Dynamic tension control method of dry-method winding and unwinding system

Abstract

The invention discloses a dynamic tension control method for a dry-method winding and unwinding system, solves the problems of high overshoot, long coil diameter sudden change adjustment time and large steady-state fluctuation in low-speed operation of the dry-method winding and unwinding system under a high-speed reversing working condition, and belongs to the technical field of composite material equipment development. The method comprises the steps of collecting data of a winding and unwinding system, identifying total rotational inertia and friction torque on line according to the data, obtaining a friction compensation item according to the friction torque, and obtaining a rotational inertia feed-forward item according to the total rotational inertia; the method comprises the following steps of S1, inputting the tension error and the tension error change rate of the winding and unwinding system into a fuzzy PID controller to obtain a PID parameter real-time adjustment amount, updating a PID parameter, and then obtaining a PID feedback control amount at the current moment k by using the PID controller and the tension error, and S3, synthesizing a friction force compensation item, a rotational inertia feed-forward item and the PID feedback control amount into a final control amount of the winding and unwinding system.

Description

Technical Field

[0001] The present application relates to a dynamic tension control method for a dry winding and unwinding system, belonging to the technical field of composite material equipment development. Background Art

[0002] In the dry winding manufacturing system of composite materials, the unwinding and rewinding unit serves as the core actuator for precise fiber prepreg delivery and interlayer stacking. Its dynamic tension stability directly determines the component interface bonding quality and fiber orientation accuracy. With the stringent requirements of aerospace and new energy storage and transportation equipment for interlayer shear strength and thickness consistency of high-performance composite components, submicron-level control of fiber tension during the winding process has become a key bottleneck to ensure product structural integrity. Inaccurate tension control within the unwinding and rewinding system will cause fiber arrangement slippage, uneven resin impregnation, and interlayer bubble defects, significantly reducing the fatigue life of the component and inducing service risks. The current industrial standard PID control scheme exposes inherent defects such as lack of parameter adaptability and insufficient dynamic anti-disturbance margin due to the multiple disturbance coupling of time-varying inertia of the winding diameter, nonlinear friction hysteresis, and material viscoelastic variation. It is difficult to achieve tension overshoot suppression in high-speed switching conditions, and it is also unable to guarantee steady-state accuracy in the low-speed stage, ultimately leading to irreversible interface delamination and discrete mechanical properties in high-end composite components. Summary of the Invention

[0003] In response to the problems of high overshoot, long adjustment time for sudden change in roll diameter, and large steady-state fluctuation in low-speed operation in a dry winding unwinding system under high-speed reversing conditions, the present application provides a dynamic tension control method for a dry winding unwinding system.

[0004] The present application provides a method for controlling the dynamic tension of a dry winding and unwinding system, comprising:

[0005] S1. Collect the data of the rewinding and unwinding system, identify the total moment of inertia and friction torque at the current k moment online based on the data, and obtain the friction compensation term M based on the friction torque at the current k moment. friction (k), the moment of inertia feedforward term M is obtained based on the total moment of inertia at the current k moment inertia (k);

[0006] S2, the tension error e of the rewinding and unwinding system at the current k moment T (k) and tension error change rate Input to the fuzzy PID controller to obtain the real-time adjustment of the PID parameters. Update the PID parameters according to the real-time adjustment of the PID parameters. Then, obtain the PID feedback control amount M at the current k moment according to the tension error and PID parameters. PID (k);

[0007] S3, M friction (k), M inertia(k), M PID (k) The final control quantity of the synthetic winding and unwinding system.

[0008] Preferably, S1 includes:

[0009] Collect data from the rewinding and unwinding system, including the braking torque M of the unwinding motor at the current time k A (k), the reel diameter r of the reel at the current k moment A (k), the reel diameter r of the unwinding shaft at time k-1 A (k-1), v A (k) is the linear velocity of the unwinding shaft at the current k moment, and the linear velocity of the unwinding shaft at k-1 moment v A (k-1);

[0010] The discretized tension model of the rewinding and unwinding system is:

[0011] z(k)=φ1(k)M FA (k)+φ2(k)J total (k)+e(k)

[0012] Among them, z(k) represents the observed value, and e(k) represents the error between the observed value and the predicted value;

[0013] T s is the sampling time, ρ is the density of the film, b is the width, T(k) represents the tension measurement value at the current moment k; the intermediate variable Intermediate variables J total (k) is the total moment of inertia at the current k moment, M FA (k) is the friction torque at the current moment k:

[0014] Define parameter vector θ = [M FA ,J total ] T and regression vector φ(k)=[φ1(k),φ2(k)] T , the tension model of the rewinding and unwinding system is simplified to: z(k)=φ T (k)θ+e(k), and estimate the parameter vector θ=[M FA ,J total ] T .

[0015] Preferably, the moment of inertia feedforward term is:

[0016]

[0017] ω(k) represents the angular velocity of the unwinding shaft at the current moment k;

[0018] The friction compensation term is:

[0019] M friction (k) = M FA (k)·sgn(ω)

[0020] in,

[0021] As a preference, the final control amount is:

[0022] M cmd (k) = M PID (k)+η J ·M inertia (k)+η F ·M friction (k)

[0023] in, K P is the scale parameter, K I is the integration parameter, K D is the differential parameter, weight η J =1.2-0.4·tanh(|Δr A (k)| / r max ), weight η F =0.8+0.3·sat(|ω(k)| / ω nom ), coil diameter change rate Δr A =|r A (k)-r A (k-1)|,r max Indicates the maximum coil diameter, ω nom is the rated angular velocity, and ω(k) represents the angular velocity of the unwinding shaft at the current moment k.

[0024] Preferably, S2 includes:

[0025] Constructing a fuzzy PID controller: The input variable of the fuzzy PID controller is e T 、 The output variable is: proportional increment ΔK P ′、Integral increment ΔK I ′, differential increment ΔK D ';

[0026] Determine the fuzzy set of the input variable at the current k moment according to the domain of the input variable at the current k moment, and use the genetic algorithm to obtain the fuzzy set of the output variable at the current k moment. The corresponding relationship between the fuzzy set of the input variable and the fuzzy set of the output variable constitutes the optimal fuzzy rule base at the current k moment.

[0027] In the genetic algorithm, the fitness function is:

[0028]

[0029] Tension overshoot t s is the adjustment time to enter the target tension ±2% error band, w1 and w2 are weight coefficients, T ref Indicates the set tension, max(T) indicates the maximum value of the tension during the control process at the time of initial control.

[0030] As a preference, in the genetic algorithm, a tournament selection strategy is adopted, and the tournament size is 5-10% of the population;

[0031] The crossover operation is a two-point crossover operation with a crossover probability p c ∈[0.75,0.90];

[0032] The mutation operation is a single-point mutation operation, and the mutation probability p m ∈[0.005,0.03];

[0033] Population size ≥ 100, number of iterations ≥ 200;

[0034] When the fitness change rate for 10 consecutive generations is less than 10 -4 When evolution ends;

[0035] Elite retention ratio ≥5%.

[0036] As a preference, the domain of the current k moments of the input variable for:

[0037]

[0038] β is the scaling gain, e max is the maximum error threshold.

[0039] As a preference, in S2, the real-time adjustment amount ΔK of the PID parameter P for:

[0040]

[0041] Where ΔK P ′ is the proportional increment of the fuzzy PID controller output.

[0042] Preferably, S2 includes:

[0043] When e T (k) Obtaining real-time adjustment of PID parameters through fuzzy PID controller within the scope of input variable domain;

[0044] when|e T (k)|>30%·T refWhen the proportional increment ΔK is significantly increased P To improve response speed;

[0045] when When , increase the differential increment to suppress overshoot;

[0046] when|e T (k)|<5%·T ref When , the differential increment dominates to eliminate the steady-state error.

[0047] The beneficial effects of the present application are as follows: the present application is a dynamic tension control method for a dry winding and unwinding system based on fuzzy adaptive PID. In view of the PID parameter mismatch problem caused by the nonlinear fluctuation of the time-varying inertia of the roll diameter in the prior art, the total rotational inertia and friction torque are identified online through the recursive least squares method to realize feedforward compensation; in view of the defect of insufficient dynamic adaptability of the fixed rule fuzzy controller, the genetic algorithm is used to optimize the fuzzy rule base and the variable domain strategy is implanted to significantly improve the overshoot suppression capability under high-speed working conditions; in view of the steady-state tension accuracy mismatch caused by the low-speed friction dead zone, an error direction adaptive gain adjustment mechanism is constructed; the present application solves the problems in the prior art such as high overshoot under high-speed reversing working conditions, long adjustment time for sudden change of roll diameter, and large steady-state fluctuation under low-speed operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a schematic diagram of the winding and unwinding unit of the winding device in this application;

[0049] Figure 2 This is a flow chart of the dynamic tension control method of the dry winding and unwinding system in this application;

[0050] Figure 3 It is an iterative flow chart of the recursive least squares method with forgetting factor in this application;

[0051] Figure 4 This is a flow chart of the algorithm for obtaining the optimal fuzzy rule base by genetic algorithm in this application;

[0052] Figure 5 It is the calculation flow chart of the fuzzy PID algorithm in this application;

[0053] Figure 6 It is a control flow chart of superimposed feedforward control in this application. DETAILED DESCRIPTION

[0054] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of this application.

[0055] It should be noted that, unless there is any conflict, the embodiments and features in the embodiments of this application can be combined with each other.

[0056] The present application will be further described below with reference to the accompanying drawings and specific embodiments, but they are not intended to limit the present application.

[0057] The dynamic tension control method for the dry winding and unwinding system of this embodiment first identifies the total moment of inertia and friction torque of the unwinding core in real time. The input signals include the motor output torque, real-time rotation speed, dynamic winding diameter, and tension sensor measurement value. Then, a fuzzy PID controller is constructed. The deviation between the actual tension value and the set value and its rate of change are input into the fuzzy system, and the adjustment of the proportional gain, integral gain, and differential gain is output in real time and the PID parameters are dynamically updated. Finally, the moment of inertia feedforward compensation and the friction torque compensation are integrated to generate a feedforward control component, which is combined with the feedback component output by the adaptive PID controller to synthesize the final control torque of the motor. Specifically, it includes:

[0058] Step 1: Collect the data of the rewinding and unwinding system, identify the total moment of inertia and friction torque at the current time k online based on the data, and obtain the friction compensation M based on the friction torque at the current time k. friction (k), the moment of inertia M is obtained based on the total moment of inertia at the current k moment inertia (k);

[0059] Specifically, the recursive least squares method can be used to identify the total moment of inertia and friction torque of the unwinding core in real time:

[0060] Figure 1 The following is a schematic diagram of the winding and unwinding unit of the winding equipment. First, the system dynamics model is established based on the winding tension model:

[0061]

[0062] Where: T: real-time tension measurement value; M A : Motor output torque; M FA : Friction torque (to be identified); J total : total moment of inertia (to be identified); ω: angular velocity (ω=v A / r A );ρ,b: film density and width.

[0063] For the discretization of system dynamics model, the time term is discretized (sampling period T s ):

[0064]

[0065] Substituting the angular velocity relationship ω(k) = v A (k) / r A (k), we get the discretized tension model:

[0066] z(k)=φ1(k)M FA +φ2(k)J total +e(k)

[0067] in:

[0068]

[0069] Derivation of the recursive least squares method

[0070] (1) Linear regression form

[0071] Define parameter vector θ = [M FA ,J total ] T and regression vector φ(k)=[φ1(k),φ2(k)] T , the model is simplified to:

[0072] z(k)=φ T (k)θ+e(k)(3)

[0073] (2) Cost function

[0074] Minimize the weighted sum of squared errors of historical data:

[0075]

[0076] Where λ∈(0,1] is the forgetting factor (the default λ=0.98).

[0077] (3) Derivation of the iterative formula of recursive least squares method

[0078] Covariance matrix update:

[0079]

[0080] Gain matrix calculation:

[0081]

[0082] Parameter estimate update:

[0083]

[0084] (4) Initialization conditions

[0085]

[0086] P(0)=αI(α=10 6 , I is a 2×2 identity matrix)

[0087] After updating, the total moment of inertia J at the current k moment is finally identified total (k) and the friction torque M at the current k moment FA (k).

[0088] Step 2: The tension error e of the rewinding and unwinding system at the current time k is T (k) and tension error change rate Input to the fuzzy PID controller to obtain the real-time adjustment of the PID parameters. Update the PID parameters according to the real-time adjustment of the PID parameters. Then, obtain the PID feedback control amount M at the current k moment according to the tension error and PID parameters. PID (k);

[0089] First, a fuzzy PID controller is constructed. The fuzzy rule base of the fuzzy PID controller can be obtained through genetic algorithm:

[0090] The performance index is defined to quantify the control performance of the fuzzy rule base. This implementation adopts two key dynamic response indicators:

[0091] Tension overshoot OS%:

[0092]

[0093] Reflects the overshoot degree of the system transient response.

[0094] Adjustment time t s : Tension first enters and remains at T ref The time within the ±2% error band (unit: seconds) represents the system stability.

[0095] The fitness function is constructed by combining the above indicators:

[0096]

[0097] Weight coefficient range: w1∈[0.5,0.7] (overshoot weight), w2∈[0.3,0.5] (adjustment time weight)

[0098] The Pareto optimal solution verified by orthogonal test is the default parameters: w1=0.6, w2=0.4

[0099] Physical meaning: A larger Fitness value indicates better rule base performance (smaller overshoot and faster response), and the maximum value approaches 1.

[0100] (2) Chromosome coding design

[0101] Each chromosome encodes a complete fuzzy rule base, which is composed of the following gene segments in series:

[0102]

[0103] Input variables Each is divided into 7 fuzzy sets: negative large (NB), negative medium (NM), negative small (NS), zero (ZO), positive small (PS), positive medium (PM), positive large (PB). The boundary position is coded with real numbers (such as e T NB / PM boundary value).

[0104] Output variable (proportional increment ΔK P , integral increment ΔK I , differential increment ΔK D ): Each is divided into 7 fuzzy sets: NB, NM, NS, ZO, PS, PM, PB, and integer coding (1-7).

[0105] Total chromosome length:

[0106] L total =5×2+25×3×log2(7)≈5×2+75×3=235 bits(10)

[0107] (3) Genetic manipulation process

[0108] Genetic operations first require the generation of an initial population, with a population size of N ≥ 100.

[0109] The initialization method is:

[0110] Input is the domain range of the input variable: in [-30% T ref ,30%T ref ] Randomly generate boundary values ​​within .

[0111] Rule table gene: uniformly randomly generate integers from 1 to 7.

[0112] After initialization, a tournament selection operation is required, where the tournament size is: k = 8% × N (the default population size is 8%), and the selection probability expression is:

[0113]

[0114] After selection, a crossover operation is performed, where the crossover probability is: p c ∈[0.75,0.90](default pc =0.85)

[0115] After crossover, a single-point mutation operation is performed, and the mutation probability is: p m ∈[0.005,0.03](default p m =0.01)

[0116] Variation method:

[0117] Input partition gene: add Gaussian noise to boundary value

[0118] Rule table gene: p m The probability is randomly reset to an integer from 1 to 7

[0119] (4) Evolutionary Termination and Elite Retention

[0120] The termination conditions must meet both of the following conditions:

[0121] Iterations ≥ 200

[0122] The fitness change rate for 10 consecutive generations is δ<10 -4 :

[0123]

[0124] The elite retention strategy is:

[0125] Elite ratio: r elite ≥5% (default r elite =6%)

[0126] Execution mechanism: Each generation directly copies the r with the highest fitness elite ×N individuals to the next generation.

[0127] According to the domain of the input variable at the current k moment, the fuzzy set of the input variable at the current k moment is determined, and the proportional increment ΔK is obtained using the above genetic algorithm. P , integral increment ΔK I , differential increment ΔK D The fuzzy set at the current k moment, the final proportional increment ΔK P , integral increment ΔK I , differential increment ΔK D The fuzzy rule base is shown in Tables 1 to 3;

[0128] Table 1: Proportional increment ΔK P Fuzzy rule base (expressed in table form)

[0129]

[0130] Design idea: ΔK PMainly focus on the current error e T When the error is large, a larger proportional gain should be used to speed up the response; when the error is small, the proportional gain should be reduced to prevent overshoot. Error change rate It has a fine-tuning effect on it. For example, when the error changes in the direction of reduction, the proportional gain can be appropriately reduced. As shown in Table 1, when the error is large (e is NB or PB), ΔK P Taking a larger positive value (PB, PM) is in line with the strategy of "large error zone"; when the error is near zero (e is ZO), ΔK P Take a smaller or negative value to reduce the P effect and avoid overshoot.

[0131] Table 2: Integral increment ΔK I Fuzzy rule base (expressed in table form)

[0132]

[0133] Design idea: ΔK I It is mainly used to eliminate steady-state errors, so it is only used when the error is small (e T When the error is large, the integral action should be limited or even cleared to prevent integral saturation. As shown in Table 2, when the error e is near ZO, NS, and PS, ΔK I Only when the error e is large (NB or PB), ΔK I Taking the value of ZO or the reverse direction effectively suppresses integral saturation.

[0134] Table 3: Differential increment ΔK D Fuzzy rule base

[0135]

[0136] Design idea: ΔK D Mainly focus on the changing trend of error Used to predict and suppress overshoot. When the error changes drastically, a large differential action should be taken; when the system tends to be stable (e T When the error rate of change |e| is large (NB, NM, PB, PM columns), ΔK D A larger value is used to enhance the system damping. When the error approaches the set value quickly (e.g. T is PB, Te is NB), ΔK D Take the maximum positive value (PB),

[0137] It plays a strong "brake" role to prevent overshoot. When the system is stable ( When ZO), ΔK DTo avoid unnecessary adjustments and noise interference.

[0138] The meaning of 4 fuzzy sets

[0139]

[0140]

[0141] Determine the optimal fuzzy rule base of the fuzzy PID controller to dynamically adjust the ratio of the PID controller in real time (K P ), integral(K I ), differential (K D ) parameters to cope with the nonlinear time-varying characteristics of the rewinding and unwinding system. Specifically:

[0142]

[0143] Fuzzy reasoning mechanism, the input fuzzy set is divided into:

[0144]

[0145] Membership function: Gaussian distribution

[0146]

[0147] where c A is the fuzzy set center, σ A is the width (determined by optimization in step 2).

[0148] The defuzzification method uses the centroid method to calculate the exact output value:

[0149]

[0150] This embodiment further proposes a variable universe adaptive strategy based on the above fuzzy reasoning mechanism, specifically:

[0151] According to the domain of discourse at k-1 moment and e T (k) Update the domain of the input variable at the current k moment

[0152]

[0153] β is the scaling gain, e max is the maximum error threshold.

[0154] The parameters in the formula are set as follows: Scaling gain β = 0.8 Maximum error threshold: e max =20%·T ref The physical meaning of this setting is as follows: When |e T |>20%T refWhen , the domain of discussion is expanded to 1.8 times, which enhances the control sensitivity in the large error range.

[0155] Based on the above fuzzy inference mechanism, this embodiment further proposes output gain direction adaptation to achieve dynamic adjustment of control variables, specifically:

[0156] Error trend determination is required during adjustment: Sign function Indicates the direction of error change: +1: the error is increasing (need to strengthen control); -1: the error is decreasing (need to weaken control). The final output proportional gain is adjusted to:

[0157]

[0158] Where ΔK P ′ is the proportional increment of the fuzzy PID controller output.

[0159] When the error increases, the actual ΔK P Increase by 20%; when the error decreases, ΔK P Reduced by 20%.

[0160] This embodiment adopts the above method to perform fuzzy PID control when the error state is within the range of the domain, and adopts the following adaptive priority when the error state is not within the range of the domain:

[0161]

[0162] e T (k) Input the PID controller to determine the final PID parameters and obtain the generated PID control quantity:

[0163]

[0164] The following preprocessing is required before outputting the control quantity: Integral anti-saturation: When |e T |>20%T ref Freeze the integral term; Differential filtering: add a first-order low-pass filter (cut-off frequency 50Hz) to suppress noise.

[0165] Step 3: M friction (k), M inertia (k), M PID (k) The final control quantity of the synthetic winding and unwinding system.

[0166] The rotational inertia feedforward term and friction compensation term generated in step 1 are dynamically synthesized to generate the final motor control torque M cmd , to solve the time-varying inertia and nonlinear friction disturbance problems of the winding and unwinding system.

[0167] The moment of inertia feedforward term is based on Newton's second law and compensates for the inertia torque caused by angular acceleration:

[0168]

[0169] Discretize the moment of inertia to obtain:

[0170]

[0171] Where T s The sampling period is 10ms by default. Feedforward control can offset the fluctuation of the moment of inertia caused by the change of the coil diameter.

[0172] The friction compensation term uses a symbolic function model to compensate for Coulomb friction:

[0173] M friction =M FA ·sgn(ω)(20)

[0174] In the above formula, the symbolic function is defined as:

[0175]

[0176] The dead zone processing should be performed for the friction compensation item as follows: when |ω|≤0.5° / s, stop compensation to avoid chattering.

[0177] Add the amount of feedforward control to generate a composite control signal:

[0178] M cmd (k) = M PID (k)+M inertia (k)+M friction (k)(22)

[0179] Among them, M PID (k) is feedback control, M inertia (k) is the inertia feedforward, M friction (k) is friction compensation

[0180] Physical constraints: Limiting is required before output control |M cmd |≤M max (M max =150%·M N )M N is the rated torque of the motor.

[0181] This implementation adopts a dynamic weight adjustment strategy to dynamically adjust the feedforward weight according to the system state:

[0182]

[0183] Parameter description: Δr A =|r A (k)-r A (k-1)|: coil diameter change rate; r max =10%;ω nom : Rated angular velocity

[0184] sat(x)=min(max(x,0),1): saturation function.

[0185] According to the dynamic weight adjustment strategy, the weight adjustment logic can be obtained as shown in the following table:

[0186]

[0187] This embodiment uses recursive least squares to identify the unwinding core's moment of inertia and friction torque in real time. The input signals include the motor output torque, real-time speed, dynamic reel diameter, and tension sensor measurements. A genetic algorithm is then used to optimize the fuzzy rule base, using the tension overshoot percentage and the adjustment time to enter the target error band as fitness evaluation criteria. The optimal fuzzy rule set is generated through a tournament selection mechanism, two-point crossover operations, and single-point mutation operations. A fuzzy inference controller is then constructed based on the optimized rule set. The deviation between the actual tension value and the set value and its rate of change are input into the fuzzy system. Adjustments to the proportional gain, integral gain, and differential gain are output in real time, and the PID parameters are dynamically updated. Finally, the moment of inertia feedforward compensation and friction torque compensation are integrated to generate a feedforward control component, which is then combined with the feedback component output by the adaptive PID controller to synthesize the motor's final control torque. This application improves tension control accuracy and collaboratively addresses the problems of traditional control methods such as high overshoot in high-speed commutation conditions, long reel diameter mutation adjustment time, and large steady-state fluctuations in low-speed operation.

[0188] Although the present application is described herein with reference to specific embodiments, it should be understood that these embodiments are merely illustrative of the principles and applications of the present application. It should be understood that many modifications may be made to the illustrative embodiments, and that other arrangements may be devised, without departing from the spirit and scope of the present application as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in ways other than those described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in conjunction with other described embodiments.

Claims

1. A dynamic tension control method for a dry winding and unwinding system, characterized in that: include: S1. Collect the data of the rewinding and unwinding system, identify the total moment of inertia and friction torque at the current k moment online based on the data, and obtain the friction compensation term M based on the friction torque at the current k moment. friction (k), the moment of inertia feedforward term M is obtained based on the total moment of inertia at the current k moment inertia (k); S2, the tension error e of the rewinding and unwinding system at the current k moment T (k) and tension error change rate Input to the fuzzy PID controller to obtain the real-time adjustment of the PID parameters. Update the PID parameters according to the real-time adjustment of the PID parameters. Then, obtain the PID feedback control amount M at the current k moment according to the tension error and PID parameters. PID (k); S3, M friction (k), M inertia (k), M PID (k) The final control quantity of the synthetic winding and unwinding system.

2. The dynamic tension control method of the dry winding and unwinding system according to claim 1, characterized in that: S1 includes: Collect data from the rewinding and unwinding system, including the braking torque M of the unwinding motor at the current time k A (k), the reel diameter r of the reel at the current k moment A (k), the reel diameter r of the unwinding shaft at time k-1 A (k-1), v A (k) is the linear velocity of the unwinding shaft at the current k moment, and the linear velocity of the unwinding shaft at k-1 moment v A (k-1); The discretized tension model of the rewinding and unwinding system is: z(k)=φ1(k)M FA (k)+φ2(k)J total (k)+e(k) Among them, z(k) represents the observed value, and e(k) represents the error between the observed value and the predicted value; T s is the sampling time, ρ is the density of the film, b is the width, and T(k) represents the tension measurement value at the current moment k; Intermediate variables Intermediate variables J total (k) is the total moment of inertia at the current k moment, M FA (k) is the friction torque at the current moment k: Define parameter vector θ = [M FA ,J total ] T and regression vector φ(k)=[φ1(k),φ2(k)] T , the tension model of the rewinding and unwinding system is simplified to: z(k)=φ T (k)θ+e(k), and estimate the parameter vector θ=[M FA ,J total ] T .

3. The dynamic tension control method of the dry winding and unwinding system according to claim 2, characterized in that: The moment of inertia feedforward term is: ω(k) represents the angular velocity of the unwinding shaft at the current moment k; The friction compensation term is: M friction (k)=M FA (k)·sgn(ω) in, 4. The dynamic tension control method of the dry winding and unwinding system according to claim 3, characterized in that: The final control amount is: M cmd (k)=M PID (k)+η J ·M inertia (k)+η F ·M friction (k) in, K P is the scale parameter, K I is the integration parameter, K D is the differential parameter, weight η J =1.2-0.4·tanh(|Δr A (k)| / r max ), weight η F =0.8+0.3·sat(|ω(k)| / ω nom ), coil diameter change rate Δr A =|r A (k)-r A (k-1)|,r max Indicates the maximum coil diameter, ω nom is the rated angular velocity, and ω(k) represents the angular velocity of the unwinding shaft at the current moment k.

5. The dynamic tension control method of the dry winding and unwinding system according to claim 1, characterized in that S2 include: Constructing a fuzzy PID controller: The input variable of the fuzzy PID controller is e T 、 The output variable is: proportional increment ΔK P ′、Integral increment ΔK I ′, differential increment ΔK D '; Determine the fuzzy set of the input variable at the current k moment according to the domain of the input variable at the current k moment, and use the genetic algorithm to obtain the fuzzy set of the output variable at the current k moment. The corresponding relationship between the fuzzy set of the input variable and the fuzzy set of the output variable constitutes the optimal fuzzy rule base at the current k moment. In the genetic algorithm, the fitness function is: Tension overshoot t s is the adjustment time to enter the target tension ±2% error band, w1 and w2 are weight coefficients, T ref Indicates the set tension, max(T) indicates the maximum value of the tension during the control process at the time of initial control.

6. The dynamic tension control method of the dry winding and unwinding system according to claim 5, characterized in that: In the genetic algorithm, a tournament selection strategy is adopted, and the tournament size is 5-10% of the population; The crossover operation is a two-point crossover operation with a crossover probability p c ∈[0.75,0.90]; The mutation operation is a single-point mutation operation, and the mutation probability p m ∈[0.005,0.03]; Population size ≥ 100, number of iterations ≥ 200; When the fitness change rate for 10 consecutive generations is less than 10 -4 When evolution ends; Elite retention ratio ≥5%.

7. The dynamic tension control method of the dry winding and unwinding system according to claim 5, characterized in that: The universe U of the input variable at the current k moment eT (k) is: YOU eT (k)=U eT (k-1)·[1+β·tanh(|e T (k)| / e max )] β is the scaling gain, e max is the maximum error threshold.

8. The dynamic tension control method of the dry winding and unwinding system according to claim 5, characterized in that: In S2, the real-time adjustment value of PID parameter ΔK P for: Where ΔK P ′ is the proportional increment of the fuzzy PID controller output.

9. The dynamic tension control method of the dry winding and unwinding system according to claim 5, characterized in that: The S2 includes: when e T (k) Obtaining real-time adjustment of PID parameters through fuzzy PID controller within the scope of input variable domain; when|e T (k)|>30%·T ref When the proportional increment ΔK is significantly increased P To improve response speed; when When , increase the differential increment to suppress overshoot; when|e T (k)|<5%·T ref When , the differential increment dominates to eliminate the steady-state error.

10. A dynamic tension control device for a dry winding and unwinding system, comprising a storage device, a processor, and a computer program stored in the storage device and executable on the processor, characterized in that: The processor executes the computer program to implement the steps of the dynamic tension control method for the dry winding and unwinding system according to any one of claims 1 to 9.

Images