Motor driving precision correction method and system based on force feedback data

By using a motor drive accuracy correction method based on force feedback data, the problem of lag in dynamic error capture in traditional methods is solved, and the accuracy maintenance and control stability of the motor drive system under complex working conditions are achieved, improving response sensitivity and correction efficiency.

CN121055857AInactive Publication Date: 2025-12-02SHEN ZHEN SAN YA TECH LTD CO
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202511599216.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-11-04
Publication Date
2025-12-02
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional motor drive accuracy correction methods rely on hardware sensors, which make it difficult to capture dynamic errors in a timely manner. In particular, during the startup phase or under sudden changes in operating conditions, correction lag is likely to occur, leading to oscillation or deviation of the drive system. Furthermore, the lack of ability to identify high-frequency disturbances affects the accuracy and reliability of the system.

Method used

A motor drive accuracy correction method based on force feedback data is adopted. By acquiring force feedback sensor data, calculating the force feedback growth rate, constructing an error early warning threshold, screening dynamic error characteristic frequency bands, and establishing a mapping table between the energy value of the characteristic frequency band and the average force feedback value during steady-state operation, a forward-looking error correction and dynamic error adjustment are realized, and the correction torque is superimposed on the drive current command.

Benefits of technology

It effectively suppresses operational deviations, improves the system's accuracy in complex environments, enhances the response sensitivity and correction efficiency to abnormal operating conditions, and ensures control stability and continuity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121055857A_ABST
    Figure CN121055857A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of error correction, in particular to a motor driving precision correction method and system based on force feedback data, and the method comprises the following steps: obtaining a force feedback and pulse number calculation growth rate peak value definition early warning threshold value, extracting multi-speed frequency domain amplitude, screening characteristic frequency bands, and constructing a mapping table; in the initial driving stage, abnormity is judged, a prospective correction torque is generated, frequency domain characteristics are extracted, energy is calculated, a mapping table is called for interpolation correction, and prospective and dynamic correction torque is superposed to adjust a current instruction. According to the invention, through combination of a force feedback sensor and frequency domain feature extraction, construction of a force feedback growth rate peak early warning threshold value, screening of a dynamic error feature frequency band, prospective identification of an abnormal state, error compensation according to frequency band energy and steady-state average force feedback mapping, and dual-path linkage, prospective correction and dynamic adjustment are completed. Frequency domain compensation inhibits transient control drift, precision is kept under multiple loads, the abnormal response and correction efficiency without hardware intervention is improved, and stability and continuity of system control are guaranteed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of error correction technology, and in particular to a method and system for correcting the accuracy of motor drives based on force feedback data. Background Technology

[0002] Error correction technology is an important interdisciplinary field in computer and control engineering. Its core aspects include detecting and correcting errors generated during the transmission and processing of digital information or control signals. This involves monitoring data integrity, locating and repairing anomalies, and adjusting deviations in the control process to ensure system stability and reliability. This technology systematically covers the entire process from data acquisition, error detection, error analysis to error correction, and its applications extend to digital data processing, mechatronics control, and complex motion control systems. Traditional motor drive accuracy correction methods involve monitoring position or speed deviations during motor operation for correction. This typically uses encoder feedback for position correction or speed measurement devices for speed correction, combined with current detection to adjust motor torque. These methods rely on comparing the measurement results from hardware sensors with the target values ​​set by the controller, and then using control algorithms to correct the drive signal to complete the correction process.

[0003] Traditional methods rely heavily on encoders and speed measuring devices to provide position and speed error information for error correction. When sensor accuracy decreases or response lags, it is difficult to capture dynamic errors in the early stages of operation or under rapid load changes. In particular, correction lag problems are prone to occur during the startup phase or under sudden changes in operating conditions, causing the drive system to oscillate or deviate in a short period of time. Current feedback is insufficient to respond to torque control under abnormal changes, and there is a lack of ability to identify high-frequency disturbances or non-steady-state changes. This makes the control system susceptible to instantaneous instability in complex operating environments. For example, under frequent start-stop or heavy-load speed change conditions, the lack of auxiliary judgment on force feedback and frequency domain characteristics makes it impossible to cut off the error transmission path in time, affecting the accuracy and reliability of the overall system. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and to propose a method and system for correcting the accuracy of motor drive based on force feedback data.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for correcting the accuracy of motor drive based on force feedback data, comprising the following steps: S1: Obtain the raw values ​​of the force feedback sensor and the cumulative pulse count of the position encoder, calculate the peak value of the force feedback growth rate, define the peak value statistics as the start-up error warning threshold, obtain the frequency domain amplitude distribution of the raw values ​​of the force feedback sensor under multiple speeds, filter the dynamic error characteristic frequency band, calculate the energy value of the characteristic frequency band, obtain the steady-state average force feedback value, and construct a force feedback characteristic-correction torque mapping table between the characteristic frequency band energy value and the steady-state average force feedback value. S2: Increase the drive current command value, obtain the original value of the force feedback sensor, calculate the force feedback growth rate, and if the force feedback growth rate exceeds the start-up error warning threshold, fit the original value of the force feedback sensor, calculate the predicted torque value, and generate a forward-looking error correction torque. S3: Collect the raw values ​​of the force feedback sensor, perform a fast Fourier transform, extract the frequency domain data of the dynamic error characteristic frequency band, calculate the energy value of the characteristic frequency band, and call the force feedback characteristic-correction torque mapping table to calculate the dynamic correction torque by linear interpolation; S4: Superimpose the forward error correction torque and the driving torque, add the dynamic correction torque to the feedforward control channel, and adjust the driving current command value.

[0006] As a further embodiment of the present invention, the force feedback feature-correction torque mapping table includes feature energy bonds and friction torque values. The forward-looking error correction torque is calculated based on polynomial fitting coefficients, time series reference points, and predicted time steps. The dynamic correction torque is calculated based on neighboring energy interpolation points, stress torque reference values, and interpolation weighting coefficients. The driving current command value includes the current corresponding to the original position command, the compensation current during the startup phase, and the compensation current during steady-state operation.

[0007] As a further aspect of the present invention, the specific steps of S1 are as follows: collecting the original values ​​of the force feedback sensor and the cumulative pulse count of the position encoder when the motor is running under multiple preset constant speeds; A fast Fourier transform is performed on the raw values ​​of the force feedback sensor under each set of constant rotational speeds to generate a frequency domain amplitude distribution; By comparing the frequency domain amplitude distributions under multiple sets of constant rotation speeds, the frequency band corresponding to the common resonance peak that exists under different rotation speeds is selected and defined as the dynamic error characteristic frequency band. Based on the formula: ; Calculate the energy within the dynamic error characteristic frequency band and generate the characteristic frequency band energy value, wherein, The energy value of the characteristic frequency band. For frequency The amplitude at that point, and These are the start frequency and cutoff frequency of the dynamic error characteristic frequency band, respectively; Calculate the time average of the raw values ​​of the force feedback sensor under each group of constant speed to obtain the steady-state average force feedback value; Using the energy value of the characteristic frequency band as the key and the average force feedback value of steady-state operation as the value, a force feedback characteristic-correction torque mapping table is established.

[0008] As a further aspect of the present invention, the specific steps of S2 are as follows: during the motor start-up phase, the original values ​​of the force feedback sensor are acquired in real time and its first-order time difference is calculated to obtain the force feedback growth rate. Determine whether the force feedback growth rate is greater than the start-up error warning threshold. If the determination result is yes, then trigger error correction. The original numerical sequence of the force feedback sensor within a preset time window before the trigger correction time is extracted, and polynomial fitting is performed using the least squares method to generate polynomial fitting coefficients. Based on the polynomial fitting coefficients, the force feedback value is predicted after one prediction time step, and the predicted torque value is obtained. The predicted torque value is used as a compensation amount to generate a forward error correction torque.

[0009] As a further aspect of the present invention, the specific steps of S3 are as follows: during the steady-state operation phase of the motor, the original values ​​of the force feedback sensor are periodically collected; Perform a fast Fourier transform on the newly acquired raw values ​​of the force feedback sensor to extract the frequency domain data corresponding to the dynamic error characteristic frequency band; Calculate the current energy value of the characteristic frequency band based on the frequency domain data corresponding to the dynamic error characteristic frequency band; Call the force feedback feature-correction torque mapping table, find two feature energy keys adjacent to the current feature frequency band energy value and their corresponding friction torque values, and obtain the nearest energy interpolation point and the stress torque reference value; Based on the formula: ; Perform linear interpolation calculations to generate dynamic correction torque; in, For the dynamic correction torque, The current characteristic frequency band energy value, and These are the two mapping points formed by the nearest energy interpolation point and the reference value of the stress moment.

[0010] As a further aspect of the present invention, the specific step of S4 is: obtaining the original driving torque output by the controller; The forward error correction torque is superimposed on the original driving torque to generate the start-up phase compensation torque; The dynamic correction torque is used as a feedforward component and added to the compensation torque during the startup phase and the feedback control torque calculated via the position loop and velocity loop to obtain the final composite torque. Based on the ratio of the final combined torque to the motor torque constant, the adjusted drive current value is calculated, and the adjusted drive current command value is generated.

[0011] As a further aspect of the present invention, the order of the polynomial fitting is preset to third order according to the motor load characteristics; The predicted torque value is calculated as follows: the current time plus the predicted time step is used as the independent variable, and then substituted into the polynomial function determined by the polynomial fitting coefficients. The calculated function value is the predicted torque value. The predicted time step is determined based on the response delay of the motor control system.

[0012] As a further aspect of the present invention, the method for determining the start-up error warning threshold is as follows: during the motor debugging stage, multiple start-up processes from a standstill to a preset speed are performed, and the original values ​​of the force feedback sensor are recorded during each start-up process. Calculate the first-order time difference of the original values ​​of the force feedback sensor during each startup process to obtain the instantaneous force feedback growth rate sequence; Extract the maximum value from each instantaneous force feedback growth rate sequence to form a force feedback growth rate peak sample set; Calculate the mean and standard deviation of the peak sample set of the force feedback growth rate; Based on the formula: ; The sum of the mean and three times the standard deviation is defined as the trigger error warning threshold; in, This represents the startup error warning threshold. The mean of the peak sample set representing the force feedback growth rate is... The standard deviation of the peak sample set representing the force feedback growth rate is given.

[0013] As a further embodiment of the present invention, in the driving current command value, the current corresponding to the original position command is calculated and generated by the position controller based on the position deviation. The start-up phase compensation current is derived from the forward error correction torque divided by the motor torque constant, and the start-up phase compensation current is superimposed on the original position command current only during the motor start-up phase and when the start-up error warning threshold condition is met. The steady-state operating compensation current is derived from the dynamic correction torque divided by the motor torque constant, and during the steady-state operation phase of the motor, the steady-state operating compensation current is added to the feedforward control channel.

[0014] A motor drive accuracy correction system based on force feedback data, the system being used to implement the aforementioned motor drive accuracy correction method based on force feedback data, the system comprising: The mapping table construction module obtains the original values ​​of the force feedback sensor and the cumulative pulse count of the position encoder, calculates the peak value of the force feedback growth rate to define the start-up error warning threshold, and, based on the frequency domain analysis of the original values ​​of the force feedback sensor under multiple speeds, filters the dynamic error characteristic frequency bands, calculates the energy value of the characteristic frequency bands, and, combined with the steady-state average force feedback value, constructs a force feedback characteristic-correction torque mapping table. The start-up correction module increases the drive current command value when the motor starts, calculates the force feedback growth rate in real time, and when the force feedback growth rate exceeds the start-up error warning threshold, it fits the original value of the force feedback sensor to calculate the predicted torque value and generates a forward-looking error correction torque. The dynamic correction module collects the raw values ​​of the force feedback sensor when the motor is running in steady state, extracts the data of the dynamic error characteristic frequency band through fast Fourier transform, calculates the real-time characteristic frequency band energy value, and calls the force feedback feature-correction torque mapping table to generate the dynamic correction torque through linear interpolation. The instruction adjustment module superimposes the forward error correction torque with the original driving torque and adds the dynamic correction torque to the feedforward control channel. Finally, it combines the feedback control torque to adjust and generate the final driving current instruction value.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by combining force feedback sensor data with frequency domain feature extraction, an error warning threshold is constructed using the peak value of the force feedback growth rate. Dynamic error characteristic frequency bands are screened based on the amplitude distribution under multiple speeds to achieve forward-looking identification and correction of abnormal states. A mapping table is established based on the energy of the characteristic frequency bands and the steady-state average force feedback to complete real-time dynamic error identification and compensation. The forward-looking error correction and dynamic error adjustment are linked in a dual-path manner to effectively suppress drive anomalies caused by the accumulation of operating deviations. The frequency domain correction compensates for the control signal drift caused by transient fluctuations, thereby maintaining the accuracy of the drive process under varying loads and complex operating conditions. This improves the system's response sensitivity and correction efficiency to abnormal conditions without additional hardware intervention, ensuring control stability and continuity. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the overall process of the motor drive accuracy correction method of the present invention. Figure 2 This is a flowchart illustrating the construction process of the error warning and force feedback feature mapping for this invention. Figure 3 This is a flowchart illustrating the generation process of the forward-looking error correction torque of the present invention. Figure 4 This is a flowchart illustrating the calculation process of the dynamic correction torque of the present invention. Figure 5 This is a flowchart illustrating the adjustment of the drive current command value according to the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the software-based technical solution is described in detail below with reference to system architecture diagrams and embodiments. It should be understood that the specific embodiments described herein are only for explaining the technical solutions of this invention and do not constitute a limitation on the scope of protection.

[0018] In the description of this invention, the system architecture relationships or data processing flows indicated by terms such as "layer," "module," "interface," "data flow," "client," and "server" are all defined based on the architecture diagram or flowchart corresponding to the embodiments. This way of describing is only used to clearly illustrate the logical relationships between the elements in the technical solution, and not to limit the physical deployment form. The term "multiple" includes two or more technical units, including but not limited to multiple data nodes, processing threads, service instances, or functional components and other scalable elements. The specific number is determined according to the actual business scenario and needs to be specifically specified.

[0019] Please see Figure 1 and Figure 2 This invention provides a technical solution: a method for correcting the accuracy of motor drive based on force feedback data, comprising the following steps: S1: Obtain the raw values ​​of the force feedback sensor and the cumulative pulse count of the position encoder, calculate the peak value of the force feedback growth rate, define the peak value statistics as the start-up error warning threshold, obtain the frequency domain amplitude distribution of the raw values ​​of the force feedback sensor under multiple speeds, filter the dynamic error characteristic frequency band, calculate the energy value of the characteristic frequency band, obtain the steady-state average force feedback value, and construct a force feedback characteristic-correction torque mapping table between the characteristic frequency band energy value and the steady-state average force feedback value. The specific steps of S1 are: to collect the original values ​​of the force feedback sensor and the cumulative number of pulses of the position encoder when the motor is running at multiple preset constant speeds; Perform a fast Fourier transform on the raw values ​​of the force feedback sensor at each constant rotation speed to generate a frequency domain amplitude distribution; By comparing the frequency domain amplitude distributions of multiple sets of constant speeds, the frequency band corresponding to the common resonance peak that exists at different speeds is selected and defined as the dynamic error characteristic frequency band. Based on the formula: ; Calculate the energy within the dynamic error characteristic frequency band, and generate the characteristic frequency band energy value, where... The energy value of the characteristic frequency band. For frequency The amplitude at that point, and These are the start and cutoff frequencies of the dynamic error characteristic frequency band, respectively. Calculate the time average of the raw force feedback sensor values ​​under constant speed for each group to obtain the steady-state average force feedback value; A force feedback characteristic-correction torque mapping table is established using the characteristic frequency band energy value as the key and the steady-state average force feedback value as the value.

[0020] The method for determining the start-up error warning threshold is as follows: during the motor debugging phase, perform multiple start-up processes from a standstill to a preset speed, and record the original values ​​of the force feedback sensor during each start-up process; Calculate the first-order time difference of the raw values ​​of the force feedback sensor during each startup process to obtain the instantaneous force feedback growth rate sequence; Extract the maximum value from each instantaneous force feedback growth rate sequence to form a peak sample set of force feedback growth rates; The mean and standard deviation of the peak sample set of the force feedback growth rate were calculated. Based on the formula: ; The sum of the mean and three times the standard deviation is defined as the threshold for triggering error warning. in, This represents the threshold for initiating error warning. The mean of the sample set representing the peak growth rate of the force feedback. The standard deviation of the peak sample set representing the force feedback growth rate.

[0021] The force feedback feature-correction torque mapping table includes feature energy bonds and friction torque values.

[0022] In a debugging scenario for the joint motor (model: JZR12-50-24A, rated torque 10Nm) of a six-axis robot arm used in precision assembly, the raw values ​​of the force feedback sensor (mounted at the connection between the motor output shaft and the load, model: FN-200, range ±20Nm, sampling frequency 1000Hz) and the cumulative pulse count of the position encoder (24-bit absolute encoder) were collected when the motor ran at multiple preset constant speeds. The set constant speed groups were 100r / min, 200r / min, 300r / min, 400r / min and 500r / min, and each speed was run for 10 seconds, collecting 10,000 force feedback data points and the corresponding encoder pulse counts.

[0023] A Fast Fourier Transform (FFT) was performed on the raw numerical sequences of 10,000 force feedback sensors acquired at a constant rotational speed (e.g., 300 r / min) for each group. This transforms the force feedback signals in the time domain. Convert to amplitude spectrum in the frequency domain This demonstrates the vibration intensity at different frequency components. For example, at a rotational speed of 300 r / min, the collected force feedback sequence is fed into the computing unit to obtain amplitude distribution data with a frequency range from 0 Hz to 500 Hz (half of the sampling frequency according to the Nyquist theorem).

[0024] The frequency domain amplitude distribution diagrams generated at five speeds (100 r / min, 200 r / min, 300 r / min, 400 r / min, and 500 r / min) were compared. A significant amplitude peak was found in the 155 Hz to 165 Hz frequency band across all speeds. This peak is caused by electromagnetic harmonics within the motor or the inherent resonance characteristics of the mechanical structure. For example, the peak value in this frequency band is 0.8 V at 100 r / min, 1.5 V at 300 r / min, and 2.3 V at 500 r / min. Since the resonance phenomenon in this frequency band is stable under different operating conditions, the 155 Hz to 165 Hz frequency band was selected and defined as the dynamic error characteristic frequency band.

[0025] Based on formula Calculate the energy within the dynamic error characteristic frequency band. Here... The characteristic frequency band energy value is the sum of the squares of the amplitudes of all frequency points within the band, reflecting the total vibration energy of that frequency band. It is a frequency point The amplitude at that point is obtained by fast Fourier transform; and These are the start and cutoff frequencies of the dynamic error characteristic frequency band, respectively, which are 155Hz and 165Hz in this embodiment; (Summarization symbol) This represents the summation of the squared amplitude terms corresponding to each integer frequency point from the starting frequency to the cutoff frequency in Hertz; absolute value symbol. Indicates taking the modulus of the amplitude; squaring operation This is for calculating energy. The entire formula's operational logic is as follows: by accumulating the energy (amplitude squared) of each frequency component within a specific frequency band, the total vibration intensity caused by the dynamic error source is quantified. The advantage of this formula is that by calculating the energy sum of a specific frequency band, it can quantify the dynamic error caused by resonance phenomena and eliminate noise interference outside the set frequency range. Taking a rotational speed of 300 r / min as an example, let... , Extract the amplitude of this frequency band from the FFT result, for example: , , ..., , ..., Substitute into the formula to calculate: The results show that at a rotational speed of 300 r / min, the energy value of the dynamic error characteristic frequency band is 8.5 V².

[0026] Calculate the time average of the raw force feedback sensor values ​​for each group at a constant rotational speed (taking 300 r / min as an example). Sum the 10,000 force feedback values ​​(unit: Nm) collected at this rotational speed, and then divide by the total number of data points, 10,000. For example, the calculated average value is 1.2 Nm, which is the steady-state average force feedback value at 300 r / min. This value mainly reflects the average frictional torque at this rotational speed.

[0027] Repeat the above energy calculation and average force feedback value calculation process to obtain data pairs at all preset speeds. Use the characteristic frequency band energy value as the key and the steady-state operating average force feedback value as the value to establish a force feedback characteristic-correction torque mapping table.

[0028] Table 1 Force Feedback Characteristics-Correction Torque Mapping Table

[0029] As shown in Table 1, this table establishes the correspondence between dynamic vibration energy and average friction torque under different steady-state operating conditions, where the characteristic frequency band energy value is used as the key and the steady-state operating average force feedback value (i.e. friction torque value) is used as the value.

[0030] The method for determining the start-up error warning threshold is as follows: During the motor debugging phase, perform 100 start-up processes from a standstill to a preset speed of 300 r / min on the same robot joint motor, and record the raw values ​​of the force feedback sensor during each start-up (0 to 500 ms). The sampling frequency is 1000 Hz, and 500 data points are obtained for each start-up. Calculate the first-order time difference of the raw values ​​of the force feedback sensor during each start-up to obtain the instantaneous force feedback growth rate sequence. For example, for a force feedback sequence of [0, 0.05, 0.12, 0.2, ...] (Nm) for a certain start-up, its growth rate sequence is [0.05, 0.07, 0.08, ...] (Nm / ms). Extract the maximum value in each instantaneous force feedback growth rate sequence to form a peak force feedback growth rate sample set containing 100 samples, for example, [8.5, 8.2, 8.8, 7.9, 8.6, ...] (Nm / s). Calculate the mean and standard deviation of this sample set. After calculation, the mean is obtained. Standard deviation Based on the formula The sum of the mean and three standard deviations is defined as the threshold for triggering error warning. This represents the start-up error warning threshold, used to determine whether there are abnormal impacts during the motor start-up process; This represents the average level of the peak force feedback growth rate during multiple normal startup processes; This represents the dispersion of these peaks. Using "mean plus three standard deviations" is based on the 3-sigma principle in statistics; for normally distributed data, approximately 99.7% of the values ​​will fall within this range. Within this range, events outside this range can be considered low-probability anomalies. The advantage of this formula is that it uses statistical methods to define a clear and reliable quantitative boundary for the initiation of "anomalies," avoiding the subjectivity and uncertainty of manually setting thresholds. Substituting numerical values ​​into the calculation: To verify the rationality of this threshold, an additional 1000 startup tests were performed. The test results showed that the peak force feedback growth rate was below 9.0 Nm / s in 998 startups, and only 2 (one of which was due to a slight manual interruption) exceeded the threshold, resulting in a false alarm rate of approximately 0.1%. These test results indicate that this threshold can be used to distinguish between normal and abnormal startups, with a false alarm rate of 0.1% across 1000 tests. Ultimately, the startup error warning threshold was determined to be 9.0 Nm / s.

[0031] Please see Figure 1 and Figure 3 S2: Increase the drive current command value, obtain the original value of the force feedback sensor, calculate the force feedback growth rate. If the force feedback growth rate exceeds the start-up error warning threshold, fit the original value of the force feedback sensor, calculate the predicted torque value, and generate the forward-looking error correction torque. The specific steps of S2 are as follows: During the motor start-up phase, the raw values ​​of the force feedback sensor are acquired in real time and its first-order time difference is calculated to obtain the force feedback growth rate. If the force feedback growth rate is greater than the error warning threshold, then error correction is triggered. The original numerical sequence of the force feedback sensor within a preset time window before the trigger correction time is extracted, and polynomial fitting is performed using the least squares method to generate polynomial fitting coefficients. Based on the polynomial fitting coefficients, the force feedback value is predicted after one prediction time step, and the predicted torque value is obtained. The predicted torque value is used as a compensation amount to generate a forward-looking error correction torque.

[0032] The forward error correction torque is calculated based on the polynomial fitting coefficients, the time series reference point, and the prediction time step.

[0033] The order of the polynomial fitting is preset to third order based on the motor load characteristics; The predicted torque value is calculated as follows: the current time plus the predicted time step is used as the independent variable, and then substituted into the polynomial function determined by the polynomial fitting coefficients. The calculated function value is the predicted torque value. The predicted time step is determined based on the response delay of the motor control system.

[0034] During the motor start-up phase, such as acceleration from a standstill to the target speed of 400 r / min, the raw values ​​are acquired in real time by a force feedback sensor, and the first-order time difference is calculated at 1 ms intervals to obtain the real-time force feedback growth rate. For example, if the force feedback value is 0.4 Nm at t=50 ms and 0.4092 Nm at t=51 ms, then the force feedback growth rate at this moment is... .

[0035] Determine if the real-time calculated force feedback growth rate exceeds the trigger error warning threshold determined in S1. Compare the current value of 9.2 Nm / s with the threshold of 9.0 Nm / s. Since 9.2 > 9.0, the determination result is yes, and the error correction process is immediately triggered.

[0036] The calibration is triggered at t=51ms. At this point, 50 raw force feedback sensor values ​​from a pre-defined time window preceding this time, from t=1ms to t=50ms, are extracted to form a numerical sequence. This time window is set to 50ms to ensure sufficient data for fitting while avoiding the introduction of outdated dynamic information. A third-order polynomial fitting is performed on these 50 points using the least squares method to generate the polynomial. The goal of the fitting process is to find a set of coefficients (a, b, c, d) such that... The value of is the smallest, among which It is the sampling time point. These correspond to the actual force feedback values. Based on these 50 data points [(1,0.002),(2,0.005),…,(50,0.4)] (time unit ms, force feedback unit Nm), the calculated polynomial fitting coefficients are as follows: , , , .

[0037] Based on this set of polynomial fitting coefficients, the force feedback value is predicted one prediction time step later. This prediction time step is determined based on the response delay of the motor control system. In this embodiment, testing shows that the system's delay from receiving a command to executing an action is approximately 10ms; therefore, the prediction time step is set to 10ms. The current time... Add prediction time step As the independent variable, that is The values ​​are then substituted into the polynomial function determined by the polynomial fitting coefficients for calculation. Predicted torque value. The calculated function value of 3.0037 Nm is the predicted torque value.

[0038] The predicted torque value of 3.0037 Nm is used as a compensation amount to generate a forward-looking error correction torque. This forward-looking error correction torque, valued at 3.0037 Nm, will be used in subsequent steps to correct drive commands and proactively offset impending torque surges. The forward-looking error correction torque is calculated based on polynomial fitting coefficients, a time series reference point, and a predicted time step. The order of the polynomial fitting is predetermined based on the typical nonlinear and variable load characteristics of motors in robotic arm applications. After comparing the fitting effects of second-, third-, and fourth-order polynomials, a third-order polynomial is chosen to achieve a specific balance between prediction accuracy and computational complexity. The predicted torque value is calculated by adding the predicted time step to the current moment as the independent variable and substituting it into the polynomial function determined by the polynomial fitting coefficients. The calculated function value is the predicted torque value. The predicted time step is determined based on the response delay of the motor control system.

[0039] The order of the polynomial fitting was set to third order, based on the following: In experimental comparisons of this motor model, the average prediction error of second-order fitting was 0.2 Nm, while the average prediction error of third-order fitting decreased to 0.17 Nm, a reduction of 15%. The average prediction error of fourth-order fitting was 0.16 Nm, showing no significant improvement compared to third-order. Regarding computation time, using the same processor, the single-run computation times for second-order, third-order, and fourth-order fitting were 0.2 ms, 0.35 ms, and 0.5 ms, respectively. The computation time of third-order fitting meets the system's real-time requirements and is 30% shorter than that of fourth-order fitting. Considering both the improvement in prediction accuracy and computational resource consumption, third-order fitting was chosen.

[0040] Please see Figure 1 and Figure 4 S3: Collect raw values ​​from the force feedback sensor, perform fast Fourier transform, extract frequency domain data of the dynamic error characteristic frequency band, calculate the energy value of the characteristic frequency band, and call the force feedback characteristic-correction torque mapping table to calculate the dynamic correction torque by linear interpolation; The specific steps of S3 are as follows: During the steady-state operation of the motor, the raw values ​​of the force feedback sensor are periodically collected. Perform a fast Fourier transform on the latest acquired raw values ​​from the force feedback sensor to extract the frequency domain data corresponding to the dynamic error characteristic frequency band; Calculate the current energy value of the characteristic frequency band based on the frequency domain data corresponding to the dynamic error characteristic frequency band; Call the force feedback feature-correction torque mapping table, find the two feature energy bonds adjacent to the current feature frequency band energy value and their corresponding friction torque values, and obtain the nearest energy interpolation point and the stress torque reference value; Based on the formula: ; Perform linear interpolation calculations to generate dynamic correction torque; in, To dynamically correct the torque, This represents the current characteristic frequency band energy value. and These are two mapping points formed by the nearest energy interpolation point and the reference value of the stress moment.

[0041] The dynamic correction torque is calculated based on the nearest energy interpolation point, the reference value of the stress torque, and the interpolation weighting coefficient.

[0042] Once the motor enters a steady-state operating phase, for example, when running at a stable speed of 400 r / min, the control system continuously acquires raw numerical sequences of 512 points from the force feedback sensor at a period of 200 ms. This acquisition period and data length are chosen to ensure timely capture of system dynamic changes and to provide sufficient resolution for the FFT.

[0043] A Fast Fourier Transform was performed on the newly acquired raw values ​​(unit: Nm) of the 512 force feedback sensors to obtain the updated frequency domain amplitude distribution. From this frequency domain data, the frequency domain data corresponding to the dynamic error characteristic frequency band (155Hz to 165Hz) defined in S1 was extracted.

[0044] Based on the frequency domain data of the extracted dynamic error characteristic frequency band (155Hz-165Hz), the same energy calculation formula as in S1 is used. Calculate the energy value of the current characteristic frequency band. Assuming the result of this calculation... .

[0045] The force feedback feature-correction torque mapping table established in S1 (see Table 1) is invoked to find the two feature energy bonds adjacent to the current feature frequency band energy value of 13.5V² and their corresponding friction torque values. The table shows that 13.5V² lies between 12.3V² and 15.8V². Therefore, the two mapping points formed by the obtained neighboring energy interpolation points and the stress torque reference value are... and .

[0046] Based on formula Perform linear interpolation calculations to generate dynamic correction torque; here... The dynamic correction torque to be determined; This represents the energy value of the currently calculated characteristic frequency band. and They were retrieved from the mapping table, and Two adjacent data points. Fractions in the formula. It calculates the relative position of the current energy value within two reference energy bond intervals, using it as an interpolation weight. The overall calculation logic of the formula is: [The following text appears to be incomplete and requires further context: "with reference torque..."] Based on this, add a torque increment that is proportional to the position of the energy value within the interval. The advantage of this formula lies in its ability to estimate correction values ​​for arbitrary intermediate states using simple linear interpolation and discrete calibration data points, achieving real-time and efficient computation without requiring calibration for every possible energy value. The obtained values ​​are then substituted into the formula for calculation: ; ; ; The calculated result of 1.6028571 Nm is the generated dynamic correction torque, which reflects the magnitude of the compensation torque required to offset dynamic errors (such as friction fluctuations) under the current steady-state operating conditions. The dynamic correction torque is calculated based on the nearest energy interpolation points (12.3V², 15.8V²), the stress torque reference values ​​(1.5Nm, 1.8Nm), and the interpolation weighting coefficient of 0.342857 calculated using the current energy value of 13.5V².

[0047] Please see Figure 1 and Figure 5 S4: Superimpose the forward error correction torque and the driving torque, add the dynamic correction torque to the feedforward control channel, and adjust the driving current command value.

[0048] The specific steps of S4 are: obtain the original driving torque output by the controller; The forward error correction torque is superimposed on the original driving torque to generate the start-up compensation torque; The dynamic correction torque is used as a feedforward component and added to the compensation torque during the start-up phase and the feedback control torque calculated through the position loop and velocity loop to obtain the final composite torque. Based on the ratio of the final resultant torque to the motor torque constant, the adjusted drive current value is calculated, and the adjusted drive current command value is generated.

[0049] The drive current command value includes the current corresponding to the original position command, the compensation current during the startup phase, and the compensation current during steady-state operation.

[0050] In the drive current command value, the current corresponding to the original position command is calculated and generated by the position controller based on the position deviation; The start-up phase compensation current is derived from the forward error correction torque divided by the motor torque constant, and the start-up phase compensation current is superimposed on the original position command current only during the motor start-up phase and when the start-up error warning threshold condition is met. The steady-state operating compensation current is derived from the dynamic correction torque divided by the motor torque constant, and during the steady-state operation of the motor, the steady-state operating compensation current is added to the feedforward control channel.

[0051] In the motor control process, the initial driving torque is first obtained by the PID algorithms of the position and speed loops within the controller, which calculate the output torque based on the deviation between the position command and the actual position. Assume that at a certain moment during the startup phase described in S2, the initial driving torque output by the controller is 2.5 Nm.

[0052] The forward-looking error correction torque of 3.0037 Nm calculated in S2 is superimposed on the original driving torque of 2.5 Nm to generate the start-up compensation torque. Start-up compensation torque = 2.5 Nm + 3.0037 Nm = 5.5037 Nm. This operation is only performed when the start-up error warning condition in S2 is triggered.

[0053] The dynamic correction torque of 1.6028571 Nm calculated in S3 is used as a feedforward component. This is added to the torque command (including the compensation torque during the startup phase) and the feedback control torque calculated via the position and speed loops to obtain the final composite torque. During steady-state operation, the forward error correction torque is 0. At this point, the final composite torque = original driving torque (feedback control torque) + dynamic correction torque (feedforward component). For example, if the original driving torque is 4.0 Nm during steady-state operation, then the final composite torque = 4.0 Nm + 1.6028571 Nm = 5.6028571 Nm.

[0054] The adjusted drive current value is calculated based on the ratio of the final combined torque to the motor torque constant. The torque constant of the motor (model JZR12-50-24A) used in this embodiment is... After consulting the manual and conducting experimental calibration, the value is determined to be 0.8 Nm / A. In the steady-state operation example, the adjusted drive current value = final resultant torque / =5.6028571Nm / 0.8Nm / A=7.003571A. This current value of 7.003571A will be sent to the motor driver as the final command.

[0055] The drive current command value includes the current corresponding to the original position command, the compensation current during startup, and the compensation current during steady-state operation. The current corresponding to the original position command is calculated by the position controller based on the position deviation. In the aforementioned steady-state operation example, the original drive torque is 4.0 Nm, and the corresponding current is 4.0 Nm / 0.8 Nm / A = 5.0 A. The compensation current during startup is derived by dividing the forward error correction torque by the motor torque constant. It is added to the current corresponding to the original position command only during the motor startup phase and when the startup error warning threshold condition is met. In the example of S2, this current is 3.0037 Nm / 0.8 Nm / A = 3.754625 A. The steady-state operating compensation current is derived by dividing the dynamic correction torque by the motor torque constant. During the steady-state operation of the motor, this compensation current is added to the feedforward control channel. In the example of S3, this current is 1.6028571Nm / 0.8Nm / A = 2.003571A. Therefore, at the moment the starting abnormality is triggered, the total drive current command is approximately 2.5Nm / 0.8Nm / A + 3.754625A = 3.125A + 3.754625A = 6.879625A. At a certain moment during steady-state operation, the total drive current command is 5.0A + 2.003571A = 7.003571A.

[0056] A motor drive accuracy correction system based on force feedback data, wherein the system is used to execute the aforementioned motor drive accuracy correction method based on force feedback data, the system comprising: The mapping table construction module obtains the original values ​​of the force feedback sensor and the cumulative pulse count of the position encoder, calculates the peak value of the force feedback growth rate to define the start-up error warning threshold, and, based on the frequency domain analysis of the original values ​​of the force feedback sensor under multiple speeds, filters the dynamic error characteristic frequency bands, calculates the energy value of the characteristic frequency bands, and, combined with the steady-state average force feedback value, constructs a force feedback characteristic-correction torque mapping table. The start-up correction module increases the drive current command value when the motor starts, calculates the force feedback growth rate in real time, and when the force feedback growth rate exceeds the start-up error warning threshold, it fits the original value of the force feedback sensor to calculate the predicted torque value and generates a forward-looking error correction torque. The dynamic correction module collects raw values ​​from the force feedback sensor during steady-state operation of the motor, extracts data of the dynamic error characteristic frequency band through fast Fourier transform, calculates the real-time characteristic frequency band energy value, and calls the force feedback characteristic-correction torque mapping table to generate the dynamic correction torque through linear interpolation. The instruction adjustment module superimposes the forward error correction torque with the original drive torque and adds the dynamic correction torque to the feedforward control channel. Finally, it combines the feedback control torque to adjust and generate the final drive current instruction value.

[0057] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the scope of protection defined by the claims of the present invention.

Claims

1. A method for correcting the accuracy of motor drives based on force feedback data, characterized in that, Includes the following steps: S1: Obtain the raw values ​​of the force feedback sensor and the cumulative pulse count of the position encoder, calculate the peak value of the force feedback growth rate, define the peak value statistics as the start-up error warning threshold, obtain the frequency domain amplitude distribution of the raw values ​​of the force feedback sensor under multiple speeds, filter the dynamic error characteristic frequency band, calculate the energy value of the characteristic frequency band, obtain the steady-state average force feedback value, and construct a force feedback characteristic-correction torque mapping table between the characteristic frequency band energy value and the steady-state average force feedback value. S2: Increase the drive current command value, obtain the original value of the force feedback sensor, calculate the force feedback growth rate, and if the force feedback growth rate exceeds the start-up error warning threshold, fit the original value of the force feedback sensor, calculate the predicted torque value, and generate a forward-looking error correction torque. S3: Collect the raw values ​​of the force feedback sensor, perform a fast Fourier transform, extract the frequency domain data of the dynamic error characteristic frequency band, calculate the energy value of the characteristic frequency band, and call the force feedback characteristic-correction torque mapping table to calculate the dynamic correction torque by linear interpolation; S4: Superimpose the forward error correction torque and the driving torque, add the dynamic correction torque to the feedforward control channel, and adjust the driving current command value.

2. The motor drive accuracy correction method based on force feedback data according to claim 1, characterized in that, The force feedback feature-correction torque mapping table includes feature energy bonds and friction torque values. The forward-looking error correction torque is calculated based on polynomial fitting coefficients, time series reference points, and predicted time steps. The dynamic correction torque is calculated based on neighboring energy interpolation points, stress torque reference values, and interpolation weighting coefficients. The driving current command value includes the current corresponding to the original position command, the start-up stage compensation current, and the steady-state operation compensation current.

3. The motor drive accuracy correction method based on force feedback data according to claim 1, characterized in that, The specific steps of S1 are: collecting the original values ​​of the force feedback sensor and the cumulative pulse count of the position encoder when the motor is running at multiple preset constant speeds; A fast Fourier transform is performed on the raw values ​​of the force feedback sensor under each set of constant rotational speeds to generate a frequency domain amplitude distribution; By comparing the frequency domain amplitude distributions under multiple sets of constant rotation speeds, the frequency band corresponding to the common resonance peak that exists under different rotation speeds is selected and defined as the dynamic error characteristic frequency band. Based on the formula: ; Calculate the energy within the dynamic error characteristic frequency band and generate the characteristic frequency band energy value, wherein, The energy value of the characteristic frequency band. For frequency The amplitude at that point, and These are the start frequency and cutoff frequency of the dynamic error characteristic frequency band, respectively; Calculate the time average of the raw values ​​of the force feedback sensor under each group of constant speed to obtain the steady-state average force feedback value; Using the energy value of the characteristic frequency band as the key and the average force feedback value of steady-state operation as the value, a force feedback characteristic-correction torque mapping table is established.

4. The motor drive accuracy correction method based on force feedback data according to claim 1, characterized in that, The specific steps of S2 are as follows: during the motor start-up phase, the original value of the force feedback sensor is acquired in real time and its first-order time difference is calculated to obtain the force feedback growth rate. Determine whether the force feedback growth rate is greater than the start-up error warning threshold. If the determination result is yes, then trigger error correction. The original numerical sequence of the force feedback sensor within a preset time window before the trigger correction time is extracted, and polynomial fitting is performed using the least squares method to generate polynomial fitting coefficients. Based on the polynomial fitting coefficients, the force feedback value is predicted after one prediction time step, and the predicted torque value is obtained. The predicted torque value is used as a compensation amount to generate a forward error correction torque.

5. The motor drive accuracy correction method based on force feedback data according to claim 1, characterized in that, The specific steps of S3 are as follows: during the steady-state operation phase of the motor, the raw values ​​of the force feedback sensor are periodically collected; Perform a fast Fourier transform on the newly acquired raw values ​​of the force feedback sensor to extract the frequency domain data corresponding to the dynamic error characteristic frequency band; Calculate the current energy value of the characteristic frequency band based on the frequency domain data corresponding to the dynamic error characteristic frequency band; Call the force feedback feature-correction torque mapping table, find two feature energy keys adjacent to the current feature frequency band energy value and their corresponding friction torque values, and obtain the nearest energy interpolation point and the stress torque reference value; Based on the formula: ; Perform linear interpolation calculations to generate dynamic correction torque; in, For the dynamic correction torque, The current characteristic frequency band energy value, and These are the two mapping points formed by the nearest energy interpolation point and the reference value of the stress moment.

6. The motor drive accuracy correction method based on force feedback data according to claim 1, characterized in that, The specific steps of S4 are: obtaining the original driving torque output by the controller; The forward error correction torque is superimposed on the original driving torque to generate the start-up phase compensation torque; The dynamic correction torque is used as a feedforward component and added to the compensation torque during the startup phase and the feedback control torque calculated via the position loop and velocity loop to obtain the final composite torque. Based on the ratio of the final combined torque to the motor torque constant, the adjusted drive current value is calculated, and the adjusted drive current command value is generated.

7. The motor drive accuracy correction method based on force feedback data according to claim 4, characterized in that, The order of the polynomial fitting is preset to third order based on the motor load characteristics; The predicted torque value is calculated as follows: the current time plus the predicted time step is used as the independent variable, and then substituted into the polynomial function determined by the polynomial fitting coefficients. The calculated function value is the predicted torque value. The predicted time step is determined based on the response delay of the motor control system.

8. The motor drive accuracy correction method based on force feedback data according to claim 3, characterized in that, The method for determining the start-up error warning threshold is as follows: during the motor debugging phase, perform multiple start-up processes from a standstill to a preset speed, and record the original values ​​of the force feedback sensor during each start-up process; Calculate the first-order time difference of the original values ​​of the force feedback sensor during each startup process to obtain the instantaneous force feedback growth rate sequence; Extract the maximum value from each instantaneous force feedback growth rate sequence to form a force feedback growth rate peak sample set; Calculate the mean and standard deviation of the peak sample set of the force feedback growth rate; Based on the formula: ; The sum of the mean and three times the standard deviation is defined as the trigger error warning threshold; in, This represents the startup error warning threshold. The mean of the peak sample set representing the force feedback growth rate is... The standard deviation of the peak sample set representing the force feedback growth rate is given.

9. The motor drive accuracy correction method based on force feedback data according to claim 2, characterized in that, In the drive current command value, the current corresponding to the original position command is calculated and generated by the position controller based on the position deviation; The start-up phase compensation current is derived from the forward error correction torque divided by the motor torque constant, and the start-up phase compensation current is superimposed on the original position command current only during the motor start-up phase and when the start-up error warning threshold condition is met. The steady-state operating compensation current is derived from the dynamic correction torque divided by the motor torque constant, and during the steady-state operation phase of the motor, the steady-state operating compensation current is added to the feedforward control channel.

10. A motor drive accuracy correction system based on force feedback data, characterized in that, The system is used to implement the motor drive accuracy correction method based on force feedback data as described in any one of claims 1-9, and the system includes: The mapping table construction module obtains the original values ​​of the force feedback sensor and the cumulative pulse count of the position encoder, calculates the peak value of the force feedback growth rate to define the start-up error warning threshold, and, based on the frequency domain analysis of the original values ​​of the force feedback sensor under multiple speeds, filters the dynamic error characteristic frequency bands, calculates the energy value of the characteristic frequency bands, and, combined with the steady-state average force feedback value, constructs a force feedback characteristic-correction torque mapping table. The start-up correction module increases the drive current command value when the motor starts, calculates the force feedback growth rate in real time, and when the force feedback growth rate exceeds the start-up error warning threshold, it fits the original value of the force feedback sensor to calculate the predicted torque value and generates a forward-looking error correction torque. The dynamic correction module collects the raw values ​​of the force feedback sensor when the motor is running in steady state, extracts the data of the dynamic error characteristic frequency band through fast Fourier transform, calculates the real-time characteristic frequency band energy value, and calls the force feedback feature-correction torque mapping table to generate the dynamic correction torque through linear interpolation. The instruction adjustment module superimposes the forward error correction torque with the original driving torque and adds the dynamic correction torque to the feedforward control channel. Finally, it combines the feedback control torque to adjust and generate the final driving current instruction value.

Citation Information

Cited By

  • Full-automatic immunofluorescence slide staining system and method

    CN121656584A