An adaptive closed-loop control method for output torque of an electric screwdriver

Abstract

The present application belongs to the technical field of industrial automation control, and particularly relates to a self-adaptive closed-loop control method for output torque of an electric screwdriver, which comprises the following steps: synchronously collecting real-time running speed and real-time running current at a preset frequency; obtaining average no-load speed and average no-load current; calculating a load mutation coefficient; obtaining corresponding meshing initial speed and meshing initial current; extracting deep meshing speed and deep meshing current again; calculating a material hardness mapping value; obtaining an adaptive target torque and an adaptive target speed; obtaining a real-time output torque; and outputting a tightening completion signal. The present application solves the technical problem that the prior art cannot perceive and adapt to dynamic load changes in real time by using fixed parameter open-loop control, resulting in unstable fastening quality and the inability to balance efficiency and safety.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology. More specifically, this invention relates to an adaptive closed-loop control method for the output torque of an electric screwdriver. Background Technology

[0002] In the field of industrial automation control, closed-loop torque control is a key technology for achieving precision assembly and motion control. As industrial scenarios demand increasing adaptability to dynamic loads, real-time adaptive adjustment of output torque is crucial for ensuring control accuracy, preventing equipment damage, and improving operational consistency. To assess fastening quality, it is essential to monitor and manage torque fluctuations, speed response, and load characteristics throughout the entire fastening process in a closed-loop manner.

[0003] In the field of electronic control technology, motor control schemes based on digital conversion are currently widely adopted. For example, Chinese patent document CN215548275U discloses an electric screwdriver with electronically adjustable torque. This screwdriver uses a brushless motor installed in the front and rear housings, and a magnetic encoder in the stator assembly, in conjunction with a controller, to achieve electronic adjustment of the motor's speed and torque based on digital conversion. Simultaneously, the converted torque and speed data are output to a display. This method allows for torque preset under ideal static conditions.

[0004] However, open-loop control relying on manually preset single fixed parameters cannot perceive and adapt to dynamic load changes during the tightening process in real time, resulting in dynamic load adaptation failure and tightening characteristic distortion. Tightening is a continuous dynamic process comprising three stages: high-speed feed under no-load, thread engagement and climbing, and head contact and pressure holding. Its physical resistance exhibits a non-linear abrupt change with the screw feed depth, and changes in workpiece material and screw type in actual working conditions further distort the load response curve. Existing technologies assume a consistent load throughout the process, failing to perceive stage-specific characteristic changes and material-related load differences. When the screw transitions from no-load rotation to the head contact stage, the fixed high torque generates a momentary impact force, easily causing screw stripping, breakage, or damage to the surface of precision workpieces due to pressure. If a low speed and low torque are preset to protect the workpiece, the feed efficiency during the no-load stage will be low. When facing mixed assembly scenarios with varying material hardness or screw pitch, the control model cannot automatically identify the workpiece type based on the real-time differential characteristics of current and rotation speed, resulting in fluctuating output torque that is either too tight (damaging) or too loose (loose). In scenarios requiring consistent fastening, the control accuracy of traditional solutions will decrease due to the lack of adaptive correction for the dynamic impedance slope, failing to meet the requirements of modern industry for adaptive closed-loop fastening. Summary of the Invention

[0005] To address the technical problem of existing technologies using fixed-parameter open-loop control failing to detect and adapt to dynamic load changes in real time, resulting in unstable fastening quality and a trade-off between efficiency and safety, this invention provides an adaptive closed-loop control method for the output torque of an electric screwdriver, comprising:

[0006] The system synchronously collects real-time operating speed and current at a preset frequency; it integrates and averages the real-time operating speed and current to obtain the average no-load speed and average no-load current; based on the phenomenon that hard physical contact can cause a sudden drop in speed and a surge in current, it calculates the load mutation coefficient based on the average no-load speed and average no-load current; based on the load mutation coefficient, it determines the moment of engagement and obtains the corresponding initial engagement speed and initial engagement current; after a preset fixed monitoring cycle, it extracts the deep engagement speed and deep engagement current again; based on the synchronous change characteristics of current and speed caused by material hardness, it calculates the material hardness mapping value by combining the deep engagement data and the initial steady-state data; based on the material hardness mapping value and the initial engagement current, it obtains the adaptive target torque and adaptive target speed; using the adaptive target speed as a feedforward reference, it records the product of the system's preset motor torque coefficient and the real-time operating current as the real-time output torque to obtain the real-time output torque; based on the real-time output torque and the adaptive target torque, it outputs a tightening completion signal.

[0007] This invention overcomes, to some extent, the inherent limitations of traditional electric screwdrivers that use fixed-parameter open-loop control and assume consistent load throughout the tightening process. Specifically, it constructs a load mutation coefficient by extracting the time-domain deviation of motor current and speed, amplifying the cross-coupling response at the moment of hard physical contact at the algorithm level. This solution accurately captures the transient mechanical characteristics of screw-workpiece engagement at the underlying electronic control logic. It introduces a material hardness mapping value to replace fixed empirical parameters for allocating target torque, and, combined with a tightening judgment mechanism based on continuous sampling verification, effectively eliminates underlying interference from transient noise and operating condition fluctuations. It also eliminates the defects of contact-moment impact damage and inconsistent tightening quality caused by fixed parameters and mixed-line assembly. This not only achieves dynamic torque adjustment throughout the entire cycle in accordance with the laws of tightening mechanics, but also realizes adaptive and refined control of the tightening process by combining the physical impedance of the workpiece and the engagement state.

[0008] Preferably, the real-time operating speed and real-time operating current are synchronously acquired at a preset frequency, including: Start the electric screwdriver to tighten the screws, and synchronously collect the real-time operating speed and real-time operating current of the motor at a preset frequency.

[0009] Preferably, obtaining the average no-load speed and average no-load current includes: Preset length is The statistical window is used to integrate and average the real-time operating speed and real-time operating current within the statistical window to obtain the average no-load speed and average no-load current.

[0010] Preferably, the load mutation coefficient satisfies the following expression: ; In the formula, Indicates the load mutation coefficient; Indicates the real-time operating current; This represents the average no-load current. This represents the average no-load speed. This indicates the real-time operating speed.

[0011] This invention calculates the load mutation coefficient by combining the product of current deviation and speed deviation. It leverages the physical characteristic that rigid physical contact can simultaneously trigger a sudden drop in speed and a surge in current to amplify the true load change signal. This two-dimensional variable cross-coordination calculation method can effectively distinguish between one-dimensional random noise fluctuations and genuine external load mutations, avoiding the false triggering or slow response problems that are prone to occur with single threshold judgments.

[0012] Preferably, obtaining the corresponding initial meshing speed and initial meshing current includes: The anti-interference threshold is obtained in advance from the system. When the load mutation coefficient is greater than the anti-interference threshold, it is determined that the screw and the workpiece start to mesh and the meshing moment is obtained. At the same time, the current meshing initial speed and meshing initial current are immediately locked and recorded.

[0013] This invention determines the instant of engagement between the screw and the workpiece by comparing the load mutation coefficient with the anti-interference threshold, and simultaneously locks and records the rotational speed and current data at this moment. This method can effectively filter interference from non-substantial contact such as brief rubbing, ensuring the accuracy of engagement state determination. The locked initial data anchors a unified time starting point for subsequent material hardness assessment, eliminating calculation errors caused by mechanical clearance during the no-load stage, and laying an accurate foundation for subsequent dynamic parameter adjustment.

[0014] Preferably, the deep engagement speed and deep engagement current are extracted again, including: Using the moment of engagement as the starting point for timing, the current feed state is maintained, and after a fixed monitoring cycle, the depth engagement speed and depth engagement current at this moment are extracted again from the real-time data stream.

[0015] This invention uses the moment of engagement as a starting point and extracts rotational speed and current data during the deep engagement phase after a fixed monitoring period. This delayed sampling method allows the screw to pass through the initial impact period of contact, eliminating noise interference from high-frequency surface oscillations, and obtaining steady-state operating data after the screw has stably cut into the workpiece substrate. The obtained deep engagement data can accurately reflect the inherent physical resistance of the workpiece material under continuous cutting, providing reliable steady-state data support for subsequent accurate calculation of material hardness.

[0016] Preferably, calculating the material hardness mapping value includes: A fixed monitoring cycle is obtained from the system; the material hardness mapping value satisfies the following expression: ; In the formula, This represents the material hardness mapping value; Indicates the depth of engagement current; Indicates the initial engagement current; Indicates the initial rotational speed at engagement; Indicates the rotational speed at deep engagement; This indicates a fixed monitoring cycle.

[0017] This invention combines the current and speed difference during the initial and deep engagement stages to obtain a material hardness mapping value through normalization calculation. The calculation utilizes the characteristic that materials of different hardness will cause synchronous changes in current and speed of varying magnitudes, thus converting changes in electrical signals into an assessment of the workpiece material hardness. This method eliminates the need for additional material detection equipment; material identification can be achieved solely through the motor's own operating data. This provides a direct basis for subsequent adaptive adjustment of torque and speed, improving the system's adaptability to workpieces of different materials.

[0018] Preferably, obtaining the adaptive target torque and adaptive target speed includes: Extract the system's preset base torque, base speed, and proportional coefficient. , , Based on the material hardness mapping value and the initial meshing current, the system uses its built-in basic proportional correlation logic to match and generate adaptive target torque and adaptive target speed for the final bonding stage.

[0019] Preferably, obtaining real-time output torque includes: The adaptive target speed is used as the feedforward reference command, and the product of the system's preset motor torque coefficient and the real-time operating current is recorded as the real-time output torque. Based on the deviation between the real-time output torque and the adaptive target torque, the PID control algorithm is used to adjust the PWM duty cycle sent to the motor in real time, forming a closed-loop regulation circuit.

[0020] Preferably, the output tightening completion signal includes: The system continuously compares the real-time output torque with the adaptive target torque. When the real-time output torque is greater than or equal to the adaptive target torque for S consecutive samplings, it determines that the screw has reached the optimal tightening state, automatically cuts off the motor power, and outputs a tightening completion signal to the main control module.

[0021] The beneficial effects of this invention are as follows: This invention improves the intelligence and adaptability of electric screwdriver torque control, providing a low-cost, high-precision torque control solution for industrial automated assembly. Compared to traditional control methods, this invention requires no additional hardware sensors; the entire control process can be completed solely through the motor's own electrical control feedback, reducing equipment manufacturing costs and maintenance difficulty. The system can automatically adapt to workpieces of different materials and specifications, eliminating the need for frequent manual parameter switching, and is suitable for industrial assembly scenarios with mixed product lines. Through precise torque control and impact suppression, it effectively protects precision workpieces, extends equipment lifespan, and improves the consistency of fastening quality and production efficiency, contributing to the standardization and automation of industrial assembly processes. Attached Figure Description

[0022] Figure 1 This schematically illustrates a flowchart of an adaptive closed-loop control method for the output torque of an electric screwdriver according to the present invention; Figure 2 The diagram illustrates the timing characteristics of rotational speed and current in this invention. Figure 3 The diagram illustrates a comparison of torque control errors under different connection conditions in this invention. Detailed Implementation

[0023] This invention discloses an adaptive closed-loop control method for the output torque of an electric screwdriver, referring to... Figure 1 This includes steps S1-S4: S1: Synchronously collect real-time operating speed and real-time operating current at a preset frequency.

[0024] It should be noted that in modern high-intensity, multi-variety industrial assembly lines, traditional electric screwdrivers typically rely on external torque sensors or complex preset parameter matrices for control. However, external sensors not only increase the manufacturing cost and size of the equipment, but are also prone to temperature drift and accuracy degradation under long-term high-frequency vibration and mechanical shock, leading to a surge in failure rates. Relying on factory-preset parameters means that the system's perception of changes in operating conditions is rigid and unable to cope with physical variables in the assembly environment caused by material tolerances, ambient temperature, or mechanical wear. Since the decrease in motor speed and the increase in current during motor operation are direct physical mappings of overcoming external mechanical resistance, this invention converts mechanical torque into an electrical signal flow solely through the underlying electronic control feedback of the drive motor itself.

[0025] Specifically, real-time operating speed and real-time operating current are synchronously collected at a preset frequency, including: Start the electric screwdriver to tighten the screws, and synchronously collect the real-time operating speed and real-time operating current of the motor at a preset frequency.

[0026] Thus, the real-time operating speed and real-time operating current of the motor during operation were obtained.

[0027] S2: Obtain the initial rotational speed and initial current of meshing.

[0028] It should be noted that during the no-load phase of the electric screwdriver's startup and high-speed movement towards the workpiece, the motor itself experiences a surge in starting current and electromagnetic and mechanical noise from internal gearbox friction. If, as with traditional control methods, only the transient current at a single moment is extracted as the judgment benchmark, it is easy to misinterpret the motor's structural noise or slight hand tremors as load resistance, causing the system to prematurely trigger deceleration or shutdown mid-air. This invention establishes a fixed-time statistical window to perform time-division calculations on the fluctuation data during the no-load period to obtain a smoothed average. Its physical significance lies in establishing a zero-point benchmark for the motor's current resistance-free operating state. Subsequently, in the initial pre-tightening stage before the screw contacts the workpiece, the system captures transient characteristics in real time. This separation and comparison of historical averages and real-time transient characteristics not only filters out system background noise but also ensures that subsequent algorithms only react to genuine surges in external physical resistance, improving the anti-interference capability and data reliability of the front-end feature extraction.

[0029] Specifically, the real-time operating speed and real-time operating current are integrated and averaged to obtain the average no-load speed and average no-load current, including: Preset length is The statistical window is used to integrate and average the real-time operating speed and real-time operating current within the statistical window to obtain the average no-load speed and average no-load current.

[0030] Thus, the average no-load speed and average no-load current were obtained.

[0031] It should be noted that within the critical milliseconds of the screw's high-speed feed and initial contact with the workpiece surface, the screw threads begin substantial engagement with the workpiece substrate, at which point the physical frictional force undergoes a rapid jump. In conventional open-loop control or single-variable monitoring systems, this initial, slight load change is easily masked by the mechanical background noise of the motor gearbox, brush commutation spikes, or slight hand tremors of the operator. This results in either a slow system response, only beginning to decelerate after the screw has already heavily impacted the workpiece, generating a destructive impact force; or frequent false triggering due to excessively low threshold settings caused by mechanical noise. To overcome this predicament, this invention does not employ a single threshold but instead constructs a two-dimensional variable cross-cooperative model based on current and speed deviations. From a physical perspective, true hard physical contact inevitably triggers both a sudden drop in motor speed and a passive surge in drive current. Multiplying these two transient deviations mathematically amplifies the synchronous physical phenomenon of speed drop and current rise. For single-dimensional random noise fluctuations, due to the lack of coordination from another dimension, the product result remains low. Therefore, this model captures and amplifies this real load surge characteristic, providing data support for the system to quickly reduce the no-load speed and smoothly transition to the bonding stage.

[0032] Preferably, based on the phenomenon that rigid physical contact can cause a sudden drop in speed and a surge in current, a load mutation coefficient is calculated based on the average no-load speed and average no-load current, including: The load mutation coefficient satisfies the following expression: ; In the formula, Indicates the load mutation coefficient; Indicates the real-time operating current; This represents the average no-load current. This represents the average no-load speed. This indicates the real-time operating speed.

[0033] In the formula, Its mathematical prototype originates from the classic formula for mechanical work, and its standard evolutionary form is: In this plan, It represents the increased electromagnetic driving force of the motor in overcoming additional friction. This makes the coefficient quiescent to the normal physical friction interference when the screw is idling, but triggers positive amplification once the screw contacts the workpiece and causes a sudden load change, thus realizing the capture and identification of weak physical load surge characteristics at the algorithm level. From the perspective of two-dimensional deviation coupling and collaborative detection, the deviation cross model constructed by this product operation maintains a balanced low reference benchmark when both current deviation and speed deviation are at low levels during the no-load stage. The deviation cross product term in this product form serves as a dynamic risk indicator. Since a sudden drop in speed is inevitably accompanied by a surge in current, the larger this value is, the more severe the sudden impact of contact occurs, which is consistent with the working principle of load jump when the screw is rigidly engaged.

[0034] Thus, the load mutation coefficient, which characterizes the slight load surge at the moment of workpiece contact, was obtained.

[0035] It should be noted that at the critical point where the screw threads contact the workpiece surface, the physical friction state is complex, accompanied by subtle cutting collisions and sliding. Traditional torque recognition algorithms often rely on setting an absolute current hard threshold. However, when faced with screws of different specifications or complex assembly environments, a fixed hard threshold is either too insensitive, causing the workpiece to be damaged before triggering, or too sensitive, leading to false triggering by motor commutation pulses. The anti-interference threshold introduced in this invention, from a physical mechanism perspective, only confirms mathematically that the screw has undergone substantial physical engagement with the material, rather than a brief frictional interference, when the amplification effect generated by the load mutation coefficient successfully breaks through this anti-interference threshold. More importantly, freezing and recording the initial rotational speed and current at the moment of engagement determination is equivalent to anchoring an accurate initial starting line for the hardness assessment of different workpieces, thereby eliminating the influence of mechanical clearance in the no-load stage on subsequent calculation errors and ensuring the accuracy and robustness of engagement state determination.

[0036] Preferably, determining the instant of meshing and obtaining the corresponding initial meshing speed and initial meshing current based on the load mutation coefficient includes: The anti-interference threshold is obtained in advance from the system. When the load mutation coefficient is greater than the anti-interference threshold, it is determined that the screw and the workpiece start to mesh and the meshing moment is obtained. At the same time, the current meshing initial speed and meshing initial current are immediately locked and recorded.

[0037] It should be noted that, in order to avoid overlap and ambiguity in the control logic, this invention divides the fastening operation into distinct stages based on the relationship between the load mutation coefficient and the anti-interference threshold: First, the operating interval within the preset statistical window after the motor starts is defined as the no-load stage; second, the interval where the load mutation coefficient is calculated in real time and is consistently less than or equal to the anti-interference threshold is defined as the pre-tightening initial stage; third, abandoning the conventional practice of tracking the lagging peak, the critical moment when the load mutation coefficient first exceeds the anti-interference threshold is defined as the engagement instant; finally, using this engagement instant as the timing starting point, abandoning the traditional method of finding the peak return steady-state interval, and after a forced delay with a fixed monitoring period, the deep engagement stage is determined.

[0038] Thus, the initial rotational speed and initial current of meshing were obtained.

[0039] S3: Calculate the material hardness mapping value.

[0040] It should be noted that after the screw breaks through the workpiece surface and enters substantial engagement, it experiences a brief impact period during which the current and rotational speed data are filled with disordered high-frequency oscillations. Directly using the abrupt changes in data at the moment of contact to determine workpiece hardness often results in distortion due to the impact disturbance. Traditional equipment, lacking time-series sampling capabilities, can only passively wait for tightening to complete, missing the opportunity for proactive adjustment. This invention forcibly maintains the feed state and introduces a fixed underlying hardware monitoring cycle. The physical purpose is to allow the screw drill bit to overcome the initial chaotic impact and penetrate deeply and stably into the workpiece matrix. The depth engagement data extracted after this fixed cycle removes surface impact noise, truly reflecting the material's inherent physical resistance and friction gradient under continuous cutting. This time-delay sampling mechanism based on a fixed time window provides core, reliable steady-state data for subsequent material hardness calculations under a unified benchmark.

[0041] Specifically, after a preset fixed monitoring cycle, the deep engagement speed and deep engagement current are extracted again, including: Using the moment of engagement as the starting point for timing, the current feed state is maintained, and after a fixed monitoring cycle, the depth engagement speed and depth engagement current at this moment are extracted again from the real-time data stream.

[0042] Thus, the deep engagement speed and deep engagement current were obtained.

[0043] It should be noted that a hard-connected workpiece will cause a sudden drop in speed and a surge in current, with both of the synchronous changes amplified. In contrast, a soft-connected workpiece will maintain low changes in both speed and current. This invention performs a product operation on the transient difference between the initial and final engagement states and normalizes it with a fixed monitoring period. This amplifies the synchronous changes corresponding to hard materials, avoiding the problem that the ratio of current change to speed change of the same motor under a fixed input voltage is approximately equal to the intrinsic constant of the motor, which is determined by the back electromotive force coefficient and the armature resistance. This enables the function of mapping electrical signals to the hardness of the workpiece material.

[0044] Preferably, based on the synchronous change characteristics of current and rotational speed caused by material hardness, and combined with depth engagement data and initial steady-state data, the material hardness mapping value is calculated, including: Obtain a fixed monitoring cycle from the system.

[0045] The material hardness mapping value satisfies the following expression: ; In the formula, This represents the material hardness mapping value; Indicates the depth of engagement current; Indicates the initial engagement current; Indicates the initial rotational speed at engagement; Indicates the rotational speed at deep engagement; This indicates a fixed monitoring cycle.

[0046] In the formula, Its mathematical prototype originates from the classic average power formula, and its standard evolutionary form is: In this plan, It characterizes the climbing driving force generated by the motor overcoming resistance when it spins deep into the workpiece; This makes the coefficient adaptable to the buffering process of slow extrusion deformation of soft workpieces, with small current and speed deviations and a low numerator product. However, when encountering hard workpieces such as metals, which cause a significant drop in speed and a simultaneous surge in current, the hardness coefficient is positively amplified, exhibiting a synergistic amplification and growth of the two deviations. Thus, at the algorithm level, the transient deviation of the electrical signal is coupled to the material hardness of the workpiece surface. From a classical mechanics perspective, the denominator... This means that as long as the sampling window remains fixed, the final calculated result will be normalized and amortized according to a unified time benchmark, ensuring that the hardness mapping values ​​under different working conditions are comparable; while the molecules contain As a dynamic stiffness index, since the hard material preventing the screw from being screwed in will inevitably cause a surge in motor current and a sharp drop in speed, the larger this value is, the stronger the hardness mapping gain will be in the final evaluation, which is in line with the working principle of electric screwdrivers to adaptively identify the workpiece material and adjust the torque.

[0047] Thus, the material hardness mapping value used to evaluate the transient physical resistance of the workpiece surface was obtained.

[0048] S4: Output tightening complete signal.

[0049] It should be noted that in mixed-assembly industrial environments, the same electric screwdriver might be tightening a hard metal chassis one second and securing a fragile plastic casing the next. If traditional manual parameter presets are used, the frequent changes in workpiece materials can easily lead to problems: excessive torque causing cracking or stripping of plastic parts, or insufficient torque resulting in loosening of metal parts. This invention solves the problem of parameter fixation based on pre-calculated material hardness mapping values ​​with physical differentiation. From a control theory perspective, this gives the system adaptive adjustment capabilities. When the system senses a low material hardness mapping value, it automatically matches a lower target torque and target speed to achieve gentle contact and pressure retention; conversely, when facing a higher material hardness mapping value, it matches a higher torque to ensure strong locking. This basic proportional correlation logic enables the control parameters to adaptively flow with the physical properties of the workpiece, solving the technical problem of inconsistent tightening quality under complex working conditions.

[0050] Specifically, based on the material hardness mapping value and the initial engagement current, adaptive target torque and adaptive target speed are obtained, including: Extract the system's preset base torque, base speed, and proportional coefficient. , , Based on the material hardness mapping value and the initial meshing current, the system uses its built-in basic proportional correlation logic to match and generate adaptive target torque and adaptive target speed for the final bonding stage.

[0051] It should be noted that the adaptive target torque satisfies the formula The adaptive target rotational speed satisfies the formula. In the formula, and This indicates the base torque and base speed. The preset proportional adjustment coefficient for the system. This characterizes the dynamic rigidity compensation term, meaning that the higher the physical hardness of the workpiece, the better. The larger the torque, the stronger the additional torque compensation applied by the system to overcome the extremely high static friction of hard materials; at the same time, by increasing the rotational speed... To improve fastening efficiency.

[0052] At this point, the adaptive target torque and adaptive target speed have been obtained.

[0053] It should be noted that in the final bonding stage before the tightening operation is completed, the motor's driving force is entirely converted into axial torque to clamp the workpiece. Traditional open-loop control systems often operate at full speed until the motor stalls during this stage. This not only generates transient mechanical damage but also shortens the lifespan of the motor and reduction gears. This invention constructs a closed-loop regulation circuit without external sensors by establishing a dual-track mechanism of target speed distribution and real-time current calculation. Physically, this means that as the motor approaches the target torque, it no longer blindly applies force but actively decelerates and brakes based on the dynamic deviation between the calculated real-time output torque and the target value. This closed-loop correction not only uses the motor itself as a torque sensing basis but also reduces the mechanical impact process at the final tightening moment, ensuring that the final torque applied to each screw smoothly and accurately approaches the preset value, thus protecting the precision workpiece.

[0054] Preferably, using the adaptive target speed as a feedforward reference to obtain the real-time output torque includes: The adaptive target speed is used as the feedforward reference command, and the product of the system's preset motor torque coefficient and the real-time operating current is recorded as the real-time output torque. Based on the deviation between the real-time output torque and the adaptive target torque, the PID control algorithm is used to adjust the PWM duty cycle sent to the motor in real time, forming a closed-loop regulation circuit.

[0055] It should be noted that in the final bonding stage before the tightening operation is completed, the motor's driving force is entirely converted into axial torque to press the workpiece. Traditional open-loop control systems often operate at full speed until the motor stalls during this stage, which not only generates transient mechanical damage but also shortens the lifespan of the motor and reduction gears. This invention constructs a sensorless composite closed-loop regulation circuit by establishing a dual-track mechanism of target speed transmission and real-time current calculation. Specifically, to avoid conflicts caused by the dual control of speed and torque in the underlying electronic control system's PWM duty cycle, this invention uses the adaptive target speed as a feedforward reference command, providing a basic feed drive benchmark for the system, while entrusting the actual dynamic control to the torque closed loop: based on the dynamic deviation between the real-time output torque and the target value, a PID algorithm is used to fine-tune and correct the PWM duty cycle sent to the motor in real time. This correction mechanism not only uses the motor itself as a torque sensing basis but also reduces the mechanical impact overshoot at the final locking moment, thus protecting the precision workpiece.

[0056] At this point, the real-time output torque was obtained.

[0057] It should be noted that in the decision-making stage of determining whether tightening is finally completed, even minor fluctuations in the power grid, brush commutation spikes, or momentary jamming of the gear set in an industrial setting can cause the real-time calculated torque data to spike abnormally within a millisecond, occasionally reaching or even exceeding the target torque. If the system adopts a single-step decision logic, it will inevitably lead to false tightening. This invention introduces a fault-tolerant filtering mechanism with continuous multiple sampling in the target comparison stage. This multi-frame continuous confirmation logic ensures that the motor is indeed in a stable pressure-holding stagnation or stable high resistance state, rather than being affected by transient noise. Only after the fact that the torque has met the target has been continuously verified over time will the system cut off the power source and output a tightening completion signal. This mechanism improves the anti-interference robustness of the entire closed-loop control system, avoids the problem of false shutdowns, and provides a tightening completion signal for automated production lines.

[0058] Preferably, based on the real-time output torque and the adaptive target torque, a tightening completion signal is output, including: The system continuously compares the real-time output torque with the adaptive target torque. When the real-time output torque is greater than or equal to the adaptive target torque for S consecutive samplings, it determines that the screw has reached the optimal tightening state, automatically cuts off the motor power, and outputs a tightening completion signal to the main control module.

[0059] It should be noted that, Figure 2 The graph shows the time-series characteristics of speed and current. The horizontal axis represents time, covering the complete tightening cycle from no-load feed to deep engagement. The upper vertical axis represents the real-time operating speed, and the lower vertical axis represents the real-time operating current. The upper curve represents the real-time operating speed change curve, and the lower curve represents the real-time operating current change curve. The dashed lines in the graph divide the no-load stage, the initial pre-tightening stage, and the deep engagement stage. The vertical lines mark the moment of engagement, and the bottom marks the fixed monitoring cycle.

[0060] It should be noted that, Figure 3 A comparison chart of torque control errors under different connection conditions shows the difference in torque control accuracy between the manually preset fixed parameter method and the method of this invention under different material connection conditions. The horizontal axis represents the connection condition type, including metal rigid connection and wood-plastic flexible connection; the vertical axis represents the torque control error; the diagonally filled bars represent the manually preset fixed parameter method, and the dark-filled bars represent the method of this invention. Under the metal rigid connection condition, the torque control error of the manually preset method is 160, while that of the method of this invention is 125; under the wood-plastic flexible connection condition, the torque control error of the manually preset method is 140, while that of the method of this invention is 108.

[0061] It should be noted that, in order to enable those skilled in the art to have a clearer understanding of the present invention, some contents of the present invention will now be described as follows: Basic torque: range 0.2 to 1.0 N·m, and 0.5 N·m is used in this embodiment. This value is the minimum tightening torque lower limit for most soft workpieces such as plastics and rubber, which can ensure that soft workpieces will not be loose due to insufficient tightening. If the value is too small, it will result in insufficient tightening force on the soft material, and if it is too large, it will easily cause the soft material to crack.

[0062] Basic speed: range 50~200rpm, 100rpm is used in this embodiment. This value takes into account both tightening efficiency and impact control, which can avoid damage to the screw head caused by excessive speed, while ensuring that the no-load feed speed meets the production line cycle requirements. If the value is too small, it will reduce production efficiency, and if it is too large, it will increase the impact force at the moment of meshing.

[0063] The statistical window length M is in milliseconds (ms), ranging from 50 to 200 ms; in this embodiment, it is 100 ms. This value covers the time period after the electric screwdriver motor starts, after it is freed from the initial surge current, and enters steady-state idling. It is specifically used to extract and establish a pure no-load zero-point reference. If the value is too small, the sample size for data integration will be insufficient, and it will be unable to effectively filter the mechanical friction noise and electromagnetic background transient fluctuations during motor startup, resulting in a distorted zero-point reference. If the value is too large, the statistical time may be too long, causing the window to extend beyond the pure no-load period to the stage where the screw contacts the workpiece, thus mixing the actual physical friction load into the no-load reference, causing the subsequent amplification model to fail.

[0064] Hardness proportional coefficient :scope N·m·ms / (A·rpm), in this embodiment, is taken as This value matches the magnitude of the material hardness mapping value in this solution. When the material hardness mapping value is 30000, it can provide a torque increment of 0.3, so that the total torque reaches the required 0.8~1.0 N·m for the metal workpiece. If the value is too large, it will cause the hard material workpiece to be too tight and strip the teeth. If it is too small, the hard material workpiece will not have enough clamping force.

[0065] Initial current compensation coefficient The physical dimension is N·m / A, ranging from 0.01 to 0.05 N·m / A, and 0.02 N·m / A is used in this embodiment. This coefficient is used to compensate for the torque of the basic frictional force in the initial stage of engagement. When the initial engagement current E is 2A, it can provide a basic compensation of 0.04 N·m. If the value is too large, it will cause the initial contact force to overload; if it is too small, it will be unable to overcome the feed resistance caused by the initial mechanical clearance.

[0066] Rotational speed hardness coefficient The range is 0.005~0.02ms / A, and in this embodiment, 0.01ms / A is used. This value can achieve adaptive adjustment of high speed for hard materials and low speed for soft materials. When the material hardness mapping value is 30000, the speed is increased to 400 to improve efficiency. When the material hardness mapping value is 500, the speed is maintained at 105 to avoid impact. If the value is too large, it will cause the hard material to generate flying chips due to excessive speed. If it is too small, the fastening efficiency of the hard material will be low.

[0067] Anti-interference threshold: range 1~10A·rpm, 5A·rpm is used in this embodiment. This value can filter interference signals generated by motor commutation spikes and slight hand tremors of the operator, while ensuring that the judgment is triggered within 1ms when the screw contacts the workpiece; if the value is too small, it will frequently trigger false triggers and cause premature deceleration, while if the value is too large, it will delay engagement recognition and increase the impact damage force.

[0068] Fixed monitoring cycle: range 50~500ms, 100ms is used in this embodiment. This value corresponds to the time it takes for the screw to be screwed into the workpiece 2~3 times, which can eliminate the high-frequency vibration noise generated by the initial collision impact and obtain stable continuous cutting data; if the value is too short, the data will contain a lot of impact noise, resulting in distortion of hardness calculation; if the value is too long, it will prolong the adaptive adjustment time and reduce production efficiency.

[0069] The number of consecutive sampling times S is 2 to 5 times, and 3 times is used in this embodiment. This value can filter out transient noise in a single frame, while ensuring that the shutdown response delay does not exceed 3ms and the system sampling period is 1ms. If the value is too small, it is easy to be triggered by transient noise, resulting in false tightening. If the value is too large, it will increase the shutdown overshoot and affect the torque control accuracy.

[0070] The system's preset motor torque coefficient represents the effective mechanical torque that the motor can output under a unit ampere current. This solution already includes the equivalent conversion of gearbox reduction ratio and transmission efficiency. The motor torque coefficient ranges from 0.05 to 0.5 N·m / A, and this embodiment uses 0.15 N·m / A. This value matches the electromagnetic characteristics and mechanical transmission ratio of the stator winding of common industrial-grade brushless electric screwdrivers, enabling the underlying current signal to be accurately and linearly converted into physical output torque. If the value is too small, the calculated real-time output torque will be too low, and the system will incorrectly continue to output a large current in order to achieve the target torque, which can easily lead to motor overload and burnout or excessive compression and damage to the workpiece. If the value is too large, the calculated torque will be artificially high, and the system will prematurely trigger the shutdown judgment before reaching the actual tightening force, resulting in loose screws.

[0071] The control method of the present invention can be implemented by an embedded microprocessor, without the need for additional hardware sensors. The entire control process can be completed solely through motor control feedback, which has the advantages of low hardware cost and high integration.

[0072] This completes the adaptive closed-loop control method for the output torque of the electric screwdriver.

[0073] While various embodiments of the invention have been shown and described in this specification, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention.

Claims

1. An adaptive closed-loop control method for the output torque of an electric screwdriver, characterized in that, include: S1: Synchronously collect real-time operating speed and real-time operating current at a preset frequency; S2: Integrate and average the real-time operating speed and real-time operating current to obtain the average no-load speed and average no-load current; based on the phenomenon that rigid physical contact will cause a sudden drop in speed and a surge in current, calculate the load mutation coefficient based on the average no-load speed and average no-load current; based on the load mutation coefficient, determine the moment of engagement and obtain the corresponding initial engagement speed and initial engagement current. S3: After a preset fixed monitoring cycle, the deep meshing speed and deep meshing current are extracted again; based on the synchronous change characteristics of current and speed caused by material hardness, the material hardness mapping value is calculated by combining the deep meshing data and the initial steady-state data. S4: Based on the material hardness mapping value and the initial meshing current, obtain the adaptive target torque and adaptive target speed; using the adaptive target speed as a feedforward reference, record the product of the system's preset motor torque coefficient and the real-time operating current as the real-time output torque to obtain the real-time output torque; Based on the real-time output torque and the adaptive target torque, a tightening completion signal is output.

2. The adaptive closed-loop control method for the output torque of an electric screwdriver according to claim 1, characterized in that, The method of synchronously acquiring real-time operating speed and real-time operating current at a preset frequency includes: Start the electric screwdriver to tighten the screws, and synchronously collect the real-time operating speed and real-time operating current of the motor at a preset frequency.

3. The adaptive closed-loop control method for the output torque of an electric screwdriver according to claim 1, characterized in that, The acquisition of the average no-load speed and average no-load current includes: Preset length is The statistical window is used to integrate and average the real-time operating speed and real-time operating current within the statistical window to obtain the average no-load speed and average no-load current.

4. The adaptive closed-loop control method for the output torque of an electric screwdriver according to claim 1, characterized in that, The load mutation coefficient satisfies the following expression: ； In the formula, Indicates the load mutation coefficient; Indicates the real-time operating current; This represents the average no-load current. This represents the average no-load speed. This indicates the real-time operating speed.

5. The adaptive closed-loop control method for the output torque of an electric screwdriver according to claim 1, characterized in that, Obtaining the corresponding initial meshing speed and initial meshing current includes: The anti-interference threshold is obtained in advance from the system. When the load mutation coefficient is greater than the anti-interference threshold, it is determined that the screw and the workpiece start to mesh and the meshing moment is obtained. At the same time, the current meshing initial speed and meshing initial current are immediately locked and recorded.

6. The adaptive closed-loop control method for the output torque of an electric screwdriver according to claim 1, characterized in that, The process of re-extracting the depth engagement speed and depth engagement current includes: Using the moment of engagement as the starting point for timing, the current feed state is maintained, and after a fixed monitoring cycle, the depth engagement speed and depth engagement current at this moment are extracted again from the real-time data stream.

7. The adaptive closed-loop control method for the output torque of an electric screwdriver according to claim 1, characterized in that, The calculation of the material hardness mapping value includes: A fixed monitoring cycle is obtained from the system; the material hardness mapping value satisfies the following expression: ； In the formula, This represents the material hardness mapping value; Indicates the depth of engagement current; Indicates the initial engagement current; Indicates the initial rotational speed at engagement; Indicates the rotational speed at deep engagement; This indicates a fixed monitoring cycle.

8. The adaptive closed-loop control method for the output torque of an electric screwdriver according to claim 1, characterized in that, The process of obtaining the adaptive target torque and adaptive target speed includes: Extract the system's preset base torque, base speed, and proportional coefficient. , , Based on the material hardness mapping value and the initial meshing current, the system uses its built-in basic proportional correlation logic to match and generate adaptive target torque and adaptive target speed for the final bonding stage.

9. The adaptive closed-loop control method for the output torque of an electric screwdriver according to claim 1, characterized in that, The process of obtaining real-time output torque includes: The adaptive target speed is used as the feedforward reference command, and the product of the system's preset motor torque coefficient and the real-time operating current is recorded as the real-time output torque. Based on the deviation between the real-time output torque and the adaptive target torque, the PID control algorithm is used to adjust the PWM duty cycle sent to the motor in real time, forming a closed-loop regulation circuit.

10. The adaptive closed-loop control method for the output torque of an electric screwdriver according to claim 1, characterized in that, The output tightening completion signal includes: The system continuously compares the real-time output torque with the adaptive target torque. When the real-time output torque is greater than or equal to the adaptive target torque for S consecutive samplings, it determines that the screw has reached the optimal tightening state, automatically cuts off the motor power, and outputs a tightening completion signal to the main control module.

Images