: Intelligent Decision-Making and Emergency Response Coordination. Based on the hierarchical coupling of risk levels, multi-dimensional early warnings and operational guidance are output, and subsequent processes are automatically interlocked or emergency shutdowns are executed according to the risk level. The early warning in this step indicates that an incident has already occurred or is about to occur, providing both reminders and solutions, and enforcing mandatory control.

[0012] In some embodiments, the specific improvements to the improved probabilistic graphical network model described in step 3 are as follows:

[0013] Modular node directional connection improvement: Breaking through the fully connected redundant architecture of general probabilistic graph networks, the root node is divided into "state safety sub-cluster, mechanical balance sub-cluster, environmental adaptation sub-cluster, and working condition sub-cluster" according to function. Each sub-cluster only establishes a directional connection with the intermediate coupling node of the corresponding lifting process, avoiding non-associated parameter interaction and adapting to the differences in risk focus of the 7 core stages of the lifting operation.

[0014] Dynamic weight learning improvement: The "offline pre-training + online dual-dimensional calibration" mechanism is adopted. The basic weights are trained offline based on data from multiple tower crane models and multiple working conditions. In the online stage, the parameter weights are adjusted in real time through the "working condition dimension" and "balancing feedback dimension". The balancing feedback dimension includes the balancing deviation output by the dynamic balancing self-calibration algorithm, and the working condition dimension includes the service life of the tower crane and the number of attached layers to adapt to changes in equipment status and balancing accuracy.

[0015] Path-based reasoning and result correction improvement: A dedicated reasoning path is preset for each lifting process (tailored to the risk focus of different processes). At the same time, the results of dynamic balancing self-calibration algorithm are introduced to perform secondary correction on the output value of intermediate coupling nodes, so as to solve the risk quantification deviation caused by insufficient balancing accuracy.

[0016] In some embodiments, the core coupling parameters include, but are not limited to: the bolt connection status between the lower slewing support and the tower column; the connection status of the new standard section; the positioning status of the safety pin of the jacking beam; the positioning status of the climbing claw; the positioning status of the introduction section; the X-axis tilt angle of the jacking frame; the Y-axis tilt angle of the jacking frame; the tower body tilt angle; the jacking height; the slewing angle; the amplitude; the wind speed; the distance between the jacking frame and the tower crane; the matching degree of the balancing status; the collaborative calculation based on amplitude / slewing angle data + tower body tilt angle data; the sensor health status is determined based on the stability of the acquired signal, transmission delay, and cross-validation error of each parameter: continuous data acquisition deviation exceeding a set threshold is determined as signal instability, data transmission delay > 50ms is determined as transmission abnormality, and multi-parameter cross-validation error > 0.3 seconds is determined as verification abnormality; all three indicators (acquisition signal stability, transmission delay, and multi-parameter cross-validation error) are normal, indicating health; any one abnormality indicates sub-health; any two or more abnormalities indicate potential health hazards or malfunctions, and when a potential health hazard or malfunction is determined, the backup sensor switching mechanism is automatically triggered. It covers all safety dimensions: state safety dimension; mechanical balance dimension; environmental adaptation dimension; coupling verification dimension; and driver state dimension.

[0017] The data acquisition units are uniformly installed via standardized magnetic bases with an adsorption pull force ≥80N. They are equipped with five standard shims (0.5mm, 1mm, 2mm, 3mm, and 5mm) to compensate for manufacturing errors on the tower crane mounting surface. These thin components can be combined to offset tower crane manufacturing errors ranging from 0.5-10mm. Simultaneously, with software parameter calibration, the system automatically corrects the coupling calibration process of multiple parameter acquisition references by inputting the tower crane model and installation position offset into the host computer, achieving compatibility with all types of tower cranes. The optimal acquisition frequency is 10Hz.

[0018] In some embodiments, step 2 includes the following steps:

[0019] Step 2-1, Data Preprocessing: The collected synchronous data is denoised and filtered. The denoising process includes data stability verification based on time-series sampling. The denoising method is to compare two sets of data after two samplings with an interval of 0.01s. If the comparison is consistent, the data is determined to be stable.

[0020] Step 2-2, Verification and Reliability Assurance:

[0021] The core parameters are validated using a multi-parameter cross-validation mechanism, and the system has the ability to automatically switch to a backup sensor in case of sensor failure.

[0022] Steps 2-3, Data Transmission and Caching: The processed and verified data is transmitted in real time via wireless communication, and locally stored during transmission using an edge caching module. This edge caching module activates a backup transmission channel in case of network interruption. Preferably, UDP wireless transmission is used via an ESP32-Pro embedded module, with the edge caching module simultaneously storing 30 minutes of critical data.

[0023] The multi-parameter cross-validation mechanism described in step 2-2 specifically achieves bidirectional validation in two ways:

[0024] Verification of parameters in the same dimension using multiple types of sensors: For core parameters such as state safety and mechanical balance, different types of sensors are used to collect data simultaneously to ensure that the data collection results for the same parameter can be mutually verified; for example, the data collection of bolt connection state parameters is mutually verified by photoelectric switches and laser-assisted sensors, and the data collection of parameters of the climbing claw in position is mutually verified by photoelectric switches and ultrasonic radar.

[0025] Verification of related dimensional parameters using similar / multi-type sensors: For parameters with logical relationships, data is collected using similar or multiple types of sensors, and the coupling relationship between parameters is used to verify the validity of the data. For example, data collection on the climbing claw's positioning status and the frame's tilt angle can be verified by observing abnormal fluctuations in the frame's tilt angle when the climbing claw is not in position.

[0026] In the two verification methods mentioned above, cross-validation error specifically refers to the "time difference between the acquisition of 'valid data that meets the basic compliance threshold' by different sensors / related parameters"; if the time difference is ≤0.3 seconds, the data is considered valid, based on the following: the core parameter acquisition frequency of this system is 10Hz, and 0.3 seconds corresponds to 3 consecutive acquisition cycles. This can filter out the occasional interference of a single acquisition and meet the real-time requirements of parameter verification for tower crane jacking operations, avoiding the impact on process efficiency due to excessive verification time; if the time difference is >0.3 seconds, the data is considered invalid, the sensor health status alarm is immediately triggered, and the fault self-switching function is activated to switch to the backup sensor.

[0027] The model training process involves using time-series data (including process labels and actual risk results) of 15 core parameters from various tower crane models (QTZ63, QTZ80, etc.) and multiple operating conditions (service life 0-15 years, number of attached layers 1-5 layers) as samples. First, the data is purified according to the preprocessing rules in step 2. Then, the modular sub-cluster design of the model is matched, and cluster-oriented training is carried out in 7 core lifting processes (only the corresponding functional sub-clusters are connected to the intermediate coupling nodes). The node association weights are iteratively optimized to make the matching degree between the model output and the actual risk results ≥92%. Finally, the basic model is formed through independent sample validation (risk quantification error ≤5%).

[0028] Online adaptation and calibration (dynamically matching actual scenarios)

[0029] By combining dynamic weight learning with the model to improve the design during the lifting operation:

[0030] The operating condition weights are calibrated based on the current service life and number of attachment layers of the tower crane to adapt to equipment aging / tower stability changes;

[0031] Based on the balancing deviation of the dynamic balancing self-calibration algorithm, the weights of balancing-related parameters are finely adjusted to correct the balancing accuracy deviation.

[0032] As the process switches, the pre-trained weights of the corresponding process are automatically invoked, and the process coupling logic is switched through smooth transition logic to avoid fluctuations in risk quantification.

[0033] In some embodiments, step 3 includes the following steps:

[0034] Step 3-1, Risk Quantification Model Construction: Based on the preprocessed multi-class parameters, a three-level risk quantification model containing root nodes, intermediate coupling nodes, and leaf nodes is constructed by combining an improved probabilistic graphical network model with a dynamic balancing self-calibration algorithm.

[0035] Step 3-2, Coupling Parameter Calculation: Based on the model in Step 3-1 and the set coupling logic, the multi-parameter coupling analysis index represented by the intermediate coupling nodes is generated by calculating the parameters of the root node.

[0036] Step 3-3, Dynamic Evaluation and Output:

[0037] The specific architecture of the three-level risk quantification model, which includes a root node, intermediate coupling nodes, and leaf nodes, in step 3-1 is as follows:

[0038] The root node consists of the core coupling parameters and operating parameters determined in step 1. The operating parameters include, but are not limited to, jacking height, number of attachment layers, and service life of the tower crane.

[0039] The intermediate nodes are multi-parameter coupled analysis indicators, including but not limited to tower overturning moment, guide wheel force balance, hanging shoe bearing safety factor, balancing state matching degree, and cross-process risk linkage coefficient;

[0040] Tower overturning moment Calculated using the following formula:

[0041]

[0042]

[0043] in, It is air density. S is the wind load shape coefficient, and S is the windward area of ​​the tower. For wind speed, H0 is the jacking height, and H0 is the reference jacking height. For amplitude This is a process correction factor. This is the working condition compensation coefficient; Based on the value taken during the jacking phase The value is determined based on the service life and the number of attachment layers. For service life less than 5 years, the value is 1.0; for every additional 5 years thereafter, the value increases by 0.1. For attachment layers less than 3 layers, the value is 1.0; for each additional layer, the value increases by 0.08. K 工 The seven core stages of the jacking process are assigned values ​​based on the differences in mechanical stress intensity and risk level at each stage: 0.9 for the pre-jacking preparation stage, 0.8 for the balancing and calibration stage, 1.2 for the jacking stage of the support frame, 1.0 for the standard section introduction stage, 1.1 for the new section installation and fastening stage, 1.05 for the support frame return stage, and 0.9 for the post-operation review stage; the tower overturning moment... The safety threshold is 50 kN·m;

[0044] Stress balance of guide wheel Calculated using the following formula:

[0045]

[0046] Where n is the number of guide wheels, F i For the actual force on a single guide wheel, F mThe average force on the guide wheel is 85%;

[0047] Safety factor of hanging boot load-bearing capacity Calculated using the following formula:

[0048]

[0049] Where F 实 For the actual load-bearing capacity of the hanging boot, F 许 Preset according to tower crane model Values ​​are assigned based on the crawler's positioning status: 1.0 for meeting the standard and 0.6 for not meeting the standard; the safety threshold is 1.3.

[0050] Balanced state matching degree Calculated using the following formula:

[0051]

[0052] This is the actual amplitude variation length. This is the theoretical amplitude reference value. This is the actual tilt angle of the tower. This is the theoretical reference value for the tower's inclination angle. This is the actual lifting height. This is the theoretical lifting height benchmark value; Cross-process linkage coefficient, The standard weighting is used, and the safety threshold is 90%.

[0053] The cross-process risk linkage coefficient is determined based on the compliance rate of the core parameters of the preceding process. The compliance rate of the core parameters of the preceding process is specifically Ntotal / Ntotal, where Ntotal is the number of compliant parameters and Ntotal is the total number of parameters that need to be verified. The specific value rules are as follows: if the preceding process is fully compliant, take 1, then Ntotal / Ntotal = 1; if there is a minor violation, take 0.5, then Ntotal / Ntotal ≥ 0.67; if there is a serious violation, take 0.1, then Ntotal / Ntotal < 0.67.

[0054] In some embodiments, step 4 includes the following steps:

[0055] Step 4-1, Stage Division and Feature Extraction: Divide the lifting operation process into multiple core stages, and extract time-series coupling features that reflect the trend of coordinated change based on multi-parameter time-series data;

[0056] Step 4-2, Temporal Risk Prediction: Based on the temporal coupling characteristics obtained in Step 4-1, predict the risk level changes within a specified time period in the future; the specified time period is preferably 3-5 seconds.

[0057] Step 4-3, Dynamic Early Warning Trigger: Based on the prediction results and combined with the dynamically adjusted risk threshold, an early warning is triggered before the risk level exceeds the threshold, preferably 2 seconds in advance.

[0058] Utilizing an offline-online framework, the core focus is on training by capturing temporal features. Using a 5-second parameter sequence, process labels, and subsequent 3-5 second risk changes as samples, dedicated temporal branches are trained for 7 processes to optimize the ability to capture temporal dependencies, forming a basic prediction model.

[0059] Online adaptation calibration: The same framework is used, and the core is to dynamically adapt to the changes in time sequence. It extracts the time sequence coupling characteristics of the current process in real time, adjusts the sensitivity of parameter change rate in combination with the working conditions, and incorporates the driver's Perclos time sequence changes to tighten the prediction threshold, so as to ensure that the time sequence prediction fits the real-time operation scenario.

[0060] In some embodiments, step 5 includes the following mechanism:

[0061] Multi-dimensional early warning and guidance mechanism: Based on the risk level, output multi-dimensional early warning information and operation guidance including sound, light, text, images and voice;

[0062] Process interlocking control mechanism: When multiple parameters of the preceding process fail to meet the standards, the hydraulic action of the next process is automatically locked, and the action command of the hydraulic pump station is cut off through electronic control signal, thus avoiding the risk transmission across processes from the mechanism.

[0063] Emergency response mechanism: An emergency shutdown mechanism is designed to be independent of the conventional control. It supports manual cancellation after secondary confirmation by the operator and requires two confirmations from the operator to cancel the linkage safety design.

[0064] In some embodiments, the multi-parameter coupling data generated in steps 1 to 5 is uploaded to a remote cloud server via a wireless communication network. Based on the aggregated coupling big data from multiple projects, the cloud server periodically updates the local coupling model parameters, dynamic weights of the multi-parameters, and the self-calibration threshold for balancing. A multi-level storage system consisting of on-site storage devices, edge cache modules, and cloud backup is employed, supporting real-time monitoring across multiple terminals, including computer terminals, mobile terminals, and tower crane operating terminals, as well as providing data interfaces to authorized third-party systems. The preferred update cycle is 15 days.

[0065] In some embodiments, driver fatigue detection is achieved through a HIKVISION smart camera deployed near the driver's operating seat. The camera captures real-time facial images of the driver and extracts key features such as eyelid opening and closing, blinking frequency, and blinking duration based on facial feature recognition algorithms. A Perclos score is calculated, which is the percentage of time the eyelids are closed within a unit of time. When a Perclos score ≥ 0.2, the driver is considered fatigued. When driver fatigue or violation is detected, the judgment is coupled with the sensor-collected parameters, increasing the warning sensitivity by 40% and tightening the risk level judgment threshold by 30%. Simultaneously, hydraulic actions are locked until the violation is resolved. The camera preferably captures facial images at a frequency of 30 frames per second.

[0066] Stable multi-parameter transmission: The sensor fault self-diagnosis transmission architecture is adopted (a collaborative transmission solution consisting of ESP32-Pro embedded module + edge cache + backup channel). The ESP32-Pro embedded module supports UDP wireless transmission, binds to local port 7788, and the target IP and port can be flexibly configured. The edge cache module stores 30 minutes of key coupling parameters and automatically switches to the backup channel when the network is interrupted, ensuring that multi-parameter transmission is uninterrupted and without loss.

[0067] Work Process Division: The jacking operation is divided into 7 core stages, with specific definitions and core tasks as follows:

[0068] Pre-lift preparation stage: The core tasks are sensor installation and calibration, tower crane status pre-inspection (bolt tightness, initial position of climbing claw / safety pin), working condition parameter input (lifting height, number of attachment layers, service years), and operator status initialization monitoring. The key parameters are: sensor health status + initial bolt connection status + working condition parameter matching degree + Perclos benchmark value.

[0069] The balancing and calibration phase: The core tasks are to adjust the position of the luffing trolley and re-measure the tower body / frame tilt angle to ensure the tower crane's center of gravity is balanced. The key parameters coupled with these parameters are: balancing status matching degree + luffing length + tower body tilt angle + jacking height + Perclos score and luffing operation coupling characteristics.

[0070] The core task of the lifting stage of the frame is to start the hydraulic system and slowly lift the frame to the preset height. The key parameters coupled with these parameters are: frame dual-axis tilt angle, lifting height change rate, wind speed, hanging shoe load safety factor, and Perclos score coupled with the lifting operation characteristics.

[0071] Standard section introduction phase: The core task is to transport the new section between the frame and the tower body through the introduction platform. The key parameters to be coupled are: the position of the introduced section + the distance between the frame and the tower crane + the slewing angle + the wind speed + the coupling characteristics of the Perclos score and the slewing operation.

[0072] New section installation and fastening stage: The core tasks are positioning and docking of the new section and bolt fastening. The corresponding parameter coupling focus is: bolt connection status + standard section positioning status + double axis tilt angle of the frame + cross-process linkage coefficient + Perclos score and fastening operation coordination characteristics.

[0073] The core task of the frame return and positioning stage is to slowly return the frame so that the new section can be fully attached to the original tower body to bear the force. The key parameters coupled are: tower body tilt angle + frame dual-axis tilt angle + lifting height return rate + guide wheel force balance + Perclos score and return operation coupling characteristics.

[0074] Post-operation review phase: The core tasks are to comprehensively check the connection status of the new section, zero-calibrate the sensor data, and record the driver status. The key parameters to be checked are: bolt connection status + sensor health status + multi-parameter cross-validation error + Perclos score time series curve.

[0075] Multi-parameter temporal feature extraction: The LSTM-lightweight edge computing model (an optimized version of the Long Short-Term Memory Neural Network, which reduces the number of parameters by simplifying the network structure, reduces the hardware operating pressure, and is good at capturing long-term dependencies between temporal data) targets different risk priorities in 7 core stages and extracts stage-specific multi-parameter collaborative change trends. For example, in the tower crane lifting stage, the focus is on extracting the "coupling feature between the lifting height change rate and the tower crane tilt angle change rate"; in the new section installation stage, the focus is on extracting the "coupling feature between the bolt tightening progress and the standard section positioning status"; and throughout the operator's operation, the focus is on extracting the "coupling feature between the Perclos score change trend and the tower crane's operating actions (amplitude / slewing angle / lifting height adjustment)" (e.g., if Perclos ≥ 0.2 and amplitude adjustment speed > 0.5 m / s, it is judged as a high-risk temporal feature), rather than relying solely on the static values ​​of a single parameter.

[0076] Multi-parameter coupling prediction and early warning: Based on the current 5-second stable multi-parameter coupling data (including Perclos score), combined with the risk characteristics of the current core stage, predict the trend of multi-parameter coordinated change and risk level change in the next 3-5 seconds, and trigger an early warning 2 seconds in advance. For example, in the lifting stage of the frame, predict the coupling trend of "frame X-axis tilt change rate 0.15° / s + lifting height change rate 5cm / s + Perclos score 0.22 and continuously rising", determine that the risk will rise from "medium risk" to "high risk" in 3 seconds, and give an early warning to reserve sufficient time for handling. The risk threshold is dynamically adjusted according to the four-dimensional coupling relationship of "multi-parameter-operating condition-stage-driver status" (e.g., the tilt angle threshold in the lifting stage is tightened by 15% compared to the balancing stage, and the thresholds in each stage are tightened by an additional 20% when the driver is fatigued).

[0077] Step 5: Intelligent Decision-Making and Emergency Response Coordination – Multi-Parameter Coupled Risk Response

[0078] This step is executed sequentially after the timing prediction is completed. It is the implementation stage of the "multi-parameter coupling" logic, and its core solution addresses the pain points of "single decision-making, insufficient control over violations, and lack of driver status handling." Specifically, it is implemented as follows:

[0079] Multi-parameter coupled differential response:

[0080] Alarm: Outputs audible and visual alarms based on multi-parameter coupled risk levels—low risk: no audible or visual alarm; medium risk: short yellow tone + "Please take a break" voice prompt; high risk: long red tone + "Stop operation immediately and check status" voice prompt; extremely high risk: high red tone + "Emergency shutdown and evacuate the site" voice prompt.

[0081] Guidance: Based on multi-parameter coupled risk scenarios, four-dimensional guidance is generated, including text, voice, animation, and host computer instructions. For example, for the coupled risk of "Perclos score 0.23 (fatigue) + amplitude adjustment speed 0.6m / s (dangerous operation)," the guidance can be pushed to "immediately stop amplitude adjustment operation → pull the handbrake → rest for 5 minutes → unlock operation after retesting Perclos score < 0.2." The guidance plan accurately addresses the "personnel-equipment" coupling problem.

[0082] Multi-parameter coupling process lock-up: When multiple parameters of the preceding process fail to meet the safety threshold, the hydraulic action of the next process is automatically locked. The lock-up can only be unlocked after rectification and "multi-parameter synchronization meets the standard". This mechanism eliminates illegal cross-process operation and fatigue operation.

[0083] Emergency Response: In case of extremely high risk, the power supply to the hydraulic pump station will be automatically cut off and the emergency brake will be activated. At the same time, a multi-parameter coupled risk report will be pushed to the remote monitoring platform, which can be manually deactivated after secondary confirmation by the operator.

[0084] The multi-parameter coupling data generated in steps 1 to 5 is uploaded to a remote cloud server via a wireless communication network. Based on the coupled big data from multiple projects, the cloud server periodically updates the local coupling model parameters, dynamic weights of multiple parameters, and self-calibration thresholds for balancing. A multi-level storage system consisting of on-site storage devices, edge cache modules, and cloud backup is adopted to meet the needs of multi-parameter coupling risk tracing and driver operation behavior analysis.

[0085] Multi-parameter coupling model update: Based on multi-project, multi-tower crane multi-parameter coupling big data (parameter coupling relationships and risk evolution patterns under different models, working conditions, stages, and driver states), the local coupling model parameters, multi-parameter dynamic weights, balance self-calibration thresholds, and driver state coupling weights are updated every 15 days in the cloud. For example, if big data reveals that "when a novice driver (operating experience < 1 year) has a Perclos score ≥ 0.2, the risk amplification coefficient is 1.2, which is higher than that of an experienced driver (1.1)," the driver state coupling weight for novice drivers will be optimized to 18%.

[0086] Remote monitoring: Mobile devices can view multi-parameter coupling data, multi-parameter time-series coupling curves, the current core stage, and the real-time curve of the Perclos score in real time. It supports alarm video playback, export of multi-parameter coupling data, and automatic generation of driver operation behavior analysis reports and compliance reports.

[0087] Beneficial effects:

[0088] Compared with the prior art, the present invention has the following significant advantages:

[0089] More precise risk quantification: Clearly define core parameters, allocate a total weight of 100%, identify key risks at each process stage, and integrate coupled analysis of driver status parameters and equipment parameters. The false alarm rate has been reduced from 35% to below 5%, the risk quantification error is ≤3%, and the early warning accuracy rate is ≥95%.

[0090] More proactive early warning: Multi-parameter temporal coupling features are extracted for seven core stages, and risk trends are predicted 3-5 seconds in advance, avoiding the problem of delayed prediction in single-parameter, single-stage, and single-human-machine coupling, and improving the accident avoidance rate by more than 85%.

[0091] More comprehensive control: Establish a coupled relationship between "preceding multi-parameters - current multi-parameters - driver status", and prevent illegal and fatigued operation through process lock-in mechanism. Improve the efficiency of novice drivers in handling complex risks with multiple parameters by 60%;

[0092] The system is more reliable: multi-parameter cross-validation and fault self-switching reduce the sensor fault false judgment rate to below 0.5%; UDP transmission and edge buffering ensure real-time coupling and transmission of multi-parameter and driver status data; the enhanced tilt sensor's temperature drift self-compensation function ensures accurate parameter acquisition under extreme temperatures; and the overall reliability is ≥99.9%.

[0093] Easier installation and adaptation: Standardized magnetic base + software parameter coupling calibration, no need for complicated physical processing, driver fatigue detection reuses existing smart cameras, no need for new hardware, adapts to different tower crane size differences and different driver operating position layouts, reduces the cost of retrofitting old tower cranes by 50% and increases installation efficiency by 60%; Attached Figure Description

[0094] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments, and the advantages of the present invention in the above and / or other aspects will become clearer.

[0095] Figure 1 This is a flowchart of the overall method of the present invention.

[0096] Figure 2 This is a flowchart of the multi-parameter coupling and early warning process of the core sub-process of this invention. Detailed Implementation

[0097] Example 1: QTZ63 tower crane (service life 3 years, lifting height 80m, number of attached layers 3, operator's operating experience 2 years)

[0098] This embodiment follows a forward flow and linkage logic based on "multi-parameter coupling":

[0099] Step 1: Multimodal data acquisition:

[0100] Hardware installation: 13 OMCH E3F-DS30N1 photoelectric switches are deployed to collect status parameters; 4 laser-assisted sensors are used for auxiliary status parameter coupling verification; 4 US-015 ultrasonic radars are used to collect distance parameters; 2 SINDT-CAN enhanced tilt sensors are used to collect tilt parameters; 20 HIKVISION smart cameras are deployed, 19 of which cover the key jacking area to achieve image-based and sensor-based parameter coupling verification; 1 camera is deployed near the operator's position to collect facial images of the operator to detect fatigue; multi-parameter acquisition benchmark coupling calibration is completed by inputting the QTZ63 tower crane model and installation position offset of 0cm into the host computer.

[0101] Data Acquisition Startup: The sensor acquires 15 types of coupling parameters, i.e., core parameters, at 10Hz. The intelligent camera at the driver's operating position acquires facial images at 30 frames per second, simultaneously monitoring status, mechanical, environmental, and driver status parameters. Among them, the SINDT-CAN enhanced tilt sensor accurately acquires the dual-axis tilt angle of the frame and the tilt angle of the tower body in a field environment of 28℃. The driver's Perclos reference value is 0.08, providing input for coupling calculation.

[0102] Step 2: Data Preprocessing and Transmission

[0103] Preprocessing: The collected data undergoes two sampling comparisons for denoising and adaptive Kalman filtering. The driver's facial image is processed by grayscale conversion and denoising to extract eyelid features, and the Perclos score is calculated to be 0.09. The core parameters are cross-validated by multiple parameters of "photoelectric switch + laser sensor". If the error is within 0.2 seconds, it is considered valid. If there is no sensor health status alarm, the multi-parameter acquisition stability coupling judgment is qualified.

[0104] Transmission: The ESP32-Pro module transmits multi-parameter data wirelessly via UDP to the target IP, i.e., the host computer (or local monitoring terminal) deployed on site. The data includes raw data, cross-validation results, Perclos scores, and intermediate data of multi-parameter coupled calculation synchronously stored in the edge cache.

[0105] Step 3: Quantifying cross-process coupling risks:

[0106] Importing basic parameters:

[0107] Operating parameters: Lifting height Hactual = 80m, Number of attachment layers = 3, Service life = 3 years (corresponding to Ksupplement = 1.0 × 1.0 = 1.0);

[0108] Acquisition and preprocessing parameters: actual amplitude length =28m, theoretical amplitude reference value =30m (QTZ63 tower crane preset); actual tower inclination angle =0.4°, theoretical tower inclination angle benchmark value =0.5°; Actual lifting height =80m, theoretical lifting height benchmark value =100m (preset according to working conditions);

[0109] Related parameters: All preceding processes are compliant (Ncombined / Ntotal = 1), cross-process linkage coefficient = 1; Standard weight =1.0; The theoretical balance benchmark value γtheoretical for QTZ63 tower crane is 4.466;

[0110] Driver status parameter: Perclos=0.09<0.2, no risk correction.

[0111] Calculation of tower overturning moment Mtilt

[0112]

[0113]

[0114] M-tilt risk contribution: When M-tilt ≤ 50 kN·m, it is... When Minclination > 50 kN·m, it is 100%.

[0115] The specific parameter value is air density. =1.225 kg / m³ (value taken under standard atmospheric pressure). The wind load shape coefficient is taken as 1.3, and the windward area of ​​the tower body S = 12m² (derived from the cross-sectional dimensions of the QTZ63 tower crane). The actual measured wind speed on site... =5m / s, amplitude L = 28m, lifting height H = 80m, reference lifting height = 100m, process correction factor = 0.8 (value taken during balancing and calibration), working condition compensation coefficient = 1.0 × 1.0 = 1.0 (1.0 is used for service years < 5 years; 1.0 is used for 3 attached layers). Substituting into the formula, the overturning moment can be calculated. The value is 5.35 kN·m, with a risk contribution rate of 10.7%.

[0116] Stress balance of guide wheel calculate

[0117]

[0118] η-lead risk contribution: 0% when η-lead ≥ 85%, and 0% when η-lead < 85%. The number of guide wheels is n=8, and the forces Fi on each guide wheel are 25, 25, 24, 26, 25, 24, 26, and 25 respectively. The average force on each guide wheel is Fm = 25 kN. Substituting these values ​​into the formula, we can obtain... It is 98% (≥85%, risk contribution rate 0%).

[0119] Safety factor of hanging boot load-bearing capacity calculate

[0120]

[0121] S 挂 Risk contribution: 0% when S_sup>≥1.3, and 0% when S_sup><1.3.

[0122] The actual load-bearing capacity of the hanging boot =100KN, allowable load capacity of the hanging shoe =100KN, climbing claw positioning coefficient Kclimb=1 (climbing claw positioning meets the standard), substitute into the formula to calculate. =1, risk contribution rate 23.08%

[0123] Balanced state matching degree The calculation uses a normalized weighted algorithm, and the specific formula is as follows:

[0124]

[0125] γ 配 Risk contribution: when γ 配 When ≥90%, it is 0%; when γ 配 <90% is

[0126] Among them, 56%, 27%, and 17% are weighting coefficients, quantified based on the Analytic Hierarchy Process (AHP), corresponding to the degree of influence of amplitude, tower tilt angle, and jacking height on the tower crane's balancing status. Substituting these values ​​into the data acquisition and preprocessing parameters yields the following results. =87.3% (<90%), indicating that there is a deviation between the actual balancing state and the theoretical benchmark, with a risk contribution of 3.11%.

[0127] Cross-process linkage coefficient risk contribution: All preceding conditions are compliant, K-link is 1, and risk contribution is 0.

[0128] Comprehensive risk probability calculation:

[0129] Weights for balancing phase: Risk contribution rate × 40% + M-bias risk contribution rate × 20% + Risk contribution rate × 15% + S-hanging risk contribution rate × 10% + cross-process linkage coefficient risk contribution rate × 15%;

[0130] The overall risk probability is calculated as (3.11% × 40%) + (10.7% × 20%) + (0% × 15%) + (23.08% × 10%) + (0% × 15%) = 5.7%, which is considered low risk.

[0131] Step 4: Phased time series prediction:

[0132] Current core stage: Balancing and calibration stage (corresponding parameter coupling focus: balancing state matching degree + luffing length + tower tilt angle + jacking height + Perclos score and luffing operation coupling characteristics).

[0133] Multi-parameter temporal feature extraction: Extract the coupled features of "X-axis tilt change rate of the frame 0.12° / s + balance state change rate 1% / s + tower tilt angle change rate 0.08° / s + Perclos score 0.18 and amplitude change rate 0.4m / s";

[0134] Coupling prediction: Based on the time-series coupling features specific to the balancing stage, the LSTM-lightweight edge computing model predicts that after 3 seconds, the balancing matching degree will be 85%, the tower tilt angle will be 0.55°, the Perclos score will be 0.23 (fatigue), and the risk probability will be 62%×1.15=71.3% (high risk), triggering an early warning 2 seconds in advance.

[0135] Current core stage: Balancing and calibration stage (corresponding parameter coupling focus: balancing state matching degree + luffing length + tower tilt angle + jacking height + Perclos score and luffing operation coupling characteristics).

[0136] Temporal feature extraction: Based on the LSTM-lightweight edge computing model, extract the temporal coupling features (tower tilt angle change rate, guide wheel force fluctuation, and hanging shoe bearing force change) during the balancing stage.

[0137] Coupling prediction: Based on the time-series coupling characteristics specific to the balancing stage, the LSTM-lightweight edge computing model predicts that after 3 seconds, the balancing matching degree will drop to 50% (core parameter deterioration), the overturning moment will rise to 48kN·m, the force balance of the guide wheel will drop to 78%, the actual bearing capacity of the hanging shoe will rise to 120kN, and the Perclos score will be 0.23 (fatigue, fatigue correction coefficient 1.4). The calculated comprehensive risk probability is 49.3%, of which parameter deterioration increases the risk, and the risk threshold of fatigue state is tightened and corrected to reach 69.02%, triggering an early warning 2 seconds in advance.

[0138] Step 5: Intelligent Decision-Making and Emergency Response Coordination:

[0139] Multi-parameter coupling response: The host computer outputs a long red alarm tone indicating high risk, and a pop-up four-dimensional operation guide says "Stop the amplitude change operation immediately → Pull the handbrake → Rest for 5 minutes → Retest the Perclos score < 0.2 and then unlock the operation", and the voice broadcast says "Please stop the operation immediately and check the status".

[0140] After rectification: The driver rested for 5 minutes, the Perclos score was 0.07, multiple parameters met the standard (the matching degree increased to 93%), the coupling risk probability was 25% (low risk), and the process was unlocked.

[0141] Simultaneously, multi-parameter coupled data is uploaded to the cloud and third-party monitoring platforms; 15 days later, the cloud pushes an update package based on multi-project multi-parameter coupled big data, optimizing the weight of laser sensors in the balancing stage to 5%, optimizing the driver status coupling weight, and modifying the operating experience of 2-year drivers to 13%.

[0142] Example 2: Old QTZ80 tower crane (service life 12 years, lifting height 100m, number of attached layers 4, operator's operating experience 10 years)

[0143] Steps 1-2: Data collection shows multi-parameter coupling anomalies—At the ambient temperature of -5℃, the SINDT-CAN enhanced tilt sensor, after correcting errors through temperature drift self-compensation, collects the following data: X-axis tilt angle of the frame is 0.7°, Y-axis tilt angle is 0.6° (abnormal mechanical parameters), the climbing claw is not in place (abnormal status parameters), and the cross-process linkage coefficient is 0.8 (minor violation of preceding bolts, multi-parameter coupling judgment). The driver's operating position camera collects facial images, and the Perclos score is calculated to be 0.26 (fatigue). After preprocessing through multi-parameter cross-validation (including Perclos and slewing operation coupling verification), the data is effectively transmitted via UDP wireless transmission.

[0144] Step 3: Multi-parameter coupling risk quantification. The "climbing claw status + tilt angle + cross-process coefficient" are coupled and calculated according to the weight of the new section installation stage (bolted connection status 45% + standard section in place status 30% + cross-process linkage coefficient 10% + double-axis tilt angle of the lifting frame 9% + distance between the lifting frame and the tower crane 6%). The risk probability is 93%. After adding the driver fatigue correction (93% × 1.15 = 106.95%), it is judged as extremely high risk.

[0145] Step 4: Multi-parameter temporal coupling prediction - Current core stage: scaffold lifting stage (corresponding parameter coupling focus: scaffold dual-axis tilt angle + lifting height change rate + wind speed + hanging shoe load-bearing safety factor + Perclos score and lifting operation coupling characteristics). Predict that after 2 seconds, the coupling risk will increase due to "dual-axis tilt angle rising to 0.8° + climbing claw not in place + lifting height continuing to rise + Perclos score 0.28", triggering an extremely high risk warning 2 seconds in advance.

[0146] Step 5: Emergency Linkage - Automatically cut off hydraulic power + emergency braking, red high-frequency alarm on the host computer, push multi-parameter coupling rectification guidance ("Clean foreign objects from the climbing claw ear plate + tighten the lower slewing support bolts + the driver should rest immediately → the climbing claw is in position and meets the standard + unlock after Perclos < 0.2"), and announce "Emergency stop, evacuate the site" in voice.

[0147] This invention provides a concept and method for multi-parameter coupled prediction, early warning, and intelligent decision-making in tower crane jacking. Many methods and approaches exist for implementing this technical solution; the above description is merely a preferred embodiment of the invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications should also be considered within the scope of protection of this invention. All components not explicitly stated in this embodiment can be implemented using existing technologies.

Claims

1. A multi-parameter coupled prediction, early warning, and intelligent decision-making method for tower crane jacking, characterized in that, Includes the following steps: Step 1: Multimodal data acquisition. Through a multimodal intelligent monitoring module integrating multiple sensors, multiple core coupled parameters covering the dimensions of state safety, mechanical balance and environmental adaptation are collected simultaneously. Step 2: Data preprocessing and transmission. The data collected in Step 1 is preprocessed, and the validity of the data is ensured through a multi-parameter cross-validation mechanism. Step 3: Cross-process coupling risk quantification. Based on the multi-class parameters preprocessed in Step 2, a three-level risk quantification model containing root nodes, intermediate coupling nodes, and leaf nodes is constructed by combining an improved probabilistic graphical network model with a dynamic balancing self-calibration algorithm. The weights of each parameter are dynamically allocated according to different lifting processes, and the hierarchical coupling risk level is calculated and output. Step 4: Phased time series prediction. Based on the risk level and its time series changes output in Step 3, the time series prediction model is used to analyze the trend of multi-parameter coordinated changes, predict the risk level changes in future periods, and trigger early warning information in advance based on the prediction results. Step 5: Intelligent decision-making and emergency response linkage. Based on the hierarchical coupling risk level, output multi-dimensional early warnings and operational guidance, and automatically interlock control or execute emergency shutdown for subsequent processes according to the risk level.

2. The multi-parameter coupled prediction, early warning, and intelligent decision-making method for tower crane jacking according to claim 1, characterized in that, The core coupling parameters include, but are not limited to: the bolt connection status between the lower slewing support and the tower column; the connection status of the new standard section; the positioning status of the safety pin of the jacking beam; the positioning status of the climbing claw; the positioning status of the introduction section; the X-axis tilt angle of the jacking frame; the Y-axis tilt angle of the jacking frame; the tower body tilt angle; the jacking height; the slewing angle; the amplitude; the wind speed; the distance between the jacking frame and the tower crane; the matching degree of the balancing status; and the health status of the sensors.

3. The multi-parameter coupled prediction, early warning, and intelligent decision-making method for tower crane jacking according to claim 1, characterized in that, Step 2 includes the following steps: Step 2-1, Data preprocessing: Denoising and filtering are performed on the collected synchronous data. The denoising process includes data stability verification based on time-series sampling. Step 2-2, Verification and Reliability Assurance: The core parameters are verified for effectiveness using a multi-parameter cross-validation mechanism, and the system has the ability to automatically switch to a backup sensor in case of sensor failure. The multi-parameter cross-validation mechanism is based on dual logic implementation: same-dimensional multi-sensor redundancy verification and related-dimensional parameter collaborative verification. Specifically, the same-dimensional multi-sensor redundancy verification is as follows: for a single core parameter, at least two different types of sensors are used to collect and compare data simultaneously for verification; related-dimensional parameter collaborative verification is as follows: for core parameters that are logically related, the validity of the target parameter is inferred from the rationality of the related parameters. Steps 2-3, Data transmission and caching: The processed and verified data is transmitted in real time via wireless communication and stored locally via an edge caching module during transmission. The edge caching module activates a backup transmission channel when the network is interrupted.

4. The multi-parameter coupled prediction, early warning, and intelligent decision-making method for tower crane jacking according to claim 1, characterized in that, Step 3 includes the following steps: Step 3-1, Risk Quantification Model Construction: Based on the preprocessed multi-class parameters, a three-level risk quantification model containing root nodes, intermediate coupling nodes, and leaf nodes is constructed by combining an improved probabilistic graphical network model with a dynamic balancing self-calibration algorithm. Step 3-2, Coupling Parameter Calculation: Based on the model in Step 3-1 and the set coupling logic, the multi-parameter coupling analysis index represented by the intermediate coupling nodes is generated by calculating the parameters of the root node. Step 3-3, Dynamic Evaluation and Output: Based on the aforementioned coupling analysis indicators and combined with the current lifting process, dynamically allocate the weights of each parameter and perform comprehensive calculations, finally outputting the corresponding risk level at the leaf node.

5. The multi-parameter coupled prediction, early warning, and intelligent decision-making method for tower crane jacking according to claim 4, characterized in that, The specific architecture of the three-level risk quantification model, which includes a root node, intermediate coupling nodes, and leaf nodes, in step 3-1 is as follows: The root node consists of the core coupling parameters and operating parameters determined in step 1. The operating parameters include, but are not limited to, jacking height, number of attachment layers, and service life of the tower crane. The intermediate nodes are multi-parameter coupled analysis indicators, including but not limited to tower overturning moment, guide wheel force balance, hanging shoe bearing safety factor, balancing state matching degree, and cross-process risk linkage coefficient; Tower overturning moment Calculated using the following formula: in, It is air density. S is the wind load shape coefficient, and S is the windward area of ​​the tower. For wind speed, H0 is the jacking height, and H0 is the reference jacking height. For amplitude, This is a process correction factor. This is the working condition compensation coefficient; Based on the value taken during the jacking phase The value is determined based on the service life and the number of attachment layers. For service life less than 5 years, the value is 1.0; for every additional 5 years thereafter, the value increases by 0.

1. For attachment layers less than 3 layers, the value is 1.0; for each additional layer, the value increases by 0.

08. K 工 The seven core stages of the jacking process are assigned values ​​based on the differences in mechanical stress intensity and risk level at each stage: 0.9 for the pre-jacking preparation stage, 0.8 for the balancing and calibration stage, 1.2 for the jacking stage of the support frame, 1.0 for the standard section introduction stage, 1.1 for the new section installation and fastening stage, 1.05 for the support frame return stage, and 0.9 for the post-operation review stage; the tower overturning moment... The safety threshold is 50 kN·m; Stress balance of guide wheel Calculated using the following formula: Where n is the number of guide wheels, F i For the actual force on a single guide wheel, F m The average force on the guide wheel is 85%; Safety factor of hanging boot load-bearing capacity Calculated using the following formula: in For the actual load-bearing capacity of the hanging boots, Preset according to tower crane model Values ​​are assigned based on the crawler's positioning status: 1.0 for meeting the standard and 0.6 for not meeting the standard; the safety threshold is 1.

3. Balanced state matching degree Calculated using the following formula: This is the actual amplitude variation length. This is the theoretical amplitude reference value. This is the actual tilt angle of the tower. This is the theoretical reference value for the tower's inclination angle. This is the actual lifting height. This is the theoretical lifting height benchmark value; This is the cross-process linkage coefficient. The standard weighting is used, and the safety threshold is 90%. The cross-process risk linkage coefficient is determined based on the compliance rate of the core parameters of the preceding process. The compliance rate of the core parameters of the preceding process is specifically Ntotal / Ntotal, where Ntotal is the number of compliant parameters and Ntotal is the total number of parameters that need to be verified. The specific value rules are as follows: if the preceding process is fully compliant, take 1, then Ntotal / Ntotal = 1; if there is a minor violation, take 0.5, then Ntotal / Ntotal ≥ 0.67; if there is a serious violation, take 0.1, then Ntotal / Ntotal < 0.

67. Leaf nodes represent four risk levels: low, medium, high, and extremely high. Risk probability < 30% is defined as low risk, 30% ≤ probability < 60% as medium risk, 60% ≤ probability < 85% as high risk, and probability ≥ 85% as extremely high risk.

6. The multi-parameter coupled prediction, early warning, and intelligent decision-making method for tower crane jacking according to claim 1, characterized in that, Step 4 includes the following steps: Step 4-1, Stage Division and Feature Extraction: Divide the lifting operation process into multiple core stages, and extract time-series coupling features that reflect the trend of coordinated change based on multi-parameter time-series data; Step 4-2, Temporal Risk Prediction: Based on the temporal coupling characteristics obtained in Step 4-1, predict the risk level changes within a specified future time period; Step 4-3, Dynamic Early Warning Trigger: Based on the prediction results and combined with dynamically adjusted risk thresholds, an early warning is triggered before the risk level exceeds the threshold.

7. The multi-parameter coupled prediction, early warning, and intelligent decision-making method for tower crane jacking according to claim 1, characterized in that, Step 5 includes the following mechanisms: Multi-dimensional early warning and guidance mechanism: Based on the risk level, output multi-dimensional early warning information and operation guidance including sound, light, text, images and voice; Process interlocking control mechanism: When multiple parameters of the preceding process fail to meet the standards, the hydraulic action of the next process is automatically locked, and the action command of the hydraulic pump station is cut off through electronic control signal, thus avoiding the risk transmission across processes from the mechanism. Emergency response mechanism: An emergency shutdown mechanism is designed to be independent of the conventional control. It supports manual cancellation after secondary confirmation by the operator and requires two confirmations from the operator to cancel the linkage safety design.

8. The multi-parameter coupled prediction, early warning, and intelligent decision-making method for tower crane jacking according to claim 1, characterized in that, The multi-parameter coupling data generated in steps 1 to 5 is uploaded to a remote cloud server via a wireless communication network. Based on the coupled big data from multiple projects, the cloud server periodically updates the local coupling model parameters, dynamic weights of multiple parameters, and self-calibration thresholds for balancing. A multi-level storage system consisting of on-site storage devices, edge cache modules, and cloud backup is adopted.

9. The multi-parameter coupled prediction, early warning, and intelligent decision-making method for tower crane jacking according to claim 1, characterized in that, In step 1, the driver's facial image is captured in real time by a camera, and the Perclos score is calculated. When the Perclos score is ≥0.2, it is determined that the driver is fatigued, the warning sensitivity is improved by 40%, and the risk level judgment threshold is tightened by 30%.

Images