A lift stage structure control system and method

Abstract

The application discloses a lifting stage structure control system and method, and relates to the technical field of stage control.The lifting stage structure control system and method comprises the following steps: S1, collecting multi-source state data of stage equipment in real time, performing data preprocessing, and constructing a multi-source state database; S2, constructing a motion state space model, predicting a time sequence motion sequence at a future time, determining whether stage equipment is overlapped, predicting a collision risk, and taking risk avoidance measures; S3, determining the obstacle avoidance priority of stage equipment, generating a stage equipment motion scheme, and issuing a control instruction; S4, executing a motion operation, evaluating the health degree of stage equipment in real time, and implementing a hierarchical early warning measure; and S5, displaying a stage situation and a risk based on the collision risk and the health degree evaluation result, and generating an analysis report.The application solves the problem that potential collision risks cannot be avoided in time due to dynamic path intersection caused by the collaborative work of multiple stage equipment in a complex stage scene.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of stage control, in particular to a lifting stage structure control system and method. BACKGROUND

[0002] Modern stage performances have increasingly complex requirements for visual effects and movement arrangements, and various moving units such as lifting platforms, moving scenery, and mechanical arms need to operate in coordination in the same space at a high frequency, which brings the risk of path intersection and potential collision. With the improvement of sensors, network communication, and computing power, multi-source data such as device position, speed, and attitude can be collected in real time, and future movements can be inferred through fusion and prediction algorithms. At the same time, the development of deep learning and optimization algorithms makes it possible to dynamically avoid in complex scenarios.

[0003] For example, the invention patent with publication number CN103543687B discloses a lifting stage control device and control method, which includes a main controller, a PLC controller, a plurality of servo motors, a plurality of lifting platforms with the same number of servo motors, a pressure sensor provided on each lifting platform, at least one position sensor provided on each lifting platform, and a speed sensor provided on each servo motor; the PLC controller is electrically connected with each servo motor, and each servo motor is electrically connected with the corresponding lifting platform; the speed sensor, the pressure sensor, and the position sensor are electrically connected with the main controller; the main controller is provided with a display screen. The present application can ensure the efficiency of stage lifting conversion and the safety needs of performers during performance, has low modification cost, and can realize various running modes of the stage in the unloaded and loaded states.

[0004] For example, the invention patent with publication number CN103235546B discloses a stage lifting mechanism reminding system, which includes a lifting platform control device and a reminding device. The lifting platform control device has a lifting instruction input unit for inputting a lifting instruction. The wireless transceiver unit of the lifting platform control device communicates wirelessly with the reminding device; wherein the control unit of the lifting platform control device is used to send a start signal to the reminding device through the wireless transceiver unit when the lifting instruction is received from the lifting instruction input unit of the lifting platform control device, and to control the lifting mechanism according to the received lifting instruction after a first predetermined time after sending the start signal to the reminding device; the reminding device continuously sends a reminder information after receiving the start signal from the wireless transceiver unit, and stops sending the reminder information after a second predetermined time after sending the reminder information; wherein the reminder information is one or more of flashing light, buzzing sound, and vibration.

[0005] However, in the process of implementing the technical scheme of the present application, the above-mentioned technology at least has the following technical problems:

[0006] In complex stage settings, multiple motion units such as lifting platforms, moving scenery, and robotic arms work together frequently in a limited space, which presents various dynamic uncertainties: the movement of actors or temporary props, changes in the performance process, and deviations in equipment execution may all lead to the risk of path intersections and collisions.

[0007] Therefore, in order to address the above problems, there is an urgent need for a control system and method for lifting stage structures. Summary of the Invention

[0008] Technical problems to be solved

[0009] To address the shortcomings of existing technologies, this invention provides a lifting stage structure control system and method, which solves the problem that potential collision risks caused by dynamic path intersections due to the collaborative operation of multiple stage devices in complex stage scenes cannot be avoided in a timely manner.

[0010] Technical solution

[0011] To achieve the above objectives, the present invention provides the following technical solution: a method for controlling a lifting stage structure, comprising the following steps: S1, real-time acquisition of multi-source state data of stage equipment using multiple types of sensors, data preprocessing of the multi-source state data, and real-time storage of the multi-source state data in a multi-source state database; S2, acquisition of multi-source state data, construction of a motion state space model, prediction of future temporal motion sequences, determination of whether stage equipment overlaps based on the predicted temporal motion sequences, prediction of collision risks for each stage equipment, and implementation of avoidance measures based on the collision risk prediction results; S3, determination of obstacle avoidance priority of stage equipment based on collision risk prediction results, generation of stage equipment motion schemes based on obstacle avoidance priority, and issuance of control commands; S4, receipt of control commands and execution of motion operations, real-time assessment of the health of stage equipment actions based on multi-source state data, and implementation of graded early warning measures based on the health assessment results; S5, display of stage status and risks based on collision risk and health assessment results, and support for traceable simulation and generation of analysis reports.

[0012] Furthermore, the process of collecting multi-source status data of stage equipment in real time through multiple types of sensors, preprocessing the multi-source status data, and constructing a multi-source status database for real-time storage is as follows: Multi-source status data of each stage device is collected in real time using various sensors. The collection process includes: deploying infrared rangefinders at key locations on each stage device to collect spatial position information; using encoders to collect real-time displacement and velocity data of each stage device; and using IMUs to collect three-axis acceleration, angular velocity, and attitude angle data. The multi-source status data includes spatial position information, displacement, velocity, acceleration, angular velocity, and attitude angle data. Simultaneously, global data is acquired in real time through cameras on the top and sides of the rising and falling stage. Visual images are used to record and measure the type and maximum external diameter of each stage device; clock offsets from multi-source state data from different sensors are corrected, and unified timestamp alignment is performed. Kalman filtering is used to remove measurement noise and glitches, and packet loss interpolation and abnormal multi-source state data are detected using the sliding window method; a unique ID and attribute description are assigned to each sensor, the data structure format of multi-source state data is unified, and the data is packaged according to the interface protocol, and the multi-source state data is standardized and normalized; a multi-source state database is built, and multi-source state data is stored in real time according to sensor ID and time order index; at the same time, environmental change events are recorded synchronously, and target point information of each stage device is recorded in real time.

[0013] Furthermore, the process of acquiring multi-source state data, constructing a motion state space model, and predicting the temporal motion sequence at future moments, and determining whether stage equipment overlaps based on the predicted temporal motion sequence, is as follows: Multi-source state data is acquired to form a temporal motion sequence, which includes spatial position information, velocity, attitude angle data, and a timestamp; a historical temporal feature matrix is ​​constructed based on historical step lengths and the temporal motion sequence; this historical temporal feature matrix is ​​input and trained using a Long Short-Term Memory (LSTM) network to construct the motion state space model, ultimately outputting the predicted temporal motion sequence for future moments, including predicted spatial position information, predicted velocity, predicted attitude angle data, and a timestamp; the output predicted temporal motion sequence is segmented according to time slices to form a dynamic path set; the OBB bounding box algorithm is used to determine whether each stage piece spatially overlaps at any given moment, outputting the pairs of spatially overlapping pieces and their time slice indices.

[0014] Furthermore, the collision risk of each stage device is predicted, and the specific process of taking avoidance measures based on the collision risk prediction results is as follows: The predicted temporal motion sequence output by the motion state space model is obtained; the Euclidean distance between two devices is calculated using the predicted spatial position information to obtain the distance between the spatial center points of device i and device j at time t in the future; the maximum outer diameter of the stage equipment is obtained and its average value is calculated to obtain the safe distance threshold between the two devices; the pairs of devices that overlap spatially and their time slice indices are obtained; the total overlap time of device i and device j is obtained by multiplying the number of time slices of spatial overlap between the two devices by the duration of the time slices; the predicted spatial position information and predicted velocity of device i and device j are obtained; the velocity vector is calculated using the difference in position, and then the velocity vector is cosine-sine-sigma ... The velocity vector angle between the two devices at time t is calculated. The distance influence value is obtained by subtracting the ratio of the distance between the spatial center points of device i and device j at future time t to the safe distance threshold using a constant. The distance influence value is kept positive; if the distance influence value is less than 0, it is set to 0; otherwise, the original distance influence value is maintained. The distance influence value is multiplied by the distance safety margin weighting factor to obtain the distance safety margin term. The overlap time influence term is obtained by multiplying the overlap time weighting factor by the ratio of the total overlap time of device i and device j to the prediction time window length. The directional risk term is obtained by subtracting the velocity vector angle between the two devices at time t using a constant and then multiplying it by the motion direction weighting factor. The collision risk prediction value is obtained by adding the distance safety margin term, the overlap time influence term, and the directional risk term.

[0015] Furthermore, the specific process of taking avoidance measures based on collision risk prediction results is as follows: The predicted collision risk value is compared in real time with the primary and secondary risk thresholds. When the predicted collision risk value is less than the primary risk threshold, it is determined to be a low collision risk, and the equipment runs along its original path. The predicted collision risk value is recorded in the background log. When the predicted collision risk value is greater than or equal to the primary risk threshold but less than the secondary risk threshold, it is determined to be a moderate collision risk. The equipment speed is automatically reduced, and the stage equipment trajectory is fine-tuned for dynamic avoidance. Using a spline path optimization local path fine-tuning algorithm, equipment with moderate collision risk makes slight detours to automatically avoid the predicted high-risk areas. High-risk areas; if multiple moderate-risk collision devices are predicted to meet in the same small area, they will coordinate to pass through each other; when the predicted collision risk value is greater than or equal to the level 2 risk threshold, it is judged as a high-risk collision area, an early warning is immediately issued, and the current action of the high-risk device is suspended; temporary waiting is set for the relevant devices; the path is dynamically adjusted to temporarily avoid the conflict area and postpone the planned action sequence; the global task queue is intelligently adjusted, and the priority of non-critical tasks is temporarily reduced; if the high-risk state is not resolved, the stage equipment will automatically roll back to the previous safe position; and after a high-risk action is triggered, the relevant stage equipment will perform a rapid health status self-check.

[0016] Further, the specific process of determining the obstacle avoidance priority of stage equipment based on the collision risk prediction results, generating a motion plan for the stage equipment based on the obstacle avoidance priority, and issuing control commands is as follows: Acquire multi-source state data; based on the spatial position information of equipment i, for equipment i, traverse all other equipment, calculate the Euclidean distance between two equipment and filter out the minimum value to obtain the minimum spatial distance between equipment i and other equipment; acquire the velocity of equipment i, calculate the velocity components in three-dimensional space using the difference method, and then calculate the magnitude to obtain the velocity magnitude of equipment i; based on the spatial position information of equipment i and the displacement of the stage equipment to the target point, calculate the distance through coordinates to obtain the straight-line distance from equipment i to the target point; divide the straight-line distance from equipment i to the target point by the velocity magnitude of equipment i to obtain the remaining completion time of the current task of equipment i; divide the minimum spatial distance between equipment i and other equipment by the product of the velocity magnitude of equipment i and the straight-line distance from equipment i to the target point to obtain the spatial velocity priority factor; for equipment... The remaining time elasticity factor is obtained by taking the natural logarithm of the sum of the remaining completion time of the current task and the constant 1. The obstacle avoidance priority value of device i is obtained by adding the spatial velocity priority factor to the remaining time elasticity factor. All devices with medium and high collision risks are sorted according to their obstacle avoidance priority values. Devices with higher obstacle avoidance priority values ​​are given priority to maintain their original path and pass through at normal speed. Devices with lower obstacle avoidance priority values ​​dynamically decelerate and automatically adjust their current trajectory, shifting slightly towards a safe area. If there is a conflict with a high-priority device, temporary waiting is performed. At the same time, for the obstacle avoidance scheme generated based on the obstacle avoidance priority value, the scheme details are optimized using dynamic window algorithm and greedy algorithm in combination with the global scheduling goal, and the final device motion scheme is output, including: motion path, speed and action plan. According to the final device motion scheme, motion control commands for each device are automatically issued, including speed, path and action delay. All adjustment processes, collision risk prediction values, obstacle avoidance priority values ​​and actual control actions are recorded throughout the process.

[0017] Further, the specific process of receiving control commands and executing motion operations, and evaluating the health of stage equipment's actions in real time based on multi-source state data, is as follows: The system receives motion control commands from the intelligent avoidance decision module, parses the target action of each stage equipment, and sends the parsed commands to each stage equipment, automatically starting the equipment to execute motion operations; it acquires the actual attitude angle data of device i when it reaches the target point, and simultaneously acquires the predicted attitude angle data of device i output by the motion state space model; it subtracts the predicted attitude angle data from the actual attitude angle data and takes the absolute value to obtain the absolute attitude error; it acquires multiple sets of historical absolute attitude errors, and calculates the average value based on the sliding time window length as the attitude error tolerance value; it acquires the command timestamp recorded when the command is issued and the completion timestamp recorded when device i actually reaches the target point, and subtracts the command timestamp from the completion timestamp to obtain the actual action completion delay; it acquires multiple sets of historical action completion delays, and calculates the average value based on the sliding time window length as the maximum action delay duration; it subtracts the ratio of the absolute attitude error to the attitude error tolerance value from a constant, and then subtracts the ratio of the actual action completion delay to the maximum action delay duration to obtain the action health assessment value of device i.

[0018] Furthermore, the specific process of implementing graded early warning measures based on health assessment results is as follows: Real-time comparison of the action health assessment value with the first-level and second-level health thresholds. When the action health assessment value is greater than or equal to the second-level health threshold, the action is judged to be in good health, and execution continues with continuous dynamic relaxation monitoring and periodic small-amplitude calibration movements to verify sensor and control accuracy. When the action health assessment value is greater than or equal to the first-level health threshold but less than the second-level health threshold, an early warning state is declared. Using log records and multi-source status data, a lightweight diagnostic algorithm is automatically run to determine whether there is sensor drift, hysteresis in dynamic response, sudden environmental changes, or mismatched control parameters, and a targeted fine-tuning plan is generated. If the equipment's action health assessment value remains within the early warning range and shows a downward trend, some tasks are automatically split and transferred to other equipment with higher action health assessment values. Based on historical data... The system dynamically adjusts the attitude error tolerance and maximum action delay based on the action execution status and environmental characteristics. When the action health assessment value is less than the first-level health threshold, it is determined to be a serious abnormality in action execution, an emergency stop operation is initiated, the abnormal equipment action is immediately isolated, and tasks are reassigned. After the emergency stop, a rapid self-check sub-process is automatically executed, and possible fault points are automatically located by combining historical logs, generating a preliminary diagnostic report. All multi-source status data, action health assessment values, control commands, and environmental information before and after the equipment abnormality are automatically summarized to construct a detailed event package. The action health assessment values ​​and execution status of the equipment are transmitted back in real time, and the action health assessment values, warnings, and effects of executed automation measures of each equipment are presented in real time. At the same time, reports are automatically generated periodically to analyze the action health assessment distribution, failure rate, self-healing success rate, and collaborative obstacle avoidance efficiency. If multiple devices of the same model are running in parallel, feedback data is shared.

[0019] Furthermore, the specific process of displaying stage status and risks based on collision risk and health assessment results, and supporting traceable simulation and analysis report generation, is as follows: The entire platform structure, stage equipment, motion obstacles, and motion paths are rendered in real time using a 3D engine, highlighting risk areas and avoidance actions; if a high-risk collision or abnormal equipment action is detected, pop-up windows and audible / visual alarms are triggered, along with a pause prompt; manual commands are supported for maintenance personnel, and all manual operations and responses are recorded; every command, risk event, and equipment status log is automatically recorded according to time and event indexes, and multi-condition retrieval is supported; simultaneously, motion simulation playback, collision reproduction, and anomaly cause analysis can be performed on any historical time period, and statistical reports and event analysis reports are automatically generated.

[0020] The second aspect of this invention provides a lifting stage structure control system, comprising: a perception and environment modeling module, a motion prediction and collision detection module, an intelligent avoidance decision-making module, a motion control and execution module, and a visualization and backtracking module; wherein the perception and environment modeling module is used to collect multi-source state data of the stage equipment in real time through multiple types of sensors, preprocess the multi-source state data, and construct a multi-source state database to store the multi-source state data in real time; the motion prediction and collision detection module is used to acquire multi-source state data, construct a motion state space model, predict the temporal motion sequence of future moments, and determine the stage equipment based on the predicted temporal motion sequence. The system includes: a collision risk prediction module to determine whether there is overlap and to predict the collision risk of each stage device; an intelligent obstacle avoidance decision module to determine the obstacle avoidance priority of stage devices based on the collision risk prediction results, generate motion plans for stage devices based on the obstacle avoidance priority, and issue control commands; a motion control and execution module to receive control commands and execute motion operations, evaluate the health of stage devices' actions in real time based on multi-source state data, and implement graded early warning measures based on the health assessment results; and a visualization and backtracking module to display the stage situation and risks based on the collision risk and health assessment results, and support backtracking simulation and analysis report generation.

[0021] Beneficial effects

[0022] The present invention has the following beneficial effects:

[0023] (1) This invention constructs a high-precision multi-source state database by integrating multiple types of sensors such as infrared rangefinders, encoders, IMUs, and cameras, and combines time-series modeling to achieve accurate prediction of the future motion state of equipment. It can identify and quantify the potential path overlap and collision risks between complex stage equipment in advance, effectively reducing the risk rate of sudden events.

[0024] (2) This invention, by calculating the obstacle avoidance priority value, combined with the minimum spatial distance, speed, target distance and remaining task time between devices, quantifies the importance and flexibility of devices, dynamically allocates conflict device avoidance strategies, effectively ensures priority passage of key devices, and improves the overall execution rhythm and scheduling efficiency.

[0025] (3) This invention constructs a motion health assessment value that integrates posture error and motion delay, and sets up a graded early warning response mechanism, so that the system can automatically perform self-check, self-adjustment, task switching and emergency stop operations according to changes in health status, which significantly improves the adaptive operation capability and safety assurance capability in scenarios of equipment status fluctuation or minor faults.

[0026] (4) This invention integrates functions such as 3D rendering, risk status highlighting, log indexing, and playback simulation to support collision reproduction and anomaly tracing for any historical time period, and automatically generates reports and fault analysis reports, providing strong support for stage safety management, equipment maintenance and system upgrades, and greatly improving operation and maintenance efficiency and data value.

[0027] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0028] Figure 1 This is a flowchart of a control method for a lifting stage structure;

[0029] Figure 2 This is a structural diagram of a lifting stage control system;

[0030] Figure 3 A bar chart showing the predicted collision risk values ​​for stage equipment pairs;

[0031] Figure 4 A simplified rendering of the movement path and collision avoidance of stage equipment. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. As those skilled in the art will understand, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] Please see Figures 1-3This invention provides a technical solution: a lifting stage structure control system and method, comprising the following steps: S1, real-time acquisition of multi-source state data of stage equipment through multiple types of sensors, data preprocessing of the multi-source state data, and real-time storage of multi-source state data in a multi-source state database; S2, acquisition of multi-source state data, construction of a motion state space model, prediction of future temporal motion sequences, determination of whether stage equipment overlaps based on the predicted temporal motion sequences, prediction of collision risk for each stage equipment, and implementation of avoidance measures based on the collision risk prediction results; S3, determination of obstacle avoidance priority of stage equipment based on collision risk prediction results, generation of stage equipment motion schemes based on obstacle avoidance priority, and issuance of control commands; S4, receipt of control commands and execution of motion operations, real-time evaluation of the health of stage equipment actions based on multi-source state data, and implementation of graded early warning measures based on the health evaluation results; S5, display of stage status and risks based on collision risk and health evaluation results, and support for traceable simulation and generation of analysis reports.

[0034] Specifically, the process of collecting multi-source status data of stage equipment in real time through multiple types of sensors, preprocessing the multi-source status data, and constructing a multi-source status database for real-time storage is as follows: Multi-source status data of each stage device is collected in real time using various sensors. The collection process includes: deploying infrared rangefinders at key parts of each stage device to collect spatial position information; using encoders to collect real-time displacement and velocity of each stage device; and using IMUs to collect three-axis acceleration, angular velocity, and attitude angle data. The multi-source status data includes spatial position information, displacement, velocity, acceleration, angular velocity, and attitude angle data. This not only enables spatial perception of static structures but also captures minute attitude changes during high-speed motion, ensuring high-precision reproduction and real-time modeling of dynamic stage scenes. Simultaneously, global visual images are acquired in real time through cameras on the top and sides of the rising stage. The type and maximum external diameter of each stage device are recorded and measured. The image information and device geometric data together construct a complete description of the device in three-dimensional space, which is used to support subsequent collision detection and spatial dynamic simulation. The clock offset of multi-source state data from different sensors is corrected, and unified timestamp alignment is performed. Kalman filtering is used to remove measurement noise and glitches. The sliding window method is used to perform packet loss interpolation and abnormal multi-source state data detection, which significantly improves the temporal consistency and anti-interference ability of the data. A unique ID and attribute description are assigned to each sensor, the data structure format of multi-source state data is unified, and the multi-source state data is packaged according to the interface protocol and standardized and normalized. A multi-source state database is built to store multi-source state data in real time according to sensor ID and time order index. At the same time, environmental change events are recorded synchronously, and the target point information of each stage device is recorded in real time.

[0035] In this implementation scheme, a unified standard multi-source state database is constructed by integrating multiple types of sensors such as infrared ranging, encoders, IMUs, and visual images, thereby improving the accuracy of spatial perception and dynamic modeling of stage equipment. Kalman filtering and sliding window compensation processing enhance data robustness and timing consistency. Standardized and normalized operations ensure efficient system parsing and compatibility, providing stable and high-quality support for subsequent path prediction, collision detection, and closed-loop control.

[0036] Specifically, the process of acquiring multi-source state data, constructing a motion state space model, and predicting future temporal motion sequences to determine whether stage equipment overlaps is as follows: First, multi-source state data is acquired to form a temporal motion sequence, including spatial position information, velocity, attitude angle data, and a timestamp. Second, a historical temporal feature matrix is ​​constructed based on historical step lengths and the temporal motion sequence. This matrix is ​​then input and trained using a Long Short-Term Memory (LSTM) network to construct the motion state space model. Finally, the predicted temporal motion sequence for future moments is output, including predicted spatial position information, predicted velocity, predicted attitude angle data, and a timestamp. This achieves short-term accuracy prediction of complex dynamic behaviors of various motion units, improving the ability to perceive potential risks in advance. The output predicted temporal motion sequence is segmented according to time slices to form a dynamic path set. The OBB bounding box algorithm is used to determine whether stage equipment spatially overlaps at any given time, outputting the pairs of spatially overlapping equipment and their time slice indices. Combining the predicted paths with geometric space modeling, a structured identification of future collision situations is achieved, significantly improving the real-time performance and accuracy of collision detection.

[0037] In this implementation scheme, by constructing a motion state space model, combining it with a long short-term memory network to predict future motion sequences, and using the OBB bounding box algorithm to achieve high-precision spatial overlap determination, not only is the prediction accuracy of dynamic collision risks of complex stage equipment improved, but the system's ability to identify and handle potential conflicts in advance is also significantly enhanced.

[0038] Specifically, the process of predicting collision risks for each stage device and taking avoidance measures based on the collision risk prediction results is as follows: First, obtain the predicted temporal motion sequence output by the motion state space model. Second, calculate the Euclidean distance between two devices using the predicted spatial position information to obtain the spatial center point distance between device i and device j at time t. Third, obtain the maximum outer diameter of the stage devices and calculate the average value to obtain the safe distance threshold between the two devices, which can dynamically adapt to changes in the actual configuration between devices and enhance adaptability to different device scales. Fourth, obtain the pairs of devices that spatially overlap and their time slice indices. Fifth, multiply the number of time slices of spatial overlap between the two devices by the duration of the time slices to obtain the total overlap duration of device i and device j, quantifying the persistence of potential risks from a time dimension, which helps to determine the urgency of the conflict event. Fifth, obtain the predicted spatial position information and predicted velocity of device i and device j. Sixth, calculate the velocity vector through position difference, and then... The angle between the velocity vectors of the two devices at time t is obtained by cosine calculation of the velocity vectors. This angle reflects the similarity of the motion directions of the two devices. If the angle is close, the risk of conflict is higher. The distance influence value is obtained by subtracting the ratio of the distance between the spatial center points of device i and device j at future time t to the safe distance threshold using a constant. The distance influence value is kept positive. If the distance influence value is less than 0, it is set to 0; otherwise, the original distance influence value is maintained. The distance influence value is multiplied by the distance safety margin weighting factor to obtain the distance safety margin term. The overlap time influence term is obtained by multiplying the overlap time weighting factor by the ratio of the total overlap time of device i and device j to the prediction time window length, which reflects the strength of the risk persistence. The directional risk term is obtained by subtracting the angle between the velocity vectors of the two devices at time t from a constant and multiplying it by the motion direction weighting factor. The collision risk prediction value is obtained by adding the distance safety margin term, the overlap time influence term, and the directional risk term.

[0039] The specific formula for the collision risk prediction value is as follows:

[0040] ;

[0041] In the formula, This represents the predicted collision risk between device i and device j at time t in the future. It quantitatively analyzes the potential collision risk of any two stage devices in a lifting stage at a certain time in the future, providing a scientific basis for collision prediction, graded early warning and subsequent intelligent avoidance decision-making. This represents the distance between the spatial center points of device i and device j at a future time t, reflecting the actual straight-line distance between the two stage devices in three-dimensional space. This represents the safe distance threshold between two devices, used to measure the actual distance from the minimum interval required for a collision to occur. This means that only positive numbers are retained. If the value inside the parentheses is less than 0, it is taken as 0; otherwise, the original value is retained. This is to ensure that the collision risk is not negative and to prevent the unreasonable phenomenon of negative risk when the distance between devices is large. Indicates the length of the prediction time window, which is the total duration of multi-step prediction; This represents the total overlap time between device i and device j, which measures the total time during which the two devices have a potential collision within the prediction time window. The longer the overlap time, the greater the risk. Let represent the angle between the velocity vectors of the two devices at time t, and The value range is [0, 2]. It takes the maximum value when the moving objects are moving in opposite directions (i.e., the included angle is 180°) and is 0 when they are moving in the same direction. The more the moving objects are moving in opposite directions or intersecting, the higher the risk of collision. The risk is lowest when the moving objects are moving in the same direction. The distance safety margin weighting factor reflects the impact of the proximity of equipment space on the overall collision risk. Based on the statistical characteristics of different equipment types, speeds, and volumes, as well as the distance data of the actual collision center point, the probability of collision events is fitted using historical risk factor data through a logistic regression algorithm. The regression coefficient is the distance safety margin weighting factor, with a value range between 0.2 and 0.6. The overlap time weighting factor measures the impact of the duration of spatial overlap on risk. It is obtained by quantitatively analyzing the correlation coefficient based on the correlation between the total historical overlap duration and the occurrence of actual risk events, and the value ranges from 0.2 to 0.5. The motion direction weight factor represents the weight of the angle between two devices on the collision risk. Based on the frequency of collision events under different velocity vector angles in historical data, as well as the actual safe distance and interference probability between devices when moving at different angles, the optimal motion direction weight factor is obtained by training with the angle and risk event labels through logistic regression algorithm. The value range is between 0.1 and 0.4.

[0042] The distance safety margin weighting factor was set to 0.4, the overlap time weighting factor to 0.3, and the motion direction weighting factor to 0.3. The safety distance threshold between the two devices was set to 2, and the prediction time window length was set to 10. The collision risk prediction value was calculated based on the spatial center point distance between different device pairs, the total overlap time, and the velocity vector angle. The collision risk prediction value data is shown in Table 1.

[0043] Table 1 Collision Risk Prediction Data Table

[0044]

[0045] like Figure 3The image shows a bar chart illustrating the predicted collision risk values ​​for stage equipment pairs provided in this embodiment of the application. It displays the predicted collision risk values ​​for different equipment pairs, reflecting the likelihood of them colliding in future paths. (Based on Table 1 and...) Figure 3 It can be seen that, when the weighting factor, safety distance threshold, and prediction time window length are the same, different devices have different predicted collision risks.

[0046] This implementation plan integrates three dimensions of features: spatial distance, overlap duration, and direction of movement, to construct a quantitative collision risk prediction value. This not only enhances the ability to detect potential conflicts among multiple devices in advance, but also supports graded response and dynamic avoidance, significantly improving the operational safety and intelligence of the lifting stage in complex collaborative environments.

[0047] Specifically, the process of taking avoidance measures based on collision risk prediction results is as follows: Real-time comparison of the predicted collision risk value with the primary and secondary risk thresholds. When the predicted collision risk value is less than the primary risk threshold, it is determined to be a low collision risk, and the equipment runs along its original path. The predicted collision risk value is recorded in the background log; this effectively reduces system resource consumption and maintains scheduling efficiency. When the predicted collision risk value is greater than or equal to the primary risk threshold but less than the secondary risk threshold, it is determined to be a moderate collision risk. The equipment speed is automatically reduced, and the stage equipment trajectory is fine-tuned for dynamic avoidance, adhering to the principles of non-interruption and low intervention to maximize operational continuity. A local path fine-tuning algorithm based on spline path optimization is used to allow equipment with moderate collision risk to make slight detours, automatically avoiding predicted high-risk areas, achieving smooth path changes, and avoiding inertial interference caused by violent turning. If multiple moderate collision risk devices are predicted to meet in the same small area, they are coordinated to pass through each other; this achieves flexible multi-device collaboration and improves space utilization. When the predicted collision risk value is greater than... When the risk level equals or equals the secondary risk threshold, it is determined to be a high collision risk. An early warning is immediately issued, and the current operation of the high-risk equipment is suspended. Temporary waiting is set for the relevant equipment; the path is dynamically adjusted to temporarily avoid the conflict area and postpone the planned action sequence; the global task queue is intelligently adjusted, and the priority of non-critical tasks is temporarily reduced to decouple risk suppression and task postponement, ensuring the safety of the main line operation; if the high-risk state is not resolved, the stage equipment will automatically roll back to the previous safe position; and after a high-risk action is triggered, the relevant stage equipment will perform a rapid health self-check to help determine whether the risk is caused by abnormal equipment status, thus forming a closed-loop linkage mechanism of risk identification, action control, and health verification.

[0048] This implementation plan utilizes a tiered response mechanism to dynamically manage collision risks, enabling real-time perception and intelligent intervention of stage equipment operating status, and possessing excellent adaptability and safety assurance capabilities. Its core advantages lie in: ensuring task continuity under low-risk conditions, achieving smooth path fine-tuning and collaborative obstacle avoidance during medium-risk phases, and rapidly triggering pauses and global scheduling reconstruction during high-risk phases. Furthermore, it incorporates a closed-loop feedback system based on health status self-checks, effectively enhancing operational stability and accident prevention capabilities in complex stage environments.

[0049] Specifically, the process of determining the obstacle avoidance priority of stage equipment based on collision risk prediction results, generating motion plans for stage equipment based on obstacle avoidance priority, and issuing control commands is as follows: First, acquire multi-source state data. Based on the spatial position information of equipment i, for equipment i, iterate through all other equipment, calculate the Euclidean distance between two equipment, and select the minimum value to obtain the minimum spatial distance between equipment i and other equipment. This reflects the degree of spatial pressure on the equipment in the current environment and is an important indicator of potential collision risk. Second, acquire the velocity of equipment i, calculate the velocity components in three-dimensional space using the finite difference method, and then calculate the magnitude to obtain the velocity magnitude of equipment i, reflecting the instantaneous operating state of the equipment. Third, based on the spatial position information of equipment i and the displacement of the stage equipment to the target point, calculate the straight-line distance from equipment i to the target point using coordinates. Fourth, divide the straight-line distance from equipment i to the target point by the velocity magnitude of equipment i to obtain the remaining completion time of the current task of equipment i, assisting in dynamic priority adjustment. Fifth, divide the minimum spatial distance between equipment i and other equipment by the product of the velocity magnitude of equipment i and the straight-line distance from equipment i to the target point to obtain the spatial velocity priority factor. The spatial velocity priority factor varies depending on the equipment's position. When there is a conflict between tasks, the relative priority of passage is reflected. The natural logarithm of the sum of the remaining completion time of the current task of device i and a constant is taken to obtain the remaining time elasticity factor. The natural logarithm processing can suppress extreme values ​​and improve the robustness of system scheduling. The spatial speed priority factor is added to the remaining time elasticity factor to obtain the obstacle avoidance priority value of device i. All devices with medium and high collision risks are sorted according to their obstacle avoidance priority values. Devices with higher obstacle avoidance priority values ​​are given priority to maintain their original path and normal speed. Devices with lower obstacle avoidance priority values ​​are dynamically decelerated and automatically adjusted to shift slightly towards a safe area to dynamically control traffic bottlenecks and prevent the conflict from escalating. If there is a conflict with a high-priority device, temporary waiting is performed. At the same time, the obstacle avoidance scheme generated based on the obstacle avoidance priority value is combined with the global scheduling goal, and the details of the scheme are optimized using dynamic window algorithm and greedy algorithm to output the final device motion scheme, including: motion path, speed and action plan. According to the final device motion scheme, motion control commands for each device are automatically issued, including speed, path and action delay. All adjustment processes, collision risk prediction values, obstacle avoidance priority values ​​and actual control actions are recorded throughout the process.

[0050] The specific formula for the obstacle avoidance priority value is as follows:

[0051] ;

[0052] In the formula, This represents the obstacle avoidance priority value of device i. When there is a risk of collision, multiple stage devices are sorted to automatically determine which has priority to go straight and which has priority to yield, providing a scientific and data-driven basis for generating safe and efficient obstacle avoidance decisions. This represents the minimum spatial distance between device i and other devices, reflecting the actual physical distance between device i and the nearest other device. It is a core parameter for measuring collision risk and space margin. The smaller the distance, the easier it is for a collision to occur. The speed modulus of device i reflects how fast device i is moving at this moment. The faster the speed, the more difficult it is to adjust and avoid obstacles. It should prioritize going straight and avoid frequent braking. This represents the straight-line distance from device i to the target point, and the shortest distance that device i still needs to travel. The closer to the target, the more priority should be given to passing through it to avoid being forcibly interrupted in the final stage. This represents the remaining completion time of the current task for device i, reflecting the urgency and timing pressure of the task. The longer the remaining time, the more likely it is to yield or wait; the shorter the remaining time, the more likely it is to be prioritized. Taking the logarithm is to reduce the impact of extreme values ​​on priority and make the priority increase more gradual.

[0053] In this implementation plan, the minimum spatial distance between devices, operating speed, and task urgency are organically integrated through quantifiable obstacle avoidance priority value calculation. This not only improves the rationality and real-time performance of dynamic obstacle avoidance decisions but also enables precise sorting and intelligent scheduling in multi-device collaborative operation scenarios. The path optimization strategy, combined with dynamic windows and greedy algorithms, makes the motion scheme more efficient and flexible. At the same time, the scheduling process and key parameters are recorded throughout, providing a data foundation for subsequent behavior backtracking, adaptive learning, and risk warning of the system, significantly improving the security, intelligence, and operational traceability of the entire platform.

[0054] Specifically, the process of receiving control commands and executing motion operations, and evaluating the health of stage equipment's actions in real time based on multi-source state data, is as follows: The system receives motion control commands from the intelligent avoidance decision module, parses the target motion of each stage device, and sends the parsed commands to each stage device, automatically starting the device to execute the motion operation; it acquires the actual attitude angle data of device i when it reaches the target point, and simultaneously acquires the predicted attitude angle data of device i output by the motion state space model; it subtracts the predicted attitude angle data from the actual attitude angle data and takes the absolute value to obtain the absolute attitude error; it acquires multiple sets of historical absolute attitude errors, calculates the average value based on the sliding time window length as the attitude error tolerance value, thereby dynamically adapting to the operation of the equipment. System errors; obtain the instruction timestamp recorded when the instruction is issued and the completion timestamp recorded when device i actually reaches the target point, and subtract the instruction timestamp from the completion timestamp to obtain the actual completion delay of the action; obtain multiple sets of historical actual completion delays of actions, calculate the average based on the sliding time window length as the maximum action delay duration, and enhance the system's adaptability to the response characteristics of different devices; subtract the ratio of absolute attitude error to attitude error tolerance value from a constant 1, and then subtract the ratio of actual completion delay of action to maximum action delay duration to obtain the action health assessment value of device i; subtract the error ratio from a constant 1 to construct a health index with a higher score for smaller errors, which facilitates system status assessment, graded early warning, and closed-loop control.

[0055] The specific formula for the movement health assessment value is as follows:

[0056] ;

[0057] In the formula, This represents the motion health assessment value of device i, used to quantify the health and reliability of a single motion execution. It evaluates the actual execution status of the device in the motion control and execution module, so as to automatically determine whether an early warning or emergency response is needed. This represents the actual attitude angle data of device i upon reaching the target point, reflecting the true execution result and execution accuracy. This represents the predicted attitude angle data of device i upon reaching the target point, reflecting the expected execution result; Indicates absolute attitude error; This indicates the attitude error tolerance value, reflecting the acceptable error range; This indicates the actual delay in completing an action, i.e., the deviation in actual execution time, reflecting the impact of execution efficiency and dynamic environment; This indicates the maximum action delay duration and the acceptable delay duration, reflecting the task timing requirements and safety tolerance.

[0058] This implementation plan integrates two key indicators—posture error and motion delay—to construct a unified motion health assessment value, enabling quantitative judgment of the equipment's motion execution status. By adaptively adjusting tolerance parameters using historical data, the accuracy of the assessment is improved. The use of a 1-minus-error ratio method enhances sensitivity to deviations, facilitating tiered early warning and precise control. Simultaneously, the entire process is data-driven and requires no human intervention, ensuring automation, real-time performance, and high robustness of the assessment process, effectively guaranteeing the reliability and safety of stage equipment movements.

[0059] Specifically, the process of implementing graded early warning measures based on health assessment results is as follows: Real-time comparison of the action health assessment value with the first-level and second-level health thresholds. When the action health assessment value is greater than or equal to the second-level health threshold, the action is deemed healthy and continues to be executed, with continuous dynamic relaxation monitoring and periodic small-amplitude calibration movements to verify sensor and control accuracy, ensuring stable system operation amidst small fluctuations. When the action health assessment value is greater than or equal to the first-level health threshold but less than the second-level health threshold, an early warning state is established. Using log records and multi-source state data, a lightweight diagnostic algorithm is automatically run to determine if there is sensor drift, dynamic response lag, environmental abrupt changes, or control parameter mismatch, and generates targeted fine-tuning solutions: If sensor error occurs, the filter parameter fine-tuning or recalibration process is automatically triggered; if dynamic lag is present, the acceleration limit is temporarily reduced or the velocity curve is changed; if environmental abrupt changes occur, the local path is automatically replanned or the attitude priority is adjusted; if control parameter problems exist, the predictive control weights of the motion state space model are automatically adjusted; if the equipment's action health assessment value remains within the early warning range and shows a downward trend, the system automatically... The system automatically splits some tasks and assigns them to other devices with high motion health assessment values, forming a flexible and redundant scheduling mechanism to ensure the continuity of stage tasks. It dynamically adjusts the attitude error tolerance and maximum motion delay time based on historical actual motion execution and environmental characteristics. When the motion health assessment value is less than the first-level health threshold, it is judged as a serious motion execution anomaly, initiating an emergency stop operation and immediately isolating the abnormal device's motion while reassigning tasks, thereby preventing the anomaly from spreading and causing systemic risks. After the emergency stop, a rapid self-check sub-process is automatically executed, and combined with historical logs, it automatically locates possible fault points, generates a preliminary diagnostic report, and improves maintenance efficiency. It automatically summarizes all multi-source status data, motion health assessment values, control commands, and environmental information before and after the device anomaly, constructing a detailed event package. It transmits the device's motion health assessment values ​​and execution status back in real time, and presents the motion health assessment values, warnings, and the effects of implemented automation measures for each device in real time. Simultaneously, it automatically generates reports periodically, analyzing motion health assessment distribution, failure rate, self-healing success rate, and collaborative obstacle avoidance efficiency. If multiple devices of the same model are running in parallel, feedback data sharing helps to form common patterns and improve group robustness.

[0060] This implementation plan achieves accurate identification and intelligent control of stage equipment execution status through real-time monitoring and tiered response of motion health assessment values. Combined with dynamic threshold adjustment, lightweight fault diagnosis, task redundancy transfer, and self-healing mechanisms, it not only improves the reliability and safety of equipment operation but also enhances the system's adaptability to anomalies such as sensor drift and control lag. Simultaneously, through fault event data backtracking and report generation, it supports system optimization and strategy iteration, significantly improving overall operation and maintenance efficiency and the intelligence level of the stage control system.

[0061] Specifically, the process of displaying stage status and risks based on collision risk and health assessment results, and supporting traceable simulation and analysis report generation, is as follows: A 3D engine renders the entire platform structure, stage equipment, obstacles, and movement paths in real time, forming a three-dimensional and intuitive dynamic situational awareness interface, highlighting risk areas and avoidance actions; if a high-risk collision or abnormal equipment action is detected, pop-up windows and audible / visual alarms are triggered, along with a pause prompt, enabling early and rapid warnings to ensure equipment and personnel safety; manual commands are supported for maintenance personnel to issue, enabling human-machine collaborative intervention, and all manual operations and responses are recorded for subsequent analysis and accountability; every command, risk event, and equipment status log is automatically recorded according to time and event indexes, supporting multi-condition retrieval to improve the efficiency of event queries and problem review; motion simulation playback, collision reproduction, and anomaly cause analysis can be performed on any historical period, helping technicians to deeply understand the fault evolution mechanism; and statistical reports and event analysis reports are automatically generated, comprehensively supporting system operation, optimization, and upgrades.

[0062] like Figure 4 The image shown is a simplified rendering of the stage equipment's motion path and collision avoidance according to an embodiment of this application. It illustrates the prediction of the stage equipment's motion path, identification of potential collision areas, and dynamic avoidance behavior during a performance. The predicted movement trajectories of each piece of equipment are marked with dashed lines of different colors, and high-risk collision areas are highlighted with red ellipses. In the image, the scenery vehicle C triggers an avoidance strategy, indicated by a blue arrow, effectively avoiding the conflict area. The visualization intuitively presents the results of integrating path prediction, collision scoring, and obstacle avoidance priority decisions.

[0063] This implementation plan significantly improves the lifting stage system's ability to perceive collision risks and equipment anomalies, as well as its response speed and diagnostic accuracy, by integrating 3D visualization rendering, risk highlighting, manual intervention support, full-process log recording and retrospective analysis. At the same time, it automatically generates statistical reports and evolution playback, enhancing the system's interpretability and maintainability, and helping to achieve intelligent operation and maintenance and continuous optimization and upgrading of stage control.

[0064] Reference Figure 2As shown, the second aspect of the present invention provides a lifting stage structure control system, applied to the aforementioned lifting stage structure control method, comprising: a perception and environment modeling module, a motion prediction and collision detection module, an intelligent avoidance decision-making module, a motion control and execution module, and a visualization and backtracking module; wherein the perception and environment modeling module is used to collect multi-source state data of the stage equipment in real time through multiple types of sensors, preprocess the multi-source state data, and construct a multi-source state database to store the multi-source state data in real time; the motion prediction and collision detection module is used to acquire multi-source state data, construct a motion state space model, predict the temporal motion sequence of future moments, and use the predicted temporal sequence... The system comprises several modules: a motion sequence module to determine if stage equipment overlaps and predict the collision risk of each piece of equipment, and to take avoidance measures based on the collision risk prediction results; an intelligent avoidance decision module to determine the obstacle avoidance priority of stage equipment based on the collision risk prediction results, generate motion plans for the stage equipment based on the obstacle avoidance priority, and issue control commands; a motion control and execution module to receive control commands and execute motion operations, to evaluate the health of the stage equipment's actions in real time based on multi-source state data, and to implement graded early warning measures based on the health assessment results; and a visualization and backtracking module to display the stage situation and risks based on the collision risk and health assessment results, and to support backtracking simulation and generate analysis reports.

[0065] In this implementation plan, by constructing an integrated lifting stage structure control system that combines perception, prediction, decision-making, execution, and visualization, the system achieves accurate acquisition of multi-source states, early prediction of dynamic collision risks, intelligent generation of differentiated obstacle avoidance strategies, real-time health assessment of action execution status, and full-process visualization and retrospective analysis, significantly improving the safety, collaboration, and intelligent operation and maintenance of stage equipment.

[0066] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0067] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. As those skilled in the art will understand, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for controlling a lifting stage structure, characterized in that, Includes the following steps: S1: Collect multi-source status data of stage equipment in real time through multiple types of sensors, preprocess the multi-source status data, and build a multi-source status database to store the multi-source status data in real time. S2, acquire multi-source state data, construct a motion state space model, predict the temporal motion sequence of future moments, determine whether stage equipment overlaps based on the predicted temporal motion sequence, predict the collision risk of each stage equipment, and take avoidance measures based on the collision risk prediction results. The specific process of predicting the collision risk of each stage device and taking avoidance measures based on the collision risk prediction results is as follows: Obtain the predicted temporal motion sequence output by the motion state space model, calculate the Euclidean distance between two devices using the predicted spatial location information to obtain the spatial center point distance between device i and device j at time t in the future; obtain the maximum outer diameter of the stage equipment and calculate the average value to obtain the safe distance threshold between the two devices; obtain the device pair and time slice index that have spatial overlap, and obtain the total overlap duration of device i and device j by multiplying the number of time slices that have spatial overlap between the two devices by the duration of the time slices; Obtain the predicted spatial position information and predicted velocity of device i and device j, calculate the velocity vector by the difference of position, and then calculate the angle between the velocity vectors of the two devices at time t by cosine calculation of the velocity vector; The distance influence value is obtained by subtracting the ratio of the distance between the spatial center points of device i and device j at future time t to the safe distance threshold from a constant. The distance influence value is then kept positive; if it is less than 0, it is set to 0; otherwise, the original distance influence value is maintained. This distance influence value is multiplied by the distance safety margin weighting factor to obtain the distance safety margin term. The overlap time influence term is obtained by multiplying the overlap time weighting factor by the ratio of the total overlap time of device i and device j to the prediction time window length. The directional risk term is obtained by subtracting the angle between the velocity vectors of the two devices at time t from a constant and then multiplying it by the motion direction weighting factor. The collision risk prediction value is obtained by adding the distance safety margin term, the overlap time influence term, and the directional risk term. S3 determines the obstacle avoidance priority of stage equipment based on the collision risk prediction results, generates a motion plan for stage equipment based on the obstacle avoidance priority, and issues control commands. S4 receives control commands and executes motion operations. It evaluates the health of the stage equipment's actions in real time based on multi-source status data and implements graded early warning measures based on the health assessment results. S5 displays the stage situation and risks based on collision risk and health assessment results, and supports traceable simulation and generation of analysis reports.

2. The method for controlling a lifting stage structure according to claim 1, characterized in that, The specific process of acquiring multi-source status data of stage equipment in real time through multiple types of sensors, preprocessing the multi-source status data, and constructing a multi-source status database to store the multi-source status data in real time is as follows: Multiple sensors are used to collect multi-source status data of various stage equipment in real time. The collection process is as follows: infrared rangefinders are deployed at key parts of each stage equipment to collect spatial position information; encoders are used to collect the real-time displacement and speed of each stage equipment. The system utilizes an IMU to collect three-axis acceleration, angular velocity, and attitude angle data; multi-source state data includes: spatial position information, displacement, velocity, acceleration, angular velocity, and attitude angle data; simultaneously, it acquires global visual images in real time through cameras on the top and sides of the lifting stage, and records and measures the type of each stage device and its maximum external diameter. Correct the clock offset of multi-source state data from different sensors, perform unified timestamp alignment, use Kalman filtering to remove measurement noise and glitches, and use the sliding window method to perform packet loss interpolation and detect abnormal multi-source state data. Assign a unique ID and attribute description to each sensor, unify the multi-source state data structure format, package the data according to the interface protocol, and standardize and normalize the multi-source state data. A multi-source status database is constructed, and multi-source status data is stored in real time according to sensor ID and time sequence index; at the same time, environmental change events are recorded synchronously, and target point information of each stage equipment is recorded in real time.

3. The method for controlling a lifting stage structure according to claim 1, characterized in that, The specific process of acquiring multi-source state data, constructing a motion state space model, predicting the temporal motion sequence at future times, and determining whether stage equipment overlaps based on the predicted temporal motion sequence is as follows: Multi-source state data is acquired to form a temporal motion sequence, which includes spatial location information, velocity, attitude angle data, and timestamps. A historical temporal feature matrix is ​​constructed based on historical step length and temporal motion sequence. The historical temporal feature matrix is ​​input and trained through a long short-term memory network to construct a motion state space model. Finally, the predicted temporal motion sequence for future moments is output, including predicted spatial location information, predicted velocity, predicted attitude angle data, and timestamps. The output predicted temporal motion sequence is segmented according to time slices to form a dynamic path set; the OBB bounding box algorithm is used to determine whether each stage device spatially overlaps at any time, and the device pairs and time slice indices that spatially overlap are output.

4. The method for controlling a lifting stage structure according to claim 1, characterized in that, The specific process for taking avoidance measures based on collision risk prediction results is as follows: The collision risk prediction value is compared with the first-level risk threshold and the second-level risk threshold in real time. When the collision risk prediction value is less than the first-level risk threshold, it is determined to be a low collision risk, and the equipment runs along the original path. The collision risk prediction value is recorded in the background log. When the predicted collision risk value is greater than or equal to the Level 1 risk threshold and less than the Level 2 risk threshold, it is judged as a moderate collision risk, and the equipment speed is automatically reduced and the stage equipment trajectory is finely adjusted for dynamic avoidance. By using a local path fine-tuning algorithm optimized by spline paths, devices with moderate collision risk can make slight detours and automatically avoid predicted high-risk areas; if multiple devices with moderate collision risk are predicted to meet in the same small area, they can coordinate to pass through each other in an alternating manner. When the predicted collision risk value is greater than or equal to the level 2 risk threshold, it is determined to be a high collision risk, an early warning is immediately issued, the current operation of the high-risk equipment is suspended, and a temporary waiting state is set for the relevant equipment. The system dynamically adjusts paths to temporarily avoid conflict areas and postpones planned actions; it intelligently adjusts the global task queue and temporarily lowers the priority of non-critical tasks; if the high-risk status persists, the stage equipment will automatically roll back to the previous safe position; and after a high-risk action is triggered, the relevant stage equipment will perform a rapid health self-check.

5. The method for controlling a lifting stage structure according to claim 1, characterized in that, The specific process of determining the obstacle avoidance priority of stage equipment based on the collision risk prediction results, generating a stage equipment motion plan based on the obstacle avoidance priority, and issuing control commands is as follows: Acquire multi-source state data. Based on the spatial location information of device i, for device i, iterate through all other devices, calculate the Euclidean distance between any two devices, and filter out the minimum value to obtain the minimum spatial distance between device i and other devices. Obtain the velocity of device i, calculate the velocity components in three-dimensional space using the finite difference method, and then calculate the velocity modulus of device i. Based on the spatial location information of device i and the displacement of the stage equipment from the target point, calculate the distance through coordinates to obtain the straight-line distance from device i to the target point. Divide the straight-line distance from device i to the target point by the velocity modulus of device i to obtain the remaining completion time of the current task of device i. The spatial velocity priority factor is obtained by dividing the minimum spatial distance between device i and other devices by the product of the velocity modulus of device i and the straight-line distance from device i to the target point; the remaining time elasticity factor is obtained by taking the natural logarithm of the sum of the remaining completion time of the current task of device i and a constant; the obstacle avoidance priority value of device i is obtained by adding the spatial velocity priority factor and the remaining time elasticity factor. All devices with medium and high collision risks are sorted by obstacle avoidance priority. Devices with higher obstacle avoidance priority maintain their original path and pass through at normal speed; devices with lower obstacle avoidance priority dynamically decelerate and automatically adjust their current trajectory, shifting slightly towards a safe area; if they conflict with higher priority devices, they temporarily wait. Meanwhile, for the obstacle avoidance scheme generated based on the obstacle avoidance priority value, the scheme details are optimized by combining the global scheduling target with the dynamic window algorithm and the greedy algorithm, and the final equipment motion scheme is output, including: motion path, speed and action plan; Based on the final equipment motion plan, motion control commands for each piece of equipment are automatically issued, including speed, path, and action delay; and all adjustment processes, collision risk prediction values, obstacle avoidance priority values, and actual control actions are recorded throughout the entire process.

6. The method for controlling a lifting stage structure according to claim 1, characterized in that, The specific process of receiving control commands and executing motion operations, and evaluating the health of the stage equipment's actions in real time based on multi-source state data, is as follows: It receives motion control commands from the intelligent avoidance decision module, analyzes the target motion of each stage device, and sends the analyzed commands to each stage device, automatically starting the device to execute motion operations; Acquire the actual attitude angle data of device i when it reaches the target point, and simultaneously acquire the predicted attitude angle data of device i output by the motion state space model. Subtract the predicted attitude angle data from the actual attitude angle data and take the absolute value to obtain the absolute attitude error. Acquire multiple sets of historical absolute attitude errors, and calculate the average value based on the sliding time window length as the attitude error tolerance value. Acquire the command timestamp recorded when the command is issued and the completion timestamp recorded when device i actually reaches the target point. Subtract the command timestamp from the completion timestamp to obtain the actual action completion delay. Acquire multiple sets of historical action completion delays, and calculate the average value based on the sliding time window length as the maximum action delay duration. Subtract the ratio of the absolute attitude error to the attitude error tolerance value from a constant, and then subtract the ratio of the actual action completion delay to the maximum action delay duration to obtain the action health assessment value of device i.

7. The method for controlling a lifting stage structure according to claim 1, characterized in that, The specific process of implementing tiered early warning measures based on health assessment results is as follows: The motion health assessment value is compared with the first-level health threshold and the second-level health threshold in real time. When the motion health assessment value is greater than or equal to the second-level health threshold, it is determined that the execution of the motion is in good health and continues to be executed. Continuous dynamic relaxation monitoring is performed, and small-amplitude calibration movements are performed periodically to verify the accuracy of the sensors and control. When the motion health assessment value is greater than or equal to the first-level health threshold and less than the second-level health threshold, it is determined to be in a warning state. Using log records and multi-source status data, a lightweight diagnostic algorithm is automatically run to determine whether there is sensor drift, dynamic response lag, environmental abrupt change, or control parameter mismatch, and generates a targeted fine-tuning plan. If the device's motion health assessment value remains within the warning range and shows a downward trend, some tasks will be automatically split and transferred to other devices with higher motion health assessment values; the attitude error tolerance value and maximum motion delay time will be dynamically adjusted based on historical actual motion execution and environmental characteristics. When the motion health assessment value is less than the first-level health threshold, it is determined to be a serious abnormality in motion execution, an emergency stop operation is initiated, the abnormal equipment motion is immediately isolated, and the task is reassigned; after the emergency stop, a rapid self-check sub-process is automatically executed, and the possible fault point is automatically located by combining historical logs, and a preliminary diagnostic report is generated; all multi-source status data, motion health assessment values, control commands, and environmental information before and after the equipment abnormality are automatically summarized to construct a detailed event package; It transmits the motion health assessment values ​​and execution status of the equipment in real time, and presents the motion health assessment values, warnings and the effects of the automated measures implemented for each device in real time; at the same time, it automatically generates reports on a regular basis to analyze the motion health assessment distribution, failure rate, self-healing success rate and collaborative obstacle avoidance efficiency; if multiple devices of the same model are running in parallel, it shares feedback data.

8. The method for controlling a lifting stage structure according to claim 1, characterized in that, The specific process of displaying the stage situation and risks based on collision risk and health assessment results, and supporting traceable simulation and generation of analysis reports, is as follows: The entire platform structure, stage equipment, obstacles, and movement paths are rendered in real time using a 3D engine, and risk areas and avoidance actions are highlighted. If a high-risk collision or abnormal equipment action is detected, pop-up windows and audio-visual alarms are triggered, and a pause prompt is given. It supports maintenance personnel in issuing manual commands and records all manual operations and responses; it automatically records every command, risk event, and equipment status log of the entire process by time and event index, and supports multi-condition retrieval; at the same time, it can perform motion simulation playback, collision reproduction, and anomaly cause analysis for any historical period, and automatically generate statistical reports and event analysis reports.

9. A lifting stage structure control system, employing the lifting stage structure control method as described in any one of claims 1-8, characterized in that, include: The module comprises a perception and environment modeling module, a motion prediction and collision detection module, an intelligent avoidance decision-making module, a motion control and execution module, and a visualization and backtracking module, characterized by: The perception and environment modeling module is used to collect multi-source state data of stage equipment in real time through multiple types of sensors, preprocess the multi-source state data, and build a multi-source state database to store the multi-source state data in real time. The motion prediction and collision detection module is used to acquire multi-source state data, construct a motion state space model, predict the temporal motion sequence of future moments, determine whether stage equipment overlaps based on the predicted temporal motion sequence, predict the collision risk of each stage equipment, and take avoidance measures based on the collision risk prediction results. The intelligent obstacle avoidance decision module is used to determine the obstacle avoidance priority of stage equipment based on the collision risk prediction results, generate a movement plan for the stage equipment based on the obstacle avoidance priority, and issue control commands. The motion control and execution module is used to receive control commands and execute motion operations, evaluate the health of the stage equipment's actions in real time based on multi-source status data, and implement graded early warning measures based on the health evaluation results. The visualization backtracking module is used to display the stage situation and risks based on collision risk and health assessment results, and supports backtracking simulation and generation of analysis reports.

Images