A method and system for controlling steel structures based on dynamic load distribution
By employing time-frequency decomposition and two-stage closed-loop calibration methods, combined with the coordinated control of temperature sensing units and piezoelectric units, the control accuracy problem of steel structures under dynamic loads was solved, achieving precise distribution of high-frequency loads and improved structural stability.
Patent Information
- Application Number
- CN202611250296.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-18
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies for controlling steel structures under dynamic loads have low control precision and cannot adapt to rapid changes in high-frequency loads, leading to amplified structural vibrations and fatigue damage. Furthermore, thermal inertia causes mismatch in the output of support forces, making it impossible to achieve precise distribution of high-frequency loads.
The load is split into high-frequency and low-frequency components using time-frequency decomposition technology. Temperature sensing units and piezoelectric units are used to undertake steady-state and transient support respectively. Combined with two-stage closed-loop calibration and advanced prediction compensation, precise support commands are generated. The structural resonant frequency is adjusted through frequency domain analysis to achieve precise load distribution across the entire frequency band.
It improves the control accuracy under dynamic loads, reduces structural vibration and fatigue damage, and enhances the operational stability and safety of steel structures.
Smart Images

Figure CN123113291A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of steel structure control technology, and in particular to a steel structure control method and system based on dynamic load distribution. Background Technology
[0002] Large steel structures, with their advantages of high strength, excellent plasticity and toughness, recyclability, and high degree of industrialization, have become the core structural form in major engineering fields such as long-span bridges, super high-rise buildings, offshore oil and gas platforms, heavy industrial plants, and large equipment support systems. Throughout their entire service life, steel structures inevitably endure repeated exposure to complex time-varying dynamic loads, including wind-induced vibrations, seismic forces, traffic dynamic loads, temperature gradient changes, and equipment vibrations. These dynamic loads can cause problems such as localized stress concentration, fatigue damage accumulation, and irreversible permanent deformation, not only reducing the installation accuracy and performance of the structure but also, in severe cases, leading to cracking, instability, or even collapse, resulting in significant casualties and property damage. Therefore, real-time monitoring and active control of the stress state of large steel structures under dynamic loads is a key technical requirement for ensuring engineering safety and extending the service life of the structure.
[0003] For example, Chinese invention patent CN120335288A discloses a method for controlling the installation of large steel structures based on dynamic load distribution, including the following steps: installing strain gauges at key nodes of the steel structure to monitor local deformation caused by dynamic loads in real time and obtain the resistance value of the strain gauges; presetting the optimal resistance value range of the strain gauges; installing support structures at key nodes of the steel structure to additionally bear the dynamic loads, the support structures being temperature-sensitive metal strips with heating elements on them; integrating multi-objective optimization algorithms and adaptive PID control, dynamically adjusting PID control parameters, and controlling the heating amount of the heating elements according to the adjusted PID control parameters, so that the resistance value of the strain gauges remains within the optimal resistance value range, thereby controlling the degree to which each temperature-sensitive metal strip bears the dynamic load.
[0004] However, the aforementioned technical solutions rely on electric heating to drive the thermal expansion of temperature-sensitive metal strips to adjust the support force. Heat conduction is a typical slow-response physical process; the temperature-sensitive metal strip requires a long response time from heating to generating the target support force, and the cooling process relying solely on natural heat dissipation takes even longer. This is completely unable to keep up with the rapid changes in high-frequency dynamic loads, resulting in a severe misalignment between the peak support force and the peak load, and even the generation of reverse support, exacerbating structural vibrations. Furthermore, the NSGA-II multi-objective optimization algorithm requires multiple iterative calculations to solve for the optimal PID control parameters. Each optimization process involves a significant computational delay, and the more nodes involved, the longer the computation time. This causes the control parameter update speed to be far lower than the frequency of high-frequency load changes, degenerating into fixed-parameter PID control in actual operation. This is unable to adapt to the nonlinear time-varying characteristics of steel structures and is prone to overshoot and oscillation problems. Furthermore, the delay superposition compresses the effective control bandwidth. High-frequency load components exceeding this bandwidth not only cannot be effectively controlled, but also generate a negative damping effect due to excessive phase difference, which amplifies structural vibration. The thermal inertia of the temperature-sensitive metal strip also leads to heating overshoot. Even after heating stops, the temperature continues to rise, causing excessive output of support force. When the load drops rapidly, it cannot cool and shrink in time, forming a continuous reverse support force. Long-term accumulation will cause additional fatigue damage and reduce the overall accuracy of dynamic load distribution.
[0005] In summary, there is an urgent need for steel structure control methods and systems based on dynamic load distribution that can continuously improve control accuracy. Summary of the Invention
[0006] This invention provides a steel structure control method and system based on dynamic load distribution, which can improve control accuracy.
[0007] To solve the above-mentioned technical problems, this application provides the following technical solution: A steel structure control system based on dynamic load distribution includes: a data acquisition module, a data processing module, an instruction generation module, and a load execution module; The data acquisition module is used to collect global perception data; The data processing module is used to perform initial state judgment based on the full-domain sensing data to obtain initial state data; to perform filtering and transcoding on the initial state data to obtain standard monitoring data; to perform time-frequency decomposition on the standard monitoring data to obtain high-frequency component data and low-frequency component data; to perform coefficient iteration on the standard monitoring data to obtain dynamic coupling data; to perform trend prediction and data repair on the standard monitoring data to obtain compensation and correction data; to perform frequency domain analysis on the standard monitoring data to obtain resonance tuning data; and to perform model optimization on the low-frequency component data and dynamic coupling data to obtain optimal PID parameter data. The instruction generation module is used to compile signals based on optimal PID parameter data to obtain low-frequency heating instructions; to calculate output based on optimal PID parameter data to obtain support-related instructions; to perform advance calculations based on high-frequency component data and compensation correction data to obtain high-frequency drive instructions; to compile frequency control based on resonance tuning data to obtain resonance adjustment instructions; and to compile welding instructions based on assembly requirements information to obtain solid welding operation instructions. The load execution module is used to drive the temperature sensing unit to work according to the low-frequency heating command, drive the piezoelectric unit to work according to the high-frequency driving command, and drive the motor jack to output hydraulic pressure according to the support-related command, so as to complete the load support operation in a coordinated manner, obtain execution feedback data, and perform initial fine-tuning based on the execution feedback data; complete the alignment and welding of the new steel structure according to the welding operation command, collect the global sensing data again, and perform secondary fine-tuning operation based on the re-collected global sensing data; and adjust the natural frequency point by using external force to stimulate the resonance of the structure according to the resonance adjustment command.
[0008] The basic principle and beneficial effects of the solution are as follows: Due to the inherent thermal inertia of a single temperature-sensing actuator, which is limited by the physical characteristics of heat conduction, the response speed under high-frequency dynamic loads is severely mismatched with the rhythm of load changes, and the peak support force is out of phase with the peak load, which can easily lead to dynamic control failure. This solution decomposes the global load into high-frequency and low-frequency components through time-frequency decomposition. The low-frequency component is supported by the temperature-sensing unit in steady state, ensuring the steady-state control accuracy of the global load distribution. The high-frequency component is compensated by the piezoelectric unit in transient state, relying on its microsecond-level response speed to match the rapid changes of the high-frequency load. The division of labor between the two covers the entire frequency band load, which can prevent dynamic tracking errors caused by insufficient bandwidth of a single actuator and improve the control and following accuracy of the entire frequency band load. Meanwhile, the preset fixed coupling coefficient is calculated based on the static mechanical model, which deviates significantly from the actual modal and force transmission characteristics of the structure under high-frequency dynamic loads. This can easily lead to a mismatch between the control parameters obtained by multi-objective optimization and the actual working conditions. This solution continuously updates the dynamic coupling data between nodes based on real-time monitoring data, so that the multi-objective optimization model always fits the current real force transmission characteristics of the structure. The optimal PID parameters obtained by the solution are highly adapted to the current working conditions, which reduces the model mismatch error from the root.
[0009] In addition, this solution generates compensation and correction data through trend prediction and data repair, performs advanced calculations based on high-frequency components, and generates high-frequency drive commands. It predicts load change trends in advance and outputs corresponding support forces, actively offsetting delays and ensuring that the support force output is phase-synchronized with load changes. This eliminates dynamic control deviations caused by lag and improves the control accuracy of high-frequency transient loads. Moreover, since single-loop control lacks execution feedback calibration and cannot cover force offsets after structural parameter changes, control errors will continue to accumulate. This solution constructs a two-stage calibration closed loop: after the initial support execution, initial fine-tuning is performed based on execution feedback data to eliminate the output deviation of the actuator and ensure the steady-state accuracy of single control. After the addition of steel structure welding and fixing, full-domain perception data is collected again and secondary fine-tuning is performed to correct the force distribution offset caused by structural parameter changes, achieving accuracy recalibration after structural changes, forming a complete closed loop of "detection-execution-feedback-calibration". Finally, resonance tuning data is obtained through frequency domain analysis, and resonance adjustment commands are generated. By using external forces such as wind and earthquakes to stimulate structural resonance, the natural frequency of the structure is actively adjusted so that the natural frequency avoids the main load frequency band, thereby reducing the amplitude and deformation of structural vibration from the source, reducing the difficulty of control and the upper limit of deviation, and improving the control accuracy and stability under extreme dynamic loads.
[0010] In summary, this solution, through time-frequency hierarchical control, dynamic model adaptation, advanced prediction compensation, two-level closed-loop calibration, and resonance frequency adjustment, can systematically eliminate control errors and improve the load distribution control accuracy of steel structures under dynamic loads.
[0011] Furthermore, the data processing module performs time-domain windowing on the standard monitoring data to obtain segmented time-domain data, performs Fourier transform on the segmented time-domain data to obtain frequency-domain distribution data, and divides the frequency-domain distribution data into frequency bands to obtain high-frequency component data and low-frequency component data.
[0012] The beneficial effects are: effectively suppressing spectral leakage errors, accurately converting to obtain complete frequency domain distribution data, accurately separating high-frequency component data from low-frequency component data, and avoiding frequency band aliasing errors.
[0013] Furthermore, the data processing module obtains time-series sample data by performing sliding window interception based on standard monitoring data, and obtains dynamic coupling data by recursive identification based on the time-series sample data; it also obtains historical time-series data by constructing sequences based on standard monitoring data, and obtains compensated and corrected data by predictive completion based on the historical time-series data.
[0014] The beneficial effects are: real-time matching of structural force transmission characteristics, eliminating model mismatch errors caused by fixed coefficients; accurate prediction of load change trends and filling of data gaps, effectively offsetting control delays and improving the real-time performance of dynamic load distribution.
[0015] Furthermore, the data processing module calculates the spectral values based on the standard monitoring data to obtain power spectrum data, identifies peak points based on the power spectrum data to obtain frequency peak feature data, and derives the resonant tuning data based on the frequency peak feature data; it constructs the target based on the low-frequency component data and the dynamic coupling data to obtain the optimization target data, iterates and optimizes based on the optimization target data to obtain the candidate parameter data, and selects the optimal PID parameter data based on the candidate parameter data.
[0016] The beneficial effects are: accurately locating the structural resonant frequency point, avoiding tuning deviations caused by frequency misjudgment; ensuring the optimality of global load distribution, eliminating control errors caused by coarse parameter solutions, and improving the overall accuracy of dynamic load distribution.
[0017] Furthermore, the instruction generation module performs parameter conversion based on the optimal PID parameter data to obtain temperature control reference data, performs pulse width modulation based on the temperature control reference data to obtain low-frequency heating instructions, performs time-series superposition of high-frequency component data and compensation correction data to obtain advanced reference data, and performs amplitude adaptation based on the advanced reference data to obtain high-frequency drive instructions.
[0018] The beneficial effects are: precise control of the output of the temperature sensing unit, reducing overshoot deviation caused by thermal inertia; precise matching of the high-frequency load change rhythm, offsetting control delay, and improving the tracking and control accuracy of high-frequency transient loads.
[0019] Furthermore, the instruction generation module performs frequency difference analysis based on the resonance tuning data to obtain tuning target data, and performs instruction encoding based on the tuning target data to obtain resonance adjustment instructions; it performs point calculation based on assembly requirement information to obtain welding coordinate data, and performs process encoding based on the welding coordinate data to obtain solid welding operation instructions.
[0020] The beneficial effects are: accurately matching the structural tuning target, avoiding tuning failure caused by frequency control deviation, and ensuring the accuracy of the inherent frequency adjustment; accurately locating the welding position and operation sequence of the newly added steel structure, reducing welding alignment error, and ensuring the control stability and accuracy of load distribution after structural changes.
[0021] Furthermore, the load execution module performs a difference comparison based on the execution feedback data to obtain output deviation data, performs correction calculation based on the output deviation data to obtain correction instruction data, and performs output fine-tuning based on the correction instruction data to obtain initial adjustment verification data, thus completing the initial fine-tuning.
[0022] The beneficial effects are: quantifying the deviation between the actual output and the target value, accurately correcting the output deviation and outputting initial calibration data, effectively eliminating the inherent output error at the actuator, forming a closed-loop calibration for a single execution, and improving the accuracy and stability of the initial support control.
[0023] Furthermore, the load execution module performs deviation comparison based on the re-collected global perception data to obtain structural offset data, recalculates parameters based on the structural offset data to obtain readjustment command data, and performs output calibration based on the readjustment command data to obtain final adjustment verification data, thus completing the secondary fine-tuning operation.
[0024] The beneficial effects are: accurately quantifying the stress offset after structural changes, matching the updated structural characteristics to generate appropriate corrections, forming a complete calibration closed loop after structural changes, and improving the control accuracy and operational stability of load distribution after welding operations.
[0025] Furthermore, the load execution module performs excitation analysis based on the resonance adjustment command to obtain vibration parameter data, performs external force vibration based on the vibration parameter data to obtain resonance response data, and performs frequency point adjustment based on the resonance response data to obtain fixed frequency calibration data, thus completing the natural frequency point adjustment.
[0026] The beneficial effects are as follows: the excitation parameters are generated by accurately matching the tuning target, resulting in highly adaptable vibration parameter data, avoiding tuning failure caused by excitation deviation; the fixed frequency calibration data is output after frequency point adjustment, which can accurately offset the structure's natural frequency point, effectively avoid the main load frequency band, reduce the structural vibration amplitude and control deviation, and improve the long-term control accuracy of dynamic load distribution. Attached Figure Description
[0027] Figure 1 This is a system block diagram of Embodiment 1 of a steel structure control system based on dynamic load distribution; Figure 2 This is a flowchart of an embodiment 2 of a steel structure control method based on dynamic load distribution. Detailed Implementation
[0028] The following detailed description illustrates the specific implementation method: Example 1 This embodiment describes a steel structure control system based on dynamic load distribution, applied to the roof support system of a single-story steel structure industrial plant with a span of 36m. The roof bears multiple dynamic loads, including daily wind loads, vibration loads from production equipment, and moving loads from maintenance personnel, with load frequencies covering the 0.1Hz~20Hz range. (See attached image.) Figure 1 As shown, the system includes a data acquisition module, a data processing module, an instruction generation module, and a load execution module. The signal output terminal of the data acquisition module is electrically connected to the signal input terminal of the data processing module; the output terminal of the data processing module is electrically connected to the input terminal of the instruction generation module; the instruction output terminal of the instruction generation module is electrically connected to the control input terminal of the load execution module; and the feedback signal output terminal of the load execution module is electrically connected to the feedback input terminal of the data acquisition module, forming a closed-loop control system. The specific implementation process is as follows: The data acquisition module consists of a resistance strain sensor, a piezoelectric accelerometer, a hydraulic pressure sensor, and a wind pressure sensor. All sensors are deployed along key locations such as the mid-span, beam-column joints, and support points of the roof steel beams, with a total of 12 strain monitoring points, 8 acceleration monitoring points, 6 pressure monitoring points, and 4 wind pressure monitoring points. All sensors achieve synchronous sampling via a synchronous acquisition card, with a sampling frequency set to 100Hz. The data acquisition module acquires the output analog signals of each sensor in real time to obtain comprehensive sensing data. This comprehensive sensing data includes four types of physical quantities: structural strain data, vibration acceleration data, support hydraulic pressure data, and roof wind pressure data, covering all dimensions of information regarding the structural stress state and external load input.
[0029] The data processing module is deployed in the edge computing controller. After receiving the global perception data, it completes the processing according to the following progressive process: Initial Status Assessment and Standardization: Based on the full-domain sensing data, an initial status assessment is performed by comparing the current structural deformation and load amplitude with preset safety thresholds to determine the current load level and safety status of the structure, thus obtaining initial status data. Based on this initial status data, a filtering and transcoding operation is performed, using a third-order Butterworth low-pass filter to remove high-frequency electrical noise. Simultaneously, unified encoding and dimensional normalization of analog to digital quantities are completed to obtain standard monitoring data. All physical quantities in the standard monitoring data are converted into standard digital signals, with a uniform sampling rate of 100Hz.
[0030] Time-frequency decomposition processing: Based on standard monitoring data, a time-domain windowing operation is performed. The Hanning window is used to segment the 1024-point time-series data into segments to suppress spectral leakage, resulting in segmented time-domain data. Based on the segmented time-domain data, a Fourier transform operation is performed. The Fast Fourier Transform (FFT) is used to convert the time-domain signal to the frequency domain. The calculation formula is as follows: in, For the segmented time-domain data, the first Values of each sampling point, This represents the total number of sampling points per segment in this embodiment. , For the first The frequency domain complex values of each frequency point are calculated to obtain frequency domain distribution data. Based on the frequency domain distribution data, frequency band division is performed. With 5Hz as the dividing frequency, the frequency band below 5Hz is divided into low frequency band and the frequency band above 5Hz is divided into high frequency band. The amplitude and phase information of the corresponding frequency bands are extracted to obtain high frequency component data and low frequency component data.
[0031] Coupling Coefficient Iteration and Trend Prediction Compensation: Based on standard monitoring data, a sliding window truncation operation is performed to select the time-series data from the most recent 200 sampling periods as the calculation sample, resulting in time-series sample data. Based on the time-series sample data, a recursive identification operation is performed, and the recursive least squares method is used to update the force transmission coupling coefficient between each support node in real time. The calculation formula is as follows: in, for The coupling coefficient vector at time step, This is the algorithm gain vector. Let covariance be the parameter. For the input data vector, This is the current observed support force, in kN. It is the identity matrix. As the forgetting factor, in this embodiment, we take... Dynamic coupling data is obtained through iterative calculation, and the dynamic coupling data characterizes the degree of mutual influence of the forces on different support points.
[0032] Based on standard monitoring data, a sequence construction operation is performed to organize the continuous monitoring data of the past 30 minutes into a regular time series, thus obtaining historical time series data. Based on the historical time series data, a prediction completion operation is performed, using an autoregressive integral moving average (ARIMA) model to predict the load change trend for the next 5 sampling periods. At the same time, linear interpolation is performed to repair the data loss period, resulting in compensated and corrected data. The order of the ARIMA model is set to ARIMA(2,1,1), and it is trained offline based on historical load data and then run online.
[0033] Resonance characteristic analysis and control parameter optimization: Based on standard monitoring data, spectral value calculation is performed, and the power spectral density of the structural vibration signal is calculated using the periodogram method to obtain power spectrum data; based on the power spectrum data, peak point identification is performed, and the three peaks with the largest amplitude in the power spectrum and their corresponding frequencies are extracted to obtain frequency peak characteristic data; based on the frequency peak characteristic data, tuning derivation is performed, and the current structural natural frequency is compared with the main load frequency to calculate the required natural frequency offset and adjustment direction, thus obtaining resonance tuning data.
[0034] Based on low-frequency component data and dynamic coupling data, a target construction operation is performed. With "minimum support force deviation and minimum structural deformation" as the optimization objectives, a multi-objective optimization function is constructed to obtain the optimization target data. The optimization objective function is as follows: in, The objective function value is a dimensionless value. , As the weighting coefficient, in this embodiment , ; For the first The actual supporting force of each support point is expressed in kN. For the first The target supporting force of each support point is expressed in kN. As the benchmark value for support force, the rated bearing capacity of a single support is taken as 100kN; For the first The structural deformation at each monitoring point is expressed in mm. The deformation reference value is taken as the maximum allowable deformation of the structure, which is 30mm. The total number of support points, This represents the total number of deformation monitoring points. Based on the target data, an iterative optimization operation is performed using the particle swarm optimization algorithm. The number of iterations is set to 50, and the population size is set to 30, resulting in multiple sets of candidate parameter data. Based on these candidate parameters, an optimization selection operation is performed, choosing the set of parameters with the smallest objective function value as the final result, thus obtaining the optimal PID parameter data. For example, in the optimal PID parameters obtained after optimization, the proportional coefficient is 2.3 and the integral coefficient is 0.06s. - ¹, The differential coefficient is 0.25s.
[0035] The instruction generation module receives various types of data output by the data processing module and converts them into control instructions that can be directly recognized by the execution end. The specific process is as follows: High and low frequency and main support drive command generation: Based on the optimal PID parameter data, a parameter conversion operation is performed to convert the PID control output into the target temperature value of the temperature sensing unit, obtaining temperature control reference data; based on the temperature control reference data, a pulse width modulation operation is performed to convert the target temperature value into a PWM drive signal with adjustable duty cycle, obtaining a low-frequency heating command; the low-frequency heating command is used to control the heating power of the temperature sensing unit. Based on the high-frequency component data and compensation correction data, a timing superposition operation is performed to align and superimpose the high-frequency load component and the predicted compensation amount on the time axis, obtaining advanced reference data; based on the advanced reference data, an amplitude adaptation operation is performed to linearly map the control quantity to the drive voltage range (0~150V) of the piezoelectric unit, obtaining a high-frequency drive command. Based on the optimal PID parameter data, an output calculation operation is performed to convert the total support force target value output by the PID into the target output force value of the motor jack, obtaining the main support drive command; the main support drive command is used to control the output of the main support hydraulic pressure by the motor jack.
[0036] In this embodiment, the support-related instructions are specifically the main support drive instructions. The generation process is completely embedded in the operation flow of the instruction generation module. Specifically: after the data operation module obtains the optimal PID parameter data covering all support points through model optimization, it transmits the parameter data to the instruction generation module. The instruction generation module first combines the steel structure load distribution weight matrix to decompose the global control target into the target support force of each motor jack at a single point. Then, it combines the rated output limit of each point and the current real-time output feedback to complete the amplitude limit verification, forming the target output sequence of each support point. Subsequently, the instruction generation module converts the target output sequence into a control quantity code adapted to the hydraulic proportional control valve, and finally generates a standardized main support drive instruction, which is transmitted to the motor jack control terminal of the load execution module through the industrial fieldbus to drive the jack to output the corresponding hydraulic pressure. The main support drive command, together with the low-frequency heating command and the high-frequency drive command, constitutes a collaborative control command system. The three of them respectively undertake the functions of main load steady-state control, low-frequency thermal deformation compensation, and high-frequency transient vibration compensation. They are output synchronously under the same time reference, and the output effect is linearly superimposed at the structural support point, jointly realizing the precise distribution of dynamic load across the entire frequency band.
[0037] In this embodiment, the time-series superposition of high-frequency component data and compensation correction data first achieves time-domain alignment and dimensional unification of the two types of data: after frequency domain division, the high-frequency component data is reconstructed into a time-domain high-frequency alternating signal with the same sampling rate and time base as the standard monitoring data through inverse Fourier transform. The sampling rate is unified to 100Hz, and each sampling point corresponds to a high-frequency load increment value. The compensation correction data is generated by the ARIMA model based on historical time-series data prediction, and the output is a time-domain predicted increment sequence with the same time base and the same sampling rate. Each sampling point corresponds to a load prediction compensation value.
[0038] Based on this, a linear weighted superposition operator is used to fuse the two types of data. The superposition formula is as follows: in, For the high-frequency component weight, this embodiment uses 0.7. To predict the compensation weight, this embodiment uses 0.3, and the leading baseline data is obtained after point-by-point calculation. Regarding the PID control closed loop, the input to the PID controller is the support force deviation. in, As a support for the target, To provide the actual support force in the feedback data; the optimal PID parameter data includes the proportional coefficient. Integral coefficient Differential coefficients The incremental PID algorithm is used to calculate and control the output, and the formula is as follows: The output control increment is accumulated and converted into the corresponding drive command, realizing a complete closed-loop control from parameter optimization to execution output.
[0039] Resonance tuning and welding command generation: Based on the resonance tuning data, frequency difference analysis is performed to calculate the difference between the current natural frequency and the target tuning frequency, determining the tuning direction and adjustment amount to obtain the tuning target data; based on the tuning target data, command encoding is performed to convert it into a frequency control command format recognizable by the tuning actuator, obtaining the resonance tuning command. Based on assembly requirements (e.g., the installation requirements of a new roof secondary beam), point calculation is performed to calculate the installation coordinates, alignment accuracy requirements, and welding process parameters of the new component, obtaining welding coordinate data; based on the welding coordinate data, process encoding is performed to plan the welding operation sequence, weld leg height, welding duration, and other process parameters, obtaining the welding operation command.
[0040] The load execution module is deployed at each support point of the steel structure and includes a temperature sensing unit, a piezoelectric unit, a motor jack, a welding operation mechanism, and a frequency tuning mechanism. The specific execution process is as follows: Collaborative support and initial fine-tuning: The temperature sensing unit is driven to work according to the low-frequency heating command. The temperature sensing unit adopts a thermal expansion steel support component, which changes the support length through electric heating to achieve steady-state load support with a response time of about 2 seconds. The piezoelectric unit is driven to work according to the high-frequency drive command. The piezoelectric unit adopts a piezoelectric ceramic stacked actuator with a response time of less than 1 ms and is used to compensate for high-frequency dynamic loads. The motor jack is driven to output hydraulic pressure according to the main support drive command to provide the main support bearing capacity. The three work together to complete the load support operation, and at the same time, the output force and displacement signals of each actuator are collected to obtain execution feedback data.
[0041] Based on the execution feedback data, a difference comparison operation is performed to calculate the difference between the actual output and the target output, thus obtaining the output deviation data. Based on the output deviation data, a correction calculation operation is performed to calculate the output correction amount based on the incremental PID algorithm, thus obtaining the correction command data. Based on the correction command data, a fine-tuning operation is performed to adjust the output force of each execution unit, thus obtaining the initial adjustment verification data and completing the initial fine-tuning.
[0042] Welding Operation and Secondary Fine-tuning: Based on the welding operation command, the automatic welding mechanism is driven to complete the alignment, fixing, and welding of the newly added steel structure. After welding, the data acquisition module collects the global perception data again, and the load execution module receives this re-collected global perception data. Based on the re-collected global perception data, a deviation comparison operation is performed to compare the structural deformation and stress distribution differences before and after welding, obtaining structural offset data. Based on the structural offset data, a parameter recalculation operation is performed to recalculate the target support force and control parameters of each support point, obtaining readjustment command data. Based on the readjustment command data, an output calibration operation is performed to adjust the output of each execution unit to the new target value, obtaining final calibration verification data, completing the secondary fine-tuning operation.
[0043] Natural frequency adjustment: Based on the resonance adjustment command, excitation analysis is performed to analyze control parameters such as excitation frequency and excitation amplitude, obtaining vibration parameter data; based on the vibration parameter data, external force vibration is performed, using a controllable electromagnetic exciter to apply a periodic external force to excite the structure to resonate, and the structural vibration response signal is collected simultaneously to obtain resonance response data; based on the resonance response data, frequency adjustment is performed, using a stepper motor to drive an additional mass block to move along the steel beam axis, and simultaneously adjusting the preload at the support end to change the support stiffness, thereby achieving continuous fine-tuning of the structure's natural frequency until the target frequency is reached, obtaining natural frequency calibration data, and completing the natural frequency adjustment.
[0044] In this embodiment, the core principle of adjusting the structure's natural frequency is to actively change the structure's dynamic parameters to shift the natural frequency, thereby avoiding the main excitation frequency band of external loads and suppressing the risk of resonance, rather than actively inducing resonance with external forces. Specifically, environmental external forces such as wind and equipment vibration are only used as passive modal excitation sources for online identification of the structure's inherent characteristics: the data acquisition module continuously collects structural vibration response data under environmental external force excitation; the data processing module identifies the structure's current actual natural frequency and mode shape through frequency domain analysis, calculates the optimal natural frequency shift and shift direction based on the main frequency distribution of external loads, and generates resonance tuning data; the command generation module generates resonance adjustment commands based on the resonance tuning data, driving the frequency tuning mechanism in the load execution module to work, using a stepper motor to move the additional mass block along the steel beam axis to change the distributed mass of the structure, and simultaneously adjusting the preload at the support end to change the support stiffness. Through the coordinated adjustment of mass and stiffness, the natural frequency of the structure is continuously changed until it shifts to the preset safe frequency band, completing the active control of the natural frequency. Resonance is only used as an observation object for identifying the structure's inherent characteristics, not as a means of active control.
[0045] This embodiment covers dynamic loads across the entire frequency band through a time-frequency hierarchical execution architecture. It improves the matching degree of the control model by combining the dynamic coupling coefficient with real-time iteration. It offsets the full-link control delay through advance prediction compensation and eliminates execution errors and structural change deviations through a two-level closed-loop fine-tuning mechanism. At the same time, it actively adjusts the inherent frequency of the structure to avoid resonance risks from the root. This can improve the control accuracy, operational stability and structural safety of dynamic load distribution in steel structures.
[0046] In this embodiment, a temperature sensing unit, a piezoelectric unit, and a motor jack are selected as the specific implementation forms of the execution end. The purpose is to provide a concrete explanation of the technical solution in conjunction with engineering application scenarios, making it easier for technical personnel to understand and reproduce the solution. In actual implementation, the execution unit of this solution can be replaced in various equivalent ways depending on the application scenario: for example, the temperature sensing unit for low-frequency steady-state support can be replaced with various execution mechanisms with low-frequency large-stroke output capabilities, such as electro-hydraulic servo actuators, shape memory alloy drive units, and magnetostrictive actuators; the piezoelectric unit for high-frequency transient compensation can be replaced with actuators with high-frequency response capabilities, such as electromagnetic actuators, magnetorheological dampers, and electrorheological dampers; the motor jack for main support can be replaced with various main load-bearing execution devices, such as hydraulic servo support cylinders, screw-type electric lifting mechanisms, and pneumatic support devices; the frequency tuning mechanism can also adopt various equivalent structures such as tuned mass dampers and active stiffness adjustment supports.
[0047] Example 2 This embodiment presents a steel structure control method based on dynamic load distribution, applied to the large-span steel structure factory roof support system described in Embodiment 1, as shown in the attached diagram. Figure 2 As shown, the specific implementation process is as follows: S1. Data Acquisition: Strain sensors, acceleration sensors, pressure sensors, and wind pressure sensors are deployed at key locations such as the mid-span, beam-column joints, and support points of the steel structure. All sensors are set to sample synchronously at a sampling frequency of 100Hz to collect real-time signals of the structural stress state and external load inputs, thereby obtaining comprehensive perception data.
[0048] S2. Data processing: After receiving the full-domain sensing data, the system sequentially completes the following operations according to progressive logic: initial state judgment, filtering and transcoding, time-frequency decomposition, coefficient iteration, trend prediction and data repair, frequency domain analysis, and model optimization. Based on the full-domain perception data, a preliminary state judgment operation is performed, and the current load level and structural safety status are compared with the preset structural safety threshold to obtain the preliminary state judgment data. Based on the initial state data, filtering and transcoding operations are performed. High-frequency noise is removed by a third-order Butterworth low-pass filter, and dimensional normalization and digital encoding are completed at the same time to obtain standard monitoring data. Based on standard monitoring data, time-frequency decomposition is performed. The high-frequency and low-frequency components of the load are separated by windowed fast Fourier transform to obtain high-frequency component data and low-frequency component data. Based on standard monitoring data, coefficient iteration is performed, and the force transmission coupling coefficients between each node are updated in real time using the recursive least squares method to obtain dynamic coupling data. Based on standard monitoring data, trend prediction and data repair operations are performed. The load change trend is predicted by time series prediction model, and missing values during the data loss period are repaired to obtain compensated and corrected data. Based on standard monitoring data, frequency domain analysis is performed, and the structural resonance peak is identified through power spectrum analysis. The target adjustment amount of the natural frequency is calculated to obtain resonance tuning data. Based on low-frequency component data and dynamic coupling data, model optimization is performed with the goal of minimizing support force deviation and structural deformation. The optimal control parameters are solved by intelligent optimization algorithm to obtain the optimal PID parameter data.
[0049] S3. Instruction generation: Based on the various parameters obtained from data processing, corresponding control instructions that can be recognized by the execution end are generated. Based on the optimal PID parameter data, a signal compilation operation is performed to convert it into a PWM heating drive signal for the temperature sensing unit, thereby obtaining a low-frequency heating command; Based on the high-frequency component data and the compensation correction data, a forward calculation operation is performed, and the predicted compensation amount is superimposed to generate a drive signal with a forward time to obtain a high-frequency drive command. The target output of the main support is calculated based on the optimal PID parameter data, and the main support drive command is obtained. Based on the resonance tuning data, a frequency control compilation operation is performed to convert it into control instructions for the frequency tuning mechanism, thus obtaining the resonance adjustment instructions; Based on the assembly requirements, welding compilation operations are performed to plan welding points, process parameters, and operation sequences, resulting in welding operation instructions.
[0050] S4, Load Execution: After receiving various control commands, it drives the corresponding actuators to complete support operations, calibration adjustments, and frequency control. The low-frequency heating command drives the temperature sensing unit to support the steady-state load, the high-frequency drive command drives the piezoelectric unit to compensate for the high-frequency dynamic load, and the main support drive command drives the motor jack to output the main support hydraulic pressure. The three work together to complete the load support operation. At the same time, the output and displacement signals of the actuator are collected to obtain the execution feedback data. The initial fine-tuning operation is performed based on the execution feedback data to eliminate the output deviation of the actuator. The alignment and welding of the newly added steel structure are completed according to the welding operation instructions. After the welding is completed, the full-domain perception data is collected again. Based on the collected full-domain perception data, a second fine-tuning operation is performed to correct the force distribution shift caused by the change in structural parameters. According to the resonance adjustment command, an external force is applied to the structure to generate resonance by using an electromagnetic vibrator. The natural frequency of the structure is changed by moving the additional mass block and adjusting the support preload to avoid the main load frequency band, thus completing the natural frequency adjustment operation.
[0051] This embodiment achieves hierarchical control and advance compensation of high and low frequency loads through a closed-loop management process of "acquisition-computation-generation-execution". It is combined with a two-level fine-tuning mechanism to ensure control accuracy after structural changes, while actively adjusting the inherent characteristics of the structure to avoid resonance risks. It can effectively adapt to multi-source dynamic load scenarios and improve the control accuracy and operational reliability of steel structure load distribution.
[0052] The above are merely embodiments of the present invention. The invention is not limited to the fields covered by these embodiments. Commonly known structures and characteristics in the solutions are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are able to access all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.
Claims
1. A steel structure control system based on dynamic load distribution, characterized in that, include: Data acquisition module, data processing module, instruction generation module, and load execution module; The data acquisition module is used to collect global perception data; The data processing module is used to perform initial state judgment based on the full-domain sensing data to obtain initial state data; to perform filtering and transcoding based on the initial state data to obtain standard monitoring data; to perform time-frequency decomposition based on the standard monitoring data to obtain high-frequency component data and low-frequency component data; to perform coefficient iteration based on the standard monitoring data to obtain dynamic coupling data; to perform trend prediction and data repair based on the standard monitoring data to obtain compensation and correction data; and to perform frequency domain analysis based on the standard monitoring data to obtain resonance tuning data. The optimal PID parameter data are obtained by model optimization based on low-frequency component data and dynamic coupling data; The instruction generation module is used to compile signals based on optimal PID parameter data to obtain low-frequency heating instructions; to calculate output based on optimal PID parameter data to obtain support-related instructions; to perform advance calculations based on high-frequency component data and compensation correction data to obtain high-frequency drive instructions; to compile frequency control based on resonance tuning data to obtain resonance adjustment instructions; and to compile welding instructions based on assembly requirements information to obtain solid welding operation instructions. The load execution module is used to drive the temperature sensing unit to work according to the low-frequency heating command, drive the piezoelectric unit to work according to the high-frequency driving command, and drive the motor jack to output hydraulic pressure according to the support-related command, so as to complete the load support operation in a coordinated manner, obtain execution feedback data, and perform initial fine-tuning based on the execution feedback data; complete the alignment and welding of the new steel structure according to the welding operation command, collect the global sensing data again, and perform secondary fine-tuning operation based on the re-collected global sensing data; and adjust the natural frequency point by using external force to stimulate the resonance of the structure according to the resonance adjustment command.
2. The steel structure control system based on dynamic load distribution according to claim 1, characterized in that, The data processing module performs time-domain windowing on the standard monitoring data to obtain segmented time-domain data, performs Fourier transform on the segmented time-domain data to obtain frequency-domain distribution data, and divides the frequency-domain distribution data into frequency bands to obtain high-frequency component data and low-frequency component data.
3. A steel structure control system based on dynamic load distribution according to claim 2, characterized in that, The data processing module obtains time-series sample data by performing sliding window interception based on standard monitoring data, and obtains dynamic coupling data by recursive identification based on the time-series sample data; it also obtains historical time-series data by constructing sequences based on standard monitoring data, and obtains compensated and corrected data by predictive completion based on historical time-series data.
4. A steel structure control system based on dynamic load distribution according to claim 3, characterized in that, The data processing module calculates the spectral value based on the standard monitoring data to obtain power spectrum data, identifies peak points based on the power spectrum data to obtain frequency peak characteristic data, and derives the resonant tuning data based on the frequency peak characteristic data. The target data for optimization is obtained by constructing the target data based on the low-frequency component data and the dynamic coupling data. The candidate parameter data is obtained by iterative optimization based on the target data. The optimal PID parameter data is obtained by selection based on the candidate parameter data.
5. A steel structure control system based on dynamic load distribution according to claim 4, characterized in that, The instruction generation module performs parameter conversion based on the optimal PID parameter data to obtain temperature control reference data, performs pulse width modulation based on the temperature control reference data to obtain low-frequency heating instructions, performs time-series superposition of high-frequency component data and compensation correction data to obtain advanced reference data, and performs amplitude adaptation based on the advanced reference data to obtain high-frequency drive instructions.
6. A steel structure control system based on dynamic load distribution according to claim 5, characterized in that, The instruction generation module obtains the tuning target data by performing frequency difference analysis based on the resonance tuning data, and obtains the resonance adjustment instruction by performing instruction encoding based on the tuning target data; it obtains the welding coordinate data by performing point calculation based on the assembly requirement information, and obtains the solid welding operation instruction by performing process encoding based on the welding coordinate data.
7. A steel structure control system based on dynamic load distribution according to claim 6, characterized in that, The load execution module obtains output deviation data by comparing the difference based on the execution feedback data, performs correction calculations based on the output deviation data to obtain correction instruction data, and performs output fine-tuning based on the correction instruction data to obtain initial adjustment verification data, thus completing the initial fine-tuning.
8. A steel structure control system based on dynamic load distribution according to claim 7, characterized in that, The load execution module performs deviation comparison based on the re-collected global perception data to obtain structural offset data, recalculates parameters based on the structural offset data to obtain readjustment command data, and performs output calibration based on the readjustment command data to obtain final adjustment verification data, thus completing the secondary fine-tuning operation.
9. A steel structure control system based on dynamic load distribution according to claim 8, characterized in that, The load execution module performs excitation analysis according to the resonance adjustment command to obtain the vibration parameter data, performs external force vibration according to the vibration parameter data to obtain the resonance response data, and performs frequency point adjustment according to the resonance response data to obtain the fixed frequency calibration data, thus completing the natural frequency point adjustment.
10. A steel structure control method based on dynamic load distribution, characterized in that, Including the following steps: S1. Data Acquisition: Acquire data from the entire sensing domain; S2. Data processing: Based on the full-domain sensing data, perform initial state judgment to obtain initial state data; based on the initial state data, perform filtering and transcoding to obtain standard monitoring data; based on the standard monitoring data, perform time-frequency decomposition to obtain high-frequency component data and low-frequency component data. Based on standard monitoring data, coefficient iteration operations are performed to obtain dynamic coupling data; Based on standard monitoring data, trend prediction and data repair operations are performed to obtain compensated and corrected data. Based on the standard monitoring data, frequency domain analysis was performed to obtain the resonance tuning data; Based on the low-frequency component data and the dynamic coupling data, a model optimization operation is performed to obtain the optimal PID parameter data; S3, Command Generation: Based on the optimal PID parameter data, perform signal compilation operations to obtain low-frequency heating commands; perform output calculations based on the optimal PID parameter data to obtain support-related commands; perform advance calculation operations based on high-frequency component data and compensation correction data to obtain high-frequency drive commands. Based on the resonance tuning data, a frequency control compilation operation is performed to obtain the resonance adjustment command; based on the assembly requirement information, a welding compilation operation is performed to obtain the solid welding operation command. S4. Load Execution: Drives the temperature sensing unit to work according to the low-frequency heating command, drives the piezoelectric unit to work according to the high-frequency drive command, and drives the motor jack to output hydraulic pressure according to the support-related command, to complete the load support operation in a coordinated manner, obtain execution feedback data, and perform the initial fine-tuning operation based on the execution feedback data; completes the alignment and welding operation of the newly added steel structure according to the welding operation command, collects the global sensing data again, and performs the second fine-tuning operation based on the re-collected global sensing data; and performs the natural frequency adjustment operation by using external force to stimulate the structure resonance according to the resonance adjustment command.
Citation Information
Patent Citations
Large steel structure installation control method and system based on dynamic load distribution
CN120335288A
