Method for dynamic measurement of material roll diameter based on hybrid filtering

Abstract

The application discloses a material roll diameter measurement method based on an encoder pulse, comprising the following steps: 1) roll diameter measurement starting condition detection; 2) encoder pulse division and measurement triggering; 3) roll length and shaft position calculation; 4) instantaneous radius calculation and storage; 5) Kalman filtering and roll diameter calculation: a hybrid algorithm of double one-dimensional Kalman filtering and one-dimensional particle filtering is adopted, independent sampling and asynchronous measurement are combined to realize optimal estimation of the roll diameter, and the hybrid algorithm comprises parameter setting, mode selection, filtering execution, weighted fusion and roll diameter compensation; and 6) output of measurement results. The application can solve the problems of low measurement precision, weak anti-interference and inadaptability to roll winding direction switching of a traditional material roll winding system, has the characteristics of high measurement precision, strong robustness and good dynamic adaptability, can be linked to closed-loop control of roll winding tension, and improves the material roll winding quality.

Description

Technical Field

[0001] This invention relates to the field of roll diameter measurement technology for material winding systems, and in particular to a high-precision material roll diameter measurement method based on encoder pulse signals and combined with a hybrid filtering algorithm, applicable to real-time roll diameter detection during the winding and unwinding process of various types of wound materials. Background Technology

[0002] In the continuous winding industry of materials such as printing, packaging, and lithium battery electrode manufacturing, the roll diameter is a core process parameter that directly affects the tension control accuracy and product quality. The measurement method based on encoder pulses has become the mainstream solution due to its simplicity, economy, and rapid response.

[0003] However, existing technologies have obvious drawbacks: fluctuations in winding speed can easily introduce measurement noise, and traditional timed sampling can easily lead to the accumulation of data errors; the limitations of a single filtering scheme are prominent; at the same time, outlier verification is imperfect and it is difficult to balance measurement accuracy, real-time performance and robustness.

[0004] Therefore, there is an urgent need to develop technologies that integrate the advantages of multiple filters and adapt to complex working conditions, to solve the problems of insufficient existing filtering strategies and weak anti-interference capabilities, and to achieve high-precision and high-stability real-time measurement of roll diameter. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a material roll diameter measurement method based on encoder pulses to solve the problems of low accuracy, weak anti-interference and unsuitability for dynamic working conditions of existing roll diameter measurement methods, so as to realize real-time, high-precision and robust measurement of roll diameter during the winding process and provide reliable data support for tension closed-loop control.

[0006] The material roll diameter measurement method based on encoder pulses of the present invention includes the following steps: 1) Encoder pulse classification and measurement triggering:

[0007] The encoder pulse count per revolution is divided into N equal parts, where the number of pulses in each part is the encoder resolution divided by N. A roll diameter measurement is triggered only when the change in the encoder pulse count reaches or exceeds the number of equal parts. Based on the encoder pulse count and shaft position, the winding length and shaft position change are calculated.

[0008] 2) Radius rationality check:

[0009] The obtained instantaneous radius value will be compared with the previously stored radius value for a validity check. Only radius values ​​that meet the roll-up conditions will be stored in the array in sequence.

[0010] 3) First-order Kalman filtering:

[0011] Prediction phase (taking roll A unwinding as an example): Based on the number of rotations and the thickness of the wound material collected by the encoder, the roll diameter change of the tested roll is predicted in real time, and this is performed once per sampling cycle.

[0012] Prior estimation of roll diameter for roll A (unwinding):

[0013]

[0014] in, The prior estimate of the volume diameter represents the number of k sampling points. The posterior (optimal) estimate of the volume diameter represents the k-1 sampling points.

[0015] Prior error covariance update:

[0016]

[0017] Let be the prior error covariance of k sampling points. Let be the posterior error covariance of the k-1 sampling points. For a one-dimensional Kalman filter, the state transition matrix is... =1.

[0018] Measurement acquisition during the correction phase: When the cumulative number of rotations of the rotary roller reaches the set number, the roll diameter of the roller under test is calculated by the number of rotations of the rotary encoder and the encoder radius, triggering the roll diameter measurement and executing the correction. The two rollers are triggered independently and do not affect each other.

[0019] One-dimensional Kalman gain calculation:

[0020]

[0021] For a one-dimensional Kalman filter, the observation matrix H=1.

[0022] Posterior estimation update (taking roll A as an example):

[0023]

[0024] After the correction is triggered, the lap counter is reset to zero.

[0025] Error covariance update (after each prediction or correction phase):

[0026]

[0027] Because of which =1, =1. When correction is not triggered, =0, so the prediction phase can be used directly. As the current error covariance.

[0028] 4) First-order particle filter:

[0029] Particle initialization: in A batch of diameter values ​​near the diameter that follow a normal distribution are obtained; these are the initial particles. Importance sampling:

[0030] Weight update:

[0031] Weight normalization:

[0032] in, This represents the normalized weight of the i-th particle at sampling point k. This represents the unnormalized weight of the i-th particle at sampling point k. State estimation: When the number of effective particles At that time, system resampling is performed.

[0033] Particle filter covariance calculation:

[0034]

[0035] 5) Second-order Kalman filter: Kalman filter output: Particle filter output: Fusion state equation: Kalman filter weights: Particle filter weights:

[0036] Roll diameter compensation and final calculation (taking roll A as an example):

[0037] in This is the final roll diameter value. This is the estimated roll diameter after hybrid filtering and fusion. This is the diameter compensation value of roller A obtained by the PID algorithm after filtering.

[0038] The beneficial effects of this invention are:

[0039] This invention presents a material roll diameter measurement method based on encoder pulses. Addressing the issues of noise interference and poor adaptability during material winding, this method employs a hybrid algorithm combining dual one-dimensional Kalman filtering and one-dimensional particle filtering, along with a two-order fusion architecture. This approach integrates pulse-based triggering sampling with dynamic compensation logic. Compared to traditional single-filter measurement methods, it reduces algorithm redundancy while ensuring real-time performance. Furthermore, it simultaneously solves problems related to: measurement noise suppression, handling sudden nonlinear interference, the impact of winding speed fluctuations, abnormal data interference, and adaptation to winding direction switching. Moreover, the resulting measurement scheme exhibits strong robustness against Gaussian white noise and nonlinear interference, adaptability to winding speed fluctuations and material characteristic differences, and high adaptability to complex conditions such as winding direction switching and sudden tension changes, effectively improving the accuracy and stability of roll diameter measurement. Attached Figure Description

[0040] Figure 1 Here is the system structure diagram: In the diagram, 1 is the unwinding motor, 2 is the drive shaft, 3 is the tension sensor, 4 is the clamping device, 5 is the rotary encoder, 6 is the drive shaft, 7 is the extrusion roller, 8 is the material tray, 9 is the drive shaft, and 10 is the rewinding motor.

[0041] Figure 2 This is the main flowchart for roll diameter measurement;

[0042] Figure 3 This is a filtering logic diagram. Detailed Implementation

[0043] The present invention will be further described below with reference to the accompanying drawings and embodiments, so that those skilled in the art can implement it based on the description.

[0044] The technical solution adopted in this invention is a material roll diameter measurement method based on encoder pulses, including the following steps: 1) Roll diameter measurement start condition detection: measurement is started after monitoring that the material tension meets the standard and is stable; 2) Encoder pulse division and measurement triggering: the number of encoder pulses per revolution is divided into N equal parts, and measurement is triggered when the pulse change meets the standard; 3) Calculation of winding length and shaft position: the winding length and the arc of the take-up roller are calculated based on encoder parameters using a specific formula; 4) Instantaneous radius calculation and storage: the instantaneous radius is calculated by the change in winding length and the change in shaft position, and stored after a reasonableness check; 5) Kalman filtering and roll diameter calculation: a hybrid algorithm of dual one-dimensional Kalman filtering and one-dimensional particle filtering is adopted, combined with independent sampling and asynchronous measurement to achieve optimal roll diameter estimation, including parameter setting, mode selection, filtering execution, weighted fusion and roll diameter compensation; 6) Output measurement results.

[0045] This invention can solve the problems of low roll diameter measurement accuracy, weak anti-interference, and inability to adapt to winding direction switching in traditional material winding systems. It has the characteristics of high measurement accuracy, strong robustness, and good dynamic adaptability. It can be linked to closed-loop control of winding tension to improve the quality of material winding.

[0046] The material roll diameter measurement method based on encoder pulses of the present invention includes the following steps:

[0047] 1) Roll diameter measurement start-up condition detection:

[0048] Monitor the material tension. When the tension is greater than or equal to the material tightening threshold (the tightening value varies depending on the extensibility of different materials; for film materials, it is generally set to 150N) and the tension is continuously dragged for a set stabilization time (the time required for the tension to reach a stable state), start the roll diameter measurement process.

[0049] 2) Encoder pulse division and measurement triggering:

[0050] The number of pulses per revolution of the encoder is divided into N equal parts. The number of pulses in each part is the encoder resolution divided by N. For example, if the encoder resolution is 1024 and N is 4, then the number of pulses in each part is 256. A roll diameter measurement will only be triggered when the change in the number of encoder pulses reaches or exceeds 256 pulses.

[0051] 3) Calculation of winding length and shaft position:

[0052] Based on the encoder pulse count and shaft position, the winding length and shaft position change are calculated; the formula for calculating the winding length is:

[0053]

[0054] Indicates the diameter of the rotary encoder. This indicates the number of pulses from the rotary encoder when the roll diameter measurement is triggered. This indicates the resolution of the rotary encoder.

[0055] The formula for calculating the shaft position is:

[0056]

[0057] This indicates the radius of curvature through which the take-up roller has rotated. This represents the angle through which the take-up roller has rotated.

[0058] 4) Instantaneous radius calculation and storage:

[0059] Based on the changes in winding length and shaft position obtained in step 3, calculate the instantaneous radius value using the following formula:

[0060]

[0061] Indicates the radius of the take-up roller. This is the absolute difference between the current winding length and the previous winding length. The instantaneous radius value is compared with the previously stored radius value to determine the shaft position of the take-up roller when the roll diameter measurement is triggered. Only radius values ​​that meet the take-up conditions are stored in the array in sequence.

[0062] Instantaneous roll diameter measurement That is, twice the radius.

[0063] 5) Filtering and Roll Diameter Calculation: A hybrid algorithm combining dual one-dimensional Kalman filtering and one-dimensional particle filtering is used to optimally estimate the instantaneous roll diameter measurements of the unwinding and take-up rolls. Combining independent sampling (sampling period equal to one rotation time for each roll) and asynchronous measurement (measurement triggered every three rotations), the optimal roll diameter estimation is achieved through a recursive process of "continuous prediction - discontinuous correction - real-time update." The filtering process includes:

[0064] (1) Kalman filter parameter settings: material single-turn thickness corresponding to roll diameter change :

[0065]

[0066] in, The change in radius of a single rotation represents the thickness of the material.

[0067] Process noise covariance :

[0068]

[0069] Measurement noise covariance :

[0070]

[0071] In actual operation, R should be less than Q, which is consistent with the characteristic that the measurement accuracy is higher than the model prediction accuracy.

[0072] Initial error covariance P:

[0073] = =

[0074] in, This is the initial roll diameter measurement error.

[0075] Initial roll diameter set up: and The initial roll diameters are obtained by asynchronous measurement of rolls A and B, respectively.

[0076] Initial value of the lap counter :

[0077] = =

[0078] (2) Particle filter parameter settings:

[0079] Number of sampled particles:

[0080] Process noise distribution: Set as with Consistent.

[0081] Measure noise distribution: Set as with Consistent.

[0082] Resampling threshold Resampling is triggered when the number of effective particles is less than 50% of the total number of particles.

[0083] Particle set initialization: from initial volume distribution Medium sampling There are 1 particle, and the particle weights are initialized to 1. .

[0084] (3) One-dimensional Kalman prediction stage (taking A releasing and B receiving as an example):

[0085] Based on the number of rotations collected by the encoder, the roll diameter change is predicted in real time, and this is performed once per sampling period:

[0086] Prior estimation of roll diameter for roll A (unwinding):

[0087]

[0088] Prior estimate of roll diameter for B roll (rewinding):

[0089]

[0090] in, and The prior estimate of the volume diameter represents the number of k sampling points. and The posterior (optimal) estimate of the volume diameter represents the k-1 sampling points.

[0091] Prior error covariance update:

[0092]

[0093] Let be the prior error covariance of k sampling points. Let be the posterior error covariance of the k-1 sampling points. For a one-dimensional Kalman filter, the state transition matrix is... =1.

[0094] Total number of cycles per sampling period:

[0095] = +1, = +1

[0096] (4) One-dimensional Kalman correction stage (performed once every three rotations)

[0097] When the rotating roller has rotated a total of three times, that is... At that time, the roll diameter measurement is triggered and correction is performed. The two rolls are triggered independently and do not affect each other.

[0098] Measurement value acquisition (taking roller A as an example):

[0099]

[0100] in, This represents the roll diameter measurement after roller A has rotated three times. This represents the total circumference of roller A after three rotations.

[0101] (5) Calculation of one-dimensional Kalman gain:

[0102]

[0103] For a one-dimensional Kalman filter, the observation matrix H=1.

[0104] Parameter update:

[0105] Posterior estimation update (taking roll A as an example):

[0106]

[0107] Triggering correction resets the rotation counter (taking roller A as an example):

[0108] =

[0109] Error covariance update (after each prediction or correction phase):

[0110]

[0111] Because of which =1, =1. When correction is not triggered, =0, so the prediction phase can be used directly. As the current error covariance.

[0112] (6) One-dimensional particle filtering process:

[0113] Particle initialization:

[0114]

[0115] in, Represents the initial state in A batch of normally distributed volume diameter values ​​obtained near the volume diameter are considered the initial particles. Importance sampling:

[0116] Weight update:

[0117] Weight normalization:

[0118] in, This represents the normalized weight of the i-th particle at sampling point k. This represents the unnormalized weight of the i-th particle at sampling point k. State estimation:

[0119] Resampling:

[0120] When the number of effective particles At that time, system resampling is performed.

[0121] Particle filter covariance calculation:

[0122]

[0123] (7) Second-order Kalman filter

[0124] Kalman filter output:

[0125] Particle filter output:

[0126] Fusion state equation:

[0127]

[0128] in .

[0129] Kalman filter weights:

[0130] Particle filter weights:

[0131] in, and These represent the estimation error variances for the two filtering methods.

[0132] (8) Roll diameter compensation and final calculation

[0133] Roll diameter compensation and final calculation (taking roll A as an example):

[0134] in This is the final roll diameter value. This is the estimated roll diameter after hybrid filtering and fusion. This is the diameter compensation value of roller A obtained by the PID algorithm after filtering.

[0135] (9) Dynamic adjustment to adapt to winding direction switching

[0136] When the winding direction is switched to "B unwind A winding", the sign of the change in the winding diameter prediction is automatically adjusted, while other filtering parameters and iteration logic remain unchanged.

[0137] 6) Output measurement results:

[0138] Output the roll diameter measurement results, including the roll diameter value and status flags.

[0139] The material roll diameter measurement method based on encoder pulses in this embodiment can simultaneously ensure the accuracy and stability of roll diameter measurement under complex working conditions such as winding speed fluctuations, nonlinear interference, and direction switching, achieving controllable measurement errors. Compared with traditional single-filter measurement methods, this method has a simpler architecture. Combining pulse-triggered sampling and two-order fusion filtering logic, it can better handle problems such as measurement noise suppression, abnormal data interference, adaptation to differences in material characteristics, and dynamic adjustment requirements for winding direction switching. In addition, it can dynamically compensate for system errors through PID algorithms, greatly reducing the algorithm redundancy and industrial deployment difficulty of roll diameter measurement under complex working conditions.

[0140] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for measuring the diameter of a material roll based on encoder pulses, characterized in that, Including the following steps: 1) Encoder pulse division and measurement triggering: The number of pulses per encoder revolution is divided into N equal parts, and the number of pulses in each part is the encoder resolution divided by N. A roll diameter measurement is triggered only when the change in the number of encoder pulses reaches or exceeds the number of equal parts. Based on the number of encoder pulses and the shaft position, the winding length and the change in shaft position are calculated. Radius rationality check: The obtained instantaneous radius value will be compared with the previously stored radius value for a validity check. Only radius values ​​that meet the roll-up conditions will be stored in the array in sequence. First-order Kalman filter: Prediction phase: Based on the number of rotations and the thickness of the wound material collected by the encoder, the roll diameter change of the tested roller is predicted in real time. This is performed once in each sampling cycle to achieve a priori estimation of the roll diameter and update the prior error covariance. Measurement value acquisition during the correction stage: When the cumulative number of rotations of the rotary roller reaches the set number of rotations, the roll diameter of the roller under test is calculated by the number of rotations of the rotary encoder and the encoder radius, triggering the roll diameter measurement and performing correction. The two rollers are triggered independently and do not affect each other. Calculate the one-dimensional Kalman gain, and update the posterior estimate and error covariance; First-order particle filtering: exist A batch of diameter values ​​that follow a normal distribution obtained near the diameter of the initial particle; Importance sampling is performed on the initial particles, the weights are updated, the weights are normalized, and the state is estimated; the covariance is calculated. 5) Second-order Kalman filter: A hybrid algorithm combining dual one-dimensional Kalman filtering and one-dimensional particle filtering is used to achieve optimal roll diameter estimation by combining independent sampling and asynchronous measurement. It includes parameter setting, mode selection, filtering execution, weighted fusion and roll diameter compensation. 6) Output the measurement results.

2. The material roll diameter measurement method based on encoder pulses according to claim 1, characterized in that, 2) The encoder resolution is divided into equal parts for sampling instead of the traditional timed sampling. The number of equal parts N can be dynamically adjusted according to the measurement accuracy requirements and set through PLC program parameters. The measurement trigger condition is that the encoder pulse change reaches the number of equal parts pulses.

3. The material roll diameter measurement method based on encoder pulses according to claim 1, characterized in that, 3) Perform a reasonableness check on the instantaneous radius value, including range check and rate of change check: the range check ensures that the radius value is between the core radius and the maximum allowable roll diameter; the rate of change check ensures that the change in adjacent radii does not exceed the product of the material thickness and the maximum number of winding turns.

4. The material roll diameter measurement method based on encoder pulses according to claim 1, characterized in that, In steps 3), 4), and 5), a hybrid algorithm combining dual one-dimensional Kalman filtering and one-dimensional particle filtering is used to independently filter the unwinding and winding rolls. The optimal roll diameter is estimated through a recursive process of "continuous prediction - discontinuous correction - real-time update". A fusion filtering architecture is established, and the Kalman filter output and the particle filter output are weighted and fused. The fusion weight is dynamically adjusted according to the estimation error variance of the two filtering methods.

5. The material roll diameter measurement method based on encoder pulses according to claim 1, characterized in that, 6) The fused roll diameter estimate is combined with the PID control algorithm to dynamically compensate and optimize the roll diameter measurement result. The compensation logic is linked with the winding tension closed-loop control and used as the final roll diameter value.

6. The material roll diameter measurement method based on encoder pulses according to claim 1, characterized in that, 5) Kalman filter output: Particle filter output: Fusion state equation: Kalman filter weights: Particle filter weights: ; Roll diameter compensation and final calculation: ; in This is the final roll diameter value. This is the estimated roll diameter after hybrid filtering and fusion. This is the diameter compensation value of roller A obtained by the PID algorithm after filtering.

Images