Intelligent control method and system for bolt tightening based on shaft force measurement

Abstract

The present application relates to the technical field of bolt tightening, and proposes a bolt tightening intelligent control method and system based on axial force measurement. The method comprises the following steps: collecting bolt identification and process parameter data, collecting temperature measurement data, obtaining sound velocity compensation parameters and elastic modulus compensation parameters. Collecting zero-load preloading data and processing to generate zero-load baseline data, obtaining a self-calibration model. Collecting axial force data and ultrasonic echo data, obtaining axial force fusion estimation results. Comparing the axial force fusion estimation results with the target axial force data. When in the target interval, fine tightening control is carried out based on model predictive control and micro-step strategy, and a controllable damping unit is driven to absorb energy and stop. Collecting the axial force fusion estimation results when the target axial force is reached, processing the holding stage data, obtaining the axial force rebound checking results, and if the axial force rebound checking results exceed the limit, secondary supplementary tightening control is carried out. The scheme realizes the refinement and traceability of the bolt tightening process.

Description

Technical Field

[0001] This invention relates to the field of bolt tightening technology, and more specifically, to an intelligent control method and system for bolt tightening based on axial force measurement. Background Technology

[0002] Bolted connections, as the most common and critical connection method in mechanical structures, directly affect the strength, sealing performance, and fatigue life of the structure due to the magnitude of their preload. In key fields such as aerospace, energy equipment, pressure vessels, and rail transportation, bolt tightening quality control is a crucial aspect of ensuring the safe operation of systems. Traditional tightening methods are mostly based on torque or rotation methods. However, due to fluctuations in factors such as thread friction coefficient, surface roughness, lubrication conditions, and material elastic modulus, there is uncertainty between the torque and the actual axial force of the bolt, making it difficult to guarantee tightening consistency and reliability. To address this, the industry has gradually developed tightening methods based on bolt axial force detection. By directly controlling the axial force through real-time detection, tightening accuracy can be improved.

[0003] However, existing technologies based on axial force detection still have shortcomings. For example, patent CN120160736A discloses a bolt tightening force control device and method based on real-time bolt axial force detection. It measures the bolt axial force in real time using an ultrasonic probe and dynamically stops the hydraulic wrench during tightening to ensure that the bolt axial force reaches the set value. Although this solution improves the controllability of bolt preload to some extent, it still has several problems: the measurement relies on the liquid coupling agent to adhere to the probe, which is easily affected by oil, vibration, and temperature fluctuations in the field environment, resulting in unstable echo signals and affecting the accuracy of axial force measurement; it uses a single ultrasonic detection path and lacks a fusion and verification mechanism with other sensor signals (such as strain or torque), making it difficult to detect measurement anomalies; the control strategy mainly relies on "stopping when the set value is reached," lacking dynamic buffering and fine-tuning measures when approaching the target value, which can easily cause axial force overshoot or rebound.

[0004] Therefore, it is necessary to design an intelligent control method and system for bolt tightening based on axial force measurement to solve the problems existing in the current technology. Summary of the Invention

[0005] In view of this, the present invention proposes an intelligent control method and system for bolt tightening based on axial force measurement, aiming to solve the problems of insufficient measurement stability, lack of multi-source fusion verification, and coarse control strategy in the current technology.

[0006] In one aspect, the present invention proposes an intelligent control method for bolt tightening based on axial force measurement, comprising: Collect bolt identification and process parameter data, collect temperature measurement data, including probe temperature data and ambient temperature data, process the process parameter data and temperature data to obtain sound velocity compensation parameters and elastic modulus compensation parameters; Collect zero-load preload data and process it to generate zero-load baseline data to obtain a self-calibration model; Axial force data and ultrasonic echo data were collected to obtain the axial force fusion estimation result; The axial force fusion estimation result is compared with the target axial force data. When the axial force fusion estimation result is lower than the first threshold of the target axial force, the actuator is controlled to drive at constant speed or constant power. When the axial force fusion estimation result is within the target range, fine-tuning control is performed based on model predictive control and micro-stepping strategy, and the controllable damping unit is driven to absorb energy and stop according to the axial force climb rate data and the angle change rate data. The axial force fusion estimation results when the target axial force is reached are collected, the data of the holding stage are processed, and the axial force springback check results are obtained. If the axial force springback check results exceed the limit, secondary tightening control is performed.

[0007] Furthermore, when processing the process parameter data and temperature data to obtain the sound velocity compensation parameters and elastic modulus compensation parameters, the process includes: Based on the material code, heat treatment state, surface roughness grade, and thread lubrication state in the process parameter data, the reference sound velocity and reference elastic modulus are retrieved from the preset calibration index table; combined with the probe temperature data and ambient temperature data, the surface temperature of the test piece is estimated, and the reference sound velocity and reference elastic modulus are temperature-corrected; based on the surface roughness grade and thread lubrication state, a coupling correction factor is introduced to correct the temperature-corrected sound velocity and elastic modulus, thereby obtaining the sound velocity compensation parameters and elastic modulus compensation parameters.

[0008] Furthermore, when acquiring zero-load preload data and processing it to generate zero-load baseline data to obtain the self-calibration model, the process includes: The control actuator enters the zero-load preload mode, and drives the axial force sensor A and axial force sensor B to establish stable contact without generating effective preload force below the preset zero-load preload upper limit threshold. Simultaneously acquire zero-load signal data from axial force sensor A, zero-load echo data from axial force sensor B, zero-position data from angle encoder, zero-point data from torque sensor, and zero-load temperature data from temperature sensor to form zero-load preload data; The zero-load preload data is filtered, denoised, and drift subtracted, and a steady-state determination is performed according to a preset steady-state determination threshold. When the steady-state determination is passed, the zero-point offset, time delay offset, noise threshold, gain correction coefficient, and coupling quality index are calculated to generate zero-load baseline data. The self-calibration model is updated using the zero-load baseline data combined with the sound velocity compensation parameters and elastic modulus compensation parameters. The self-calibration model is used for axial force fusion estimation and anomaly diagnosis. When the steady-state determination fails or the coupling quality index is lower than the threshold, the zero-load preload and contact posture are automatically adjusted and the data is repeatedly collected and processed until zero-load baseline data that meets the threshold conditions is generated and stored in conjunction with the bolt identification.

[0009] Furthermore, when acquiring axial force data and ultrasonic echo data to obtain the axial force fusion estimation result, the following is included: Axial force correction data is obtained from axial force sensor A; ultrasonic axial force estimation data is obtained from axial force sensor B; within a sliding time window, the axial force correction data and the ultrasonic axial force estimation data are time-aligned, steady-state determined, and abnormal samples are removed; the confidence level is determined based on the signal-to-noise ratio, coupling quality index, and temperature stability; the two types of data are weighted and fused to obtain the axial force fusion estimation result; when the signal-to-noise ratio or coupling quality index of axial force sensor B is lower than the threshold, the weight of axial force sensor B is reduced or the output of axial force sensor A is switched to single-channel output; when axial force sensor A saturates or drifts beyond the limit, the output of axial force sensor B is used and the re-establishment of zero-load baseline data is triggered.

[0010] Furthermore, when comparing the axial force fusion estimation result with the target axial force data, the following steps are included: When the axial force fusion estimation result is lower than the first threshold of the target axial force, the actuator is controlled to drive at constant speed or constant power. When the axial force fusion estimation result is within the target range, fine-tuning control is performed based on model predictive control and micro-stepping strategy, and the controllable damping unit is simultaneously driven to absorb energy and stop according to the axial force climbing rate and the angle change rate. When the axial force fusion estimation result is higher than or equal to the second threshold and lower than the overload threshold, overshoot correction control is performed. The overshoot correction control includes stopping the actuator output, driving the controllable damping unit to absorb energy and implementing small-angle reversal or micro-step back to bring the axial force fusion estimation result back to the target range, and then performing fine-tuning control after returning to the target range. When the axial force fusion estimation result is determined to be higher than the overload threshold, the actuator output is immediately stopped, the controllable damping unit is released, and an alarm is issued. Here, the target range is the range between the first threshold and the second threshold, and the tolerance band is the allowable deviation range centered on the target axial force.

[0011] Furthermore, when the axial force fusion estimation result is lower than a first threshold of the target axial force, controlling the actuator to drive at constant speed or constant power includes: The system collects coarse-tightening control results and corresponding speed or power setpoints; it also collects angular velocity data from the angular encoder, torque data from the torque sensor, and temperature data from the temperature sensor; processes the angular velocity data and the speed setpoints to obtain speed error and power error; and obtains protection criteria based on a temperature rise threshold. When the coarse-tightening control result is constant speed, a speed closed-loop control result is obtained, which is used to adjust the actuator output so that the angular velocity tracks the speed setpoint and is controlled by a preset torque limit. When the coarse-tightening control result is constant power, a power closed-loop control result is obtained, which is used to adjust the actuator output so that the power tracks the power setpoint and is controlled by a preset angular velocity limit.

[0012] Furthermore, when performing fine-tuning control based on model predictive control and micro-stepping strategies, it includes: Collect axial force fusion estimation results, torque measurement data, rotation angle measurement data and temperature measurement data to obtain state vector data, axial force climb rate data and rotation angle change rate data; Based on the self-calibration model and historical response data within the sliding time window, the controlled object prediction model data is updated online. Cost function data is constructed, which includes a target axial force tracking error term, an overshoot penalty term, and a control increment penalty term. Torque upper limit, angular velocity upper limit, temperature rise upper limit, and predicted overshoot boundary are set. Candidate control sequence data are obtained by solving the problem, and the first control quantity data is obtained. The first control quantity data is quantized into several microstepping unit data through a microstepping strategy. Each microstepping unit data includes microstepping increment data and microstepping duration data, which are applied to the actuator in sequence to form a fine-tuning control result. After each microstepping unit data is executed, the measurement data is re-acquired and the prediction model data and cost function data are updated on a rolling basis.

[0013] Furthermore, when the controllable damping unit performs energy absorption and braking based on the axial force climb rate data and the rotation angle change rate data, it includes: When the axial force fusion estimation result data is within the target range and the axial force climb rate data is greater than the preset threshold, the damping command data in the pre-charge state or the gradual state is obtained to increase the damping proportionally; when the prediction model data indicates overshoot risk or the axial force fusion estimation result data is close to the second threshold, the damping command data in the brake state is obtained to stop the movement for a short time; when the axial force fusion estimation result data falls out of the target range or the axial force climb rate data and the angle change rate data both fall below the threshold, the damping command data in the release state is obtained to reduce the additional load.

[0014] Furthermore, the axial force fusion estimation results when the target axial force is reached are collected, the data during the holding phase is processed, and the axial force springback check results are obtained. If the axial force springback check results exceed the limit, a second tightening control is performed, including: Record the axial force fusion estimation result at the start of the holding period as the holding baseline data, control the controllable damping unit to briefly apply the brake and stop the actuator; within the preset holding period, collect the axial force fusion estimation result and temperature measurement data at a preset sampling frequency, preprocess the holding phase data to obtain holding correction data; at the end of the holding period, compare the holding correction data with the holding baseline data to obtain the axial force springback check result and compare it with the tolerance band: when the axial force springback check result is within the tolerance band, the holding period ends and is archived; when the axial force springback check result exceeds the tolerance band but is below the overload threshold, execute secondary tightening control, the secondary tightening control includes: placing the controllable damping unit in the precharge state, driving the actuator to tighten at a small angle according to the micro-step unit, and immediately checking the axial force fusion estimation result and tolerance band after each micro-step unit is completed, until entering the tolerance band or reaching the upper limit of the tightening step and the upper limit of the tightening angle; when the axial force springback check result reaches the overload threshold or the secondary tightening triggers the temperature rise limit, enter the protection control and record the event.

[0015] Compared with existing technologies, the advantages of this invention are as follows: By introducing multi-source acquisition and processing of bolt identification and process parameter data, temperature measurement data, and zero-load preload data, a self-calibration model is established, including sound velocity compensation parameters, elastic modulus compensation parameters, and zero-load baseline data, thereby effectively eliminating the influence of material batch differences and environmental fluctuations on the accuracy of axial force measurement; simultaneously, through the joint acquisition of axial force sensors and ultrasonic probes, a confidence-weighted axial force fusion estimation result is obtained, achieving higher robustness and reliability than a single ultrasonic detection path; in the control stage, not only is the staged control of coarse and fine tightening achieved based on the comparison between the axial force fusion estimation result and the target axial force data, but also, combined with model predictive control and micro-stepping strategies, a controllable damping unit is used for dynamic energy absorption and stopping when approaching the target range, thereby suppressing overshoot and springback, ensuring rapid, stable, and high-precision convergence of bolt preload; after the target axial force is reached, a holding stage data processing and springback verification mechanism is introduced, automatically performing secondary tightening when necessary, enabling the bolt tightening quality to have self-checking and compensation capabilities, ensuring the consistency and long-term reliability of the preload. It solves the problems of unstable ultrasonic coupling, lack of multi-source fusion, and coarse control strategy in existing technologies, and realizes the refinement and traceability of the bolt tightening process.

[0016] On the other hand, this application also provides a bolt tightening intelligent control system based on axial force measurement, for applying the above-mentioned bolt tightening intelligent control method based on axial force measurement, including: Controller, actuator, shaft force sensor A, shaft force sensor B, angle encoder, torque sensor, temperature sensor, controllable damping unit, and identification read / write unit; among which, The axial force sensor A is used to output axial force data; the axial force sensor B is used to output ultrasonic echo data; the rotation encoder is used to output rotation measurement data and angular velocity data; the torque sensor is used to output torque measurement data; the temperature sensor is used to output probe temperature data and ambient temperature data; the identification reading and writing unit is used to read bolt identification and process parameter data. The controller is electrically connected to the actuator, shaft force sensor A, shaft force sensor B, angle encoder, torque sensor, temperature sensor, controllable damping unit, and identification reading and writing unit; The controller is configured to collect bolt identification and process parameter data, collect probe temperature data and ambient temperature data, process the process parameter data and temperature data, and obtain sound velocity compensation parameters and elastic modulus compensation parameters. Collect zero-load preload data and process it to generate zero-load baseline data to obtain a self-calibration model; Axial force data and ultrasonic echo data were collected to obtain the axial force fusion estimation result; The axial force fusion estimation result is compared with the target axial force data. When the axial force fusion estimation result is lower than the first threshold of the target axial force, the actuator is controlled to drive at constant speed or constant power. When the axial force fusion estimation result is within the target range, fine-tuning control is performed based on model predictive control and micro-stepping strategy, and the controllable damping unit is driven to absorb energy and stop according to the axial force climb rate data and the angle change rate data. The axial force fusion estimation results when the target axial force is reached are collected, the data of the holding stage are processed, and the axial force springback check results are obtained. If the axial force springback check results exceed the limit, secondary tightening control is performed.

[0017] It is understandable that the above-mentioned intelligent control method and system for bolt tightening based on axial force measurement have the same beneficial effects, and will not be elaborated further here. Attached Figure Description

[0018] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart of a bolt tightening intelligent control method based on axial force measurement provided in an embodiment of the present invention; Figure 2The structural block diagram of the intelligent bolt tightening control system based on axial force measurement provided in the embodiments of the present invention is shown. Detailed Implementation

[0019] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] In the traditional bolt tightening control process, the measurement path of a single sensor is easily affected by environmental interference, which leads to inaccurate axial force detection. The lack of a fusion mechanism for multi-source sensor data can cause misjudgment of abnormal working conditions. The lack of dynamic buffering when switching between the control modes of the coarse tightening stage and the fine tightening stage can cause axial force overshoot or springback.

[0021] For example, in the assembly scenario of bolts on the rotor disk of a high-pressure compressor in an aero-engine, the bolt material is a nickel-based superalloy with a surface roughness grade of Ra0.8, and the thread lubrication is a dry molybdenum disulfide coating. During the assembly process, the ambient temperature fluctuates within a range of 25±15℃. Due to vibration, the coupling agent distribution at the contact interface between the ultrasonic probe and the bolt end face is uneven, resulting in an echo signal amplitude attenuation of up to 30%. During the coarse tightening stage, when driven at a constant speed, the axial force creep rate exceeds 5kN per second. When approaching the target axial force, the inertial torque is not reduced in time, causing the actual axial force to exceed the target value by 12%. During the holding stage, due to material creep and temperature hysteresis, the axial force rebound reaches 8% of the initial preload, exceeding the allowable range of the tolerance zone.

[0022] If the above problems are not resolved, the consistency of assembly quality will not meet the requirement of a coaxiality tolerance of 0.05mm for aero-engine rotor disks, and fretting wear may occur at the bolt connection interface, leading to a decrease in high-cycle fatigue life. Excessive axial force may cause plastic deformation of the threaded pair, reducing the reliability of repeated assembly. Excessive axial force rebound will cause localized stress relaxation on the flange sealing surface, inducing a risk of gas leakage under high-temperature and high-pressure conditions.

[0023] For this, please refer to Figure 1 As shown, this application proposes an intelligent control method for bolt tightening based on axial force measurement, comprising: S100: Collects bolt identification and process parameter data, collects temperature measurement data including probe temperature data and ambient temperature data, processes process parameter data and temperature data, and obtains sound velocity compensation parameters and elastic modulus compensation parameters.

[0024] S200: Collect zero-load preload data and process it to generate zero-load baseline data to obtain a self-calibration model.

[0025] S300: Acquires axial force data and ultrasonic echo data to obtain axial force fusion estimation results.

[0026] S400: Compares the axial force fusion estimation result with the target axial force data. When the axial force fusion estimation result is lower than the first threshold of the target axial force, it controls the actuator to drive at constant speed or constant power. When the axial force fusion estimation result is within the target range, it performs fine-tuning control based on model predictive control and micro-stepping strategy, and drives the controllable damping unit to absorb energy and stop according to the axial force climb rate data and the angle change rate data.

[0027] S500: Collects the axial force fusion estimation result when the target axial force is reached, processes the data during the holding phase, and obtains the axial force springback check result. If the axial force springback check result exceeds the limit, a second tightening control is performed.

[0028] Specifically, bolt identification refers to a unique code or label used to identify the bolt type or specification. This can be achieved using QR codes, RFID tags, or laser engraving, facilitating rapid retrieval of corresponding process parameters by the system. Process parameter data includes material code, heat treatment state, surface roughness grade, and thread lubrication state, which can be obtained from a preset database or process documents to determine the bolt's physical properties and friction conditions. Temperature measurement data includes probe temperature and ambient temperature, which can be acquired using thermocouples or infrared sensors to correct for temperature drift errors in sound velocity and elastic modulus. Sound velocity compensation parameters are correction coefficients that adjust the ultrasonic wave propagation speed based on temperature and material properties. These can be generated using a calibration index table and temperature correction algorithm to eliminate the influence of temperature on ultrasonic measurements. Elastic modulus compensation parameters correct for changes in the bolt material's elastic modulus due to temperature. These can be calculated using a thermal expansion coefficient model combined with real-time temperature data to ensure the accuracy of axial force estimation. Zero-load preload data refers to sensor reference data acquired under no-preload conditions, specifically including the axial force sensor zero-point signal and ultrasonic echo baseline, used to eliminate sensor drift and contact noise. The self-calibration model is a dynamic compensation model built based on zero-load data. It is generated through filtering, denoising, and steady-state determination, and is used to correct measurement errors and diagnose anomalies in real time. Before collecting axial force and ultrasonic echo data, various sensors are calibrated to ensure that their accuracy meets relevant sensor standards, thus guaranteeing the reliability of bolt detection. Axial force data refers to the bolt axial force directly measured by strain or pressure sensors, specifically resistance strain gauges or piezoelectric sensors, providing direct force feedback. Ultrasonic echo data refers to the axial force estimate derived from the reflected ultrasonic signal, specifically by measuring bolt length changes using the pulse-echo method to indirectly calculate the axial force. The axial force fusion estimation result is a weighted output combining axial force and ultrasonic echo data, generated through time alignment, confidence assessment, and a weighted fusion algorithm to improve measurement reliability and accuracy. The first threshold of the target axial force is the lower limit of the axial force triggering the coarse-tightening stage, determined based on the target axial force and a preset tolerance band, used to distinguish between the coarse-tightening and fine-tightening control stages. Among them, constant speed or constant power drive refers to the control mode of the actuator during the coarse tightening stage, specifically employing closed-loop speed or power control algorithms to quickly approach the target axial force range. Model predictive control refers to a control method based on a dynamic model to predict future axial force changes. Specifically, it generates a control sequence by updating the predictive model online and optimizing the cost function, achieving precise adjustment during the fine tightening process. Micro-stepping strategy refers to an execution method that decomposes the control quantity into tiny step increments, specifically employing stepper motor or servo motor microstepping drive technology to reduce overshoot risk.The controllable damping unit refers to a mechanical braking device with adjustable damping force, specifically employing an electromagnetic clutch or magnetorheological damper, which dynamically absorbs energy or stops the movement based on the rate of change of axial force. The axial force springback check result refers to the assessment value of the axial force attenuation during the holding phase, specifically generated by comparing the holding baseline data with real-time measurement data to determine whether secondary tightening is needed. Secondary tightening control refers to the compensation operation for excessive axial force springback, specifically employing micro-step increments and coordinated adjustment of controllable damping to ensure that the axial force remains stable within the tolerance range.

[0029] This application achieves high-precision closed-loop control of bolt axial force through multi-source data fusion and a phased dynamic control strategy. By integrating axial force sensors and ultrasonic measurement data, combined with temperature compensation and a self-calibration model, the robustness of axial force estimation is improved. Constant speed / constant power drive in the coarse-tightening stage and model predictive control in the fine-tightening stage, along with the energy absorption and stopping functions of a controllable damping unit, effectively suppresses overshoot and springback. Based on the axial force springback verification and secondary tightening mechanism in the holding stage, the final axial force is ensured to remain stable within the target tolerance range.

[0030] The working process and principle of this application are as follows: Bolt identification and process parameter data are collected, along with temperature measurement data, including probe temperature data and ambient temperature data. The process parameter data and temperature data are processed to obtain sound velocity compensation parameters and elastic modulus compensation parameters. This step, by considering the influence of temperature, corrects the sound velocity and elastic modulus, improving the accuracy of subsequent measurements.

[0031] Zero-load preload data is collected and processed to generate zero-load baseline data, thus obtaining a self-calibration model. Baseline data is established under zero-load conditions to provide a reference for subsequent axial force measurements.

[0032] Axial force data and ultrasonic echo data are collected to obtain a fused estimation result of axial force. By fusing data from the two measurement methods, the accuracy and reliability of axial force estimation are improved.

[0033] The axial force fusion estimation result is compared with the target axial force data, and different control strategies are adopted according to the comparison result. When the axial force fusion estimation result is lower than the first threshold of the target axial force, the actuator is controlled to drive at constant speed or constant power to quickly approach the target axial force. When the axial force fusion estimation result is within the target range, fine-tuning control is performed based on model predictive control and micro-stepping strategy to achieve precise control. At the same time, the controllable damping unit is driven to absorb energy and stop according to the axial force climb rate data and the angle change rate data to prevent overshoot.

[0034] The axial force fusion estimation results when the target axial force is reached are collected, the data during the holding phase are processed, and the axial force springback check results are obtained. If the axial force springback check result exceeds the limit, secondary tightening control is performed to ensure that the final axial force reaches the expected target.

[0035] As a preferred embodiment, the solution of this application is specifically implemented as follows: During the bolt assembly process of the high-pressure compressor rotor disk of an aero-engine, bolt markings and process parameter data were first collected. The process parameters included that the bolt material was a nickel-based high-temperature alloy, the surface roughness grade was Ra0.8, and the thread lubrication was a dry molybdenum disulfide coating. Temperature measurement data was also collected, with the ambient temperature at 25±15℃.

[0036] Processing process parameters and temperature data yields sound velocity compensation parameters and elastic modulus compensation parameters. Based on a preset calibration index table, the reference sound velocity and reference elastic modulus are corrected using temperature data.

[0037] Collect zero-load preload data and process it to generate zero-load baseline data. Control the actuator to enter zero-load preload mode to establish stable contact with the axial force sensor. Simultaneously collect zero-load signal data from each sensor, perform filtering, noise reduction, and drift subtraction to generate zero-load baseline data.

[0038] Axial force data and ultrasonic echo data were collected to obtain axial force fusion estimation results. Time alignment, steady-state determination, and outlier sample removal were performed on the two types of data. The confidence level was determined based on the signal-to-noise ratio, coupling quality index, and temperature stability, and confidence-weighted fusion was performed.

[0039] The axial force fusion estimation result is compared with the target axial force data. When the axial force fusion estimation result is lower than the first threshold of the target axial force, the actuator is controlled to drive at a constant speed. When entering the target range, fine-tuning control is performed based on model predictive control and micro-stepping strategy. At the same time, the controllable damping unit is driven to absorb energy and stop according to the axial force rise rate and the rate of change of rotation angle.

[0040] The axial force fusion estimation result when the target axial force is reached is collected, and the data during the holding phase is processed. The axial force fusion estimation result at the start of the holding phase is recorded as the holding baseline data, and the controllable damping unit is used to control the brake for a short time. Data is collected within the preset holding time to obtain the axial force rebound verification result. If the tolerance zone is exceeded, secondary tightening control is executed, and the actuator is driven by the micro-stepping unit to tighten at a small angle until the tolerance zone is entered.

[0041] Through the above-described scheme, this application solves the problem of inaccurate axial force detection caused by environmental interference in single-sensor measurement paths. By introducing a dual-axial force sensor collaborative measurement and confidence-weighted strategy, the measurement robustness under abnormal operating conditions is improved. The use of model predictive control combined with a micro-stepping strategy achieves a smooth transition between the coarse and fine tightening stages, avoiding axial force overshoot or springback. By establishing a self-calibration model under zero-load preload mode, interference from environmental temperature fluctuations and unstable coupling interfaces is effectively addressed. The designed data verification and secondary tightening mechanism during the holding stage solves the problem of excessive axial force springback. These improvements ensure the accuracy and consistency of the preload of bolted connections, improving the assembly quality and reliability in key areas such as aero-engines.

[0042] This application further proposes a method for processing process parameter data and temperature data to obtain sound velocity compensation parameters and elastic modulus compensation parameters, including: retrieving reference sound velocity and reference elastic modulus from a preset calibration index table based on the material code, heat treatment state, surface roughness grade, and thread lubrication state in the process parameter data; estimating the surface temperature of the test piece by combining probe temperature data and ambient temperature data, and performing temperature correction on the reference sound velocity and reference elastic modulus; and introducing a coupling correction factor based on the surface roughness grade and thread lubrication state to correct the temperature-corrected sound velocity and elastic modulus, thereby obtaining the sound velocity compensation parameters and elastic modulus compensation parameters.

[0043] The material code and heat treatment status are used to match the pre-stored material property parameters in the calibration index table, ensuring that the initial values ​​of the reference sound velocity and elastic modulus match the actual bolt material. The surface roughness grade and thread lubrication status are used to determine the coupling correction factor to compensate for the influence of the contact surface condition on ultrasonic signal propagation. The probe temperature and ambient temperature are estimated using a weighted average or thermal conduction model to determine the surface temperature of the test piece, and then linear or nonlinear corrections are applied to the reference parameters based on the temperature coefficient.

[0044] Specifically, the calibration index table establishes a mapping relationship between material properties and process parameters using historical experimental data, reducing human error when retrieving benchmark parameters. During temperature correction, the probe temperature reflects the sensor's own thermal drift, while the ambient temperature reflects the external thermal environment. Combining these two factors allows for a more accurate estimation of the surface temperature of the measured part, avoiding correction deviations caused by a single temperature source. The coupling correction factor adjusts the sound velocity according to the contact gap corresponding to the surface roughness level and the elastic modulus according to the lubrication state. For example, a negative correction is introduced for high-roughness surfaces to compensate for sound wave attenuation, while a positive correction is introduced for well-lubricated surfaces to reflect the effect of reduced contact friction. Through this multi-dimensional compensation mechanism, the sound velocity and elastic modulus parameters can adapt to different working conditions, providing high-precision input for subsequent axial force fusion estimation.

[0045] As a preferred embodiment, the solution of this application is specifically implemented as follows: When processing process parameter data and temperature data to obtain sound velocity compensation parameters and elastic modulus compensation parameters, the following steps are included: First, based on the material designation, heat treatment state, surface roughness grade, and thread lubrication state in the process parameter data, the reference sound velocity and reference elastic modulus are retrieved from a preset calibration index table. For example, for a high-strength alloy steel bolt with the material designation 40CrNiMoA, heat treatment state of quenching and tempering, surface roughness grade of Ra1.6, and thread lubrication state of dry lubrication, the reference sound velocity of this bolt is found to be 5930 m / s and the reference elastic modulus is 210 GPa, according to the preset calibration index table.

[0046] Secondly, by combining probe temperature data with ambient temperature data, the surface temperature of the test piece is estimated, and temperature corrections are applied to the reference sound velocity and reference elastic modulus. Specifically, a heat conduction model is used to calculate the surface temperature of the test piece, and then the reference sound velocity and reference elastic modulus are corrected based on the material's temperature coefficient of sound velocity and temperature coefficient of elastic modulus.

[0047] Finally, a coupling correction factor is introduced based on the surface roughness grade and thread lubrication condition to correct the temperature-corrected sound velocity and elastic modulus, thus obtaining the sound velocity compensation parameters and elastic modulus compensation parameters. For example, for a surface roughness grade Ra1.6 and a dry lubrication condition, a coupling correction factor of 0.98 is introduced. The temperature-corrected sound velocity and elastic modulus are multiplied by this factor to obtain the final sound velocity compensation parameters and elastic modulus compensation parameters.

[0048] Through the above technical solution, this application can accurately obtain sound velocity compensation parameters and elastic modulus compensation parameters, thereby improving the accuracy of bolt axial force measurement. By considering multiple influencing factors such as material properties, heat treatment state, surface roughness, lubrication state, and temperature, the compensation parameters more closely reflect actual working conditions. This compensation method effectively reduces the impact of environmental factors and process parameter variations on axial force measurement, providing reliable basic data for subsequent axial force fusion estimation.

[0049] In some of the above-mentioned schemes in this application, when collecting zero-load preload data and generating zero-load baseline data, due to environmental temperature fluctuations, sensor drift or contact instability, zero-point offset, time delay or noise interference are not effectively eliminated, affecting the accuracy of the self-calibration model, and thus reducing the reliability of the axial force fusion estimation results.

[0050] This application further proposes a control mechanism to enter a zero-load preload mode, driving axial force sensor A and axial force sensor B to establish stable contact without generating effective preload force below a preset zero-load preload upper limit threshold. Zero-load signal data from axial force sensor A, zero-load echo data from axial force sensor B, zero-position data from the angle encoder, zero-point data from the torque sensor, and zero-load temperature data from the temperature sensor are simultaneously acquired to form zero-load preload data. The zero-load preload data is filtered, denoised, and drift subtracted, and a steady-state determination is performed according to a preset steady-state determination threshold. If the steady-state determination passes, the zero-point offset, time delay offset, noise threshold, gain correction coefficient, and coupling quality index are calculated to generate zero-load baseline data. The self-calibration model is updated using the zero-load baseline data combined with sound velocity compensation parameters and elastic modulus compensation parameters. The self-calibration model is used for axial force fusion estimation and anomaly diagnosis. When the steady-state determination fails or the coupling quality index is below the threshold, the zero-load preload and contact posture are automatically adjusted, and the acquisition and processing are repeated until zero-load baseline data meeting the threshold conditions is generated and bound to the bolt identification for storage.

[0051] The zero-load preload mode eliminates gap errors by limiting the output force of the drive mechanism, allowing axial force sensor A and axial force sensor B to make physical contact without effective preload. Simultaneous acquisition of multi-sensor data covers the combined effects of zero-point drift, temperature drift, and mechanical delay. Filtering and denoising employ a combination of moving average and low-pass filtering, with a preset cutoff frequency of 10Hz. Steady-state determination is based on variance analysis, with a preset variance threshold of 0.5% of full scale. Zero-point offset is calculated using an arithmetic mean, and time delay offset uses a cross-correlation algorithm to calculate signal alignment offset. Coupling quality indicators are calculated from ultrasonic echo amplitude and signal-to-noise ratio, with a preset threshold of 60dB. Automatic adjustment of zero-load preload includes adjusting actuator pressure or contact angle, with a maximum of 5 repeated acquisitions. Binding storage associates zero-load baseline data with bolt identification, forming a calibration database.

[0052] Specifically, the actuator drives the bolt to contact the sensor with a preload of less than 5N, simultaneously acquiring the voltage signal from axial force sensor A, the peak time of the ultrasonic echo from axial force sensor B, the zero-position pulse from the angle encoder, the zero-point voltage from the torque sensor, and the real-time temperature data from the temperature sensor. The acquired raw data is processed by a second-order Butterworth low-pass filter with a cutoff frequency set to 50Hz to remove high-frequency noise. Drift subtraction uses a linear fitting method to eliminate baseline drift of the sensors over time. In the steady-state determination phase, the variance of the data over 10 consecutive sampling periods is calculated; if the variance is less than 0.05N², it is determined to be in a steady state. Based on the steady-state determination data, the zero-point offset is calculated to be 1.2mV, the time delay offset to be 0.8ms, the noise threshold to be 0.3mV, the gain correction coefficient to be 1.05, and the coupling quality index to be 65dB. When the coupling quality index is lower than 60dB, the actuator adjusts the contact angle in 0.5° steps and re-acquires data until the index meets the standard. The updated self-calibration model applies zero-point offset and gain correction coefficients to the real-time signal of axial force sensor A, uses time delay offset for time alignment of the ultrasonic echo signal, and employs a noise threshold as the basis for outlier removal. The bound and stored calibration data can be directly retrieved when tightening bolts of the same type subsequently, reducing repetitive calibration time.

[0053] As a preferred embodiment, the solution of this application is specifically implemented as follows: The actuator is controlled to enter a zero-load preload mode, driving the axial force sensor A and axial force sensor B to establish stable contact without generating effective preload force below a preset zero-load preload upper limit threshold. For example, if the zero-load preload upper limit threshold is set to 100N, the actuator is controlled to slowly load at a rate of 5N / s until the contact force reaches 80N.

[0054] Simultaneously acquire zero-load signal data from axial force sensor A, zero-load echo data from axial force sensor B, zero-position data from the angle encoder, zero-point data from the torque sensor, and zero-load temperature data from the temperature sensor to form zero-load preload data. Specifically, continuously acquire data for 5 seconds at a sampling frequency of 1kHz to obtain 5000 sets of data samples.

[0055] The zero-load preload data is filtered, denoised, and has drift subtracted, and a steady-state determination is performed according to a preset steady-state determination threshold. Further, a 20Hz low-pass filter is used to filter the original data, median filtering is used to remove impulse noise, and the linear drift trend is fitted and subtracted using the least squares method. The steady-state determination threshold is set to a signal standard deviation not exceeding 0.1% of full scale.

[0056] Upon successful steady-state determination, zero-point bias, time delay bias, noise threshold, gain correction coefficient, and coupling quality index are calculated to generate zero-load baseline data. The zero-point bias is obtained by taking the mean of 5000 samples; the time delay bias is obtained through cross-correlation analysis of the two sensor signals; the noise threshold is three times the signal standard deviation; the gain correction coefficient is determined by the linear regression slope of the two sensor outputs; and the coupling quality index is characterized by the ratio of the ultrasonic echo amplitude to the baseline value.

[0057] The self-calibration model is updated using zero-load baseline data combined with sound velocity compensation parameters and elastic modulus compensation parameters. This model is used for axial force fusion estimation and anomaly diagnosis. When the steady-state condition fails or the coupling quality index falls below a threshold, the zero-load preload and contact attitude are automatically adjusted, and data acquisition and processing are repeated until zero-load baseline data meeting the threshold conditions is generated and bound to bolt identifiers for storage. For example, if the coupling quality index is below 0.8, the preload is increased by 10N and data is reacquired. If the standard is still not met after three consecutive adjustments, a manual intervention process is triggered.

[0058] Through the above technical solution, this application achieves automatic acquisition and processing of zero-load baseline data during bolt tightening, improving the accuracy and reliability of axial force measurement. By introducing multiple compensation parameters and a self-calibration model, the influence of environmental factors and changes in contact state on the measurement results is effectively overcome. Simultaneously, an iterative optimization strategy ensures the acquisition of high-quality baseline data, laying the foundation for subsequent axial force fusion estimation. This method enhances the intelligence level and control precision of the bolt tightening process, making it particularly suitable for high-reliability applications such as aerospace.

[0059] This application further proposes a method for acquiring axial force data and ultrasonic echo data to obtain axial force fusion estimation results, including: obtaining axial force correction data from axial force sensor A; obtaining ultrasonic axial force estimation data from axial force sensor B; performing time alignment, steady-state determination, and outlier removal on the axial force correction data and ultrasonic axial force estimation data within a sliding time window; determining the confidence level based on signal-to-noise ratio, coupling quality index, and temperature stability; and performing confidence-weighted fusion of the two types of data to obtain the axial force fusion estimation result. When the signal-to-noise ratio or coupling quality index of axial force sensor B is lower than a threshold, the weight of axial force sensor B is reduced or the output of axial force sensor A is switched to single-channel output. When axial force sensor A experiences saturation or drift exceeds limits, the output of axial force sensor B is used, triggering the re-establishment of zero-load baseline data.

[0060] In this process, time alignment is achieved by matching the sampling times of the two types of data within a sliding time window, eliminating the phase difference caused by sensor response delay. Steady-state determination is based on whether the variance of the data within the window is below a preset threshold; windows with variance exceeding the threshold are marked as abnormal samples and removed. In the confidence-weighted fusion, the signal-to-noise ratio (SNR) is obtained by calculating the ratio of signal amplitude to background noise; the coupling quality index is derived from the contact posture parameters in the self-calibration model; and temperature stability is evaluated based on the deviation between real-time temperature sensor data and a preset temperature drift curve. Weight allocation uses linear interpolation; SNR, coupling quality index, and temperature stability each correspond to three independent weight coefficients, and their product is used as the final fusion weight.

[0061] Specifically, the axial force correction data output by axial force sensor A undergoes zero-point offset and gain correction processing, while the ultrasonic axial force estimation data output by axial force sensor B undergoes sound velocity compensation and elastic modulus compensation correction. Within the sliding time window, the timestamps of both types of data are synchronized to the same reference, and the window length is dynamically adjusted according to the drive frequency of the actuator. During steady-state determination, if the fluctuation range of the axial force correction data within the window exceeds the preset steady-state threshold, it is determined to be a dynamic process and fusion is paused, retaining only the steady-state window data for subsequent processing. Outlier removal employs an outlier detection algorithm, marking data points that deviate from the mean by more than three standard deviations as outliers. During confidence calculation, when the signal-to-noise ratio is below 30dB or the coupling quality index is below 0.7, the weight of axial force sensor B is set to 0.3. When the temperature fluctuation exceeds ±5℃, the weight is further reduced by 0.1. During the fusion process, if the measured value of axial force sensor A exceeds 90% of its range or the zero drift exceeds 2% of the full scale, sensor B will be triggered to enter single-channel output mode and simultaneously send a command to the self-calibration model to re-establish zero-load baseline data.

[0062] As a preferred embodiment, the solution of this application is specifically implemented as follows: When acquiring axial force data and ultrasonic echo data to obtain the axial force fusion estimation result, the following steps are included: First, axial force correction data is obtained from axial force sensor A. Axial force sensor A can be a strain gauge or piezoelectric sensor, installed on the bolt head or the bottom of the nut, to directly measure the axial compressive force.

[0063] Secondly, ultrasonic axial force estimation data is obtained from axial force sensor B. Axial force sensor B is an ultrasonic probe that indirectly estimates the axial force by measuring the change in the propagation time of ultrasonic waves in the bolt.

[0064] Then, within a sliding time window, the axial force correction data and ultrasonic axial force estimation data are time-aligned, steady-state determined, and outlier samples are removed. The time window can be set to 0.5 seconds, sliding once every 50 milliseconds. Time alignment uses a cross-correlation algorithm, steady-state determination is based on analysis of variance, and outlier sample removal uses the 3σ criterion.

[0065] Next, the confidence level was determined based on the signal-to-noise ratio, coupling quality index, and temperature stability. The signal-to-noise ratio threshold was set to 20 dB, the coupling quality index threshold was 0.8, and the temperature stability requirement was a fluctuation within ±1°C.

[0066] Subsequently, confidence-weighted fusion was performed on the two types of data to obtain the axial force fusion estimation result. A Kalman filter algorithm was used for data fusion, with the fusion weights dynamically adjusted according to their respective confidence levels.

[0067] Furthermore, when the signal-to-noise ratio (SNR) or coupling quality index of axial force sensor B falls below a threshold, the weight of axial force sensor B is reduced or the output of axial force sensor A is switched to single-channel output. The SNR threshold is 15 dB, and the coupling quality index threshold is 0.6.

[0068] Finally, when axial force sensor A saturates or drifts beyond its limit, axial force sensor B outputs and triggers the re-establishment of zero-load baseline data. The saturation threshold is 95% of full scale, and the drift threshold is 1% / hour.

[0069] Through the above technical solutions, this application achieves fusion estimation of axial force data and ultrasonic echo data, improving the accuracy and reliability of axial force measurement. The stability of measurement results is enhanced through sliding time window processing, multiple judgments, and anomaly removal. The mechanism of dynamically adjusting fusion weights and switching measurement channels improves the system's adaptability and robustness. The automatic reconstruction function of zero-load baseline data ensures measurement accuracy during long-term use. These improvements effectively solve the problems of susceptibility to interference and large accuracy fluctuations associated with single measurement methods, providing more reliable axial force feedback for the bolt tightening process.

[0070] In some of the solutions described above in this application, when the axial force fusion estimation result is close to the target axial force, the use of a single control strategy may lead to overshoot or insufficient backoff, making it impossible to balance accuracy and stability during dynamic adjustment, thereby affecting the consistency of the final preload.

[0071] This application further proposes the following steps when comparing the axial force fusion estimation result with the target axial force data: When the axial force fusion estimation result is lower than a first threshold of the target axial force, control the actuator to drive at constant speed or constant power. When the axial force fusion estimation result is within the target range, perform fine-tuning control based on model predictive control and micro-stepping strategy, and simultaneously drive the controllable damping unit to absorb energy and stop according to the axial force climb rate and the angle change rate. When the axial force fusion estimation result is higher than or equal to a second threshold but lower than an overload threshold, perform overshoot correction control. Overshoot correction control includes stopping the actuator output, driving the controllable damping unit to absorb energy, and implementing small-angle reversal or micro-step back to bring the axial force fusion estimation result back to the target range. After returning to the target range, perform fine-tuning control. When the axial force fusion estimation result is determined to be higher than the overload threshold, immediately stop the actuator output, release the controllable damping unit, and issue an alarm. Here, the target range is the range between the first threshold and the second threshold, and the tolerance band is the allowable deviation range centered on the target axial force.

[0072] In this system, constant speed or constant power drive achieves rapid axial force increase during the roughing stage through speed or power closed-loop control. Model predictive control combines the predicted model of the controlled object with a cost function to optimize the control sequence online. The micro-stepping strategy decomposes the control quantity into several micro-stepping units, each containing quantized parameters for increment and duration. The controllable damping unit dynamically adjusts the damping state based on the axial force climb rate and the rate of change of rotation angle, switching between pre-charge, gradual entry, brake holding, and release states. Overshoot correction control suppresses axial force overshoot by combining reverse rotation or micro-stepping with damping energy absorption. The tolerance band is set based on the target axial force and the allowable deviation range and is used to check axial force rebound.

[0073] Specifically, during the roughing stage, when the axial force fusion estimate is below the first threshold, the actuator is driven at constant speed or constant power, rapidly increasing the axial force through a speed closed loop or power closed loop, while being constrained by torque and angular velocity protection thresholds. Upon entering the target range, model predictive control generates a control sequence based on the predictive model and cost function. The micro-stepping strategy decomposes the control sequence into multiple micro-stepping units, with the execution time and increment of each unit quantized to reduce control jitter. The controllable damping unit adjusts the damping state according to the real-time axial force climb rate and angle change rate. When the rate exceeds the threshold, damping is increased; when overshoot risk is predicted, the brake is triggered. In the event of overshoot, the actuator stops outputting, the controllable damping unit absorbs energy, and the axial force is brought back to the target range through small-angle reversal or micro-step back. If the axial force exceeds the overload threshold, the actuator immediately stops, the damping is released, and an alarm is triggered. The tolerance band, centered on the target axial force, allows for a range of permissible deviations to maintain the axial force rebound during the stage verification, ensuring the final preload is within the allowable error range.

[0074] As a preferred embodiment, the solution of this application is specifically implemented as follows: When comparing the axial force fusion estimation results with the target axial force data, the following steps are included: When the axial force fusion estimation result is lower than a first threshold of the target axial force, the actuator is controlled to drive at a constant speed or constant power. For example, if the first threshold is set to 80% of the target axial force, when the axial force fusion estimation result is lower than this threshold, the actuator is controlled to drive at a constant speed of 2000 rpm or a constant power of 500W.

[0075] When the axial force fusion estimation result is within the target range, fine-tuning control is performed based on model predictive control and a micro-stepping strategy. Specifically, a model predictive control algorithm with a rolling time domain of 10s is adopted, with a prediction step size of 0.5s and a control cycle of 0.1s. Within each control cycle, the predicted control quantity is divided into 5 micro-stepping units, each with a duration of 20ms. Simultaneously, the controllable damping unit is driven to absorb energy and stop according to the axial force climb rate and the rate of change of rotation angle.

[0076] When the axial force fusion estimation result is higher than or equal to the second threshold and lower than the overload threshold, overshoot correction control is performed. The second threshold is set to 98% of the target axial force, and the overload threshold is set to 110% of the target axial force. Overshoot correction control includes stopping the actuator output, driving the controllable damping unit to absorb energy, and implementing a small-angle reverse or micro-step back to bring the axial force fusion estimation result back to the target range. After returning to the target range, fine-tuning control is performed. For example, when overshoot is detected, a 5° reverse micro-step back is executed, with each step being 0.5° and an interval of 10ms.

[0077] When the axial force fusion estimation result is determined to be higher than the overload threshold, the actuator output is immediately stopped, the controllable damping unit is released, and an alarm is issued. The target range is the interval between the first threshold and the second threshold, and the tolerance band is the allowable deviation range of ±2% centered on the target axial force.

[0078] Through the above technical solution, this application achieves refined control of the bolt tightening process. Differentiated control strategies are employed at different tightening stages to effectively avoid axial force overshoot or undershoot. The introduction of a controllable damping unit improves the system's response speed to axial force changes and enhances the stability of the tightening process. The combination of model predictive control and a micro-stepping strategy allows the system to approach the target axial force more smoothly, reducing oscillations and overshoot. An overshoot correction control mechanism further ensures tightening accuracy, correcting even slight overshoot in a timely manner. Furthermore, setting an overload protection threshold effectively prevents bolts from being damaged by excessive axial force. Overall, this solution improves the accuracy and reliability of bolt tightening, providing strong support for demanding industrial applications.

[0079] In some of the solutions described above in this application, during the constant speed or constant power drive process in the coarse tightening stage, there is a lack of dynamic adjustment and protection mechanisms for the output state of the actuator, which may lead to overload or excessive temperature rise, affecting control accuracy and equipment safety.

[0080] This application further proposes a method for controlling the actuator to drive at constant speed or constant power when the axial force fusion estimation result is lower than a first threshold of the target axial force. This includes: acquiring the coarse-tightening control result and the corresponding speed setpoint or power setpoint; acquiring angular velocity data from the angular encoder, torque data from the torque sensor, and temperature data from the temperature sensor; processing the angular velocity data and speed setpoint to obtain the speed error; obtaining the power error; and obtaining a protection criterion based on the temperature rise threshold. When the coarse-tightening control result is constant speed, a speed closed-loop control result is obtained, which is used to adjust the actuator output so that the angular velocity tracks the speed setpoint and is controlled by a preset torque limit. When the coarse-tightening control result is constant power, a power closed-loop control result is obtained, which is used to adjust the actuator output so that the power tracks the power setpoint and is controlled by a preset angular velocity limit.

[0081] The speed closed-loop control generates an error signal by comparing the measured angular velocity value with the set value in real time. This error signal is processed by a proportional-integral algorithm and output to the actuator. The actuator adjusts the input voltage or current of the drive motor according to the adjustment signal, gradually converging the actual angular velocity to the set value range. The power closed-loop control calculates real-time power by collecting motor current and voltage data. It compares the measured power value with the set value to generate an error signal. This error signal is then limited and drives the actuator to adjust the output torque, ensuring the power remains stable within the allowable range. The temperature rise threshold protection criterion predicts the internal temperature change trend of the actuator by monitoring temperature sensor data and historical temperature rise curves. When the predicted temperature exceeds a preset safety threshold, it triggers a load reduction or shutdown command to prevent overheating damage.

[0082] Specifically, during the constant-speed drive phase of the coarse-tightening stage, the angular encoder provides real-time feedback of angular velocity data. The controller inputs the measured angular velocity value and the set value into the error calculation module to generate a speed error signal. This error signal is processed by the proportional-integral controller to generate a drive signal, which drives the motor of the actuator to adjust its output speed. Simultaneously, the torque sensor monitors the output torque in real-time. When the torque exceeds a preset upper limit, the torque limiting module is triggered, forcibly reducing the amplitude of the drive signal to prevent mechanical overload. In constant-power drive mode, the power calculation module obtains the real-time power based on the product of the motor current and voltage. After comparing this with the set power, a power error signal is generated. After amplitude limiting, the motor is driven to adjust its output torque to ensure stable power. Temperature sensor data is input into the temperature rise prediction model. When the predicted temperature approaches the threshold, the controller gradually reduces the duty cycle or frequency of the drive signal to reduce heat generation. By dynamically adjusting the output state of the actuator, both coarse-tightening efficiency and equipment damage caused by overload and excessive temperature rise are ensured. As a preferred embodiment, the solution of this application is implemented as follows: When the axial force fusion estimation result is lower than the first threshold of the target axial force, the control actuator is driven at constant speed or constant power. Specifically, firstly, the coarse turning control result and the corresponding speed setpoint or power setpoint are collected. Then, angular velocity data from the angle encoder, torque data from the torque sensor, and temperature data from the temperature sensor are collected. Next, the angular velocity data and speed setpoint are processed to obtain the speed error, the power error, and a protection criterion based on the temperature rise threshold.

[0083] Furthermore, when the coarse-tightening control result is constant speed, the speed closed-loop control result is obtained. The speed closed-loop control result is used to adjust the actuator output so that the angular velocity tracks the speed setpoint and is controlled by a preset torque limit. For example, a PID controller can be used to implement speed closed-loop control, where the proportional gain is set to 0.5, the integral gain to 0.1, and the derivative gain to 0.05. The preset torque limit can be set to 80% of the rated torque.

[0084] When the coarse-tightening control result is constant power, the power closed-loop control result is obtained. The power closed-loop control result is used to adjust the actuator output so that the power tracks the power setpoint and is controlled by a preset angular velocity limit. As a preferred implementation, a fuzzy PID controller can be used to implement power closed-loop control, where the fuzzy rules adjust the PID parameters online based on the power error and the rate of change of the power error. The preset angular velocity limit can be set to 90% of the rated angular velocity.

[0085] Through the above technical solution, this application achieves precise control during the rough tightening stage. This allows for flexible selection of constant speed or constant power drive modes according to actual needs, improving the adaptability of the tightening process. Furthermore, by introducing closed-loop control of speed and power errors, as well as torque and angular velocity limits, the stability and safety of the tightening process are ensured. Simultaneously, a protection criterion based on a temperature rise threshold effectively prevents overheating of the actuator. This intelligent control method improves the accuracy and reliability of bolt tightening, laying a solid foundation for the subsequent fine tightening stage.

[0086] This application further proposes to collect axial force fusion estimation results, torque measurement data, rotation angle measurement data, and temperature measurement data to obtain state vector data, axial force climb rate data, and rotation angle change rate data. Based on the self-calibration model and historical response data within the sliding time window, the controlled object prediction model data is updated online. Cost function data is constructed, including the target axial force tracking error term, overshoot penalty term, and control increment penalty term, and upper limits for torque, angular velocity, temperature rise, and prediction overshoot boundary are set. Candidate control sequence data is obtained by solving the problem, and the first control quantity data is obtained. The first control quantity data is quantized into several microstepping unit data through a microstepping strategy. Each microstepping unit data contains microstepping increment data and microstepping duration data, which are applied sequentially to the actuator to form a fine-tuning control result. After each microstepping unit data is executed, measurement data is re-acquired, and the prediction model data and cost function data are updated continuously.

[0087] The state vector data comprises axial force fusion estimation results, torque measurement data, rotation angle measurement data, and temperature measurement data, used to characterize the current tightening state. The controlled object prediction model data is updated online using historical response data within a sliding time window to reflect dynamic changes. The cost function data introduces overshoot and control increment penalty terms, ensuring that the constraint torque and angular velocity do not exceed safety thresholds. The microstepping unit data discretizes the continuous control quantity into small step sizes, reducing the amplitude of single-step actions.

[0088] Specifically, during the fine-tuning stage, state vector data is collected in real time and input into the prediction model. The prediction model dynamically adjusts parameters based on historical data to improve prediction accuracy. During the optimization process of the cost function, the target axial force tracking error term ensures that the axial force converges to the target range, the overshoot penalty term suppresses the risk of exceeding the second threshold, and the control increment penalty term limits sudden changes in torque and angular velocity. After solving for the candidate control sequence, only the first control variable is used, and it is decomposed into multiple micro-stepping units using a micro-stepping strategy. Each unit contains a small rotational increment and execution time. For example, a control variable can be decomposed into five micro-stepping units, each driving the actuator to rotate 0.1 degrees for 10 milliseconds. After each micro-stepping unit is completed, data is re-collected and the prediction model and cost function are updated, forming a closed-loop control. The rolling optimization mechanism enables the prediction model to adapt to dynamic changes in the system in real time, and the micro-stepping strategy reduces the amplitude of single-step movements, avoids overshoot, and improves control stability.

[0089] As a preferred embodiment, the solution of this application is specifically implemented as follows: When performing precision control based on model predictive control and micro-stepping strategy, the following steps are first taken: axial force fusion estimation results, torque measurement data, rotation angle measurement data, and temperature measurement data are collected to obtain state vector data, axial force climb rate data, and rotation angle change rate data. For example, the acquisition frequency is set to 1000Hz, and the acquisition time window is 0.5 seconds.

[0090] Furthermore, based on the self-calibration model and historical response data within the sliding time window, the predicted model data of the controlled object is updated online. Specifically, recursive least squares method is used to estimate the model parameters in real time, and the predicted model adopts a second-order linear model.

[0091] Therefore, cost function data is constructed, including a target axial force tracking error term, an overshoot penalty term, and a control increment penalty term, and upper limits are set for torque, angular velocity, temperature rise, and predicted overshoot boundary. For example, the weight of the target axial force tracking error term is set to 1, the weight of the overshoot penalty term is set to 10, and the weight of the control increment penalty term is set to 0.1. The upper limit for torque is set to 120% of the rated torque, the upper limit for angular velocity is set to 5 rad / s, the upper limit for temperature rise is set to 50°C, and the predicted overshoot boundary is set to 105% of the target axial force.

[0092] The candidate control sequence data is obtained by solving the problem, and the first control variable data is obtained. Specifically, a quadratic programming algorithm is used to solve the optimization problem, with the prediction time domain length set to 10 steps and the control time domain length set to 3 steps.

[0093] The initial control input data is quantized into several microstepping unit data using a microstepping strategy. Each microstepping unit data includes microstepping increment data and microstepping duration data, which are applied sequentially to the actuator to form a fine-tuning control result. For example, the control input is divided into 5 microstepping units, with each unit having a microstepping increment of 0.2° and a microstepping duration of 50ms.

[0094] After each microstepping unit completes its data execution, measurement data is reacquired, and the prediction model data and cost function data are updated continuously. This achieves closed-loop feedback control, ensuring control accuracy and robustness.

[0095] Through the above technical solution, this application realizes intelligent control of bolt tightening based on axial force measurement. By employing model predictive control and a micro-stepping strategy, the future state of the system can be accurately predicted and optimally controlled, effectively avoiding problems such as axial force overshoot or under-tightening. Simultaneously, the micro-stepping strategy enables refined control, improving tightening accuracy. Furthermore, by updating the predictive model and cost function in real time, the system can adapt to different working conditions and environmental changes, improving the robustness and adaptability of the control. Ultimately, this method improves the accuracy and reliability of bolt tightening, providing an effective solution for bolt connection quality control in critical areas.

[0096] This application further proposes a method for driving the controllable damping unit to absorb energy and stop based on axial force climb rate data and angle change rate data, including: when the axial force fusion estimation result data is within the target range and the axial force climb rate data is greater than a preset threshold, obtaining pre-charge or gradual damping command data to proportionally increase damping; when the prediction model data indicates overshoot risk or the axial force fusion estimation result data is close to a second threshold, obtaining brake-state damping command data for short-term stopping; when the axial force fusion estimation result data falls outside the target range or the axial force climb rate data and angle change rate data simultaneously drop below the threshold, obtaining release-state damping command data to reduce additional load.

[0097] The trigger conditions for the pre-charge or gradual damping command are set to the axial force ramp rate exceeding 5%-10% of the target axial force per second, at which point the damping coefficient increases linearly or exponentially. The trigger conditions for the brake-activated damping command are configured to the overshoot probability output by the predictive model exceeding 60% or the axial force value reaching 95% of the second threshold, at which point the damping coefficient instantaneously increases to its maximum value and remains for 50-200 milliseconds. The trigger conditions for the release-activated damping command are set to the axial force ramp rate and angle change rate simultaneously falling below 2% of the target axial force per second or without an upward trend detected for 3-5 control cycles, at which point the damping coefficient decreases to its initial value with a preset slope.

[0098] Specifically, during the fine-tuning control phase, the axial force fusion estimation result and the overshoot risk output by the prediction model are simultaneously input into the damping control module. When the axial force climb rate exceeds 8% of the target axial force per second, the pre-charge damping is activated, and the damping coefficient increases at a rate of 0.5% per millisecond, causing the kinetic energy of the actuator to be partially converted into heat energy. If the prediction model calculates an overshoot risk within the next three control cycles based on historical data, the brake-state damping is immediately activated, completely absorbing the rotational inertia within 80 milliseconds through the electromagnetic brake, preventing the axial force from exceeding the second threshold. When the axial force value falls back to the target range and both the climb rate and the rate of change of angle are less than 1.5% of the target axial force per second, the release-state damping restores the damping coefficient to the initial level within 100 milliseconds, avoiding residual damping interference with subsequent micro-stepping operations. This process controls the axial force overshoot amplitude within ±2% of the tolerance zone by real-time matching of damping strength and dynamic changes in axial force.

[0099] As a preferred embodiment, the solution of this application is implemented as follows: During the fine-tuning control stage, when the axial force fusion estimation result enters the target range and the axial force creep rate exceeds 30 kN / s, the controller generates a pre-charge damping command, causing the hydraulic valve of the controllable damping unit to open to 20%. As the axial force creep rate continues to increase to 50 kN / s, the damping command switches to a gradual entry state, and the hydraulic valve opening linearly increases to 60%. When the prediction model identifies an overshoot risk or the axial force fusion estimation result reaches 95% of the target axial force, the controller issues a brake-holding command, the hydraulic valve closes instantaneously, and the electromagnetic brake applies a braking torque of 80 Nm for 0.5 seconds. After the brake-holding state ends, if the axial force creep rate and the rate of change of rotation angle both decrease to below 10 kN / s and 5 degrees / s respectively, the controller switches the damping unit to a release state, the hydraulic valve opening returns to 10%, and the electromagnetic brake is released. When the axial force fusion estimation result falls out of the target range due to rebound, the controller immediately switches the damping unit to the pre-charge state and re-executes micro-step control.

[0100] Through the above technical solution, this application achieves precise matching between the dynamic response of the damping unit and the trend of axial force variation. During the rapid rise of axial force, inertial overshoot is effectively suppressed by increasing damping in stages. When approaching the target axial force threshold, a short-term braking action eliminates residual kinetic energy, avoiding reverse adjustment caused by overshoot. Once the axial force enters a stable phase, reducing damping can reduce the impact of additional loads on measurement accuracy. This solution solves the problem of axial force fluctuation caused by single threshold control in existing technologies, and improves the control stability during the fine-tuning stage.

[0101] This application further proposes a method for driving a controllable damping unit to absorb energy and stop based on axial force climbing rate data and rotation angle change rate data.

[0102] The controllable damping unit includes four damping command states: pre-charge, gradual entry, brake engagement, and release. The pre-charge state corresponds to a damping ratio coefficient of 30%-50% of the initial value; the gradual entry state corresponds to a damping ratio coefficient increasing to 70%-90% of the initial value; the brake engagement state corresponds to a damping ratio coefficient reaching 150%-200% of the initial value and maintaining a short-term stop; and the release state corresponds to a damping ratio coefficient returning to 10%-20% of the initial value. The preset threshold for axial force creep rate is 5-15 N / s, and the preset threshold for angle change rate is 0.5-2 rad / s. Overshoot risk assessment is based on the prediction model outputting an axial force change gradient exceeding 1.2-1.8 times the difference between the target axial force and the current axial force. The short-term stop time window is 0.1-0.5 seconds.

[0103] Specifically, when the axial force fusion estimation result enters the target range and the axial force rise rate exceeds 10 N / s, the controllable damping unit switches to the gradual entry state, increasing the damping proportional coefficient to 80% of the initial value to absorb excess energy generated by the inertia of the actuator. When the prediction model detects that the axial force may exceed the second threshold within the next 0.3 seconds, the controllable damping unit enters the brake state, the damping proportional coefficient instantly increases to 180% of the initial value, and suppresses the axial force rise within 0.2 seconds. When the axial force fusion estimation result falls back to the target range and the rotation angle change rate is less than 1 rad / s, the controllable damping unit switches to the release state, the damping proportional coefficient drops to 15% of the initial value, reducing the reverse load on the actuator. By dynamically adjusting the damping state, axial force overshoot is effectively suppressed and the rebound amplitude is reduced, allowing the axial force to converge stably within the tolerance range.

[0104] As a preferred embodiment, the solution of this application is implemented as follows: After the bolt axial force reaches the target value, the baseline data during the holding phase is recorded at a sampling frequency of 500 times per second, and the controllable damping unit performs a 0.5-second brake action to stop the actuator. The holding phase lasts for 30 seconds, during which the axial force fusion data is preprocessed using a moving average filter to eliminate high-frequency noise. When the difference between the holding correction data and the baseline data exceeds the tolerance band ±3%, a secondary tightening procedure is triggered: the controllable damping unit switches to the pre-charge state, and the actuator tightens the bolt step by step in increments of 0.1°, with an axial force check performed after every 2° of tightening. If the bolt has not entered the tolerance band after 5 tightening cycles, or if the temperature sensor detects a temperature rise exceeding 50°C, a protection program is triggered to cut off the power output and generate an abnormal log.

[0105] Through the above technical solution, this application effectively solves the problem of decreased assembly accuracy caused by bolt preload rebound. By maintaining dynamic monitoring and intelligent tightening mechanism during the holding phase, it ensures that the axial force remains within the process requirements after elastic deformation recovery. The synergistic effect of controllable damping unit and micro-stepping control avoids the risk of thread damage caused by over-tightening in traditional methods and prevents material performance degradation due to excessive temperature rise, thus improving the stability and reliability of axial force control in high-precision assembly scenarios.

[0106] In the above embodiments, multi-source acquisition and processing of bolt identification and process parameter data, temperature measurement data, and zero-load preload data are introduced to establish a self-calibration model that includes sound velocity compensation parameters, elastic modulus compensation parameters, and zero-load baseline data. This effectively eliminates the impact of material batch differences and environmental fluctuations on the accuracy of axial force measurement. Simultaneously, through joint acquisition by axial force sensors and ultrasonic probes, a confidence-weighted axial force fusion estimation result is obtained, achieving higher robustness and reliability than a single ultrasonic detection path. In the control phase, not only is staged control of coarse and fine tightening implemented based on the comparison between the axial force fusion estimation result and the target axial force data, but model predictive control and micro-stepping strategies are also combined. When approaching the target range, a controllable damping unit is used for dynamic energy absorption and stopping, thereby suppressing overshoot and springback, ensuring rapid, stable, and high-precision convergence of the bolt preload. After the target axial force is reached, a holding phase data processing and springback verification mechanism is introduced. If necessary, secondary tightening is automatically performed, enabling the bolt tightening quality to have self-checking and compensation capabilities, ensuring the consistency and long-term reliability of the preload. It solves the problems of unstable ultrasonic coupling, lack of multi-source fusion, and coarse control strategy in existing technologies, and realizes the refinement and traceability of the bolt tightening process.

[0107] In another preferred embodiment based on the above embodiments, see [reference] Figure 2 As shown, this embodiment provides a bolt tightening intelligent control system based on axial force measurement, used to apply the above-mentioned bolt tightening intelligent control method based on axial force measurement, including: The system includes a controller, actuator, shaft force sensor A, shaft force sensor B, angle encoder, torque sensor, temperature sensor, controllable damping unit, and identification read / write unit. Axial force sensor A outputs axial force data. Axial force sensor B outputs ultrasonic echo data. The rotation encoder outputs rotation angle measurement data and angular velocity data. The torque sensor outputs torque measurement data. The temperature sensor outputs probe temperature data and ambient temperature data. The identification reading and writing unit reads bolt markings and process parameter data.

[0108] The controller is electrically connected to the actuator, shaft force sensor A, shaft force sensor B, angle encoder, torque sensor, temperature sensor, controllable damping unit, and identification reading and writing unit.

[0109] The controller is configured to collect bolt identification and process parameter data, collect probe temperature data and ambient temperature data, process process parameter data and temperature data, and obtain sound velocity compensation parameters and elastic modulus compensation parameters.

[0110] Collect zero-load preload data and process it to generate zero-load baseline data to obtain a self-calibration model.

[0111] Axial force data and ultrasonic echo data were collected to obtain the axial force fusion estimation result.

[0112] The axial force fusion estimation result is compared with the target axial force data. When the axial force fusion estimation result is lower than the first threshold of the target axial force, the actuator is controlled to drive at constant speed or constant power. When the axial force fusion estimation result is within the target range, fine-tuning control is performed based on model predictive control and micro-stepping strategy, and the controllable damping unit is driven to absorb energy and stop according to the axial force climb rate data and the angle change rate data.

[0113] The axial force fusion estimation results when the target axial force is reached are collected, the data of the holding stage are processed, and the axial force springback check results are obtained. If the axial force springback check results exceed the limit, secondary tightening control is performed.

[0114] It is understandable that the above-mentioned intelligent control method and system for bolt tightening based on axial force measurement have the same beneficial effects, and will not be elaborated further here.

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for intelligent control of bolt tightening based on axial force measurement, characterized in that, include: Collect bolt identification and process parameter data, collect temperature measurement data, including probe temperature data and ambient temperature data, process the process parameter data and temperature data to obtain sound velocity compensation parameters and elastic modulus compensation parameters; Collect zero-load preload data and process it to generate zero-load baseline data to obtain a self-calibration model; Axial force data and ultrasonic echo data were collected to obtain the axial force fusion estimation result; The axial force fusion estimation result is compared with the target axial force data. When the axial force fusion estimation result is lower than the first threshold of the target axial force, the actuator is controlled to drive at constant speed or constant power. When the axial force fusion estimation result is within the target range, fine-tuning control is performed based on model predictive control and micro-stepping strategy, and the controllable damping unit is driven to absorb energy and stop according to the axial force climb rate data and the angle change rate data. The axial force fusion estimation results when the target axial force is reached are collected, the data of the holding stage are processed, and the axial force springback check results are obtained. If the axial force springback check results exceed the limit, secondary tightening control is performed.

2. The intelligent control method for bolt tightening based on axial force measurement according to claim 1, characterized in that, When processing the process parameter data and temperature data to obtain the sound velocity compensation parameters and elastic modulus compensation parameters, the process includes: Based on the material code, heat treatment state, surface roughness grade, and thread lubrication state in the process parameter data, the reference sound velocity and reference elastic modulus are retrieved from the preset calibration index table; combined with the probe temperature data and ambient temperature data, the surface temperature of the test piece is estimated, and the reference sound velocity and reference elastic modulus are temperature-corrected; based on the surface roughness grade and thread lubrication state, a coupling correction factor is introduced to correct the temperature-corrected sound velocity and elastic modulus, thereby obtaining the sound velocity compensation parameters and elastic modulus compensation parameters.

3. The intelligent control method for bolt tightening based on axial force measurement according to claim 2, characterized in that, When acquiring zero-load preload data and processing it to generate zero-load baseline data to obtain a self-calibration model, the process includes: The control actuator enters the zero-load preload mode, and drives the axial force sensor A and axial force sensor B to establish stable contact without generating effective preload force below the preset zero-load preload upper limit threshold. Simultaneously acquire zero-load signal data from axial force sensor A, zero-load echo data from axial force sensor B, zero-position data from angle encoder, zero-point data from torque sensor, and zero-load temperature data from temperature sensor to form zero-load preload data; The zero-load preload data is filtered, denoised, and drift subtracted, and a steady-state determination is performed according to a preset steady-state determination threshold. When the steady-state determination is passed, the zero-point offset, time delay offset, noise threshold, gain correction coefficient, and coupling quality index are calculated to generate zero-load baseline data. The self-calibration model is updated using the zero-load baseline data combined with the sound velocity compensation parameters and elastic modulus compensation parameters. The self-calibration model is used for axial force fusion estimation and anomaly diagnosis. When the steady-state determination fails or the coupling quality index is lower than the threshold, the zero-load preload and contact posture are automatically adjusted and the data is repeatedly collected and processed until zero-load baseline data that meets the threshold conditions is generated and stored in conjunction with the bolt identification.

4. The intelligent control method for bolt tightening based on axial force measurement according to claim 3, characterized in that, When acquiring axial force data and ultrasonic echo data to obtain the axial force fusion estimation result, the following is included: Axial force correction data is obtained from axial force sensor A; ultrasonic axial force estimation data is obtained from axial force sensor B; within a sliding time window, the axial force correction data and the ultrasonic axial force estimation data are time-aligned, steady-state determined, and abnormal samples are removed; the confidence level is determined based on the signal-to-noise ratio, coupling quality index, and temperature stability; the two types of data are weighted and fused to obtain the axial force fusion estimation result; when the signal-to-noise ratio or coupling quality index of axial force sensor B is lower than the threshold, the weight of axial force sensor B is reduced or the output of axial force sensor A is switched to single-channel output; when axial force sensor A saturates or drifts beyond the limit, the output of axial force sensor B is used and the re-establishment of zero-load baseline data is triggered.

5. The intelligent control method for bolt tightening based on axial force measurement according to claim 1, characterized in that, When comparing the axial force fusion estimation result with the target axial force data, the following is included: When the axial force fusion estimation result is lower than the first threshold of the target axial force, the actuator is controlled to drive at constant speed or constant power. When the axial force fusion estimation result is within the target range, fine-tuning control is performed based on model predictive control and micro-stepping strategy, and the controllable damping unit is simultaneously driven to absorb energy and stop according to the axial force climbing rate and the angle change rate. When the axial force fusion estimation result is higher than or equal to the second threshold and lower than the overload threshold, overshoot correction control is performed. The overshoot correction control includes stopping the actuator output, driving the controllable damping unit to absorb energy and implementing small-angle reversal or micro-step back to bring the axial force fusion estimation result back to the target range, and then performing fine-tuning control after returning to the target range. When the axial force fusion estimation result is determined to be higher than the overload threshold, the actuator output is immediately stopped, the controllable damping unit is released, and an alarm is issued. Here, the target range is the range between the first threshold and the second threshold, and the tolerance band is the allowable deviation range centered on the target axial force.

6. The intelligent control method for bolt tightening based on axial force measurement according to claim 5, characterized in that, When the axial force fusion estimation result is lower than a first threshold of the target axial force, the control of the actuator to drive at constant speed or constant power includes: The system collects coarse-tightening control results and corresponding speed or power setpoints; it also collects angular velocity data from the angular encoder, torque data from the torque sensor, and temperature data from the temperature sensor; processes the angular velocity data and the speed setpoints to obtain speed error and power error; and obtains protection criteria based on a temperature rise threshold. When the coarse-tightening control result is constant speed, a speed closed-loop control result is obtained, which is used to adjust the actuator output so that the angular velocity tracks the speed setpoint and is controlled by a preset torque limit. When the coarse-tightening control result is constant power, a power closed-loop control result is obtained, which is used to adjust the actuator output so that the power tracks the power setpoint and is controlled by a preset angular velocity limit.

7. The intelligent control method for bolt tightening based on axial force measurement according to claim 6, characterized in that, When performing fine-tuning control based on model predictive control and micro-stepping strategy, it includes: Collect axial force fusion estimation results, torque measurement data, rotation angle measurement data and temperature measurement data to obtain state vector data, axial force climb rate data and rotation angle change rate data; Based on the self-calibration model and historical response data within the sliding time window, the controlled object prediction model data is updated online. Cost function data is constructed, which includes a target axial force tracking error term, an overshoot penalty term, and a control increment penalty term. Torque upper limit, angular velocity upper limit, temperature rise upper limit, and predicted overshoot boundary are set. Candidate control sequence data are obtained by solving the problem, and the first control quantity data is obtained. The first control quantity data is quantized into several microstepping unit data through a microstepping strategy. Each microstepping unit data includes microstepping increment data and microstepping duration data, which are applied to the actuator in sequence to form a fine-tuning control result. After each microstepping unit data is executed, the measurement data is re-acquired and the prediction model data and cost function data are updated on a rolling basis.

8. The intelligent control method for bolt tightening based on axial force measurement according to claim 7, characterized in that, When the controllable damping unit performs energy absorption and braking based on the axial force climb rate data and the rotation angle change rate data, it includes: When the axial force fusion estimation result data is within the target range and the axial force climb rate data is greater than the preset threshold, the damping command data in the pre-charge state or the gradual state is obtained to increase the damping proportionally; when the prediction model data indicates overshoot risk or the axial force fusion estimation result data is close to the second threshold, the damping command data in the brake state is obtained to stop the movement for a short time; when the axial force fusion estimation result data falls out of the target range or the axial force climb rate data and the angle change rate data both fall below the threshold, the damping command data in the release state is obtained to reduce the additional load.

9. The intelligent control method for bolt tightening based on axial force measurement according to claim 8, characterized in that, The process involves collecting the axial force fusion estimation result when the target axial force is reached, processing the data during the holding phase, obtaining the axial force springback check result, and performing secondary tightening control if the axial force springback check result exceeds the limit. This includes: Record the axial force fusion estimation result at the start of the holding period as the holding baseline data, control the controllable damping unit to briefly apply the brake and stop the actuator; within the preset holding period, collect the axial force fusion estimation result and temperature measurement data at a preset sampling frequency, preprocess the holding phase data to obtain holding correction data; at the end of the holding period, compare the holding correction data with the holding baseline data to obtain the axial force springback check result and compare it with the tolerance band: when the axial force springback check result is within the tolerance band, the holding period ends and is archived; when the axial force springback check result exceeds the tolerance band but is below the overload threshold, execute secondary tightening control, the secondary tightening control includes: placing the controllable damping unit in the precharge state, driving the actuator to tighten at a small angle according to the micro-step unit, and immediately checking the axial force fusion estimation result and tolerance band after each micro-step unit is completed, until entering the tolerance band or reaching the upper limit of the tightening step and the upper limit of the tightening angle; when the axial force springback check result reaches the overload threshold or the secondary tightening triggers the temperature rise limit, enter the protection control and record the event.

10. A bolt tightening intelligent control system based on axial force measurement, used to apply the bolt tightening intelligent control method based on axial force measurement as described in any one of claims 1-9, characterized in that, include: Controller, actuator, shaft force sensor A, shaft force sensor B, angle encoder, torque sensor, temperature sensor, controllable damping unit, and identification read / write unit; among which, The axial force sensor A is used to output axial force data; the axial force sensor B is used to output ultrasonic echo data; the rotation encoder is used to output rotation measurement data and angular velocity data; the torque sensor is used to output torque measurement data; the temperature sensor is used to output probe temperature data and ambient temperature data; the identification reading and writing unit is used to read bolt identification and process parameter data. The controller is electrically connected to the actuator, shaft force sensor A, shaft force sensor B, angle encoder, torque sensor, temperature sensor, controllable damping unit, and identification reading and writing unit; The controller is configured to collect bolt identification and process parameter data, collect probe temperature data and ambient temperature data, process the process parameter data and temperature data, and obtain sound velocity compensation parameters and elastic modulus compensation parameters. Collect zero-load preload data and process it to generate zero-load baseline data to obtain a self-calibration model; Axial force data and ultrasonic echo data were collected to obtain the axial force fusion estimation result; The axial force fusion estimation result is compared with the target axial force data. When the axial force fusion estimation result is lower than the first threshold of the target axial force, the actuator is controlled to drive at constant speed or constant power. When the axial force fusion estimation result is within the target range, fine-tuning control is performed based on model predictive control and micro-stepping strategy, and the controllable damping unit is driven to absorb energy and stop according to the axial force climb rate data and the angle change rate data. The axial force fusion estimation results when the target axial force is reached are collected, the data of the holding stage are processed, and the axial force springback check results are obtained. If the axial force springback check results exceed the limit, secondary tightening control is performed.

Images