A method and system for controlling the twisting of an anti-kink multi-strand enameled wire

Abstract

The application discloses a kind of anti-self-twist multi-strand enameled stranded wire twisting control method and system, it is related to cable processing technical field, method includes: obtaining the real-time tension data of each strand and preset target tension data and pre-processing, constructs error sequence, forms multi-channel time series dataset;Quantify the coupling strength between each strand, calculate the enhancement base weight of each strand;Calculate the dynamic importance coefficient of each strand, correct the enhancement base weight of each strand, obtain the final dynamic synchronous compensation weight of each strand;Obtain the independent PID control amount of each strand and combine the final dynamic synchronous compensation weight of each strand, calculate the total control amount of each strand;The total control amount of each strand is converted into drive signal to adjust the pay-off tension of each strand.The application can realize the adaptive collaborative control of multi-strand tension, solve the finished product self-twist problem caused by uneven tension of each strand during the twisting process of multi-strand enameled wire.

Description

Technical Field

[0001] This invention relates to the field of cable processing technology, and in particular to a method and system for controlling the stranding of anti-self-twisting multi-strand enameled stranded wire. Background Technology

[0002] In multi-strand stranding processes, the consistency of tension among each strand is a core factor determining the quality and reliability of the finished product. Uneven tension distribution can easily generate unbalanced torque at the stranding points, leading to self-twisting of the finished cable and adversely affecting its mechanical, electrical, and long-term stability. To control the tension of individual strands, industrial applications often employ independent control algorithms based on PID (Proportion-Integral-Differential) control. In this method, each strand typically undergoes closed-loop adjustment based solely on the error between its actual tension and the target setpoint, with each controller operating independently.

[0003] However, in actual stranding processes, the multiple strands are closely physically connected and not isolated. The strands influence each other through mechanical transmission mechanisms, a shared spindle, friction at the stranding point, and vibration transmission, forming a complex dynamic coupling relationship. Traditional independent PID control strategies do not consider this dynamic coupling relationship, resulting in limited control effectiveness under complex operating conditions. Specifically, when the coupling relationship of the multi-strand stranding control system of cable processing equipment shifts due to speed fluctuations, changes in lubrication conditions, or external disturbances, the independent adjustments of each controller may conflict, easily adversely affecting the control accuracy of multi-strand tension. This leads to a significant decrease in overall control effectiveness and stability, making it difficult to achieve synchronous and coordinated control of multi-strand tension. Summary of the Invention

[0004] To overcome the limitations of existing independent PID control strategies in dealing with dynamic coupling and nonlinear time-varying characteristics, and to achieve adaptive and coordinated control of multi-strand tension, thereby solving the problem of self-torsion of the finished product caused by uneven tension of each strand during the stranding process of multi-strand enameled wire, this invention provides a stranding control method and system for anti-self-torsion multi-strand enameled wire, the technical solution of which is as follows: In a first aspect, the present invention provides a stranding control method for anti-self-torsion multi-strand enameled stranded wire, comprising the following steps: synchronously acquiring real-time tension data and preset target tension data of each strand and preprocessing them to construct an error sequence of each strand, forming a multi-channel time-series dataset; quantifying the coupling strength between each strand based on the multi-channel time-series dataset and the prediction residual method; defining a group anomaly coefficient and fusing the coupling strength to calculate the reinforcing base weight of each strand; calculating the dynamic importance coefficient of each strand based on the error sequence of each strand, and then correcting the reinforcing base weight of each strand to obtain the final dynamic synchronization compensation weight of each strand; obtaining the independent PID control quantity of each strand based on a PID control algorithm, and calculating the total control quantity of each strand in combination with the final dynamic synchronization compensation weight of each strand; and converting the total control quantity of each strand into a drive signal to adjust the unwinding tension of each strand. The process involves setting the length and step size of a sliding window, using the error sequence in the multi-channel time series dataset as the sample window for the latest sampling time, employing a first-order linear regression model as the prediction model, and using recursive least squares to update the model parameters based on the sample window. Two arbitrary lines are selected as target one and target two. Based on the error sequence of target two and the corresponding prediction model, the prediction error sequence of target one is obtained. The absolute difference between the corresponding data values ​​in the error sequence and the prediction error sequence is used as the prediction residual of target one relative to target two at the corresponding sampling time.

[0005] Preferably, tension data of each strand is synchronously acquired in real time using multiple tension sensors at a fixed sampling frequency to obtain the original tension sequence of each strand. The preset target tension data of each strand is read from the control system of the equipment, and each strand is numbered with the number corresponding to the position of the physical laying unit. For the original tension sequence of each strand, linear interpolation is used to fill the short-term missing data in each sequence, and outlier detection and replacement are performed based on the Laida rule. After filtering out high-frequency noise in each sequence, Z-Score normalization is finally performed to obtain the preprocessed tension data sequence of each strand. The difference between the target tension data of each strand and the data value in the corresponding tension data sequence at each sampling time is calculated to obtain the error sequence of each strand. The set of error sequences of each strand is used as a multi-channel time series dataset.

[0006] Preferably, a scale parameter is set, and the ratio between the prediction residual and the scale parameter at a certain sampling time is back-mapped using the natural exponential function to obtain the prediction capability value of target one relative to target two at that sampling time; similarly, the prediction capability value of target two relative to target one at that sampling time is calculated, and the average of the two prediction capability values ​​is used as the coupling strength between target one and target two at that sampling time, thereby obtaining the coupling strength between any two lines at the current sampling time.

[0007] Preferably, the coupling strength between each line at the same sampling time is normalized, and the normalized value is used as the basic dynamic synchronization compensation weight between each line at the same sampling time. The predictive ability value of target one relative to the other lines at the same sampling time is extracted and the mean is calculated. The difference between 1 and the mean is used as the group anomaly coefficient corresponding to target one at the same sampling time. The ratio between the group anomaly coefficient corresponding to target one and the group anomaly coefficient corresponding to target two at the same sampling time is used as the anomaly correction factor of target one relative to target two at the corresponding sampling time. The basic dynamic synchronization compensation weight between each line at the corresponding sampling time is weighted based on the anomaly correction factor, and the weighted basic dynamic synchronization compensation weight is normalized to obtain the enhanced basic weight between each line at the same sampling time.

[0008] Preferably, the error sequence of each stock line is extracted, and the maximum absolute value of the data values ​​in the error sequence of each stock line at the same sampling time is obtained. The ratio between the absolute value of the data value of a stock line at a certain sampling time and the corresponding maximum value is used as the error amplitude factor of the stock line at that sampling time. The absolute difference between the data value at a certain sampling time and the data value at the previous sampling time in the error sequence is used as the error change rate at that sampling time. The maximum value of the error change rate of each stock line at the same sampling time is obtained. The ratio between the error change rate of a stock line at a certain sampling time and the corresponding maximum value is used as the change rate factor of the stock line at that sampling time. The variance of each stock line is calculated based on the sample window at a certain sampling time, and the maximum value is obtained. The ratio between the variance of a stock line at that sampling time and the corresponding maximum value is used as the volatility factor of the stock line at that sampling time.

[0009] Preferably, weight coefficients for three factors are set, and the sum of the products of the three factors and the corresponding weight coefficients of a certain line at the same sampling time is taken as the dynamic importance coefficient of the line at the corresponding sampling time. The dynamic importance coefficients of target one and target two at the same sampling time are extracted, and the ratio between the dynamic importance coefficient of target one and the dynamic importance coefficient of target two is taken as the importance correction coefficient of target one relative to target two. The enhanced basic weight of target one at the corresponding sampling time is extracted, and the product between the enhanced basic weight and the importance correction coefficient is taken as the final dynamic synchronization compensation weight of target one relative to target two at the corresponding sampling time. Similarly, the final dynamic synchronization compensation weights between each line at the current sampling time are obtained.

[0010] Preferably, the classic PID control algorithm is used to obtain the independent PID control quantity of each line, calculate the difference between the data value in the error sequence of target one and the data value in the error sequence of target two at the same sampling time, extract the final dynamic synchronization compensation weight of target one relative to target two at the corresponding sampling time, and take the product between the difference and the corresponding final dynamic synchronization compensation weight as the compensation component of target two to target one at the corresponding sampling time. Similarly, the compensation components of the other lines to target one at the same sampling time are obtained.

[0011] Preferably, a fixed synchronization gain coefficient is set, and the product between the cumulative value of the compensation component of each line to target one at the sampling time and the fixed synchronization gain coefficient is used as the synchronization compensation control quantity of target one at the sampling time. The sum of the independent PID control quantity and the synchronization compensation control quantity of target one at the same sampling time is used as the total control quantity of target one at the corresponding sampling time. Similarly, the total control quantity of each line at the current sampling time is obtained.

[0012] Preferably, the total control quantity of each strand at the current sampling time is limited to ensure that the total control quantity of each strand is always within the physical allowable range of the equipment control system. The total control quantity of each strand after the limit is converted into an analog voltage signal or pulse signal by the digital-to-analog converter module in the equipment and output to the equipment control system. In this way, the speed of the wire feeding motor or the excitation current of the brake in the equipment is adjusted in real time according to the input drive signal, thereby changing the wire feeding tension of each strand.

[0013] Secondly, the present invention provides a stranding control system for anti-self-twisting multi-strand enameled stranded wire, used to implement the above-mentioned stranding control method for anti-self-twisting multi-strand enameled stranded wire, comprising: a processor, a memory, a communication interface, a data acquisition device, and a wire-laying execution mechanism. The processor stores computer program instructions for implementing the above-mentioned stranding control method for anti-self-twisting multi-strand enameled stranded wire, and the communication interface is communicatively connected to the data acquisition device and the wire-laying execution mechanism.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: Compared to traditional independent PID control or fixed-weight collaborative control, this invention upgrades from independent PID control to intelligent collaborative control by quantifying the coupling strength between individual strands and calculating the final dynamic synchronous compensation weights of each strand. It can adaptively quantify the complex time-varying coupling relationships between individual strands and sense the abnormal states and urgency levels of each strand in real time, thereby dynamically adjusting the collaborative weights. This allows the control algorithm to prioritize the allocation of control resources to the most tightly coupled strands and those most in need of intervention. While maintaining basic single-strand tracking performance, it efficiently and intelligently achieves global synchronous adjustment of multi-strand tension, effectively suppressing cable self-torsion caused by uneven tension from a physical perspective. This significantly improves the finished cable quality, the stability of the stranding process, and the robustness of the control system. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the implementation of a method for controlling the stranding of anti-self-torsion multi-strand enameled stranded wire according to an embodiment of the present invention. Figure 2 This is a structural block diagram of a stranding control system for anti-self-twisting multi-strand enameled stranded wire according to an embodiment of the present invention. Detailed Implementation

[0016] The technical features of the present invention will be further described in detail below with reference to the accompanying drawings so that those skilled in the art can understand them.

[0017] A method for controlling the stranding of anti-self-torsion multi-strand enameled stranded wire, the implementation process of which is as follows: Figure 1 As shown, the specific implementation steps are as follows: Step S1: Synchronously collect real-time tension data and preset target tension data of each strand and preprocess them to construct the error sequence of each strand and form a multi-channel time series dataset.

[0018] This step involves acquiring real-time data using synchronous tension sensors corresponding to multiple strands of wire, and then standardizing the data to provide a high-quality data foundation for subsequent dynamic feature extraction and collaborative control.

[0019] Specifically, multiple tension sensors synchronously collect tension data of each strand in real time at a fixed sampling frequency to obtain the original tension sequence of each strand. The preset target tension data of each strand is read from the control system of the equipment. Each strand is numbered, and the number corresponds to the position of the physical laying unit. For the original tension sequence of each strand, linear interpolation is used to fill the short-term missing data in each sequence, and outlier detection and replacement are performed based on the Laida rule. After filtering out high-frequency noise in each sequence, Z-score normalization is performed to obtain the preprocessed tension data sequence of each strand. The difference between the target tension data of each strand and the data value in the corresponding tension data sequence at each sampling time is calculated to obtain the error sequence of each strand. The set of error sequences of each strand is used as a multi-channel time series dataset, and the error sequence of each strand is updated in real time based on the sampling data.

[0020] The sampling frequency of the tension sensor can be set to 100Hz. The number of each strand corresponds one-to-one with the position of the physical wire laying unit. The preset target tension data of each strand is a process preset constant, which is dynamically adjusted according to the production process requirements and the cable type in the actual application scenario. For example, the filter used for noise reduction can be a first-order low-pass filter with a cutoff frequency of 10 Hz.

[0021] Step S2: Based on the multi-channel time series dataset and the prediction residual, quantify the coupling strength between each line.

[0022] This step aims to mine the dynamic coupling relationship between the individual lines from the multi-channel time series dataset, providing a data foundation for the subsequent calculation of the basic weights for collaborative control.

[0023] During the stranding process of multi-strand wires, the coupling between the strands originates from complex factors such as mechanical transmission, friction at the stranding point, and vibration transmission, and has time-varying and nonlinear characteristics. Therefore, the mutual prediction residual method can be used to measure the coupling strength between the strands. If one strand can accurately predict the tension error of another strand, it indicates that there is a strong coupling relationship between the two, and vice versa.

[0024] Specifically, the length and step size of the sliding window are set, and the error sequence in the multi-channel time series dataset is extracted in real time as a sample window. A first-order linear regression model is used as the prediction model. As the sample window is updated, the linear regression coefficients of the prediction model are updated in real time using the recursive least squares method. Two lines with arbitrary numbers are selected as target one and target two. A prediction model of target one relative to target two is constructed. Based on the error sequence of target two and the corresponding prediction model, the prediction error sequence of target one is obtained. The absolute difference between the error sequence of target one and the corresponding data values ​​in the prediction error sequence is used as the prediction residual of target one relative to target two at each sampling time. A scale parameter is set, and the ratio between the prediction residual and the scale parameter at a certain sampling time is back-mapped using the natural exponential function to obtain the prediction capability value of target one relative to target two at that sampling time. Similarly, the prediction capability value of target two relative to target one at that sampling time is calculated. The mean of the two prediction capability values ​​is used as the coupling strength between target one and target two at that sampling time, and thus the coupling strength between any two lines at each sampling time is obtained.

[0025] The length of the sliding window is set according to the cable production rate, and can be set to 100 sampling points. The sliding step size is set to 1. The recursive least squares method can track the parameter changes of the prediction model with extremely low computational cost, thus efficiently adapting to the time-varying nature of the coupling relationship. The larger the value of the prediction residual, the smaller the corresponding prediction capability value. When the value of the prediction residual is zero, the corresponding prediction capability value is 1. The scale parameter is used to adjust the decay rate of the prediction capability value. It can be adjusted according to the noise level of the data in the actual application scenario. The higher the noise level and the worse the data quality, the smaller the value of the scale parameter needs to be adjusted. For example, the value of the scale parameter can be set to 5% of the full scale of the tension sensor. The role of using the natural exponential function is to provide a smooth transition, avoid abrupt changes in the calculation results, and limit the prediction capability value to a certain level. Within the range; Since the coupling relationship between each line should be mutual, the reference weights need to be symmetrical to avoid control oscillations. Therefore, the average value of the bidirectional prediction capability between two lines is taken as the coupling strength between the two lines. The smaller the prediction residual value, the larger the corresponding coupling strength value, indicating that the coupling strength between the two lines is stronger and the coupling relationship is tighter. Conversely, the larger the prediction residual value, the weaker the coupling relationship between the two lines.

[0026] It should be noted that the present invention uses a first-order linear regression model as the prediction benchmark, not assuming that the coupling relationship between the strands is strictly linear. On the contrary, when the actual coupling exhibits nonlinear characteristics, the prediction residual of the linear model will naturally increase. This characteristic is cleverly utilized by the present invention to transform the limitation of the linear model into the advantage of perceiving nonlinearity. The increase in the value of the prediction residual means that "under the linear benchmark corresponding to the current sample window, the coupling relationship between the two strands weakens", thereby automatically reducing the coupling weight. When the nonlinear coupling relationship between the strands of the equipment is strengthened, the control system of the device weakens the coordination strength, thereby effectively avoiding control oscillations caused by forced coordination.

[0027] Step S3: Define the group anomaly coefficient and fuse the coupling strength to calculate the enhanced basic weight of each line.

[0028] Specifically, the coupling strength between each line at the same sampling time is normalized, and the normalized value is used as the basic dynamic synchronization compensation weight between each line at that sampling time. The predictive ability value of Target 1 relative to the other lines at the same sampling time is extracted and the mean is calculated. The difference between 1 and the mean is used as the group anomaly coefficient corresponding to Target 1 at that sampling time. The ratio between the group anomaly coefficient corresponding to Target 1 and the group anomaly coefficient corresponding to Target 2 at the same sampling time is used as the anomaly correction factor of Target 1 relative to Target 2 at the corresponding sampling time. The basic dynamic synchronization compensation weight between each line at the corresponding sampling time is weighted based on the anomaly correction factor. The weighted basic dynamic synchronization compensation weight is summed and normalized to obtain the enhanced basic weight between each line at that sampling time.

[0029] The basic dynamic synchronization compensation weight is calculated by summation and normalization. This weight represents the inherent coupling relationship between the two lines. For example, the basic dynamic synchronization compensation weight between the i-th and j-th lines at sampling time t is... Its expression is as follows: In the formula, This represents the coupling strength between the i-th and j-th lines at sampling time t. This represents the cumulative value of the coupling strength between the i-th line and the other lines at sampling time t; at the same sampling time, satisfying... .

[0030] The physical meaning of the group anomaly coefficient is as follows: If any line is selected as the target line, and none of the lines other than the target line can accurately predict the current state of the target line, it indicates that the behavior of the target line deviates from the group and is in an abnormal state, requiring higher attention to be paid to the abnormal line in collaborative control; when the target line is perfectly predicted, the prediction capability value approaches 1, and the corresponding group anomaly coefficient value approaches 0, indicating that the target line is not abnormal at the corresponding sampling time; when the target line cannot be predicted at all, the prediction capability value approaches 0, and the corresponding group anomaly coefficient value approaches 1, indicating that the target line is highly abnormal at the corresponding sampling time; the anomaly correction factor is used to reflect the comparison of the degree of anomaly of the target line with the other lines at the same sampling time.

[0031] For example, the enhanced basic weight between the i-th line and the j-th line at sampling time t is: Its expression is as follows: In the formula, This represents the basic dynamic synchronization compensation weight between the i-th and j-th lines at sampling time t. This represents the anomaly correction factor of the i-th line relative to the j-th line at sampling time t. This represents the accumulated value of the weighted basic dynamic synchronous compensation weights of the i-th line relative to the other lines at sampling time t; enhanced basic weights. The larger the value, the greater the influence of the j-th line on the i-th line at sampling time t, and the higher the degree to which the j-th line needs to be referenced in collaborative control.

[0032] Step S4: Based on the error sequence of each stock line, calculate the dynamic importance coefficient of each stock line, and then correct the enhanced basic weight of each stock line to obtain the final dynamic synchronous compensation weight of each stock line.

[0033] This step aims to use the real-time tension error data at the current sampling moment to calculate the dynamic importance coefficient of each line, thereby correcting the enhanced basic weights obtained in step S3. This allows the collaborative control algorithm to prioritize the lines that currently require the most intervention, namely lines with large errors, rapid changes, and drastic fluctuations, thereby eliminating disturbances in the equipment control system more quickly and improving control accuracy and response speed.

[0034] Specifically, the error sequence of each stock line is extracted, and the maximum absolute value of the data values ​​in the error sequence of each stock line at the same sampling time is obtained. The ratio between the absolute value of the data value of a stock line at a certain sampling time and the corresponding maximum value is used as the error amplitude factor of the stock line at that sampling time. The absolute difference between the data value at a certain sampling time and the data value at the previous sampling time in the error sequence is used as the error change rate at that sampling time. The maximum value of the error change rate of each stock line at the same sampling time is obtained, and the ratio between the error change rate of a stock line at a certain sampling time and the corresponding maximum value is used as the change rate factor of the stock line at that sampling time. The variance of each stock line is calculated based on the sample window at a certain sampling time, and the maximum value is obtained. The ratio between the variance of a stock line at that sampling time and the corresponding maximum value is used as the volatility factor of the stock line at that sampling time. The weight coefficients of the three factors are set, and the sum of the products of the three factors of a stock line at the same sampling time and the corresponding weight coefficients is used as the dynamic importance coefficient of the stock line at the corresponding sampling time.

[0035] In addition, the dynamic importance coefficients of target one and target two at the same sampling time are extracted. The ratio between the dynamic importance coefficient of target one and the dynamic importance coefficient of target two is used as the importance correction coefficient of target one relative to target two. The enhanced basic weight of target one at the corresponding sampling time is extracted. The product between the enhanced basic weight and the importance correction coefficient is used as the final dynamic synchronization compensation weight of target one relative to target two at the corresponding sampling time. Similarly, the final dynamic synchronization compensation weight between each line at the current sampling time is obtained.

[0036] Among them, the error amplitude factor reflects the relative degree of deviation of the target line from the target tension value at the same sampling time; the rate of change factor reflects the relative trend and speed of the target line tension fluctuation at the same sampling time; and the volatility factor reflects the relative severity of the target line data fluctuation within the sample window at the same sampling time. The sum of the weight coefficients of the three factors is 1. The values ​​of the weight coefficients are manually set according to the requirements of adjustment accuracy and adjustment stability in the actual application scenario. When the requirement for adjustment accuracy is high, the weight coefficient of the error amplitude factor needs to be increased; when the requirement for adjustment stability is high, the weight coefficients of the rate of change factor and the volatility factor need to be increased. For example, the weight coefficients of the three factors can be set to 0.4, 0.3, and 0.3 respectively to comprehensively consider the current deviation, the trend of change, and the historical volatility, and to fully measure the real-time importance of each line.

[0037] Step S5: Based on the PID control algorithm, obtain the independent PID control quantity of each stock, and calculate the total control quantity of each stock by combining the final dynamic synchronization compensation weight of each stock.

[0038] This step aims to perform a weighted summation of the tension errors of each strand based on the final dynamic synchronous compensation weights obtained in step S4. The weights are automatically adjusted according to the real-time coupling strength, so that the control quantity of each strand not only considers its own error but also the error difference with the other strands. This results in the calculation of an additional collaborative control component, i.e., a compensation component, for each strand, which is then combined with the independent PID control quantity to obtain the total control quantity for each strand. The core objective of this step is to drive the tension errors of each strand to be consistent, achieve synchronous collaborative control of multi-strand tension, eliminate the unbalanced torque caused by tension differences at the twisting point, and suppress cable self-torsion from a physical source.

[0039] Specifically, the classic PID control algorithm is used to obtain the independent PID control quantity of each line. The difference between the data value in the error sequence of target one and the data value in the error sequence of target two at the same sampling time is calculated. The final dynamic synchronization compensation weight of target one relative to target two at the corresponding sampling time is extracted. The product of the difference and the corresponding final dynamic synchronization compensation weight is used as the compensation component of target two to target one at the corresponding sampling time. Similarly, the compensation components of the other lines to target one at the same sampling time are obtained.

[0040] The classic PID control algorithm is an existing technology and will not be elaborated on in this step. The independent PID control quantity is used to ensure the basic tracking performance of the single line to the target tension value.

[0041] In addition, a fixed synchronization gain coefficient is set, and the product between the cumulative value of each line's compensation component to target one at the sampling time and the fixed synchronization gain coefficient is used as the synchronization compensation control quantity of target one at the sampling time. The sum of the independent PID control quantity and the synchronization compensation control quantity of target one at the same sampling time is used as the total control quantity of target one at the corresponding sampling time. Similarly, the total control quantity of each line at the current sampling time is obtained.

[0042] Wherein, the synchronous compensation control quantity of the i-th line at the t-th sampling time is The calculation formula is as follows: In the formula, This represents the fixed synchronization gain coefficient. This represents the final dynamic synchronization compensation weight of the i-th line relative to the j-th line at sampling time t. This represents the data value at sampling time t within the error sequence of the i-th line. This represents the data value at sampling time t within the error sequence of the j-th line.

[0043] also, The overall strength of the synchronization compensation control quantity is adjusted, and its value is set based on the physical allowable range of the equipment control system. For example, it can be set to 0.3. The smaller the physical allowable upper limit of the equipment control system, the better. It is the difference in instantaneous tension error between the i-th and j-th strands at the t-th sampling time, used to reflect the degree and direction of inconsistency between the tension of the two strands and the target tension value; The value of determines the extent to which the control amount of the i-th line should reference the state of the j-th line. The larger the value, the higher the proportion of contribution from the j-th line in the synchronous compensation control quantity of the i-th line.

[0044] When the tension error of the i-th strand is greater than that of the j-th strand, i.e. the i-th strand is looser than the j-th strand, the difference in tension error is positive, and the corresponding compensation component is positive. This compensation component is superimposed on the independent PID control quantity, which drives the wire feeding actuator of the i-th strand to increase the tension, making the i-th strand taut, and at the same time making the tension of the i-th strand move closer to the tension level of the j-th strand.

[0045] Step S6: Convert the total control quantity of each strand into a drive signal to adjust the tension of each strand.

[0046] Specifically, the total control quantity of each strand at the current sampling time is limited to ensure that the total control quantity of each strand is always within the physical allowable range of the equipment control system. The total control quantity of each strand after being limited is converted into an analog voltage signal or pulse signal by the digital-to-analog converter module in the equipment and output to the equipment control system. In this way, the speed of the wire feeding motor or the excitation current of the brake in the equipment is adjusted in real time according to the input drive signal, thereby changing the wire feeding tension of each strand.

[0047] If the upper limit of the control voltage of the equipment control system is 10V, then the corresponding drive signal amplitude limit is 10V.

[0048] This invention also discloses a stranding control system for anti-self-twist multi-strand enameled wire, used to implement the above-mentioned stranding control method for anti-self-twist multi-strand enameled wire. The system structure is as follows: Figure 2 As shown, it includes: a processor, a memory, a communication interface, a data acquisition device, and a wire-laying actuator. The processor stores computer program instructions for implementing the above-mentioned method for controlling the stranding of anti-self-twisting multi-strand enameled wire. The communication interface is communicatively connected to the data acquisition device and the wire-laying actuator.

[0049] The data acquisition device is a tension sensor, and the wire feeding actuator is a servo driver or magnetic powder brake inside the equipment.

[0050] The embodiments included in this invention are descriptions of preferred embodiments of the invention and are not limited to the precise structures already described above and shown in the accompanying drawings. Various modifications and changes can be made without departing from the scope of protection. All variations and improvements made by those skilled in the art to the technical solutions of this invention without departing from the design concept of this invention should fall within the scope of protection of this invention.

Claims

1. A method for controlling the stranding of anti-self-twisting multi-strand enameled stranded wire, characterized in that: Real-time tension data and preset target tension data of each strand are collected synchronously and preprocessed to construct error sequences for each strand, forming a multi-channel time-series dataset. Based on the multi-channel time-series dataset and the prediction residual method, the coupling strength between each strand is quantified, including: setting a scale parameter, using the natural exponential function to inversely map the ratio between the prediction residual and the scale parameter at a certain sampling time to obtain the prediction capability value of target one relative to target two at that sampling time; similarly, the prediction capability value of target two relative to target one at that sampling time is calculated, and the average of the two prediction capability values ​​is taken as the coupling strength between target one and target two at that sampling time, thereby obtaining the coupling strength between any two strands at the current sampling time. Define the group anomaly coefficient and fuse the coupling strength to calculate the enhanced basic weights of each stock line. This includes: normalizing the coupling strength between each stock line at the same sampling time, using the normalized value as the basic dynamic synchronization compensation weight between each stock line at that sampling time; extracting the prediction ability value of target one relative to the other lines at the same sampling time and calculating the mean; using the difference between 1 and the mean as the group anomaly coefficient corresponding to target one at that sampling time; using the ratio between the group anomaly coefficient corresponding to target one and the group anomaly coefficient corresponding to target two at the same sampling time as the anomaly correction factor of target one relative to target two at the corresponding sampling time; weighting the basic dynamic synchronization compensation weights between each stock line at the corresponding sampling time based on the anomaly correction factor; normalizing the weighted basic dynamic synchronization compensation weights to obtain the enhanced basic weights between each stock line at that sampling time. Based on the error sequence of each strand, the dynamic importance coefficient of each strand is calculated, and then the enhanced basic weight of each strand is corrected to obtain the final dynamic synchronous compensation weight of each strand. Based on the PID control algorithm, the independent PID control quantity of each strand is obtained, and combined with the final dynamic synchronous compensation weight of each strand, the total control quantity of each strand is calculated. The total control quantity of each strand is converted into a drive signal to adjust the tension of each strand. The process involves setting the length and step size of a sliding window, using the error sequence in the multi-channel time series dataset as the sample window for the latest sampling time, employing a first-order linear regression model as the prediction model, and using recursive least squares to update the model parameters based on the sample window. Two arbitrary lines are selected as target one and target two. Based on the error sequence of target two and the corresponding prediction model, the prediction error sequence of target one is obtained. The absolute difference between the corresponding data values ​​in the error sequence and the prediction error sequence is used as the prediction residual of target one relative to target two at the corresponding sampling time.

2. The method for controlling the stranding of anti-self-twisting multi-strand enameled stranded wire according to claim 1, characterized in that, The construction of error sequences for each strand to form a multi-channel time-series dataset includes: synchronously acquiring tension data of each strand in real time using multiple tension sensors at a fixed sampling frequency to obtain the original tension sequence of each strand; reading the preset target tension data of each strand from the control system of the equipment; numbering each strand and corresponding the number to the physical laying unit position; for the original tension sequence of each strand, using linear interpolation to fill the short-term missing data in each sequence, and performing outlier detection and replacement based on Laida's rule; filtering out high-frequency noise in each sequence; and finally performing Z-Score normalization to obtain the preprocessed tension data sequence of each strand; calculating the difference between the target tension data of each strand and the data value in the corresponding tension data sequence at each sampling time to obtain the error sequence of each strand; and using the set of error sequences of each strand as a multi-channel time-series dataset.

3. The method for controlling the stranding of anti-self-twisting multi-strand enameled stranded wire according to claim 1, characterized in that, The calculation of the dynamic importance coefficient of each stock line based on the error sequence of each stock line includes: extracting the error sequence of each stock line; obtaining the maximum absolute value of the data values ​​in the error sequence of each stock line at the same sampling time; taking the ratio between the absolute value of the data value of a stock line at a certain sampling time and the corresponding maximum value as the error amplitude factor of the stock line at that sampling time; taking the absolute difference between the data value at a certain sampling time and the data value at the previous sampling time in the error sequence as the error change rate at that sampling time; obtaining the maximum value of the error change rate of each stock line at the same sampling time; taking the ratio between the error change rate of a stock line at a certain sampling time and the corresponding maximum value as the change rate factor of the stock line at that sampling time; calculating the variance of each stock line based on the sample window at a certain sampling time and obtaining the maximum value among them; taking the ratio between the variance of a stock line at that sampling time and the corresponding maximum value as the volatility factor of the stock line at that sampling time.

4. The method for controlling the stranding of anti-self-twisting multi-strand enameled stranded wire according to claim 3, characterized in that, The calculation of the dynamic importance coefficient of each stock line, and the subsequent correction of the enhanced basic weight of each stock line to obtain the final dynamic synchronization compensation weight of each stock line, includes: setting the weight coefficients of three factors; taking the sum of the products of the three factors of a stock line and the corresponding weight coefficients at the same sampling time as the dynamic importance coefficient of that stock line at the corresponding sampling time; extracting the dynamic importance coefficients of target one and target two at the same sampling time; taking the ratio between the dynamic importance coefficient of target one and the dynamic importance coefficient of target two as the importance correction coefficient of target one relative to target two; extracting the enhanced basic weight of target one at the corresponding sampling time; taking the product between the enhanced basic weight and the importance correction coefficient as the final dynamic synchronization compensation weight of target one relative to target two at the corresponding sampling time; and similarly obtaining the final dynamic synchronization compensation weight between each stock line at the current sampling time.

5. The method for controlling the stranding of anti-self-twisting multi-strand enameled stranded wire according to any one of claims 1 to 4, characterized in that, The calculation of the total control quantity of each line includes: obtaining the independent PID control quantity of each line using the classic PID control algorithm; calculating the difference between the data value in the error sequence of target one and the data value in the error sequence of target two at the same sampling time; extracting the final dynamic synchronization compensation weight of target one relative to target two at the corresponding sampling time; and taking the product between the difference and the corresponding final dynamic synchronization compensation weight as the compensation component of target two to target one at the corresponding sampling time. Similarly, the compensation components of the remaining lines to target one at the same sampling time are obtained.

6. The method for controlling the stranding of anti-self-twisting multi-strand enameled stranded wire according to claim 5, characterized in that, The calculation of the total control quantity of each line also includes: setting a fixed synchronization gain coefficient, using the product between the accumulated value of each line's compensation component to target one at the sampling time and the fixed synchronization gain coefficient as the synchronization compensation control quantity of target one at the sampling time, and using the sum of the independent PID control quantity and the synchronization compensation control quantity of target one at the same sampling time as the total control quantity of target one at the corresponding sampling time. Similarly, the total control quantity of each line at the current sampling time is obtained.

7. The method for controlling the stranding of anti-self-twisting multi-strand enameled stranded wire according to claim 6, characterized in that, The step of converting the total control quantity of each strand into a drive signal to adjust the tension of each strand includes: limiting the total control quantity of each strand at the current sampling time to ensure that the total control quantity of each strand is always within the physical allowable range of the equipment control system; converting the limited total control quantity of each strand into an analog voltage signal or pulse signal through the digital-to-analog converter module in the equipment and outputting it to the equipment control system; thereby adjusting the speed of the wire-laying motor or the excitation current of the brake in the equipment in real time according to the input drive signal to change the tension of each strand.

8. A stranding control system for anti-self-twisting multi-strand enameled stranded wire, characterized in that, include: The device includes a processor, a memory, a communication interface, a data acquisition device, and a wire-laying actuator. The processor stores computer program instructions for implementing the stranding control method for anti-self-twisting multi-strand enameled stranded wire as described in any one of claims 1 to 7. The communication interface is communicatively connected to the data acquisition device and the wire-laying actuator.

Images