Lithium ion battery diaphragm cutting and positioning device and method

Abstract

The invention discloses a lithium ion battery diaphragm cutting and positioning device and method, and the method comprises the steps: collecting the multi-modal sensing data of a diaphragm relaxation region in real time after a diaphragm is cut, constructing the three-dimensional form of the diaphragm relaxation region through data fusion based on the multi-modal sensing data, and calculating the basic length compensation amount of the diaphragm according to the three-dimensional form; analyzing the three-dimensional form, and identifying the wrinkle type and characteristic parameters of the diaphragm; calculating a wrinkle correction amount, and combining the basic length compensation amount to obtain a total length compensation amount; based on the total length compensation amount, diaphragm positioning is completed, and the diaphragm is cut off in the next period; the dimension transition of compensation amount calculation from one-dimensional line estimation to three-dimensional surface integral is realized, and the problem of fixed-length dimension fluctuation is effectively solved.

Description

Technical Field

[0001] This invention relates to a method for cutting and positioning a lithium-ion battery separator, and more particularly to a lithium-ion battery separator cutting and positioning device and method. Background Technology

[0002] In the manufacturing process of lithium-ion batteries, the precise cutting and positioning of the separator is one of the key processes affecting battery consistency and safety performance. As a porous polymer film, the separator is prone to complex sagging and wrinkling in the width direction during transportation due to its flexibility and uneven tension distribution. These local deformations cause the actual length of the separator on the transportation path to deviate from the theoretical length. If this cannot be accurately compensated, it will directly affect the accuracy of the subsequent cutting position, resulting in poor electrode assembly or even the risk of internal short circuits in the battery.

[0003] Currently, most cutting and positioning devices used in the industry rely on single-point or single-line measurement methods, such as using laser rangefinders or vision sensors to detect sag height at the center or a fixed position along the width of the separator. However, such methods can only reflect the local sag at the measurement point or line, and cannot fully capture the complex three-dimensional deformation distribution of the separator across its entire width due to factors such as wrinkles and stress concentration. Since the wrinkles and sag of the separator are often non-uniform along the width, the data obtained from traditional single-point measurements cannot accurately reflect the actual relaxation and extra length of the separator during transport, leading to deviations in subsequent length compensation calculations and affecting the accuracy of cutting and positioning. Therefore, there is an urgent need for a cutting and positioning method and device that can achieve full-field deformation detection and precise length compensation of the separator to improve the overall precision and reliability of battery manufacturing. Summary of the Invention

[0004] This invention overcomes the shortcomings of the prior art and provides a lithium-ion battery separator cutting and positioning device and method.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is: a method for cutting and positioning a lithium-ion battery separator, comprising the following steps:

[0006] S1: After the diaphragm is cut, multimodal sensing data of the diaphragm relaxation area are acquired in real time.

[0007] Among them, the multimodal sensing data includes laser ranging data and ultrasonic ranging data from multiple measurement points distributed along the width of the diaphragm;

[0008] S2: Based on multimodal sensing data, the three-dimensional morphology of the diaphragm relaxation region is constructed through data fusion, and the basic length compensation of the diaphragm is calculated accordingly.

[0009] S3: Analyze the three-dimensional morphology and identify the fold type and characteristic parameters of the diaphragm;

[0010] S4: Calculate the wrinkle correction amount based on the identified wrinkle type and feature parameters, and combine it with the basic length compensation amount to obtain the total length compensation amount;

[0011] S5: Based on the total length compensation, perform two-stage compensation control on the diaphragm to complete the diaphragm positioning and proceed with the next cycle of diaphragm cutting.

[0012] The two-stage compensation control includes: in the first stage, controlling the drive mechanism to move rapidly to eliminate the main slack; in the second stage, controlling the drive mechanism to move slowly to smoothly increase the diaphragm tension to the target value.

[0013] In a preferred embodiment of the present invention, in S1, the acquisition of multimodal sensing data includes: acquiring laser ranging data and ultrasonic ranging data for each measurement point;

[0014] The signal strength validity of laser ranging data is verified, and the signal-to-noise ratio validity of ultrasonic ranging data is verified.

[0015] For data that passes the validity verification, weighted fusion is performed according to the preset confidence level weights to obtain the fused height data of the measurement point.

[0016] In a preferred embodiment of the present invention, step S2 further includes: constructing a three-dimensional curved surface of the diaphragm relaxation region based on the fused height data of each measurement point at different times;

[0017] Calculate the actual surface area of ​​a three-dimensional curved surface and its projected area under ideal flat conditions;

[0018] Calculate the length expansion coefficient based on the actual surface area and the projected area;

[0019] The basic length compensation is calculated based on the coefficient of length expansion, the theoretical length of the diaphragm, and the tensile coefficient of the material.

[0020] In a preferred embodiment of the present invention, S3 further includes: extracting a height profile sequence of the three-dimensional morphology in the width direction of the diaphragm, performing spectral analysis on each profile, determining whether there are transverse wrinkles based on the energy of the high-frequency components, and extracting their wavelength characteristics.

[0021] The height time series of each measurement point along the diaphragm conveying direction is extracted, and spectrum analysis is performed. Based on the energy in a specific frequency band, it is determined whether there are longitudinal wrinkles and their main frequency characteristics are extracted.

[0022] The type of fold is determined by comprehensively considering the presence of both horizontal and vertical folds.

[0023] In a preferred embodiment of the present invention, step S4 further includes step S41: performing fault diagnosis and cause analysis based on the wrinkle features identified in step S3;

[0024] Fault diagnosis and cause analysis include: determining the potential equipment fault type based on the characteristic pattern of the wrinkles and calculating the equipment health index; if the equipment health index is normal, assessing the matching degree between the current diaphragm material characteristics and the system control parameters, and adaptively adjusting the control parameters when the matching degree is lower than the threshold.

[0025] In a preferred embodiment of the present invention, in S5, the first stage adopts a position control mode, and the motion target of the drive mechanism is a part of the total length compensation amount;

[0026] The second stage employs a torque control mode based on tension feedback, where the controller parameters are adaptively adjusted according to the type and severity of wrinkles identified by S3.

[0027] In a preferred embodiment of the present invention, in S4, calculating the wrinkle correction amount includes: if there are transverse wrinkles, calculating a first correction component based on their wavelength characteristics and severity;

[0028] If longitudinal wrinkles exist, the second correction component is calculated based on their main frequency characteristics, diaphragm delivery speed, and severity.

[0029] The total length compensation is the sum of the basic length compensation and the wrinkle correction, and the wrinkle correction is the weighted sum of the first correction component and the second correction component.

[0030] In a preferred embodiment of the present invention, the adaptive parameter adjustment in the second stage includes: when the fold type is mainly horizontal folds, adjusting the proportional, integral and derivative parameters of the controller according to the first rule;

[0031] When the wrinkle type is predominantly longitudinal, adjust the proportional, integral, and derivative parameters of the controller according to the second rule.

[0032] When there are no wrinkles, use the nominal control parameters.

[0033] A lithium-ion battery separator cutting and positioning device, comprising:

[0034] The sensor fusion and detection module is used to acquire laser and ultrasonic multimodal sensing data after the diaphragm is cut, and to perform data fusion processing.

[0035] The 3D reconstruction and compensation calculation module is used to reconstruct the 3D relaxation morphology of the diaphragm based on the fused data and calculate the basic length compensation amount.

[0036] The wrinkle feature extraction and recognition module is used to analyze three-dimensional morphology and identify the type, severity, and feature parameters of wrinkles.

[0037] The intelligent diagnosis and parameter adjustment module is used to diagnose equipment faults and evaluate material properties based on wrinkle characteristics, and adaptively adjust system control parameters.

[0038] The two-stage compensation control module is used to generate two-stage control commands based on the basic length compensation amount, the wrinkle correction amount, and the adjusted control parameters, driving the actuator to complete the precise compensation and positioning of the diaphragm.

[0039] In a preferred embodiment of the present invention, the sensor fusion and detection module includes: a plurality of measurement units arranged at equal intervals along the width direction of the diaphragm, each measurement unit including a laser displacement sensor and an ultrasonic ranging sensor; and a synchronous acquisition unit, used to synchronously trigger and acquire the measurement data of all sensors within a set time period after receiving a cut-off signal.

[0040] This invention addresses the shortcomings of the prior art and has the following beneficial effects:

[0041] (1) This invention provides a lithium-ion battery separator cutting and positioning device and method. By fusing data through a heterogeneous sensor array, the true three-dimensional curved surface morphology of the loose area of ​​the separator after cutting is reconstructed. By calculating the ratio of the true surface area to the projected area of ​​the curved surface, the basic length compensation amount required to eliminate wrinkles and sagging in all directions is accurately calculated. Furthermore, the system deeply identifies the type and quantitative characteristics of wrinkles through spectrum analysis and calculates the targeted dynamic wrinkle correction amount accordingly. The two are fused to form the final total length compensation amount. In the prior art, traditional single-point or single-line measurement can only reflect the local sagging height and cannot capture the extra length caused by the complex distribution of wrinkles in the width direction of the separator. However, this invention treats the separator as a three-dimensional entity, whose true length is equal to the path integral along its complex surface contour. This realizes the dimensional leap from one-dimensional line estimation to three-dimensional area integral in the compensation amount calculation, effectively solving the problem of fixed-length dimension fluctuation.

[0042] (2) The present invention provides a lithium-ion battery separator cutting and positioning device and method. By using an embedded model, it can determine whether the potential root cause is a mechanical failure of the equipment or a mismatch between the process parameters and the current material properties. Based on this, it can automatically trigger a maintenance alarm or adaptively adjust the control parameters. In the prior art, it can only passively compensate for the error in the current time, but cannot prevent the continuous effect of the error source. The present invention can detect hidden faults such as slight deviation of the guide roller in advance through real-time feature extraction and diagnosis, or automatically optimize the control parameters when a new batch of materials is used, so that the system always runs near the optimal state, and prevents the systematic size drift caused by a single small deviation being continuously reproduced and amplified in thousands of cutting cycles.

[0043] (3) The present invention provides a lithium-ion battery separator cutting and positioning device and method. By adopting a two-stage adaptive servo control strategy and dynamically adjusting the controller parameters according to the real-time identified wrinkle type, the system first efficiently eliminates the main slack through the two-stage division, so that the separator is safely close to the tension critical state, and completely avoids the internal microstructure stretching damage or local stress concentration tearing caused by the servo motor moving too fast or too violently. Compared with the single speed or single mode tensioning method commonly used in the prior art, the present invention protects the physical integrity of the separator as a fragile thin film material through two-stage tensioning, and fundamentally improves the safety and consistency of the battery. Attached Figure Description

[0044] To more clearly illustrate the technical solutions 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 recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart of a preferred embodiment of the present invention.

[0046] Figure 2 This is a logic diagram of the preferred embodiment of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein. Therefore, the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0049] In the description of this application, it should be understood that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting the scope of protection of this application. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, features defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0050] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this application based on the specific circumstances.

[0051] As shown in the figure, a method for cutting and positioning a lithium-ion battery separator includes the following steps:

[0052] S1: After the diaphragm is cut, multimodal sensing data of the diaphragm relaxation area are acquired in real time.

[0053] Among them, the multimodal sensing data includes laser ranging data and ultrasonic ranging data from multiple measurement points distributed along the width of the diaphragm;

[0054] S2: Based on multimodal sensing data, the three-dimensional morphology of the diaphragm relaxation region is constructed through data fusion, and the basic length compensation of the diaphragm is calculated accordingly.

[0055] S3: Analyze the three-dimensional morphology and identify the fold type and characteristic parameters of the diaphragm;

[0056] S4: Calculate the wrinkle correction amount based on the identified wrinkle type and feature parameters, and combine it with the basic length compensation amount to obtain the total length compensation amount;

[0057] S5: Based on the total length compensation, perform two-stage compensation control on the diaphragm to complete the diaphragm positioning and proceed with the next cycle of diaphragm cutting.

[0058] The two-stage compensation control includes: in the first stage, controlling the drive mechanism to move rapidly to eliminate the main slack; in the second stage, controlling the drive mechanism to move slowly to smoothly increase the diaphragm tension to the target value.

[0059] It should be noted that, after each cut of the diaphragm, this invention simultaneously acquires ranging data of the relaxed region using an ultrasonic-laser heterogeneous sensor array arranged along its width. The system first performs confidence assessment and weighted fusion of the two types of data, effectively overcoming the inherent defects of optical or acoustic detection due to the transparency and reflectivity of the diaphragm material, thereby obtaining reliable information on the diaphragm surface height. Based on this spatiotemporal data, the system reconstructs the three-dimensional morphological surface of the relaxed region of the cut diaphragm, and calculates the basic length compensation required to eliminate sagging by calculating the ratio of the actual surface area to the projected area of ​​this three-dimensional surface, i.e., the length expansion coefficient.

[0060] Furthermore, to address wrinkling issues caused by uneven tension, equipment deviations, or material property fluctuations, the system performs in-depth analysis of the three-dimensional morphology. Through spectral analysis of the width-direction cross-sectional sequence, transverse wrinkles are detected and their wavelength characteristics are extracted; through spectral analysis of the height time series at each measurement point, longitudinal fluctuations are detected and their dominant frequency characteristics are extracted. This spectral feature extraction method based on three-dimensional morphology can accurately distinguish the type and severity of wrinkles. The system calculates dynamic correction amounts related to wrinkle type and severity using these quantified features, integrating them with basic compensation amounts to form the final total accuracy compensation amount; secondly, it serves as the basis for intelligent diagnosis. Furthermore, by analyzing the characteristic patterns of wrinkles, i.e., wavelength or frequency, the system determines whether the potential cause is equipment malfunction (non-parallel cutter, skewed guide roller) or material parameter mismatch, and accordingly triggers maintenance alarms or adaptively adjusts control parameters.

[0061] Ultimately, the system employs a two-stage adaptive servo control strategy to perform compensation. In the first stage, the servo motor rotates rapidly in position control mode, eliminating most of the slack and bringing the diaphragm to a near-tightening critical state. In the second stage, the system switches to torque control mode based on real-time tension feedback and dynamically adjusts the controller parameters according to the previously identified wrinkle types. For example, it enhances the integral action to uniform tension when dealing with transverse wrinkles and enhances the derivative action to suppress oscillations when dealing with longitudinal fluctuations, thereby smoothly and accurately raising the diaphragm tension to the set value, completing the fixed-length tensioning and positioning. This method and device systematically solve the problems of diaphragm damage, unstable fixed-length dimensions, and error accumulation caused by inaccurate compensation in traditional processes.

[0062] S1: After the diaphragm is cut, multimodal sensing data of the diaphragm relaxation area is collected in real time. The multimodal sensing data includes laser ranging data and ultrasonic ranging data of multiple measurement points distributed along the width direction of the diaphragm.

[0063] In this invention, in step S1, the acquisition of multimodal sensing data includes: acquiring laser ranging data and ultrasonic ranging data for each measurement point; verifying the signal strength validity of the laser ranging data and verifying the signal-to-noise ratio validity of the ultrasonic ranging data; and for the data that passes the validity verification, performing weighted fusion according to a preset confidence weight to obtain the fused height data of the measurement point.

[0064] It should be noted that in step S1 of the present invention, the acquisition and fusion mechanism of multimodal sensing data constitutes the sensing front end of the entire diaphragm intelligent positioning system. The diaphragm is often an extremely thin and semi-transparent film, and its surface is coated or matte-treated, and it is in a high-speed shaking state at the moment of cutting.

[0065] If traditional high-precision laser triangulation is used alone, its precise optical triangulation principle relies on the formation of a stable, clear, high-contrast laser spot on the measured surface. However, when facing a semi-transparent membrane, some of the laser energy will penetrate the membrane material, causing the sensor to actually measure the distance to the rear guide roller, resulting in a positive deviation. If there is a specular reflection area on the membrane surface, or if the reflected spot deviates from the field of view of the receiving lens due to severe shaking, wrinkles, or inclined surfaces, the laser signal will attenuate sharply or even be lost.

[0066] Conversely, if only ultrasonic ranging is relied upon, its measurement principle based on the time of flight of sound waves is not sensitive to the optical properties of the measured object. It can stably obtain a ranging value, but its physical beam is wide and its spatial resolution is low. It returns the average distance of reflective objects in the detection area and cannot accurately capture the sharp contours of wrinkles. Moreover, its sound wave propagation speed is affected by the ambient temperature and humidity, and there is a slow drift.

[0067] Therefore, this step constructs a heterogeneous data fusion model based on complementary physical mechanisms and driven by intelligent algorithms. The core algorithm of this model first performs data validity gating. For laser data, the algorithm monitors the intensity amplitude of the signal at the receiving end in real time. This amplitude is generated by the conversion of light spot energy by photoelectric components. When it falls below a threshold set based on historical noise statistics, the algorithm determines that the measurement is unreliable due to the aforementioned optical interference and marks it accordingly. For ultrasonic data, the algorithm calculates the signal-to-noise ratio of its echo signal to distinguish between valid distance reflection peaks and environmental mechanical vibrations or airflow noise. This step assigns a health status label to each frame of raw data, isolating outliers caused by their respective physical limitations at the source and ensuring the fundamental quality of the data for subsequent fusion.

[0068] For valid data that passes the gate, the algorithm enters the core confidence-weighted fusion. The preset confidence weights here are not fixed constants, but an embedded parameter model. During the initial calibration of the system, the sensor array measures a series of calibration diaphragms with known heights, different transmittances and surface conditions, and statistically analyzes the systematic deviations and random error distributions of laser and ultrasound in this specific production line environment.

[0069] Based on this, the algorithm pre-sets a set of basic weights (w) for each typical material mode. L , w U ), where w L + w U =1. More importantly, this weighting model has dynamic contextualization capabilities. When online real-time monitoring detects that the overall distribution of laser signal intensity of a batch of diaphragms is lower than the historical average, while the ultrasonic data remains stable, the algorithm can infer that the current material has high light transmittance and automatically adjust w upwards based on the built-in mapping function. U , lower w L This dynamic adjustment mechanism enables the fusion model to adapt to fluctuations in material properties, ensuring that under any operating condition, the system tends to trust the sensor that is more reliable under the current conditions.

[0070] Finally, the fused height data h of each measurement point i at time t is... i (t) is generated by the algorithm according to the model, h i (t) = w L (t)×L i (t)+w U (t)×U i (t), where L and U represent the effective laser and ultrasonic data, respectively. In the event of both failures, the algorithm activates spatial interpolation protection, using the heights of adjacent effective measurement points for estimation to ensure data continuity. This fusion process essentially combines the high spatial resolution of laser with the environmental robustness of ultrasound at the algorithmic level, resulting in fused height data that removes systematic errors from single sensors and provides a more robust estimate of the diaphragm's true pose.

[0071] This step provides a clean and reliable spatial point cloud for the subsequent S2 3D morphology reconstruction, avoiding the presence of a large amount of jump noise and bias in the reconstructed 3D surface, which would cause the calculated basic length compensation to be completely inaccurate. At the same time, the S2 step also provides a high-quality feature extraction source for S3 wrinkle recognition. The subtle wavelengths of lateral wrinkles and the specific frequencies of longitudinal wrinkles must be accurately captured by spectral analysis from precise and smooth height change curves.

[0072] S2: Based on multimodal sensing data, the three-dimensional morphology of the diaphragm relaxation region is constructed through data fusion, and the basic length compensation of the diaphragm is calculated accordingly.

[0073] In this invention, step S2 further includes: constructing a three-dimensional surface of the diaphragm relaxation region based on the fused height data of each measurement point at different times; calculating the actual surface area of ​​the three-dimensional surface and its projected area under an ideal flat state; calculating the length expansion coefficient based on the actual surface area and the projected area; and calculating the basic length compensation amount based on the length expansion coefficient, the theoretical length of the diaphragm, and the material tensile coefficient.

[0074] It should be noted that in S2 of this invention, the process of constructing a three-dimensional shape and calculating the basic length compensation based on multimodal sensing data transforms the discrete data sensed by the front end into a precise quantification of the macroscopic physical state of the diaphragm. By utilizing the fused height data point array output from step S1, which is discretely distributed in space (width direction X-axis) and time (conveying direction virtual Y-axis), the algorithm reconstructs the continuous and complete three-dimensional curved surface shape of the diaphragm in the free relaxation state after cutting. Based on the principle of curved surface geometry, the basic length of the diaphragm required to compensate for eliminating this relaxation state is calculated.

[0075] The underlying mechanism lies in the fact that the relaxation after the diaphragm is cut essentially manifests as the deformation of its actual surface from an ideally flat plane into a three-dimensional curved surface with complex undulations. This results in the actual path length of the diaphragm material along its surface contour—its true length—being much greater than its projected straight-line length within a fixed projection distance. Traditional methods estimate the compensation amount by measuring the sag height of a single cross-section or point, which cannot accurately reflect the true length increment caused by complex two-dimensional wrinkles and deformation. This is the fundamental reason for inaccurate compensation calculations, leading to dimensional fluctuations or diaphragm damage.

[0076] The algorithm model used in this step accurately solves for this true length increment. The algorithm first constructs a three-dimensional surface, with the input being a discrete set of spatiotemporal points (x). i ,y k ,z i,k ), where z i,k =h i (t k ) is the fusion height obtained in step S1, y k =v·t k It is a spatial axis mapped from the known transport speed v to the time axis. The algorithm uses bilinear interpolation or polynomial surface fitting methods to fit these discrete points into a continuous surface function z=f(x,y).

[0077] In this construction process, the algorithm parameters, namely the fitting order and smoothing factor, must fully consider the physical properties of the diaphragm material. Since the diaphragm has a certain degree of elasticity and does not undergo severe bending, the surface should be continuous and first-order differentiable. By introducing smoothing constraints, the algorithm can effectively suppress sensor noise from being amplified into false sharp ripples during surface reconstruction, thereby obtaining a smoother surface that better reflects physical reality.

[0078] After obtaining the continuous surface f(x,y), the core of the algorithm is to calculate its actual surface area S. a This is the most crucial application of the mathematical model in the entire process, and its calculation formula is as follows:

[0079] Where Ω is the projection of the diaphragm relaxation region onto the XY plane. The integrand is... The geometric meaning of is the ratio of the area of ​​the infinitesimal element on the surface to the area of ​​the projected infinitesimal element. The algorithm numerically approximates this double integral by discrete summation:

[0080]

[0081] This calculation process directly affects the 3D point cloud data z. i,k It cleverly utilizes the height difference (z) between adjacent points. i+1,k -z i,k ) and (z i,k+1 -z i,k Numerical approximation of the surface partial derivative f x and f y This ensures that regardless of whether the diaphragm produces transverse wrinkles, longitudinal undulations, or mixed wrinkles, as long as it causes surface undulations (f... x ,f y If ≠ 0, then the contribution of this fluctuation to the length must be... Items are captured and accumulated into the total area S. a This makes the algorithm global, automatically and completely calculating the length increase caused by folds and sags in all directions, thus completely overcoming the shortcomings of single-point or single-line measurements.

[0082] Subsequently, the algorithm calculates the projected area S0=(x) of the region under ideal flat conditions. N -x1)×(vT). The length expansion coefficient η is defined as the ratio of the two:

[0083] η is a dimensionless number whose value is always greater than 1, intuitively reflecting the length magnification factor caused by three-dimensional relaxation deformation. Finally, the basic length compensation amount ΔL basic Calculated by the following formula:

[0084] Here, L0 is the theoretical geometric length of the diaphragm between two fixed points, and the material tensile coefficient α is a parameter of the physical properties of the connection in this algorithm model. It is not a universal constant, but is obtained through calibration experiments for diaphragms of different materials, thicknesses, and batches. During the process of the diaphragm being stretched, the material itself will undergo elastic elongation. This coefficient α transforms the geometric length expansion (η−1)L0 into an effective compensation length that requires additional winding by the servo motor. This compensates for the elastic effect of the material at the algorithm level, allowing the calculation to transition from the pure geometric domain to the physical mechanics domain, ensuring the physical accuracy of the compensation amount command.

[0085] S3: Analyze the three-dimensional morphology and identify the fold type and characteristic parameters of the diaphragm;

[0086] In a preferred embodiment of the present invention, S3 further includes: extracting a height profile sequence of the three-dimensional morphology in the width direction of the diaphragm, performing spectral analysis on each profile, determining whether there are transverse wrinkles based on the energy of the high-frequency components, and extracting their wavelength characteristics.

[0087] The height time series of each measurement point along the diaphragm conveying direction is extracted, and spectrum analysis is performed. Based on the energy in a specific frequency band, it is determined whether there are longitudinal wrinkles and their main frequency characteristics are extracted.

[0088] The type of fold is determined by comprehensively considering the presence of both horizontal and vertical folds.

[0089] It should be noted that in S3 of the present invention, by performing depth signal analysis on the three-dimensional curved surface data reconstructed in step S2 that characterizes the true physical morphology of the diaphragm, it is automatically determined whether the diaphragm has transverse wrinkles that are periodically distributed along the width direction, longitudinal wrinkles that fluctuate and propagate along the conveying direction, or a mixture of the two, and its key physical characteristic parameters are precisely quantified.

[0090] Different types of wrinkles are essentially modal manifestations of diaphragmatic instability in different dimensions, leaving unique features in three-dimensional height data that can be captured by spectral analysis. Lateral wrinkles manifest as periodic fluctuations in height values ​​along the spatial width (X-axis), with frequencies corresponding to spatial wavelengths; longitudinal wrinkles, on the other hand, manifest as oscillating changes in height values ​​at specific points over time along the Y-axis, with frequencies corresponding to temporal frequencies. By separating and analyzing the spectral features in these two orthogonal dimensions, the type and severity of wrinkles can be clearly decoupled and quantified.

[0091] For lateral wrinkle recognition, the algorithm first extracts height profiles from the 3D data at equal time intervals (Y coordinate positions) along the width direction (X-axis), forming a spatial sequence set. For each spatial profile vector h... k =[h1(t k ),h2(tk ),...,h N (t k )] T The algorithm performs a Discrete Fourier Transform (DFT) to obtain its spatial spectrum. The algorithm focuses on features through weighted analysis, rather than treating all spectral components equally. Instead, it uses an exponential weighting function m. β Where β>0, it is usually set to 1 to enhance the weight of high-frequency components. This is because small, dense wrinkles, corresponding to high spatial frequencies, pose a greater threat to diaphragm flatness, but have lower energy and are easily drowned out by noise. After enhancing their weight, the algorithm defines the transverse wrinkle energy index. This formula directly represents the spatial height distribution h measured by the sensor. i This is transformed into a scalar value E that characterizes the severity of the profile undulations. T (k).

[0092] E is obtained by averaging the exponents of all profiles (k=1 to M). T And compared with the empirical threshold E based on historical data statistics T ,t h By comparison, it can be objectively determined whether significant lateral folds exist. If they do, the algorithm further calculates the average spatial spectrum. Its peak position m peak The characteristic wavelength is calculated by directly corresponding to the dominant spatial frequency. This wavelength parameter quantifies the abstract concept of wrinkles into a concrete physical scale, providing a quantitative basis for subsequent diagnosis of equipment faults. For example, a wavelength that is too small may indicate a problem with the cutter, while a wavelength that is related to the roll width may indicate a problem with the guide roll.

[0093] For vertical wrinkle recognition, the algorithm changes the analysis dimension, extracting the sequence g of height change over time for each fixed width measurement point i. i =[h i (t1),h i (t2),...,h i (t M )] T The time spectrum is obtained by performing a DFT. The algorithm model exhibits bandpass filtering energy integration characteristics. The free vibration frequency of the diaphragm after cutting and the fluctuation frequency caused by tension control instability fall within a specific physical range of 5-20Hz. Therefore, the algorithm does not calculate the full-band energy, but instead defines the longitudinal fold energy. , where P i (f) is the power spectral density. This design eliminates slow droop at extremely low frequencies and sensor noise interference at extremely high frequencies, focusing on identifying harmful periodic longitudinal fluctuations. Similarly, the average energy across all measurement points is... and with threshold E L,th The algorithm compares and determines the existence of longitudinal wrinkles. If they exist, the algorithm searches for the comprehensive power spectrum. The peak position is used to extract the main frequency feature f. peak This frequency directly reflects the dynamic characteristics of the system; for example, a higher f... peak This could point to mechanical resonance or high-frequency oscillation in the servo system, while specific low frequencies might be related to the response speed of the tension loop.

[0094] Finally, the algorithm performs a rule-based comprehensive determination of the wrinkle type. This is a simple logical decision model whose input is the binary decision result I. T and I L The output is a discrete wrinkle type identifier: 1 if present, 0 otherwise. While the logic is simple, it builds upon the complex spectral feature extraction described earlier, achieving a mapping from continuous signals to discrete classifications. More importantly, the algorithm outputs not only the type but also a quantized severity indicator, such as... and These two ratios, both greater than 1, intuitively represent the multiple by which the wrinkles exceed the normal threshold.

[0095] This S3 step, by identifying the problem location and severity on the highway using the aforementioned algorithm, ensures accurate identification of wrinkle types and enables the calculation of targeted wrinkle corrections and adaptive control parameter adjustments. For example, the compensation strategy and PID parameter tuning for lateral wrinkles are significantly different from those for longitudinal wrinkles. Secondly, the extracted feature parameter wavelength λ... x and main frequency f peak This is the core input of the fault diagnosis module in S4, enabling the system to go beyond simple alarms about wrinkles and achieve intelligent analysis of why wrinkles occur, such as associating specific wavelengths with faults on specific rollers. Therefore, step S3, through its two-dimensional spectrum analysis algorithm, translates the three-dimensional morphological data into a fault feature language with clear engineering significance.

[0096] S4: Calculate the wrinkle correction amount based on the identified wrinkle type and feature parameters, and combine it with the basic length compensation amount to obtain the total length compensation amount;

[0097] In a preferred embodiment of the present invention, step S4 further includes step S41: performing fault diagnosis and cause analysis based on the wrinkle features identified in step S3;

[0098] Fault diagnosis and cause analysis include: determining the potential equipment fault type based on the characteristic pattern of the wrinkles and calculating the equipment health index; if the equipment health index is normal, assessing the matching degree between the current diaphragm material characteristics and the system control parameters, and adaptively adjusting the control parameters when the matching degree is lower than the threshold.

[0099] In S4, the calculation of the wrinkle correction amount includes: if there are transverse wrinkles, the first correction component is calculated based on their wavelength characteristics and severity; if there are longitudinal wrinkles, the second correction component is calculated based on their main frequency characteristics, diaphragm conveying speed and severity; the total length compensation amount is the sum of the basic length compensation amount and the wrinkle correction amount, and the wrinkle correction amount is the weighted sum of the first correction component and the second correction component.

[0100] It should be noted that in S4 of this invention, by calculating the correction amount based on the wrinkle characteristics and performing intelligent diagnosis, the additional length compensation requirement caused by a specific wrinkle pattern beyond the basic relaxation, i.e., the wrinkle correction amount, is quantitatively calculated by using the precisely quantified wrinkle type and characteristic parameters output from step S3. This is then combined with the basic length compensation amount obtained from step S2 to form the final precise command used to drive the servo system—the total length compensation amount. Furthermore, the observed wrinkle characteristics are used as diagnostic signals. Through the built-in expert knowledge model, it is determined whether the root cause is a mechanical failure of the equipment or a mismatch between process parameters and materials, thereby triggering a maintenance alarm or initiating parameter self-adjustment, thus realizing the self-sensing and self-optimization of the production system.

[0101] Based on the above, a dual mapping relationship is established from the wrinkle feature space to the compensation amount space and the fault cause space. Different types of wrinkles, whether transverse or longitudinal, and their characteristic parameters wavelength / frequency and severity, affect the effective length and tension response characteristics of the diaphragm in different physical ways, and are compensated for through different mathematical models. At the same time, these characteristic parameters are also sensitive indicators of the equipment operating status and the matching degree of material processes, containing rich system status information.

[0102] Calculating wrinkle correction amount Instead of using traditional fixed empirical offsets, the algorithm models the wrinkles based on their physical nature.

[0103] For transverse folds, the characteristic wavelength Directly reflects the spatial period of the folds, and the severity S T This reflects the amplitude of the fluctuation. The algorithm calculates its correction component: .

[0104] In the aforementioned transverse correction component algorithm, the shorter wavelength... Small wavelengths indicate dense folds, with a high number of folds per unit length. While the excess length contributed by each fold is small, the sheer number of folds is significant. Longer wavelengths, on the other hand, indicate gentle, large waves, with each fold contributing a large amount of excess length. (Coefficient) Through extensive experimental data calibration, it was found that the geometric feature of wavelength and severity is linearly mapped to an additional length compensation.

[0105] For longitudinal folds, their characteristic dominant frequency This reflects the speed of the fluctuation. Combined with the diaphragm transport velocity v, a characteristic fluctuation spatial scale v / can be obtained. The algorithm calculates its correction component: This converts time frequency into spatial wavelength, and then determines its severity S. L Converted to length compensation. Coefficient. It is also obtained through calibration.

[0106] Finally, the total length compensation is calculated using the formula. Given, among which, Weight , The value is 0 or 1, or somewhere in between, depending on whether the wrinkle type is purely horizontal, purely vertical, or a mixture. This correction calculation model directly and linearly outputs the abstract spectral features (wavelength, frequency) extracted in step S3 into the compensation length required for the servo motor to rotate additionally, through physically meaningful conversion coefficients. This allows the control command to accurately offset the additional length error caused by a specific wrinkle pattern.

[0107] In the fault diagnosis and cause analysis section S41, the algorithm model combines the aforementioned knowledge model with the health measurement model. Its input is the complete feature set obtained in S3: The algorithm first calculates the device health index. Its formula is:

[0108] in, It is the lateral wrinkle asymmetry index, used to detect uneven tension on one side. This is the severity threshold. This formula normalizes the severity and asymmetry of multi-dimensional, different-dimension wrinkle features and merges them into an overall health score between 0 and 1. A score below the threshold (0.8) is considered a suspected equipment malfunction.

[0109] More importantly, the algorithm has an embedded diagnostic rule base, that is, if If the anomaly is small and stable, it indicates unevenness or wear of the cutting blade; if A T A high value indicates a deviation in the parallelism of the guide rollers; if... If the frequency is close to the mechanical resonant frequency, it indicates that the servo drive gain is too high or the mechanical parts are loose.

[0110] If the equipment health index is normal, the root cause of the problem points to material properties or process parameters. The algorithm initiates a material property matching assessment. The system maintains a material parameter database. The formula for calculating the matching degree is:

[0111] This model compares current running parameters. Reference parameters corresponding to the current material ID in the database Calculate the Gaussian weighted distance to obtain the matching degree M between 0 and 1. mat If M mat <M th This indicates that the current parameter set is not suitable for the current material. The algorithm will automatically load the optimal parameters from the database or initiate an online fine-tuning based on gradient descent. The cost function By minimizing the severity of wrinkles and tension error, the control parameter p is optimized online.

[0112] Through the S4 step, the wrinkle correction amount is precisely calculated using a physical model, ensuring that even under complex wrinkle conditions, the total length compensation accurately matches the actual excess length of the diaphragm, solving the problems of dimensional instability and error accumulation. Secondly, the intelligent diagnostic function transforms traditional reactive maintenance and experience-based parameter tuning into predictive maintenance and adaptive processes. By quantifying health indices and matching degrees, it proactively exposes hidden equipment faults and process drift, reducing wrinkle generation at its source, thereby significantly improving yield and reducing overall operation and maintenance costs. Therefore, the S4 step, through its hybrid intelligent decision-making model, not only optimizes the accuracy of individual control actions but also endows the entire production system with the long-term evolutionary capability of self-awareness, diagnosis, and optimization.

[0113] S5: Based on the total length compensation, perform two-stage compensation control on the diaphragm to complete the diaphragm positioning and proceed with the next cycle of diaphragm cutting.

[0114] The two-stage compensation control includes: in the first stage, controlling the drive mechanism to move rapidly to eliminate the main slack; in the second stage, controlling the drive mechanism to move slowly to smoothly increase the diaphragm tension to the target value.

[0115] In this invention, in S5, the first stage adopts a position control mode, and the motion target of the drive mechanism is a part of the total length compensation amount; the second stage adopts a torque control mode based on tension feedback, and the parameters of the controller are adaptively adjusted according to the wrinkle type and severity identified in S3.

[0116] In this invention, the adaptive parameter adjustment in the second stage includes: when the fold type is mainly horizontal folds, adjusting the proportional, integral, and derivative parameters of the controller according to the first rule; when the fold type is mainly vertical folds, adjusting the proportional, integral, and derivative parameters of the controller according to the second rule; and when there are no folds, using the nominal control parameters.

[0117] It should be noted that in S5 of this invention, the final instruction calculated in step S4, which integrates the basic relaxation amount and the wrinkle correction amount—the total length compensation amount—is used. This is transformed into a motion sequence for the servo motor. This sequence consists of two phases, separated in time and with distinct control objectives. In the first phase, the drive mechanism rotates rapidly in position control mode, aiming to efficiently and extensively wind up the diaphragm, eliminating most of the geometric slack and bringing it from a fully relaxed state to a near-tension critical point. In the second phase, the drive mechanism immediately switches to torque control mode based on real-time tension feedback, continuing to wind with precise and gentle movements, aiming to smoothly and without overshoot increase the diaphragm tension from near zero to the process-set target value T. set This allows for precise positioning, preparing the diaphragm for cutting in the next work cycle. This avoids the problem of the diaphragm being subjected to extremely high instantaneous acceleration and impact stress due to a single, high-speed pulling motion during the relaxation to tightening process, which could damage the internal microstructure. It also avoids the problem of oscillation or overstretching when approaching the target tension, which could lead to dimensional instability.

[0118] The algorithm model implemented in this step is first based on the total compensation amount. The total angle that the servo motor needs to compensate for is calculated using the drive roller radius R and the reduction ratio i. The control objective of the first phase is to complete a large part of it quickly, with the target angle set as follows: Here, k1 is a coefficient between 0.7 and 0.9, ensuring a margin for tension adjustment in the second stage. This stage's algorithm employs S-shaped velocity planning for position control, with the planned maximum velocity v... max1 and acceleration a max1 It is not a fixed value, but rather an adaptive derating based on the wrinkle type identified in step S3. Its rule model can be expressed as: if the wrinkle type is no wrinkles, then the nominal maximum speed v is used. max,o If it is a horizontal fold, then v max1 =0.7v max,0 If it is a vertical fold, then v max1 =0.8V max,0 If it is a mixed fold, then v max1 =0.6v max,0 .

[0119] Lateral wrinkles indicate uneven stress along the width of the diaphragm. Excessive pulling can exacerbate sliding and friction, potentially causing the wrinkles to jam or tear laterally. Therefore, the speed must be reduced to the maximum extent possible. Longitudinal wrinkles indicate ripples in the diaphragm along the conveying direction. Excessive pulling can induce resonance, so the speed also needs to be reduced appropriately. By directly mapping the perceived wrinkle type characteristics to the kinematic constraints of the first stage, the algorithm fundamentally prevents physical damage that may be caused by excessive tension.

[0120] When the servo motor reaches the first-stage target position Upon that, the system immediately and seamlessly switches to the second stage, where the input is the tension setpoint. Real-time feedback value from tension sensor error The output is a command used for motor torque control. The PID parameters (K) of the algorithm p ,K i ,K d It is no longer fixed, but rather based on the wrinkle type and severity output by step S3 (S T ,S L This is dynamically adjusted. This constitutes a feature-parameter mapping model oriented towards control performance optimization.

[0121] Specifically, when the system determines that the wrinkles are predominantly horizontal, the adjustment rule is as follows: , , The root cause of transverse wrinkles is uneven tension in the width direction, which can be addressed by increasing the integral coefficient K. i It can enhance the system's ability to eliminate steady-state errors, promoting a more uniform establishment of tension in the width direction; at the same time, it slightly reduces the proportional coefficient K. p and differential coefficient K d This is to avoid overreacting to instantaneous changes in tension error, thereby smoothly ironing out transverse wrinkles.

[0122] When the condition is determined to be predominantly vertical folds, the adjustment rule is as follows: , , Longitudinal wrinkles exhibit periodic fluctuations in tension, essentially constituting a dynamic stability problem, which significantly enhances the differential coefficient K. d This is equivalent to adding electrical damping to the system, which can effectively predict and suppress the trend of tension changes, thereby rapidly attenuating fluctuations; at the same time, it appropriately reduces the proportional and integral action to prevent the introduction of new overshoot or steady-state offset when suppressing fluctuations.

[0123] If there are no wrinkles, then use a set of optimized nominal parameters (K). p0 ,K i0 ,K d0 To achieve the fastest response, the PID controller is adjusted according to these parameters, transforming it from a general-purpose regulator into one that can control specific physical defects.

[0124] This S5 step safely and accurately applies the analysis and calculation results from all previous steps to the diaphragm entity. Through a two-stage division, it fundamentally avoids the transient impact from relaxation to tension, thus solving the diaphragm damage problem. The second stage, based on tight-loop tension control and parameter adaptation, ensures the accuracy and stability of tension establishment, thereby guaranteeing the consistency of the absolute length of the diaphragm after tensioning in each cycle and resolving the dimensional fluctuation problem.

[0125] More importantly, this adaptive control based on real-time state folding enables the system to actively suppress the impact of various disturbances on tension stability, avoiding the propagation and accumulation of errors in each cycle.

[0126] A lithium-ion battery separator cutting and positioning device, comprising:

[0127] The sensor fusion and detection module is used to acquire multimodal sensor data of laser and ultrasonic waves after the diaphragm is cut, and to perform data fusion processing; the 3D reconstruction and compensation calculation module is used to reconstruct the 3D relaxation morphology of the diaphragm based on the fused data and calculate the basic length compensation amount; the wrinkle feature extraction and recognition module is used to analyze the 3D morphology and identify the type, severity and characteristic parameters of the wrinkles; the intelligent diagnosis and parameter adjustment module is used to perform equipment fault diagnosis and material property evaluation based on the wrinkle features, and to adaptively adjust the system control parameters; the two-stage compensation control module is used to generate two-stage control commands based on the basic length compensation amount, the wrinkle correction amount and the adjusted control parameters, to drive the actuator to complete the precise compensation and positioning of the diaphragm.

[0128] In this invention, the sensor fusion and detection module includes: multiple measurement units arranged at equal intervals along the width direction of the diaphragm, each measurement unit including a laser displacement sensor and an ultrasonic ranging sensor; and a synchronous acquisition unit, used to synchronously trigger and acquire the measurement data of all sensors within a set time period after receiving a cut-off signal.

[0129] It should be noted that the lithium-ion battery separator cutting and positioning device of the present invention comprises five functional modules connected in series, forming a complete closed-loop system of perception-cognition-decision-execution from the acquisition of original physical signals to the final execution of mechanical actions.

[0130] Specifically, the sensor fusion and detection module is characterized by multiple laser-ultrasonic heterogeneous measurement units arranged at equal intervals along the width of the diaphragm. This paired arrangement is based on the principle of complementary physical properties. The laser displacement sensor, based on triangulation, provides high-resolution, high-precision point distance measurement; however, its measurement relies on stable optical reflection from the target surface, making it prone to failure when the laser partially penetrates the semi-transparent diaphragm or when the reflected light spot deviates from the complex folds and slopes. The ultrasonic ranging sensor, based on the time-of-flight principle, is insensitive to the optical properties of the material surface and can stably obtain ranging values; however, its wide acoustic beam results in low spatial resolution, making it difficult to capture fine contours.

[0131] In this module, the algorithm model ensures that all sensor data have a unified timestamp through a synchronous acquisition unit. For the acquired raw data, the algorithm is a confidence-based weighted fusion model. This model first verifies the signal strength validity of the laser data and the signal-to-noise ratio validity of the ultrasonic data, filtering out invalid values ​​caused by physical limitations. Subsequently, for valid data, the algorithm dynamically weights the data according to preset confidence weights that correlate with material properties (e.g., assigning higher weights to ultrasonic waves for highly transparent materials), generating reliable fused height data for each measurement point.

[0132] The 3D reconstruction and compensation calculation module receives the fused discrete point cloud data, reconstructs a continuous 3D surface, and calculates geometric compensation. The algorithm maps the data to the spatial domain by multiplying the time series by the diaphragm velocity, obtaining a discrete point set.

[0133] Subsequently, bilinear interpolation or polynomial surface fitting algorithms are used to fit these points into a continuous surface function z=f(x,y). The fitting order and smoothing coefficient of this algorithm are optimized for the physical characteristics of the diaphragm to ensure that the reconstructed surface both fits the data and suppresses noise. Based on this surface, the algorithm executes the core area integral calculation model: The actual surface area is solved by numerical integration.

[0134] This model integrates all excess length caused by lateral and longitudinal folds and sagging into the area S. a Then, the length expansion coefficient η is obtained by comparing it with the projected area S0. Finally, the basic length compensation amount is... The calculation correlates the geometric shape with the tensile coefficient of the material's physical properties through an algorithm model, achieving a precise and global conversion from three-dimensional morphology to one-dimensional compensation, overcoming the limitations of single-point measurement.

[0135] The wrinkle feature extraction and recognition module receives three-dimensional morphological data and incorporates a dual-channel parallel spectrum analysis algorithm. For transverse wrinkles, the algorithm extracts the cross-sectional sequence along the width direction for spatial spectrum analysis (DFT) and designs a weighting function to enhance the weight of high-frequency wrinkle components, calculating the transverse wrinkle energy. For longitudinal wrinkles, the algorithm extracts the time series for time spectrum analysis and integrates the energy E within the characteristic frequency band of 5-20Hz. L The algorithm automatically determines the existence of wrinkles by setting a preset threshold and extracts key feature parameters: lateral wavelength and longitudinal dominant frequency.

[0136] The intelligent diagnosis and parameter tuning module includes a rule-based diagnosis engine and a parameter matching evaluation model. The diagnosis engine bases its decisions on feature patterns, such as... The problem of the little finger pointing to the cutter, f peak Anomalies are identified as servo oscillations; the equipment health index is calculated to provide early warning of faults. The parameter matching model calculates the degree of matching between the current operating parameters and the calibration parameters in the material database. If the matching degree is low, the system will automatically retrieve the optimal parameters or initiate online fine-tuning. The algorithm of this module links feature parameters and control parameters through empirical rules and mathematical models, enabling the system to have the ability to self-diagnose and adaptively optimize.

[0137] The final two-stage compensation control module receives the total compensation command and the adjusted control parameters, and internally incorporates a hybrid control state machine algorithm. The first stage employs position control, and its S-shaped velocity planning parameters adaptively reduce speed based on the type of fold; for example, for transverse folds, v... max1 =0.7v max,0 To prevent further lateral slippage, the second stage switches to torque control with a tension closed loop, where the parameters of the variable-parameter PID controller are dynamically adjusted according to the type of wrinkle. For example, for lateral wrinkles, the algorithm increases the integral coefficient K. i To homogenize the tension in the width direction; for longitudinal wrinkles, the differential coefficient K is significantly increased. d This algorithm suppresses tension oscillations. It directly maps the wrinkle features sensed by the front end to the gain strategy of the controller, realizing online and targeted optimization of the control strategy.

[0138] Based on the preferred embodiments of the present invention described above, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for cutting and positioning a lithium-ion battery separator, characterized in that, Includes the following steps: S1: After the diaphragm is cut, multimodal sensing data of the diaphragm relaxation area are acquired in real time. Among them, the multimodal sensing data includes laser ranging data and ultrasonic ranging data from multiple measurement points distributed along the width of the diaphragm; S2: Based on multimodal sensing data, the three-dimensional morphology of the diaphragm relaxation region is constructed through data fusion, and the basic length compensation of the diaphragm is calculated accordingly. S3: Analyze the three-dimensional morphology and identify the fold type and characteristic parameters of the diaphragm; S4: Calculate the wrinkle correction amount based on the identified wrinkle type and feature parameters, and combine it with the basic length compensation amount to obtain the total length compensation amount; S5: Based on the total length compensation, perform two-stage compensation control on the diaphragm to complete the diaphragm positioning and proceed with the next cycle of diaphragm cutting. The two-stage compensation control includes: in the first stage, controlling the drive mechanism to move rapidly to eliminate the main slack; in the second stage, controlling the drive mechanism to move slowly to smoothly increase the diaphragm tension to the target value.

2. The lithium-ion battery separator cutting and positioning device and method according to claim 1, characterized in that: In S1, acquiring multimodal sensing data includes: acquiring laser ranging data and ultrasonic ranging data for each measurement point; The signal strength validity of laser ranging data is verified, and the signal-to-noise ratio validity of ultrasonic ranging data is verified. For data that passes the validity verification, weighted fusion is performed according to the preset confidence level weights to obtain the fused height data of the measurement point.

3. The lithium-ion battery separator cutting and positioning device and method according to claim 1, characterized in that: S2 also includes: constructing a three-dimensional surface of the diaphragm relaxation region based on the fused height data of each measurement point at different times; Calculate the actual surface area of ​​a three-dimensional curved surface and its projected area under ideal flat conditions; Calculate the length expansion coefficient based on the actual surface area and the projected area; The basic length compensation amount is calculated based on the length expansion coefficient, the theoretical length of the diaphragm, and the material tensile coefficient.

4. The lithium-ion battery separator cutting and positioning device and method according to claim 1, characterized in that: S3 further includes: extracting the height profile sequence of the three-dimensional morphology in the width direction of the diaphragm, performing spectral analysis on each profile, determining whether there are transverse wrinkles based on the energy of the high-frequency components, and extracting their wavelength characteristics. The height time series of each measurement point along the diaphragm conveying direction is extracted, and spectrum analysis is performed. Based on the energy in a specific frequency band, it is determined whether there are longitudinal wrinkles and their main frequency characteristics are extracted. The type of fold is determined by comprehensively considering the presence of both horizontal and vertical folds.

5. The lithium-ion battery separator cutting and positioning device and method according to claim 1, characterized in that: The S4 also includes step S41: based on the wrinkle features identified in step S3, perform fault diagnosis and cause analysis; Fault diagnosis and cause analysis include: determining potential equipment fault types based on the characteristic patterns of wrinkles, and calculating the equipment health index; If the equipment health index is normal, assess the matching degree between the current membrane material characteristics and the system control parameters, and adaptively adjust the control parameters when the matching degree is lower than the threshold.

6. The lithium-ion battery separator cutting and positioning device and method according to claim 1, characterized in that: In S5, the first stage adopts a position control mode, and the motion target of the drive mechanism is a part of the total length compensation amount; The second stage employs a torque control mode based on tension feedback, where the controller parameters are adaptively adjusted according to the type and severity of wrinkles identified by S3.

7. The lithium-ion battery separator cutting and positioning device and method according to claim 1, characterized in that: In S4, calculating the wrinkle correction amount includes: if there are transverse wrinkles, calculating a first correction component based on their wavelength characteristics and severity; If longitudinal wrinkles exist, the second correction component is calculated based on their main frequency characteristics, diaphragm delivery speed, and severity. The total length compensation is the sum of the basic length compensation and the wrinkle correction, and the wrinkle correction is the weighted sum of the first correction component and the second correction component.

8. The lithium-ion battery separator cutting and positioning device and method according to claim 6, characterized in that: The second stage of adaptive parameter adjustment includes: when the wrinkle type is mainly horizontal wrinkles, adjusting the proportional, integral, and derivative parameters of the controller according to the first rule; When the wrinkle type is predominantly longitudinal, adjust the proportional, integral, and derivative parameters of the controller according to the second rule. When there are no wrinkles, use the nominal control parameters.

9. A lithium-ion battery separator cutting and positioning device, based on the method according to any one of claims 1-8, characterized in that, include: The sensor fusion and detection module is used to acquire laser and ultrasonic multimodal sensing data after the diaphragm is cut, and to perform data fusion processing. The 3D reconstruction and compensation calculation module is used to reconstruct the 3D relaxation morphology of the diaphragm based on the fused data and calculate the basic length compensation amount. The wrinkle feature extraction and recognition module is used to analyze three-dimensional morphology and identify the type, severity, and feature parameters of wrinkles. The intelligent diagnosis and parameter adjustment module is used to diagnose equipment faults and evaluate material properties based on wrinkle characteristics, and adaptively adjust system control parameters. The two-stage compensation control module is used to generate two-stage control commands based on the basic length compensation amount, the wrinkle correction amount, and the adjusted control parameters, driving the actuator to complete the precise compensation and positioning of the diaphragm.

10. A lithium-ion battery separator cutting and positioning device according to claim 9, characterized in that: The sensor fusion and detection module includes: multiple measurement units arranged at equal intervals along the width of the diaphragm, each measurement unit including a laser displacement sensor and an ultrasonic ranging sensor; and a synchronous acquisition unit, used to synchronously trigger and acquire measurement data of all sensors within a set time period after receiving a cutoff signal.

Images