Self-adaptive compensation method for stiffness change rate of single-shaft driving system

Abstract

The invention discloses a self-adaptive compensation device and method for the stiffness change rate of a single-axis driving system based on current-position response, and belongs to the field of precise driving control. The method aims at solving the problems that an existing compensation technology is poor in adaptability to thermal deformation and load sudden change, rigidity identification delay is large, and phase error suppression is insufficient. The device mainly comprises an FPGA, an ARM controller, a servo driver, a servo motor, an absolute value encoder, a current acquisition module and the like. The core logic is that an encoder position signal and a current signal are collected and then connected to the FPGA, the stiffness change rate is calculated by the FPGA, the compensation amount is generated by combining a PI-FF algorithm, and closed-loop control is achieved through cooperative processing of the ARM and the FPGA. According to the method, the rigidity change under thermal deformation and load sudden change can be identified, compared with the prior art, the phase error and disturbance recovery time are both reduced by more than or equal to 60%, the hardware cost is reduced by 30%-50%, and the method is suitable for high-end precision equipment such as a semiconductor photoetching machine.

Description

Technical Field

[0001] This invention belongs to the field of electromechanical control technology, specifically relating to a method and device for online measurement and adaptive compensation of stiffness change rate of a single-axis drive system based on current-position response. Background Technology

[0002] In the field of precision motion control, the phase synchronization accuracy of a multi-axis system directly determines the dynamic performance of the equipment. Traditional phase compensation methods mainly rely on two technical approaches: The first approach involves deploying external sensors such as strain gauges and laser interferometers in the mechanical transmission chain to monitor deformation in real time. While this can achieve static error compensation, it faces three major bottlenecks: limited installation space, significant temperature drift interference, and signal transmission delays greater than 5 milliseconds. These problems cause the system to lag in response to time-varying factors such as bearing wear and thermal deformation. The second approach is a sensorless solution based on motor current and encoder feedback. Although this reduces hardware complexity, it can only estimate the average static stiffness and cannot capture the key parameter characterizing dynamic characteristics—the rate of change of stiffness dK / dt. Under sudden load changes or mechanical resonance conditions, this deficiency can lead to phase overshoot exceeding 30% or response delays greater than 15 milliseconds.

[0003] Existing stiffness compensation strategies suffer from three core drawbacks: First, static stiffness mapping tables require pre-calibration and cannot adapt to stiffness drift during operation. Taking semiconductor packaging equipment as an example, thermal deformation caused by temperature rise after two hours of continuous operation deteriorates the phase synchronization error between multiple axes by more than 50%. Second, frequency domain analysis methods, with computational delays exceeding 50 milliseconds, struggle to suppress trajectory deviations caused by sudden stiffness changes. Finally, adding sensors to the load side compromises the compactness of single-axis systems, increasing wiring costs and failure risks. These shortcomings are particularly pronounced in high-speed, precision equipment, severely hindering performance improvements in high-end equipment. Summary of the Invention

[0004] The purpose of this invention is to solve the problems of phase compensation lag and insufficient response to sudden disturbances caused by the inability of existing single-axis drive systems to sense dynamic changes in stiffness in real time, and to provide a method and device for online measurement and adaptive compensation of stiffness change rate based on current-position response.

[0005] To solve the above problems, the present invention is achieved through the following technical solution:

[0006] A method for online measurement and adaptive compensation of stiffness change rate of a single-axis drive system, characterized by the following steps:

[0007] Step 1: Real-time acquisition of motor current signal With encoder position signal The motor-load displacement difference is calculated using a virtual load observer. : in: 1. = The output torque of the motor ( Torque constant) 2. =Nominal stiffness of the drive chain 3. = Rotor inertia;

[0008] Step 2: Based on displacement difference With torque Construct a stiffness change rate observer to calculate the dynamic stiffness change rate in real time. : Hardware acceleration implementation: The five-point central difference method is embedded in the FPGA to calculate the differential term, with a calculation period of ≤100μs.

[0009] Step 3: Design a feedforward-feedback composite compensator and generate stiffness change rate compensation. :

[0010] Step 4: Adjust the compensation amount The position command is superimposed and output to the servo driver. Limiting function definition: in This represents the original position command, the angle input by the servo system. This is the dynamic phase compensation amount calculated in real time. This is the maximum allowable amplitude for compensation.

[0011] Step 5: When a sudden change in stiffness is detected (| |> ), Activate adaptive gain adjustment: in:

[0012] Compared with existing technologies, this invention achieves the following without additional sensors: improved phase synchronization accuracy, reduced stiffness change rate response delay, and reduced hardware costs through an innovative stiffness change rate observer and feedforward-feedback composite compensation architecture. Attached Figure Description

[0013] The invention will now be further described with reference to the accompanying drawings.

[0014] Figure 1 shows a comparison of thermal errors before and after compensation in the thermal error compensation experiment.

[0015] Figure 2 shows a comparison of the positioning accuracy and repeatability before and after compensation using a laser interferometer.

[0016] Figure 3 For simulated lead screw testing. Detailed Implementation

[0017] The present invention will be further illustrated below with reference to specific embodiments.

[0018] The system hardware includes: Mitsubishi MELSERVO-J5 servo drive, FPGA controller, ARM Cortex-M7 controller, 23-bit absolute encoder, current signal isolation amplifier module, and encoder signal conversion module.

[0019] The specific connection details are as follows: The servo driver is connected to the servo motor via the power terminal. The motor's built-in 23-bit absolute encoder is connected to the driver's CN2 port via a shielded twisted-pair cable. The driver's current sampling signal is led out from the CN3 analog output terminal, converted into an LVDS differential signal by a current isolation amplifier, and the LVDS current and position signals are respectively connected to the FPGA's dedicated differential I / O pairs. The FPGA is connected to the ARM controller via a 16-bit FSMC bus. The ARM's EtherCAT physical layer interface is connected to the IN port of the EtherCAT slave module (Beckhoff EK1100) via a twisted-pair shielded cable, and the OUT port is connected to the driver's CN4 EtherCAT interface.

[0020] The implementation process consists of three core stages: parameter calibration, algorithm compensation, and testing and verification.

[0021] The parameter calibration steps are as follows: 1. Nominal stiffness measurement: Lock the worktable, apply a 5 N·m step torque to the motor, and read the average steady-state deflection angle of the encoder. =0.025 rad; Calculate the stiffness value = / =5 / 0.025=200N·m / rad 2. Rotor inertia measurement: Under no-load conditions, a trapezoidal velocity curve (acceleration 1000 rad / s²) was executed, and the peak torque was collected. peak = 1.2 N·m, inertia calculated = peak / α =1.2 / 1000=0.0012kg·m2

[0022] Real-time execution flow of the compensation algorithm (loop every 100μs):

[0023] Step 1: Perform current and position sampling, reading the current every 100μs. and encoder position The current and position signals need to be filtered by a first-order low-pass filter to prevent high-frequency noise from causing the differential operation to diverge, and to perform anti-interference processing.

[0024] Step 2: Calculate the displacement difference using a virtual observer:

[0025] Step 3: Calculate the rate of change of stiffness based on the five-point center difference method

[0026] Generate compensation quantities, perform anti-saturation processing on the integral term in the FPGA, adaptive gain, and synthesize compensation quantities;

[0027] Limiting output: Final instruction = Original position instruction + (Limited to ±0.05 × speed range)

[0028] Test and Verification: Shaft Thermal Error Compensation Experiment To verify the correctness and effectiveness of the compensation scheme and module, as well as the accuracy of the model, a thermal error compensation experiment was designed. The thermal error compensation model was embedded into the compensation module, and the thermal error compensation function was enabled after setting the parameters. The results before and after compensation are shown in Figure 1(a) and Figure 1(b), respectively. The results show that the thermal error range has been reduced to within 10 μm after compensation, indicating that the compensation is effective and reduces the error to improve accuracy.

[0029] Under the same load, the same motion trajectory, and the same environment, the positioning accuracy before and after compensation was tested using a laser interferometer. The laser interferometer data shows that the positioning accuracy and repeatability of the system before and after compensation for single-axis drive are shown in Figure 2(a) and Figure 2(b).

[0030] Simulation testing: Simulating the temperature rise of the lead screw from 25℃ to 80℃ under infrared heating; calculating the axial elongation and torsional stiffness reduction caused by thermal expansion of the lead screw; simulating the thermal deformation phase error of the single-axis drive system, with key analysis dimensions including temperature-stiffness curves, phase error comparison, and load change tests, as follows. Figure 3 As shown.

Claims

1. A device for online measurement and adaptive compensation of stiffness change rate of a single-axis drive system, characterized in that, The system includes an FPGA controller, an ARM Cortex-M7 controller, a Mitsubishi MELSERVO-J5 servo driver, a permanent magnet synchronous servo motor, a ball screw, a 23-bit absolute encoder, a current signal isolation and amplification module, an encoder signal conversion module, a high-speed ADC acquisition module, an infrared temperature measurement module, and a magnetic powder brake. The 23-bit absolute encoder is mounted at the end of the ball screw, and its output signal is converted to an LVDS signal by the encoder signal conversion module before being transmitted to the FPGA controller. The current signal isolation and amplification module is connected to the current sensor of the servo motor, and its output signal is converted to a digital signal by the high-speed ADC acquisition module before being transmitted to the FPGA controller. The infrared temperature measurement module collects the temperature of the ball screw, which is then transmitted to the FPGA controller via the high-speed ADC acquisition module. The FPGA controller communicates bidirectionally with the ARM Cortex-M7 controller, outputting control commands to the Mitsubishi MELSERVO-J5 servo driver. The servo driver drives the permanent magnet synchronous servo motor to rotate the ball screw, which is connected to the magnetic powder brake to form the load. The servo driver provides real-time feedback of its operating status to the FPGA controller, achieving closed-loop control of "detection-control-drive-load".

2. The apparatus according to claim 1, characterized in that, The FPGA controller uses a Xilinx Artix-7XC7A50T chip to perform high-speed real-time tasks: calculating the motor angular acceleration through encoder signals, realizing online measurement of stiffness change rate based on current-position response, generating adaptive compensation and outputting it to the servo driver; the ARM Cortex-M7 controller is used to perform low-speed logic tasks: handling host computer communication, parameter configuration, data storage and fault alarms, and working in conjunction with the FPGA controller to realize control strategy linkage.

3. The apparatus according to claim 1, characterized in that, The encoder signal conversion module is a TTL to LVDS conversion module, which converts the TTL signal output by the 23-bit absolute encoder into a differential LVDS signal, reducing signal transmission interference and ensuring the accuracy of angular position acquisition.

4. The apparatus according to claim 1, characterized in that, The online measurement of the stiffness change rate adopts the five-point center difference method. The angular position signal of the motor is acquired in real time through the FPGA controller. After calculating the angular acceleration, the real-time torsional stiffness and stiffness change rate of the screw are obtained by combining the q-axis current acquired by the current signal isolation amplification module.

5. The apparatus according to claim 1, characterized in that, The logic for generating the adaptive compensation amount is as follows: a composite compensation model is constructed based on the stiffness change rate, phase error, and error integral, a stiffness change rate feedforward coefficient β is introduced, and the adaptive compensation amount is calculated through the proportional-integral-feedforward (PIF) algorithm. The compensation amount is output to the servo driver after being limited, thereby achieving phase error suppression.

6. The apparatus according to claim 1, characterized in that, The infrared temperature measurement module is a non-contact infrared sensor that collects ball screw temperature data in real time to construct a "temperature-stiffness" attenuation model, providing a temperature compensation basis for online measurement of stiffness change rate.

7. The apparatus according to claim 1, characterized in that, The magnetic powder brake is used to simulate sudden load changes and can achieve step load output from 50 N·m to 200 N·m, verifying the adaptive compensation effect of the device under dynamic load disturbance.

8. The apparatus according to claim 1, characterized in that, It also includes an emergency stop button and a safety relay. The emergency stop button and the safety relay are connected in series to the servo power supply circuit. When triggered, they can directly cut off the main power supply and control power supply of the Mitsubishi MELSERVO-J5 servo drive to ensure system safety.

Images