A precision automatic compensation control method in steel structure assembly process
By acquiring and processing various parameters during the steel structure assembly process, constructing the correspondence between design and measurement coordinates, and combining temperature and stress corrections, the actuator displacement increment is generated and limited, thus solving the control deviation problems caused by reference drift, thermal expansion and contraction, and stress rebound in steel structure assembly, and improving assembly accuracy and stability.
Patent Information
- Application Number
- CN202611025203.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-10
- Publication Date
- 2026-11-10
AI Technical Summary
During the steel structure assembly process, existing technologies are unable to accurately handle the assembly compensation control state under the combined effects of reference drift, thermal expansion and contraction error, stress rebound and actuator boundary constraints, resulting in the control output deviating from the actual assembly state.
By acquiring parameters such as the coordinates of the steel structure assembly control points, the coordinates of the reference target, the component temperature, the actuator position, and the equivalent force, the correspondence between the measured coordinates and the design coordinates is constructed. Combined with temperature correction and the influence of force rebound, the actuator displacement increment is generated, and the amplitude is limited to output the three-state locking control state.
It achieves precise assembly compensation control under the influence of factors such as reference drift, thermal expansion and contraction and force rebound, improves the accuracy and consistency of steel structure assembly and the stability of control output, and avoids control inaccuracy caused by deviation of a single measuring point or position of a single actuator.
Smart Images

Figure CN123067097A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of steel structure assembly control technology, specifically to an automatic accuracy compensation control method in the steel structure assembly process. Background Technology
[0002] During the on-site assembly of steel structure components, it is usually necessary to measure the coordinates of control points, adjust the position of temporary supports or actuators, and lock them after the deviation meets the requirements. The existing processing methods mostly rely on the direct comparison between the measurement results and the design coordinates, and then the actuator displacement is adjusted manually or by the control system according to the magnitude of the deviation.
[0003] However, in large steel structure assembly scenarios, the measurement benchmark may be affected by slight movement of the instrument support, overall deformation of the jig, or local offset of the benchmark target. If the on-site measurement coordinates are used directly for deviation judgment, the coordinate system drift is easily mixed into the actual assembly deviation of the component. At the same time, the temperature change of the component will cause reversible expansion and contraction along the main direction of the component. If the thermal expansion and contraction component and the assembly deviation are not distinguished, the control output may deviate from the actual assembly state.
[0004] Furthermore, the actuator exerts a directional force during pushing, pulling, or holding processes. The difference between the current force and the target holding force will manifest as release rebound during subsequent unloading, locking, or force redistribution. If the existing control method only outputs displacement commands based on geometric deviations without simultaneously considering the actuator's direct displacement response, the influence of force rebound, stroke boundaries, force boundaries, and single-cycle displacement boundaries, it is easy to cause a lack of consistent state basis between compensation actions and locking determination.
[0005] Therefore, there is an urgent need for an automatic accuracy compensation control method in the steel structure assembly process to solve the problem that it is difficult to accurately determine the assembly compensation control state under the combined effects of reference drift, thermal expansion and contraction error, force rebound and actuator boundary constraints in the existing technology. Summary of the Invention
[0006] This invention provides an automatic accuracy compensation control method for the assembly process of steel structures, which helps to solve the problems mentioned in the background art.
[0007] This invention provides the following technical solution: an automatic accuracy compensation control method for steel structure assembly process, comprising:
[0008] Obtain the coordinates of the steel structure assembly control points, the coordinates of the reference target, the component temperature, the actuator position, the equivalent force of the actuator with sign, and the system preset parameters;
[0009] The validity of the reference is determined based on the reference target coordinates. When the reference is valid, the correspondence between the measured coordinates and the design coordinates is constructed. When the reference is invalid, the actuator position is maintained and the lock-free and manual verification status is output.
[0010] Unify the coordinates of the control points to the design assembly coordinate system, and obtain temperature-corrected coordinates by combining the component temperature and thermal expansion parameters.
[0011] The temperature correction coordinates are compared with the nominal coordinates of the control points to obtain the actual assembly deviation;
[0012] Based on the difference between the current signed equivalent force of the actuator and the signed target holding force, and combined with the force rebound influence data, a release rebound prediction is generated;
[0013] The actual assembly deviation and the predicted amount of springback are combined into an effective compensation error, and an equivalent influence matrix including the direct displacement response and the springback change response is constructed.
[0014] The actuator displacement increment is generated based on the effective compensation error and the equivalent influence matrix, and the amplitude is limited by combining the actuator stroke, force boundary and single-cycle displacement boundary;
[0015] When the reference is valid, the actuator position command is output according to the displacement increment after the limit, the predicted force and predicted physical residual after execution are updated to output the assembly risk index, and the three-state locking control state is output in combination with the remaining mechanical executable margin and the executable action scale of this cycle.
[0016] Optionally, obtaining the coordinates of the steel structure assembly control points, the coordinates of the reference target, the component temperature, the actuator position, the equivalent force of the actuator with signs, and the system preset parameters specifically includes:
[0017] During each control cycle, the three-dimensional coordinates of each control point, the three-dimensional coordinates of three non-collinear reference targets, the temperature of components adjacent to the control point, the current position of each actuator, and the signed equivalent force of each actuator are collected.
[0018] Read the nominal coordinates of the control point, the nominal coordinates of the reference target, the direction of thermal expansion, the equivalent length of thermal expansion, the coefficient of linear expansion of the steel, the reference temperature, the allowable direction of assembly, the allowable deviation along the assembly direction, and the allowable deviation perpendicular to the assembly direction;
[0019] Read the actuator's signed target holding force, maximum allowable force, equivalent stiffness, upper and lower limits of stroke, maximum displacement in a single cycle, displacement response calibration data, data on the influence of force rebound, measurement resolution, force sensor resolution, and displacement sensor resolution;
[0020] The positive direction of the actuator's signed equivalent force is set to be consistent with the direction of the force increase caused by the actuator's positive displacement. Non-negative pressure data is converted into the actuator's signed equivalent force according to the pressure difference, effective pressure area, and direction of action. Torque data is converted into the actuator's signed equivalent force according to the torque direction and transmission conversion relationship.
[0021] Optionally, the step of determining the validity of the reference based on the reference target coordinates and constructing the correspondence between the measured coordinates and the design coordinates when the reference is valid specifically includes:
[0022] Using the nominal reference target and the current measurement reference target as processing objects respectively, the same reference point is selected as the starting point, and the first reference edge vector and the second reference edge vector are constructed pointing to the other two reference points respectively. The cross product operation is performed on the two reference edge vectors and the magnitude of the cross product result is extracted. The magnitude of the first reference edge vector is combined with the measurement resolution to form the stabilized reference length. Then, the magnitude of the cross product result is compared with the stabilized reference length to obtain the non-collinear scale representing the degree of non-collinearity of the two reference edges.
[0023] The first scale value and the second scale value are calculated by comparing the length of the first reference side with the non-collinear scale. The smaller of the first scale value and the second scale value are then selected as the usable reference scale.
[0024] Repeat the above calculation process for the nominal datum and the current measurement datum respectively to obtain the available scales for the nominal datum and the measurement datum.
[0025] The nominal reference is compared with the measurement resolution. If the nominal reference meets or exceeds the measurement resolution, it is determined that the nominal reference meets the validity condition; otherwise, it is determined to be invalid.
[0026] The measurement reference is compared with the measurement resolution. If the comparison reaches or exceeds the measurement resolution, the measurement reference is deemed to meet the valid condition; otherwise, it is deemed invalid.
[0027] When both the nominal datum and the measurement datum meet the validity conditions, the datum validity flag is set to valid; otherwise, it is set to invalid.
[0028] When the reference is valid, the first reference edge vector is normalized to obtain the first unit direction, the cross product of the two reference edge vectors is normalized to obtain the third unit direction, and the second unit direction is obtained based on the cross product relationship between the first unit direction and the third unit direction.
[0029] The first unit direction, the second unit direction, and the third unit direction are combined into an orthogonal basis matrix to form the design coordinate orthogonal basis and the measurement coordinate orthogonal basis, respectively. The correspondence between the measurement coordinates and the design coordinates is formed by using the two sets of orthogonal bases and the geometric center of the reference target.
[0030] Optionally, unifying the control point coordinates to the design assembly coordinate system and obtaining temperature-corrected coordinates by combining component temperature and thermal expansion parameters specifically includes:
[0031] Based on the correspondence between measured coordinates and design coordinates, the control point coordinates are mapped to the design assembly coordinate system to obtain unified control point coordinates;
[0032] Subtract the reference temperature from the component temperature to obtain the temperature difference.
[0033] The temperature difference is multiplied by the linear expansion parameter of the steel and the equivalent length of thermal expansion, and the result is distributed along the direction of thermal expansion to obtain the thermal expansion correction amount for the corresponding control point.
[0034] The temperature-corrected coordinates are obtained by subtracting the corresponding thermal expansion correction from the unified control point coordinates.
[0035] Optionally, comparing the temperature correction coordinates with the nominal coordinates of the control points to obtain the actual assembly deviation specifically includes:
[0036] The temperature correction coordinates of each control point are matched one by one with the nominal coordinates of the corresponding control points;
[0037] Perform difference processing on the temperature correction coordinates and nominal coordinates of each control point in the corresponding direction to obtain the position offset of the control point in the design assembly coordinate system;
[0038] The positional offset of each control point in the design assembly coordinate system is taken as the actual assembly deviation, and the actual assembly deviation is used as the input for generating effective compensation error and predicting physical residual generation.
[0039] Optionally, the step of generating a release rebound prediction based on the difference between the current signed equivalent force of the actuator and the signed target holding force, combined with the force rebound influence data, specifically includes:
[0040] The current signed equivalent force of each actuator is subtracted from the corresponding signed target holding force one by one to obtain the force deviation of the corresponding actuator.
[0041] The force deviation of each actuator is mapped to the corresponding force rebound influence data to obtain the rebound displacement contribution of each actuator to each control point.
[0042] The springback displacement contribution of each actuator corresponding to the same control point is accumulated to obtain the predicted release springback amount of that control point.
[0043] Optionally, the step of combining the actual assembly deviation and the predicted springback amount into an effective compensation error, and constructing an equivalent influence matrix including the direct displacement response and the springback change response, specifically includes:
[0044] The actual assembly deviation at each control point is superimposed with the predicted release and rebound at the corresponding control point to generate the effective compensation error for each control point. All effective compensation errors are then combined into an effective compensation error vector in the order of control points.
[0045] Construct the projection relationship along the assembly direction according to the assembly allowable direction, and construct the projection relationship perpendicular to the assembly direction according to the complete three-dimensional space reference relationship and the projection relationship along the assembly direction.
[0046] Using the allowable deviation along the assembly direction and the allowable deviation perpendicular to the assembly direction as scale boundaries, the projection relationship along the assembly direction and the projection relationship perpendicular to the assembly direction are scaled and unified to form a local direction normalization matrix.
[0047] The equivalent stiffness of the actuator is combined with the force rebound influence data corresponding to each control point to form the rebound change response column data corresponding to each actuator, and the rebound change response column data is combined into a rebound change response matrix according to the actuator order.
[0048] The rebound change response matrix is superimposed with the displacement response calibration data to generate an equivalent influence matrix that includes the direct displacement response and the rebound change response. This equivalent influence matrix is then used as the input for generating the actuator displacement increment.
[0049] Optionally, the step of generating the actuator displacement increment based on the effective compensation error and equivalent influence matrix, and limiting the amplitude in combination with the actuator stroke, force boundary, and single-cycle displacement boundary, specifically includes:
[0050] Using the combined effective compensation error vector as input, the effective compensation error vector is subjected to scale unification processing using the local direction normalization matrix corresponding to each control point to obtain the normalized effective compensation error vector.
[0051] The local orientation normalization matrix corresponding to each control point is arranged sequentially according to the control point order. The block-diagonal form of the overall orientation normalization data is filled in block by block. The overall orientation normalization data is combined with the equivalent influence matrix. At the same time, the combination result is processed according to the single-cycle displacement scale of the actuator to obtain the normalized actuator equivalent influence matrix.
[0052] Based on the proportional relationship between the measurement resolution and the allowable deviation along the assembly direction and the allowable deviation perpendicular to the assembly direction, the resolution ratio along the assembly direction and the resolution ratio perpendicular to the assembly direction are obtained respectively. The resolution ratios are squared and averaged according to the number of control points and the number of directional degrees of freedom to generate a regularization quantity.
[0053] The normalized actuator equivalent influence matrix, regularization quantity, and normalized effective compensation error vector are combined with the closed matrix solution process of the regularization quantity to obtain the original displacement increment of each actuator.
[0054] Based on the maximum allowable force, the signed equivalent force of the actuator, the equivalent stiffness of the actuator, the resolution of the force sensor, and the resolution of the displacement sensor, calculate the force margin for the positive displacement and the force margin for the negative displacement, respectively.
[0055] Calculate the forward position margin and the reverse position margin based on the current position of the actuator and the travel boundary.
[0056] The maximum displacement in a single cycle, the force margin of the positive displacement, and the position margin of the positive position are selected by minimizing the values to obtain the upper limit of the available positive displacement. The maximum displacement in a single cycle, the force margin of the negative displacement, and the position margin of the negative position are selected by minimizing the values to obtain the upper limit of the available negative displacement.
[0057] The original displacement increment of each actuator is limited to the upper limit of the reverse available displacement and the upper limit of the forward available displacement to generate the limited displacement increment.
[0058] Optionally, the step of outputting the actuator position command based on the displacement increment after limiting, updating the predicted force and predicted physical residual after execution to output the assembly risk index, and combining the remaining mechanical executable margin and the executable action scale of this cycle to output the three-state locking control state, specifically includes:
[0059] When the reference validity flag is valid, the current position of the actuator is added to the displacement increment after the limit is applied to generate the actuator position command;
[0060] When the reference validity flag is valid, the signed equivalent force of the actuator is superimposed with the force change corresponding to the equivalent stiffness of the actuator and the displacement increment after the limit, to obtain the predicted force after execution.
[0061] When the benchmark validity flag is valid, the predicted force after execution is subtracted from the marked target holding force one by one to obtain the force deviation of each actuator after execution;
[0062] The force deviation after execution of each actuator is multiplied and mapped with the corresponding force rebound influence data to obtain the rebound displacement contribution of each actuator to each control point.
[0063] The rebound displacement contributions of all the same control point are accumulated to generate the predicted amount of rebound after execution.
[0064] When the reference validity flag is valid, the actuator direct displacement response is generated based on the displacement response calibration data and the displacement increment after the limit. The actual assembly deviation, the predicted amount of springback after execution, and the actuator direct displacement response are combined and superimposed to generate the predicted physical residual.
[0065] When the benchmark validity flag is valid, the predicted physical residual is decomposed according to the assembly allowable direction to obtain the predicted residual along the assembly direction and the predicted residual perpendicular to the assembly direction. Based on the allowable deviation along the assembly direction, the allowable deviation perpendicular to the assembly direction and the maximum allowable force, the predicted residual and the predicted force after execution are scaled uniformly to generate the assembly risk index.
[0066] When the benchmark validity flag is valid, multiple normalized occupancy values are calculated based on the ratio of the predicted residual along the assembly direction to the allowable deviation along the assembly direction, the ratio of the predicted residual perpendicular to the assembly direction to the allowable deviation perpendicular to the assembly direction, and the ratio of the predicted force after execution to the maximum allowable force. The multiple normalized occupancy values are compared item by item, and the maximum occupancy value is selected as the comprehensive boundary occupancy index.
[0067] When the reference validity flag is valid, the remaining force margin is calculated based on the difference between the predicted force after execution and the maximum allowable force. The ratio of the force sensor resolution to the displacement sensor resolution is processed to obtain the resolution compensation stiffness. The resolution compensation stiffness and the equivalent stiffness of the actuator are combined into the stabilized stiffness. Then, the remaining force margin is converted into the corresponding executable displacement margin based on the stabilized stiffness.
[0068] Calculate the remaining position margin based on the distance between the actuator position command and the actuator travel boundary;
[0069] The minimum values of the executable displacement margin, position residual margin, and maximum displacement in a single cycle corresponding to the force are selected to form the positive residual displacement margin and the reverse residual displacement margin, respectively.
[0070] The forward and reverse residual displacement margins of the same actuator are merged, and the maximum merged margin is selected among all actuators to generate the remaining mechanically executable margin.
[0071] When the reference validity flag is valid, the ratio of the displacement increment after the amplitude limit to the resolution of the corresponding displacement sensor is calculated to obtain the motion resolution multiple of each actuator, and the maximum value of the motion resolution multiple of all actuators is selected to generate the executable motion scale for this cycle.
[0072] When the reference validity flag is invalid, the actuator maintains its current position and outputs a lock-lock prohibited and manual verification status.
[0073] When the baseline validity flag is valid and the comprehensive boundary occupancy index does not exceed the normalized boundary, the lock-in state is allowed.
[0074] When the baseline valid flag is valid, the comprehensive boundary occupancy index exceeds the normalized boundary, the remaining mechanical executable margin reaches the displacement sensor resolution, and the executable action scale of this cycle reaches the action scale corresponding to the displacement sensor resolution, the output continues to compensate.
[0075] When the baseline validity flag is valid, the comprehensive boundary occupancy index exceeds the normalized boundary, and the remaining mechanical executable margin does not reach the displacement sensor resolution or the executable action scale in this cycle does not reach the action scale corresponding to the displacement sensor resolution, the output will prohibit locking and manual verification status.
[0076] The present invention has the following beneficial effects:
[0077] 1. By synchronously acquiring control point coordinates, reference target coordinates, component temperature, actuator position, actuator force state, and system preset parameters in each control cycle, and converting the pressure, tension, or torque data of different types of actuators into signed equivalent forces under a unified direction, the field geometry, thermal state, force state, and actuator boundary conditions are all incorporated into subsequent processing within the same control cycle. This avoids subsequent steps relying solely on single measurement point deviations or single actuator positions for control judgments. At the same time, parameter validity verification excludes control cycles lacking physical boundary conditions, providing a consistent input basis for reference validity judgment, temperature correction, springback prediction, displacement calculation, and three-state locking output.
[0078] 2. By constructing reference side vectors for the nominal reference target and the current measurement reference target respectively, extracting non-collinear scales, and forming usable reference scales, the measurement reference is subject to the common constraints of side length scales and spatial non-collinearity before entering coordinate unification. This enables the identification of cases where the reference is degraded, the reference measurement is abnormal, or the reference target layout does not meet the requirements for constructing an orthogonal basis. When the reference is valid, the correspondence between the measurement coordinates and the design coordinates is further established using the unit direction, the orthogonal basis matrix, and the reference geometric center, so that the subsequent control point coordinate transformation is based on the valid spatial reference. When the reference is invalid, the current position of the actuator is directly maintained and the manual verification status is output to prevent abnormal references from continuing to be transmitted to the compensation control chain.
[0079] 3. By calling the design coordinate orthogonal base, measurement coordinate orthogonal base, and geometric centers of the two types of references formed when the reference is valid, the coordinates of the control points measured on site are mapped to the design assembly coordinate system, so that the subsequent deviation calculation no longer directly depends on the spatial position under the original measurement coordinate system; at the same time, by combining the component temperature, reference temperature, coefficient of linear expansion of steel, equivalent length of thermal expansion and contraction, and direction of thermal expansion and contraction, the thermal expansion correction amount corresponding to the temperature difference is extracted, and the correction amount is deducted from the unified control point coordinates, so that the reversible thermal expansion component caused by environmental temperature difference, welding residual heat or sunlight is not mistakenly included as assembly deviation in the subsequent compensation solution.
[0080] 4. By matching the temperature-corrected coordinates of each control point with their corresponding nominal coordinates one by one, and extracting the position offset in the corresponding direction of the design assembly coordinate system, the actual assembly deviation is formed after the reference drift processing and thermal expansion peeling, which can characterize the actual assembly deviation state of the control point relative to the design target. This actual assembly deviation serves as both the geometric basis for effectively compensating for error generation and as the pre-input for generating predicted physical residuals after execution. This ensures that subsequent springback prediction, equivalent influence matrix solution, and locking determination all revolve around the actual deviation after coordinate unification and temperature correction, avoiding repeated compensation of previous reversible errors.
[0081] 5. By comparing the current signed equivalent force of each actuator with the signed target holding force, the force deviation of the actuator is extracted. Based on the force rebound influence data, the force deviation is converted into the rebound displacement contribution at each control point, so that the release rebound caused by unloading, locking or force redistribution during the steel structure assembly process is characterized in advance before the control command is generated. The release rebound prediction quantity is formed by accumulating the rebound contributions of multiple actuators at the same control point, so that the subsequent effective compensation error not only reflects the current geometric position deviation, but also reflects the expected influence of the current force state on the position of the control point after locking, so that the displacement solution is closer to the physical state after execution.
[0082] 6. By superimposing the actual assembly deviation with the predicted springback amount to form an effective compensation error, and forming a local direction normalization matrix based on the assembly allowable direction, perpendicular assembly direction, and corresponding tolerance boundary, deviations at different control points, in different directions, and at different tolerance levels can be solved for compensation on a unified scale. At the same time, the equivalent stiffness of the actuator and the force-induced springback influence data are combined into a springback change response matrix, and superimposed with the displacement response calibration data to form an equivalent influence matrix. This allows the direct geometric response of the actuator displacement to the control point and the subsequent springback response caused by force changes to be simultaneously incorporated into the control model, avoiding residual shifts after execution caused by generating compensation amounts solely based on displacement response calibration.
[0083] 7. By using the local direction normalization matrix to form global direction normalization data, and by performing scale unification and column-scale processing on the effective compensation error and equivalent influence matrix respectively, the error vector and actuator influence relationship before closed-loop solution are in a matching direction scale; further, a regularization quantity is generated based on the measurement resolution and assembly tolerance, and the error projection term is processed in the direction of canceling the effective compensation error to obtain the original displacement increment; then, the force margin, position margin, single-cycle maximum displacement and non-negative lower limit are combined to form positive and negative usable displacement boundaries, so that the final limited displacement increment simultaneously meets the error compensation requirements and the field actuator executable constraints;
[0084] 8. By converting the displacement increment after limiting into the actuator position command when the reference is valid, and synchronously updating the predicted force after execution, the predicted amount of release springback after execution, the actuator direct displacement response, and the predicted physical residual, the controller can obtain the residual prediction corresponding to the actual execution state after outputting the command. Then, based on the predicted physical residual and the predicted force after execution, the assembly risk index, the comprehensive boundary occupancy index, the remaining mechanical executable margin, and the executable action scale of this cycle are generated respectively. This makes the three-state output of allowing locking, continuing compensation, and prohibiting locking and manually verifying it jointly constrained by residual occupancy, force boundary, stroke margin, and action resolution capability, so that the locking judgment and compensation capability are consistent.
[0085] 9. This solution addresses the control scenario in steel structure assembly where datum drift, thermal expansion and contraction, actuator force rebound, and mechanical boundary conditions act simultaneously, making it prone to inaccuracies when relying solely on measurement point deviations to output compensation commands. It integrates datum validity, coordinate unification, temperature correction, force rebound prediction, equivalent influence matrix, displacement limiting, and three-state locking determination into a single control chain. Specifically, the datum target first determines whether the measured coordinates can enter the normal compensation path, preventing subsequent deviations from being mixed into the measurement coordinate system drift. Temperature correction further separates reversible thermal expansion and contraction from control point offsets, making the actual assembly deviation the basis for compensation solutions. Force rebound prediction transforms the current holding force state of the actuator into the future release displacement of the control point, ensuring the control target... The static geometric alignment is transformed into effective compensation error considering the residual after locking. The equivalent influence matrix further incorporates the actuator's direct displacement response and springback change response into the displacement increment solution, avoiding secondary offset after force release caused by separate geometric deviation compensation. The amplitude limiting process uses stroke, force, and single-cycle action boundary to jointly constrain the compensation amount, making the control command executable on site. After execution, the predicted residual, assembly risk index, remaining mechanical executable margin, and action scale jointly determine the locking state. Thus, the assembly state identification, compensation control, and locking determination are progressively generated based on the same set of physical constraints, improving the accuracy consistency, risk judgment reliability, and control output stability of the steel structure assembly compensation process. Attached Figure Description
[0086] Figure 1 This is the main flowchart of the automatic compensation control method for steel structure assembly accuracy of the present invention.
[0087] Figure 2 This is a schematic diagram of the physical scenario of an automatic compensation control system for steel structure assembly accuracy.
[0088] Figure 3 This is a schematic diagram illustrating the spatial relationship between the reference target and the coordinate system of this invention.
[0089] Figure 4 This is a schematic diagram illustrating the normalization determination of the predicted physical residual and assembly tolerance boundary in this invention.
[0090] Figure 5 This is a schematic diagram of the multi-boundary constraint limiting principle for the displacement increment of the actuator in this invention. Detailed Implementation
[0091] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0092] Example: Figure 1 As shown, an automatic accuracy compensation control method for steel structure assembly process is applied to control point measurement, temperature error correction, force rebound compensation, actuator displacement solution, boundary limiting and three-state locking control of steel structure components in the assembly station. The method includes the following steps one to eight.
[0093] like Figure 2 As shown, step one: obtain the coordinates of the steel structure assembly control points, the coordinates of the reference target, the component temperature, the actuator position, the equivalent force of the actuator with the sign, and the system preset parameters;
[0094] The actuator is a drive device used to apply displacement or force to steel structure components to achieve position adjustment, including but not limited to hydraulic jacks, electric push rods or lead screw drive mechanisms. It can output controllable displacement and force under the action of control commands, and feed back the current position and force status through sensors.
[0095] During each control cycle, the three-dimensional coordinates of the steel structure assembly control point, the three-dimensional coordinates of the three non-collinear reference targets, the temperature of the components adjacent to the control point, the current position of each actuator, and the force data of each actuator are collected.
[0096] The three-dimensional coordinates of the control points are the spatial position data of each control point in the measurement coordinate system under the current assembly state, which will be used for coordinate unification, temperature correction and calculation of actual assembly deviation.
[0097] The three-dimensional coordinates of the reference target include the current measurement coordinates of three non-collinear reference targets, which are subsequently used to determine whether the current measurement reference is valid, and to construct the correspondence between the measurement coordinates and the design coordinates when the reference is valid;
[0098] The component temperature is the current temperature of the area adjacent to each control point, and is subsequently used together with the reference temperature, steel linear expansion parameters, thermal expansion equivalent length and thermal expansion direction to calculate the thermal expansion correction amount.
[0099] The actuator position is the current position of each actuator at the beginning of the current control cycle, which is subsequently used to generate actuator position commands, calculate position margin and remaining mechanically executable margin;
[0100] The actuator force data is converted into the actuator's signed equivalent force according to the actuator type. If the actuator is directly equipped with a tension / compression sensor, the directional force value output by the tension / compression sensor is used as the actuator's signed equivalent force.
[0101] If the actuator is hydraulically driven, the rodless chamber pressure, rod chamber pressure, effective pressure area of the rodless chamber, effective pressure area of the rod chamber, and the direction of action of the actuator are obtained respectively. The rodless chamber pressure and the effective pressure area of the rodless chamber are multiplied to obtain the rodless chamber thrust, and the rod chamber pressure and the effective pressure area of the rod chamber are multiplied to obtain the rod chamber reverse thrust. The difference between the rodless chamber thrust and the rod chamber reverse thrust is extracted and converted into the signed equivalent force of the actuator according to the direction of action of the actuator.
[0102] If the actuator uses a torque interface, the actuator torque, torque direction and transmission conversion relationship are obtained. The torque and transmission conversion relationship are multiplied and converted, and then the signed equivalent force of the actuator is obtained according to the torque direction and the actuator action direction.
[0103] The positive direction of the signed equivalent force of the actuator is set to be consistent with the direction of the force increase caused by the positive displacement of the actuator. This direction setting is used for the subsequent calculation of the current force deviation, the predicted force after execution, the force margin, and the three-state locking control state.
[0104] Within the same control cycle, the nominal coordinates of the control point, the nominal coordinates of the reference target, the direction of thermal expansion, the equivalent length of thermal expansion, the coefficient of linear expansion of the steel, the reference temperature, the allowable direction of assembly, the allowable deviation along the assembly direction, the allowable deviation perpendicular to the assembly direction, the holding force of the actuator with the marked target, the maximum allowable force, the equivalent stiffness, the upper and lower limits of the stroke, the maximum displacement in a single cycle, the displacement response calibration data, the data on the influence of force rebound, the measurement resolution, the force sensor resolution, and the displacement sensor resolution are read.
[0105] The nominal coordinates of the control points are the target positions of each control point in the design assembly coordinate system, which are subsequently used to compare with the temperature correction coordinates to generate the actual assembly deviation.
[0106] The nominal coordinates of the reference target are the reference positions of the three reference targets in the design coordinate system, which will be used to generate the reference valid mark and coordinate correspondence with the current measurement reference target;
[0107] The thermal expansion direction, thermal expansion equivalent length, steel linear expansion coefficient, and reference temperature are the system preset parameters required for temperature correction, which are subsequently used to peel off the positional offset caused by the thermal expansion of the component.
[0108] The assembly allowable direction, the allowable deviation along the assembly direction, and the allowable deviation perpendicular to the assembly direction are assembly tolerance parameters, which are subsequently used to construct the direction normalization matrix, the assembly risk index, and the comprehensive boundary occupancy index.
[0109] The actuator's signed target holding force, maximum allowable force, equivalent stiffness, upper and lower limits of stroke, and maximum displacement per single cycle are the actuator's boundary parameters, which are subsequently used for springback prediction, force boundary constraints, position boundary constraints, and amplitude limiting.
[0110] Displacement response calibration data represents the direct displacement response of each actuator unit displacement to each control point, while force rebound influence data represents the influence of force changes on each actuator on the release rebound displacement of each control point. Both are obtained from one-time equipment calibration or station calibration and are subsequently used to construct the equivalent influence matrix.
[0111] The measurement resolution is the smallest resolvable spatial scale of the control point coordinates and the reference target coordinates, which is subsequently used for non-collinear scale stabilization, reference validity judgment, local orientation normalization matrix construction, and regularization quantity generation.
[0112] The force sensor resolution and displacement sensor resolution are the minimum resolvable quantities for force detection and displacement detection of the actuator, respectively, which are subsequently used for force margin conversion, stabilization stiffness generation, and determination of the executable action scale in this cycle.
[0113] After reading the system preset parameters, the system preset parameters are verified for legality. The allowable deviation along the assembly direction and the allowable deviation perpendicular to the assembly direction are both positive and greater than the measurement resolution. The maximum allowable force is positive and greater than the force sensor resolution. The actuator equivalent stiffness is positive. The maximum displacement in a single cycle is positive. The upper limit of the stroke is greater than the lower limit of the stroke. The measurement resolution, force sensor resolution, and displacement sensor resolution are all positive.
[0114] When any system preset parameter does not meet the corresponding legality condition, it will not enter the normal compensation control calculation chain of this control cycle, maintain the current position of each actuator, and output the prohibition of locking and manual verification status.
[0115] When all the system preset parameters meet the corresponding legality conditions, this step outputs the control point coordinates, reference target coordinates, component temperature, actuator position, actuator signed equivalent force, and all system preset parameters for the current control cycle. The output is used for reference validity judgment in step two, for coordinate unification and temperature correction in step three, and for force rebound compensation, actuator displacement solution, amplitude limiting, post-execution prediction, and three-state locking control in steps five to eight.
[0116] By synchronously acquiring control point coordinates, benchmark target coordinates, component temperature, actuator position, actuator force state, and system preset parameters in each control cycle, and converting the pressure, tension, or torque data of different types of actuators into signed equivalent forces under a unified directional convention, the field geometric state, thermal state, force state, and actuator boundary conditions are all incorporated into subsequent processing within the same control cycle. This avoids subsequent steps from relying solely on single measurement point deviations or single actuator positions for control judgments. At the same time, parameter validity verification eliminates control cycles that lack physical boundary conditions, providing a consistent input basis for benchmark validity judgment, temperature correction, springback prediction, displacement calculation, and three-state locking output.
[0117] like Figure 3 As shown, step two: determine the validity of the benchmark based on the benchmark target coordinates, and construct the correspondence between the measured coordinates and the design coordinates when the benchmark is valid;
[0118] Taking the nominal reference target and the current measurement reference target as the processing objects respectively, the same reference point is selected as the starting point, and the starting point is pointed to the other two reference points respectively to construct the first reference side vector and the second reference side vector.
[0119] For a nominal reference target, the first reference edge vector is obtained by subtracting the coordinates of the first nominal reference target from the coordinates of the second nominal reference target, and the second reference edge vector is obtained by subtracting the coordinates of the first nominal reference target from the coordinates of the third nominal reference target.
[0120] For the current measurement reference target, the first reference side vector is obtained by subtracting the coordinates of the first current measurement reference target from the coordinates of the second current measurement reference target, and the second reference side vector is obtained by subtracting the coordinates of the first current measurement reference target from the coordinates of the third current measurement reference target;
[0121] Perform a cross product on the first reference edge vector and the second reference edge vector, and extract the magnitude of the cross product result. The magnitude of the cross product result is used to characterize the spatial deviation strength between the two reference edges.
[0122] The magnitude of the first reference side vector is superimposed with the measurement resolution to obtain the stabilized reference length. The stabilized reference length is used as a scale reference for non-collinear scale calculation to avoid the ratio calculation being unstable when the first reference side length is close to zero.
[0123] The modulus of the cross product result is used as the numerator, and the stabilized reference length is used as the denominator for ratio processing to obtain the non-collinear scale, which represents the degree to which the two reference edges are not collinear.
[0124] The magnitude of the first reference edge vector is used as the first scale value, and the non-collinear scale is used as the second scale value. The first scale value and the second scale value are compared one by one, and the smaller one is selected as the usable reference scale.
[0125] Perform the above calculation process on the nominal reference target and the current measurement reference target respectively to obtain the available scale of the nominal reference and the available scale of the measurement reference.
[0126] The available scale of the nominal reference is compared with the measurement resolution. If the available scale of the nominal reference reaches or exceeds the measurement resolution, the nominal reference is deemed to meet the validity condition. If the available scale of the nominal reference is less than the measurement resolution, the nominal reference is deemed invalid.
[0127] The available scale of the measurement reference is compared with the measurement resolution. If the available scale of the measurement reference is equal to or greater than the measurement resolution, the measurement reference is deemed to meet the validity condition. If the available scale of the measurement reference is less than the measurement resolution, the measurement reference is deemed invalid.
[0128] When both the nominal datum and the measurement datum meet the validity conditions, the datum validity flag is set to valid; when either the nominal datum or the measurement datum does not meet the validity conditions, the datum validity flag is set to invalid.
[0129] When the reference valid flag is invalid, maintain the current position of each actuator, output the state of prohibiting locking and manual verification, do not enter the normal compensation control calculation chain described in steps three to eight, and do not generate predicted physical residual, assembly risk index, comprehensive boundary occupancy index, remaining mechanical executable margin and executable action scale of this cycle.
[0130] When the reference validity flag is valid, the first reference edge vector is normalized to obtain the first unit direction, wherein the normalization process uses the magnitude of the first reference edge vector itself as the scale reference.
[0131] The cross product of the first reference edge vector and the second reference edge vector is normalized to obtain the third unit direction, wherein the normalization process uses the magnitude of the cross product itself as the scale reference.
[0132] The second unit direction is obtained based on the cross product relationship between the third unit direction and the first unit direction.
[0133] Combine the first unit direction, the second unit direction, and the third unit direction column by column to form an orthogonal basis matrix;
[0134] Perform the above orthogonal basis construction process on the nominal reference target to obtain the design coordinate orthogonal basis matrix, and perform the above orthogonal basis construction process on the current measurement reference target to obtain the measurement coordinate orthogonal basis matrix;
[0135] The design reference geometric center is obtained by averaging the coordinates of the three nominal reference targets, and the measurement reference geometric center is obtained by averaging the coordinates of the three current measurement reference targets.
[0136] This step outputs the valid reference flag, the orthogonal basis matrix of the design coordinates, the orthogonal basis matrix of the measurement coordinates, the geometric center of the design reference, and the geometric center of the measurement reference. The output is used in step three to unify the coordinates of the control points to the design assembly coordinate system, and in step eight to determine the three-state locking control state.
[0137] By constructing reference side vectors for the nominal reference target and the current measurement reference target respectively, extracting non-collinear scales, and forming usable reference scales, the measurement reference is subject to the joint constraints of side length scales and spatial non-collinearity before entering coordinate unification. This enables the identification of cases where the reference is degraded, the reference measurement is abnormal, or the reference target layout does not meet the requirements for constructing an orthogonal basis. When the reference is valid, the correspondence between the measurement coordinates and the design coordinates is further established using the unit direction, the orthogonal basis matrix, and the reference geometric center, so that the subsequent control point coordinate transformation is based on the valid spatial reference. When the reference is invalid, the current position of the actuator is directly maintained and the manual verification status is output to prevent abnormal references from continuing to be transmitted to the compensation control chain.
[0138] Step 3: Unify the coordinates of the control points to the design assembly coordinate system, and obtain the temperature correction coordinates by combining the component temperature and thermal expansion parameters;
[0139] When the datum validity flag is valid, call the design coordinate orthogonal basis matrix, measurement coordinate orthogonal basis matrix, design datum geometric center and measurement datum geometric center obtained in step two, and transform the control point measurement coordinates collected in step one to the design assembly coordinate system;
[0140] For each control point, first subtract the three-dimensional coordinates of the geometric center of the measurement reference from the three-dimensional coordinates of the control point in the measurement coordinate system to obtain the relative coordinates of the control point with respect to the center of the measurement reference.
[0141] The relative coordinates are decomposed according to the orthogonal basis matrix of the measurement coordinates to obtain the local coordinate components of the control point in the orthogonal direction of the measurement reference.
[0142] Then, the local coordinate components are projected onto each basis vector in the orthogonal basis matrix of the design coordinates, and then recombined according to the direction of each basis vector to obtain the relative coordinates of the control points in the orthogonal direction of the design coordinates.
[0143] The relative coordinates in the orthogonal direction of the design coordinates are superimposed with the geometric center of the design datum to obtain the control point coordinates unified to the design assembly coordinate system.
[0144] Subsequently, the thermal expansion correction amount is calculated based on the component temperature collected in step one and the system's preset thermal expansion parameters;
[0145] For each control point, the temperature difference between the temperature of the adjacent components and the reference temperature is extracted to obtain the temperature difference amount;
[0146] The thermal expansion ratio per unit length is obtained by multiplying the temperature difference with the coefficient of linear expansion of the steel.
[0147] The thermal expansion / contraction ratio per unit length is multiplied by the equivalent thermal expansion / contraction length corresponding to the control point to obtain the thermal expansion / contraction amplitude of the control point along the thermal expansion / contraction direction.
[0148] The thermal expansion amplitude is distributed along the thermal expansion direction corresponding to the control point to obtain the thermal expansion correction amount of the control point.
[0149] The temperature-corrected coordinates are obtained by subtracting the thermal expansion correction amount corresponding to the control point from the coordinates of the control point after unifying them to the design assembly coordinate system.
[0150] This step outputs the temperature correction coordinates of each control point. The temperature correction coordinates are compared with the nominal coordinates of the control points in step four to generate the actual assembly deviation.
[0151] By calling the design coordinate orthogonal base, measurement coordinate orthogonal base, and two types of reference geometric centers formed when the reference is valid, the coordinates of the control points measured on site are mapped to the design assembly coordinate system, so that the subsequent deviation calculation no longer directly depends on the spatial position under the original measurement coordinate system; at the same time, by combining the component temperature, reference temperature, coefficient of linear expansion of steel, equivalent length of thermal expansion and contraction, and direction of thermal expansion, the thermal expansion correction amount corresponding to the temperature difference is extracted, and the correction amount is subtracted from the unified control point coordinates, so that the reversible thermal expansion component caused by environmental temperature difference, welding residual heat or sunlight is not mistakenly used as assembly deviation in the subsequent compensation solution.
[0152] Step 4: Compare the temperature correction coordinates with the nominal coordinates of the control points to obtain the actual assembly deviation;
[0153] Call the temperature correction coordinates of each control point obtained in step three, and call the nominal coordinates of each control point read in step one;
[0154] For each control point, the temperature correction coordinates of that control point are matched one by one with the nominal coordinates of the corresponding control point;
[0155] For each pair of matched temperature correction coordinates and nominal coordinates of the control point, perform difference processing in the corresponding direction of the design assembly coordinate system, and use the offset of the temperature correction coordinates relative to the nominal coordinates as the position offset of the control point.
[0156] The positional offset of the control point is taken as the actual assembly deviation, which represents the actual assembly deviation state after removing the differences in the reference coordinates and thermal expansion errors.
[0157] Arrange the actual assembly deviations of all control points in sequence according to the control point order to form the actual assembly deviation vector;
[0158] This step outputs the actual assembly deviation and the actual assembly deviation vector for each control point. The actual assembly deviation is used to generate the effective compensation error in step six and to generate the predicted physical residual in step eight.
[0159] By matching the temperature-corrected coordinates of each control point with their corresponding nominal coordinates one by one, and extracting the position offset in the corresponding direction of the design assembly coordinate system, the actual assembly deviation is formed after the reference drift processing and thermal expansion peeling, which can characterize the actual assembly deviation state of the control point relative to the design target. This actual assembly deviation serves as both the geometric basis for effectively compensating for error generation and as the pre-input for generating predicted physical residuals after execution. This ensures that subsequent springback prediction, equivalent influence matrix solution, and locking determination all revolve around the actual deviation after coordinate unification and temperature correction, avoiding repeated compensation of previous reversible errors.
[0160] Step 5: Based on the difference between the current signed equivalent force of the actuator and the signed target holding force, and combined with the force rebound influence data, generate the release rebound prediction amount;
[0161] Call the current signed equivalent force of each actuator obtained in step one, and call the system's preset signed target holding force of each actuator;
[0162] For each actuator, the current signed equivalent force is compared with the corresponding signed target holding force, and the difference between the two is calculated to obtain the force deviation of the actuator.
[0163] Call the force rebound influence data read in step one. The force rebound influence data is used to represent the effect of the unit force deviation of each actuator on the release rebound displacement of each control point.
[0164] For each control point and for each actuator, firstly, read the force-rebound influence data of the actuator corresponding to the control point. This data is used to describe the displacement response relationship generated by a unit force change at the control point. Then, take the force deviation of the actuator as the input quantity and calculate the correspondence with the force-rebound influence data. Convert the force deviation into the displacement change at the control point proportionally to obtain the rebound displacement contribution of the actuator to the control point.
[0165] The springback displacement contribution of all actuators corresponding to the same control point is accumulated to obtain the predicted release springback of that control point.
[0166] According to the order of control points, the release rebound prediction values of all control points are arranged sequentially to form the release rebound prediction vector.
[0167] This step outputs the release springback prediction amount and release springback prediction vector for each control point. The release springback prediction amount is combined with the actual assembly deviation in step six to form an effective compensation error.
[0168] By comparing the current signed equivalent force of each actuator with the signed target holding force, the force deviation of the actuator is extracted. Based on the force rebound influence data, the force deviation is converted into the rebound displacement contribution at each control point. This allows the release rebound caused by unloading, locking, or force redistribution during steel structure assembly to be characterized in advance before the control command is generated. The release rebound prediction is formed by accumulating the rebound contributions of multiple actuators at the same control point. This ensures that the subsequent effective compensation error not only reflects the current geometric position deviation but also the expected influence of the current force state on the position of the control point after locking. As a result, the displacement solution is closer to the physical state after execution.
[0169] Step 6: Combine the actual assembly deviation and the predicted springback amount into an effective compensation error, and construct an equivalent influence matrix that includes the direct displacement response and the springback change response;
[0170] Call the actual assembly deviation obtained in step four, and call the release springback prediction obtained in step five;
[0171] For each control point, the actual assembly deviation of the control point is superimposed with the corresponding predicted release springback to obtain the effective compensation error of the control point.
[0172] According to the order of control points, the effective compensation errors of all control points are arranged sequentially to form an effective compensation error vector;
[0173] Call the assembly allowable direction read in step one. For each control point, first represent the assembly allowable direction as a unit direction vector, and then perform an outer product operation on the unit direction vector and its own transpose vector to obtain the projection matrix used to describe the component extracted along the assembly direction of the control point, thus forming the projection relationship along the assembly direction.
[0174] Call the complete three-dimensional spatial reference relationship, which represents the basic spatial reference relationship covering all three-dimensional directions within the design assembly coordinate system;
[0175] The complete three-dimensional spatial reference relationship is compared with the projection relationship along the assembly direction, and the component along the assembly direction is removed from the complete three-dimensional spatial reference relationship to obtain the projection relationship perpendicular to the assembly direction.
[0176] Call up the allowable deviation along the assembly direction, the allowable deviation perpendicular to the assembly direction, and the measurement resolution respectively;
[0177] The allowable deviation along the assembly direction is numerically superimposed with the measurement resolution to form a stabilized scale boundary along the assembly direction.
[0178] The allowable deviation in the vertical assembly direction is numerically superimposed with the measurement resolution to form a stable scale boundary in the vertical assembly direction.
[0179] The projection relationship along the assembly direction is scaled using the stabilized scale boundary along the assembly direction as the scale reference for scale unification.
[0180] The projection relationship in the vertical assembly direction is uniformly processed by using the stabilized scale boundary in the vertical assembly direction as a scale reference.
[0181] The projection relationships after unifying the two scales are superimposed to form a local orientation normalization matrix;
[0182] For each actuator, retrieve the actuator's equivalent stiffness and the force-rebound impact data corresponding to each control point;
[0183] The equivalent stiffness of the actuator is multiplied one by one with the force rebound influence data of the actuator corresponding to each control point, and arranged in the order of each control point to obtain the response column data of the actuator to the rebound change of each control point.
[0184] Arrange the springback response data of each actuator in sequence according to the actuator order to form a springback response matrix;
[0185] The displacement response calibration data read in step one is called up. The displacement response calibration data represents the direct displacement response of each actuator unit displacement to each control point.
[0186] The rebound change response matrix is superimposed with the displacement response calibration data to generate an equivalent influence matrix that includes both the direct displacement response and the rebound change response.
[0187] This step outputs an effective compensation error vector, a local direction normalization matrix, and an equivalent influence matrix. These outputs are used in step seven to generate the actuator displacement increment.
[0188] By superimposing the actual assembly deviation with the predicted springback amount to form an effective compensation error, and forming a local direction normalization matrix based on the assembly allowable direction, perpendicular assembly direction, and corresponding tolerance boundary, deviations at different control points, in different directions, and at different tolerance levels can be solved for compensation on a unified scale. At the same time, the equivalent stiffness of the actuator and the force-induced springback influence data are combined into a springback change response matrix, and superimposed with the displacement response calibration data to form an equivalent influence matrix. This allows the direct geometric response of the actuator displacement to the control point and the subsequent springback response caused by force changes to be simultaneously incorporated into the control model, avoiding residual shifts after execution caused by generating compensation amounts solely based on displacement response calibration.
[0189] like Figure 5 As shown, step seven: Generate the actuator displacement increment based on the effective compensation error and equivalent influence matrix, and limit the amplitude by combining the actuator stroke, force boundary and single-cycle displacement boundary;
[0190] Call the local orientation normalization matrix of each control point obtained in step six, arrange the local orientation normalization matrix in sequence according to the control point order, and fill the main diagonal area of the overall matrix block by block to form the overall orientation normalization data.
[0191] The effective compensation error vector obtained in step six is called, and the overall directional normalized data is used as the directional weight. The directional components in the effective compensation error vector are weighted separately. Then, the sum of squares of the weighted directional components is calculated and the square root is taken as the scale factor. Each directional component is then divided by the scale factor to normalize the overall length of the processed vector to a uniform scale, thereby obtaining the normalized effective compensation error vector.
[0192] Call the maximum single-cycle displacement of each actuator read in step one, and form the single-cycle displacement scale data of the actuators in the order of the actuators;
[0193] Call the equivalent influence matrix obtained in step six, use the overall directional normalized data as directional weights to apply to the influence coefficients between the corresponding control points and actuators in the equivalent influence matrix, weight and superimpose the directional components to form a comprehensive influence relationship with consistent direction, and then process the combined result according to the single-cycle displacement scale of the actuator to obtain the normalized actuator equivalent influence matrix.
[0194] Call the measurement resolution, allowable deviation along the assembly direction, and allowable deviation perpendicular to the assembly direction read in step one;
[0195] For each control point, the allowable deviation of the control point along the assembly direction is superimposed with the measurement resolution to form a stabilized scale boundary along the assembly direction;
[0196] For each control point, the allowable deviation in the vertical assembly direction of the control point is superimposed with the measurement resolution to form a stable scale boundary in the vertical assembly direction;
[0197] The ratio of the measurement resolution to the stable scale boundary along the assembly direction is processed to obtain the resolution ratio along the assembly direction.
[0198] The ratio of the measurement resolution to the stabilized scale boundary in the vertical assembly direction is processed to obtain the resolution ratio in the vertical assembly direction.
[0199] The resolution percentage values along the assembly direction and perpendicular to the assembly direction are squared respectively.
[0200] For each control point, the squared result of the resolution proportion along the assembly direction is added to the squared result of the resolution proportion in the vertical assembly direction corresponding to the two degrees of freedom in the vertical direction, and the comprehensive squared value of the resolution proportion of the control point is obtained.
[0201] After summing up the normalized error components of all control points in the three directions, divide by the product of the total number of control points and the number of degrees of freedom in the three-dimensional directions, and average the summed result to obtain the regularization amount.
[0202] Perform matrix multiplication on the normalized actuator equivalent influence matrix and its transpose to obtain the autocorrelation matrix of the normalized actuator equivalent influence;
[0203] The regularization amount is filled as a diagonal element into an identity matrix with the same dimension as the number of executors, so that each element on the diagonal of the identity matrix is equal to the regularization amount, thereby forming a regularization matrix.
[0204] The autocorrelation matrix and the regularization matrix are superimposed to obtain the stable solution matrix;
[0205] The transpose of the normalized actuator equivalent influence matrix is multiplied row by row with the normalized effective compensation error vector and then accumulated. The influence column corresponding to each actuator is then multiplied with the error vector to obtain the error projection amount in each actuator direction. These are then summarized to form the error projection term.
[0206] The direction of the error projection term is reversed while its amplitude remains unchanged to form a compensation amount opposite to the direction of the original error projection term, resulting in a reverse error projection term. The reverse error projection term is used to indicate that the actuator displacement increment acts in the direction that cancels the effective compensation error.
[0207] The stabilization solution matrix is solved by closed-form matrix solving to obtain the initial normalized displacement solution. The initial normalized displacement solution is then superimposed element by element with the back-incorrect error projection term to obtain the normalized displacement increment.
[0208] The normalized displacement increment is converted according to the single-cycle displacement scale of the actuator to obtain the original displacement increment of each actuator. The original displacement increment is used to form the actuator position command after subsequent limiting processing.
[0209] Call the maximum allowable force, the actuator's signed equivalent force, the actuator's equivalent stiffness, the force sensor resolution, and the displacement sensor resolution read in step one;
[0210] The ratio of the force sensor resolution to the displacement sensor resolution is used to obtain the resolution-compensated stiffness.
[0211] The resolution compensation stiffness is superimposed with the actuator equivalent stiffness to form a stable stiffness;
[0212] For each actuator, the maximum permissible force is subtracted from the current signed equivalent force to calculate the difference, obtaining the remaining force between the current force and the positive maximum permissible force boundary. Using the stabilization stiffness as a scale reference, this positive force difference is converted into a positive displacement force candidate margin. The positive displacement force candidate margin is compared with the non-negative lower limit. If the positive displacement force candidate margin is lower than the non-negative lower limit, the positive displacement force margin is taken as the non-negative lower limit. If the positive displacement force candidate margin is not lower than the non-negative lower limit, the positive displacement force candidate margin is taken as the positive displacement force margin.
[0213] For each actuator, first determine the direction of the current signed equivalent force, and assign the same sign to the maximum allowable force according to the direction. Then, calculate the difference between the current signed equivalent force and the force to obtain the remaining force of the current force from the negative maximum allowable force boundary. Using the stabilization stiffness as a scale reference, convert the difference of the reverse force into a candidate margin of reverse displacement force. Compare the candidate margin of reverse displacement force with the non-negative lower limit. If the candidate margin of reverse displacement force is lower than the non-negative lower limit, then the reverse displacement force margin is taken as the non-negative lower limit. If the candidate margin of reverse displacement force is not lower than the non-negative lower limit, then the candidate margin of reverse displacement force is taken as the reverse displacement force margin.
[0214] Call the executor's current position and executor travel boundary read in step one;
[0215] For each actuator, the upper limit of the stroke is subtracted from the current position to calculate the difference between the two, resulting in a positive position candidate margin. The positive position candidate margin is compared with the non-negative lower limit. If the positive position candidate margin is lower than the non-negative lower limit, the positive position margin is taken as the non-negative lower limit. If the positive position candidate margin is not lower than the non-negative lower limit, the positive position candidate margin is taken as the positive position margin.
[0216] For each actuator, the current position is subtracted from the lower limit of the travel to calculate the difference between the two, resulting in a reverse position candidate margin. The reverse position candidate margin is compared with the non-negative lower limit. If the reverse position candidate margin is lower than the non-negative lower limit, the reverse position margin is taken as the non-negative lower limit. If the reverse position candidate margin is not lower than the non-negative lower limit, the reverse position candidate margin is taken as the reverse position margin.
[0217] For each actuator, the minimum values of the maximum displacement in a single cycle, the positive displacement force margin, and the positive position margin are selected to obtain the upper limit of the available positive displacement.
[0218] For each actuator, the minimum values of the maximum displacement in a single cycle, the force margin of the reverse displacement, and the position margin of the reverse displacement are selected to obtain the upper limit of the usable reverse displacement.
[0219] For each actuator, the original displacement increment is limited between the upper limit of the reverse available displacement and the upper limit of the forward available displacement. If the original displacement increment exceeds the upper limit of the forward available displacement, the upper limit of the forward available displacement is taken as the displacement increment after limiting. If the original displacement increment is lower than the reverse boundary corresponding to the upper limit of the reverse available displacement, the reverse boundary is taken as the displacement increment after limiting. If the original displacement increment is between the two boundaries, the original displacement increment is retained as the displacement increment after limiting.
[0220] This step outputs the displacement increment of each actuator after limiting. The displacement increment after limiting is used in step eight to generate actuator position commands, predicted forces after execution, and predicted physical residuals.
[0221] By utilizing the local orientation normalization matrix to form global orientation normalization data, and by performing scale unification and column-oriented scaling on the effective compensation error and equivalent influence matrix respectively, the error vector and actuator influence relationship before closed-loop solution are made to be at a matching orientation scale. Furthermore, a regularization quantity is generated based on the measurement resolution and assembly tolerance, and the error projection term is processed in the direction of offsetting the effective compensation error to obtain the original displacement increment. Subsequently, the force margin, position margin, single-cycle maximum displacement and non-negative lower limit are combined to form positive and negative usable displacement boundaries, so that the final limited displacement increment simultaneously meets the error compensation requirements and the field actuator's executable constraints.
[0222] Step 8: Output the actuator position command based on the displacement increment after the limit, update the predicted force and predicted physical residual after execution, and output the assembly risk index and three-state locking control status.
[0223] When the reference validity flag is valid, call the current position of each actuator obtained in step one, and call the displacement increment of each actuator after limiting obtained in step seven;
[0224] For each actuator, the current position is superimposed with the displacement increment after the limit is applied to generate the actuator position command;
[0225] Call the signed equivalent force and equivalent stiffness of the actuator obtained in step one;
[0226] For each actuator, the equivalent stiffness is multiplied by the displacement increment after the amplitude is limited to obtain the change in force on the actuator.
[0227] The current signed equivalent force of the actuator is superimposed with the change in force on the actuator to obtain the predicted force after execution;
[0228] Call the signed target holding force read in step one;
[0229] For each actuator, the predicted force after execution is compared with the corresponding signed target holding force, and the difference between the two is calculated to obtain the force deviation after execution;
[0230] Call the force-rebound impact data read in step one;
[0231] For each control point, the force deviation after execution of each actuator is matched item by item with the force rebound influence data of the actuator corresponding to the control point. According to the pre-calibrated proportional relationship in the force rebound influence data, the force deviation after execution is converted into the displacement change in the corresponding direction and amplitude, thereby completing the conversion process from force deviation to rebound displacement and obtaining the rebound displacement contribution of each actuator to the control point after execution.
[0232] The rebound displacement contribution after execution of all operations at the same control point is accumulated to generate the predicted amount of rebound after execution.
[0233] Call the limited displacement increment obtained in step seven and combine them in the order of the actuators to form a set of limited displacement increments;
[0234] The displacement response calibration data is called, and the displacement response calibration data is matched item by item with the set of displacement increments after the amplitude limit is set. The displacement contribution of each actuator to the control point is calculated according to the corresponding relationship to obtain the direct displacement response of the actuator.
[0235] For each control point, the actual assembly deviation obtained in step four, the predicted amount of springback after execution obtained in the current step, and the direct displacement response of the actuator are called.
[0236] The actual assembly deviation, the predicted amount of springback after execution, and the direct displacement response of the actuator are combined and superimposed to obtain the predicted physical residual.
[0237] Call the assembly allowable direction, allowable deviation along the assembly direction, allowable deviation perpendicular to the assembly direction, and maximum allowable force read in step one;
[0238] like Figure 4 As shown, for each control point, the predicted physical residual is projected along the assembly allowable direction, and the absolute value of the projection result is taken. Then, the ratio is processed using the allowable deviation along the assembly direction as a scale reference to obtain the normalized predicted residual occupancy along the assembly direction.
[0239] For each control point, the actual direction component of the control point in the three-dimensional space is first determined according to the complete three-dimensional spatial reference relationship. Then, the projection component along the assembly direction is extracted according to the assembly allowable direction. Subsequently, the projection component along the assembly direction is subtracted from the three-dimensional direction component to obtain the vertical assembly direction component orthogonal to the assembly direction. Based on this, the vertical assembly direction projection relationship is constructed, the predicted physical residual is projected to the vertical assembly direction, and the modulus of the vertical assembly direction projection result is extracted. Then, the vertical assembly direction allowable deviation is used as a scale reference for ratio processing to obtain the normalized predicted residual occupancy of the vertical assembly direction.
[0240] For each actuator, the absolute value of the predicted force after execution is taken, and the ratio is processed with the maximum allowable force as the scale reference to obtain the normalized force occupancy.
[0241] The normalized prediction residual occupancy of all control points along the assembly direction is squared, and the squared results are summed one by one. The normalized prediction residual occupancy of all control points perpendicular to the assembly direction is squared, and the squared results are summed one by one. The summed results of the squares of the two types of residual occupancy are added together, and the average is processed according to the number of control points. Then the square root of the average is processed to obtain the geometric residual root mean square term.
[0242] The normalized force occupancy of all actuators is squared, and the squared results of each actuator are added together to obtain a total. The total is then averaged according to the number of actuators, and the averaged result is squared to obtain the root mean square term of the force.
[0243] The assembly risk index is obtained by adding the root mean square term of the geometric residual to the root mean square term of the stress.
[0244] The maximum occupancy is selected as the comprehensive boundary occupancy index by comparing all normalized prediction residual occupancy along the assembly direction, all normalized prediction residual occupancy perpendicular to the assembly direction, and all normalized force occupancy.
[0245] The normalized boundary is a uniform scale boundary, representing the boundary occupancy state when the predicted physical residual reaches the corresponding assembly tolerance or the predicted force after execution reaches the corresponding maximum allowable force.
[0246] When calculating the remaining mechanically executable margin, for each actuator, the maximum permissible force is compared with the predicted force after execution, and the difference between the maximum permissible force and the predicted force after execution is calculated to obtain the remaining force of the predicted force after execution from the positive maximum permissible force boundary.
[0247] For each actuator, the maximum permissible force and the predicted force after execution are compared numerically in the same direction and the difference is calculated to obtain the remaining force of the predicted force after execution from the negative maximum permissible force boundary.
[0248] The ratio of the force sensor resolution to the displacement sensor resolution is used to obtain the resolution-compensated stiffness.
[0249] The resolution compensation stiffness is superimposed with the actuator equivalent stiffness to obtain the stabilized stiffness after execution.
[0250] Using the post-execution stabilization stiffness as a benchmark, the remaining difference in positive force is converted into the remaining candidate displacement margin of positive force. The remaining candidate displacement margin of positive force is compared with the non-negative lower limit. If the remaining candidate displacement margin of positive force is lower than the non-negative lower limit, the remaining displacement margin of positive force after execution is taken as the non-negative lower limit. If the remaining candidate displacement margin of positive force is not lower than the non-negative lower limit, the remaining candidate displacement margin of positive force is taken as the remaining displacement margin of positive force after execution.
[0251] Using the post-execution stabilization stiffness as a benchmark, the residual difference of reverse force is converted into the residual candidate displacement margin of reverse force. The residual candidate displacement margin of reverse force is compared with the non-negative lower limit. If the residual candidate displacement margin of reverse force is lower than the non-negative lower limit, the post-execution residual displacement margin of reverse force is taken as the non-negative lower limit. If the residual candidate displacement margin of reverse force is not lower than the non-negative lower limit, the residual candidate displacement margin of reverse force is taken as the post-execution residual displacement margin of reverse force.
[0252] For each actuator, the upper limit of the stroke is subtracted from the actuator position command to obtain the candidate remaining margin for the positive position after execution. The candidate remaining margin for the positive position after execution is compared with the non-negative lower limit. If the candidate remaining margin for the positive position after execution is lower than the non-negative lower limit, the candidate remaining margin for the positive position after execution is taken as the non-negative lower limit. If the candidate remaining margin for the positive position after execution is not lower than the non-negative lower limit, the candidate remaining margin for the positive position after execution is taken as the remaining margin for the positive position after execution.
[0253] For each actuator, the difference between the actuator position command and the lower limit of travel is calculated to obtain the candidate remaining margin of the reverse position after execution. The candidate remaining margin of the reverse position after execution is compared with the non-negative lower limit. If the candidate remaining margin of the reverse position after execution is lower than the non-negative lower limit, the candidate remaining margin of the reverse position after execution is taken as the non-negative lower limit. If the candidate remaining margin of the reverse position after execution is not lower than the non-negative lower limit, the candidate remaining margin of the reverse position after execution is taken as the remaining margin of the reverse position after execution.
[0254] For each actuator, the minimum values of the maximum displacement in a single cycle, the remaining displacement margin under positive force after execution, and the remaining position margin after execution are selected to obtain the remaining positive displacement margin.
[0255] For each actuator, the minimum values of the maximum displacement in a single cycle, the remaining displacement margin under reverse force after execution, and the remaining position margin after execution are selected to obtain the remaining displacement margin in reverse.
[0256] The positive and negative residual displacement margins of the same actuator are added together to obtain the combined residual displacement margin of the actuator.
[0257] The maximum value is selected from the combined remaining displacement margin of all actuators to obtain the remaining mechanically executable margin for the current control cycle;
[0258] When calculating the action scale that can be executed in this cycle, for each actuator, the absolute value of the displacement increment after the amplitude limit is taken, and the ratio is processed with the resolution of the displacement sensor as the scale reference to obtain the action resolution multiple of the actuator.
[0259] The maximum value is selected from the motion resolution multiples of all actuators to obtain the executable motion scale for this cycle.
[0260] The three-state locking control status is output based on the benchmark valid flag, comprehensive boundary occupancy index, remaining mechanical executable margin, and executable action scale of this cycle.
[0261] When the reference validity flag is invalid, maintain the current position of the actuator and output a lock-lock prohibited and manual verification status;
[0262] When the baseline validity flag is valid and the comprehensive boundary occupancy index is not greater than the normalized boundary, the lock-in state is allowed.
[0263] When the baseline validity flag is valid, the comprehensive boundary occupancy index is greater than the normalized boundary, the remaining mechanical executable margin reaches or exceeds the displacement sensor resolution, and the executable action scale in this cycle reaches or exceeds the action scale corresponding to the displacement sensor resolution, the output continues to compensate.
[0264] When the benchmark validity flag is valid, the comprehensive boundary occupancy index is greater than the normalized boundary, and the remaining mechanical executable margin is less than the displacement sensor resolution or the executable action scale in this cycle is less than the action scale corresponding to the displacement sensor resolution, the output is locked and manual verification status is prohibited.
[0265] This step ultimately outputs the actuator position command, predicted physical residual, assembly risk index, remaining mechanical executable margin, executable motion scale for this cycle, and three-state locking control status.
[0266] In this embodiment, the allowable deviation along the assembly direction, the allowable deviation perpendicular to the assembly direction, the maximum allowable force, the upper and lower limits of the stroke, and the maximum displacement in a single cycle are all based on the design drawings, assembly process documents, actuator rated parameters, or one-time equipment calibration results.
[0267] When the allowable deviation along the assembly direction or the allowable deviation perpendicular to the assembly direction is small, the corresponding normalization occupancy will increase, and the three-state output will be more likely to enter the continued compensation state or the prohibited locking and manual verification state.
[0268] When the allowable deviation along the assembly direction or the allowable deviation perpendicular to the assembly direction is too large, the corresponding normalization occupancy will decrease, and the sensitivity of the state determination to the predicted residual will weaken accordingly.
[0269] When the maximum displacement in a single cycle is small, the displacement increment after limiting is more easily constrained by the single cycle boundary, and the single cycle compensation amplitude decreases accordingly.
[0270] When the maximum displacement in a single cycle is too large, the displacement increment after limiting is more constrained by the force boundary and the travel boundary.
[0271] In this embodiment, the regularization amount is calculated from the measurement resolution and assembly tolerance, the resolution compensation stiffness is calculated from the force sensor resolution and the displacement sensor resolution, and the three-state locking control state is determined by the reference valid flag, the comprehensive boundary occupancy index, the remaining mechanical executable margin, and the executable action scale of this cycle.
[0272] By converting the displacement increment after limiting into an actuator position command when the reference is valid, and simultaneously updating the predicted force after execution, the predicted amount of release springback after execution, the actuator direct displacement response, and the predicted physical residual, the controller can obtain a residual prediction corresponding to the actual execution state after outputting the command. Then, based on the predicted physical residual and the predicted force after execution, the assembly risk index, the comprehensive boundary occupancy index, the remaining mechanical executable margin, and the executable action scale of this cycle are generated respectively. This ensures that the three-state output of allowing locking, continuing compensation, and prohibiting locking and manually verifying is jointly constrained by residual occupancy, force boundary, stroke margin, and action resolution capability, thereby keeping the locking determination consistent with the compensation capability.
[0273] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0274] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. A method for automatic accuracy compensation control during steel structure assembly, characterized in that, include: Obtain the coordinates of the steel structure assembly control points, the coordinates of the reference target, the component temperature, the actuator position, the equivalent force of the actuator with sign, and the system preset parameters; The validity of the reference is determined based on the reference target coordinates. When the reference is valid, the correspondence between the measured coordinates and the design coordinates is constructed. When the reference is invalid, the actuator position is maintained and the lock-free and manual verification status is output. Unify the coordinates of the control points to the design assembly coordinate system, and obtain the temperature correction coordinates by combining the component temperature and thermal expansion parameters. The temperature correction coordinates are compared with the nominal coordinates of the control points to obtain the actual assembly deviation; Based on the difference between the current signed equivalent force of the actuator and the signed target holding force, and combined with the force rebound influence data, a release rebound prediction is generated; The actual assembly deviation and the predicted amount of springback are combined into an effective compensation error, and an equivalent influence matrix including the direct displacement response and the springback change response is constructed. The actuator displacement increment is generated based on the effective compensation error and the equivalent influence matrix, and the amplitude is limited by combining the actuator stroke, force boundary and single-cycle displacement boundary; When the reference is valid, the actuator position command is output according to the displacement increment after the limit, the predicted force and predicted physical residual after execution are updated to output the assembly risk index, and the three-state locking control state is output in combination with the remaining mechanical executable margin and the executable action scale of this cycle.
2. The method for automatic accuracy compensation control in the steel structure assembly process according to claim 1, characterized in that, The acquisition of the coordinates of the steel structure assembly control points, the coordinates of the reference target, the component temperature, the actuator position, the equivalent force of the actuator with sign, and the system preset parameters specifically includes: During each control cycle, the three-dimensional coordinates of each control point, the three-dimensional coordinates of three non-collinear reference targets, the temperature of components adjacent to the control point, the current position of each actuator, and the signed equivalent force of each actuator are collected. Read the nominal coordinates of the control point, the nominal coordinates of the reference target, the direction of thermal expansion, the equivalent length of thermal expansion, the coefficient of linear expansion of the steel, the reference temperature, the allowable direction of assembly, the allowable deviation along the assembly direction, and the allowable deviation perpendicular to the assembly direction; Read the actuator's signed target holding force, maximum allowable force, equivalent stiffness, upper and lower limits of stroke, maximum displacement in a single cycle, displacement response calibration data, data on the influence of force rebound, measurement resolution, force sensor resolution, and displacement sensor resolution; The positive direction of the actuator's signed equivalent force is set to be consistent with the direction of the force increase caused by the actuator's positive displacement. Non-negative pressure data is converted into the actuator's signed equivalent force according to the pressure difference, effective pressure area, and direction of action. Torque data is converted into the actuator's signed equivalent force according to the torque direction and transmission conversion relationship.
3. The method for automatic accuracy compensation control during steel structure assembly according to claim 1, characterized in that, The process of determining the validity of the benchmark based on the benchmark target coordinates and constructing the correspondence between the measured coordinates and the design coordinates when the benchmark is valid specifically includes: Using the nominal reference target and the current measurement reference target as processing objects respectively, the same reference point is selected as the starting point, and the first reference edge vector and the second reference edge vector are constructed pointing to the other two reference points respectively. The cross product operation is performed on the two reference edge vectors and the magnitude of the cross product result is extracted. The magnitude of the first reference edge vector is combined with the measurement resolution to form the stabilized reference length. Then, the magnitude of the cross product result is compared with the stabilized reference length to obtain the non-collinear scale representing the degree of non-collinearity of the two reference edges. The first scale value and the second scale value are calculated by comparing the length of the first reference side with the non-collinear scale. The smaller of the first scale value and the second scale value are then selected as the usable reference scale. Repeat the above calculation process for the nominal datum and the current measurement datum respectively to obtain the available scales for the nominal datum and the measurement datum. The nominal reference is compared with the measurement resolution. If the nominal reference meets or exceeds the measurement resolution, it is determined that the nominal reference meets the validity condition; otherwise, it is determined to be invalid. The measurement reference is compared with the measurement resolution. If the comparison reaches or exceeds the measurement resolution, the measurement reference is deemed to meet the valid condition; otherwise, it is deemed invalid. When both the nominal datum and the measurement datum meet the validity conditions, the datum validity flag is set to valid; otherwise, it is set to invalid. When the reference is valid, the first reference edge vector is normalized to obtain the first unit direction, the cross product of the two reference edge vectors is normalized to obtain the third unit direction, and the second unit direction is obtained based on the cross product relationship between the first unit direction and the third unit direction. The first unit direction, the second unit direction, and the third unit direction are combined into an orthogonal basis matrix to form the design coordinate orthogonal basis and the measurement coordinate orthogonal basis, respectively. The correspondence between the measurement coordinates and the design coordinates is formed by using the two sets of orthogonal bases and the geometric center of the reference target.
4. The method for automatic accuracy compensation control during steel structure assembly according to claim 1, characterized in that, The process of unifying the control point coordinates to the design assembly coordinate system and obtaining temperature-corrected coordinates by combining component temperature and thermal expansion parameters specifically includes: Based on the correspondence between measured coordinates and design coordinates, the control point coordinates are mapped to the design assembly coordinate system to obtain unified control point coordinates; Subtract the reference temperature from the component temperature to obtain the temperature difference. The temperature difference is multiplied by the linear expansion parameter of the steel and the equivalent length of thermal expansion, and the result is distributed along the direction of thermal expansion to obtain the thermal expansion correction amount for the corresponding control point. The temperature-corrected coordinates are obtained by subtracting the corresponding thermal expansion correction from the unified control point coordinates.
5. The method for automatic accuracy compensation control during steel structure assembly according to claim 1, characterized in that, The step of comparing the temperature correction coordinates with the nominal coordinates of the control points to obtain the actual assembly deviation specifically includes: The temperature correction coordinates of each control point are matched one by one with the nominal coordinates of the corresponding control points; Perform difference processing on the temperature correction coordinates and nominal coordinates of each control point in the corresponding direction to obtain the position offset of the control point in the design assembly coordinate system; The positional offset of each control point in the design assembly coordinate system is taken as the actual assembly deviation, and the actual assembly deviation is used as the input for generating effective compensation error and predicting physical residual generation.
6. The method for automatic accuracy compensation control during steel structure assembly according to claim 1, characterized in that, The method of generating a release rebound prediction based on the difference between the current signed equivalent force of the actuator and the signed target holding force, combined with the force rebound influence data, specifically includes: The current signed equivalent force of each actuator is subtracted from the corresponding signed target holding force one by one to obtain the force deviation of the corresponding actuator. The force deviation of each actuator is mapped to the corresponding force rebound influence data to obtain the rebound displacement contribution of each actuator to each control point. The springback displacement contribution of each actuator corresponding to the same control point is accumulated to obtain the predicted release springback amount of that control point.
7. The method for automatic accuracy compensation control in the steel structure assembly process according to claim 1, characterized in that, The process of combining the actual assembly deviation and the predicted springback amount into an effective compensation error, and constructing an equivalent influence matrix that includes the direct displacement response and the springback change response, specifically includes: The actual assembly deviation at each control point is superimposed with the predicted release and rebound at the corresponding control point to generate the effective compensation error for each control point. All effective compensation errors are then combined into an effective compensation error vector in the order of control points. Construct the projection relationship along the assembly direction according to the assembly allowable direction, and construct the projection relationship perpendicular to the assembly direction according to the complete three-dimensional space reference relationship and the projection relationship along the assembly direction. Using the allowable deviation along the assembly direction and the allowable deviation perpendicular to the assembly direction as scale boundaries, the projection relationship along the assembly direction and the projection relationship perpendicular to the assembly direction are scaled and unified to form a local direction normalization matrix. The equivalent stiffness of the actuator is combined with the force rebound influence data corresponding to each control point to form the rebound change response column data corresponding to each actuator, and the rebound change response column data is combined into a rebound change response matrix according to the actuator order. The rebound change response matrix is superimposed with the displacement response calibration data to generate an equivalent influence matrix that includes the direct displacement response and the rebound change response. This equivalent influence matrix is then used as the input for generating the actuator displacement increment.
8. The method for automatic accuracy compensation control in the steel structure assembly process according to claim 1, characterized in that, The process of generating actuator displacement increments based on effective compensation errors and equivalent influence matrices, and then limiting the amplitude by combining actuator stroke, force boundary, and single-cycle displacement boundary, specifically includes: Using the combined effective compensation error vector as input, the effective compensation error vector is subjected to scale unification processing using the local direction normalization matrix corresponding to each control point to obtain the normalized effective compensation error vector. The local orientation normalization matrix corresponding to each control point is arranged sequentially according to the control point order. The block-diagonal form of the overall orientation normalization data is filled in block by block. The overall orientation normalization data is combined with the equivalent influence matrix. At the same time, the combination result is processed according to the single-cycle displacement scale of the actuator to obtain the normalized actuator equivalent influence matrix. Based on the proportional relationship between the measurement resolution and the allowable deviation along the assembly direction and the allowable deviation perpendicular to the assembly direction, the resolution ratio along the assembly direction and the resolution ratio perpendicular to the assembly direction are obtained respectively. The resolution ratios are squared and averaged according to the number of control points and the number of directional degrees of freedom to generate a regularization quantity. The normalized actuator equivalent influence matrix, regularization quantity, and normalized effective compensation error vector are combined with the closed matrix solution process of the regularization quantity to obtain the original displacement increment of each actuator. Based on the maximum allowable force, the signed equivalent force of the actuator, the equivalent stiffness of the actuator, the resolution of the force sensor, and the resolution of the displacement sensor, calculate the force margin for the positive displacement and the force margin for the negative displacement, respectively. Calculate the forward position margin and the reverse position margin based on the current position of the actuator and the travel boundary. The maximum displacement in a single cycle, the force margin of the positive displacement, and the position margin of the positive position are selected by minimizing the values to obtain the upper limit of the available positive displacement. The maximum displacement in a single cycle, the force margin of the negative displacement, and the position margin of the negative position are selected by minimizing the values to obtain the upper limit of the available negative displacement. The original displacement increment of each actuator is limited to the upper limit of the reverse available displacement and the upper limit of the forward available displacement to generate the limited displacement increment.
9. The method for automatic accuracy compensation control during steel structure assembly according to claim 1, characterized in that, The process involves outputting actuator position commands based on the displacement increment after amplitude limiting, updating the predicted force and predicted physical residual after execution to output an assembly risk index, and combining the remaining mechanical executable margin and the executable action scale of the current cycle to output a three-state locking control state, specifically including: When the reference validity flag is valid, the current position of the actuator is added to the displacement increment after the limit is applied to generate the actuator position command; When the reference validity flag is valid, the signed equivalent force of the actuator is superimposed with the force change corresponding to the equivalent stiffness of the actuator and the displacement increment after the limit, to obtain the predicted force after execution. When the benchmark validity flag is valid, the predicted force after execution is subtracted from the marked target holding force one by one to obtain the force deviation of each actuator after execution; The force deviation after execution of each actuator is multiplied and mapped with the corresponding force rebound influence data to obtain the rebound displacement contribution of each actuator to each control point. The rebound displacement contributions of all the same control point are accumulated to generate the predicted amount of rebound after execution. When the reference validity flag is valid, the actuator direct displacement response is generated based on the displacement response calibration data and the displacement increment after the limit. The actual assembly deviation, the predicted amount of springback after execution, and the actuator direct displacement response are combined and superimposed to generate the predicted physical residual. When the benchmark validity flag is valid, the predicted physical residual is decomposed according to the assembly allowable direction to obtain the predicted residual along the assembly direction and the predicted residual perpendicular to the assembly direction. Based on the allowable deviation along the assembly direction, the allowable deviation perpendicular to the assembly direction and the maximum allowable force, the predicted residual and the predicted force after execution are scaled uniformly to generate the assembly risk index. When the benchmark validity flag is valid, multiple normalized occupancy values are calculated based on the ratio of the predicted residual along the assembly direction to the allowable deviation along the assembly direction, the ratio of the predicted residual perpendicular to the assembly direction to the allowable deviation perpendicular to the assembly direction, and the ratio of the predicted force after execution to the maximum allowable force. The multiple normalized occupancy values are compared item by item, and the maximum occupancy value is selected as the comprehensive boundary occupancy index. When the reference validity flag is valid, the remaining force margin is calculated based on the difference between the predicted force after execution and the maximum allowable force. The ratio of the force sensor resolution to the displacement sensor resolution is processed to obtain the resolution compensation stiffness. The resolution compensation stiffness and the equivalent stiffness of the actuator are combined into the stabilized stiffness. Then, the remaining force margin is converted into the corresponding executable displacement margin based on the stabilized stiffness. Calculate the remaining position margin based on the distance between the actuator position command and the actuator travel boundary; The minimum values of the executable displacement margin, position residual margin, and maximum displacement in a single cycle corresponding to the force are selected to form the positive residual displacement margin and the reverse residual displacement margin, respectively. The forward and reverse residual displacement margins of the same actuator are merged, and the maximum merged margin is selected among all actuators to generate the remaining mechanically executable margin. When the reference validity flag is valid, the ratio of the displacement increment after the amplitude limit to the resolution of the corresponding displacement sensor is calculated to obtain the motion resolution multiple of each actuator, and the maximum value of the motion resolution multiple of all actuators is selected to generate the executable motion scale for this cycle. When the reference validity flag is invalid, the actuator maintains its current position and outputs a lock-lock prohibited and manual verification status. When the baseline validity flag is valid and the comprehensive boundary occupancy index does not exceed the normalized boundary, the lock-in state is allowed. When the baseline valid flag is valid, the comprehensive boundary occupancy index exceeds the normalized boundary, the remaining mechanical executable margin reaches the displacement sensor resolution, and the executable action scale of this cycle reaches the action scale corresponding to the displacement sensor resolution, the output continues to compensate. When the baseline validity flag is valid, the comprehensive boundary occupancy index exceeds the normalized boundary, and the remaining mechanical executable margin does not reach the displacement sensor resolution or the executable action scale in this cycle does not reach the action scale corresponding to the displacement sensor resolution, the output will prohibit locking and manual verification status.
