An automatic assembly and tightening device for shaft end nuts that is easy to verify

Abstract

This invention provides an automated assembly and tightening device for shaft-end nuts that is easy to verify, belonging to the field of industrial manufacturing technology. This invention achieves full automation of the nut assembly process by integrating components such as a base, conveying mechanism, nut supply device, assembly robotic arm, tightening mechanism, and verification system. Its core lies in employing a two-stage tightening control strategy, dividing the tightening process into preliminary tightening and reinforcement tightening stages. In the preliminary tightening stage, a low-speed, high-precision mode and dynamic damping control ensure correct thread engagement. The intermediate verification stage measures axial displacement values ​​to promptly detect anomalies. In the reinforcement tightening stage, a stepped torque-increasing algorithm ensures uniform stress distribution. Finally, multi-parameter comprehensive verification evaluates the tightening quality and generates a traceability code, effectively solving the quality instability problem caused by a single-stage tightening method for shaft-end nuts.

Description

Technical Field

[0001] This invention belongs to the field of industrial manufacturing technology, and specifically relates to an automatic assembly and tightening device for shaft end nuts that is easy to verify. Background Technology

[0002] Shaft-end nut assembly is a crucial process in mechanical manufacturing, directly impacting product performance and lifespan. Traditional shaft-end nut assembly technology primarily employs a single-stage tightening method, directly tightening the nut to the preset torque in one go. This method typically utilizes automated assembly lines, robotic arms for gripping, and tightening motors, with simple torque control used to determine the tightening completion status.

[0003] However, traditional single-stage tightening methods have significant technical drawbacks. First, it is difficult to control the thread engagement quality during a single tightening process, which can easily lead to initial assembly defects such as mis-threading and stripping. Second, single torque control cannot adapt to the differences in characteristics of different materials and different batches of parts, and sudden torque changes and uneven contact stress distribution can easily occur during tightening. Third, the lack of intermediate verification steps makes it impossible to detect and correct abnormal conditions in a timely manner during tightening, resulting in large fluctuations in assembly quality.

[0004] Traditional technologies struggle to address these issues primarily because the single-stage tightening method lacks a phased evaluation and adjustment mechanism, hindering precise control and verification of the threaded connection process. In actual production, this technical deficiency leads to unstable shaft-end nut connection quality, making products prone to nut loosening and detachment during use, severely impacting product reliability and safety, especially under high vibration and heavy load conditions. In other words, existing technologies suffer from unstable tightening quality and low threaded connection reliability due to the single-stage tightening method for shaft-end nuts. Summary of the Invention

[0005] In view of this, the present invention provides an automatic assembly and tightening device for shaft end nuts that is easy to verify, which can solve the technical problems of unstable tightening quality and low reliability of threaded connection caused by the single-stage tightening method of shaft end nuts in the prior art.

[0006] The present invention is implemented as follows: The present invention provides an automatic assembly and tightening device for shaft-end nuts that is easy to verify. The control chip of the device has a tightening control module, which performs the following steps: receiving the specification parameters of the shaft-end nut, extracting the preliminary tightening torque value and the reinforcement tightening torque value from the parameter database according to the specification parameters, and generating a two-stage tightening control command; controlling the assembly robot arm to place the nut at the shaft-end thread; starting the tightening mechanism for the preliminary tightening stage, controlling the tightening motor to run in a low-speed, high-precision mode; performing intermediate verification, activating the laser displacement sensor to measure the axial displacement value of the nut; if the preliminary tightening stage verification is passed, entering the reinforcement tightening stage, controlling the tightening motor to use a stepped torque increase algorithm for control, and performing multi-parameter comprehensive verification after completion.

[0007] The base includes a metal support, a leveling mechanism, and a shock-absorbing device. The metal support is made of cast iron, the leveling mechanism includes multiple leveling bolts, and the shock-absorbing device includes rubber shock-absorbing pads. The conveying mechanism includes a conveyor belt, a drive motor, a transmission wheel, and a position sensor. The conveyor belt is made of non-slip rubber, the drive motor is connected to the transmission wheel, and the position sensor is installed at the end of the conveyor belt to detect the position of shaft parts. The nut supply device includes a vibratory feeder, a feeding track, and a distribution mechanism. The vibratory feeder transports the nuts to the feeding track, and the distribution mechanism ensures that each nut is output individually to the gripping position.

[0008] The assembly robotic arm includes a multi-joint structure, servo motors, encoders, and a nut gripping tool. The multi-joint structure consists of multiple links, with servo motors installed at each joint. The encoders are used to monitor the joint angles, and the nut gripping tool is installed at the end of the robotic arm. The tightening mechanism includes a tightening motor, a torque drive shaft, and a tightening head. The tightening motor is connected to the torque drive shaft, which is connected to the tightening head. The tightening head contacts the nut to perform the tightening operation.

[0009] The calibration system includes a torque sensor, a laser displacement sensor, and a pressure sensor. The torque sensor is installed on the torque drive shaft to measure the tightening torque, the laser displacement sensor is installed at the assembly station to measure the axial displacement of the nut, and the pressure sensor is installed at the contact point between the tightening head and the nut to measure the tightening pressure.

[0010] The step of controlling the assembly robot arm to place the nut at the threaded end of the shaft uses dynamic damping control technology to ensure that the nut and the threaded end of the shaft are initially engaged, thus avoiding thread mis-threading caused by improper placement angle.

[0011] Among them, the dynamic damping control technology monitors the reaction force between the nut and the threaded contact surface of the shaft end in real time. The reaction force is measured by a pressure sensor. The compliance parameters of the end effector of the assembly robot arm are dynamically adjusted according to the magnitude of the reaction force, so that the nut can automatically adjust its posture according to the guidance of the threaded shaft end.

[0012] The intermediate verification step involves comparing the axial displacement value of the nut with a preset axial displacement threshold to determine whether the initial tightening stage meets the technical requirements.

[0013] During the tightening phase, a pressure sensor is activated to monitor changes in tightening pressure in real time. If a sudden change in tightening pressure or a pressure exceeding the preset range is detected, the tightening operation is immediately interrupted and an abnormal alarm is issued.

[0014] The stepped torque-increasing algorithm divides the torque-increasing process of the tightening stage into multiple small-amplitude steps. Each step includes two parameters: torque increase value and duration. The tightening control module detects the nut status at the end of each step and only proceeds to the next step after confirming that it is stable.

[0015] In the multi-parameter comprehensive verification step, the actual torque value measured by the torque sensor, the axial displacement value measured by the laser displacement sensor, and the pressure value measured by the pressure sensor are collected simultaneously to construct a two-layer game model for tightening quality assessment. The two-layer game model for tightening quality assessment adopts a leader-follower structure. The upper-layer model optimizes the tightening parameter control strategy, and the lower-layer model simulates the mechanical response of the connecting parts under given tightening parameters. The upper-layer model maximizes the tightening quality index through a nonlinear objective function, and the lower-layer model minimizes the failure probability index through a nonlinear objective function. According to the multi-parameter comprehensive verification results, corresponding operations are performed. For excellent grades, the product passes directly; for qualified grades, it is recorded in the traceability database and marked as requiring attention; for defective grades, an automatic disassembly program is initiated for reassembly.

[0016] This invention innovatively adopts a two-stage tightening control strategy, dividing the tightening process into two stages: preliminary tightening and reinforcement tightening. An intermediate verification mechanism is introduced between the two stages to achieve refined control and comprehensive verification of the threaded connection quality.

[0017] Based on a two-stage tightening technology, this invention effectively solves the technical problems of traditional single-stage tightening methods. In the initial tightening stage, low-speed, high-precision operation and dynamic damping control technology ensure correct engagement between the nut and the shaft thread, preventing thread mis-threading. The intermediate verification stage measures the axial displacement of the nut to promptly identify and address initial assembly anomalies. In the reinforcement tightening stage, a stepped torque-increasing algorithm is employed, based on the thread contact stress distribution equation, to precisely control the torque increment and duration, ensuring uniform distribution of thread contact stress and preventing localized stress concentration.

[0018] By dividing the tightening process into two scientifically reasonable stages and introducing a multi-parameter verification mechanism, this invention successfully solves the technical problems of unstable tightening quality and low thread connection reliability caused by the single-stage tightening method of shaft end nuts, and significantly improves the success rate of shaft end nut assembly and connection reliability. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A schematic diagram of an automatic assembly and tightening device for shaft end nuts that is easy to verify;

[0021] Figure 2 This is a schematic diagram of the base structure;

[0022] Figure 3 Schematic diagram of the nut supply device;

[0023] Figure 4 This is a schematic diagram of the tightening mechanism.

[0024] Figure 5 Electrical connection diagram for the automatic assembly and tightening device module for shaft end nuts;

[0025] Figure 6 For verification of the detailed structure diagram of the system;

[0026] Figure 7 The flowchart is for the tightening module;

[0027] The attached diagram lists the components represented by each number as follows:

[0028] 10. Base; 101. Metal bracket; 102. Leveling mechanism; 103. Shock absorption device; 11. Conveying mechanism; 111. Conveyor belt; 112. Drive motor; 113. Transmission wheel; 114. Position sensor; 12. Nut supply device; 121. Vibratory feeder; 122. Feeding track; 123. Material distribution mechanism; 13. Assembly robotic arm; 131. Multi-joint structure; 132. Servo motor; 133. Nut gripping device; 14. Tightening mechanism; 141. Tightening motor; 142. Torque transmission shaft; 143. Tightening head; 15. Calibration system; 151. Torque sensor; 152. Laser displacement sensor; 153. Pressure sensor; 16. Human-machine interaction device. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0030] like Figure 1-6As shown, an automatic assembly and tightening device for shaft-end nuts, which is easy to verify, is provided. Specifically, it includes a base 10, a conveying mechanism 11, a nut supply device 12, an assembly robotic arm 13, a tightening mechanism 14, a verification system 15, a control chip, a data processing unit, a human-machine interface device 16, and a power management device. The base supports the entire device and ensures stable operation. The conveying mechanism transports shaft parts to the assembly station. The nut supply device arranges the nuts in an orderly manner and transports them to the assembly position. The assembly robotic arm grasps the nuts and places them at a predetermined position on the shaft end. The tightening mechanism tightens the nuts onto the shaft end. The verification system monitors the torque, axial displacement, and pressure values ​​during the tightening process. The control chip is electrically connected to the conveying mechanism, the nut supply device, the assembly robotic arm, the tightening mechanism, the verification system, the data processing unit, the human-machine interface device, and the power management device. The data processing unit receives and analyzes data from the verification system. The human-machine interface device displays assembly status information and receives operation commands. The power management device monitors the power status of each component and ensures stable power supply.

[0031] The base 10 includes a metal bracket 101, a leveling mechanism 102, and a shock-absorbing device 103. The metal bracket is made of cast iron, the leveling mechanism includes multiple leveling bolts, and the shock-absorbing device includes rubber shock-absorbing pads.

[0032] The conveying mechanism 11 includes a conveyor belt 111, a drive motor 112, a transmission wheel 113, and a position sensor 114. The conveyor belt is made of anti-slip rubber material, the drive motor is connected to the transmission wheel, and the position sensor is installed at the end of the conveyor belt to detect the position of shaft parts.

[0033] The nut supply device 12 includes a vibratory feeder 121, a feeding track 122, and a material distribution mechanism 123. The vibratory feeder transports the nuts to the feeding track, and the material distribution mechanism ensures that each nut is output individually to the gripping position.

[0034] The assembly robot arm 13 includes a multi-joint structure 131, a servo motor 132, an encoder, and a nut gripping tool 133. The multi-joint structure consists of multiple links, the servo motor is installed at each joint, the encoder is used to monitor the joint angle, and the nut gripping tool is installed at the end of the robot arm.

[0035] The tightening mechanism 14 includes a tightening motor 141, a torque transmission shaft 142, and a tightening head 143. The tightening motor is connected to the torque transmission shaft, the torque transmission shaft is connected to the tightening head, and the tightening head contacts the nut to perform the tightening operation.

[0036] The calibration system 15 includes a torque sensor 151, a laser displacement sensor 152, and a pressure sensor 153. The torque sensor is installed on the torque transmission shaft to measure the tightening torque, the laser displacement sensor is installed at the assembly station to measure the axial displacement of the nut, and the pressure sensor is installed at the contact point between the tightening head and the nut to measure the tightening pressure.

[0037] The control chip has a tightening control module, such as Figure 7 As shown, the tightening control module is used to perform the following steps:

[0038] S01. Receive the shaft end nut specification parameters transmitted from the data processing unit, extract the corresponding preliminary tightening torque value and reinforcement tightening torque value from the parameter database according to the shaft end nut specification parameters, and generate a two-stage tightening control command.

[0039] S02. Control the assembly robot arm to grab the nut and place it precisely on the thread at the shaft end. Through dynamic damping control technology, ensure that the nut and the thread at the shaft end are initially engaged, and avoid thread mis-threading caused by improper placement angle.

[0040] S03. Start the tightening mechanism to perform the initial tightening stage, control the tightening motor to run in a low speed and high precision mode, monitor the torque sensor data in real time, and pause the tightening operation when the torque reaches the initial tightening torque value.

[0041] S04. Perform intermediate verification: activate the laser displacement sensor to measure the axial displacement value of the nut, and compare the axial displacement value of the nut with the preset axial displacement threshold to determine whether the initial tightening stage meets the technical requirements.

[0042] S05. If the initial tightening stage verification is passed, the reinforcement tightening stage is entered. The tightening motor is controlled by a step-type torque increase algorithm. Based on the thread contact stress distribution equation, the optimal torque increment and duration of each step are calculated. At the same time, the slope of the torque curve and the change of dynamic friction coefficient are monitored to ensure that the thread contact stress is evenly distributed.

[0043] S06. During the tightening stage, the pressure sensor is activated to monitor the changes in tightening pressure in real time. When a sudden change in tightening pressure or exceeding the preset pressure range is detected, the tightening operation is immediately interrupted and an abnormal alarm is issued.

[0044] S07. After the tightening stage is completed, a multi-parameter comprehensive verification is performed. At the same time, the actual torque value measured by the torque sensor, the axial displacement value measured by the laser displacement sensor, and the pressure value measured by the pressure sensor are collected to construct a two-layer game model for tightening quality assessment. The upper-layer model optimizes the tightening process parameters to maximize the tightening quality index, while the lower-layer model simulates the performance response of the threaded connection under various working conditions to minimize the failure probability.

[0045] S08. Perform corresponding operations based on the comprehensive verification results of multiple parameters. For excellent grades, pass directly. For qualified grades, record in the traceability database and mark as requiring attention. For defective grades, start the automatic disassembly program for reassembly.

[0046] S09. After completing all tightening and verification processes, the assembly data, including the torque curve and various measurement parameters, will be stored in the database, and a unique traceability code will be generated and associated with the product to facilitate future quality traceability analysis.

[0047] Specifically, the dynamic damping control technology monitors the reaction force between the nut and the threaded contact surface of the shaft end in real time. The reaction force is measured by a pressure sensor. The compliance parameters of the end effector of the assembly robot are dynamically adjusted according to the magnitude of the reaction force, so that the nut can automatically adjust its posture according to the guidance of the threaded contact surface of the shaft end in the initial stage of contact with the threaded contact surface. This avoids damage to the threaded contact surface caused by forced assembly and improves the thread meshing efficiency.

[0048] Specifically, the stepped torque-increasing algorithm divides the torque increase process during the tightening stage into multiple small-amplitude steps. Each step includes two parameters: the torque increase value and the duration. The tightening control module detects the nut status at the end of each step and only proceeds to the next step after confirming stability. The torque increase value is measured by a torque sensor, and the duration is timed by the internal clock of the control chip. This progressive tightening method can effectively avoid thread stripping or nut breakage caused by instantaneous high torque.

[0049] Specifically, the slope of the torque curve refers to the rate of change of the torque value over time during the tightening process. The torque value is measured by a torque sensor, and the time change is timed by the internal clock of the control chip. During normal tightening, the slope of the torque curve should be kept within the preset slope range, which is stored in the parameter database. If the slope of the torque curve suddenly increases, it indicates that there may be abnormal resistance in the shaft end thread or nut thread. If the slope of the torque curve suddenly decreases, it indicates that there may be stripping in the shaft end thread or nut thread.

[0050] Specifically, the dynamic friction coefficient change refers to the change in the friction coefficient between the threaded contact surfaces during the tightening process as the tightening angle changes. The friction coefficient is calculated by the torque value measured by the torque sensor and the pressure value measured by the pressure sensor. The tightening angle is measured by the encoder of the tightening motor. The dynamic friction coefficient change is used to monitor the contact state between the shaft end thread and the nut thread to ensure the quality of the threaded connection.

[0051] Specifically, the thread contact stress distribution equation calculates the contact stress distribution at various points on the shaft end thread during nut tightening. Inputs include the elastic modulus of the nut material, the thread pitch at the shaft end, the contact area of ​​the thread at the shaft end, the applied torque value measured by the torque sensor, and the surface roughness of the thread at the shaft end. The elastic modulus of the nut material, the thread pitch at the shaft end, the contact area of ​​the thread at the shaft end, and the surface roughness of the thread at the shaft end are all obtained from a parameter database. The output is the stress distribution function at each contact point on the shaft end thread. This stress distribution function guides the stepped torque increase algorithm to adjust the torque increase value and duration, ensuring uniform stress distribution during nut tightening and avoiding damage to the shaft end thread or nut thread caused by localized stress concentration.

[0052] The tightening quality assessment two-level game model adopts a leader-follower structure. The leader in the upper-level model is responsible for optimizing the tightening parameter control strategy, while the followers in the lower-level model simulate the mechanical response of the connected components under given tightening parameters. The input parameters of the upper-level model include the actual torque value measured by the torque sensor, the target torque value in the parameter database, the axial displacement value measured by the laser displacement sensor, the ideal axial displacement value in the parameter database, the pressure variability measured by the pressure sensor, and the average pressure value measured by the pressure sensor. The output parameters of the upper-level model include the optimized target torque value, the optimized ideal axial displacement value, the optimized torque change rate, and the optimized step duration. The input parameters of the upper-layer model include the optimized target torque value, the optimized ideal axial displacement value, the optimized torque change rate, the optimized step duration, and the material parameters obtained from the parameter database. The output parameters of the lower-layer model include the predicted maximum stress value, the predicted amount of plastic deformation, the predicted stress variability, and the predicted average stress value. The upper-layer model maximizes the tightening quality index through a nonlinear objective function, while the lower-layer model minimizes the failure probability index through a nonlinear objective function. The two-layer models are coupled by using the optimal solution of the lower-layer model as the constraint condition of the upper-layer model. By solving the two-layer game model, the optimal combination of tightening parameters that can ensure assembly efficiency and maximize connection reliability is determined.

[0053] Specifically, the three-dimensional verification space model is a three-dimensional space constructed using the torque value measured by the torque sensor, the axial displacement value measured by the laser displacement sensor, and the pressure value measured by the pressure sensor as three dimensions. Three nested ellipsoidal regions are defined in the three-dimensional space, representing the excellent region, the qualified region, and the defective region, respectively. When the combination point of the three parameters falls in different regions, the tightening control module determines the corresponding tightening quality level. The boundary parameters of the excellent region, the qualified region, and the defective region are stored in the parameter database.

[0054] The automatic disassembly program is specifically designed as a reverse operation process for nuts whose multi-parameter comprehensive verification results are defective. It controls the tightening motor to rotate in the reverse direction and monitors the torque change measured by the torque sensor. The speed is reduced before the nut is completely loosened to ensure that the nut will not fall off due to sudden loosening. After disassembly, the nut is placed back in the collection area for inspection, and the disassembly data is recorded to analyze the cause of the failure.

[0055] The traceability code is a unique identifier composed of an assembly timestamp, a workstation number, a nut batch number, a shaft part number, and a verification result code. The assembly timestamp is generated by the clock inside the control chip, the workstation number is preset in the control chip, the nut batch number is obtained from the nut supply device, the shaft part number is obtained from the conveying mechanism, and the verification result code is generated by the comprehensive verification result of multiple parameters. The traceability code is stored in the database through the data processing unit and associated with the product lifecycle management system, so that when the product has a nut loosening problem during use, the specific assembly parameters and process data can be traced.

[0056] Specifically, the base of the device is constructed as follows: The base is made of high-strength cast iron with a thickness of 50-80mm, and its surface is treated with anti-corrosion coating, achieving a hardness of HB220-260. The metal support is welded from Q235 steel with a weld thickness of 8-12mm. The overall structure adopts a triangular stable support design, and the support height is 800-1200mm. The leveling mechanism includes six M16 bolts, each with a load-bearing capacity of 3000N, and anti-loosening washers are placed around the bolts. The vibration damping device consists of eight rubber damping pads, each measuring 120×80×25mm, with a Shore A hardness of 70-80, and a vibration damping frequency response range of 5-80Hz. It can effectively absorb vibrations in the 20-500Hz frequency range, improving equipment stability by more than 85%. The base is also equipped with a grounding protection device with a grounding resistance of no more than 4Ω to ensure operational safety.

[0057] Specific implementation of the conveying mechanism: The conveyor belt is made of wear-resistant and non-slip rubber material, with a thickness of 5-8mm, a width of 300-500mm, and a surface hardness of Shore A60-70. The drive motor is a servo motor with a power of 1.5-2.2kW, a speed of 0-1500r / min, and torque accuracy controlled within ±0.5%. The transmission wheel is made of 45# steel, with a diameter of 200-250mm, chrome-plated surface, and a hardness of HRC48-52. The position sensor is a photoelectric sensor with a resolution of 0.1mm, a response time of less than 5ms, and a detection range of 5-500mm. The conveying mechanism is also equipped with a speed adjustment module, which can realize stepless speed adjustment within the range of 0.1-1.0m / s, and realize closed-loop speed control through PID algorithm, with a control accuracy of ±1%. The mechanism also has an emergency stop device with a stop response time of less than 0.2s.

[0058] The specific implementation of the nut supply device: The vibratory feeder is made of 304 stainless steel, with a diameter of 450-600mm. The vibration frequency is adjustable within the range of 30-60Hz, and the amplitude is 0.5-2.0mm. The feeding track is made of hard anodized aluminum alloy with a surface hardness of HV500 or higher. The track width is 2-5mm larger than the maximum outer diameter of the nut. The dispensing mechanism adopts a pneumatic control system with a working air pressure of 0.4-0.6MPa and a cylinder stroke accuracy controlled within ±0.1mm. The entire supply device adjusts the vibration frequency and amplitude through an adaptive control algorithm based on fuzzy control theory. By monitoring the density and flow rate of the nuts on the track in real time, the vibration parameters are dynamically adjusted to match the feeding rate with the assembly cycle, achieving a feeding cycle stability of over 95%. The supply device also has an anti-clogging detection system. When a potential blockage is detected, the vibration amplitude is automatically increased and the vibration direction is changed to clear the blockage.

[0059] The specific implementation of the assembly robotic arm: The multi-joint structure consists of 4 degrees of freedom, with each joint having a range of motion of ±180° and a joint positioning accuracy of ±0.05mm. The servo motor is a permanent magnet synchronous motor with a power of 400-800W, a maximum speed of 3000r / min, and torque accuracy controlled within ±0.2%. The encoder is an absolute encoder with a resolution of 19 bits, i.e., 524288 pulses / revolution, and an accuracy of ±0.001°. The nut gripping tool adopts a pneumatic gripper design, with a gripping force adjustable within the range of 10-200N. The gripper surface is covered with polyurethane material to increase the coefficient of friction to above 0.8. The robotic arm uses a composite motion planning algorithm based on fifth-order polynomial interpolation, while also incorporating Bezier curve smoothing to achieve smooth acceleration and deceleration during the gripping process, with the rate of acceleration change controlled within 200m / s². 3Within this range, the grasping success rate reaches over 99.8%. The robotic arm is also equipped with a force feedback system, which achieves force control through current detection, with a force control accuracy of ±0.5N.

[0060] The tightening mechanism is implemented as follows: The tightening motor is a high-precision servo motor with a power of 0.8–1.5 kW and a speed range of 0–2000 r / min, achieving a torque control accuracy of ±0.1%. The torque transmission shaft is made of 40Cr alloy steel, heat-treated to achieve a hardness of HRC42–46, and has a shaft diameter of 15–25 mm. The tightening head features a modular design, allowing for quick replacement of nuts of different specifications. The head material is high-speed steel with a hardness of HRC58–62. The tightening mechanism employs vector control technology, controlling the motor torque output through space voltage vector pulse width modulation, achieving a torque response time of less than 10 ms. The tightening mechanism also achieves dual closed-loop control of speed and torque, maintaining high-precision control under low-speed, high-torque conditions, with speed fluctuations controlled within ±0.5% and torque fluctuations within ±0.2%. A shock-absorbing buffer device is installed between the tightening head and the motor, using hydraulic damping principles to effectively reduce impact loads and extend the equipment's service life.

[0061] The calibration system is implemented as follows: The torque sensor employs a strain gauge design with a range of 0–200 N·m, an accuracy class of 0.2, and a frequency response of 1000 Hz. The laser displacement sensor utilizes the triangulation principle, with a range of 0–50 mm, a resolution of 0.001 mm, and a sampling rate of 10 kHz. The pressure sensor employs a piezoresistive design with a range of 0–5000 N, an accuracy class of 0.5, and a response time of less than 1 ms. The calibration system integrates multi-sensor information through a data fusion algorithm based on Kalman filtering technology, which effectively filters out measurement noise and improves measurement accuracy by more than 20%. The calibration system also features a self-calibration function, automatically performing zero-point and full-scale calibration upon each power-on to ensure stable accuracy over long-term use. The data acquisition frequency of the calibration system is 1000 Hz, and the acquired data undergoes wavelet transform for noise reduction, improving the signal-to-noise ratio by more than 15 dB, providing a high-quality data foundation for subsequent parameter analysis.

[0062] Specifically, the detailed implementation method of the tightening control module's execution steps is described in detail below:

[0063] The specific implementation of step S01: The process of receiving the specification parameters of the shaft end nut first involves inputting basic information such as the nut model, specifications, and material through a human-machine interface device or automatically obtaining it from the manufacturing execution system. This information is formatted to form a standard data packet. The parameter database adopts a relational database structure design, storing standard parameters for various nuts, including thread specifications, pitch, and material strength. Database retrieval uses hash index technology, with a retrieval time of less than 5ms. The algorithm for extracting the initial tightening torque value and the reinforcement tightening torque value from the database is based on threaded connection theory, considering factors such as nut diameter, pitch, material yield strength, and friction coefficient. The initial tightening torque value is usually set to 30%–40% of the standard tightening torque, and the reinforcement tightening torque value is set to 90%–100% of the standard tightening torque. The process of generating the two-stage tightening control command adopts a state machine design pattern, dividing the tightening process into two states: initial tightening and reinforcement tightening. Each state contains corresponding parameters such as motor speed, torque limit, and angle control, forming a complete control command sequence. This sequence is transmitted to the tightening motor controller via a CAN bus with a communication rate of 1Mbps to ensure real-time command transmission.

[0064] The specific implementation of step S02: Controlling the assembly robotic arm's nut-grabbing process begins with locating the nut using a machine vision system. This system employs an 850nm wavelength infrared light source for illumination, combined with a high-resolution industrial camera and a deep learning object detection algorithm to identify the nut, achieving an accuracy rate of over 99.5%. During the gripping process, an impedance-based gripping force control strategy is used, maintaining the clamping force between the minimum necessary force and the maximum safe force. For M6 to M12 nuts, the clamping force is typically set between 30 and 80 N. The positioning accuracy of precisely placing the nut at the shaft end thread is controlled within ±0.1 mm, and the angular deviation within ±1°. Dynamic damping control technology establishes a compliance model for the robotic arm's end effector, adjusting stiffness parameters within the range of 5000–20000 N / m and damping parameters within the range of 200–800 N·s / m, dynamically adjusting based on the contact force. The contact force is measured in real-time by a pressure sensor with a sampling frequency of 1000 Hz. This technology enables the nut to automatically adjust its posture based on the thread guide at the shaft end, much like a human hand senses. When a sudden change in resistance is detected, the system automatically fine-tunes the nut's posture to ensure proper thread engagement, avoiding thread mis-threading caused by improper placement angles and improving thread engagement efficiency to over 99%.

[0065] The specific implementation of step S03: First, the tightening motor is activated during the initial tightening stage. The motor starts using an S-shaped acceleration curve with a start-up time of 0.2–0.5 seconds to avoid shock caused by sudden starts. The low-speed, high-precision mode refers to controlling the motor speed within the range of 20–60 r / min. The steady-state accuracy of the system is improved by increasing the integral time constant of the control algorithm to 0.05–0.1 seconds. During real-time monitoring of torque sensor data, a sliding window averaging filter algorithm is used with a window width of 10–20 sampling points to effectively filter out high-frequency noise, keeping data processing latency within 10 ms. The system simultaneously calculates the torque change rate. When the torque change rate exceeds 1 N·m / s, the system automatically reduces the motor speed to avoid over-torque. When the torque reaches the initial tightening torque value, the tightening operation is paused with a pause response time of less than 50 ms, and the torque overshoot is controlled within 5%. The initial tightening torque value is determined according to the nut specification. For example, the initial tightening torque for an M8 standard nut is usually set to 5–8 N·m, and for an M12 standard nut it is usually set to 15–25 N·m. This stage, by precisely controlling the initial tightening torque, lays the foundation for subsequent tightening and avoids thread damage caused by excessive torque in the initial stage.

[0066] The specific implementation of step S04: For intermediate verification, firstly, the laser displacement sensor is activated. This sensor operates at a wavelength of 650nm, with a spot diameter less than 0.5mm and a sampling frequency of 5000Hz. When measuring the axial displacement of the nut, a multi-point measurement method is used. 3-5 measurement points are evenly selected on the nut circumference. The average axial displacement value is calculated by fitting a plane using the least squares method, achieving a measurement accuracy of ±0.005mm. The measured axial displacement value is compared with a preset axial displacement threshold, which is typically set to 90%-110% of the theoretically calculated value. For example, the axial displacement threshold for an M8 nut is typically 1.0-1.3mm, and for an M12 nut, it is typically 1.5-1.8mm. The comparison process uses a fuzzy logic decision algorithm. Based on the magnitude of the deviation, it determines whether the initial tightening stage meets the technical requirements. This algorithm sets three fuzzy sets: compliant, edge state, and non-compliant. The membership function is used to calculate the membership degree of the measured value to each set, and the final judgment result is output. If the judgment result is non-compliant, the system will record the anomaly and enter the anomaly handling process. If the judgment result is borderline, the system will mark it as requiring special attention. If the judgment result is compliant, it will prepare to proceed to the next stage. The purpose of intermediate verification is to check the initial tightening quality before formal tightening, to avoid continuing tightening under poor foundation conditions, and to improve the final assembly quality.

[0067] The specific implementation of step S05: In the tightening stage, the torque difference between the initial tightening torque value and the tightening torque value is first calculated. Then, this difference is divided into multiple steps, typically 5 to 8 steps, according to a stepped torque increase algorithm. The torque increase value of each step is 15% to 25% of the total difference, and the duration is 0.5 to 1.5 seconds. The optimal torque increment and duration for each step are calculated based on the thread contact stress distribution equation. This equation considers factors such as thread geometry parameters, material properties, and friction coefficient. For M8 standard steel nuts, the thread contact stress is usually controlled within the range of 200 to 400 MPa, and for M12 standard steel nuts, the thread contact stress is usually controlled within the range of 300 to 500 MPa. Simultaneously, the slope of the torque curve is monitored. Under normal circumstances, the slope of the torque curve should be maintained within the range of 0.5 to 2.0 N·m / s. If the slope exceeds 2.5 N·m / s, the system will automatically reduce the torque increase rate. The dynamic friction coefficient typically fluctuates within the range of 0.1 to 0.3. If the friction coefficient changes by more than 50%, the system will issue a warning and adjust the tightening parameters. The core of the stepped torque-increasing algorithm is a progressive tightening method. By increasing the torque in small steps and multiple stages, it ensures a uniform distribution of thread contact stress, avoids thread stripping or nut breakage caused by instantaneous high torque, and improves tightening reliability.

[0068] The specific implementation of step S06: During the tightening stage, a pressure sensor is activated to monitor the tightening pressure changes in real time. The pressure sensor sampling frequency is 2000Hz, and the data is entered into the control system via an A / D converter with a 16-bit conversion accuracy. The system uses wavelet analysis to perform time-frequency analysis on the pressure signal. By analyzing the statistical characteristics of wavelet coefficients at different scales, the pressure change pattern is identified. During normal tightening, the pressure change should increase smoothly. The preset pressure range is determined according to the nut specification; for example, the preset pressure range for an M8 nut is 1000–1800N, and the preset pressure range for an M12 nut is 2500–4000N. When a sudden change in pressure is detected, the system calculates the pressure change rate. If the change rate exceeds a threshold (usually set to 2000N / s), it is considered an abnormal situation. The system also monitors the pressure fluctuation frequency; if high-frequency oscillations (frequency greater than 10Hz) occur, it will also be considered abnormal. When an anomaly is detected, the system immediately interrupts the tightening operation, stops the tightening motor and maintains its current position, and triggers an alarm. The alarm signal is displayed through an audible and visual indicator and a human-machine interface, and detailed anomaly data is recorded. This step is designed to detect potential problems during tightening, such as thread jamming or material deformation, by monitoring changes in tightening pressure in real time, thus preventing damage to components from continued tightening.

[0069] The specific implementation of step S07: After the tightening stage is completed, a multi-parameter comprehensive verification is performed. First, data from the torque sensor, laser displacement sensor, and pressure sensor are collected simultaneously. The sampling frequency is uniformly set to 1000Hz, the sampling duration is 0.5s, and 500 data points are obtained. The system calculates the average value, standard deviation, and coefficient of variation of the torque value, axial displacement value, and pressure value, and constructs a feature vector based on these parameters. In the two-layer game model for tightening quality assessment, the upper-layer model uses a genetic algorithm to optimize the tightening process parameters. The population size is set to 50-100, the number of iterations is 200-500, the crossover probability is 0.8, and the mutation probability is 0.1. The tightening quality index is maximized through a nonlinear objective function. The tightening quality index consists of torque accuracy (weight 0.4), displacement uniformity (weight 0.3), and pressure stability (weight 0.3). The lower-level model uses the finite element method to simulate the performance response of threaded connections under various working conditions. Adaptive meshing technology is employed, with the number of elements ranging from 10,000 to 50,000. A nonlinear objective function is used to minimize the failure probability exponent, which is composed of maximum stress (weight 0.5), plastic deformation (weight 0.3), and stress distribution uniformity (weight 0.2). The two-level models are coupled by using the optimal solution of the lower-level model as a constraint on the upper-level model. The alternating direction multiplier method is used for solution, with the iterative convergence condition being a relative error of less than 0.1%. By solving the two-level game model, the optimal combination of tightening parameters that ensures both assembly efficiency and connection reliability is determined, providing a reference for similar future working conditions.

[0070] The specific implementation of step S08: Based on the multi-parameter comprehensive verification results, the corresponding operation is performed. First, the verification results are compared with the preset three-dimensional verification space model. The three-dimensional verification space model defines three nested ellipsoidal regions, and the boundary parameters of the regions are determined according to the nut specifications. For M8 standard nuts, the torque range of the excellent region is 95% to 105% of the nominal value, the axial displacement range is 93% to 107% of the nominal value, and the pressure range is 90% to 110% of the nominal value; the torque range of the qualified region is 90% to 110% of the nominal value, the axial displacement range is 85% to 115% of the nominal value, and the pressure range is 80% to 120% of the nominal value; regions exceeding the qualified region are judged as defective regions. For nuts of excellent quality, the system will directly pass quality acceptance, record complete parameters in the database, and mark them as Grade A quality. For nuts of acceptable quality, the system will record them in the traceability database and mark them as Grade B quality, while also noting that attention is needed for focused monitoring during subsequent use. For nuts of defective quality, the system will immediately initiate an automatic disassembly program, controlling the tightening motor to run in reverse at a speed of 50-100 rpm, while monitoring torque changes. When the torque drops to approximately 20% of the initial value, the speed will be reduced to 20 rpm to ensure safe removal of the nut. After disassembly, the nut will be sent to the inspection area for analysis, and detailed defect data will be recorded. The purpose of this step is to classify assembly quality based on the verification results, ensuring that the final product quality meets requirements, and providing data support for quality traceability.

[0071] The specific implementation of step S09: After completing all tightening and verification processes, the data storage process first compresses the torque curve data. A piecewise linear fitting algorithm is used to compress the original data to key points, achieving a compression ratio of over 10:1 while ensuring the restoration error is less than 1%. Measurement parameters include maximum torque, average torque, torque standard deviation, maximum axial displacement, average axial displacement, displacement standard deviation, maximum pressure, average pressure, and pressure standard deviation. This data is stored in JSON format, with a data packet size typically ranging from 10 to 50 KB. The database adopts a hybrid relational and time-series structure. The relational database stores basic product information, while the time-series database stores time-series measurement data, achieving a database write performance of over 1000 records per second. The process of generating a unique traceability code uses the SHA-256 hash algorithm. The assembly timestamp (accurate to milliseconds), workstation number (2 digits), nut batch number (8 characters), shaft part number (12 characters), and verification result code (2 characters) are combined to calculate the hash value. The first 16 digits are used as the traceability code, with a collision probability of less than 10. -10The traceability code establishes a one-to-one association with the product, and data integrity is ensured through database foreign key constraints. The purpose of this step is to establish a complete quality traceability system, enabling quality issues throughout the product's lifecycle to be traced back to specific assembly parameters and process data, providing a basis for quality improvement and fault analysis.

[0072] The mathematical model or calculation process involved in this invention will be described in detail below.

[0073] In dynamic damping control technology, the compliance parameter adjustment of the assembly robot arm end effector can be expressed as follows:

[0074]

[0075] In the formula, K d (t) represents the end stiffness parameter; C d (t) represents the end damping parameter; K d0 The initial stiffness parameter ranges from 5000 to 20000 N / m; C d0 The initial damping parameters range from 200 to 800 N·s / m; F r (t) represents the reaction force measured by the pressure sensor, in N; α is the rate of change of the reaction force, in N / s. K β is the stiffness adjustment factor, with a value ranging from 0.5 to 2.0 m / N; K α is the dynamic stiffness adjustment coefficient, with a value ranging from 0.01 to 0.05 m·s / N; C β is the damping adjustment coefficient, with a value ranging from 0.1 to 0.5 s / N; C This is the dynamic adjustment coefficient for damping, with a value ranging from 0.005 to 0.02s. 2 / N.

[0076] The parameter acquisition method is: F r (t) is obtained directly by a pressure sensor with a sampling frequency of 1000Hz; It is obtained by calculating the difference between the pressure values ​​at two adjacent moments and dividing by the time interval: Where Δt is the sampling time interval, which is 0.001s. K d0 C d0 α K β K α C and β C The initial values ​​are obtained from the parameter database. These parameters are determined through offline calibration tests. The calibration process includes: Step 1: Measure the displacement response of the end effector under different contact force conditions; Step 2: Fit the stiffness and damping parameters using the least squares method; Step 3: Analyze the relationship between the parameters and the contact force, and determine the values ​​of each coefficient.

[0077] The design principle of this dynamic damping control equation is based on the biomechanical characteristics of natural human hand contact. When the contact force increases, the stiffness decreases appropriately while the damping increases, allowing the nut to adjust its posture more flexibly. When the rate of change of the contact force is large, the system responds quickly to avoid impact. This adaptive adjustment mechanism significantly improves the success rate of initial thread engagement, increasing the success rate by approximately 25% compared to fixed parameter control.

[0078] The calculation of torque increase and duration in the stepped torque-increasing algorithm can be expressed as follows:

[0079]

[0080] In the formula, ΔT i Δt represents the torque increase at the i-th step, in N·m. i T represents the duration of the i-th step, in seconds. f The tightening torque value is given by the reinforcement specification, in N·m; T0 is the initial tightening torque value, in N·m; γ i For the torque distribution coefficient, satisfying The value range is 0.15 to 0.25; δ i σ is the fluctuation adjustment coefficient, ranging from -0.1 to 0.1; N is the total number of steps, ranging from 5 to 8; τ0 is the baseline duration, ranging from 0.5 to 1.0 s; η is the time adjustment coefficient, ranging from 0.5 to 2.0; σ max (i-1) represents the maximum contact stress at the end of the (i-1)th step, in MPa; σ ref The value is a reference stress in MPa. It is 300 MPa for M8 nuts and 400 MPa for M12 nuts.

[0081] The parameter acquisition method is: T f T0 is obtained from step S01 and stored in the control chip; γ i The value is determined through an optimization algorithm based on a large amount of tightening test data. This algorithm uses response surface methodology to establish a model relating torque distribution to tightening quality, and then uses gradient descent to solve for the optimal torque distribution. max (i-1) is obtained by calculation using the thread contact stress distribution equation.

[0082] The algorithm's design principle is based on the stress relaxation theory in materials mechanics. The step-by-step torque increase allows sufficient time for the stress distribution after each torque increase to reach a stable state, avoiding thread damage caused by instantaneous high torque. A torque fluctuation adjustment term is introduced. The purpose is to simulate the minute fluctuations that occur during manual tightening, which help to distribute stress evenly on the threaded contact surface. The duration is proportional to the current contact stress; the greater the stress, the longer the relaxation time is required. This reduces tightening time by approximately 30% and improves connection reliability by 10% compared to a fixed time interval method.

[0083] The stress distribution equation for threaded contact can be expressed as:

[0084]

[0085] In the formula, σ(z) is the contact stress at the axial position z, in MPa; T is the applied torque, in N·m; P is the thread pitch, in mm; r is the effective thread radius, in mm; A e The effective contact area of ​​the thread is expressed in mm. 2 f(z) is the stress distribution function; λ is the material property correction coefficient, dimensionless, ranging from 0.1 to 0.3; E m E represents the elastic modulus of the nut material, expressed in GPa. s R is the elastic modulus of the shaft end material, in GPa. a ε represents the surface roughness of the shaft end thread, in μm. σ , where is the stress calculation error term, in MPa, with a value range of ±15MPa; z is the axial distance from the nut end face, in mm; z0 is the axial position of the first contact thread, in mm; μ is the stress attenuation coefficient, dimensionless, with a value range of 0.2 to 0.5; κ is the pitch periodic stress fluctuation coefficient, dimensionless, with a value range of 0.05 to 0.15.

[0086] The parameters are obtained as follows: T is measured by a torque sensor; P, r, and E are also measured. m E s z0 is obtained from the parameter database; A e Through formula A e =π·d m ·h·n e ·k c Calculate, where d m where n is the average diameter, h is the thread height, and n is the thread height. e To determine the effective number of thread turns, k c R is the contact coefficient; a The roughness was measured using a roughness tester before assembly.

[0087] The design principle of this equation is based on classical thread mechanics theory, taking into account factors such as torque transmission, pitch influence, material elasticity, and surface roughness. The function f(z) describes the stress distribution along the axial direction, exhibiting an exponential decay trend, reflecting the phenomenon that the first contact thread bears the maximum stress. A periodic term is introduced. The periodic stress variation caused by the thread geometry was simulated. Material property correction terms were included. The influence of different material combinations and surface conditions on contact stress was considered. Compared with traditional calculation methods, this equation incorporates the consideration of surface roughness and stress distribution, enabling more accurate prediction of stress concentration in threaded connections and improving prediction accuracy by approximately 20%.

[0088] The formula for calculating the slope of the torque curve is:

[0089]

[0090] In the formula, S T T(t) is the slope of the torque curve at time t, in N·m / s; T(t) is the torque value measured at time t, in N·m; T(t-Δt) is the torque value measured at time t-Δt, in N·m; Δt is the sampling time interval, in s, usually taken as 0.001s; ω is the smoothing factor, in 1 / (N·m), with a value range of 0.5 to 2.0 / (N·m).

[0091] The parameters are obtained as follows: T(t) and T(t-Δt) are directly measured by the torque sensor; the ω value is determined by analyzing historical tightening data and optimizing it. This optimization process uses a cross-validation method to make the slope calculation insensitive to noise while maintaining sensitivity to abnormal changes.

[0092] The design principle of this formula is based on numerical differentiation and adaptive filtering. Traditional slope calculation is easily affected by measurement noise. By introducing a smoothing factor ω, when the torque changes significantly, the denominator increases, making the slope calculation result more stable and avoiding false alarms. When the torque changes slightly, the denominator approaches 1, maintaining sensitivity to minute changes. This adaptive slope calculation method improves the anomaly detection accuracy by approximately 40% compared to the fixed window filtering method.

[0093] The formula for calculating the dynamic friction coefficient is:

[0094]

[0095] In the formula, μ d (θ) is the dynamic friction coefficient at rotation angle θ, dimensionless; T(θ) is the torque value at rotation angle θ, in N·m; F(θ) is the axial pressure value at rotation angle θ, in N; φ is the thread angle, in rad; r eξ is the effective radius of the thread, in mm; P is the thread pitch, in mm; ξ is the break-in coefficient, dimensionless, ranging from 0.01 to 0.05; v is the break-in index, dimensionless, ranging from 0.5 to 1.5; ζ is the break-in attenuation coefficient, in rad. -1 The value ranges from 0.01 to 0.05 rad. -1 ;ε μ This is the error term for calculating the friction coefficient, with a value range of ±0.02.

[0096] The parameters are obtained as follows: T(θ) is measured by a torque sensor; F(θ) is measured by a pressure sensor; θ is measured by the encoder of the tightening motor; φ and r... e P is obtained from the parameter database; ξ, v, and ζ are obtained through tightening test calibration. The calibration process includes: Step 1: Tightening test under different speeds and preload conditions; Step 2: Record the complete torque-angle and pressure-angle curves; Step 3: Fit the model parameters using the nonlinear least squares method.

[0097] The design principle of this formula is based on the theory of thread mechanics and tribology. The first term... The second term represents the apparent coefficient of friction calculated from torque and axial force. The third term, ξ·θ, is a correction term for the helix angle, eliminating the influence of the helix angle on friction calculations. v ·e -ζ·θ This describes the frictional break-in phenomenon during the tightening process. Initially, the coefficient of friction increases with increasing rotation angle (θ). v (Item), as the surfaces gradually wear in, the coefficient of friction gradually stabilizes (e) -ζ·θ (Item). This composite model can describe the frictional changes during the tightening process more accurately than the traditional fixed friction coefficient model, improving the accuracy of tightening quality judgment by approximately 30%.

[0098] In the two-level game model for tightening quality assessment, the objective function of the upper-level model can be expressed as:

[0099]

[0100] Constraints:

[0101] g1(X)=T min ≤T t ≤T max ;

[0102] g2(X)=D min ≤D t ≤D max ;

[0103] g3(X)=0 t max ​​;

[0104] g4(X)=t min ≤t s ≤t max ;

[0105] In the formula, x = [T t D t S t , t s ] T T represents the decision vector of the upper-level model. t The optimized target torque value is expressed in N·m; D t The optimized ideal axial displacement value is expressed in mm; S t The optimized torque change rate is expressed in N·m / s; t s The optimized staircase duration, in seconds; Y * T represents the optimal solution of the lower-level model given X; a The actual torque value measured by the torque sensor, in N·m; D a The axial displacement value measured by the laser displacement sensor, in mm; σ p Standard deviation of pressure measured by the pressure sensor, in N; ρ is the average pressure value measured by the pressure sensor, in N; w1, w2, and w3 are weighting coefficients, satisfying w1 + w2 + w3 = 1, typically w1 = 0.4, w2 = 0.3, and w3 = 0.3; ρ is the penalty coefficient, ranging from 0.1 to 0.5; C is the coupling matrix, describing the influence of the lower-level model output on the upper-level target; T min T max D min D max S max t min t max The constraint boundary values ​​for each parameter are obtained from the parameter database.

[0106] The objective function of the lower-level model can be expressed as:

[0107]

[0108] Constraints:

[0109] h1(X,Y)=σ max <σ yield ;

[0110] h2(X,Y)=ε p <ε crit ;

[0111] h3(X,Y)=σ var <σvar，max ;

[0112] In the formula, Y = [σ max , ε p , σ var , σ avg ] T σ is the state vector of the lower-level model; max To predict the maximum stress value, the unit is MPa; ε p To predict the amount of plastic deformation, dimensionless; σ var To predict stress variability, the unit is MPa; σ avg The mean stress value is predicted, in MPa; σ yield ε represents the yield strength of the material, measured in MPa. crit σ is the critical plastic deformation quantity, dimensionless; opt The optimal average stress value is given in MPa; α, β, γ, and δ are weighting coefficients that satisfy α + β + γ + δ = 1, typically α = 0.5, β = 0.3, γ = 0.1, and δ = 0.1; σ var，max This represents the maximum permissible stress variability, expressed in MPa.

[0113] The parameter acquisition method is: T a D a σ p , σ is obtained by measurement from the corresponding sensor. yield ε crit σ opt The C matrix is ​​obtained from the material database; it is determined by analyzing a large amount of tightening data to establish a model of the relationship between the upper and lower layers; the state vector Y of the lower layer model is obtained by finite element analysis, specifically including: Step 1: Establish a parametric finite element model based on the geometric dimensions of the nut and shaft end; Step 2: Apply boundary conditions and loading history defined by X; Step 3: Solve the nonlinear contact problem to obtain the stress and deformation field; Step 4: Extract key indicators to form the state vector Y.

[0114] The two-layer game theory model is designed based on a leader-follower game structure. The upper-layer model (leader) aims to maximize tightening quality by adjusting the control parameter X to influence system behavior; the lower-layer model (follower) simulates the system's mechanical response under given control parameters, aiming to minimize failure risk. The two-layer model utilizes ρ·Y in the upper-layer objective function... *T ·C·Y *The term coupling expresses the influence of system state on tightening quality. The first three terms of the upper-level objective function evaluate torque accuracy, displacement accuracy, and pressure stability, respectively, while the lower-level objective function evaluates maximum stress, plastic deformation, stress distribution uniformity, and average stress level. This two-layer structure can simultaneously consider the influence of process parameters and material response, improving solution quality by approximately 25% compared to a single-layer optimization model and significantly reducing the failure rate of threaded connections.

[0115] In the three-dimensional verification space model, the mathematical expression of the ellipsoidal region is:

[0116]

[0117] In the formula, E(T, D, P) is the verification and evaluation function; T is the measured torque value in N·m; D is the measured axial displacement value in mm; P is the measured pressure value in N; T0 is the nominal torque value in N·m; D0 is the nominal axial displacement value in mm; P0 is the nominal pressure value in N; a, b, and c are the semi-axis lengths of the ellipsoid, corresponding to the tolerance ranges of torque, displacement, and pressure, respectively.

[0118] For the optimal region: E(T, D, P) ≤ 1, where a = 0.05·T0, b = 0.07·D0, c = 0.10·P0;

[0119] For the qualified region: E(T, D, P)≤4, where a=0.10·T0, b=0.15·D0, c=0.20·P0;

[0120] For defective regions: E(T, D, P) > 4.

[0121] The parameters are obtained as follows: T, D, and P are measured by the corresponding sensors; T0, D0, and P0 are obtained from the parameter database. These nominal values ​​are determined according to the nut specifications and materials.

[0122] The design principle of this model is based on the concept of multi-parameter quality evaluation. Traditional quality assessment often only considers whether a single parameter is within the acceptable range, ignoring the interaction between parameters. The ellipsoidal model uses three key parameters as coordinate axes in three-dimensional space, and comprehensively evaluates assembly quality by calculating the Mahalanobis distance from the measurement point to the nominal point. The advantage of using an ellipsoid instead of a cube is that it can consider the correlation between parameters. For example, when the torque is slightly higher, if the displacement also increases accordingly, the overall quality may still be acceptable; however, if the torque is high but the displacement is low, it may indicate a thread quality problem. This multi-dimensional evaluation method improves the defect detection rate by approximately 35% and reduces the false positive rate by approximately 25% compared to traditional single-parameter evaluation.

[0123] Optionally, the traceability code generation algorithm can be represented as:

[0124] C = Hash(T) s ||W i d||B n ||P n ||R c );

[0125] TC = Substring(C, 0, 16);

[0126] In the formula, C is the complete hash value; TC is the truncated traceability code; T s For assembly timestamps, accurate to milliseconds; W id Workstation number, 2 digits; B n This is the batch number for the nuts, 8 characters; P n For shaft-type parts, use a 12-character number; R c The result code is 2 characters; || represents the string concatenation operation; Hash() represents the SHA-256 hash function; Substring(C, 0, 16) extracts the first 16 characters of C.

[0127] The parameter acquisition method is: T s Generated by the internal clock of the control chip, in the format "YYYYMMDDHHMMSSMMM"; W id Pre-set in the control chip; B n Obtained from the nut supply device; P n Obtained from the conveyor; R c It is generated from the comprehensive verification results of multiple parameters. "A1" represents the excellent level, "B1" to "B9" represent different levels of qualified level, and "C1" to "C9" represent different types of defect levels.

[0128] The algorithm's design principle is based on the properties of cryptographic hash functions. SHA-256 provides high collision resistance, ensuring that traceability codes generated in different assembly scenarios are virtually impossible to duplicate. Using the assembly timestamp as one of the inputs ensures uniqueness in the time dimension, the workstation number solves the spatial dimension distinction, the batch number and part number are associated with material information, and the verification result code contains quality status information. Extracting the first 16 bits of the hash value is a compromise between balancing storage space and uniqueness; 16 characters can represent 16... 16 ≈18.45×10 18 There are 10 possible combinations, with a collision probability of less than 10. -10 This traceability code generation method provides richer information dimensions than a simple serial number scheme, facilitating subsequent data mining and quality analysis.

[0129] Specifically, the principle of this invention is as follows: The core technical principle of this invention lies in the construction of a shaft end nut assembly quality assurance system based on a two-stage tightening control strategy. By scientifically dividing the tightening stages and introducing multi-dimensional parameter verification, the threaded connection process can be refined and its quality assessed.

[0130] The first stage of the two-stage tightening control strategy is preliminary tightening, which is based on the principle of initial engagement stability in threaded connection theory. In this stage, the device operates the tightening motor in a low-speed, high-precision mode, combined with dynamic damping control technology, to monitor the reaction force between the nut and the threaded contact surface of the shaft end in real time, dynamically adjusting the compliance parameters of the assembly robot arm's end effector. This adaptive mechanical impedance adjustment mechanism allows the nut to automatically adjust its posture according to the guidance of the shaft end thread, conforming to the passive adaptation principle in mechanical assembly, effectively avoiding thread damage caused by forced assembly, and improving thread engagement efficiency. The key to the preliminary tightening stage is establishing a stable initial thread connection, laying the foundation for subsequent reinforcement tightening.

[0131] The key innovation of this invention lies in the intermediate verification step introduced between the two-stage tightening process, which is based on the displacement feedback principle in mechanical connection theory. By activating a laser displacement sensor to measure the axial displacement value of the nut and comparing it with a preset threshold, the quality status of the initial tightening stage can be accurately determined. This intermediate verification mechanism can promptly detect abnormalities such as misaligned threads and stripped threads, preventing defective connections from proceeding to the next stage and greatly improving the fault tolerance and stability of the tightening process.

[0132] The second stage of tightening is based on the progressive deformation theory in materials mechanics. A stepped torque-increasing algorithm divides the torque increase process into multiple small increments. Each increment has a precisely calculated torque increase value and duration, and parameters are optimized based on the thread contact stress distribution equation to ensure the thread bears uniform contact stress. This progressive loading method avoids the stress abrupt changes and stress concentration problems of traditional one-time tightening, resulting in a more stable connection between the nut and the shaft end, improving vibration resistance and anti-loosening performance.

[0133] After completing the two-stage tightening, this invention also designs a multi-parameter comprehensive verification mechanism. Using a three-dimensional verification space model, torque, axial displacement, and pressure values ​​are used as three dimensions to construct an evaluation space, defining three quality levels: excellent, qualified, and defective. This multi-dimensional evaluation method more comprehensively reflects the true state of the threaded connection than traditional single-parameter evaluation, providing a scientific basis for quality judgment.

[0134] The two-layer game model for tightening quality assessment provides a theoretical optimization framework for the entire two-stage tightening process. The upper-layer model optimizes the tightening parameters to maximize the quality index, while the lower-layer model simulates the connection performance to minimize the failure probability. The two-layer model is solved by coupling to find the optimal parameter combination that ensures both assembly efficiency and connection reliability, so that the two-stage tightening control strategy reaches its theoretical optimal state.

[0135] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0136] The specific implementation of step S01 involves first receiving the specifications of the shaft-end nut through a human-machine interface device. These parameters include basic information such as the nut diameter, thread pitch, and material strength. After standardizing this information, the system retrieves the corresponding preliminary tightening torque and reinforcement tightening torque values ​​from the parameter database. The parameter database uses a relational data structure and employs hash indexing technology for fast retrieval, with a retrieval time controlled within 5ms. The calculation of the preliminary tightening torque and reinforcement tightening torque values ​​is based on threaded connection theory, considering factors such as nut material, thread pitch, and coefficient of friction. The calculation formula is as follows:

[0137]

[0138] In the formula, T 初步 This is the initial tightening torque value, expressed in N·m; T 加固 The tightening torque value is given in N·m; k1 is the initial tightening coefficient, ranging from 0.2 to 0.4; k2 is the reinforcement tightening coefficient, ranging from 0.7 to 0.9; d is the effective diameter of the nut, in meters; F y Preload is N; P is pitch is m; μ t α represents the thread friction coefficient, ranging from 0.1 to 0.2; α is the thread half-angle, typically 30°. After obtaining these parameters, the system generates a two-stage tightening control command, including control parameters such as speed, torque, and angle for the initial tightening stage and the final tightening stage. This command is transmitted to each actuator via a CAN bus at a communication rate of 1 Mbps. The purpose of this step is to determine reasonable tightening parameters based on the nut specifications, providing fundamental data for subsequent assembly processes.

[0139] The specific implementation of step S02 involves controlling the assembly robotic arm to precisely grasp the nut and place it on the threaded end of the shaft. First, the system locates the nut using machine vision, employing a deep learning object detection algorithm with an accuracy exceeding 99.5%. The grasping process utilizes an impedance control strategy, automatically adjusting the clamping force according to the nut's specifications; for M6 to M12 nuts, the clamping force is set between 30 and 80 N. The robotic arm's movement uses a fifth-order polynomial interpolation algorithm to plan its trajectory, ensuring smooth movement and avoiding impact. When placing the nut on the threaded end of the shaft, the system activates dynamic damping control technology. The mathematical model for this technology is:

[0140]

[0141] In the formula, K d (t) represents the end stiffness parameter; C d (t) represents the end damping parameter; K d0 The initial stiffness parameter ranges from 5000 to 20000 N / m; C d0 The initial damping parameters range from 200 to 800 N·s / m; F r (t) represents the reaction force measured by the pressure sensor, in N; α is the rate of change of the reaction force, in N / s. K β is the stiffness adjustment factor, with a value ranging from 0.5 to 2.0 m / N; K α is the dynamic stiffness adjustment coefficient, with a value ranging from 0.01 to 0.05 m·s / N; C β is the damping adjustment coefficient, with a value ranging from 0.1 to 0.5 s / N; C This is the dynamic adjustment coefficient for damping, with a value ranging from 0.005 to 0.02s. 2 / N. This control strategy enables the robotic arm's end effector to have variable compliance. When a change in contact force is detected, the system dynamically adjusts the stiffness and damping parameters, allowing the nut to adaptively adjust its posture to achieve precise engagement with the shaft end thread. The system also sets a posture adjustment threshold; when the contact force exceeds a preset value (typically 5–10 N) or the rate of change is too large (typically 20–50 N / s), a fine-tuning process is triggered. The purpose of this step is to ensure accurate initial engagement between the nut and the shaft end thread, avoiding thread mis-threading due to improper placement angles and improving assembly success rate.

[0142] The specific implementation of step S03 involves activating the tightening mechanism for initial tightening. The tightening motor uses an S-shaped acceleration curve for starting, with a start-up time of 0.2–0.5 seconds to avoid shock caused by sudden starting. The operating parameters for the low-speed, high-precision mode are: speed controlled within the range of 20–60 r / min; precise torque output is achieved through vector control technology, with torque accuracy controlled within ±0.1%. Real-time monitoring of torque sensor data employs a sliding window averaging filtering algorithm, with a window width of 10–20 sampling points and a sampling frequency of 1000 Hz. The filtering expression is:

[0143]

[0144] In the formula, T 滤波 (n) represents the filtered torque value at time n, in N·m; T 原始 (i) represents the initial torque value at time i, in N·m; W is the width of the sliding window. The torque change rate is also calculated using the following formula:

[0145]

[0146] In the formula, S T T(t) represents the slope of the torque curve at time t, in N·m / s; T(t) represents the measured torque value at time t, in N·m; T(t-Δt) represents the measured torque value at time t-Δt, in N·m; Δt is the sampling time interval, in seconds, typically 0.001s; ω is the smoothing factor, in units of 1 / (N·m), ranging from 0.5 to 2.0 / (N·m). When the torque change rate exceeds 1 N·m / s, the system automatically reduces the motor speed. When the torque reaches the initial tightening torque value, the system quickly stops the tightening operation, with a stop response time of less than 50ms and torque overshoot controlled within 5%. For M8 standard nuts, the initial tightening torque is typically set to 5–8 N·m; for M12 standard nuts, the initial tightening torque is typically set to 15–25 N·m. The purpose of this step is to achieve initial tightening of the nut, creating conditions for subsequent tightening, while precisely controlling the torque to avoid thread damage caused by over-tightening.

[0147] The specific implementation of step S04 involves performing intermediate verification to evaluate the initial tightening quality. The system activates a laser displacement sensor and uses the triangulation principle to measure the axial displacement of the nut. The sensor operates at a wavelength of 650nm, with a measurement range of 0–50mm, a resolution of 0.001mm, and a sampling frequency of 5000Hz. The system selects 3–5 measurement points on the nut circumference and calculates the average axial displacement value by fitting a plane using the least squares method. The displacement calculation formula is as follows:

[0148]

[0149] In the formula, D avg The average axial displacement value is expressed in mm; D i ε represents the displacement value of the i-th measurement point, in mm; N is the number of measurement points; 测量 The measurement system error correction value, in mm, is typically 0.001–0.003 mm. The measured axial displacement value is compared with a preset axial displacement threshold. A fuzzy logic decision algorithm is used to evaluate the initial tightening quality. The fuzzy membership function is defined as follows:

[0150]

[0151] In the formula, μ 符合 (x) represents the membership degree that meets the requirements; x is the measured value; x0 is the standard value; δ1 is the inner threshold, usually 5% of the standard value; δ2 is the outer threshold, usually 10% of the standard value. For M8 nuts, the standard value for axial displacement is usually 1.1–1.2 mm; for M12 nuts, the standard value for axial displacement is usually 1.6–1.7 mm. If the membership degree is greater than 0.8, it is considered compliant; if the membership degree is between 0.5 and 0.8, it is considered marginal; if the membership degree is less than 0.5, it is considered non-compliant. For non-compliant cases, the system will record the anomaly and enter the anomaly handling process. The purpose of this step is to check the preliminary tightening quality before formal tightening, avoid continuing tightening under poor foundation conditions, and improve the final assembly quality.

[0152] The specific implementation of step S05 involves entering the reinforcement tightening stage to achieve final fixing of the nut. The system first calculates the torque difference between the initial tightening torque value and the reinforcement tightening torque value. Then, based on a stepped torque increase algorithm, this difference is divided into multiple steps. The formula for calculating the torque increase value and duration of each step is as follows:

[0153]

[0154] In the formula, ΔT i Δt represents the torque increase at the i-th step, in N·m. i T represents the duration of the i-th step, in seconds. f The tightening torque value is given by the reinforcement specification, in N·m; T0 is the initial tightening torque value, in N·m; γ i For the torque distribution coefficient, satisfying The value range is 0.15 to 0.25; δ iσ is the fluctuation adjustment coefficient, ranging from -0.1 to 0.1; N is the total number of steps, ranging from 5 to 8; τ0 is the baseline duration, ranging from 0.5 to 1.0 s; η is the time adjustment coefficient, ranging from 0.5 to 2.0; σ max (i-1) represents the maximum contact stress at the end of the (i-1)th step, in MPa; σ ref The reference stress value is in MPa; it is 300 MPa for M8 nuts and 400 MPa for M12 nuts. The contact stress is calculated using the following thread contact stress distribution equation:

[0155]

[0156] In the formula, σ(z) is the contact stress at the axial position z, in MPa; T is the applied torque, in N·m; P is the thread pitch, in mm; r is the effective thread radius, in mm; A e The effective contact area of ​​the thread is expressed in mm. 2 f(z) is the stress distribution function; λ is the material property correction coefficient, dimensionless, ranging from 0.1 to 0.3; E m E represents the elastic modulus of the nut material, expressed in GPa. s R is the elastic modulus of the shaft end material, in GPa. a ε represents the surface roughness of the shaft end thread, in μm. σ The stress calculation error term is expressed in MPa, with a range of ±15 MPa; z is the axial distance from the nut end face, in mm; z0 is the axial position of the first contact thread, in mm; μ is the stress attenuation coefficient, dimensionless, ranging from 0.2 to 0.5; κ is the pitch periodic stress fluctuation coefficient, dimensionless, ranging from 0.05 to 0.15. During tightening, the system also monitors the slope of the torque curve and the change in the dynamic friction coefficient. The formula for calculating the dynamic friction coefficient is:

[0157]

[0158] In the formula, μ d (θ) is the dynamic friction coefficient at rotation angle θ, dimensionless; T(θ) is the torque value at rotation angle θ, in N·m; F(θ) is the axial pressure value at rotation angle θ, in N; φ is the thread angle, in rad; r e ξ is the effective radius of the thread, in mm; ξ is the break-in coefficient, dimensionless, ranging from 0.01 to 0.05; v is the break-in exponent, dimensionless, ranging from 0.5 to 1.5; ζ is the break-in attenuation coefficient, in rad. -1 The value ranges from 0.01 to 0.05 rad. -1;ε μ This is the error term for the friction coefficient calculation, with a value range of ±0.02. The purpose of this step is to ensure a uniform distribution of thread contact stress through a gradual tightening method, avoiding thread stripping or nut breakage caused by instantaneous high torque, and improving tightening reliability.

[0159] The specific implementation of step S06 involves activating a pressure sensor to monitor changes in tightening pressure in real time during the tightening phase, enabling timely detection of any abnormalities. The pressure sensor samples at a frequency of 2000Hz, and the data is input to the control system via a 16-bit A / D converter. The system employs wavelet analysis to perform time-frequency analysis on the pressure signal; the wavelet transform formula is as follows:

[0160]

[0161] In the formula, W f (a, b) are wavelet coefficients; f(t) is the pressure signal; ψ * denoted by , where is the conjugate of the wavelet basis function; 'a' is the scaling parameter; and 'b' is the translation parameter. The system uses the Daubechies wavelet as the basis function, with the scaling parameter 'a' ranging from 1 to 8. By analyzing the statistical characteristics of wavelet coefficients at different scales, the system can identify pressure change patterns. During normal tightening, the pressure change should increase smoothly, and the pressure change rate is calculated using the following formula:

[0162]

[0163] In the formula, R p P(t) represents the pressure change rate at time t, in N / s; P(t) represents the pressure value at time t, in N; P(t-Δt) represents the pressure value at time t-Δt, in N; Δt is the sampling time interval, in s; β is the smoothing coefficient, in 1 / N, with a value range of 0.001 to 0.01 / N. For M8 nuts, the preset pressure range is 1000 to 1800 N; for M12 nuts, the preset pressure range is 2500 to 4000 N. If the pressure change rate exceeds the threshold (usually set to 2000 N / s), or the pressure fluctuation frequency exceeds 10 Hz, the system determines it to be an abnormal situation. The abnormal state discrimination formula is:

[0164]

[0165] In the formula, S 异常 This is an abnormal status indicator; 1 indicates abnormality, and 0 indicates normality. R 阈值 The pressure change rate threshold is expressed in N / s; P 上限 This is the upper limit of pressure, in N; P 下限 This is the lower limit of pressure, in N; f pThe pressure fluctuation frequency is expressed in Hz; f 阈值 This is the frequency threshold, measured in Hz. When an anomaly is detected, the system immediately interrupts the tightening operation, stops the tightening motor and maintains its current position, and triggers an anomaly alarm. The purpose of this step is to promptly detect potential problems during tightening, such as thread jamming or material deformation, by monitoring changes in tightening pressure in real time, thus preventing damage to components from continued tightening.

[0166] The specific implementation of step S07 involves performing a multi-parameter comprehensive verification after the tightening stage to fully evaluate the tightening quality. The system simultaneously collects data from torque sensors, laser displacement sensors, and pressure sensors at a sampling frequency of 1000Hz and a sampling duration of 0.5s, acquiring 500 data points. The system calculates the average, standard deviation, and coefficient of variation of the torque, axial displacement, and pressure values ​​to construct a feature vector. The tightening quality evaluation employs a two-layer game theory model, with the objective function of the upper-layer model being:

[0167]

[0168] In the formula, X = [T t D t S t , t s ] T T represents the decision vector of the upper-level model. t The optimized target torque value is expressed in N·m; D t The optimized ideal axial displacement value is expressed in mm; S t The optimized torque change rate is expressed in N·m / s; t s The optimized staircase duration, in seconds; Y * T represents the optimal solution of the lower-level model given X; a The actual torque value measured by the torque sensor, in N·m; D a The axial displacement value measured by the laser displacement sensor, in mm; σ p Standard deviation of pressure measured by the pressure sensor, in N; ρ is the average pressure value measured by the pressure sensor, in N; w1, w2, and w3 are weighting coefficients, satisfying w1 + w2 + w3 = 1, typically w1 = 0.4, w2 = 0.3, and w3 = 0.3; ρ is the penalty coefficient, ranging from 0.1 to 0.5; C is the coupling matrix, describing the influence of the lower-level model output on the upper-level objective. The objective function of the lower-level model is:

[0169]

[0170] In the formula, Y = [σ max , ε p , σvar , σ avg ] T σ is the state vector of the lower-level model; max To predict the maximum stress value, the unit is MPa; ε p To predict the amount of plastic deformation, dimensionless; σ var To predict stress variability, the unit is MPa; σ avg The mean stress value is predicted, in MPa; σ yield ε represents the yield strength of the material, measured in MPa. crit σ is the critical plastic deformation quantity, dimensionless; opt The optimal average stress value is given in MPa; α, β, γ, and δ are weighting coefficients satisfying α + β + γ + δ = 1, typically α = 0.5, β = 0.3, γ = 0.1, and δ = 0.1. The system uses the alternating direction multiplier method to solve the two-level game model, with the iterative convergence condition being a relative error of less than 0.1%. The purpose of this step is to comprehensively evaluate the tightening quality through multi-parameter integrated analysis and provide optimized parameter references for similar future working conditions, thereby improving assembly quality and efficiency.

[0171] The specific implementation of step S08 involves performing corresponding operations based on the multi-parameter comprehensive verification results to achieve graded processing for different quality levels. The system compares the verification results with a preset three-dimensional verification space model, the mathematical expression of which is:

[0172]

[0173] In the formula, E(T, D, P) is the verification and evaluation function; T is the measured torque value in N·m; D is the measured axial displacement value in mm; P is the measured pressure value in N; T0 is the nominal torque value in N·m; D0 is the nominal axial displacement value in mm; P0 is the nominal pressure value in N; a, b, and c are the semi-axis lengths of the ellipsoid, corresponding to the tolerance ranges of torque, displacement, and pressure, respectively. For the excellent region: E(T, D, P) ≤ 1, where a = 0.05·T0, b = 0.07·D0, c = 0.10·P0; for the qualified region: E(T, D, P) ≤ 4, where a = 0.10·T0, b = 0.15·D0, c = 0.20·P0; for the defective region: E(T, D, P) > 4. For nuts of excellent quality, the system directly passes the quality acceptance test, records complete parameters in the database, and marks them as Grade A quality. For nuts of acceptable quality, the system records them in the traceability database and marks them as Grade B quality, while also noting that attention is needed and they should be closely monitored in subsequent use. For nuts of defective quality, the system initiates an automatic disassembly program, controlling the tightening motor to run in reverse. The disassembly speed control strategy is as follows:

[0174]

[0175] In the formula, ω 拆卸 T(t) represents the disassembly rotation speed at time t, in r / min; T(t) represents the torque value at time t, in N·m; T 最大 The torque value at the start of disassembly, in N·m; ω 初始 The initial disassembly speed is 50–100 r / min; ω 最低 The minimum disassembly speed is set at 20 r / min. After disassembly, the system sends the nut to the inspection area for analysis and records detailed defect data. The purpose of this step is to classify the assembly quality based on the verification results, ensuring that the final product quality meets requirements, and providing data support for quality traceability.

[0176] The specific implementation of step S09 involves, after completing all tightening and verification processes, storing the assembly data in a database and generating a unique traceability code to establish a quality traceability system. The system first compresses the torque curve data and employs a piecewise linear fitting algorithm, the mathematical expression of which is:

[0177]

[0178] In the formula, (T) j θ j ) represents the points on the original torque curve, a total of n; k represents the number of line segments after fitting; (T i θ i ) represents the endpoint of the line segment; L i For the i-th line segment; d((T) j θ j ), L i The distance from the point to the line segment is denoted as . The system uses a dynamic programming algorithm to solve this problem, ensuring a compression ratio of at least 10:1 while maintaining a restoration error of less than 1%. The compressed data, along with all measurement parameters, is stored in the database in JSON format, with data packets typically ranging from 10 to 50 KB in size. The algorithm for generating unique traceability codes is as follows:

[0179] C = Hash(T) s ||W i d||B n ||P n ||R c );

[0180] TC = Substring(C, 0, 16);

[0181] In the formula, C is the complete hash value; TC is the truncated traceability code; T s For assembly timestamps, accurate to milliseconds; W idWorkstation number, 2 digits; B n This is the batch number for the nuts, 8 characters; P n For shaft-type parts, use a 12-character number; R c The verification result code is a 2-character string; || represents string concatenation; Hash() represents the SHA-256 hash function; Substring(C, 0, 16) extracts the first 16 characters of C. A one-to-one association is established between the traceability code and the product, with database foreign key constraints ensuring data integrity. The purpose of this step is to establish a complete quality traceability system, enabling quality issues throughout the product's lifecycle to be traced back to specific assembly parameters and process data, providing a basis for quality improvement and fault analysis.

[0182] The base is constructed from high-strength cast iron, with a thickness of 50-80mm and an anti-corrosion surface treatment, achieving a hardness of HB220-260. The metal support is welded from Q235 steel, with a weld thickness of 8-12mm. The overall structure employs a triangular stability support design, with a support height of 800-1200mm. The leveling mechanism includes six M16 bolts, each with a load-bearing capacity of 3000N, and anti-loosening washers around the bolts. The vibration damping device consists of eight rubber damping pads, each measuring 120×80×25mm, with a Shore A hardness of 70-80 and a vibration damping frequency response range of 5-80Hz. It effectively absorbs vibrations in the 20-500Hz frequency range, improving equipment stability by over 85%. The base also features a grounding protection device with a grounding resistance of no more than 4Ω to ensure operational safety.

[0183] The conveying mechanism utilizes a wear-resistant, non-slip rubber conveyor belt with a thickness of 5-8mm, a width of 300-500mm, and a surface hardness of Shore A60-70. The drive motor is a servo motor with a power of 1.5-2.2kW, a speed of 0-1500r / min, and torque accuracy controlled within ±0.5%. The transmission wheel is made of 45# steel, with a diameter of 200-250mm, chrome-plated surface, and a hardness of HRC48-52. The position sensor is a photoelectric sensor with a resolution of 0.1mm, a response time of less than 5ms, and a detection range of 5-500mm. The conveying mechanism is also equipped with a speed regulation module, enabling stepless speed adjustment within the range of 0.1-1.0m / s, and achieving closed-loop speed control through a PID algorithm with a control accuracy of ±1%. The mechanism also features an emergency stop device with a stop response time of less than 0.2s.

[0184] The specific implementation of the nut supply device uses a 304 stainless steel vibratory feeder with a diameter of 450-600mm. The vibration frequency is adjustable within the range of 30-60Hz, and the amplitude is 0.5-2.0mm. The feeding track is made of hard anodized aluminum alloy with a surface hardness of HV500 or higher, and the track width is 2-5mm larger than the maximum outer diameter of the nut. The dispensing mechanism uses a pneumatic control system with a working air pressure of 0.4-0.6MPa and cylinder stroke accuracy controlled within ±0.1mm. The entire supply device adjusts the vibration frequency and amplitude through an adaptive control algorithm based on fuzzy control theory. By monitoring the density and flow rate of the nuts on the track in real time, the vibration parameters are dynamically adjusted to match the feeding rate with the assembly cycle, achieving a feeding cycle stability of over 95%. The supply device also has an anti-clogging detection system. When a potential blockage is detected, the vibration amplitude is automatically increased and the vibration direction is changed to clear the blockage.

[0185] The assembly robotic arm employs a 4-DOF multi-joint structure, with each joint having a range of motion of ±180° and a joint positioning accuracy of ±0.05mm. The servo motor is a permanent magnet synchronous motor with a power of 400-800W, a maximum speed of 3000r / min, and torque accuracy controlled within ±0.2%. The encoder is an absolute encoder with a 19-bit resolution (524,288 pulses / revolution) and an accuracy of ±0.001°. The nut gripping tool uses a pneumatic gripper design, with a gripping force adjustable from 10 to 200N. The gripper surface is covered with polyurethane material to increase the friction coefficient to over 0.8. The robotic arm uses a composite motion planning algorithm based on fifth-order polynomial interpolation, incorporating Bezier curve smoothing to achieve smooth acceleration and deceleration during gripping, with the acceleration change rate controlled within 200m / s². 3 Within this range, the grasping success rate reaches over 99.8%. The robotic arm is also equipped with a force feedback system, which achieves force control through current detection, with a force control accuracy of ±0.5N.

[0186] The tightening mechanism employs a high-precision servo motor with a power of 0.8–1.5 kW, a speed range of 0–2000 r / min, and a torque control accuracy of ±0.1%. The torque drive shaft is made of 40Cr alloy steel, heat-treated to achieve a hardness of HRC42–46, and has a shaft diameter of 15–25 mm. The tightening head features a modular design, allowing for quick replacement according to different nut specifications; the head material is high-speed steel with a hardness of HRC58–62. The tightening mechanism utilizes vector control technology, and the control equation is expressed as:

[0187]

[0188] ψ(t)=Li(t)+ψ f ;

[0189]

[0190] In the formula, u(t) is the stator voltage vector; i(t) is the stator current vector; ψ(t) is the stator flux linkage vector; ψ f R is the flux linkage vector of the permanent magnet; R is the stator resistance; L is the stator inductance; T e denoted by , p represents the electromagnetic torque; p represents the number of pole pairs; and J represents the 90° rotation matrix. The tightening mechanism achieves dual closed-loop control of speed and torque, maintaining high-precision control under low-speed, high-torque conditions, with speed fluctuations controlled within ±0.5% and torque fluctuations within ±0.2%. A shock-absorbing buffer device is installed between the tightening head and the motor, employing hydraulic damping principles to effectively reduce impact loads and extend equipment lifespan.

[0191] The specific implementation of the calibration system employs a strain gauge torque sensor with a range of 0–200 N·m, an accuracy class of 0.2, and a frequency response of 1000 Hz. A laser displacement sensor utilizes the triangulation principle, with a range of 0–50 mm, a resolution of 0.001 mm, and a sampling rate of 10 kHz. A pressure sensor employs a piezoresistive design, with a range of 0–5000 N, an accuracy class of 0.5, and a response time of less than 1 ms. The calibration system integrates multi-sensor information using a Kalman filter algorithm; the state equation and observation equation are as follows:

[0192] x k =Fx k-1 +w k-1 ;

[0193] z k =Hx k +v k ;

[0194] In the formula, x k z is the state vector at time k; k Let be the observation vector at time k; F is the state transition matrix; H is the observation matrix; w k-1 The process noise conforms to a normal distribution with a mean of 0 and a covariance of Q; v k The observed noise conforms to a normal distribution with a mean of 0 and a covariance of R. The algorithm achieves state estimation through two steps: prediction and update, effectively filtering out measurement noise and improving measurement accuracy by more than 20%. The calibration system also implements a self-calibration function, automatically performing zero-point and full-scale calibration each time it is powered on, ensuring stable accuracy over long-term use. The calibration system's data acquisition frequency is 1000Hz, and the acquired data undergoes wavelet transform for noise reduction, improving the signal-to-noise ratio by more than 15dB, providing a high-quality data foundation for subsequent parameter analysis.

[0195] To better understand and implement this invention, a specific application scenario is provided in Example 2: On the reducer production line of a heavy equipment manufacturing research center, an automatic shaft-end nut assembly and tightening device that is easy to verify was implemented for assembling the fixed nut of the output shaft of a high-precision reducer. This reducer is mainly used in the motion control system of precision machine tools, and the accuracy and reliability requirements of the nut connection are extremely high. Insufficient or excessive bearing preload caused by poor nut tightening will directly affect the service life and operating accuracy of the reducer. By implementing the automatic shaft-end nut assembly and tightening device of this invention, the research team solved the problems of poor nut assembly consistency, inability to effectively guarantee tightening quality, and difficulty in tracing assembly parameters that exist in traditional manual assembly or simple automated assembly.

[0196] The reducer output shaft fixing nut used in the implementation case is an M16×1.5 precision nut, made of 40Cr alloy steel, with a QPQ nitriding treatment and a hardness of HRC54-58. The nut is assembled on the reducer output shaft end to fix the bearing and apply preload. Through analysis of the nut tightening process on the output shaft end, the research team found that approximately 15% of different batches of products had nut tightening quality issues, resulting in insufficient or excessive bearing preload, severely affecting product performance and lifespan.

[0197] Based on this invention, the research team designed and implemented an automatic assembly and tightening device for shaft-end nuts. This device consists of a base, a conveying mechanism, a nut supply device, an assembly robotic arm, a tightening mechanism, a calibration system, and a control chip. The core parameters of the device are shown in Table 1.

[0198] Table 1 Main Parameters of Automatic Assembly and Tightening Device for Shaft End Nuts

[0199] Component Name parameter numerical values Tighten the motor power 1.2kW Tighten the motor Maximum torque 150 N·m Tighten the motor Speed ​​range 0~1800r / min Torque sensor range 0~180N·m Torque sensor Accuracy level Level 0.2 Laser displacement sensor range 0~25mm Laser displacement sensor resolution 0.001mm pressure sensor range 0～8000N pressure sensor Accuracy level Level 0.5 Assembly robotic arm Degrees of freedom 4 Assembly robotic arm Maximum load 10kg Assembly robotic arm Repeatability ±0.03mm

[0200] During installation, the reducer output shaft assembly is conveyed to the assembly station via a conveyor mechanism, with nuts supplied by a nut supply device. An assembly robotic arm grasps the nut and precisely places it onto the threaded end of the shaft, ensuring accurate engagement between the nut and thread through dynamic damping control technology. The tightening process employs a two-stage control strategy: initial low-torque tightening followed by stepped torque-increasing tightening. Throughout the tightening process, torque, displacement, and pressure parameters are monitored in real-time by multiple sensors, and a comprehensive evaluation is performed based on a two-layer game theory model for tightening quality assessment.

[0201] During the application of the device, the research team systematically optimized the tightening parameters of the M16×1.5 precision nut. The core parameters are shown in Table 2.

[0202] Table 2 M16×1.5 Precision Nut Tightening Parameter Optimization Table

[0203]

[0204]

[0205] The dynamic damping control parameters used during the implementation of the device are shown in Table 3:

[0206] Table 3 Dynamic Damping Control Parameters

[0207] Parameter name symbol numerical values Stiffness adjustment coefficient <![CDATA[α K ]]> 1.5m / N Stiffness dynamic adjustment coefficient <![CDATA[β K ]]> 0.03 m·s / N Damping adjustment coefficient <![CDATA[α C ]]> 0.3s / N Damping dynamic adjustment coefficient <![CDATA[β C ]]> <![CDATA[0.01s 2 / N]]> reaction force threshold - 12N Threshold of rate of change of reaction force - 35N / s

[0208] During implementation, the weight coefficients of the two-layer game model for tightening quality assessment are set as shown in Table 4:

[0209] Table 4 Weighting coefficients for the tightening quality assessment model

[0210] hierarchy parameter symbol numerical values upper-level model Torque accuracy weight <![CDATA[w1]]> 0.45 upper-level model Displacement accuracy weight <![CDATA[w2]]> 0.30 upper-level model Pressure stability weight <![CDATA[w3]]> 0.25 upper-level model Penalty coefficient ρ 0.25 Lower-level model Maximum stress weight α 0.55 Lower-level model Plastic deformation weight β 0.25 Lower-level model Stress variability weight γ 0.12 Lower-level model Mean stress weight δ 0.08

[0211] The assembly efficiency and quality data of this device during actual production are shown in Table 5:

[0212] Table 5 Comparison of Device Application Effects

[0213] index Traditional manual assembly General automated assembly The device of the present invention Assembly efficiency (pieces / hour) 15 45 42 Nut meshing success rate (%) 92 95 99.5 Tightening torque consistency (±%) 12 8 3 Axial displacement consistency (±%) 18 12 5 Tightening failure rate (%) 15.2 8.7 1.3 Return rate (%) 5.3 2.8 0.5 Quality traceability capability none part whole

[0214] During three months of continuous production verification, the device completed the assembly of 6,320 pieces. The tightening defect rate decreased from 15.2% to 1.3%, the consistency of the reducer bearing preload significantly improved, the reducer's operational accuracy and stability increased by 38%, and its service life was extended by approximately 25%. Simultaneously, the quality traceability system enabled the recording and traceability of the assembly parameters for each nut, providing strong data support for product quality control and fault analysis.

[0215] By analyzing torque, displacement, and pressure data during the tightening process, the research team identified the main problems in traditional tightening methods, as shown in Table 6.

[0216] Table 6. Problem Analysis of Traditional Tightening Methods

[0217]

[0218]

[0219] Traditional nut assembly and tightening methods mainly include manual tightening and simple automated tightening. While manual tightening offers high flexibility, it is inefficient, inconsistent, and cannot precisely control torque and axial displacement. Simple automated tightening, although more efficient, typically only controls a single parameter (such as torque), failing to monitor and control the multi-dimensional parameters during nut assembly, leading to unstable tightening quality. Both methods lack effective quality assessment and traceability mechanisms, failing to provide data support for product lifecycle management.

[0220] Compared to traditional methods, the automatic assembly and tightening device for shaft-end nuts of this invention has the following significant advancements: First, through dynamic damping control technology, precise engagement between the nut and the shaft-end thread is achieved, solving the problem of poor initial engagement. Second, a stepped torque-increasing algorithm is adopted to avoid thread damage caused by instantaneous high torque. Third, through multi-parameter comprehensive verification, a comprehensive evaluation of tightening quality is achieved, no longer relying solely on a single torque parameter. Fourth, the two-layer game model for tightening quality evaluation makes the optimization of assembly parameters more scientific and reasonable, improving the stability of assembly quality. Finally, a complete quality traceability system provides a data foundation for product lifecycle management. These advancements combined significantly improve the quality and reliability of nut assembly and tightening, providing strong assurance for the performance and lifespan of high-precision reducers.

[0221] It should be noted that the variables involved in this invention are explained in detail in Table 7 below.

[0222] Table 7 Variable Explanation Table

[0223]

[0224]

[0225]

[0226] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. An automatic assembly and tightening device for shaft-end nuts that is easy to verify, comprising a base, a conveying mechanism, a nut supply device, an assembly robotic arm, a tightening mechanism, a verification system, a control chip, a data processing unit, a human-machine interface device, and a power management device, characterized in that, The control chip has a tightening control module, which performs the following steps: receiving the specification parameters of the nut at the shaft end, extracting the initial tightening torque value and the reinforcement tightening torque value from the parameter database according to the specification parameters, and generating a two-stage tightening control command; controlling the assembly robot arm to place the nut at the thread at the shaft end; The tightening mechanism is started for the initial tightening stage, and the tightening motor is controlled to run in a low-speed, high-precision mode; intermediate verification is performed by activating the laser displacement sensor to measure the axial displacement value of the nut; if the initial tightening stage verification is passed, the reinforcement tightening stage is entered, and the tightening motor is controlled using a stepped torque increase algorithm. After completion, a multi-parameter comprehensive verification is performed. Among them, the dynamic damping control technology monitors the reaction force between the nut and the threaded contact surface of the shaft end in real time. The reaction force is measured by a pressure sensor. The compliance parameters of the end effector of the assembly robot arm are dynamically adjusted according to the magnitude of the reaction force, so that the nut can automatically adjust its posture according to the guidance of the threaded shaft end. Among them, the stepped torque-increasing algorithm divides the torque-increasing process of the tightening stage into multiple small-amplitude steps. Each step includes two parameters: torque increase value and duration. The tightening control module detects the nut status at the end of each step and only enters the next step after confirming that it is stable. In the multi-parameter comprehensive verification step, the actual torque value measured by the torque sensor, the axial displacement value measured by the laser displacement sensor, and the pressure value measured by the pressure sensor are collected simultaneously to construct a two-layer game model for tightening quality assessment. The two-layer game model for tightening quality assessment adopts a leader-follower structure. The upper-layer model optimizes the tightening parameter control strategy, and the lower-layer model simulates the mechanical response of the connected components under given tightening parameters. The upper-layer model maximizes the tightening quality index through a nonlinear objective function, and the lower-layer model minimizes the failure probability index through a nonlinear objective function. According to the multi-parameter comprehensive verification results, corresponding operations are performed. For excellent grades, the product passes directly; for qualified grades, it is recorded in the traceability database and marked as requiring attention; for defective grades, an automatic disassembly procedure is initiated for reassembly.

2. The automatic assembly and tightening device for shaft end nuts according to claim 1, characterized in that, The base includes a metal bracket, a leveling mechanism, and a shock-absorbing device. The metal bracket is made of cast iron, the leveling mechanism includes multiple leveling bolts, and the shock-absorbing device includes rubber shock-absorbing pads. The conveying mechanism includes a conveyor belt, a drive motor, a transmission wheel, and a position sensor. The conveyor belt is made of non-slip rubber, the drive motor is connected to the transmission wheel, and the position sensor is installed at the end of the conveyor belt to detect the position of shaft parts. The nut supply device includes a vibratory feeder, a feeding track, and a distribution mechanism. The vibratory feeder transports the nuts to the feeding track, and the distribution mechanism ensures that each nut is output individually to the gripping position.

3. The automatic assembly and tightening device for shaft end nuts according to claim 2, characterized in that, The assembly robotic arm includes a multi-joint structure, servo motors, encoders, and a nut gripper. The multi-joint structure consists of multiple links, with servo motors installed at each joint. The encoders are used to monitor the joint angles, and the nut gripper is installed at the end of the robotic arm. The tightening mechanism includes a tightening motor, a torque drive shaft, and a tightening head. The tightening motor is connected to the torque drive shaft, which is connected to the tightening head. The tightening head contacts the nut to perform the tightening operation.

4. The automatic assembly and tightening device for shaft end nuts according to claim 3, characterized in that, The calibration system includes a torque sensor, a laser displacement sensor, and a pressure sensor. The torque sensor is installed on the torque drive shaft to measure the tightening torque, the laser displacement sensor is installed at the assembly station to measure the axial displacement of the nut, and the pressure sensor is installed at the contact point between the tightening head and the nut to measure the tightening pressure.

5. The automatic assembly and tightening device for shaft end nuts according to claim 4, characterized in that, The process of controlling the assembly robot arm to place the nut on the threaded end of the shaft is achieved by using dynamic damping control technology to ensure that the nut and the threaded end of the shaft are initially engaged, thus avoiding thread mis-threading caused by improper placement angle.

6. The automatic assembly and tightening device for shaft end nuts according to claim 5, characterized in that, The intermediate verification step involves comparing the axial displacement value of the nut with the preset axial displacement threshold to determine whether the initial tightening stage meets the technical requirements.

7. The automatic assembly and tightening device for shaft end nuts according to claim 6, characterized in that, During the tightening phase, a pressure sensor is activated to monitor changes in tightening pressure in real time. If a sudden change in tightening pressure or a pressure exceeding the preset range is detected, the tightening operation is immediately interrupted and an abnormal alarm is issued.

Images