PID (Proportion Integration Differentiation) adjusting method based on position data of slide block automatically wound by wire winding machine

Abstract

The invention relates to the technical field of PID (Proportion Integration Differentiation) control, and discloses a PID adjusting method based on position data of a sliding block automatically wound by a wire winding machine, which comprises the following steps of: firstly, acquiring a main shaft angular velocity and two-way transverse movement time of the wire winding machine, and calculating a transverse movement and rotation ratio; then, rational approximation is carried out on the ratio to obtain a target rational number, and a discrete lag amount is obtained by combining a dual-frequency related parameter and a sampling period; further collecting a sliding block position reference signal and an actual signal to generate a position error sequence, and executing normalized self-correlation calculation under the discrete lag quantity to obtain a characteristic quantity; performing statistical analysis on the characteristic quantity to form an entry threshold and an exit threshold; dynamically scheduling proportional, integral and differential adjustment parameters and integral states according to a comparison result of the characteristic quantity and a threshold; and finally, the controlled quantity is calculated and output by utilizing the scheduled proportion, integral and differential adjusting parameters and the integral state so as to drive the position of the sliding block of the wire winding machine to realize accurate and automatic adjustment.

Description

Technical Field

[0001] This invention relates to the field of PID control technology, and more specifically, to a PID adjustment method based on slider position data of an automatically wound wire winding machine. Background Technology

[0002] Automatic wire winding machines require the guide wire to be evenly arranged along the surface of the drum during operation. The slider is responsible for driving the guide wire laterally, and its speed must maintain a certain ratio with the spindle's rotational angular velocity to maintain the winding pitch. As the winding process progresses, the wire layers stack up, and the drum diameter gradually increases. To maintain a constant winding angle or linear speed, the control system continuously adjusts the ratio of the spindle speed to the slider's lateral movement speed. Theoretically, this ratio should change continuously, but in actual equipment, when the ratio is close to an integer or a low fraction, the guide wire will repeatedly pass through the previous trajectory (i.e., a banded locking zone), causing localized overlap or gaps, thus forming a regular banded pattern on the wrapping surface. This phenomenon not only affects the uniformity of coil density but may also cause unstable wire tension, leading to a decrease in winding quality.

[0003] Currently, the slider position control of winding machines mostly adopts fixed-parameter PID regulation, which has the advantages of simple structure and high real-time performance. However, under the conditions of changes in coil diameter and slider stroke period, the dynamic parameters of the system are not constant. Especially in the strip locking range, the slider position error will form repetitive fluctuations in a specific period. The integral element in conventional PID regulation will treat this low-frequency error as a steady-state deviation and continuously compensate for it, resulting in integral accumulation. Integral accumulation not only fails to reduce the error, but also produces overshoot when the error reverses, causing rhythmic oscillations and deviations in the slider movement. After multiple layers of winding, this error is amplified layer by layer, forming obvious pattern deviation bands.

[0004] To ensure uniform distribution of the winding layers, the control system needs to be able to identify the periodicity of the slider error within the strip-locked range and automatically adjust the PID parameters when the error cycle is detected to suppress rather than amplify the periodic deviation under this special operating condition. Traditional solutions lack this targeted identification and dynamic adjustment mechanism, often relying on manual experience to reset the gain or interrupting the winding process for recalibration, which is both time-consuming and unstable. As automatic winding machines develop towards higher precision and automation, how to detect the periodic changes of the slider error in real time in the control loop and adjust the PID parameters accordingly has become a key issue in improving the quality and stability of winding. Summary of the Invention

[0005] This invention provides a PID adjustment method based on slider position data of an automatic winding machine, which solves the technical problem of how to achieve real-time identification and dynamic adjustment of the periodic repetition and band-like locking phenomenon of slider error in the slider position control process of the winding machine, so as to eliminate the problem of periodic error solidification caused by fixed parameters.

[0006] This invention provides a PID control method based on slider position data from an automatic winding machine, comprising: The dual-frequency correlation parameters of the winding machine are obtained at a fixed sampling period, and the transverse displacement and rotation ratio are calculated. The target rational number is obtained by rationally approximating the transverse and rotation ratios, and the discrete hysteresis is calculated by combining the dual-frequency correlation parameters and the sampling period. Obtain the slider position correlation signal, generate the position error sequence, and perform normalized autocorrelation calculation on the error sequence based on the discrete hysteresis to obtain the feature quantity; Perform statistical analysis on the feature values ​​to generate entry and exit thresholds; Based on the comparison results between the characteristic quantity and the entry or exit threshold, the scheduling ratio, integral and derivative adjustment parameters and integral state are determined; Based on the proportional, integral, and derivative adjustment parameters after scheduling and the integral state, the control quantity is calculated and output to drive the adjustment of the slider position of the winding machine.

[0007] The beneficial effects of this invention are as follows: by extracting the dual-frequency parameters of the winding machine in real time under a fixed sampling period and calculating the lateral and rotation ratios, and combining rational approximation and discrete hysteresis calculation, normalized autocorrelation analysis of the slider position error sequence is achieved, thereby dynamically identifying the periodic error characteristics that occur during the winding process; furthermore, through statistical analysis and threshold comparison of characteristic quantities, the proportional, integral and derivative adjustment parameters and integral states are automatically scheduled, so that the control system can still maintain stable and high-precision control when the lateral and rotation relationships of the winding machine change, significantly reducing interlayer stacking and pattern offset problems, and improving winding uniformity, control accuracy and automatic adaptive capability. Attached Figure Description

[0008] Figure 1 This is a flowchart of the PID adjustment method based on the slider position data of the automatic winding machine according to the present invention. Detailed Implementation

[0009] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0010] like Figure 1 As shown, the PID control method based on the slider position data of the automatic winding machine includes: The dual-frequency correlation parameters of the winding machine are obtained at a fixed sampling period, and the transverse displacement and rotation ratio are calculated. The target rational number is obtained by rationally approximating the transverse and rotation ratios, and the discrete hysteresis is calculated by combining the dual-frequency correlation parameters and the sampling period. Obtain the slider position correlation signal, generate the position error sequence, and perform normalized autocorrelation calculation on the error sequence based on the discrete hysteresis to obtain the feature quantity; Perform statistical analysis on the feature values ​​to generate entry and exit thresholds; Based on the comparison results between the characteristic quantity and the entry or exit threshold, the scheduling ratio, integral and derivative adjustment parameters and integral state are determined; Based on the proportional, integral, and derivative adjustment parameters after scheduling and the integral state, the control quantity is calculated and output to drive the adjustment of the slider position of the winding machine.

[0011] In one embodiment of the present invention, the dual-frequency related parameters of the winding machine include the spindle angular velocity and the two-way traverse time.

[0012] The spindle angular velocity is the rate of change of the rotation angle of the spindle of the winding machine per unit time. The spindle angular velocity reflects the rotation speed of the spindle, and the unit of the spindle angular velocity is radians per second. The spindle angular velocity is obtained in real time through the encoder matched with the spindle. The acquisition frequency of the spindle angular velocity is consistent with the control cycle (i.e., the sampling cycle) of the winding machine to ensure the real-time performance of the spindle angular velocity parameter.

[0013] The two-way traverse time is the complete round-trip time for the slider of the winding machine to move from one end of the traverse stroke to the other end and then back to the initial end. The two-way traverse time can reflect the traverse speed of the slider, and the unit of the two-way traverse time is seconds. The two-way traverse time needs to be obtained through a slider position sensor (such as a linear encoder), specifically by recording the time difference between the slider starting from the starting point, reaching the end point, and returning to the starting point.

[0014] In one embodiment of the present invention, calculating the lateral displacement to rotation ratio includes: Calculate the product of pi and two, and use the quotient of the principal axis angular velocity and the product as the principal axis rotation frequency; The fundamental frequency of the transverse shift is obtained by dividing one by the two-way transverse shift time. Divide the spindle rotation frequency by the fundamental transverse frequency to obtain the transverse-rotation ratio.

[0015] The spindle rotation frequency is the number of times the spindle of the winding machine completes rotation per unit time, reflecting the frequency characteristics of the spindle rotation. The unit of spindle rotation frequency is Hertz. Specifically, the spindle rotation frequency is equal to the spindle angular velocity divided by (pi multiplied by two).

[0016] The fundamental frequency of transverse movement is the number of times the slider of the winding machine completes a double transverse movement per unit time, reflecting the frequency characteristics of the slider's transverse movement. The unit of the fundamental frequency of transverse movement is Hertz; specifically, the fundamental frequency of transverse movement is equal to one divided by the double transverse movement time.

[0017] The transverse-to-rotation ratio is the ratio of the spindle rotation frequency to the fundamental transverse frequency, reflecting the frequency coupling relationship between the spindle rotation and the transverse movement of the winding machine; specifically, the transverse-to-rotation ratio is equal to the spindle rotation frequency divided by the fundamental transverse frequency.

[0018] In one embodiment of the present invention, a target rational number is obtained by rationally approximating the lateral and rotation ratios, and the discrete hysteresis is calculated by combining dual-frequency correlation parameters and the sampling period, including: The target fraction is constructed based on rational number numerators and rational number denominators; wherein the rational number denominator is less than or equal to the range of the upper bound of the rational approximation denominator, the rational number numerator is an integer, and the rational number denominator is a positive integer; The absolute value of the difference between the transverse shift and the rotation ratio and the target score is used as the rational approximation error; The rational approximation error is minimized to determine the target rational number numerator and denominator corresponding to the minimized rational approximation error. The discrete hysteresis is obtained by dividing the denominator of the target rational number by the product of the quotient of the transverse fundamental frequency and the sampling period, and then rounding the product.

[0019] The target fraction is a fraction consisting of a rational numerator and a rational denominator, used to compare with the transverse and rotation ratios to calculate the rational approximation error; specifically, the target fraction is equal to the rational numerator divided by the rational denominator.

[0020] The preset upper bound of the rational approximation denominator is a pre-set positive integer constant used to limit the maximum value of the denominator of rational numbers. The preset upper bound of the rational approximation denominator is used to balance the rational approximation accuracy and computational efficiency, and to avoid the calculation time being increased due to the denominator being too large. The commonly used value range in industry is eight to twelve, which can be adjusted according to the type of wire used in the winding machine (such as copper wire or steel wire). When the wire hardness is high, a smaller value (such as eight) is used, and when the hardness is low, a larger value (such as twelve) is used.

[0021] Rational approximation error is a parameter that measures how close the target fraction is to the shift and rotation ratio. The smaller the error value, the better the target fraction approximates the shift and rotation ratio. The rational approximation error is a non-negative real number. The target rational numerator is the rational number that minimizes the rational approximation error under the constraint of the upper bound of the preset rational approximation denominator. The target rational numerator is an integer. Specifically, the first step is to iterate through all positive integer denominators that satisfy the condition that the rational denominator is less than or equal to the upper bound of the preset rational approximation denominator. The second step is to calculate the integer numerator that minimizes the absolute value of the difference between the shift and rotation ratio and the fraction corresponding to that denominator for each denominator. The third step is to select the combination with the smallest error among all combinations of (numerator and denominator). The numerator in this combination is the target rational numerator. In short, the rational approximation error is equal to the absolute value of the difference between the shift and rotation ratio and the target fraction.

[0022] The target rational denominator is the rational denominator that minimizes the rational approximation error under the constraint of the upper bound of the pre-set rational approximation denominator. The target rational denominator is a positive integer and is less than or equal to the upper bound of the pre-set rational approximation denominator. Specifically, the first step is to iterate through all positive integer denominators that satisfy the condition that the rational denominator is less than or equal to the upper bound of the pre-set rational approximation denominator. The second step is to calculate the integer numerator that minimizes the absolute value of the difference between the sum of the horizontal shift and rotation ratios (the fraction corresponding to the denominator) for each denominator. The third step is to select the combination with the smallest error among all combinations of (numerator and denominator), and the denominator in this combination is the target rational denominator.

[0023] The sampling period is a fixed signal acquisition and calculation cycle of the winding machine control system, reflecting the time interval for parameter updates. The unit of the sampling period is seconds. The commonly used industrial value range is one millisecond to ten milliseconds, which needs to be matched with the acquisition frequency of the spindle encoder and slider position sensor (e.g., when the sensor acquisition frequency is 1 kilohertz, the sampling period is one millisecond).

[0024] Discrete lag is a parameter that converts the time period corresponding to the target rational number into the number of sampling points. It is used for subsequent normalized autocorrelation calculation of the position error sequence. The value of discrete lag is a positive integer, and the unit is the number of sampling points. Specifically, discrete lag is equal to the product of the quotient of (the denominator of the target rational number divided by the basic frequency of the transverse shift) and the sampling period, and then the product is rounded to the nearest integer.

[0025] In one embodiment of the present invention, acquiring slider position-related signals and generating a position error sequence includes: At each sampling time, the slider position reference signal and the slider position actual signal are acquired respectively; Calculate the difference between the slider position reference signal and the actual slider position signal at each sampling time to form a position error sequence.

[0026] The sampling moment is a specific point in time when the winding machine control system collects and calculates signals according to a fixed sampling period. The time interval between each sampling moment and the previous sampling moment is equal to the sampling period. The number of sampling moments increases continuously with the winding process and is used to synchronously acquire the slider position related signals.

[0027] The slider position reference signal is a target position signal that the slider should reach, generated by the winding machine control system according to the preset winding process (such as the variation law of winding pitch and coil diameter). The unit of the slider position reference signal is millimeters. The signal source is the process control program of the winding machine. It needs to be generated synchronously with the sampling time to ensure that there is a corresponding target position value at each sampling time.

[0028] The actual slider position signal is the signal of the actual position of the slider acquired by the slider position detection device at each sampling moment. The unit of the actual slider position signal is millimeters. The acquisition device is a linear encoder matched with the slider (with a resolution of not less than 0.01 millimeters to ensure position detection accuracy). The acquisition time must be completely synchronized with the generation time of the slider position reference signal (same sampling moment) to avoid error calculation deviation caused by time difference.

[0029] The position error sequence is a dataset formed by arranging the slider position errors at each sampling time in chronological order. Each element in the position error sequence reflects the deviation between the actual slider position and the target position at the corresponding sampling time. The unit of the position error sequence is millimeters, and the sequence length is equal to the number of sampling times performed. Specifically, the error value at each sampling time in the position error sequence is equal to the slider position reference signal value at that sampling time minus the actual slider position signal value at that sampling time. Arranging the error values ​​at all sampling times in chronological order of sampling time forms the position error sequence.

[0030] In one embodiment of the present invention, a characteristic quantity is obtained by performing normalized autocorrelation calculation on the error sequence based on the discrete hysteresis, including: Determine the sample range with the current sampling time as the termination time, and cover the preset window length; Within this sample range, the element at each sampling time in the position error sequence is multiplied by the element corresponding to the discrete hysteresis shifted backward at that sampling time, and the numerator is obtained by summing all the multiplication results; Within this sample range, the elements at all sampling times in the position error sequence are squared and summed to obtain the denominator; Divide the numerator by the denominator to obtain the characteristic quantity.

[0031] The preset window length is a pre-defined positive integer used to select the number of position error sequence samples. The preset window length determines the amount of error data participating in the normalized autocorrelation calculation and must meet the constraint of not less than three times the discrete lag (to ensure that the sample covers at least three error periods and improves the stability of the calculation). The commonly used value range in industry is 128 to 512 sampling points, which can be adjusted according to the sampling period (take the larger value when the sampling period is small, such as 512 points when the sampling period is one millisecond).

[0032] The sample range is a set of continuous sampling times starting from the current sampling time minus the preset window length plus one sampling time and ending at the current sampling time. The number of position error sequence elements contained in this range is equal to the preset window length, and it is the only data source for normalized autocorrelation calculation. Specifically, the starting sampling time of the sample range is equal to the current sampling time minus the preset window length plus one sampling time, and the ending sampling time is the current sampling time. All sampling times from the start to the end constitute the sample range.

[0033] The numerator is the cumulative value of the correlation term in the normalized autocorrelation calculation, reflecting the similarity between the position error sequence within the sample range and its sequence after offset by discrete lag. Specifically, the first step is to select the position error sequence element corresponding to each sampling time within the sample range; the second step is to find the position error sequence element corresponding to the sampling time offset by discrete lag by sampling times for each selected element (if the offset time is outside the sample range, it is not included in the calculation); the third step is to multiply each pair of matched elements, and then add all the multiplication results together, and the sum is the numerator.

[0034] The denominator is the cumulative value of the energy term in the normalized autocorrelation calculation, which reflects the total energy of the position error sequence within the sample range. Specifically, within the sample range, the position error sequence elements corresponding to each sampling time are squared, and then all the squared results are added together to obtain the denominator.

[0035] The feature quantity is the final result of the normalized autocorrelation calculation and is used to characterize the strength of the periodic component in the position error sequence. The value of the feature quantity ranges from negative one to positive one. The closer the feature quantity is to positive one, the stronger the periodicity of the position error sequence (the higher the risk of band locking). The closer it is to zero or negative one, the weaker the periodicity. Specifically, the feature quantity is equal to the quotient of the numerator divided by the denominator.

[0036] In one embodiment of the present invention, statistical analysis is performed on the feature quantities to generate entry and exit thresholds, including: Set a smoothing coefficient, an entry threshold coefficient, and an exit threshold coefficient; where the entry threshold coefficient is greater than the exit threshold coefficient, and the smoothing coefficient ranges from zero to one. Calculate the exponential moving average of the updated feature by multiplying the historical exponential moving average by the complement of the smoothing coefficient, and then adding the feature value at the current sampling time multiplied by the smoothing coefficient. The exponential moving absolute deviation of the updated feature is calculated by multiplying the historical exponential moving absolute deviation by the complement of the smoothing coefficient, and then adding the absolute difference between the feature at the current sampling time and the exponential moving mean at the current sampling time by the smoothing coefficient. The entry threshold is obtained by adding the entry threshold coefficient to the exponential moving average at the current sampling time and the absolute deviation of the exponential moving average at the current sampling time; The exit threshold is obtained by adding the product of the exponential moving average at the current sampling time and the absolute deviation of the exponential moving average at the current sampling time to the exit threshold coefficient.

[0037] The smoothing coefficient is a parameter used to balance the influence of historical data and current data on the exponential moving statistic of a feature quantity. The smoothing coefficient ranges from zero to one, and commonly used values ​​are 0.05 to 0.2 (smaller values ​​are used when the sampling frequency is high, such as 0.05 when the sampling frequency is 1 kilohertz; larger values ​​are used when the sampling frequency is low, such as 0.2 when the sampling frequency is 100 kilohertz). The smaller the value, the greater the influence of historical data, and the more stable the statistic is.

[0038] The entry threshold coefficient is a positive real constant used to amplify the exponential sliding absolute deviation of the feature quantity and generate the entry threshold. The value of the entry threshold coefficient must be greater than the exit threshold coefficient. The commonly used value is three. This coefficient determines the sensitivity of the entry threshold. The larger the coefficient, the higher the entry threshold and the more difficult it is to trigger parameter scheduling.

[0039] The exit threshold coefficient is a positive real constant used to amplify the exponential sliding absolute deviation of the characteristic quantity and generate the exit threshold. The value of the exit threshold coefficient must be less than the entry threshold coefficient. The commonly used value is two. This coefficient determines the sensitivity of the exit threshold. The smaller the coefficient, the lower the exit threshold and the easier it is to exit parameter scheduling.

[0040] The complement of the smoothing coefficient is a parameter used to assign weights to historical data when calculating the exponential moving average. The complement of the smoothing coefficient is a non-negative real number; specifically, the complement of the smoothing coefficient is equal to one minus the smoothing coefficient.

[0041] The historical exponential moving average is the exponential moving average of the feature quantity calculated at the previous sampling time. The initial value is set to the feature quantity at the first sampling time (to ensure that the calculation starts without bias).

[0042] The exponential moving average of a feature is a statistic that reflects the long-term trend of the feature. This statistic can filter out short-term fluctuations in the feature and is updated at each sampling time. Specifically, the exponential moving average of the feature at the current sampling time is equal to the historical exponential moving average multiplied by the complement of the smoothing coefficient, plus the feature at the current sampling time multiplied by the smoothing coefficient.

[0043] The historical exponential sliding absolute deviation is the characteristic quantity exponential sliding absolute deviation calculated at the previous sampling time. The unit of the historical exponential sliding absolute deviation is the same as that of the characteristic quantity, and the initial value is set to zero (there is no historical data at the first sampling time, and the deviation accumulates from zero).

[0044] The absolute difference between the feature quantity at the current sampling time and the exponential moving average at the current sampling time is a parameter that measures the degree to which the current feature quantity deviates from its long-term trend. The larger the difference, the more significant the fluctuation of the current feature quantity. Specifically, the absolute difference between the feature quantity at the current sampling time and the exponential moving average at the current sampling time is equal to the feature quantity at the current sampling time minus the exponential moving average at the current sampling time, and then the absolute value of the difference is taken.

[0045] The exponential moving absolute deviation of a feature quantity is a statistical measure that reflects the long-term trend of the fluctuation range of the feature quantity. The larger the value, the more drastic the overall fluctuation of the feature quantity. Specifically, the exponential moving absolute deviation of the feature quantity at the current sampling time is equal to the historical exponential moving absolute deviation multiplied by the complement of the smoothing coefficient, plus the absolute difference between the feature quantity at the current sampling time and the exponential moving mean at the current sampling time multiplied by the smoothing coefficient.

[0046] The entry threshold is the threshold for determining whether a feature quantity has entered the banded locking risk zone. When the feature quantity at the current sampling time is greater than or equal to the entry threshold, PID parameter scheduling is triggered. Specifically, the entry threshold is equal to the exponential moving average of the feature quantity at the current sampling time plus the entry threshold coefficient multiplied by the absolute deviation of the exponential moving average of the feature quantity at the current sampling time.

[0047] The exit threshold is the threshold for determining whether a feature quantity has exited the risk zone of the band lock. When the feature quantity at the current sampling time is less than or equal to the exit threshold, the PID parameter scheduling is stopped and the nominal parameters are restored. Specifically, the exit threshold is equal to the exponential moving average of the feature quantity at the current sampling time plus the exit threshold coefficient multiplied by the absolute deviation of the exponential moving average of the feature quantity at the current sampling time.

[0048] The strip-shaped locking risk zone is a risky area that easily produces strip-shaped defects during the automatic winding process of a wire winding machine, formed when the frequency coupling relationship between the slider's lateral movement and the spindle's rotation enters a specific working condition. Specifically, it refers to the working condition range where, when the lateral movement to rotation ratio (the ratio of the lateral movement's basic frequency to the spindle's rotation frequency) sweeps through the neighborhood of integers or low-denominator rational numbers (such as 1 / 1, 2 / 3, etc.), the guide wire tends to repeatedly overlap at the same position on the drum surface, forming a regular strip-shaped pattern. The triggering of this risk zone is directly related to the dual-frequency parameters of the wire winding machine (spindle angular velocity, double-pass lateral movement time), which will cause periodic repetitive fluctuations in the slider's position error. If a fixed-parameter PID control is used, the integral stage will amplify such low-frequency errors, further causing uneven coil density, unstable tension, and even aggravating the wear of components such as gears and motors. It is necessary to avoid the risk by real-time monitoring of the rational approximation of the lateral movement to rotation ratio and dynamic scheduling of PID parameters (such as suppressing integral accumulation and fine-tuning the proportional / derivative coefficients).

[0049] In one embodiment of the present invention, based on the comparison result of the characteristic quantity with the entry threshold or exit threshold, the scheduling ratio, integral and derivative adjustment parameters, and integral state include: The feature values ​​are compared with the entry and exit thresholds, respectively. When the feature value is greater than or equal to the entry threshold, the scheduling intensity is obtained by dividing the difference between the feature value and the entry threshold by one. When the feature quantity is less than or equal to the exit threshold, the scheduling intensity is obtained by dividing the difference between the negative exit threshold and the feature quantity by the exit threshold. When the feature value is between the exit threshold and the entry threshold, the scheduling intensity is set to zero; The larger value between the scheduling intensity and zero is taken as the entry area scheduling share; The larger of the negative value of the scheduling intensity and zero is taken as the scheduling share of the exit zone; Set the nominal proportional control parameter, nominal integral control parameter, nominal derivative control parameter, entry zone scheduling coefficient, exit zone recovery coefficient, and integral discharge coefficient; The updated proportional adjustment parameter is obtained by multiplying the nominal proportional adjustment parameter by one and subtracting the product of the entry zone scheduling share and the entry zone scheduling coefficient, and then adding the product of the exit zone scheduling share and the exit zone restoration coefficient. The updated integral adjustment parameters are obtained by multiplying the nominal integral adjustment parameters by one and subtracting the product of the entry zone scheduling share and the entry zone scheduling coefficient, and then adding the product of the exit zone scheduling share and the exit zone restoration coefficient. The updated differential adjustment parameter is obtained by multiplying the nominal differential adjustment parameter by one, adding the product of the entry zone scheduling share and the entry zone scheduling coefficient, and then subtracting the product of the exit zone scheduling share and the exit zone restoration coefficient. The updated integral state is obtained by multiplying the integral state at the previous sampling time by one and subtracting the product of the entry area scheduling share and the integral discharge coefficient, and then adding the updated integral adjustment parameters, the position error at the current sampling time, and the sampling period.

[0050] Scheduling intensity is a parameter that quantifies the degree to which a feature value deviates from a threshold. It is used to allocate scheduling shares between the entry and exit zones, and its value ranges from -1 to +1. Specifically, in the first case, if the feature value is greater than or equal to the entry threshold, the scheduling intensity is equal to the difference between (feature value minus entry threshold) and (-minus entry threshold). In the second case, if the feature value is less than or equal to the exit threshold, the scheduling intensity is equal to the negative difference between (exit threshold minus feature value) and the exit threshold. In the third case, if the feature value is between the exit threshold and the entry threshold, the scheduling intensity is directly set to zero.

[0051] The entry zone scheduling share is a non-negative weight used to trigger PID parameter scheduling. Its value ranges from zero to one. The larger the value, the greater the parameter scheduling amplitude. Specifically, the scheduling intensity is compared with zero, and the larger value between the two is selected. This larger value is the entry zone scheduling share.

[0052] The exit zone scheduling share is a non-negative weight used to stop PID parameter scheduling and restore the nominal parameters. The value ranges from zero to one. The larger the value, the greater the parameter restoration. Specifically, the negative value of the scheduling intensity is calculated first, and then the negative value is compared with zero. The larger value is selected as the exit zone scheduling share.

[0053] The nominal proportional adjustment parameter is the default proportional coefficient of the PID controller when there is no risk of strip locking in the winding machine. It is used to quickly respond to the current position error. The commonly used value range is 0.5 to 2 (it can be adjusted according to the wire tension requirements. When the tension is sensitive, take a small value, such as 0.5; when the tension is stable, take a large value, such as 2).

[0054] The nominal integral adjustment parameter is the default integral coefficient of the PID controller when there is no risk of strip locking in the winding machine. It is used to eliminate steady-state position error. The commonly used value range is 0.1 to 0.5 (take the smaller value when the sampling period is small, such as 0.1 when the sampling period is 1 millisecond; take the larger value when the sampling period is large, such as 0.5 when the sampling period is 10 milliseconds).

[0055] The nominal differential adjustment parameter is the default differential coefficient of the PID controller when there is no risk of strip locking in the winding machine. It is used to suppress the rate of change of position error. The commonly used value range is 0.01 to 0.1 (take a larger value, such as 0.1, when the error fluctuates greatly; take a smaller value, such as 0.01, when the error is stable).

[0056] The entry zone scheduling coefficient is a non-negative real number used to control the adjustment range of PID parameters when entering a locked risk zone. The value range of the entry zone scheduling coefficient is 0.1 to 0.3. The larger the coefficient, the more the proportional and integral coefficients decrease and the more the derivative coefficient increases when entering the zone.

[0057] The exit zone recovery coefficient is a non-negative real number used to control the recovery magnitude of PID parameters to the nominal value when exiting the locked risk zone. The range of the exit zone recovery coefficient is 0.1 to 0.3. The larger the coefficient, the faster the parameters recover to the nominal value when exiting the zone.

[0058] The integral release coefficient is a non-negative real number used to control the rate of integral state release when entering a locked risk zone. The integral release coefficient ranges from 0.3 to 0.7. The larger the coefficient, the faster the integral state releases when entering the zone, and the better it can suppress integral accumulation.

[0059] The updated proportional adjustment parameter is a proportional coefficient adjusted based on the entry / exit zone scheduling share, used to adapt to the current locked risk conditions; specifically, the updated proportional adjustment parameter is equal to the nominal proportional adjustment parameter multiplied by (one minus the product of the entry zone scheduling share and the entry zone scheduling coefficient), plus (the product of the exit zone scheduling share and the exit zone recovery coefficient).

[0060] The updated integral adjustment parameters are integral coefficients adjusted based on the entry / exit zone scheduling share, used to adapt to the current locked risk conditions. The unit of the updated integral adjustment parameters is per second. Specifically, the updated integral adjustment parameters are equal to the nominal integral adjustment parameters multiplied by (one minus the product of the entry zone scheduling share and the entry zone scheduling coefficient), plus (the product of the exit zone scheduling share and the exit zone recovery coefficient).

[0061] The updated differential adjustment parameter is a differential coefficient adjusted based on the entry / exit zone scheduling share, used to adapt to the current locking risk condition. The unit of the updated differential adjustment parameter is seconds. Specifically, the updated differential adjustment parameter is equal to the nominal differential adjustment parameter multiplied by (one plus the product of the entry zone scheduling share and the entry zone scheduling coefficient), and then subtracted by (the product of the exit zone scheduling share and the exit zone recovery coefficient).

[0062] The integral state at the previous sampling time is the cumulative PID integral value calculated in the previous sampling period, with the initial value set to zero (there is no historical integral data at the first sampling time, so the accumulation starts from zero).

[0063] The updated integral state is the final cumulative value of the PID integral link at the current sampling time, which is used to calculate the control quantity in the subsequent calculation. Specifically, the updated integral state is equal to the integral state at the previous sampling time multiplied by (one minus the product of the entry zone scheduling share and the integral discharge coefficient), plus the product of (the updated integral adjustment parameter multiplied by the position error at the current sampling time multiplied by the sampling period).

[0064] In one embodiment of the present invention, based on the scheduled proportional, integral, and derivative adjustment parameters and the integral state, a control quantity is calculated and output to drive the adjustment of the winding machine slider position, including: The proportional term is obtained by multiplying the adjusted proportional parameter after scheduling with the value of the position error sequence at the current sampling time; Calculate the difference between the position error sequence at the current sampling time and the value at the previous sampling time, divide the difference by the sampling period, and then multiply it by the differential adjustment parameter after scheduling to obtain the differential term; The control quantity is obtained by adding the proportional term, the updated integral state, and the differential term.

[0065] The proportional term is the output of the proportional element in PID control, used to quickly respond to the position error at the current sampling time; specifically, the proportional term is equal to the proportional adjustment parameter after scheduling multiplied by the value of the position error sequence at the current sampling time.

[0066] The value of the position error sequence at the previous sampling time is the specific value of the position error sequence at the previous sampling time. When there is no historical data at the first sampling time, this value is set as the value of the position error sequence at the current sampling time (to avoid initial fluctuations in the differential element).

[0067] The derivative term is the output of the derivative element in PID control, used to suppress the rate of change of position error and avoid oscillation of the slider motion. Specifically, the first step is to calculate the difference between the value of the position error sequence at the current sampling time and the value at the previous sampling time; the second step is to divide the difference by the sampling period to obtain the rate of change of position error; the third step is to multiply the rate of change of position error by the scheduled derivative adjustment parameter to obtain the derivative term.

[0068] The control quantity is the final output of the PID controller, used to drive the actuator (such as a servo motor) for adjusting the position of the winding machine slider. It must meet the rated range of the actuator (e.g., when the rated voltage of the servo motor is zero to ten volts, the control quantity must be limited to this range). Specifically, the control quantity equals the proportional term plus the updated integral state, plus the derivative term.

[0069] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A PID control method based on slider position data from an automatic winding machine, characterized in that, include: The dual-frequency correlation parameters of the winding machine are obtained at a fixed sampling period, and the transverse displacement and rotation ratio are calculated. The target rational number is obtained by rationally approximating the transverse and rotation ratios, and the discrete hysteresis is calculated by combining the dual-frequency correlation parameters and the sampling period. Obtain the slider position correlation signal, generate the position error sequence, and perform normalized autocorrelation calculation on the error sequence based on the discrete hysteresis to obtain the feature quantity; Perform statistical analysis on the feature values ​​to generate entry and exit thresholds; Based on the comparison results between the characteristic quantity and the entry or exit threshold, the scheduling ratio, integral and derivative adjustment parameters and integral state are determined; Based on the proportional, integral, and derivative adjustment parameters after scheduling and the integral state, the control quantity is calculated and output to drive the adjustment of the slider position of the winding machine.

2. The PID adjustment method based on slider position data of automatic winding machine according to claim 1, characterized in that, Dual-frequency parameters of the winding machine include spindle angular velocity and double-stroke traverse time.

3. The PID adjustment method based on slider position data of automatic winding machine according to claim 2, characterized in that, Calculate the lateral displacement to rotation ratio, including: Calculate the product of pi and two, and use the quotient of the principal axis angular velocity and the product as the principal axis rotation frequency; The fundamental frequency of the transverse shift is obtained by dividing one by the two-way transverse shift time. Divide the spindle rotation frequency by the fundamental transverse frequency to obtain the transverse-rotation ratio.

4. The PID adjustment method based on slider position data of automatic winding machine according to claim 3, characterized in that, The target rational number is obtained by rationally approximating the transverse and rotation ratios. The discrete hysteresis is then calculated by combining the dual-frequency correlation parameters and the sampling period, including: The target fraction is constructed based on rational number numerators and rational number denominators; wherein the rational number denominator is less than or equal to the range of the upper bound of the rational approximation denominator, the rational number numerator is an integer, and the rational number denominator is a positive integer; The absolute value of the difference between the transverse shift and the rotation ratio and the target score is used as the rational approximation error; The rational approximation error is minimized to determine the target rational number numerator and denominator corresponding to the minimized rational approximation error. The discrete hysteresis is obtained by dividing the denominator of the target rational number by the product of the quotient of the transverse fundamental frequency and the sampling period, and then rounding the product.

5. The PID adjustment method based on slider position data of automatic winding machine according to claim 4, characterized in that, Acquire slider position-related signals and generate a position error sequence, including: At each sampling time, the slider position reference signal and the slider position actual signal are acquired respectively; Calculate the difference between the slider position reference signal and the actual slider position signal at each sampling time to form a position error sequence.

6. The PID adjustment method based on slider position data of automatic winding machine according to claim 5, characterized in that, Based on the discrete lag, normalized autocorrelation of the error sequence is calculated to obtain characteristic quantities, including: Determine the sample range with the current sampling time as the termination time, and cover the preset window length; Within this sample range, the element at each sampling time in the position error sequence is multiplied by the element corresponding to the discrete hysteresis shifted backward at that sampling time, and the numerator is obtained by summing all the multiplication results; Within this sample range, the elements at all sampling times in the position error sequence are squared and summed to obtain the denominator; Divide the numerator by the denominator to obtain the characteristic quantity.

7. The PID adjustment method based on slider position data of automatic winding machine according to claim 6, characterized in that, Perform statistical analysis on the feature quantities to generate entry and exit thresholds, including: Set a smoothing coefficient, an entry threshold coefficient, and an exit threshold coefficient; where the entry threshold coefficient is greater than the exit threshold coefficient, and the smoothing coefficient ranges from zero to one. Calculate the exponential moving average of the updated feature by multiplying the historical exponential moving average by the complement of the smoothing coefficient, and then adding the feature value at the current sampling time multiplied by the smoothing coefficient. The exponential moving absolute deviation of the updated feature is calculated by multiplying the historical exponential moving absolute deviation by the complement of the smoothing coefficient, and then adding the absolute difference between the feature at the current sampling time and the exponential moving mean at the current sampling time by the smoothing coefficient.

8. The PID adjustment method based on slider position data of automatic winding machine according to claim 7, characterized in that, Perform statistical analysis on the feature quantities to generate entry and exit thresholds, including: The entry threshold is obtained by adding the entry threshold coefficient to the exponential moving average at the current sampling time and the absolute deviation of the exponential moving average at the current sampling time; The exit threshold is obtained by adding the product of the exponential moving average at the current sampling time and the absolute deviation of the exponential moving average at the current sampling time to the exit threshold coefficient.

9. The PID adjustment method based on slider position data of automatic winding machine according to claim 8, characterized in that, Based on the comparison results between the characteristic quantity and the entry or exit threshold, the scheduling proportional, integral, and derivative adjustment parameters and the integral state are determined, including: The feature values ​​are compared with the entry and exit thresholds, respectively. When the feature value is greater than or equal to the entry threshold, the scheduling intensity is obtained by dividing the difference between the feature value and the entry threshold by one. When the feature quantity is less than or equal to the exit threshold, the scheduling intensity is obtained by dividing the difference between the negative exit threshold and the feature quantity by the exit threshold. When the feature value is between the exit threshold and the entry threshold, the scheduling intensity is set to zero; The larger value between the scheduling intensity and zero is taken as the entry area scheduling share; The larger of the negative value of the scheduling intensity and zero is taken as the scheduling share of the exit zone; Set the nominal proportional control parameter, nominal integral control parameter, nominal derivative control parameter, entry zone scheduling coefficient, exit zone recovery coefficient, and integral discharge coefficient; The updated proportional adjustment parameter is obtained by multiplying the nominal proportional adjustment parameter by one and subtracting the product of the entry zone scheduling share and the entry zone scheduling coefficient, and then adding the product of the exit zone scheduling share and the exit zone restoration coefficient. The updated integral adjustment parameters are obtained by multiplying the nominal integral adjustment parameters by one and subtracting the product of the entry zone scheduling share and the entry zone scheduling coefficient, and then adding the product of the exit zone scheduling share and the exit zone restoration coefficient. The updated differential adjustment parameter is obtained by multiplying the nominal differential adjustment parameter by one, adding the product of the entry zone scheduling share and the entry zone scheduling coefficient, and then subtracting the product of the exit zone scheduling share and the exit zone restoration coefficient. The updated integral state is obtained by multiplying the integral state at the previous sampling time by one and subtracting the product of the entry area scheduling share and the integral discharge coefficient, and then adding the updated integral adjustment parameters, the position error at the current sampling time, and the sampling period.

10. The PID adjustment method based on slider position data of automatic winding machine according to claim 9, characterized in that, Based on the adjusted proportional, integral, and derivative control parameters and integral state, the control quantity is calculated and output to drive the adjustment of the winding machine slider position, including: The proportional term is obtained by multiplying the adjusted proportional parameter after scheduling with the value of the position error sequence at the current sampling time; Calculate the difference between the position error sequence at the current sampling time and the value at the previous sampling time, divide the difference by the sampling period, and then multiply it by the differential adjustment parameter after scheduling to obtain the differential term; The control quantity is obtained by adding the proportional term, the updated integral state, and the differential term.

Images