Automatic winding and coating equipment for annular transformer magnetic core and tension control method

Abstract

The present application relates to transformer winding control technical field, specifically relates to a kind of annular transformer magnetic core automatic winding rubber coating equipment and tension control method, solve the technical problems that the inertia time-varying of phase shift and consumable consumption due to belt drive in prior art, leading to difficult to realize constant tension control in winding process.The method comprises: the torque current and mechanical angle of servo motor are collected, the torque current and mechanical angle of collection time are associated, real-time current array is constructed, and historical current array is obtained by latching real-time current array according to the rotation period of servo motor;Waveform feature analysis is carried out on historical current array, interval division result and adaptive gain coefficient are determined;According to the mechanical angle and real-time speed of current time, determine the look-up index in advance, and read the corresponding current prediction value from historical current array;According to current prediction value, interval division result and adaptive gain coefficient, generate feedforward compensation instruction.

Description

Technical Field

[0001] This invention relates to the field of transformer winding control technology, specifically to an automatic winding and coating equipment for a toroidal transformer core and a tension control method. Background Technology

[0002] Toroidal transformers are widely used in precision instruments and audio equipment. The core technology of these devices is to tightly wind enameled wire onto a toroidal core. This process is generally completed by a toroidal winding machine. The winding machine's wire storage ring carries a certain length of consumable material. During high-speed rotation, the wire is released by a slider to achieve the winding operation.

[0003] In the actual winding production of toroidal transformers, belt-driven winding machines have become one of the mainstream equipment. However, the tension control aspect of these winding machines is always a key challenge in production and a major factor restricting the quality and efficiency of the winding process. The technical difficulties related to tension control urgently need to be addressed.

[0004] Belt-driven winding machines face two main problems in tension control: the wire storage ring is not an ideal rigid connection, and the belt drive itself has elastic hysteresis and slight slippage, causing a nonlinear dynamic deviation between the actual rotation angle of the wire storage ring and the mechanical angle of the servo motor. At the same time, the consumable material in the wire storage ring is gradually consumed during winding, and the total rotational inertia of the system decreases significantly. Under the same driving torque, the system response under no-load is more sensitive than under full load. The existing technologies using fixed mechanical angle tension compensation, fixed gain proportional integral derivative (PID) or constant torque compensation cannot effectively address the above problems, resulting in poor tension control performance. Summary of the Invention

[0005] To address the technical problem in existing technologies where phase drift in belt drives and time-varying inertia from material consumption make it difficult to achieve constant tension control during winding, this invention aims to provide an automatic winding and coating device for toroidal transformer cores and a tension control method. The specific technical solution adopted is as follows: Firstly, a tension control method for automatic winding of a toroidal transformer core is provided, comprising: acquiring the torque current and mechanical angle of a servo motor; associating the torque current and mechanical angle at the acquisition time to construct a real-time current array indexed by the mechanical angle; latching the real-time current array according to the rotation cycle of the servo motor to obtain a historical current array; performing waveform feature analysis on the historical current array to determine the interval division result used to characterize the current load phase information and the adaptive gain coefficient used to characterize the current inertia state of the system; determining a look-up table index pointing to the load state at a future time based on the current mechanical angle and real-time rotation speed, and reading the current prediction value corresponding to the look-up table index from the historical current array; and generating a feedforward compensation command for superimposing on the torque setpoint of the servo driver based on the current prediction value, the interval division result, and the adaptive gain coefficient.

[0006] Based on the above technical solution, in the tension control method for automatic winding of toroidal transformer core provided by this invention, a corresponding current array is constructed and latched by associating torque current with mechanical angle. The adaptive gain coefficient for load phase and inertia adaptation is determined by combining waveform feature analysis. Then, the current prediction value is obtained by looking up the table index based on real-time mechanical angle and rotation speed. Finally, relevant parameters are integrated to generate feedforward compensation commands. This can effectively solve the phase drift caused by belt drive and the time-varying system inertia caused by winding material consumption. At the same time, it can offset the response lag of the servo system, realize precise and dynamic control of the winding tension of toroidal transformer core, and ensure that the timing and force of tension compensation can match the actual winding conditions. This effectively suppresses tension fluctuations during the winding process and ensures the stability and accuracy of winding tension control.

[0007] In conjunction with the first aspect above, in one possible implementation, the method for constructing the real-time current array specifically includes: synchronously acquiring torque current and mechanical angle in the high-frequency interrupt service routine of the servo driver; performing rounding and modulo operations on the mechanical angle to obtain the target index corresponding to the current moment in the real-time current array; the length of the real-time current array corresponding to the resolution of the mechanical angle within one rotation cycle; writing the torque current into the position corresponding to the target index in the real-time current array; if the mechanical angle interval between two adjacent interrupt service routines is greater than the resolution, using interpolation to fill the missing index positions in the real-time current array.

[0008] In conjunction with the first aspect above, in one possible implementation, the method for determining the interval division result specifically includes: based on the magnitude characteristics of the current values ​​in the historical current array, using a preset classification algorithm to divide the current values ​​in the historical current array into a high-load stretching zone dataset and a low-load shrinking zone dataset, as the interval division result.

[0009] In conjunction with the first aspect above, in one possible implementation, the method for determining the adaptive gain coefficient specifically includes: statistically analyzing the statistical variance of the concentrated current values ​​in the high-load stretching region data; determining the adaptive gain coefficient based on the statistical variance and a preset calibration constant; and the adaptive gain coefficient being negatively correlated with the statistical variance.

[0010] In conjunction with the first aspect above, in one possible implementation, the method for determining the preset calibration constant specifically includes: determining the full-load gain of the system when it is operating in a critical stable state under the standard operating condition of the storage loop being fully loaded, and obtaining the full-load variance of the stretching region dataset under the standard operating condition; and determining the calibration constant based on the full-load gain and the full-load variance.

[0011] In conjunction with the first aspect above, in one possible implementation, the method further includes: determining the base friction current based on the current values ​​in the low-load retraction zone dataset; the base friction current is used to characterize the average current required by the system to overcome mechanical friction, belt resistance, and gravity components.

[0012] In conjunction with the first aspect above, in one possible implementation, the method for generating a feedforward compensation command to be superimposed on the torque setpoint of the servo drive based on the current prediction value, the interval division result, and the adaptive gain coefficient specifically includes: removing the basic friction current from the current prediction value to obtain the tension torque increment used to characterize the tension fluctuation component; fusing the tension torque increment with the adaptive gain coefficient to obtain the adjusted compensation amplitude; and inverting the adjusted compensation amplitude to generate the feedforward compensation command.

[0013] In conjunction with the first aspect above, in one possible implementation, the method for determining the look-up table index pointing to the load state at a future time based on the current mechanical angle and real-time rotation speed specifically includes: determining the look-up compensation angle used to offset the system response lag based on the inherent total lag time constant of the system and the real-time rotation speed; adding the current mechanical angle and the look-up compensation angle and performing a modulo operation to obtain the look-up table index.

[0014] In conjunction with the first aspect mentioned above, in one possible implementation, the method further includes: prohibiting the output of feedforward compensation commands before completing the data acquisition of the first rotation cycle; and after completing the data acquisition of the first rotation cycle, if the range of the historical current array is less than a preset fluctuation threshold, determining that the current cycle data is invalid and setting the adaptive gain coefficient to zero.

[0015] In a second aspect, an automatic winding and coating device for a toroidal transformer core is provided, comprising a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement the tension control method for automatic winding of a toroidal transformer core as described in the first aspect.

[0016] The present invention has the following beneficial effects: By constructing and latching a corresponding current array by associating torque current with mechanical angle, and combining waveform feature analysis to determine the adaptive gain coefficient for load phase and inertia adaptation, and then determining the current prediction value by looking up a table based on real-time mechanical angle and speed, and finally integrating relevant parameters to generate feedforward compensation commands, this method can effectively solve the phase drift caused by belt drive and the time-varying inertia problem caused by winding material consumption. At the same time, it can offset the response lag of the servo system, realize precise and dynamic control of the winding tension of the toroidal transformer core, and ensure that the timing and force of tension compensation can match the actual winding conditions, effectively suppress tension fluctuations during the winding process, and ensure the stability and accuracy of winding tension control. Attached Figure Description

[0017] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a tension control method for automatic winding of a toroidal transformer core, provided in one embodiment of the present invention; Figure 2 This is a schematic diagram of the hardware structure of an automatic winding and coating device for a toroidal transformer core, provided as an embodiment of the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an automatic winding and coating device and tension control method for a toroidal transformer core according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] The following description, in conjunction with the accompanying drawings, details the specific scheme of the automatic winding and coating equipment and tension control method for a toroidal transformer core provided by the present invention.

[0022] Please see Figure 1The diagram illustrates a method flowchart for tension control of an automatic winding of a toroidal transformer core according to an embodiment of the present invention. The method includes: S1. Collect the torque current and mechanical angle of the servo motor, associate the torque current and mechanical angle at the time of collection, construct a real-time current array indexed by the mechanical angle, and latch the real-time current array according to the rotation cycle of the servo motor to obtain the historical current array.

[0023] In some implementations, during the power-on initialization phase of the servo driver, the processor first allocates a dedicated data buffer in random access memory and constructs two floating-point arrays as a real-time current array and a historical current array. The array length is set according to the sampling resolution of the mechanical angle. In this embodiment, a length of 360 is used, corresponding to a 1° resolution for mechanical angles from 0° to 359°. If higher precision is required, the array length can also be preset to 720, corresponding to a 0.5° resolution, or 3600, corresponding to a 0.1° resolution. At the same time, the processor performs an initialization operation on all elements of the two arrays, setting their values ​​to 0.0 to avoid interference from residual random data in memory on subsequent tension control logic. A Boolean-type first-cycle sampling completion flag is also defined and its initial value is set to False. This flag is used to determine whether the system has accumulated valid historical data that supports feedforward compensation calculations. Before completing the data acquisition for the first rotation cycle, the processor will prohibit the output of feedforward compensation instructions based on this flag, thereby ensuring the safe operation of the system.

[0024] After the servo driver enters real-time operation, the processor continuously performs data acquisition and correlation operations within a preset high-frequency interrupt service routine. In this embodiment, the execution cycle of the high-frequency interrupt service routine is preset to 62.5μs. Each time an interrupt is triggered, the processor synchronously reads the absolute mechanical position of the servo motor from the position sensor and the q-axis torque current from the current sensor. The q-axis torque current directly reflects the driving torque required by the wire loop to overcome frictional resistance and wire tension at the current moment. Subsequently, the processor performs rounding down and modulo 360 operations on the read absolute mechanical position to obtain the target index corresponding to the current moment in the real-time current array. This calculation method ensures that the target index always falls within the valid range of 0 to 359, achieving precise correlation between torque current and mechanical angle.

[0025] After completing the index calculation, the processor writes the currently collected q-axis torque current into the position corresponding to the target index in the real-time current array. For different speed conditions of the servo motor, the processor will also perform differentiated data optimization operations: Under low-speed conditions, the motor may be sampled multiple times within the same 1° mechanical angle range. In this embodiment, a latest value overwrite strategy can be preset to directly overwrite the historical data at the same index position with the newly collected torque current. Alternatively, a cumulative averaging strategy can be preset according to actual needs to calculate the average value of multiple samples before writing, thereby simplifying the calculation and ensuring data timeliness. Under high-speed conditions, if the motor speed exceeds the preset 3000rpm threshold, the angle that the motor rotates between two adjacent interrupted samples will exceed 1°, resulting in unfilled index holes in the real-time current array. At this time, the processor will check the difference between the current target index and the index at the previous moment. If the difference is greater than 1, a linear interpolation algorithm will be used to calculate and fill the data at the missing index position using the current value at the previous index position and the current value at the current target index position, ensuring the data continuity of the real-time current array and avoiding the impact of data holes on subsequent analysis. Simultaneously, the system performs cross-cycle interpolation. When it detects that the current index is less than the previous index, it is determined to be a mechanical angle cross-cycle jump (such as a jump from 359° to 1°). At this time, the current index is added to 360 and then the previous index is subtracted to calculate the interpolation distance. Then, linear interpolation is performed to fill the data at the missing index position.

[0026] The processor monitors the Z-phase pulse signal of the servo motor encoder in real time. This signal serves as the zero-position detection basis for the motor's rotation cycle. When the rising edge of the Z-phase pulse signal is detected, it is determined that the servo motor has completed a complete rotation cycle. At this time, the processor immediately executes a high-priority memory copy atomic operation, calling memory copy functions (memcpy) and other memory copy instructions to completely copy all 360 floating-point data in the real-time current array to the historical current array. This allows the historical current array to store the complete load current waveform data of the previous rotation cycle, providing a static and complete input source for subsequent feature analysis and feedforward lookup.

[0027] After the historical current array is latched, the processor updates the system state and sets the first sampling completion flag to True, indicating that the system has successfully established the first valid historical reference data. Subsequent control logic can then activate the feedforward compensation function based on this data. At the same time, the current sampling value is retained in the real-time current array and the corresponding index position is normally overwritten. The array is only updated point by point through real-time sampling in the new rotation cycle.

[0028] After acquiring and latching the historical current array for the first rotation cycle, the processor performs a data validity check on the historical current array. First, it iterates through the historical current array to find the maximum and minimum current values, calculates the difference between them to obtain the current range, and then compares this current range with a preset minimum load fluctuation threshold. In this embodiment, this threshold is preset to 5%~10% of the reference current corresponding to the rated tension. This threshold corresponds to the minimum tension fluctuation range under normal winding conditions and can adapt to the tension control requirements of different specifications of enameled wire. If the current range is less than the preset fluctuation threshold, the processor determines that the current cycle data is invalid, indicating that the system may be in an idling or disconnected state. In this case, it will directly set the adaptive gain coefficient to zero to avoid outputting incorrect control commands.

[0029] S2. Perform waveform feature analysis on the historical current array to determine the interval division result used to characterize the current load phase information and the adaptive gain coefficient used to characterize the current inertia state of the system.

[0030] In some implementations, after verifying the validity of the historical current array and determining that the data is valid, the processor performs waveform feature analysis on the historical current array. The core objective is to decouple the interval division result that represents the current load phase information and the adaptive gain coefficient that represents the current inertia state of the system. The entire process relies on a preset iterative classification algorithm to complete the interval division, and then obtains the basic friction current and adaptive gain coefficient through statistical calculation and calibration rules, providing core parameter support for the generation of subsequent feedforward compensation instructions.

[0031] The processor first performs interval partitioning based on waveform amplitude to address the nonlinear drift of the coil phase relative to the motor angle caused by belt slippage, abandoning the traditional partitioning method that relies on absolute mechanical angles. In this embodiment, the preset iterative classification algorithm uses a simplified K-means clustering algorithm. The processor first sorts all current values ​​in the historical current array from largest to smallest, selecting the arithmetic mean of the first 10% of the sorted data as the initial center of the stretching zone, and the arithmetic mean of the last 10% of the sorted data as the initial center of the shrinking zone, thus initializing the cluster centers and eliminating interference from occasional current spike noise. Subsequently, the processor iterates through each current value in the historical current array, calculating the Euclidean distance from each current value to the two initial cluster centers. By comparing the two distances, the processor determines the interval to which the current value belongs. If the distance from the current value to the center of the stretching zone is smaller, the current value is assigned to the high-load stretching zone dataset; if the distance from the current value to the center of the shrinking zone is smaller or equal, the current value is assigned to the low-load shrinking zone dataset, completing the first round of sample classification.

[0032] After completing the first round of sample classification, the processor calculates the arithmetic mean of the two datasets respectively, and uses the calculated average as the new center of the stretching region and the center of the retraction region to update the cluster centers. Next, the processor compares the change in the new and old cluster centers. If the change is less than a preset convergence threshold (in this embodiment, the convergence threshold is preset to 0.01A), or if the number of iterations reaches a preset upper limit (in this embodiment, the upper limit of the number of iterations is preset to 10), then the iteration is considered to have converged and the operation is stopped. If the convergence condition is not met, the sample classification and center update operation is repeated using the updated cluster centers. After the iteration is completed, the final high-load stretching region dataset and low-load retraction region dataset are the interval division results representing the current load phase information. This process does not rely on the absolute position of the mechanical angle; it locks the true physical phase of the conductor stretching and retraction during the winding process only through the current amplitude characteristics, effectively avoiding the phase misalignment problem caused by belt slippage.

[0033] After obtaining the interval division results, the processor immediately performs the extraction operation of the basic frictional current. The purpose is to extract the DC bias required by the system to overcome mechanical friction, belt resistance, and gravity components from the original current waveform, laying the zero-point reference for subsequent compensation only for the tension fluctuation component. The processor traverses all current values ​​in the low-load retraction zone dataset, accumulates these current values, and then divides the accumulation result by the number of elements in the dataset to calculate the arithmetic mean of the low-load retraction zone dataset. This average value is assigned to the global variable basic frictional current. This parameter accurately represents the average load level of the system at the easiest moment during the winding process and becomes the zero-point reference for subsequent current waveform processing.

[0034] Subsequently, the processor performs inertia state observation and adaptive gain calculation operations. This operation addresses the time-varying inertia problem caused by consumable consumption during winding. It indirectly observes the current inertia state using the statistical characteristics of the current waveform and adjusts the compensation gain accordingly. The processor first calculates the statistical variance of the high-load stretching region dataset. The calculation begins by obtaining the mean of the dataset, which is the center of the stretching region obtained by the iterative classification algorithm. Then, the squares of the differences between each current value and the mean are accumulated. Finally, the accumulated result is divided by the number of elements in the dataset. The resulting statistical variance characterizes the degree of glitches in the current waveform of the high-load stretching region, indirectly reflecting the rotational inertia state of the system. In engineering practice, the less enameled wire consumable in the storage ring, the smaller the system rotational inertia, the more sensitive the dynamic response under the same mechanical disturbance, the more high-frequency glitches in the current waveform, and the larger the statistical variance.

[0035] Before calculating the adaptive gain coefficient, offline calibration of the calibration constant must be completed. The calibration process is carried out under the standard operating condition of full load on the wire storage ring: First, full load preparation is performed by loading the wire storage ring of the toroidal winding machine with a full reel of enameled wire to reach the maximum design material consumption, corresponding to the maximum rotational inertia state of the system; then, the equipment is started and set to the rated operating speed, which is preset to 800 rpm in this embodiment, and the feedforward compensation function is temporarily turned off, relying only on the servo's own speed loop PID for control; next, gain tuning is performed by gradually and manually increasing the feedforward gain coefficient and observing the current waveform. The system's stability and the feel of the enameled wire's tension are assessed. When the gain increases to a certain level and the system begins to exhibit slight oscillations or high-frequency whistling, the gain value at this point is recorded. This value is then reduced by 10% to 20% to represent the optimal gain under full load, at which point the system is in a state of near-stability with minimal tension fluctuations. While maintaining stable operation under the optimal full load gain, the system automatically executes variance calculation logic, recording the statistical variance value of the data in the stretching zone at this point, denoted as the full load variance. This value reflects the baseline glitches in the current waveform under full load and high inertia. Finally, based on the optimal full load gain... With full load variance The calibration constant is calculated using a preset formula. , represented as: In the formula, is a calibration constant with dimensions A², representing the optimal compensation gain reference under full-load conditions; The optimal gain under full load (i.e., full load gain) is dimensionless and characterizes the feedforward gain under the critical steady state under full load. The full-load variance, with dimensions A², characterizes the reference glitches in the current waveform under full-load and large inertia conditions. It is an extremely small positive number, such as 1×10-6A², which is a zero-prevention parameter adjustment coefficient used to avoid calculation anomalies when the denominator is zero.

[0036] The calculated calibration constant is written into the non-volatile memory of the servo driver. In subsequent actual production, regardless of how much consumables are consumed, the system will use this fixed constant to automatically back-calculate the current optimal gain based on the real-time variance, without the need for manual intervention.

[0037] After obtaining the calibration constants, the processor calculates the adaptive gain coefficients based on the statistical variance and the calibration constants. , represented as: In the formula, is a calibration constant with dimensions A², and is a reference parameter obtained from offline calibration, which characterizes the optimal compensation gain reference of the system under full load conditions. , where A² is the statistical variance, representing the current statistical variance of the high-load stretching region dataset. It characterizes the degree of current waveform glitches and indirectly reflects the system's rotational inertia. It is a very small positive number, such as 1 × 10 -6 A² is the zero-prevention parameter adjustment coefficient, used to avoid calculation anomalies when the denominator is zero; Dimensionless, representing the scaling ratio of the load fluctuation amplitude, used to adapt the tension compensation intensity to the current system inertia state.

[0038] After the adaptive gain coefficient is calculated, the system executes gain saturation protection logic to prevent the calculated gain coefficient from becoming infinitely large due to extremely stable operating conditions with very small current variance, which could cause system oscillation. If the calculated adaptive gain coefficient exceeds the preset threshold (such as 1.5 times the full-load gain of offline calibration), it will be forcibly clamped to the upper limit value to avoid the gain value from exploding and causing overcompensation.

[0039] S3. Based on the current mechanical angle and real-time rotation speed, determine the lookup table index pointing to the load state at a future time, and read the current prediction value corresponding to the lookup table index from the historical current array.

[0040] In some implementations, after completing feature analysis and obtaining interval division results, basic friction current, and adaptive gain coefficient, the processor enters the feedforward instruction generation stage. The core task is to use real-time mechanical angle and rotation speed to determine the lookup table index pointing to the load state at a future moment, read the corresponding current prediction value from the historical current array, and provide load prediction basis for subsequent generation of feedforward compensation instructions. The whole process first performs safety interlock verification, then offsets the inherent lag of the system through dynamic lead angle calculation, and finally achieves accurate prediction of the future load state.

[0041] The processor first performs a safety interlock check. Since feedforward control heavily relies on historical data and characteristic parameters from the previous cycle, the relevant data may not be established or may be incomplete during the first rotation cycle after equipment startup. To prevent the output of incorrect torque commands in a data-deficient state from causing equipment overrun or disconnection, the processor reads the first-cycle sampling completion flag from the global state variables at the beginning of each high-frequency current loop control cycle. If this flag is False, it indicates that the system is in the first-cycle data acquisition phase or initialization state. The processor forcibly sets the feedforward torque command to 0.0, and the system maintains basic operation solely based on the conventional speed and position loops, without performing additional feedforward intervention. If this flag is True, it indicates that the system has successfully established valid historical reference data, and the key parameters have been updated in the S200 stage, allowing entry into the subsequent feedforward calculation process.

[0042] After completing the safety interlock verification, the processor performs dynamic advance angle calculation and table lookup index generation operations. This aims to address the inherent lag problem in the servo system from receiving commands to outputting actual electromagnetic torque, including communication delay, current loop response time, and mechanical inertia, transforming post-event feedback into pre-event prevention. The system's total lag time constant is a fixed parameter calibrated offline. This parameter is the sum of the servo system's communication delay, driver calculation delay, current loop response delay, and phase lag generated by the signal filter. Its value depends on hardware performance; in this embodiment, it is preset to 2ms, obtained through step testing. The motor's real-time speed is fed back by the encoder. The processor first performs a first-order low-pass filter on the encoder's feedback of the motor's real-time speed to remove encoder differential noise. Since the motor's real-time speed dynamically changes with the variable frequency speed control, the fixed angle compensation cannot be adapted. Therefore, the processor uses a preset formula to dynamically calculate the advance compensation angle. The calculation formula is: In the formula, The filtered real-time motor speed is expressed in revolutions per minute (rpm). It is a fixed constant, representing the total angle (360°) of a complete rotation cycle. It is a fixed constant used to convert the unit of rotational speed from "revolutions per minute" to "revolutions per second" (1 minute = 60 seconds); is the total system lag time constant, measured in seconds (s), and is a fixed parameter for offline calibration, representing the total delay time from receiving a command to outputting the actual electromagnetic torque in the servo system; Round is a rounding function, a numerical processing function, specifically named the rounding function, used to convert angle values ​​to integers to avoid invalid indexes; The dimension is degrees, representing the angular offset required to compensate for the inherent lag of the servo system.

[0043] Meanwhile, to prevent the calculated lead angle from being too large at extremely high speeds, which could lead to phase misalignment, the system sets the maximum value of the lead compensation angle to not exceed a preset empirical threshold of 15°. If the value exceeds this threshold, the lead compensation angle will be forcibly clamped to 15°.

[0044] After calculating the advance compensation angle, the processor reads the current real-time mechanical angle, which ranges from 0 to 359° and is directly fed back by the position sensor. Then, the current mechanical angle is added to the advance compensation angle, and the result is modulo 360 to generate an advance lookup index j pointing to the load data at future times. The calculation formula is as follows: In the formula, This is the current real-time machine angle; It is the leading compensation angle; Add the current mechanical angle to the advance compensation angle to obtain the mechanical angle value at the future time. Modulo operation, a number theory operation, is specifically called modulo 360 operation. It is used to map angle values ​​to a valid range of 0 to 359° to achieve cyclic indexing. j represents the position of the load state in the historical current array corresponding to the future time. The load current prediction value can be read through this index to realize the early prediction of tension fluctuations and avoid the compensation timing error caused by servo lag.

[0045] S4. Based on the current prediction value, the interval division result, and the adaptive gain coefficient, generate a feedforward compensation command to be superimposed on the torque setpoint of the servo driver.

[0046] In some implementations, after generating and reading the predicted current value by looking up the table, the processor enters the synthesis and output stage of the reverse compensation torque. The core is to remove the interference from the mechanical base, dynamically adjust the compensation force, and generate a feedforward compensation instruction that can be directly superimposed on the torque given by the servo driver. The whole process relies on the characteristic parameters and executes the DC removal operation, gain adjustment operation, and reverse synthesis operation in sequence to finally complete the instruction superposition and achieve precise cancellation of tension fluctuations.

[0047] The processor first performs a data read operation, using a lookup table index to read the corresponding current value from the historical current array. This current value is the predicted current value, representing the load current reference at future moments, and providing a predictive basis for subsequent compensation calculations.

[0048] The subsequent DC removal operation aims to remove the DC bias required to overcome mechanical friction, belt resistance, and gravity components from the original predicted current, ensuring that subsequent compensation targets only the tension fluctuation component. The processor subtracts the base friction current from the read current prediction value to obtain the effective tension torque increment containing only the tension fluctuation component. This step eliminates constant load interference, allowing the compensation torque to be precisely focused on the dynamically changing tension portion, avoiding ineffective compensation for constant loads.

[0049] Next, a gain adjustment operation is performed, multiplying the tension torque increment by the adaptive gain coefficient to obtain the adjusted compensation amplitude. The adaptive gain coefficient changes dynamically with the consumption of consumables. When consumables are sufficient (full load), the system inertia is large, the current waveform is smooth, the variance is small, the adaptive gain coefficient is large, and the compensation force is sufficient. When consumables are exhausted (no load), the system inertia is small, the current waveform has many glitches, the variance is large, the adaptive gain coefficient automatically decreases, and the compensation force weakens. This step ensures that the compensation torque amplitude automatically weakens in the no-load state, preventing overcompensation from causing system oscillation or wire breakage.

[0050] Next, the reverse synthesis and output operation is performed. To counteract tension fluctuations, the direction of the compensation torque must be opposite to the direction of the disturbance. The processor multiplies the adjusted compensation amplitude with the sign value of the motor's real-time speed (i.e., the sign function of the motor's real-time speed, which is 1 when the speed is forward and -1 when it is reverse, used to ensure the damping characteristics of the compensation torque during forward and reverse switching), and then inverts the product to generate the final feedforward torque command. Finally, the command superposition operation is performed, directly superimposing the calculated feedforward torque command onto the torque command signal output by the servo driver's speed loop. Through this operation, the servo motor will generate a precisely matched reverse torque at the system lag time before the tension peak arrives, actively counteracting the pulling effect caused by the slider movement and ensuring that the wire tension remains constant throughout the high-speed winding process.

[0051] The method for determining the system lag time is as follows: First, a step test is performed, where the motor is stationary and a momentary torque step command is given. In this embodiment, the preset value is a jump from 0 to 50% of the rated torque. Then, using an oscilloscope or high-frequency data acquisition card, the digital signal waveform at the moment the command is issued and the analog signal waveform at the moment the actual motor current rises are monitored simultaneously. Next, the time difference between the rising edges of the two waveforms is measured, and this time difference is the current loop response lag time. Finally, considering the communication cycle of the servo bus, this embodiment uses Ethernet for control automation technology (EtherCAT) bus, with a preset communication cycle of 1ms to compensate for the current loop response lag time. The final system lag time is set as the measured response time plus 1.5 times the bus communication cycle. This parameter is usually a fixed value after the servo system selection is determined. In this embodiment, it is preset to 2ms~5ms and can be directly written into the program constant.

[0052] Based on the above technical solution, by constructing and latching the corresponding current array by associating torque current with mechanical angle, and determining the adaptive gain coefficient for load phase and inertia adaptation by combining waveform feature analysis, and then obtaining the current prediction value by determining the lookup table index based on real-time mechanical angle and speed, and finally generating feedforward compensation commands by integrating relevant parameters, it can effectively solve the phase drift caused by belt drive and the time-varying system inertia caused by winding material consumption. At the same time, it can offset the response lag of the servo system, realize precise and dynamic control of the winding tension of the toroidal transformer core, and ensure that the timing and force of tension compensation can match the actual winding conditions, effectively suppress tension fluctuations during the winding process, and ensure the stability and accuracy of winding tension control.

[0053] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0054] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0055] In this embodiment of the invention, the automatic winding and coating equipment for toroidal transformer cores can be divided into functional units according to the above method example. For example, each function can be divided into its own functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or software. It should be noted that the unit division in this embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0056] This invention also provides a schematic diagram of the hardware structure of an automatic winding and coating device for toroidal transformer cores, see [link / reference]. Figure 2 The automatic winding and coating equipment 200 for the toroidal transformer core includes a processor 201, and optionally, a memory 202 connected to the processor 201.

[0057] In the first possible implementation, see Figure 2 The automatic winding and coating equipment 200 for toroidal transformer cores also includes a transceiver 203. The processor 201, memory 202, and transceiver 203 are connected via a bus. The transceiver 203 is used to communicate with other devices or communication networks. Optionally, the transceiver 203 may include a transmitter and a receiver. The device in the transceiver 203 that implements the receiving function can be considered as a receiver, which is used to perform the receiving steps in the embodiments of the present invention. The device in the transceiver 203 that implements the transmitting function can be considered as a transmitter, which is used to perform the transmitting steps in the embodiments of the present invention.

[0058] Based on the first possible implementation method Figure 2 The structural diagram shown can be used to illustrate the structure of the automatic winding and coating equipment for the toroidal transformer core involved in the above embodiments.

[0059] in, Figure 2The diagram can also illustrate the system chip in the automatic winding and coating equipment for toroidal transformer cores. In this case, the actions performed by the aforementioned automatic winding and coating equipment for toroidal transformer cores can be implemented by this system chip. Specific actions performed can be found above and will not be repeated here.

[0060] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings and the disclosure, will understand and implement other variations of the disclosed embodiments in carrying out the claimed invention. In this invention, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several of the functions listed in this invention.

[0061] Although the invention has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Accordingly, this specification and drawings are merely illustrative of the invention and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications of the invention fall within the scope of the invention and its equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A tension control method for automatic winding of a toroidal transformer core, characterized in that, include: The torque current and mechanical angle of the servo motor are collected, the torque current at the time of collection is associated with the mechanical angle, a real-time current array indexed by the mechanical angle is constructed, and the real-time current array is latched according to the rotation cycle of the servo motor to obtain the historical current array. Waveform feature analysis is performed on the historical current array to determine the interval division result used to characterize the current load phase information and the adaptive gain coefficient used to characterize the current inertia state of the system. Based on the current mechanical angle and real-time rotation speed, determine the lookup table index pointing to the load state at a future time, and read the current prediction value corresponding to the lookup table index from the historical current array; Based on the predicted current value, the interval division result, and the adaptive gain coefficient, a feedforward compensation command is generated for superimposing onto the torque setpoint of the servo driver.

2. The tension control method for automatic winding of a toroidal transformer core according to claim 1, characterized in that, Constructing a real-time current array includes: In the high-frequency interrupt service routine of the servo driver, the torque current and mechanical angle are synchronously acquired; The mechanical angle is rounded and moduloed to obtain the target index corresponding to the real-time current array at the current moment; the length of the real-time current array corresponds to the resolution of the mechanical angle within one rotation cycle; Write the torque current into the position corresponding to the target index in the real-time current array; If the mechanical angle interval between two consecutive interrupt service routines is greater than the resolution, interpolation is used to fill the missing index positions in the real-time current array.

3. The tension control method for automatic winding of a toroidal transformer core according to claim 1, characterized in that, Determine the interval partitioning results, including: Based on the magnitude characteristics of the current values ​​in the historical current array, a preset classification algorithm is used to divide the current values ​​in the historical current array into a high-load stretching zone dataset and a low-load shrinking zone dataset, which are used as the interval division results.

4. The tension control method for automatic winding of a toroidal transformer core according to claim 3, characterized in that, Determining the adaptive gain coefficients includes: Calculate the statistical variance of the concentrated current values ​​in the high-load stretching zone dataset; The adaptive gain coefficient is determined based on the statistical variance and a preset calibration constant; the adaptive gain coefficient is negatively correlated with the statistical variance.

5. The tension control method for automatic winding of a toroidal transformer core according to claim 4, characterized in that, Determine the preset calibration constants, including: Under the standard operating condition of full load on the storage loop, determine the full load gain when the system is operating in a critical steady state, and obtain the full load variance of the stretching region dataset under the standard operating condition. The calibration constant is determined based on the full-load gain and the full-load variance.

6. The tension control method for automatic winding of a toroidal transformer core according to claim 3, characterized in that, Also includes: The base friction current is determined based on the current values ​​in the low-load retraction zone dataset. The basic triboelectric current is used to characterize the average current required by the system to overcome mechanical friction, belt resistance, and the gravitational component.

7. The tension control method for automatic winding of a toroidal transformer core according to claim 6, characterized in that, Based on the predicted current value, the interval division result, and the adaptive gain coefficient, a feedforward compensation command is generated for superimposing onto the torque setpoint of the servo driver, including: The base friction current is removed from the current prediction value to obtain the tension torque increment used to characterize the tension fluctuation component; The adjusted compensation amplitude is obtained by fusing the tension torque increment with the adaptive gain coefficient. The adjusted compensation amplitude is inverted to generate the feedforward compensation command.

8. The tension control method for automatic winding of a toroidal transformer core according to claim 1, characterized in that, Based on the current mechanical angle and real-time rotational speed, determine the lookup table index pointing to the load state at a future time, including: Based on the system's inherent total lag time constant and the real-time rotational speed, determine the lead compensation angle used to offset the system response lag; The current mechanical angle is added to the advance compensation angle, and then a modulo operation is performed to obtain the advance lookup table index.

9. The tension control method for automatic winding of a toroidal transformer core according to claim 1, characterized in that, Also includes: Before the data acquisition for the first rotation cycle is completed, the output of feedforward compensation commands is prohibited; After completing the data acquisition for the first rotation cycle, if the range of the historical current array is less than the preset fluctuation threshold, the current cycle data is determined to be invalid, and the adaptive gain coefficient is set to zero.

10. An automatic winding and coating equipment for a toroidal transformer core, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement the tension control method for automatic winding of the toroidal transformer core as described in any one of claims 1 to 9.

Images