Dead zone optimization method and device for servo motor, equipment and storage medium

Abstract

The invention provides a dead zone optimization method and device for a servo motor, equipment and a storage medium, and the method comprises the steps: carrying out the multi-frequency-domain excitation of the servo motor, and obtaining a dead zone threshold value and a dead zone slope; detecting the position error, the speed error, the rotating speed and the acceleration of the servo motor at the current position; performing disturbance compensation amount calculation on the position error and the speed error according to the dead zone threshold and the dead zone slope to obtain a compensation gain; determining a compensation coefficient according to the compensation gain, the rotating speed and the acceleration; and according to the compensation coefficient, the original control signal of the servo motor is superposed to obtain a correction control signal, and the servo motor is controlled based on the correction control signal.

Description

Technical Field

[0001] The present application relates to the field of motor control technology, and in particular to a dead zone optimization method, device, equipment and storage medium for a servo motor. Background Art

[0002] Currently, servo motors commonly experience dead zones during operation. This nonlinearity means that when the control signal varies within a certain range, the motor output barely changes. This severely impacts the system's control accuracy and dynamic response. Existing compensation algorithms often ignore the impact of the motor's current operating state (such as speed and acceleration) on dead zone compensation. This hinders servo system performance in high-precision, high-dynamic response applications and prevents true adaptive optimization. Summary of the Invention

[0003] The present application provides a dead zone optimization method, apparatus, device and storage medium for a servo motor, which are used to adaptively optimize the dead zone effect of the servo motor.

[0004] In a first aspect, an embodiment of the present application provides a dead zone optimization method for a servo motor, the method comprising: Perform multi-frequency domain excitation on the servo motor to obtain the dead zone threshold and dead zone slope; Detecting the position error, velocity error, rotation speed and acceleration of the servo motor at the current position; Calculating a disturbance compensation amount for the position error and the speed error according to the dead zone threshold and the dead zone slope to obtain a compensation gain; determining a compensation coefficient according to the compensation gain, the rotational speed, and the acceleration; The original control signal of the servo motor is superimposed according to the compensation coefficient to obtain a modified control signal, and the servo motor is controlled based on the modified control signal.

[0005] In a second aspect, an embodiment of the present application provides a dead zone optimization device for a servo motor, wherein the dead zone optimization device for a servo motor is configured to execute the dead zone optimization method for a servo motor as described in any one of the embodiments of the present application, and the dead zone optimization device for a servo motor includes: The excitation detection module is used to perform multi-frequency domain excitation on the servo motor to obtain the dead zone threshold and dead zone slope; A parameter acquisition module is used to detect the position error, speed error, speed and acceleration of the servo motor at the current position; a compensation calculation module, configured to calculate a disturbance compensation amount for the position error and the speed error according to the dead zone threshold and the dead zone slope to obtain a compensation gain; a coefficient determination module, configured to determine a compensation coefficient according to the compensation gain, the rotational speed, and the acceleration; The signal correction module is used to superimpose the original control signal of the servo motor according to the compensation coefficient to obtain a corrected control signal, and control the servo motor based on the corrected control signal.

[0006] In a third aspect, an embodiment of the present application provides an electronic device, the electronic device including a memory and a processor; The memory is used to store computer programs; The processor is configured to execute the computer program and implement the dead zone optimization method for a servo motor as described in any one of the embodiments of the present application when executing the computer program.

[0007] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the processor implements the dead zone optimization method for a servo motor as described in any one of the embodiments of the present application.

[0008] The embodiment of the present application provides a dead zone optimization method for a servo motor, the method comprising: performing multi-frequency domain excitation on the servo motor to obtain a dead zone threshold and a dead zone slope; detecting the position error, velocity error, speed and acceleration of the servo motor at the current position; calculating the disturbance compensation amount for the position error and velocity error according to the dead zone threshold and the dead zone slope to obtain a compensation gain; determining a compensation coefficient according to the compensation gain, speed and acceleration; superimposing the original control signal of the servo motor according to the compensation coefficient to obtain a corrected control signal, and controlling the servo motor based on the corrected control signal. In the above method, the dead zone threshold and slope parameters are accurately obtained through multi-frequency domain excitation, the four key state quantities of position error, velocity error, speed and acceleration are detected in real time, the compensation gain is determined by using a disturbance compensation amount calculation method weighted by dead zone characteristics, the compensation coefficient is adaptively adjusted according to the operating state, and the original control signal is smoothly corrected by signal superposition, thereby achieving accurate matching of the dead zone compensation with the actual operating conditions of the motor, solving the problems of low compensation accuracy and poor adaptability, and improving the control accuracy and dynamic response performance of the servo motor in the full operating range. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1A schematic flow chart of a dead zone optimization method for a servo motor provided in an embodiment of the present application; Figure 2 A schematic block diagram of a dead zone optimization device for a servo motor provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0012] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, combined, or partially merged, so the actual execution order may vary depending on the actual situation.

[0013] It should also be understood that the terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit the present application. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0014] It should be further understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0015] See also Figure 1 , Figure 1 FIG. 1 is a schematic flow chart of a dead zone optimization method for a servo motor provided in an embodiment of the present application. Figure 1 As shown, the specific steps of the dead zone optimization method for the servo motor include: S101-S105.

[0016] S101 , performing multi-frequency domain excitation on the servo motor to obtain a dead zone threshold and a dead zone slope.

[0017] Exemplarily, the frequency components of the composite sinusoidal excitation signal cover a multi-frequency range of 0.1Hz to 100Hz, and the signal amplitude gradually increases from a tiny signal to 10% of the rated range. The designed composite sinusoidal excitation signal is input to the control terminal of the servo motor, and the data acquisition module is simultaneously activated to record the position response data and velocity response data of the servo motor at a sampling frequency of 1kHz. The collected position response data and velocity response data are processed by sliding slices, with each slice containing 1024 sampling points and an overlap of 50% between adjacent slices. A fast Fourier transform is performed on each data slice to extract amplitude-frequency characteristic data and phase-frequency characteristic data. By analyzing the amplitude-frequency characteristic data, the amplitude attenuation characteristics at different frequencies are identified; by analyzing the phase-frequency characteristic data, the phase lag characteristics are identified. Based on the principle of nonlinear characteristic analysis, parameter identification is performed on the amplitude-frequency characteristic data and phase-frequency characteristic data. Specifically, a critical point where the output response begins to appear is searched within the low-amplitude input range. The input amplitude corresponding to this critical point is the dead zone threshold. The dead zone thresholds include positive and negative dead zone thresholds, corresponding to the dead zone boundaries for forward and reverse motion, respectively. By calculating the slopes of the input-output relationship curve inside and outside the dead zone, we obtain the slope inside and outside the dead zone. The slope inside the dead zone reflects the weak response characteristics within the dead zone, while the slope outside the dead zone reflects the linear response characteristics in the normal operating range.

[0018] S102 , detecting the position error, speed error, rotation speed, and acceleration of the servo motor at the current position.

[0019] For example, the position error is calculated by reading the current position feedback value from the servo motor encoder and calculating the difference between it and the position command value. The position error is calculated as follows: Position error = Position command value - Current position feedback value. The actual speed value is calculated by taking the difference between the position feedback values ​​of two consecutive sampling periods and dividing it by the sampling interval. The speed error is calculated by taking the difference between the speed command value and the actual speed value. The speed error is calculated as follows: Speed ​​error = Speed ​​command value - Actual speed value. The rotational speed is obtained by reading the encoder pulse frequency and converting it. The specific conversion method is: Speed ​​= (Encoder pulse frequency × 60) / (Encoder resolution × Reduction ratio). Acceleration is obtained by taking the second-order difference of the speed values ​​of three consecutive sampling periods. The calculation formula is: Acceleration = (Speed ​​value [t] - 2 × Speed ​​value [t-1] + Speed ​​value [t-2]) / (Sampling period squared), where t is the time. To improve detection accuracy, the position error, velocity error, rotational speed, and acceleration are each subjected to a sliding average filter, with a filter window length set to 5 sampling periods.

[0020] S103 , calculating disturbance compensation amounts for the position error and the speed error according to the dead zone threshold and the dead zone slope to obtain compensation gains.

[0021] Exemplarily, a first weight coefficient for position error is determined based on the dead zone threshold, calculated as follows: first weight coefficient = 1 / (1 + |dead zone threshold|). This weight coefficient reflects the extent of the dead zone's impact on position control accuracy. A larger dead zone threshold indicates a smaller weight coefficient, indicating a need for stronger compensation. A second weight coefficient for velocity error is determined based on the dead zone slope, calculated as follows: second weight coefficient = slope outside dead zone / slope inside dead zone. This weight coefficient reflects the difference in response characteristics inside and outside the dead zone. The position error is multiplied by the first weight coefficient to obtain a position compensation component: position compensation component = position error × first weight coefficient. The velocity error is multiplied by the second weight coefficient to obtain a velocity compensation component: velocity compensation component = velocity error × second weight coefficient. The position compensation component and the velocity compensation component are summed to obtain an initial compensation gain. To prevent overcompensation, the initial compensation gain is amplitude-limited. When the absolute value of the initial compensation gain exceeds a preset amplitude limit, it is limited to within the range [upper amplitude, lower amplitude]. The result after the limiting process is the compensation gain, which comprehensively considers the error effects of both position and speed dimensions.

[0022] S104: Determine a compensation coefficient according to the compensation gain, the rotation speed, and the acceleration.

[0023] For example, the operating state of the servo motor is determined by analyzing the numerical characteristics of the speed and acceleration. When the absolute value of the speed is less than 0.1rad / s and the absolute value of the acceleration is less than 0.5rad / s², it is determined to be in stop-hold mode; when the absolute value of the speed is less than 5rad / s and the absolute value of the acceleration is greater than 10rad / s², it is determined to be in start-up mode; when the absolute value of the acceleration is less than 0.5rad / s² and the speed is stable near the set value, it is determined to be in uniform speed mode; other cases are determined to be in acceleration and deceleration mode. For the start-up mode, the compensation coefficient = preset excitation coefficient × compensation gain, and the excitation coefficient ranges from 1.5 to 2.0, which is used to enhance the driving ability in the startup phase. For the uniform speed mode, the linear coefficient of the speed is calculated = |speed| / rated speed, and the compensation coefficient = linear coefficient × compensation gain to ensure stability during uniform speed operation. For the acceleration and deceleration mode, a quadratic function is established: f(V,A)=k1×V²+k2×V²+k3×V×A; Where k1, k2, and k3 are preset weight parameters, V is the speed, and A is the acceleration. The compensation coefficient = f(V, A) × compensation gain. For stop-hold mode, the compensation coefficient = micro-vibration coefficient × compensation gain. The micro-vibration coefficient ranges from 0.1 to 0.3 and is used for minor adjustments to maintain position lock. This state-specific compensation coefficient calculation method achieves adaptive compensation for different operating conditions.

[0024] S105 , superimposing the original control signal of the servo motor according to the compensation coefficient to obtain a corrected control signal, and controlling the servo motor based on the corrected control signal.

[0025] Exemplarily, the compensation coefficient is decomposed into a feedforward compensation component and a feedback compensation component. The feedforward compensation component is calculated by obtaining the servo motor's position and velocity commands, performing first-order difference operations on each of the position and velocity commands to determine the command change rate. The command change rate for the current cycle and the command change rates for the previous three cycles are weighted averaged, with weights assigned as [0.4, 0.3, 0.2, 0.1], to obtain the predicted command increment. The feedforward compensation component = predicted command increment × feedforward allocation coefficient × compensation coefficient. The typical value of the feedforward allocation coefficient is 0.6. The feedback compensation component is calculated by weighting the position error and velocity error according to preset error weight coefficients (position error weight 0.7, velocity error weight 0.3) to obtain a weighted error value. The feedback compensation component = weighted error value × feedback allocation ratio × compensation coefficient. The typical value of the feedback allocation ratio is 0.4. The feedforward and feedback compensation components are weighted and combined to generate a first compensation signal. The first compensation signal is optimized by limiting its amplitude and rate of change to ensure smoothness, generating a second compensation signal. This second compensation signal is numerically superimposed with the original control signal to produce a corrected control signal. The corrected control signal equals the original control signal plus the second compensation signal. This corrected control signal is then output to the servo motor driver, achieving precise control of the servo motor.

[0026] In order to more clearly introduce the technical solution of the present application, the technical solution of the present application will be introduced through specific embodiments below. It should be noted that the specific embodiments are used to expand the technical solution of the present application, but are not intended to limit the present application.

[0027] In some embodiments, multi-frequency domain excitation is performed on the servo motor to obtain a dead zone threshold and a dead zone slope, including: designing a composite sinusoidal wave excitation signal, the composite sinusoidal wave excitation signal containing multi-frequency domain components from 0.1 Hz to 100 Hz, and the amplitude increasing from a tiny signal to 10% of the rated range; applying the composite sinusoidal wave excitation signal to the servo motor, and collecting position response data and speed response data of the servo motor; performing sliding slicing and fast Fourier transform on the position response data and the speed response data to obtain amplitude-frequency characteristic data and phase-frequency characteristic data; establishing a nonlinear model based on the amplitude-frequency characteristic data and the phase-frequency characteristic data, performing parameter identification on the nonlinear model, and obtaining a parameter identification result; extracting the dead zone threshold and the dead zone slope based on the parameter identification result, the dead zone threshold including: a positive dead zone threshold and a negative dead zone threshold, and the dead zone slope including: a slope inside the dead zone and a slope outside the dead zone.

[0028] For example, designing a composite sinusoidal excitation signal requires creating a signal sequence containing multiple frequency components. Initialize a signal array with a length of 100,000 sampling points, and set the sampling frequency to 10 kHz. Generate a frequency list, starting from 0.1 Hz and increasing logarithmically to 100 Hz, with a total of 50 frequency points. For each frequency point, calculate the corresponding sinusoidal component: sin(2π×frequency×time). Set the initial amplitude of each frequency component to 0.001 times the rated value as the starting value of the small signal. Create an amplitude increment sequence, linearly increasing from 0.001 times the rated value to 0.1 times the rated value, with an increment step of 0.001. For each amplitude level, scale all frequency components according to the amplitude and perform superposition operations to form a composite sinusoidal excitation signal at the current amplitude.

[0029] The generated composite sinusoidal excitation signal is output to the control input of the servo motor through a digital-to-analog converter. Configure the sampling parameters of the data acquisition card, set the sampling frequency to 10kHz, and the sampling channels include the position feedback channel and the velocity feedback channel. Start the data acquisition program and synchronously record the moment when the excitation signal is applied. Continuously collect position response data and velocity response data. Each data point contains a timestamp, position value, and velocity value. After completing the excitation of one amplitude level, wait for the motor response to stabilize. The stability criterion is that the position change of 100 consecutive sampling points is less than 0.001 radians. Record the complete response data at this amplitude level and store it as a separate data file. Repeat the above process until the test of all amplitude levels is completed.

[0030] Preprocess the collected position and velocity response data to remove the DC bias component. Set the sliding window length to 2048 sampling points with a window overlap of 75%. Starting from the data start position, extract the data slice within the first window. Apply the Hanning window function to this data slice to reduce spectral leakage. Perform a fast Fourier transform and calculate the complex spectrum. Extract the amplitude spectrum and phase spectrum from the complex spectrum. Calculate the amplitude of each frequency point: sqrt(real part² + imaginary part²). Calculate the phase of each frequency point: arctan(imaginary part / real part). Store the amplitude-frequency feature data and phase-frequency feature data of the current window in the feature matrix. Move the sliding window forward 512 sampling points (25% of the window length) to extract the next data slice. Repeat the transformation and feature extraction process until the entire data sequence is traversed.

[0031] Analyze the nonlinear characteristics in the amplitude-frequency characteristic data and identify the relationship curve between the input amplitude and the output amplitude. Search for the starting point of the output response in the low-amplitude range with a search step size of 0.0001 times the rated value. When the output amplitude first exceeds the noise threshold (0.0005 times the rated value), record the corresponding input amplitude as a candidate dead-zone threshold. Continue to increase the input amplitude and monitor the rate of change of the output response. When the output response shows a linear growth characteristic, record the input amplitude at this turning point and confirm it as the positive dead-zone threshold. Apply the excitation signal in the reverse direction and repeat the above detection process to obtain the negative dead-zone threshold.

[0032] Calculate the slope of the input-output relationship curve within the dead zone, selecting data points within a ±10% range around the dead zone threshold. Use the least squares method to fit the data points within this range. The slope of the fitted line is the slope within the dead zone. Calculate the slope of the input-output relationship curve outside the dead zone, selecting data points that exceed the dead zone threshold by more than 20%. Use the least squares method to fit the slope outside the dead zone. Verify the accuracy of the recognition results and calculate the fitting residual. If the residual exceeds the threshold, re-adjust the data interval and recalculate.

[0033] Through the above process, a sampling inspection of the entire batch of servo motors or an advance inspection of the servo motors in key positions is performed to obtain the corresponding dead zone threshold and dead zone slope, which are used to improve the accuracy of subsequent real-time control of the servo motors.

[0034] In some embodiments, a disturbance compensation amount is calculated for the position error and the speed error based on the dead zone threshold and the dead zone slope to obtain a compensation gain, including: determining a first weight coefficient for the position error based on the dead zone threshold, and determining a second weight coefficient for the speed error based on the dead zone slope; calculating the product of the position error and the first weight coefficient to obtain a position compensation component; calculating the product of the speed error and the second weight coefficient to obtain a speed compensation component; summing the position compensation component and the speed compensation component, and limiting the amplitude according to a preset limiting condition to obtain the compensation gain.

[0035] Exemplarily, the identified dead zone threshold parameters are read, including the positive dead zone threshold and the negative dead zone threshold, and the average value of the dead zone threshold is calculated. The first weight coefficient of the position error is determined based on the average value of the dead zone threshold, and an inverse proportional relationship is adopted: the first weight coefficient = reference value / (reference value + average value of the dead zone threshold), where the reference value is determined according to the rated parameters of the motor, and a typical value is 0.01 radians. When the dead zone threshold is large, the first weight coefficient is reduced accordingly, indicating that a stronger position compensation effect is required. The first weight coefficient is normalized to ensure that its value range is within the range of [0.1, 1.0].

[0036] Read the deadband slope parameters and the deadband slope parameters. Calculate the slope ratio: Second Weighting Factor = Slope Outside Deadband / Slope Inside Deadband. This ratio reflects the difference in response characteristics between inside and outside the deadband. When the deadband slope approaches zero, the minimum deadband slope is limited to 0.001 to avoid division-by-zero errors. The second weighting factor is capped at a maximum of 10 to prevent overcompensation. Store the second weighting factor in the compensation parameter table for subsequent calculations.

[0037] Obtain the position error value for the current sampling period, which is provided in real time by the position control loop. Multiply the position error by the first weighting coefficient: Position compensation component = Position error × first weighting coefficient. Check the sign of the position compensation component to ensure that the compensation direction aligns with the error direction. If the position error is positive, the position compensation component should also be positive; if the position error is negative, the position compensation component should also be negative. Initially limit the position compensation components to prevent any single component from being too large.

[0038] Obtain the speed error value for the current sampling period, which is provided in real time by the speed control loop. Multiply the speed error by the second weighting coefficient: speed compensation component = speed error × second weighting coefficient. The calculation of the speed compensation component must take into account the dynamic characteristics of the motor. When the speed error changes rapidly, increase the compensation strength appropriately. Filter the speed compensation component using a first-order low-pass filter with a time constant of 0.01 seconds. This filtered speed compensation component suppresses the effects of high-frequency noise.

[0039] Perform an addition operation on the position compensation component and the velocity compensation component to obtain the initial compensation gain. Set the compensation gain limit conditions, with the upper limit being 20% ​​of the rated control amount and the lower limit being -20% of the rated control amount. When the initial compensation gain exceeds the upper limit, it is limited to the upper limit; when the initial compensation gain is lower than the lower limit, it is limited to the lower limit. The limiting process is implemented using a saturation function to ensure output continuity. Check the rate of change of the compensation gain after limiting. The change in the compensation gain between two adjacent sampling cycles should not exceed 5%. If the rate of change is too large, smoothing is performed, using a weighted average of 70% of the current value and 30% of the previous value.

[0040] In some embodiments, the compensation coefficient is determined based on the compensation gain, rotational speed and acceleration, including: determining the operating state based on the rotational speed and acceleration, the operating states including: start mode, uniform speed mode, acceleration / deceleration mode and stop / hold mode; when the operating state is the start mode, the compensation coefficient is determined based on the product of a preset excitation coefficient and the compensation gain; when the operating state is the uniform speed mode, the linear coefficient of the rotational speed is calculated, and the compensation coefficient is determined based on the product of the linear coefficient and the compensation gain; when the operating state is the acceleration / deceleration mode, a quadratic function of the rotational speed and acceleration is established, and the quadratic function value is calculated, and the compensation coefficient is determined based on the product of the quadratic function value and the compensation gain; when the operating state is the stop / hold mode, the compensation coefficient is determined based on the product of a preset micro-vibration coefficient and the compensation gain.

[0041] For example, the current speed and acceleration values ​​are read. These two parameters are provided in real time by the motor status detection module. Threshold parameters for determining the operating state are set: the speed threshold for stop-and-hold mode is 0.1 rad / s², and the acceleration threshold is 0.5 rad / s²; the speed threshold for start-up mode is 5 rad / s², and the acceleration threshold is 10 rad / s²; the acceleration threshold for constant speed mode is 0.5 rad / s², and the speed deviation threshold is 2% of the rated speed. The operating state determination logic is executed: the absolute value of the speed is compared with 0.1 rad / s², and the absolute value of the acceleration is compared with 0.5 rad / s². If both are less than the corresponding thresholds, the state is determined to be stop-and-hold mode. The absolute value of the speed is compared with 5 rad / s², and the absolute value of the acceleration is compared with 10 rad / s². If the speed is less than the threshold and the acceleration is greater than the threshold, the state is determined to be start-up mode. The deviation rate between the current speed and the set speed is calculated. If the deviation rate is less than 2% and the absolute value of the acceleration is less than 0.5 rad / s², the state is determined to be constant speed mode. If none of the above conditions are met, the state is determined to be acceleration / deceleration mode.

[0042] When the startup mode is determined, the preset excitation coefficient is read from the parameter table. The value range of this coefficient is 1.5 to 2.0. The specific value of the excitation coefficient is determined by the motor's inertia characteristics. Larger inertia loads correspond to larger excitation coefficients. A multiplication operation is performed: compensation coefficient = excitation coefficient × compensation gain. The compensation coefficient in startup mode is used to enhance the motor's starting torque and overcome the effects of static friction and dead zone. The calculated compensation coefficient is checked for plausibility to ensure it does not exceed the motor's maximum allowable compensation range.

[0043] When the motor is in constant speed mode, the linear coefficient of the speed is calculated. The rated speed of the motor is read and the division operation is performed: linear coefficient = |current speed| / rated speed. The linear coefficient reflects the proportional relationship between the current speed and the rated speed. The multiplication operation is performed: compensation coefficient = linear coefficient × compensation gain. The compensation coefficient in constant speed mode is proportional to the speed. Higher speeds require more compensation. The compensation coefficient is smoothed to avoid unnecessary disturbances during constant speed operation.

[0044] When acceleration / deceleration mode is determined, the quadratic function coefficients k1, k2, and k3 are defined. k1 corresponds to the weight of the squared speed term, with a typical value of 0.001; k2 corresponds to the weight of the squared acceleration term, with a typical value of 0.002; and k3 corresponds to the weight of the cross term between speed and acceleration, with a typical value of 0.0005. The quadratic function calculation is: quadratic function value = k1 × V² + k2 × A² + k3 × V × A. This quadratic function comprehensively considers the independent effects of speed and acceleration as well as their coupling. The multiplication operation is: compensation coefficient = quadratic function value × compensation gain. The compensation coefficient in acceleration / deceleration mode can adapt to the dynamic operating conditions of the motor.

[0045] When the stop-and-hold mode is determined, the preset micro-vibration coefficient is read from the parameter table. The value range of this coefficient is 0.1 to 0.3. The micro-vibration coefficient provides a small compensation when the motor is stationary to overcome static friction and maintain position accuracy. The multiplication operation is: Compensation coefficient = Micro-vibration coefficient × Compensation gain. The compensation coefficient in stop-and-hold mode is small to avoid unnecessary vibration when the motor is stationary. The compensation coefficient is processed with a zero dead zone, and its absolute value is reset to zero when it is less than the minimum resolution.

[0046] The determined compensation coefficient is written to the output buffer for use in subsequent signal superposition. The current operating status and compensation coefficient value are recorded for status monitoring and fault diagnosis. The historical record queue of the compensation coefficient is updated, retaining data from the last 100 sampling cycles for trend analysis and performance evaluation.

[0047] In some embodiments, the original control signal of the servo motor is superimposed according to the compensation coefficient to obtain a corrected control signal, including: decomposing the compensation coefficient into a feedforward compensation component and a feedback compensation component, the feedforward compensation component is generated based on instruction prediction, and the feedback compensation component is generated based on error feedback; weighted synthesis of the feedforward compensation component and the feedback compensation component to generate a first compensation signal, optimizing the first compensation signal according to a preset amplitude restriction condition and a preset rate of change restriction condition to obtain a second compensation signal; numerically superimposing the second compensation signal with the original control signal to output a corrected control signal.

[0048] For example, the compensation coefficient value for the current control cycle is read. This value was provided by the previous calculation step and stored in the compensation parameter register. The feedforward compensation component variable and the feedback compensation component variable are initialized, with their initial values ​​set to zero. The decomposition ratio parameters are defined, with the feedforward allocation ratio set to 0.6 and the feedback allocation ratio set to 0.4, with the sum of the two equal to 1.0. The compensation coefficient decomposition operation is performed, and the compensation coefficient is assigned to the feedforward channel and feedback channel respectively according to the allocation ratio.

[0049] The generation of the feedforward compensation component requires a command prediction mechanism. Read the current position and velocity command values ​​from the command buffer. Create a command history array to store the command data for the last 10 control cycles. Perform a first-order difference operation on the current position command: Position command rate of change = (current position command - previous cycle position command) / control cycle. Perform the same difference operation on the current velocity command: Velocity command rate of change = (current velocity command - previous cycle velocity command) / control cycle. Store the calculated command rate of change in the rate of change buffer. Extract the command rate of change data for the previous three historical cycles from the rate of change buffer. Set the weighting coefficient array [0.5, 0.3, 0.2] to correspond to the weights for the current cycle, previous cycle, and previous two cycles. Perform a weighted average operation: Predicted command increment = 0.5 × current rate of change + 0.3 × previous cycle rate of change + 0.2 × previous two cycle rate of change. Read the preset feedforward allocation coefficient, which is stored in the configuration parameter table. Perform continuous multiplication operations: feedforward compensation component = predicted instruction increment × feedforward allocation coefficient × (compensation coefficient × 0.6).

[0050] The feedback compensation component is generated based on the error feedback mechanism. Read the current position error and velocity error from the error register. Set the error weight coefficients to 0.7 for position error and 0.3 for velocity error. Perform a weighted sum operation: Weighted error value = position error × 0.7 + velocity error × 0.3. Read the preset feedback allocation ratio parameters. Perform a continuous multiplication operation: Feedback compensation component = Weighted error value × Feedback allocation ratio × (compensation coefficient × 0.4).

[0051] During the weighted synthesis of the feedforward and feedback compensation components, the synthesis weight parameters are defined: the feedforward weight is set to 0.6, and the feedback weight is set to 0.4. The weighted synthesis operation is performed: first compensation signal = feedforward compensation component × 0.6 + feedback compensation component × 0.4. The amplitude of the first compensation signal is checked, and the preset amplitude limit conditions are read. The upper limit is 30% of the rated control value, and the lower limit is -30%. If the first compensation signal exceeds the upper limit, it is limited to the upper limit; if it falls below the lower limit, it is limited to the lower limit.

[0052] Apply rate-of-change limiting to the limited signal. Calculate the difference between the current compensation signal and the previous compensation signal: Signal change = Current first compensation signal - Previous compensation signal. Read the preset rate-of-change limit conditions; the maximum allowable rate of change is 5% of the rated value per control cycle. When the signal change exceeds the maximum allowable rate of change, limit the current output to: Second compensation signal = Previous compensation signal + Sign (Signal change) × Maximum allowable rate of change. When the signal change is within the allowable range, the second compensation signal equals the first compensation signal. Store the second compensation signal in the compensation signal buffer for use in the next cycle's rate-of-change calculation.

[0053] The original control signal, the result of the position and velocity loop calculations, is read from the controller output. A numerical superposition operation is performed: Corrected control signal = Original control signal + Second compensation signal. A plausibility check is performed on the corrected control signal to ensure it is within the servo drive's input range. The corrected control signal is written to the output buffer, awaiting readout by the digital-to-analog converter. The control signal history is updated, saving the last 100 cycles for performance analysis. The output update flag is triggered, notifying the drive to read the new control signal.

[0054] In some embodiments, the compensation coefficient is decomposed into a feedforward compensation component and a feedback compensation component, the feedforward compensation component is generated based on instruction prediction, and the feedback compensation component is generated based on error feedback, including: obtaining the position instruction and speed instruction of the servo motor, performing a first-order difference operation on the position instruction and the speed instruction to obtain the instruction change rate; performing a weighted average of the instruction change rate and the instruction change rate of the previous three historical cycles to obtain a predicted instruction increment; multiplying the predicted instruction increment with a preset feedforward allocation coefficient, and then multiplying it with the compensation coefficient to obtain a feedforward compensation component; performing a weighted sum operation on the position error and the speed error according to a preset error weight coefficient to obtain a weighted error value; multiplying the weighted error value with a preset feedback allocation ratio, and then multiplying it with the compensation coefficient to obtain a feedback compensation component.

[0055] For example, the servo motor's position and speed commands are read from the command interface. These commands are issued in real time by the upper-level motion controller. A command storage array is created to store historical command data. The currently read position command is stored in the latest position of the position command array, with the array index automatically incremented. The currently read speed command is stored in the latest position of the speed command array to maintain synchronization with the position command.

[0056] Perform a first-order difference operation on the position command, and the same difference operation on the speed command: speed command change rate [k] = (speed command [k] - speed command [k-1]) / sampling period. The calculated command change rate is stored in the change rate array, which also holds 100 historical data points.

[0057] Create historical change rate extraction indexes to point to the data locations of the current cycle, previous cycle, previous two cycles, and previous three cycles respectively. Perform vector dot product operations based on the extraction indexes and instruction change rates to obtain the predicted instruction increment: predicted instruction increment = 0.4 × current change rate + 0.3 × first historical change rate + 0.2 × second historical change rate 2 + 0.1 × third historical change rate.

[0058] Read the preset feedforward allocation coefficient from the configuration register. The default value of this coefficient is 0.65 and can be adjusted according to the actual application. Read the current compensation coefficient value, which is provided by the compensation coefficient determination step mentioned above. Perform a two-level multiplication operation: intermediate value = predicted command increment × feedforward allocation coefficient, feedforward compensation component = intermediate value × compensation coefficient. Check the validity of the feedforward compensation component to ensure that its value is within a reasonable range. Read the preset error weight coefficient. The position error weight defaults to 0.7, and the speed error weight defaults to 0.3. Perform a weighted summation operation on the errors: weighted error value = position error × position error weight + speed error × speed error weight. This weighting method reflects the control strategy that prioritizes position accuracy.

[0059] Read the preset feedback allocation ratio from the configuration register; the default value is 0.35. Perform a two-stage multiplication operation: intermediate feedback value = weighted error value × feedback allocation ratio, and feedback compensation component = intermediate feedback value × compensation coefficient. Verify the proportional relationship between the feedforward compensation component and the feedback compensation component, ensuring that their sum does not exceed the total compensation limit. Write the feedforward compensation component to the feedforward channel buffer, and write the feedback compensation component to the feedback channel buffer. Update the component calculation completion flag to notify subsequent synthesis stages to read data. Record the current decomposition results, including the feedforward compensation component, feedback compensation component, predicted instruction increment, and weighted error value, for debugging and optimization analysis.

[0060] In some embodiments, a nonlinear model is established based on the amplitude-frequency characteristic data and the phase-frequency characteristic data, and parameters of the nonlinear model are identified to obtain parameter identification results, including: normalizing the amplitude-frequency characteristic data and the phase-frequency characteristic data to obtain normalized characteristic data; segmenting the normalized characteristic data according to multiple preset frequency bands to obtain segmented characteristic data; respectively calculating the linear slope, intercept parameter and coordinate value of the connection point of the segmented characteristic data in each frequency band to obtain first fitting parameters; setting the initial population of the genetic algorithm according to the first fitting parameters, and performing iterative mutation based on the initial population to obtain second fitting parameters; reconstructing the nonlinear model using the second fitting parameters, and the nonlinear model is described by a piecewise linear function; calculating the residual sum of squares between the output data of the nonlinear model and the actual measurement data, stopping the iteration when the residual sum of squares is less than a preset residual threshold, and obtaining target model parameters; extracting the slope value, inflection point coordinates and linear interval range of each frequency band from the target model parameters as parameter identification results.

[0061] For example, the maximum and minimum values ​​of the amplitude-frequency characteristic data are calculated to determine the dynamic range of the data. The maximum and minimum values ​​of the phase-frequency characteristic data are calculated to determine the phase variation range. A normalization operation is performed on the amplitude-frequency characteristic data: normalized amplitude-frequency data = (original amplitude-frequency data - minimum value) / (maximum value - minimum value). The same normalization operation is performed on the phase-frequency characteristic data: normalized phase-frequency data = (original phase-frequency data - minimum value) / (maximum value - minimum value). The normalized data is stored in a normalized characteristic data array.

[0062] Define the frequency band partitioning parameters, dividing the frequency range from 0.1Hz to 100Hz into five intervals: [0.1-1Hz], [1-10Hz], [10-30Hz], [30-60Hz], and [60-100Hz]. Create a segment index array to record the start and end positions of each frequency band in the data array. Traverse the normalized feature data and assign each data point to the corresponding frequency band based on its corresponding frequency value. Create an independent data subset for each frequency band to form a segmented feature data structure.

[0063] Select a set of data points from the first frequency band, in the format (frequency, amplitude) or (frequency, phase). Construct a linear equation: y = ax + b, where x is the frequency, y is the amplitude or phase, a is the slope, and b is the intercept. Solve the linear equations using the least squares method. The calculations include: calculating the mean of x and y, the covariance of x and y, and the variance of x. The slope a = covariance / variance, and the intercept b = mean y - a × mean x. Record the slope and intercept for this frequency band.

[0064] Calculate the end point of the first frequency band: x1 = the upper frequency band limit, y1 = a1 × x1 + b1. Calculate the starting point of the second frequency band: x2 = the lower frequency band limit, y2 = a2 × x2 + b2. Take the average of the two points for the connection point coordinates: connection point x = (x1 + x2) / 2, connection point y = (y1 + y2) / 2. Repeat the above process to calculate the connection points for all adjacent frequency bands. Combine the linear slopes, intercept parameters, and connection point coordinates for all frequency bands to form the first set of fitting parameters.

[0065] The first fitting parameters were used as the individual genes for the initial population, with a population size of 50 individuals, a crossover probability of 0.8, and a mutation probability of 0.1. An iterative calculation was performed with the minimum fitting error as the optimization goal to obtain the second fitting parameters. Specifically, the genetic algorithm population was initialized based on the first fitting parameters, and an individual data structure was created, with each individual containing the slope, intercept, and tie point parameters for all frequency bands. Fifty initial individuals were generated, of which the first individual used the first fitting parameters directly, and the remaining 49 individuals had ±10% random perturbations added to the first fitting parameters. The genetic algorithm operating parameters were set as follows: maximum number of iterations 1000, population size 50, crossover probability 0.8, and mutation probability 0.1.

[0066] A piecewise linear function is constructed using the individual parameters and its predicted values ​​are calculated for all data points. The difference between the predicted values ​​and the actual measured values ​​is calculated, and the sum of the squares of these differences is taken as the fitting error. The fitness value is defined as 1 / (1 + fitting error); the smaller the error, the higher the fitness.

[0067] Sort the population by fitness and select the 25 individuals with the highest fitness as the parents. Perform a crossover: Randomly select two parents and swap some of their genes with a probability of 0.8, generating two offspring. Perform a mutation: Randomly perturb each gene position with a probability of 0.1, within a range of ±5% of the current value. Merge the parents and offspring, and select the 50 individuals with the highest fitness as the next generation population. Record the best individual of each generation and the corresponding fitting error.

[0068] Iterations were terminated when the number of iterations reached 1000 or when the fitting error of the optimal individual showed no improvement after 10 consecutive generations. The parameters of the optimal individual were extracted as the second fitting parameters. The second fitting parameters were used to construct a nonlinear model in the form of a piecewise linear function. The output of this model was calculated at all measurement points and compared with the actual measurement data. After calculating the sum of squared residuals, a preset residual threshold of 0.001 was set. Parameter identification was considered successful when the sum of squared residuals was less than this threshold.

[0069] Traverse each frequency band and extract the corresponding slope values. These slopes reflect the gain characteristics of different frequency bands. Extract the coordinates of all inflection points. The inflection point locations indicate the transition frequencies of the nonlinear characteristics. Determine the effective range of each linear interval, i.e., [start frequency, end frequency]. Organize the extracted slope values, inflection point coordinates, and linear interval ranges into a structured parameter identification result. Save the parameter identification results to a configuration file for use by the dead zone compensation algorithm.

[0070] See also Figure 2 , Figure 2 is a schematic block diagram of a dead zone optimization device for a servo motor provided by an embodiment of the present application. The dead zone optimization device 200 for a servo motor is used to execute the aforementioned dead zone optimization method for a servo motor. The dead zone optimization device 200 for a servo motor can be configured in a server.

[0071] Among them, the server can be an independent server, a server cluster, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0072] like Figure 2 As shown, the dead zone optimization device 200 for a servo motor includes: an excitation detection module 201 , a parameter acquisition module 202 , a compensation calculation module 203 , a coefficient determination module 204 and a signal correction module 205 .

[0073] The excitation detection module 201 is used to perform multi-frequency domain excitation on the servo motor to obtain a dead zone threshold and a dead zone slope.

[0074] The parameter acquisition module 202 is used to detect the position error, speed error, rotation speed and acceleration of the servo motor at the current position.

[0075] The compensation calculation module 203 is used to calculate the disturbance compensation amount for the position error and the speed error according to the dead zone threshold and the dead zone slope to obtain a compensation gain.

[0076] The coefficient determination module 204 is configured to determine a compensation coefficient according to the compensation gain, the rotational speed, and the acceleration.

[0077] The signal correction module 205 is configured to superimpose the original control signal of the servo motor according to the compensation coefficient to obtain a corrected control signal, and control the servo motor based on the corrected control signal.

[0078] An embodiment of the present application provides an electronic device, which includes a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement a dead zone optimization method for a servo motor as described in any one of the embodiments of the present application when executing the computer program.

[0079] An embodiment of the present application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor implements any one of the dead zone optimization methods for a servo motor according to the embodiments of the present application.

[0080] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present application, and such modifications or substitutions should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A dead zone optimization method for a servo motor, characterized in that: The method comprises: Perform multi-frequency domain excitation on the servo motor to obtain the dead zone threshold and dead zone slope; Detecting the position error, velocity error, rotation speed and acceleration of the servo motor at the current position; Calculating a disturbance compensation amount for the position error and the speed error according to the dead zone threshold and the dead zone slope to obtain a compensation gain; determining a compensation coefficient according to the compensation gain, the rotational speed, and the acceleration; The original control signal of the servo motor is superimposed according to the compensation coefficient to obtain a modified control signal, and the servo motor is controlled based on the modified control signal.

2. The dead zone optimization method for a servo motor according to claim 1, wherein: The multi-frequency domain excitation of the servo motor is performed to obtain a dead zone threshold and a dead zone slope, including: Design a composite sine wave excitation signal; Applying the composite sinusoidal wave excitation signal to the servo motor, and collecting position response data and speed response data of the servo motor; Performing sliding slicing and fast Fourier transform on the position response data and the velocity response data to obtain amplitude-frequency characteristic data and phase-frequency characteristic data; Establishing a nonlinear model according to the amplitude-frequency characteristic data and the phase-frequency characteristic data, performing parameter identification on the nonlinear model, and obtaining a parameter identification result; The dead zone threshold and the dead zone slope are extracted according to the parameter identification result. The dead zone threshold includes a positive dead zone threshold and a negative dead zone threshold. The dead zone slope includes an inner dead zone slope and an outer dead zone slope.

3. The dead zone optimization method for a servo motor according to claim 1, wherein: The calculating the disturbance compensation amount for the position error and the speed error according to the dead zone threshold and the dead zone slope to obtain the compensation gain includes: Determine a first weight coefficient of the position error according to the dead zone threshold, and determine a second weight coefficient of the speed error according to the dead zone slope; Calculating the product of the position error and the first weight coefficient to obtain a position compensation component; Calculating the product of the speed error and the second weight coefficient to obtain a speed compensation component; The position compensation component and the speed compensation component are summed, and the amplitude is limited according to a preset limiting condition to obtain a compensation gain.

4. The dead zone optimization method for a servo motor according to claim 1, wherein: The determining of the compensation coefficient according to the compensation gain, the rotational speed, and the acceleration includes: Determine an operating state according to the rotational speed and the acceleration, the operating state including: a start mode, a constant speed mode, an acceleration / deceleration mode, and a stop / hold mode; When the operating state is the startup mode, determining a compensation coefficient according to a product of a preset excitation coefficient and the compensation gain; When the operating state is the uniform speed mode, calculating the linear coefficient of the rotational speed, and determining the compensation coefficient according to the product of the linear coefficient and the compensation gain; When the operating state is the acceleration / deceleration mode, a quadratic function of the rotational speed and the acceleration is established, a quadratic function value is calculated, and a compensation coefficient is determined according to a product of the quadratic function value and the compensation gain; When the operating state is the stop-and-hold mode, the compensation coefficient is determined according to the product of a preset micro-vibration coefficient and the compensation gain.

5. The dead zone optimization method for a servo motor according to claim 1, wherein: The method of superimposing the original control signal of the servo motor according to the compensation coefficient to obtain a corrected control signal includes: Decomposing the compensation coefficient into a feedforward compensation component and a feedback compensation component, wherein the feedforward compensation component is generated based on instruction prediction and the feedback compensation component is generated based on error feedback; Performing weighted synthesis of the feedforward compensation component and the feedback compensation component to generate a first compensation signal, and optimizing the first compensation signal according to a preset amplitude constraint and a preset rate-of-change constraint to obtain a second compensation signal; The second compensation signal is numerically superimposed on the original control signal to output a modified control signal.

6. The dead zone optimization method for a servo motor according to claim 5, characterized in that: Decomposing the compensation coefficient into a feedforward compensation component and a feedback compensation component, wherein the feedforward compensation component is generated based on instruction prediction and the feedback compensation component is generated based on error feedback, comprises: Obtaining a position command and a speed command of the servo motor, performing a first-order difference operation on the position command and the speed command to obtain a command change rate; Taking a weighted average of the instruction change rate and the instruction change rates of the previous three historical cycles to obtain a predicted instruction increment; Multiplying the predicted instruction increment by a preset feedforward allocation coefficient, and then performing a product operation with the compensation coefficient to obtain a feedforward compensation component; Performing a weighted sum operation on the position error and the speed error according to a preset error weight coefficient to obtain a weighted error value; The weighted error value is multiplied by the preset feedback allocation ratio, and then multiplied by the compensation coefficient to obtain a feedback compensation component.

7. The dead zone optimization method for a servo motor according to claim 2, wherein: The step of establishing a nonlinear model based on the amplitude-frequency characteristic data and the phase-frequency characteristic data, and performing parameter identification on the nonlinear model to obtain a parameter identification result includes: Normalizing the amplitude-frequency characteristic data and the phase-frequency characteristic data to obtain normalized characteristic data; Segmenting the normalized feature data according to a plurality of preset frequency band intervals to obtain segmented feature data; Calculating the linear slope, intercept parameter and coordinate value of the connection point of the segmented feature data in each frequency band interval respectively to obtain a first fitting parameter; Setting an initial population of the genetic algorithm according to the first fitting parameter, and performing iterative mutation based on the initial population to obtain a second fitting parameter; Reconstructing the nonlinear model using the second fitting parameters, where the nonlinear model is described using a piecewise linear function; Calculating the residual sum of squares between the output data of the nonlinear model and the actual measurement data, stopping the iteration when the residual sum of squares is less than a preset residual threshold, and obtaining the target model parameters; The slope value, inflection point coordinates and linear interval range of each frequency band interval are extracted from the target model parameters as parameter identification results.

8. A dead zone optimization device for a servo motor, characterized in that: The dead zone optimization device for a servo motor is used to execute the dead zone optimization method for a servo motor according to any one of claims 1 to 7, and the dead zone optimization device for a servo motor includes: The excitation detection module is used to perform multi-frequency domain excitation on the servo motor to obtain the dead zone threshold and dead zone slope; A parameter acquisition module is used to detect the position error, speed error, speed and acceleration of the servo motor at the current position; a compensation calculation module, configured to calculate a disturbance compensation amount for the position error and the speed error according to the dead zone threshold and the dead zone slope to obtain a compensation gain; a coefficient determination module, configured to determine a compensation coefficient according to the compensation gain, the rotational speed, and the acceleration; The signal correction module is used to superimpose the original control signal of the servo motor according to the compensation coefficient to obtain a corrected control signal, and control the servo motor based on the corrected control signal.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor; The memory is used to store computer programs; The processor is configured to execute the computer program and implement the dead zone optimization method for a servo motor according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, enables the processor to implement the dead zone optimization method for a servo motor according to any one of claims 1 to 7.

Images