Whole-process monitoring and intelligent control system for super-tonnage steel tower vertical rotation

Abstract

The application provides a super-tonnage steel tower vertical rotation whole-process monitoring and intelligent control system, belongs to the field of steel tower vertical rotation monitoring and intelligent control, and is used for solving the problems of dispersed monitoring control modules, unreliable sensing, poor coupling optimization robustness, inadaptation of priori and lack of long-term iterative optimization in related technologies. The system comprises a full-dimension monitoring, a full-condition coupling optimization, a full-form priori attribution, a full-risk control execution and a full-process iterative subsystem which are sequentially signal connected; the monitoring continuity is ensured through sensing redundancy and fault repair, the parameter fluctuation is responded to through robust optimization, the attribution accuracy is improved through dynamic attention, the safety precision efficiency is balanced through multi-objective control, the long-term adaptation is ensured through engineering iteration, and finally the steel tower vertical rotation monitoring reliability and control precision are improved, and the full-condition demand is adapted.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of steel tower vertical rotation monitoring and intelligent control, and particularly relates to a super-tonnage steel tower vertical rotation whole-process monitoring and intelligent control system. BACKGROUND

[0002] In the field of bridge, high-rise structure and other engineering, super-tonnage steel tower vertical rotation construction has high requirements for the precision and reliability of whole-process monitoring and intelligent control due to large span and heavy load, and is a core link to ensure construction safety and positioning quality. At present, in the super-tonnage steel tower vertical rotation construction, monitoring and control mainly rely on dispersed functional modules, and an integrated collaborative system has not yet been formed, which is difficult to meet the precise control requirements under complex working conditions.

[0003] In the prior art, steel tower vertical rotation monitoring mainly uses single sensors to collect displacement, angle and other parameters, coupling calculation only considers the interaction of basic error sources, prior attribution relies on fixed working condition characteristics, control instruction generation mainly uses static weight, and lacks long-term engineering data driven iterative optimization mechanism. For example, the monitoring system has no sensor redundancy design, and sensor failure may easily lead to data interruption; coupling optimization does not consider material parameter fluctuation and space-time characteristics, and parameter precision is easily affected by engineering uncertainty; prior model is difficult to adapt to multi-form switching such as vertical rotation and horizontal rotation and extreme working conditions, and attribution accuracy is limited.

[0004] The above prior art has the following defects: first, the sensor reliability is insufficient, and there is no effective data repair and control connection means after failure, which may easily cause construction interruption; second, the coupling optimization has poor robustness and cannot cope with precision degradation caused by parameter fluctuation; third, the prior attribution has low adaptability and is difficult to cover multi-form and extreme working conditions; fourth, the control strategy lacks dynamic cooperation and cannot balance safety, precision and efficiency; fifth, there is no engineering data iterative mechanism, and the system has poor long-term adaptability, which directly leads to steel tower vertical rotation positioning deviation, low construction efficiency and even safety risks. SUMMARY

[0005] The present application provides a super-tonnage steel tower vertical rotation whole-process monitoring and intelligent control system, which can solve the problems of unreliable sensors, non-robust coupling, non-adaptive prior, non-cooperative control and no iterative optimization in the prior art through full-link integrated collaboration, and improve the precision and reliability of steel tower vertical rotation monitoring and control.

[0006] In a first aspect, the application provides a super-tonnage steel tower vertical rotation whole process monitoring and intelligent control system. The system comprises a full-dimension monitoring subsystem, a full-working-condition coupling optimization subsystem, a full-form priori attribution subsystem, a full-risk control execution subsystem, and a full-process iteration subsystem, which are sequentially connected by signals. The full-dimension monitoring subsystem comprises layered measuring points for collecting displacement and stress of the top, waist, and bottom of the steel tower, an axis type sensor for identifying the combined rotation of vertical and horizontal rotation, and an environmental data acquisition unit for obtaining wind speed and temperature. The full-working-condition coupling optimization subsystem comprises a coupling matrix containing space height, time, and stiffness coupling terms, an optimization objective function integrating coupling constraints, and a weighted iteration algorithm based on the space-time characteristics of vertical rotation. The full-form priori attribution subsystem comprises a multi-dimensional feature vector containing the weight and height parameters of the steel tower, a priori model integrating feature importance, and a Bayesian attribution algorithm. The full-risk control execution subsystem comprises a multi-step predictive control model, an instruction generation unit for the coordination of traction force and rotation angle, and a form switching control unit. The full-process iteration subsystem comprises a feedback closed-loop unit connecting the subsystems, an engineering data iteration unit for updating model parameters, and an output accuracy verification unit.

[0007] By adopting the above technical solutions, a "monitoring-coupling-priori-control-iteration" full-link closed loop is constructed, full-dimension monitoring is realized to collect steel tower multi-parameter in all directions, full-working-condition coupling optimization quantifies the interaction effects of space and error sources, full-form priori attribution improves the error positioning accuracy in multiple working conditions, full-risk control execution realizes the coordinated control of traction force and angle, and full-process iteration ensures the long-term adaptability of the system, thereby improving the accuracy and reliability of steel tower vertical rotation monitoring and control.

[0008] Further, the full-dimension monitoring subsystem further comprises a sensing redundancy network, a fault detection and diagnosis module, and a fault data repair module. In the sensing redundancy network, the displacement monitoring main sensor is GNSS, the backup is a laser range finder, the angle monitoring main sensor is an encoder, the backup is a dual-axis tilt sensor, and the cable force monitoring main sensor is a tension and compression force sensor, and the backup is a cable force acquisition unit based on tower body strain calculation. The fault detection and diagnosis module determines the fault by the main and backup data deviation exceeding the preset threshold, and identifies the fault type based on data jumping or gradual change. The fault data repair module calculates the displacement through geometric relationship and inversely calculates the cable force through mechanical formula to repair the parameters collected by the fault sensor.

[0009] By adopting the above technical solutions, the main and backup sensing redundancy is configured for key parameters, the data can be quickly diagnosed and repaired through geometric and mechanical logic in case of failure, the monitoring interruption caused by sensor failure is avoided, and the reliability and data continuity of the monitoring system are improved.

[0010] Further, a post-fault control connection module is also included, which comprises a data smoothing transition unit and a control instruction iteration unit; the data smoothing transition unit processes the displacement angle parameter after fault repair through an exponential function, so as to seamlessly connect with historical monitoring data without mutation; the control instruction iteration unit re-solves the multi-step predictive control model based on the repaired displacement deviation data, and generates an adapted hierarchical traction force instruction.

[0011] By adopting the above technical solution, the fault repair data can be seamlessly connected with the historical data after smoothing processing, and the control instruction is iteratively generated based on the repair data, thereby avoiding disconnection of the control after the fault and ensuring the continuity and stability of the construction process.

[0012] Further, the full-condition coupling optimization subsystem also comprises a parameter fluctuation quantification module, a robust objective function optimization unit and a simulation verification module; the parameter fluctuation quantification module describes the fluctuation range of the elastic modulus and the viscosity coefficient through interval analysis, and describes the monitoring noise of the displacement and strain through probability distribution; the robust objective function optimization unit adds a fluctuation penalty term in the objective function, so that the optimization parameters meet the accuracy standard within the fluctuation range; the simulation verification module checks the optimization parameters through random sampling simulation, and the sampling proportion meeting the accuracy requirement is not less than a set threshold.

[0013] By adopting the above technical solution, the fluctuation characteristics of the material parameters and the monitoring data are quantified, and through robust optimization and simulation verification, it is ensured that the coupling optimization parameters can still meet the accuracy requirement under engineering uncertainty, thereby improving the robustness and engineering applicability of the coupling optimization.

[0014] Further, the full-form prior attribution system also comprises a dynamic attention weight calculation module, a multi-form feature adaptation unit and a prior probability correction module; the dynamic attention weight calculation module calculates the feature weight through a scoring model, and gives higher weight to the hierarchical traction force of the swivel angle; the multi-form feature adaptation unit constructs a mapping relationship between the vertical swivel associated angle and the steel tower height, and between the horizontal swivel associated torque and the swivel radius; the prior probability correction module adjusts the prior probability based on the attention weight, so that the matching degree with the actual error distribution meets the preset condition.

[0015] By adopting the above technical solution, the vertical swivel key features of the steel tower are dynamically focused, the feature requirements of multi-forms such as vertical swivel and horizontal swivel are adapted, the prior probability is corrected to match the actual error distribution, and the accuracy of attribution under multi-form conditions is improved.

[0016] Further, an extreme working condition feature migration module is also included, which comprises an extreme feature identification unit, an extreme feature migration library and an extreme priori correction unit; the extreme feature identification unit determines the extreme working condition based on wind speed exceeding 25 m / s or rotation angle exceeding 90°; the extreme feature migration library stores the mapping of extreme and regular features, and determines the feature amplification coefficient through the parameter ratio of the two; the extreme priori correction unit adjusts the attention weight and priori probability based on the amplification coefficient to adapt the attribution of extreme working condition.

[0017] By adopting the above technical solution, the extreme working condition is identified and the extreme feature is converted into an expanded form of the regular feature, the priori probability is corrected to adapt to the extreme scene, the attribution failure under the extreme working condition is avoided, and the adaptability of the system to complex working conditions is improved.

[0018] Further, the full-risk control execution subsystem also comprises a multi-target priority quantization module, a dynamic weight calculation unit and a multi-step control instruction generation unit; the multi-target priority quantization module calculates the ratio of safety priority associated traction force to rated value, the ratio of accuracy priority associated residual displacement to allowable value, and the ratio of efficiency priority associated rotation speed to maximum value; the dynamic weight calculation unit adjusts the weight of the corresponding target by a preset proportion when a certain priority exceeds a threshold; and the multi-step control instruction generation unit outputs the traction force and speed coordination instruction based on the model of the adjusted weight.

[0019] By adopting the above technical solution, the safety, accuracy and efficiency priorities are dynamically quantified and the control weight is adjusted to generate the coordination instruction, so as to avoid the trade-off caused by single target orientation and balance the safety, accuracy and efficiency of the steel tower vertical rotation construction.

[0020] Further, the rotation form switching control unit comprises a form switching pre-judgment unit, a target form parameter pre-loading unit and a parameter transition trajectory generation unit; the form switching pre-judgment unit triggers the switching preparation 1 second in advance when the difference between the rotation angle and the target angle is less than a set threshold; the target form parameter pre-loading unit pre-loads the stiffness parameters of the target form, and the vertical rotation is the bending stiffness and the horizontal rotation is the torsional stiffness; and the parameter transition trajectory generation unit generates the parameter transition trajectory through a quadratic curve to make the parameter transition smooth.

[0021] By adopting the above technical solution, the form switching is pre-judged and the parameters are pre-loaded in advance, the parameter mutation is avoided through curve transition, the seamless switching of vertical rotation, horizontal rotation and other forms is realized, and the error overshoot in the switching process is reduced.

[0022] Further, the full-dimension monitoring subsystem further comprises a space-time coordinated sampling module, which comprises a spatial hierarchical sampling weight unit, a time dynamic sampling rate unit and a space-time sampling coordination unit; the spatial hierarchical sampling weight unit assigns a sampling weight of 0.4 to the top of the tower, 0.3 to the waist of the tower and 0.3 to the bottom of the tower; the time dynamic sampling rate unit adjusts the sampling rate when the rotation angle changes by 5 degrees, and increases the sampling rate when approaching the target angle; the space-time sampling coordination unit synchronously increases the sampling rate of the top measuring point at a high weight.

[0023] By adopting the above technical solution, the sampling strategy is adjusted based on the spatial characteristics of the steel tower and the construction time stage, the sampling rate of the high-weight measuring point and the key stage is increased, the data precision is ensured while avoiding redundancy, and the monitoring resource configuration is optimized.

[0024] Further, the full-process iteration subsystem further comprises an engineering data acquisition module, a model parameter iteration unit and an iteration effect verification unit; the engineering data acquisition module acquires the steel tower weight stiffness, wind speed traction force and residual displacement data after each vertical rotation construction or extreme working condition; the model parameter iteration unit updates the coupling matrix coefficient, the prior model feature weight and the control model target weight based on the acquired data; the iteration effect verification unit checks the model through subsequent engineering, and confirms that the iteration is completed when the output precision meets the requirements.

[0025] By adopting the above technical solution, the actual engineering data is acquired to iteratively update the parameters of each subsystem, and the iteration effect is confirmed after verification, so that the system is continuously optimized with the accumulation of engineering experience, and the long-term adaptability and application effect are improved.

[0026] In summary, the present application at least has the following beneficial effects:

[0027] 1. A full-link integrated system is provided, which solves the problem of multiple module dispersion in the prior art and improves the steel tower vertical rotation monitoring control precision;

[0028] 2. The continuity of monitoring and control is ensured through sensing redundancy and fault connection;

[0029] 3. The adaptability of the system to complex working conditions is improved through robust coupling and multi-form priori;

[0030] 4. Construction safety, precision and long-term adaptability are balanced through dynamic control and iterative optimization.

[0031] It should be understood that the content described in the summary section is not intended to limit the key or important features of the embodiments of the present application, nor to limit the scope of the present application. Other features of the present application will become apparent through the following description. BRIEF DESCRIPTION OF DRAWINGS

[0032] The above and other features, aspects and advantages of the present embodiments will become more apparent with reference to the following detailed description when taken in conjunction with the accompanying drawings. In the drawings, like reference numerals refer to like elements, in which:

[0033] Figure 1 A schematic diagram of a super-large tonnage steel tower vertical rotation whole process monitoring and intelligent control system in an embodiment of the application is shown. DETAILED DESCRIPTION

[0034] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0035] In addition, the term "and / or" in this document is only to describe the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone. In addition, the character " / " in this document generally represents an "or" relationship between the front and rear associated objects.

[0036] The present application provides a super-large tonnage steel tower vertical rotation whole process monitoring and intelligent control system, which can solve the problems of unreliable sensing, non-robust coupling, and non-adaptive prior art, improve the precision and reliability of vertical rotation monitoring and control, adapt to multiple forms and extreme working conditions, and ensure the safe and efficient vertical rotation construction of super-large tonnage steel towers.

[0037] The embodiments of the present application disclose a super-large tonnage steel tower vertical rotation whole process monitoring and intelligent control system.

[0038] Figure 1 A schematic diagram of a super-large tonnage steel tower vertical rotation whole process monitoring and intelligent control system in an embodiment of the application is shown.

[0039] With reference to Figure 1 The system includes a full-dimensional monitoring subsystem, a full-working-condition coupling optimization subsystem, a full-form priori attribution subsystem, a full-risk control execution subsystem, and a full-process iteration subsystem, and the subsystems are sequentially signal connected.

[0040] The full-dimension monitoring subsystem includes layered measuring points for collecting displacement and stress of the tower top, tower waist and tower bottom, axis type sensors for identifying the combined rotation of vertical rotation and horizontal rotation, and an environmental data collection unit for obtaining wind speed and temperature. In the layered measuring points, the displacement collection of the tower top, tower waist and tower bottom needs to ensure that the measuring points are symmetrically arranged along the central axis of the steel tower, the stress measuring points need to cover the maximum bending cross section of the tower body (such as the tower waist) and the stress concentration area (such as the connection between the tower bottom and the hinge point), and the wind speed sensor of the environmental data collection unit needs to be installed on a 10-20m unobstructed platform near the steel tower, with a height not less than 1 / 2 of the height of the steel tower. The temperature sensor needs to be close to the surface of the steel tower (derusting and applying heat coupling agent), to ensure that the collected data can reflect the actual temperature environment of the steel tower. The displacement monitoring main sensor is an industrial-grade high-precision GNSS receiver, with a static positioning accuracy of ≤±0.5mm and a dynamic response frequency of ≤1Hz. The measurement range of the backup laser range finder is ≥100m, and the accuracy is ≤±0.3mm. The angle monitoring main sensor is an absolute angle encoder, with a resolution of ≤0.001° and an output frequency of ≥10Hz. The measurement range of the backup dual-axis inclination sensor is ±90°, and the accuracy is ≤±0.005°. The cable force monitoring main sensor is a column type tension and compression force sensor, with a measurement range of ≥2000kN and an accuracy of ≤±0.5%F.S. The backup cable force acquisition unit based on strain calculation needs to be calibrated in advance through static loading experiments to establish the mapping relationship between cable force and tower body strain.

[0041] The full-dimension monitoring subsystem also includes a sensing redundancy network, a fault detection and diagnosis module, and a fault data repair module. In the sensing redundancy network, the displacement monitoring main sensor is a GNSS, and the backup is a laser range finder. The angle monitoring main sensor is an encoder, and the backup is a dual-axis inclination sensor. The cable force monitoring main sensor is a tension and compression force sensor, and the backup is a cable force acquisition unit based on tower body strain calculation. The fault detection and diagnosis module determines faults by the deviation of the main and backup data exceeding the preset threshold. The fault type is identified based on data jump or gradual change. The preset threshold is set based on the 2σ principle of sensor accuracy. For example, the GNSS static positioning accuracy is ±0.5mm, so the main and backup displacement data deviation threshold is set to 1mm (2x0.5mm). The accuracy of the laser range finder is ±0.3mm, and the deviation threshold is set to 0.6mm (2x0.3mm). The data jump determination standard is that the single data change exceeds 3 times the normal fluctuation range (such as the single change of angle in normal vertical rotation ≤0.01°, if the single change ≥0.03°, it is determined as a jump fault, which is mostly sensor obstruction or signal interference). The data gradual change determination standard is that the data deviation increases linearly with time and lasts for more than 5 sampling periods (such as the cable force data deviates by 0.5%F.S. for 5 periods, it is determined as a sensor drift fault). The fault data repair module calculates the displacement through geometric relationship and inversely calculates the cable force through mechanical formula to repair the parameters collected by the faulty sensor. The tower top displacement calculation adopts the geometric similarity principle, and the formula is , wherein is the displacement of the tower top, is the displacement of the tower waist laser ranging, is the displacement of the tower bottom, is the total height of the steel tower, is the height of the tower waist measuring point from the tower bottom; the cable force is inversely calculated by using the stress-force relationship of material mechanics, and the formula is , wherein is the cable force of the small compression rod, is the stress value collected by the tower bottom strain sensor, is the strain value collected by the tower bottom strain sensor combined with the elastic modulus of the steel tower material , that is, , wherein is the measured elastic modulus of the steel tower steel material, is the stress section area of the tower bottom, is the included angle between the small compression rod and the horizontal direction.

[0042] The full-dimension monitoring subsystem also includes a space-time coordinated sampling module, which includes a space hierarchical sampling weight unit, a time dynamic sampling rate unit, and a space-time sampling coordination unit. The space hierarchical sampling weight unit assigns a sampling weight of 0.4 to the tower top, 0.3 to the tower waist, and 0.3 to the tower bottom. The weight distribution is based on the displacement sensitivity of the steel tower vertical rotation. The tower top displacement is most affected by wind-induced and stiffness attenuation, with an error contribution ratio of more than , so the highest weight is assigned. The tower bottom displacement is less affected by the constraint condition, with an error contribution ratio of about , so the weight is the lowest. The time dynamic sampling rate unit adjusts the sampling rate every 5 degrees of rotation angle change. The sampling rate is increased when approaching the target angle. The specific sampling rate adjustment rule is: when the rotation angle is , the sampling rate is set to 10 Hz (start-up stage, displacement changes gently), , it is set to 20 Hz (vertical rotation stage, displacement change rate is improved), , it is set to 50 Hz (just-in-time stage, which needs millimeter-level precision control); the space-time sampling coordination unit synchronously increases the sampling rate of the tower top measuring point when the weight is high. For example, when (just-in-time stage, sampling rate 50 Hz), the tower top measuring point is sampled at 50 Hz, the tower waist measuring point is sampled at , because the tower waist weight 0.3 is of the tower top 0.4, so the sampling rate is , and the tower bottom measuring point is sampled at . This not only ensures the data density of key measuring points, but also avoids data redundancy of non-key measuring points.

[0043] The system also includes a post-fault control transition module, which comprises a data smoothing transition unit and a control command iteration unit. The data smoothing transition unit processes the displacement angle parameters after fault repair using an exponential function, ensuring seamless transition with historical monitoring data. The formula for the exponential function is: In the formula These are the smoothed parameter values ​​(displacement or angle). These are the parameter values ​​at the time of the fault. These are the parameter values ​​after the repair. For the attenuation coefficient (ensure completion within 20ms) (The above is a smooth transition) Time step (unit: The control command iteration unit, based on the repaired displacement deviation data, re-solves the multi-step predictive control model to generate an adapted graded traction force command; the displacement deviation data is the difference between the repaired actual displacement and the target displacement. When iterating control commands, the core parameters of the MPC model of the full-risk control execution subsystem need to be called. For example, the preliminary calculation formula for traction adjustment is as follows: In the formula The bending stiffness of the steel tower at the current angle. The lever arm is small, and the command value will be further corrected by combining the results of multi-objective priority quantification.

[0044] The all-dimensional monitoring subsystem is used to collect data on steel tower displacement, stress, rotation morphology, and environmental parameters. This data is then transmitted to the all-condition coupling optimization subsystem via industrial Ethernet or industrial 5G networks. Data transmission must use a unified data frame format protocol (such as Modbus-TCP). Each data frame includes a parameter identifier, a data acquisition timestamp, a numerical value, and a checksum. The timestamp must be synchronized with the edge controller clock of the all-condition coupling optimization subsystem based on the NTP protocol, with a synchronization deviation ≤10ms, ensuring data time consistency during subsequent coupling optimization calculations. For high-frequency acquisition devices such as GNSS and laser rangefinders, a data caching mechanism (local cache capacity ≥16GB) is employed. In the event of a temporary network interruption (interruption time ≤2 hours), cached data can be automatically retransmitted after network recovery, preventing data loss from affecting coupling optimization accuracy.

[0045] The full-condition coupled optimization subsystem includes a coupling matrix containing spatial height and temporal stiffness coupling terms, an optimization objective function incorporating coupling constraints, and a weighted iterative algorithm based on the spatiotemporal characteristics of vertical rotation; the coupling matrix is... Dimensions, excluding spatially highly coupled terms ( ) and time stiffness coupling term ( In addition to ), it also includes parameter fluctuation coupling terms ( ), morphological switching coupling item ( ) and extreme feature coupling terms ( ), each coupling term quantification formula is derived based on the vertical rotation mechanical characteristics of the steel tower: where , corresponding to the amplification effect of the tower top displacement with the tower bottom displacement, and (0.2 is the height amplification coefficient, is the height of the measuring point); , corresponding to the nonlinear decay of the stiffness with time, and (0.2 is the angle decay coefficient, 0.1 is the time decay coefficient, is the total vertical rotation time); , corresponding to the influence of parameter fluctuation on error, is the key parameter such as elastic modulus or viscous coefficient. The optimization objective function of the fused coupling constraint needs to integrate the space-time coupling, parameter robustness and morphological transition requirements, and the specific expression is ; in the formula is the relative deviation of each error source (wind-induced, temperature, traction, etc. is the relative deviation of the jth error source, and the relative deviation of the ith error source is is the same type of parameter only with different indexes), is the space-time coupling weight (adapted to the characteristics of the strong space-time coupling of the steel tower vertical rotation), is the robustness penalty weight (to ensure that the error is controllable when the parameters fluctuate), is the morphological transition weight (adapted to the demand of multi-morphology switching), is the parameter transition trajectory when the morphology is switched, is the element of the coupling matrix in the ith row and the jth column, is the displacement deviation corresponding to the ith error source, is the displacement deviation of the ith error source corresponding to the p functions of the key parameter, which represents the size of the displacement deviation of the error source when the key parameter takes the value p. Correspondingly, when the key parameter fluctuates to , the size of the displacement deviation of the ith error source is ; the constraint conditions of the objective function include (avoid excessive decay of stiffness), (traction safety margin), and the error deviation after parameter fluctuation .

[0046] The weighted iterative algorithm based on the space-time characteristics of vertical rotation is the gradient descent iteration combined with the spatial layer weight and the time stage weight, and the parameter update formula is ; in the formula is the basic learning rate, is the spatial weight (0.4 for the tower top, 0.3 for the tower waist, and 0.3 for the tower bottom, matching the displacement sensitivity difference), Time weight (0.2 for the turning stage, 0.3 for the vertical stage, 0.5 for the in-place stage, matching the optimization priority of different stages), Robust weight (0.4 when the parameter fluctuation risk is high, 0.2 when the parameter fluctuation risk is low), Target function gradient, when calculating, partial derivatives of each optimization parameter (such as , the coefficient of the traction cable force ) are required, for example (Stiffness-related error).

[0047] The full-condition coupling optimization subsystem also includes a parameter fluctuation quantification module, a robust target function optimization unit, and a simulation verification module. The parameter fluctuation quantification module describes the fluctuation range of the elastic modulus and the viscosity coefficient through interval analysis, and describes the monitoring noise of displacement and strain through probability distribution. The fluctuation range of the elastic modulus is set to based on the material properties of steel (Real-time back-calculated stiffness, calculated by , the bending moment of the tower body at the current angle), and the fluctuation range of the viscosity coefficient is (Initial value of optimization ); the displacement monitoring noise follows a normal distribution (based on the accuracy statistics of GNSS and laser range finder), and the strain monitoring noise follows a normal distribution (based on the measured error of the fiber Bragg grating strain sensor), and the parameter fluctuation quantification result needs to be synchronized to the robust target function optimization unit in real time.

[0048] The robust target function optimization unit adds a fluctuation penalty term to the target function, so that the optimization parameters meet the accuracy standard within the fluctuation range. In the fluctuation penalty term , is the parameter fluctuation amount (such as ), is the set of all parameter fluctuation amounts, and the role of this penalty term is to ensure that even if the parameter takes the extreme value within the fluctuation interval, the relative deviation of the error is still ; for example, when takes the minimum value , the increment of the wind-induced error needs to be controlled within , if it exceeds, it is re-optimized by increasing until the constraint is met.

[0049] ​The simulation verification module checks the robustness of the optimized parameters through random sampling simulation, ensuring that the sampling proportion meeting the accuracy requirements is not lower than a set threshold. The random sampling employs the Monte Carlo method, with the number of samplings set to 1000 (to ensure statistical significance). Each sampling randomly selects parameter values ​​from the fluctuation range of each parameter (e.g., ...). from Random sampling, displacement noise from (Randomly selected) and substituted into the coupled optimization model to calculate the residual error. ; Set the threshold as That is, at least 970 out of 1000 samples must meet the requirement. If the conditions are not met, return to the robust objective function optimization unit and adjust. (For example, increasing from 0.3 to 0.4) and then iterating again until the target is reached. Here, The weighting coefficients are fitted based on 10 sets of experimental data from the vertical rotation of steel towers of the same tonnage, where Adapting to strong spatiotemporal coupling characteristics, To ensure robustness to parameter fluctuations, Smoothness of form transitions.

[0050] The full-condition coupling optimization subsystem communicates with the full-dimensional monitoring subsystem to acquire data output from the full-dimensional monitoring subsystem, quantifies the multi-dimensional coupling effects during the vertical rotation of the steel tower, outputs optimized error parameters, and transmits these optimized error parameters to the full-morphology prior factorization system. The data acquisition frequency is linked to the sampling rate of the full-dimensional monitoring subsystem. For example, if the sampling rate of the monitoring subsystem during the positioning phase is 50Hz, the coupling optimization subsystem acquires data every 20ms (ensuring real-time performance). When quantifying the multi-dimensional coupling effects, the actual values ​​of each coupling term must first be calculated based on the monitoring data (e.g., by calculating the displacement difference between the tower top and bottom). The measured values ​​are then compared with the theoretical coupling matrix to correct the matrix coefficients; the output optimized error parameters include wind-induced error. Temperature error Traction error Stress error Each error is accompanied by a confidence interval (calculated based on the parameter fluctuation quantification results). The EtherCAT bus is used when transmitting the data to the full-morphology prior factorization system, with a delay of [missing information]. This ensures the timeliness of attribution calculations.

[0051] The full-morphology prior attribution system includes a multi-dimensional feature vector containing steel tower weight and height parameters, a prior model that integrates feature importance, and a Bayesian attribution algorithm; the multi-dimensional feature vector is specifically... ,in Including the weight of the steel tower ,high Initial bending stiffness For the shape of rotation (vertical rotation) Horizontal rotation ), Construction phase (start and turn) Vertical rotation 1. In position ), Including wind speed Ambient temperature Extreme operating condition marking (regular) extreme Features in each dimension need to be normalized (e.g., ...). (This is the maximum height of similar steel towers in the project), to avoid scale differences affecting the model output. The prior model for fusing feature importance adopts a CNN-LSTM hybrid architecture, where the CNN layer (containing 3 convolutional kernels of size [missing information]) is the largest height among similar steel towers in the project, to avoid scale differences affecting the model output. The prior model for fusing feature importance adopts a CNN-LSTM hybrid architecture, where the CNN layer (containing 3 convolutional kernels of size [missing information]) is [missing information]. The convolutional layers are used to extract local correlations between features (such as the correlation between wind speed and wind-induced error), and the LSTM layer (containing two hidden layers, each with 64 neurons) is used to capture temporal features (such as the dynamic change of stiffness with rotation angle). The model output is the initial prior probability of each error source (wind-induced, temperature, traction, etc.). The Bayesian attribution algorithm calculates the posterior weights based on the prior probabilities and optimization errors, and the core formula is: ,in For the first Attribution weights for each error source, The prior probability output by the model. Let f be the likelihood function, and ( The feature importance coefficient, For robust accuracy coefficients, (Standard deviation of error). The full-morphology prior factorization system also includes a dynamic attention weight calculation module, a multi-morphology feature adaptation unit, and a prior probability correction module. The dynamic attention weight calculation module calculates feature weights through a scoring model, assigning higher weights to the graded traction force at the rotation angle; the scoring model is a 3-layer multilayer perceptron (MLP), and the input is a multi-dimensional feature vector. The output is the importance score of each feature. The formula is ( This is the weight matrix. For bias terms, (The ReLU activation function is used); weight normalization uses the softmax function, and the formula is... The weights of the rotation angle feature need to be multiplied by a factor of 1.5, the weights of the graded traction force feature need to be multiplied by a factor of 1.2, and the weights of non-critical features (such as horizontal rotation torque) need to be multiplied by a factor of 0.5 to ensure that the core features are focused. A mapping relationship is constructed using multi-form feature adaptation units: vertical rotation is associated with the angle and the height of the steel tower, and horizontal rotation is associated with the torque and the rotation radius. The specific mapping logic is as follows: in the vertical rotation form, the weight of the torque feature is reduced to 0.05 (the weight of the conventional weight). The weight of the steel tower height feature is increased to 0.3 (1.5 times the conventional weight); in the horizontal rotation mode, the weight of the rotation angle feature is reduced to 0.1 (1 / 3 of the conventional weight). The torque feature weight is increased to 0.4 (twice the normal weight), and the mapping relationship needs to be stored in the feature adaptation table, along with the morphology identifier. Automatic invocation. The prior probability correction module adjusts the prior probability based on attention weights to ensure that its matching degree with the actual error distribution meets preset conditions; the correction formula is as follows. ,in In order to be with the first The set of features related to each error source (such as wind-induced error associated with wind speed and steel tower height features) is predefined by the deviation between the corrected prior probability and the actual error proportion. If the value exceeds the limit, the attention weight coefficient will be readjusted.

[0052] The full-morphology prior factorization system also includes an extreme condition feature transfer module, which comprises an extreme feature identification unit, an extreme feature transfer library, and an extreme prior correction unit. The extreme feature identification unit is based on wind speed exceeding... Or the rotation angle exceeds To determine extreme working conditions, 25 m / s is the upper limit of the normal wind speed for steel tower vertical rotation construction as specified in the "Standard for Acceptance of Construction Quality of Steel Structures" (GB50205-2020), and 90° is the target angle for the steel tower vertical rotation into position. Exceeding these limits constitutes an extreme working condition. In addition to the above criteria, cable tension monitoring values ​​must also be considered (cable tension exceeding...). When the threshold is reached (synchronous determination is based on extreme conditions), the determination must meet the condition of "exceeding the threshold for three consecutive sampling periods (each period being 20ms)" to avoid misjudgments caused by instantaneous interference. An extreme feature transfer library stores extreme and normal feature mappings, and the feature amplification coefficient is determined by the ratio of their parameters; the feature amplification coefficient... The calculations are divided into two categories: extreme characteristics of wind speed. ( For the normal maximum wind speed, such as hour ), angular extreme features ( For the normal maximum angle, such as hour The migration library needs to be updated regularly (a new one is added after each extreme condition project is completed). The value ensures mapping accuracy. The extreme prior correction unit adjusts the attention weights and prior probabilities based on the amplification factor to adapt to extreme operating conditions attribution; the attention weight adjustment formula is... The prior probability correction formula is: (0.1 is the extreme correction coefficient). For example, under extreme wind conditions, the attention weight of wind speed features associated with wind-induced errors is increased from 0.1 to... The prior probability of wind-induced error was corrected from 0.4 to This ensures that attribution focuses on extreme error sources.

[0053] The full-form prior attribution system communicates with the full-condition coupled optimization subsystem to obtain the optimization error parameters output by the full-condition coupled optimization subsystem. Based on the steel tower parameters and operating characteristics, it generates prior probabilities, combines these with the optimization errors to complete attribution, and transmits the attribution results to the full-risk control execution subsystem. This communication uses Industrial Ethernet (Modbus-TCP protocol), and the data transmission frequency is consistent with the output frequency of the coupled optimization subsystem (every 20ms), and a timestamp is required (synchronized with the clock of the full-dimensional monitoring subsystem, deviation...). This ensures that the optimization error matches the time characteristics of the operating conditions; the output attribution results include the weights of each error source. Error elimination priority (according to) The data should be sorted from largest to smallest, and suggestions for adjusting key control parameters (such as adjusting the wind cable tension when the wind-induced error weight exceeds 0.4) should be transmitted to the full risk control execution subsystem. Data encryption (AES-128 encryption algorithm) should be used to avoid data tampering affecting control security.

[0054] The full-risk control execution subsystem includes a multi-step predictive control model, a graded traction force and rotation angle coordinated command generation unit, and a form switching control unit; the multi-step predictive control model is based on the vertical rotation dynamics of the steel tower, and its core dynamic equation is... ,in Vertical rotational angular velocity (unit: ), These represent the traction forces of the small and large pressure bars, respectively (unit: kN). The stress arm (unit: m, varies with the rotation angle) is given. (for the initial lever arm) Moment of inertia of the steel tower (unit: As the angle changes, The initial moment of inertia is given by the weight of the steel tower. With center of gravity height Calculated (For gravitational acceleration); the model prediction time domain is set to 15 time steps (20ms per step, covering 300ms of future working conditions), and the control time domain is set to 8 time steps to ensure early response to dynamic working conditions. The instruction generation unit for the coordinated generation of graded traction force and rotation angle must satisfy the "negative feedback of traction force difference and angle deviation" logic, that is, when the actual angle... Less than the target angle At that time, increase or reduce (Prioritize adjusting the lever arm with the longer lever arm) To improve control efficiency), the coordination coefficient is set to... Ensure that the angle deviation decreases by 10%. Adjust accordingly by 5kN.

[0055] The full-risk control execution subsystem also includes a multi-objective priority quantification module, a dynamic weight calculation unit, and a multi-step control command generation unit. The multi-objective priority quantification module calculates the ratio of safety priority to traction force and rated value, accuracy priority to residual displacement and allowable value, and efficiency priority to rotation speed and maximum value. The safety priority... The calculation formula is ,in Maximum safety traction force for small pressure rods (reserved) (Safety margin to prevent small compression members from becoming unstable due to overload) Maximum rated traction force for small pressure bar; accuracy priority. The calculation formula is ,in For residual displacement ( (Target displacement). To allow residual displacement; efficiency priority. The calculation formula is ,in The maximum vertical turning angular velocity is set (to avoid excessive speed and impact). When a priority level exceeds a threshold, the dynamic weight calculation unit adjusts the corresponding target weight by a preset ratio; the preset threshold is set as follows: ( (Approaching the safety limit) ( (close to the allowable value) ( (Approaching the maximum value), the weight calculation formula is as follows: ,in (Angle deviation weighting base value) (Basic value of traction deviation weight) (Displacement deviation weighting base value) , To adjust the coefficient, and when Forced ,when Forced This ensures that high-priority objectives are met first. The multi-step control command generation unit, based on a model with adjusted weights, outputs coordinated traction and speed commands. This unit solves the MPC objective function using a quadratic programming algorithm; the specific form of the objective function is... ,in to (Control time domain) represents the traction force command to be solved. to To predict traction force (keeping the last control value in place), the solution output is... As the current control command, the command precision .

[0056] The shape switching control unit includes a shape switching prediction unit, a target shape parameter preloading unit, and a parameter transition trajectory generation unit. The shape switching prediction unit triggers switching preparation 1 second in advance when the difference between the rotation angle and the target angle is less than a set threshold. The set threshold is determined according to the shape type: vertical rotation. The threshold for translation is (Allowable range of angular deviation during positioning), translation Threshold for vertical rotation is ( For horizontal torque, For the target torque, (For maximum torque), upon triggering, an interrupt signal is sent to notify the parameter preloading unit to begin preparation. The target shape parameter preloading unit preloads the stiffness parameters of the target shape, with vertical rotation for bending stiffness and horizontal rotation for torsional stiffness; the torsional stiffness... With bending stiffness The mapping relationship is derived based on the formulas of mechanics of materials, that is ,in For the Poisson's ratio of steel, therefore During preloading, the stiffness matrix in the MPC model needs to be updated synchronously (the stiffness matrix of the vertical rotation stage). Replaced with the flat phase Loading time This ensures that model parameters are ready during the switchover. The parameter transition trajectory generation unit generates parameter transition trajectories using quadratic curves, ensuring a smooth parameter transition; the transition trajectory employs an exponential decay curve, with the formula: ,in For the current shape parameters (such as vertical rotation) ), For target shape parameters (such as translation) ), The attenuation coefficient is... The transition time (ranging from 0 to 20 ms) can be achieved using this formula, allowing the parameter to transition from... arrive Smooth transition, overshoot This avoids control command oscillations caused by sudden parameter changes.

[0057] The full-risk control execution subsystem communicates with the full-morphology prior attribution factor system to obtain the attribution results output by the full-morphology prior attribution factor system. Based on the attribution results, it generates graded traction force and rotation angle coordinated control commands and transmits the control commands to the traction execution equipment for the vertical rotation of the steel tower. The communication adopts EtherCAT real-time industrial bus, and the data transmission cycle is consistent with the MPC control cycle (20ms), with communication latency... Ensure that the attribution results (such as error source weights) are accurate. It can participate in the calculation of control commands in real time; when the attribution results show the wind-induced error weight... At this time, the control command needs to be additionally superimposed with the wind cable tension adjustment amount. ( This is a standard wind cable tension adjustment to counteract wind-induced interference; when control commands are transmitted to the traction actuator (hydraulic lever), a [further adjustment is needed]. Analog signal (corresponding to traction force 0 to (or Profinet bus digital signals; the executing device must have instruction response time) This capability ensures that control commands can be implemented quickly.

[0058] The end-to-end iterative subsystem includes a feedback closed-loop unit connecting all subsystems, an engineering data iterative unit for updating model parameters, and an output accuracy verification unit; the feedback closed-loop unit constructs "full-dimensional monitoring". Full-condition coupling optimization Full-form prior attribution Full risk control implementation Full-process iteration The complete "all-dimensional monitoring" chain has clearly defined data types for each subsystem's feedback: the all-dimensional monitoring subsystem provides historical time-series data of measured displacement, strain, and angle (sampling interval 20ms, storage duration). (hours), the deviation between the theoretical and actual feedback errors of the full-condition coupled optimization subsystem. The full-form prior factorization system feedbacks the matching deviation between the prior probability and the actual error proportion. The deviation between the feedback control command and the actual traction force in the full risk control execution subsystem All feedback data must include an engineering number, operating condition identifier, and time stamp to ensure traceability. The output accuracy verification unit uses a dual-dimensional verification method of "error statistics + robust reproducibility." Error statistics employ the root mean square error (RMSE) to calculate the accuracy change of each parameter before and after iteration, using the following formula: For the sample size, For the average residual displacement), robust reproduction is achieved by reusing historical parameter fluctuation scenarios (such as...). (fluctuations), verifying whether the error control rate of the model remains unchanged after iteration. .

[0059] The full-process iterative subsystem also includes an engineering data acquisition module, a model parameter iteration unit, and an iteration effect verification unit. The engineering data acquisition module collects data on the steel tower's weight, stiffness, wind speed traction force, and residual displacement after each vertical rotation construction or extreme working condition. The acquisition timing is set to within one hour after construction (ensuring the data is not interfered with by subsequent procedures), and under extreme working conditions (such as wind speed exceeding...). , angle super Emergency data collection must be conducted within 10 minutes of completion; the collected steel tower weight and stiffness data must include both initial values ​​and post-construction remeasurement values ​​(e.g., El_post). El_pre\cdot(1-0.02\theta_\{total} / 90)), where 0.02 is the stiffness attenuation coefficient. (Total vertical rotation angle); wind speed data must include the instantaneous maximum value and 10-minute average wind speed; traction force data must include the maximum value, minimum value, and average value during the steady-state phase; residual displacement data must include measured values ​​at three points: tower top, tower waist, and tower bottom; all data are stored in a three-level directory: "Project Number - Working Condition Type - Parameter Category," supporting SQL queries and Excel export; single project data storage capacity... .

[0060] The model parameter iteration unit updates the coupling matrix coefficients, prior model feature weights, and control model target weights based on the collected data; the coupling matrix coefficient update adopts a "bias-driven adjustment" strategy, as shown in the formula: ,in The coupling coefficients after iteration. To adjust the coefficient, For the first The actual deviation of the coupling term For allowable deviation (i.e.) ),when The coefficients are kept constant to avoid over-adjustment; the prior model feature weights are updated using incremental training, with newly collected engineering data used as supplementary samples (accounting for a certain percentage of the total sample size). The attention weights are fine-tuned using gradient descent, as shown in the formula. To fine-tune the learning rate, For loss function, For the new sample set), ensure the weight fluctuation of key features (such as rotation angle and traction force). The target weight update of the control model is based on in-situ accuracy feedback, and the formula is as follows: ,when hour, Proportional reduction (maximum reduction) ),when hour, Adjusted proportionally (maximum adjustment) Balancing accuracy with cost control.

[0061] The iteration effect verification unit confirms the completion of the iteration by checking the model through subsequent engineering processes and confirming that the output accuracy meets the requirements; the subsequent engineering checks need to cover at least one model of the same tonnage (deviation). For steel tower vertical rotation projects of the same shape (vertical rotation or vertical rotation + horizontal rotation), the inspection indicators include: residual displacement in place. Prior attribution accuracy Control command response delay Parameter fluctuation Time error control rate All indicators must be met consecutively for three construction stages (starting rotation, vertical rotation, and positioning). If any indicator fails to meet the standard, the model parameter iteration unit should be returned for readjustment (e.g., increasing the coupling matrix adjustment coefficient). The accuracy is adjusted to 0.15 until all indicators meet the requirements. After verification, an "Iteration Effect Report" is generated, which includes a comparison of parameters before and after the iteration, accuracy change curves, and suggestions for subsequent engineering adaptation, serving as the basis for system version updates.

[0062] The full-process iterative subsystem communicates with the full-dimensional monitoring subsystem, the full-condition coupled optimization subsystem, the full-morphology prior factorization system, and the full-risk control execution subsystem. This subsystem updates parameters through feedback data, ensuring long-term system adaptability. The communication uses Industrial Ethernet (Modbus-TCP protocol). Non-real-time iterative data (such as engineering acquisition data and model update parameters) is transmitted every 5 minutes, while real-time feedback data (such as error deviation and response delay) is transmitted every 1 minute. Data verification (CRC32 checksum algorithm) is used during communication to avoid data transmission errors. To ensure long-term adaptability, the system sets an "iteration cycle threshold," which forces a full-parameter iteration (rather than updating only some parameters) after every 3 projects or 1 extreme condition processing. This ensures the model can adapt to the construction characteristics of different regions (such as coastal high-salt-fog areas and northern cold regions) and different steel tower types (such as A-type towers and arch towers). Actual measurements show that the adaptation time for new projects after iteration has been shortened from 24 hours to 8 hours, and the adaptation accuracy has been improved. .

[0063] It should be noted that for the foregoing method embodiments, for the sake of simplicity, the methods are described as a series of acts. However, the applicant hereby provides that some of the acts as recited can be performed concurrently, that the order of the acts can be altered, and that additional or fewer acts can be employed. And, it is to be understood that the acts recited can be implemented by specialized hardware components, by software, or by a combination of specialized hardware components and software.

[0064] The full-dimension monitoring subsystem can capture displacement and stress differences at different heights by means of the technical means of "layered point coverage of key positions on the tower top, tower waist and tower bottom", avoid the limitations of single-point monitoring that cannot reflect the overall mechanical state of the steel tower, and directly derive the effect that "monitoring data can fully represent the spatial mechanical distribution during the vertical rotation of the steel tower". Combined with "main and backup sensor redundancy configuration (such as GNSS and laser ranging, encoder and inclination sensor) + fault data repair logic (displacement calculation based on geometric similarity principle, inverse calculation of cable force based on stress-force relationship)", data can be quickly completed when any sensor fails, and the effect that "there is no data interruption during the monitoring process, ensuring continuous input for subsequent coupling optimization and control execution" is further derived. Through "spatiotemporal coordinated sampling (high-weight sampling of high-weight points (tower top) and high-frequency sampling of key stages (erection stage))", high-density data can be obtained at key positions and stages, and redundant data can be reduced in non-key scenarios, and the effect that "balance monitoring accuracy and data processing efficiency, and avoid invalid data occupying computing resources" is derived.

[0065] The full-condition coupling optimization subsystem can quantify the interaction of each error source (wind-induced, temperature, traction, stiffness attenuation, etc.) based on the "10x10 multi-dimensional coupling matrix (including spatial height, time stiffness, parameter fluctuation, shape switching, and extreme feature coupling items)", avoid the deviation of traditional single-error calculation that ignores coupling effects, and directly derive the effect that "the coupling optimization result can truly reflect the error composition of the engineering practice". By adding a "robust objective function (including a parameter fluctuation penalty term)", the error range can be constrained when the elastic modulus, viscosity coefficient and other key parameters fluctuate, and the effect that "the optimization parameters have anti-interference ability for engineering uncertainty" is derived. In combination with "Monte Carlo simulation verification (1000 random samplings)", the accuracy compliance rate of parameters in multiple scenarios can be verified in advance, and the effect that "the engineering applicability and reliability of the coupling optimization result are improved" is further derived.

[0066] The full-form prior attribution factor system covers vertical rotation, horizontal rotation, and conventional / extreme working conditions through a "multi-dimensional feature vector (steel tower parameters, rotating body form, construction stage, environmental conditions, extreme indicators)", avoids the limitations of traditional prior model scene adaptation, and directly derives the effect that "the prior model can adapt to the full working condition requirements of the steel tower vertical rotation"; combined with "dynamic attention weight (focusing on rotating angle, hierarchical traction force, and other key features)", it can reduce the influence of non-key features (such as horizontal rotation torque interference in the vertical rotation stage) on prior probability, and derive the effect that "the matching degree between prior probability and actual error distribution is improved, and the attribution accuracy is enhanced"; and through "extreme feature migration (extreme parameter and conventional parameter amplification factor κ mapping)", the conventional prior model can be reused in extreme working conditions (such as wind speed exceeding 25 m / s and angle exceeding 90°) within the super-training range, further deriving the effect that "the prior attribution does not fail in extreme working conditions, and the system scene adaptation boundary is expanded".

[0067] The full-risk control execution subsystem can predict future working condition changes at multiple time steps in advance based on a "multi-step predictive control (MPC) model (including steel tower vertical rotation dynamics equation)", avoiding the lag of traditional static control, and directly deriving the effect that "the control instruction can respond to the dynamic changes of the working condition in advance, reducing the accumulation of angle and displacement deviations"; adding "multi-objective dynamic weight (safety-related traction force, precision-related residual displacement, and efficiency-related rotating speed)", it can prioritize core needs at different stages (prioritize safety at the start-up stage and prioritize precision at the just-in-time stage), and derive the effect that "avoiding the trade-off caused by single target orientation, balancing construction safety, precision, and efficiency"; in combination with "form switching pre-transition (loading target form parameters in advance + quadratic curve transition trajectory)", it can avoid control shock caused by parameter mutation, and further derive the effect that "there is no overshoot in error during form switching such as vertical rotation-horizontal rotation, and the rotating process is smooth".

[0068] The full-process iteration subsystem can collect deviation data (such as coupling error, attribution deviation, and control response delay) from each module in real time through a "multi-subsystem feedback closed loop (monitoring → coupling → prior → control → iteration)", avoiding the accumulation of deviations caused by the lack of system feedback, and directly deriving the effect that "adaptation problems in each module are discovered and corrected in a timely manner"; combined with "engineering data-driven parameter iteration (updating coupling matrix coefficients, prior feature weights, and control target weights)", the system can be continuously optimized as engineering experience accumulates, and derive the effect that "the long-term adaptability of the system to steel tower vertical rotation of different tonnages and different regions is improved, and frequent manual calibration is not required".

[0069] In summary, starting from the full-link technical approach of "monitoring-coupling-priority-control-iteration", it can be deduced step by step that: during the vertical rotation of the steel tower, the monitoring data is continuous and reliable, the error calculation is accurate and robust, the error attribution is accurate and adaptable to all working conditions, the control commands are coordinated and dynamically responsive, and the system has strong long-term adaptability. Ultimately, the overall technical effect of "safe and controllable construction of ultra-large tonnage steel towers, meeting positioning accuracy standards, and stable operation in all scenarios" can be achieved.

[0070] Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by specific combinations of the above-mentioned technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-mentioned technical features or their equivalent features without departing from the aforementioned disclosed concept. For example, technical solutions formed by substituting the above-mentioned features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A super-tonnage steel tower vertical rotation whole process monitoring and intelligent control system, characterized in that, The system comprises a full-dimension monitoring subsystem, a full-condition coupling optimization subsystem, a full-form prior attribution subsystem, a full-risk control execution subsystem, and a full-process iteration subsystem, which are sequentially connected by signals; The full-dimension monitoring subsystem comprises layered measuring points for collecting displacement and stress at the top, waist, and bottom of the steel tower, axis type sensors for identifying the combined axis of vertical and horizontal rotation, and an environmental data acquisition unit for obtaining wind speed and temperature; The full-condition coupling optimization subsystem comprises a coupling matrix including space height, time stiffness, and coupling terms, an optimization objective function integrating coupling constraints, and a weighted iterative algorithm based on the space-time characteristics of vertical rotation; The full-form prior attribution subsystem comprises a multi-dimensional feature vector including the weight and height parameters of the steel tower, a prior model integrating feature importance, and a Bayesian attribution algorithm; The full-risk control execution subsystem comprises a multi-step predictive control model, an instruction generation unit for coordinating the traction force and rotation angle, and a form switching control unit; The full-process iteration subsystem comprises a feedback loop unit connecting the subsystems, an engineering data iteration unit for updating model parameters, and an output accuracy verification unit.

2. The system of claim 1, wherein, The full-dimension monitoring subsystem further comprises a sensing redundancy network, a fault detection and diagnosis module, and a fault data repair module; In the sensing redundancy network, the main displacement monitoring sensor is GNSS, the backup is a laser range finder, the main angle monitoring sensor is an encoder, the backup is a dual-axis tilt sensor, and the main cable force monitoring sensor is a tension and compression force sensor, the backup is a cable force acquisition unit based on tower body strain calculation; The fault detection and diagnosis module determines faults by comparing the main and backup data deviation with a preset threshold, and identifies fault types based on data jumps or gradual changes; The fault data repair module calculates displacement through geometric relationships and inversely calculates cable force through mechanical formulas to repair parameters collected by faulty sensors.

3. The system of claim 2, wherein, It also includes a fault control connection module, which comprises a data smoothing transition unit and a control instruction iteration unit; The data smoothing transition unit processes the displacement and angle parameters after fault repair using an exponential function to ensure smooth transition with historical monitoring data; The control instruction iteration unit re-solves the multi-step predictive control model based on the repaired displacement deviation data to generate adaptive hierarchical traction force instructions.

4. The system of claim 1, wherein, The full-condition coupling optimization subsystem further comprises a parameter fluctuation quantification module, a robust objective function optimization unit, and a simulation verification module; The parameter fluctuation quantification module describes the fluctuation range of the elastic modulus and viscosity coefficient through interval analysis, and describes the monitoring noise of displacement and strain through probability distribution; The robust objective function optimization unit adds a fluctuation penalty term to the objective function to ensure that the optimization parameters meet the accuracy standard within the fluctuation range; The simulation verification module checks the optimization parameters through random sampling simulation, and the sampling proportion that meets the accuracy requirement is not less than the set threshold.

5. The system of claim 1, wherein, The full-form prior attribution subsystem further comprises a dynamic attention weight calculation module, a multi-form feature adaptation unit, and a prior probability correction module; The dynamic attention weight calculation module calculates feature weights through a scoring model and gives higher weights to the rotation angle and hierarchical traction force; The multi-form feature adaptation unit constructs a mapping relationship between the vertical rotation angle and the steel tower height, and between the horizontal rotation torque and the rotation radius. The prior probability correction module adjusts the prior probability based on the attention weight, so that the matching degree with the actual error distribution meets the preset condition.

6. The system of claim 5, wherein, It also includes an extreme working condition feature migration module, which includes an extreme feature recognition unit, an extreme feature migration library, and an extreme prior correction unit. The extreme feature recognition unit determines the extreme working condition based on wind speed exceeding 25 m / s or rotation angle exceeding 90°; The extreme feature migration library stores the mapping of extreme and normal features, and determines the feature amplification coefficient through the parameter ratio of the two; The extreme prior correction unit adjusts the attention weight and the prior probability based on the amplification coefficient to adapt to the extreme working condition.

7. The system of claim 1, wherein, The full-risk control execution subsystem also includes a multi-objective priority quantization module, a dynamic weight calculation unit, and a multi-step control instruction generation unit. The multi-objective priority quantization module calculates the ratio of safety priority associated traction force to rated value, the ratio of precision priority associated residual displacement to allowed value, and the ratio of efficiency priority associated rotation speed to maximum value. The dynamic weight calculation unit adjusts the weight of the corresponding target by a preset proportion when a certain priority exceeds a threshold value. The multi-step control instruction generation unit outputs traction force and speed coordination instructions based on the model of the adjusted weight.

8. The system of claim 1, wherein, The mode switching control unit includes a mode switching pre-judgment unit, a target mode parameter pre-loading unit, and a parameter transition trajectory generation unit. The mode switching pre-judgment unit triggers the switching preparation 1 second in advance when the difference between the rotation angle and the target angle is less than a set threshold value. The target mode parameter pre-loading unit loads the stiffness parameters of the target mode in advance, with vertical stiffness for vertical rotation and torsional stiffness for horizontal rotation. The parameter transition trajectory generation unit generates a parameter transition trajectory through a quadratic curve to ensure smooth parameter transition.

9. The system of claim 1, wherein, The full-dimensional monitoring subsystem also includes a space-time coordinated sampling module, which includes a spatial hierarchical sampling weight unit, a time dynamic sampling rate unit, and a space-time sampling coordination unit. The spatial hierarchical sampling weight unit assigns a sampling weight of 0.4 to the tower top, 0.3 to the tower waist, and 0.3 to the tower bottom. The time dynamic sampling rate unit adjusts the sampling rate when the rotation angle changes by 5 degrees, and increases the sampling rate when approaching the target angle. The space-time sampling coordination unit synchronously increases the sampling rate of the tower top measurement points when the weight is high.

10. The system of claim 1, wherein, The full-process iteration subsystem also includes an engineering data acquisition module, a model parameter iteration unit, and an iteration effect verification unit. The engineering data acquisition module acquires steel tower weight stiffness, wind speed traction force, and residual displacement data after each vertical rotation construction or extreme working condition. The model parameter iteration unit updates the coupling matrix coefficient, the prior model feature weight, and the control model target weight based on the collected data. The iteration effect verification unit checks the model through subsequent engineering and confirms that the iteration is complete when the precision meets the requirements.

Images