Dry method pole piece forming control method based on AI algorithm and pole piece forming composite all-in-one machine
By combining neural network algorithms and PID control in the calender control system, a linear regression model was established to optimize roll gap adjustment, solving the problem of insufficient control of existing calenders in nonlinear and multivariable systems, and achieving more efficient electrode forming and quality control.
Patent Information
- Application Number
- CN202411510092.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-28
- Publication Date
- 2025-12-12
- Estimated Expiration
- 2044-10-28
AI Technical Summary
Existing calender control systems have poor control performance when dealing with nonlinear, time-varying, or multivariable systems, which affects the quality of the electrode sheets and the stability of the system.
By combining neural network algorithms with PID control, a linear regression model is established by integrating the DRNN neural network algorithm with the PID controller. The roll gap is adjusted according to the priority level, and the output is fitted by combining expert knowledge base and historical data to construct a lightweight output quantity, which is then input into the PID controller for real-time adjustment.
It improves the system's control performance and stability, enhances its ability to handle complex dynamic performance and multivariable coupling, and improves the production efficiency and product quality of the calender.
Smart Images

Figure CN119114645B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application relates to the technical field of calendering and compounding of pole pieces, and in particular to a dry-method pole piece forming control method based on an AI algorithm and a pole piece forming and compounding all-in-one machine. BACKGROUND
[0002] A calender is applied to the industries of calendering, film stretching and rolling, and in the control system of the calender, multiple structural units are related to feedback control, namely, a roll gap adjusting unit, a pole piece deviation correcting unit and a pole piece tension control unit. The roll gap adjusting structure is used for adjusting the roll gap between the main calendering rollers to complete the adjustment of the rolling force and the thickness of the pole piece; the pole piece deviation correcting structure is used for adjusting the deviation of the pole piece to avoid the phenomenon of tower-shaped winding caused by the deviation of the pole piece and affecting the quality of the pole piece; and the pole piece tension control system is used for maintaining the stability of the tension of the pole piece to ensure that the rolled pole piece is flat and avoid the phenomenon of waves caused by uneven extension of the pole piece.
[0003] The control system of the existing calender usually adopts PID control (proportional integral derivative control), and the traditional PID control directly inputs the collected parameters or signals into the PID controller, and outputs the control amount by linear combination according to the proportion, integral and derivative to control the controlled object. The output of the traditional PID control is composed of linear combination, and when facing a nonlinear, time-varying or multivariable system, the traditional PID control is insufficient in processing the complex dynamic performance and multivariable coupling, and lacks self-adaptability and robustness, resulting in poor control performance, affecting the control stability and accuracy of the system, and thus affecting the quality of the pole piece. SUMMARY
[0004] In view of the above-mentioned shortcomings of the existing calender control system, the present applicant provides a reasonable dry-method pole piece forming control method based on an AI algorithm and a pole piece forming and compounding all-in-one machine, and introduces a neural network algorithm combined with PID control to improve the stability and accuracy of the system.
[0005] The technical scheme adopted by the application is as follows:
[0006] A dry-method pole piece forming control method based on an AI algorithm, wherein the calendering control system adjusts each roll gap according to the thickness of the finished film, roll gap displacement information, position information of the double rods of the hydraulic cylinder and angle information of the driving motor of the screw wedge assembly.
[0007] The control center of the calendering control system adopts a DRNN neural network algorithm to calculate the signals to obtain a light-weight output, a linear regression model is established among the light-weight output, the output of the expert knowledge base and the fitting output of the historical data, and the linear regression formula of the three is: , k1, k2, k3 are weights of the three outputs of light weight output O, expert knowledge base P, and historical data fitting Q, b is a constant, the control amount calculated by the linear regression model is used as the input vector of the PID controller, the output of the PID controller is output to the corresponding actuator, and the corresponding part is adjusted.
[0008] As a further improvement of the above technical scheme:
[0009] The adjustment priority of each roll gap is: in the total direction of the material, the priority of the later roll gap is higher than that of the former roll gap.
[0010] The weight values of the linear regression formula are k1=0.4, k2=0.3, and k3=0.3.
[0011] After the adjustment of each roll gap, one of the remaining roll gaps is greater than or equal to the previous roll gap, except for the first and last roll gaps.
[0012] The control process of the roll gap includes: after the control center converts and calculates the offline detection data of the film produced by the calender unit, the parameter set {K P1 , T I1 , T D1 , e1(t)} is output, the expert knowledge base and the historical data fitting formula output the parameter set {K P2 , T I2 , T D2 , e2(t)} and {K P3 , T I3 , T D3 , e3(t)}, and the three parameter sets are calculated by the linear regression model to output the parameter set {K P , T I , T D , e(t)}; one or more parameter sets {K P , T I , T D , e(t)} that meet the roll gap size relationship are output to one or more PIDs that control different roll gaps, and the PID calculates through a discrete algorithm to issue a displacement vector instruction to the servo motor and servo hydraulic valve group of the screw wedge assembly, and the servo motor and hydraulic cylinder of the screw wedge assembly run synchronously to adjust the roll gap.
[0013] After the roll gap adjustment is completed, the offline detection method is used to detect the thickness and surface density of the film, if the offline detection thickness is within ±5%, the roll gaps are fixed, if it exceeds ±5%, the calculation is returned, and the adjustment steps are repeated until the thickness of the film is within ±5%.
[0014] The PID controller uses a discrete-time system; the transfer function of the PID is s is a Laplace variable (s = a + jw), K p is an amplification factor, T I is an integration time, T D is a derivation time; the adjustment function of the PID is The integral approximation of e(t) is a rectangular sum The approximation of e'(t) is a straight line The control expression of the PID is .
[0015] The calender control system adjusts the rotating speed of the servo motor according to the rotating speed information of the servo motor, adjusts the unwinding tension of the metal foil and the winding tension of the pole piece according to the tension information collected by the tension sensor, and corrects the position of the metal foil and the pole piece according to the displacement information detected by the ultrasonic sensor.
[0016] A pole piece forming and compounding all-in-one machine adopts the dry-method pole piece forming control method based on AI algorithm in claim 1 for control, comprising at least one calender set, one unwinding set and one winding set, the unwinding set and the winding set being arranged on the pole piece compounding and forming side of the calender set; the calender set comprises a plurality of calender rollers arranged in sequence along the main conveying direction X, and each two adjacent calender rollers have a roller gap therebetween.
[0017] As a further improvement of the above technical solution:
[0018] The roller gaps between each two adjacent calender rollers gradually decrease from front to back along the main conveying direction X, and one of the roller gaps is equal to the previous one except the first and the last roller gaps.
[0019] A bearing seat is correspondingly arranged on each calender roller, a screw and wedge assembly is arranged between the bearing seats corresponding to adjacent calender rollers, the screw and wedge assembly is matched with the bearing seat through a slope surface, and the screw and wedge assembly moves along the axial direction of the calender roller to adjust the roller gap width between the two calender rollers; a roller gap displacement sensor is further arranged between the bearing seats corresponding to adjacent calender rollers.
[0020] A back roller is arranged on the side opposite to the main conveying direction X of the first calender roller, the back roller is tightly attached to the first calender roller, a hydraulic cylinder is arranged on the side opposite to the main conveying direction X of the back roller and on the side same as the main conveying direction X of the last calender roller respectively, and the hydraulic cylinder controls the displacement of the back roller or the last calender roller by controlling the extension and retraction of the double rods thereof through a valve group.
[0021] The roller diameter of the back roller and the roller diameter of the last two calender rollers are both greater than the roller diameter of the other calender rollers; each calender roller is separately connected to an independent motor and driven by the independent motor, and the rotating speed of each calender roller gradually increases from front to back along the main conveying direction X.
[0022] The first and second calendering rollers near the back roller have a smaller diameter than the other calendering rollers and the back roller.
[0023] Each calendering roller can be independently adjusted in a direction perpendicular to the main conveying direction X; the axes of the calendering rollers are located in the same plane or staggered, and when staggered, the extension lines of the axes of each pair of adjacent calendering rollers intersect at a point with the extension line of the axes of the adjacent other pair of adjacent calendering rollers; at least one calendering roller is a heatable roller with a heating temperature of 150-200 DEG C.
[0024] The unwinding unit is provided with a metal foil unwinding shaft, and between the metal foil unwinding shaft and the roll gap of the last two calendering rollers of the calendering unit, an interface platform, an unwinding tension sensor, a tension swing roller, an unwinding deviation correction assembly and an unwinding ultrasonic sensor are arranged in sequence along the unwinding direction of the metal foil, and unwinding guide rollers are arranged between adjacent components.
[0025] The unwinding unit is provided with a film unwinding shaft, and a plurality of unwinding guide rollers are arranged between the film unwinding shaft and the roll gap of the last two calendering rollers of the calendering unit.
[0026] The winding unit is provided with a winding shaft, and between the winding shaft and the roll gap of the last two calendering rollers of the calendering unit, a first winding tension sensor, a first swing roller assembly, a tension blocking assembly, a second winding tension sensor, a second swing roller assembly, a winding deviation correction assembly and a winding ultrasonic sensor are arranged in sequence along the winding direction of the pole piece, and winding guide rollers are arranged between adjacent components.
[0027] The beneficial effects of the present application are as follows:
[0028] The calendering control system of the present application introduces a neural network algorithm combined with a PID controller, and the neural network algorithm also combines an expert knowledge base output and a historical data fitting output to construct a linear regression model, obtains a linear regression output, and then inputs the PID controller, and the PID controller automatically adjusts the control amount in real time according to the linear regression output, which can more accurately control the roll gap, the tape speed, the tension and the deviation correction, and when facing a nonlinear, time-varying or multivariable system, the complex dynamic performance and multivariable coupling can also be accurately processed, the adaptability and robustness are better, the control performance of the system is improved, the stability and accuracy of the system are improved, and the production efficiency and product quality of the calendering machine are also improved.
[0029] The PID controller of the present application uses a discrete-time system, which can suppress noise and interference through digital filtering and other technologies, has better robustness than a continuous-time system and is suitable for real-time control applications, and can realize complex control strategies through simple algorithms. BRIEF DESCRIPTION OF DRAWINGS
[0030] Figure 1 Figure 1 is a front view of the structure of the first embodiment of the pole piece forming and composite integrated machine of the present application.
[0031] Figure 2 Figure 2 is a top view of the structure of the calender unit of the present application.
[0032] Figure 3 Figure 3 is a schematic diagram of the roll gap adjustment of the calender unit.
[0033] Figure 4 Figure 4 is a front view of the structure of the second embodiment of the pole piece forming and composite integrated machine of the present application.
[0034] Figure 5 Figure 5 is a front view of the structure of the third embodiment of the pole piece forming and composite integrated machine of the present application.
[0035] Figure 6 Figure 6 is a front view of the structure of the fourth embodiment of the pole piece forming and composite integrated machine of the present application.
[0036] Figure 7 Figure 7 is a control flow chart of the calender control system.
[0037] Figure 8 Figure 8 is a control flow chart of the control center.
[0038] Figure 9 Figure 9 is a flow chart of the neural network to the PID controller to the actuator.
[0039] In the figure:
[0040] 100, calender unit; 11, calender roll; 12, back roll; 13, hydraulic cylinder; 14, motor; 15, bearing seat; 16, screw and wedge assembly; 17, roll gap displacement sensor; 10, roll gap; 101, roll gap; 102, roll gap; 103, roll gap; 104, roll gap; 105, roll gap; 106, roll gap; 107, roll gap;
[0041] 200, unwinding unit; 21, metal foil unwinding shaft; 22, film unwinding shaft; 23, unwinding tension sensor; 24, tension swing roller; 25, unwinding deviation correction assembly; 26, unwinding ultrasonic sensor; 27, belt receiving platform; 28, unwinding guide roller;
[0042] 300, winding unit; 31, winding shaft; 32, first winding tension sensor; 33, first swing roller assembly; 34, tension cutoff assembly; 35, second winding tension sensor; 36, second swing roller assembly; 37, winding deviation correction assembly; 38, winding ultrasonic sensor; 39, winding guide roller. DETAILED DESCRIPTION
[0043] The specific embodiments of the present invention will now be described with reference to the accompanying drawings.
[0044] Example 1 of an integrated electrode forming and composite machine:
[0045] like Figure 1 As shown, this embodiment is a single-sided double-sided electrode forming composite integrated machine, including a calendering unit 100, an unwinding unit 200 and a winding unit 300. The unwinding unit 200 and the winding unit 300 are arranged on the electrode composite forming side of the calendering unit 100.
[0046] like Figures 1 to 3As shown, the calender set 100 comprises a plurality of calender rollers 11 arranged in sequence along the main conveying direction X, in this embodiment, eight calender rollers 11 are provided, each calender roller 11 can be independently adjusted in the direction perpendicular to the main conveying direction X to improve its adaptability, the shaft center of each calender roller 11 can be located in the same plane, the shaft centers of each calender roller 11 can also be staggered, when staggered, the extension line of the shaft center connecting line between each pair of adjacent two calender rollers 11 intersects with the extension line of the shaft center connecting line between the adjacent other pair of adjacent two calender rollers 11 at a point, at least one calender roller 11 is a heated roller, the heating temperature is between 150°C and 200°C; there is a roller gap 10 between adjacent two calender rollers 11, along the total material advancing direction, in sequence, the roller gaps are 101-107, the roller gap 10 between each adjacent calender roller 11 gradually decreases from front to back along the main conveying direction X, to meet the thickness requirement of multiple calendering, the film thickness is gradually reduced step by step, at the same time, except the first and last roller gaps 10, one of the remaining roller gaps 10 is greater than or equal to the previous roller gap 10, such as: roller gap 101> roller gap 102≥ roller gap 104≥ roller gap 103> roller gap 105> roller gap 106> roller gap 107, in this way, the internal stress of the film generated by multiple calendering is released, so that the compactness of the film and the thickness uniformity thereof can be improved; the film material is input from the side of the roller gap 10 of the first two calender rollers 11, and is formed into a film through multiple rolling of the calender rollers 11. Along the main conveying direction X of the material, the roll diameter of the last two calender rollers 11 is greater than the roll diameter of the other calender rollers 11, the roll diameter of the last two calender rollers 11 is larger, which has greater rigidity and can withstand greater rolling force, which is beneficial to improve the uniformity of the formed electrode sheet. The first calender roller 11 is provided with a back roller 12 on the side opposite to the main conveying direction X, the back roller 12 is closely attached to the first calender roller 11, the roll diameter of the back roller 12 and the roll diameter of the last two calender rollers 11 are both greater than the roll diameter of the other calender rollers 11, the roll diameter of the back roller 12 is larger, so that the calender roller 11 can be provided with greater rigid support, so that the calender roller 11 can withstand greater rolling force, and the shear force of the material is improved to improve the uniformity of the electrode sheet. The back roller 12 is provided with a hydraulic cylinder 13 on the side opposite to the main conveying direction X, and the last calender roller 11 is provided with a hydraulic cylinder 13 on the side same as the main conveying direction X, the hydraulic cylinder 13 controls the extension and contraction of the double rods thereof through the valve group to control the displacement of the back roller 12 or the last calender roller 11, so as to adjust the roller gap 10 between each calender roller 11. Each calender roller 11 is respectively connected to an independent motor 14, driven by the independent motor 14, and independently controlled in rotation speed and torque. The rotation speed of each calender roller 11 gradually increases from front to back along the main conveying direction X, so as to improve the adhesion of the film on the back roller and prevent the film from piling up; the speed increment of adjacent calender rollers 11 can be the same or different, which is set according to the actual processing requirement.Corresponding bearing seat 15 is arranged on each calendering roller 11, and a screw wedge assembly 16 is arranged between the corresponding bearing seats 15 of adjacent calendering rollers 11. The screw wedge assembly 16 is matched with the bearing seat 15 through a slope, and the width of the roll gap 10 between the two calendering rollers 11 can be adjusted by moving the screw wedge assembly 16 in the axial direction of the calendering roller 11. A roll gap displacement sensor 17 is also arranged between the corresponding bearing seats 15 of adjacent calendering rollers 11, which is used to monitor the distance between the bearing seats 15 to indirectly reflect the width of the roll gap 10, so as to determine the pressure of the hydraulic cylinder 13 and the rotation angle of the motor on the screw wedge assembly 16. Compared with the laser and infrared roll gap sensors for direct measurement, the roll gap displacement sensor 17 has the advantages of low cost, simple structure, small space occupation ratio, etc. A bearing anti-clearance mechanism is arranged on each calendering roller 11 and the back roller 12.
[0047] As shown in Figure 1 The unwinding machine group 200 is provided with a metal foil unwinding shaft 21 and a diaphragm unwinding shaft 22. The metal foil unwinding shaft 21 is used to place the metal foil coiled material, and the diaphragm unwinding shaft 22 is used to place the pre-produced diaphragm coiled material. Between the metal foil unwinding shaft 21 and the roll gap 10 of the last two calendering rollers 11 of the calendering machine group 100, an interface platform 27, an unwinding tension sensor 23, a tension swing roller 24, an unwinding deviation correction assembly 25, and an unwinding ultrasonic sensor 26 are arranged in sequence along the unwinding direction of the metal foil. Unwinding guide rollers 28 are arranged between adjacent components. After the metal foil coiled material is pulled out from the metal foil unwinding shaft 21, it passes through the interface platform 27, the unwinding tension sensor 23, the tension swing roller 24, the unwinding deviation correction assembly 25, and the unwinding ultrasonic sensor 26 in sequence, and is then introduced into the roll gap 10 between the last two calendering rollers 11 of the calendering machine group 100. A plurality of unwinding guide rollers 28 are arranged between the diaphragm unwinding shaft 22 and the roll gap 10 of the last two calendering rollers 11 of the calendering machine group 100. After the pre-produced diaphragm coiled material is pulled out from the diaphragm unwinding shaft 22, it passes through the plurality of unwinding guide rollers 28 and is then introduced into the roll gap 10 between the last two calendering rollers 11 of the calendering machine group 100. The diaphragm formed by the calendering machine group 100, the metal foil diaphragm guided by the metal foil unwinding shaft 21, and the diaphragm guided by the diaphragm unwinding shaft 22 are rolled and compounded into a pole piece by the last two calendering rollers 11. The metal foil unwinding shaft 21, the diaphragm unwinding shaft 22, and the tension swing roller 24 are respectively connected to driving motors and driven by the driving motors. The unwinding tension sensor 23 is used to monitor the unwinding tension of the metal foil in real time, the tension swing roller 24 is used to adjust the tension of the metal foil to ensure the accuracy of the unwinding tension, the unwinding ultrasonic sensor 26 is used to monitor the edge position of the metal foil in real time, and the unwinding deviation correction assembly 25 is used to correct the position of the metal foil to ensure that the position of the metal foil is not deviated.
[0048] As shown in Figure 1As shown, the winding unit 300 is equipped with a winding shaft 31, which is used to wind the composite-formed electrode sheet into a roll. Between the winding shaft 31 and the roll gap 10 between the last two calendering rolls 11 of the calender unit 100, a first winding tension sensor 32, a first oscillating roll assembly 33, a tension isolation assembly 34, a second winding tension sensor 35, a second oscillating roll assembly 36, a winding correction assembly 37, and a winding ultrasonic sensor 38 are arranged sequentially along the winding direction of the electrode sheet. Winding guide rollers 39 are respectively arranged between adjacent components. The first winding tension sensor 32 and the second winding tension sensor 35 are used to monitor the winding tension of the electrode sheet in real time. The first oscillating roll assembly 33 and the second oscillating roll assembly 36 are used to adjust the tension of the electrode sheet to ensure the accuracy of the winding tension. The winding ultrasonic sensor 38 is used to monitor the edge position of the electrode sheet in real time, and the winding correction assembly 37 is used to correct the position of the electrode sheet to ensure that the position of the electrode sheet is without deviation. The winding unit 300 is equipped with two sets of tension control components (winding tension sensor and swing roller assembly) to monitor and adjust the winding tension of the electrode sheets, making the tension adjustment more accurate and the winding more stable.
[0049] Example 2 of the electrode forming composite integrated machine:
[0050] like Figure 4 As shown, this embodiment is a double-sided, double-sided electrode forming and composite integrated machine, including two sets of calendering units 100 arranged symmetrically from left to right, one set of unwinding unit 200, and one set of winding unit 300. The main conveying directions of the two sets of calendering units 100 are opposite, and they share the last two calendering rolls 11. One set of calendering units 100 calenders multiple times from left to right along the main conveying direction X1 and then introduces the calendering roll gap 10 of the last two calendering rolls 11. The other set of calendering units 100 calenders multiple times from right to left along the main conveying direction X2 and then introduces the calendering roll gap 10 of the last two calendering rolls 11. The films calendered by the two sets of calendering units 100 and the metal foil films guided by the unwinding unit 200 are rolled and composited into electrode sheets by the last two calendering rolls 11. The double-sided double-sided electrode forming and laminating machine directly laminates the films on both sides after continuous online forming, saving the film unwinding device of the unwinding unit by 200 and saving the film pre-production time, resulting in higher electrode production efficiency.
[0051] like Figure 4As shown, in this embodiment, the unwinding unit 200 and the winding unit 300 are respectively arranged on the upper and lower sides of the two calendering units 100. The unwinding unit 200 is provided with a metal foil unwinding shaft 21. Between the metal foil unwinding shaft 21 and the roll gap 10 of the last two calendering rolls 11 of the calendering unit 100, a strip receiving platform 27, an unwinding tension sensor 23, a tension swing roller 24, an unwinding correction assembly 25, and an unwinding ultrasonic sensor 26 are arranged sequentially along the unwinding direction of the metal foil. Unwinding guide rollers 28 are respectively arranged between adjacent components. Between the winding shaft 31 of the winding unit 300 and the roll gap 10 of the last two calendering rolls 11 of the calender unit 100, a first swing roller assembly 33, a first winding tension sensor 32, a tension isolation assembly 34, a second winding tension sensor 35, a second swing roller assembly 36, a winding correction assembly 37, and a winding ultrasonic sensor 38 are arranged sequentially along the winding direction of the electrode sheet. A winding guide roller 39 is provided between adjacent components.
[0052] Example 3 of the electrode forming composite integrated machine:
[0053] like Figure 5 As shown, this embodiment is a single-sided double-sided electrode forming composite integrated machine. The difference between this embodiment and Embodiment 1 is that the calendering unit 100 in this embodiment is equipped with six calendering rolls. The diameter of the first and second calendering rolls 11 closest to the back roll 12 is smaller than the diameters of the other calendering rolls 11 and the back roll 12. The size of each roll gap 10 also satisfies the following condition: except for the first and last roll gaps 10, one of the remaining roll gaps 10 is greater than or equal to the previous roll gap 10, such as: roll gap 101 > roll gap 102 ≥ roll gap 104 ≥ roll gap 103 > roll gap 105.
[0054] Example 4 of the electrode forming composite integrated machine:
[0055] like Figure 6 As shown, this embodiment is a double-sided, double-sided electrode forming composite integrated machine. The difference between this embodiment and Embodiment 2 is that each calendering mill 100 in this embodiment is equipped with six calendering rolls. The diameter of the first and second calendering rolls 11 closest to the back roll 12 is smaller than the diameters of the other calendering rolls 11 and the back roll 12. The size of each roll gap 10 also satisfies the following condition: except for the first and last roll gaps 10, one of the remaining roll gaps 10 is greater than or equal to the previous roll gap 10, such as: roll gap 101 > roll gap 102 ≥ roll gap 104 ≥ roll gap 103 > roll gap 105.
[0056] The electrode forming composite integrated machine of the present invention is monitored and its parameters are adjusted in real time by a calendering control system to ensure the stability and accuracy of calendering and composite forming.
[0057] like Figure 7 As shown, the control method of the calendering control system mainly includes the following steps:
[0058] 1) The calender unit 100, unwinding unit 200 and winding unit 300 transmit detection signals required for roll gap control, detection signals required for speed control, detection signals required for tension control and detection signals required for deviation correction control to the control center: the detection signals required for roll gap control include roll gap 10 displacement information collected by roll gap displacement sensor 17, double rod position information of hydraulic cylinder 13, angle information of the drive motor of screw block assembly 16 and finished product film thickness information; the detection signals required for speed control include the rotational speed information of each calender roll 11, metal foil unwinding shaft 21, film unwinding shaft 22 and winding shaft 31, which are based on the rotational speed of one of the calender rolls 11, in this embodiment, the rotational speed of the second last calender roll 11, and the rotational speed of the remaining rolls or shafts is fed back to the control center by the connecting motor; the detection signals required for tension control include tension information detected by the tension sensor, the tension information of the unwinding unit 200 is the tension information collected by unwinding tension sensor 23, and the tension information of the winding unit 300 is the tension information collected by first winding tension sensor 32 and / or second winding tension sensor 35; the detection signals required for deviation correction control include displacement information detected by the ultrasonic sensor, the displacement information of the unwinding unit 200 is the position difference information between the metal foil edge and its preset value detected by unwinding ultrasonic sensor 26, and the displacement information of the winding unit 300 is the position difference information between the pole piece edge and its preset value detected by winding ultrasonic sensor 38;
[0059] 2) The control center performs fitting calculation on the received signals based on the neural network algorithm: for roll gap control, fitting calculation is performed on the roll gap 10 displacement information collected by roll gap displacement sensor 17, the double rod position information of hydraulic cylinder 13 and the angle information of the drive motor of screw block assembly 16; for speed control, the rotational speed of each roll or shaft is calculated; for tension control, the tension information of the tension sensor is calculated; for deviation correction control, the displacement information of the ultrasonic sensor is calculated;
[0060] 3) The execution instructions calculated by the control center are sent to the corresponding execution components: the roll gap control execution instructions are sent to the drive motor of hydraulic cylinder 13 and / or screw block assembly 16 of the calender unit 100; the speed control execution instructions are sent to the motor drive of each roll or shaft of each unit; the tension control execution instructions of the unwinding unit 200 are sent to tension swing roll 24, and the tension control execution instructions of the winding unit 300 are sent to first swing roll assembly 33 and / or second swing roll assembly 36; the deviation correction control execution instructions of the unwinding unit 200 are sent to unwinding deviation correction assembly 25, and the deviation correction control execution instructions of the winding unit 300 are sent to winding deviation correction assembly 37;
[0061] 4) Each execution component adjusts according to the received execution instruction: the hydraulic cylinder 13 and / or the screw wedge assembly 16 of the calendering unit 100 adjust the roll gap 10 according to the execution instruction, the hydraulic cylinder 13 can reduce the roll gap 10, and the screw wedge assembly 16 can enlarge the roll gap 10 to ensure the stability of the roll gap 10, thereby ensuring the compaction density and thickness uniformity of the film; the motor driver of each roll or shaft of each unit adjusts the rotation speed of each motor in real time according to the execution instruction to ensure that the rotation speeds of each motor are matched; the tension swing roller 24 of the unwinding unit 200, the first swing roller assembly 33 and / or the second swing roller assembly 36 of the winding unit 300 adjust the torque of the swing roller motor according to the execution instruction, thereby adjusting the unwinding tension of the metal foil and the winding tension of the pole piece; the unwinding deviation correction assembly 25 of the unwinding unit 200 and the winding deviation correction assembly 37 of the winding unit 300 adjust the torsion angle of the roller according to the execution instruction, thereby correcting the position of the metal foil and the pole piece to ensure that the position of the metal foil and the pole piece is deviation-free.
[0062] The control center of the calendering control system uses a DRNN (lightweight deep recurrent neural network) neural network algorithm to perform fitting calculation or analysis calculation on the signals to obtain a lightweight output, and introduces an expert knowledge base and historical data fitting. A linear regression model is established among the lightweight output, the expert knowledge base output, and the historical data fitting output, and the linear regression formula of the three is: k1, k2, and k3 are the weights of the outputs of the lightweight output O(x), the expert knowledge base P(x), and the historical data fitting Q(x), and b is a constant. In this embodiment, k1 is 0.4, k2 is 0.3, and k3 is 0.3. The calendering control system mainly uses a deep recurrent neural network, while introducing an expert knowledge base and historical data fitting, and giving them a relatively large proportional weight (40%, 30%, and 30% respectively), to ensure stable operation of the equipment, improve the precision of control, reduce control deviation caused by insufficient total amount of database data, and obtain more accurate and reasonable control quantity. The neural network structure includes an input layer, a hidden layer, and an output layer, and the neural network has an activation function.
[0063] Figure 8 A control flowchart of the control center of the calendering control system is shown. The control center uses a neural network algorithm combined with a linear regression model and a PID controller. The control quantity obtained by the neural network algorithm is used as the input vector of the PID controller, and the output quantity calculated by the PID controller is output to the corresponding execution mechanism to adjust the corresponding components.
[0064] The neural network model needs to be trained before being embedded in the control system, which learns the mapping function X→Y, X being the input vector and Y being the output vector. The neural network model is trained by expert using historical data, which includes the roll gap information, motor speed, membrane and pole piece tension information, swing roller angle, position and other key parameter data required by the control system. The collected information is preprocessed before being input into the neural network model, which includes noise removal, data normalization and other operation processing, which is beneficial to improve the training effect of the neural network model. The neural network model can also be updated and adjusted regularly, and retrained when the working condition changes, to adapt to the changes of the working environment and the evolution of the system. The neural network model can also store and self-learn multiple processes to obtain the function of flexible self-adjustment. In the training process of the neural network model, optimization algorithms such as gradient descent, Adam, etc. can be used to continuously adjust the weights and biases of the neural network to minimize the error value between the predicted output and the actual output, and improve the accuracy and stability of the predicted output.
[0065] After the neural network model is embedded in the control system, the specific control method of the control center is as follows:
[0066] 1) The neural network model starts running;
[0067] 2) Each execution component is initialized: the hydraulic cylinder 13 and the screw wedge assembly 16 are initialized to the initial position; each motor driver is initialized to a constant speed; the swing roller motor of each swing roller assembly is initialized to a constant torque; the unwinding and rectifying assembly 25 and the winding and rectifying assembly 37 are initialized to the initial position; the initial position of each execution component is pre-set during installation;
[0068] 3) The neural network model predicts the initial control amount of the execution component according to the difference between the actual parameter value and the set parameter value of each execution component;
[0069] 4) The actual parameter value detected by each detection component in real time is input into the neural network model as a lightweight input vector;
[0070] For roll gap control, the input vector is the actual displacement vector of each roll gap displacement sensor 17 and the offline detected membrane thickness. After the roll gap parameters of each roll gap displacement sensor 17 are input, priority screening is also needed. The logic of screening is: from front to back along the main conveying direction of the membrane, the priority gradually increases, that is, the closer to the roll gap 10 at the end of the calendering, the more priority is given to the adjustment, which can reduce the impact on other roll gaps 10 when adjusting a certain roll gap 10, making the adjustment more accurate and efficient;
[0071] For speed control, the input vector is the speed of each motor;
[0072] For tension control, the input vector is the swing roller angle of the swing roller assembly (actual angle and set angle), the speed and acceleration of the swing roller motor;
[0073] For deviation correction control, the input vector is the displacement of the metal foil and the pole piece deviation;
[0074] 5) The neural network model calculates a lightened output vector;
[0075] For roll gap control, the output vector is the extension and retraction amount of the double rod of the hydraulic cylinder 13;
[0076] For strip speed control, the output vector is the motor speed of the reference roller;
[0077] For tension control, the output vector is the speed and acceleration of the swing roller motor of the swing roller assembly;
[0078] For deviation correction control, the output vector is the twist angle of the roller of the deviation correction assembly;
[0079] 6) The lightened output vector, the expert knowledge base output, and the historical data fitting output are combined to obtain a linear regression output quantity, and the parameters of the linear regression output quantity are the same as those of the lightened output vector;
[0080] 7) The linear regression output quantity is input into a PID controller, and the PID controller adjusts the output quantity in real time according to the output vector (prediction control quantity) of the neural network model;
[0081] The adjustment function of the PID controller is: u'(t) = u(t) + f(Y) (where u(t) is usually a function based on the error e(t), including proportional, integral, and differential terms, and the error e(t) refers to the difference between the actual parameter value and the set parameter value; f(Y) is an empirical adjustment function).
[0082] 8) According to the values detected by each detection component, it is determined whether the signal is in a stable state: if it is stable, the control is completed; if it is not stable, it returns to the start operation and runs again until the signal is stable.
[0083] The calender control system of the present application introduces a neural network algorithm combined with a PID controller, and the neural network algorithm also combines an expert knowledge base output and a historical data fitting output to construct a linear regression model, obtains a linear regression output, and then inputs the PID controller, and the PID controller automatically adjusts the control amount in real time according to the linear regression output, so that the roll gap 10, the belt speed, the tension and the deviation correction can be more accurately controlled, when facing a nonlinear, time-varying or multivariable system, the complex dynamic performance and the multivariable coupling can also be accurately processed, the adaptability and the robustness are better, the control performance of the system is improved, the stability and the accuracy of the system are improved, and the production efficiency and the product quality of the calender are also improved.
[0084] As shown in Figure 9 , in the present application, a discrete-time system is used in the PID controller, the discrete-time system can suppress noise and interference through digital filtering and other technologies, is more robust than a continuous-time system and is suitable for real-time control applications, and can realize a complex control strategy through an easy algorithm. In the present application, the PID uses an Euler transformation discretization control algorithm: where P is a proportional band, the smaller the P is, the larger the amplification value K P of the error signal (e(t)) is; I represents an integral part of the control algorithm, which is used to eliminate the “residual error”, and if this part is not present, the set value and the process value may be different even in a stable state; finally, D represents a differential part of the algorithm, which has two goals, it makes the system more flexible to sudden changes in the set value or disturbances, but most importantly, it helps to stabilize the system.
[0085] The transfer function of the PID is:
[0086]
[0087] where s is the Laplace variable (s=α+jω), K p is the amplification coefficient, T I is the integral time, T D is the derivative time, and {K p , T I , T D} is a parameter set of the PID, which depends on the physical properties in the process.
[0088] The adjustment function of the PID is:
[0089]
[0090] where the integral of e(t) can be approximated as a rectangle and is
[0091]
[0092] where e'(t) can be approximated as a straight line, i.e.
[0093]
[0094] Therefore, PID in using Euler transformation discrete control algorithm, get the following expression:
[0095]
[0096] The roll gap control of the single-sided double-sided pole piece forming composite integrated machine of example one is taken as an example, and the detailed control process is as follows:
[0097] When the calender set 100 is installed, the calender rolls 11 are tangent to each other, the calender rolls 11 are tangent to the back roll 12, and the roll gaps 101-107 are adjusted according to the initial position; the thickness of the film produced by the calender set 100 is detected offline, that is, after the calender set 100 starts to run stably, part of the film is cut off for weighing and thickness measurement, and the thickness and area density are detected, after detection, the thickness data is input into the control center, at the same time, the control center converts the thickness information into x(t) and inputs it into the DRNN, the DRNN outputs the parameter set {K P1 ,T I1 ,T D1 ,e1(t)} after deep cycle calculation, at the same time, the expert knowledge base and the historical data fitting formula also output the parameter sets {K P2 ,T I2 ,T D2 ,e2(t)} and {K P3 ,T I3 ,T D3 ,e3(t)}, the three parameter sets are calculated through a linear regression model to output the parameter set {K P ,T I ,T D ,e(t)}.
[0098] After the parameter set {K P ,T I ,T D ,e(t)} is output, the adjusted roll gaps 101-107 need to be compared in size to meet the relationship of roll gap 101> roll gap 102≥ roll gap 104≥ roll gap 103> roll gap 105> roll gap 106> roll gap 107, through the optimal control theory, the priority control is performed against the total material advancing direction (i.e. from roll gap 101 to roll gap 107, the priority level is improved in turn, the priority level of the subsequent roll gap 10 is better than that of the previous roll gap), if the condition is not met, the DRNN is returned for recalculation, if the size order relationship is met, the next step is performed.
[0099] The one or more parameter sets {K PT I ,T D The output of the DRNN {e(t)} is sent to one or more PIDs that control different roll gaps. The PIDs calculate the displacement vector command through a discrete algorithm and send the command to the servo motors of the screw wedge assembly 16 and the servo hydraulic valve group of the hydraulic cylinder 13. The servo motors of the screw wedge assembly 16 and the servo hydraulic valve group of the hydraulic cylinder 13 operate synchronously to quickly and accurately adjust the roll gap 10. After the motion is completed, the roll gap displacement sensor 17 feeds back the displacement of each roll gap to the PID and the DRNN to ensure the accuracy of the execution process. If the accuracy is met, the servo motors of the screw wedge assembly 16 and the servo hydraulic valve group of the hydraulic cylinder 13 stop operating and remain in the current state.
[0100] After the adjustment is completed, the thickness and area density of the produced film are detected using the offline detection method. If the thickness information detected offline meets the fluctuation range (within ±5%), the roll gaps are fixed. If the fluctuation range is not met (exceeding ±5%), the DRNN is continuously inputted for calculation, and the above steps are repeated until the thickness of the produced film is within the fluctuation range (±5%). At this time, the single roll gap control process is completed, and each execution component remains in the current state.
[0101] The above description is an explanation of the present application, not a limitation. The present application can be modified in any form without departing from the spirit of the present application.
Claims
1. An AI algorithm-based dry pole piece forming control method, characterized in that: The calender control system adjusts each roll gap (10) according to the finished film thickness, roll gap (10) displacement information, hydraulic cylinder (13) double rod position information and angle information of the driving motor of the screw wedge block assembly (16); after the adjustment of each roll gap (10), one of the remaining roll gaps (10) is greater than or equal to the previous roll gap (10) except the first and last roll gaps (10); The control center of the calendering control system uses a DRNN neural network algorithm to calculate the signals to obtain a lightweight output. A linear regression model is established for the lightweight output, an expert knowledge base output, and a historical data fitting output. A linear regression formula of the three is: k1, k2, and k3 are weights of the outputs of the three, b is a constant, a control amount calculated by the linear regression model is used as an input vector of a PID controller, an output amount calculated by the PID controller is output to a corresponding actuator, and a corresponding part is adjusted. The control process of the roll gap includes: the control center converts and calculates the offline detection data of the film produced by the calender unit (100), outputs a parameter set {K P1 ,T I1 ,T D1 ,e1(t)}, the expert knowledge base and the historical data fitting formula output a parameter set {K P2 ,T I2 ,T D2 ,e2(t)} and {K P3 ,T I3 ,T D3 ,e3(t)}, three parameter sets output a parameter set {K P ,T I ,T D ,e(t)} through a linear regression model; one or more parameter sets {K P ,T I ,T D ,e(t)} meeting the roll gap size relationship are output to one or more PIDs controlling different roll gaps, the PID outputs a displacement vector instruction to the servo motor of the screw wedge assembly (16) and the servo hydraulic valve group of the hydraulic cylinder (13) through a discrete algorithm, and the servo motor of the screw wedge assembly (16) and the hydraulic cylinder (13) are synchronously operated to adjust the roll gap (10).
2. The AI algorithm-based dry pole piece forming control method according to claim 1, characterized in that: The adjustment priority of each roll gap (10) is that the priority of the roll gap (10) after the material in the total direction of movement is higher than the priority of the roll gap (10) before the material. 3.The AI algorithm-based dry lamination sheet forming control method of claim 1, wherein: The weight values k1, k2 and k3 of the linear regression formula are 0.4, 0.3 and 0.3 respectively.
4. The AI algorithm-based dry pole piece forming control method according to claim 1, characterized in that: After the adjustment of the roll gap (10) is completed, the thickness and area density of the film are detected by an offline detection method; if the offline detected thickness is within ±5%, the roll gaps (10) are fixed; if the offline detected thickness is out of the range of ±5%, the calculation is returned and the adjustment steps are repeated until the thickness of the film is within the range of ±5%. 5.The AI algorithm-based dry lamination sheet forming control method of claim 1, wherein: PID controllers use discrete time systems; the transfer function of a PID is , s is the Laplace variable (s = a + jco), K p is the amplification factor, T I is the integration time, T D is the derivation time; The adjustment function of the PID is , e(t) is approximately integrated as a rectangle and , e'(t) is approximately a straight line , the control expression of the PID is .
6. The AI algorithm-based dry pole piece forming control method according to claim 1, characterized in that: The calender control system adjusts the speed of the servo motor according to the speed information of the servo motor, adjusts the unwinding tension of the metal foil and the winding tension of the pole piece according to the tension information collected by the tension sensor, and adjusts the position of the metal foil and the pole piece according to the displacement information detected by the ultrasonic sensor.
7. An electrode sheet forming and compounding all-in-one machine, characterized by: The AI algorithm-based dry method pole piece forming control method of claim 1 is used for control, including at least one set of calender unit (100), one set of unwinding unit (200) and one set of winding unit (300), the unwinding unit (200) and the winding unit (300) are arranged on one side of the calender unit (100) for pole piece composite forming; the calender unit (100) includes a plurality of calender rollers (11) arranged in sequence along the main conveying direction X, and each two adjacent calender rollers (11) have a roll gap (10) therebetween.
8. The pole piece forming and combining integrator according to claim 7, characterized in that: The roll gap (10) between each adjacent calender roller (11) gradually decreases from front to back along the main conveying direction X, and one of the remaining roll gaps (10) is equal to the previous roll gap (10) except the first and last roll gaps (10).
9. The pole piece forming and combining integrator according to claim 7, characterized in that: A bearing seat (15) is correspondingly arranged on each calender roller (11), a screw wedge block assembly (16) is arranged between the corresponding bearing seats (15) of adjacent calender rollers (11), the screw wedge block assembly (16) is matched with the bearing seat (15) through a slope, the screw wedge block assembly (16) moves along the axial direction of the calender roller (11) to adjust the width of the roll gap (10) between the two calender rollers (11); a roll gap displacement sensor (17) is further arranged between the corresponding bearing seats (15) of adjacent calender rollers (11).
10. The pole piece forming and combining integrator according to claim 7, characterized in that: The first calender roller (11) is provided with a back roller (12) on the side opposite to the main conveying direction X, the back roller (12) is closely attached to the first calender roller (11), the back roller (12) and the last calender roller (11) on the side same as the main conveying direction X are respectively provided with a hydraulic cylinder (13) on the side opposite to the main conveying direction X, the hydraulic cylinder (13) controls the displacement of the back roller (12) or the last calender roller (11) by controlling the extension and retraction of the double rods thereof through a valve group.
11. The pole piece forming and combining integrator according to claim 7, characterized in that: The roller diameter of the back roller (12) and the last two calender rollers (11) is greater than the roller diameter of the other calender rollers (11); each calender roller (11) is respectively connected with an independent motor (14) and is driven by the independent motor (14), and the rotation speed of each calender roller (11) gradually increases from front to back along the main conveying direction X.
12. The pole piece forming and integrating machine of claim 7, wherein: The roller diameter of the first and second calender rollers (11) close to the back roller (12) is smaller than the roller diameter of the other calender rollers (11) and the back roller (12).
13. The pole piece forming and integrating machine of claim 7, wherein: Each calender roller (11) can be independently adjusted in the direction perpendicular to the main conveying direction X; the shaft centers of the calender rollers (11) are located in the same plane or the shaft centers of the calender rollers (11) are staggered, when the shaft centers are staggered, the extension lines of the shaft center connecting lines between each pair of adjacent two calender rollers (11) intersect with the extension line of the shaft center connecting line between the adjacent other pair of adjacent two calender rollers (11) at a point; at least one calender roller (11) is a heatable roller, and the heating temperature is between 150 DEG C and 200 DEG C.
14. The pole piece forming and combining integrator according to claim 7, characterized in that: The unwinding set (200) is provided with a metal foil unwinding shaft (21), between the metal foil unwinding shaft (21) and the nip (10) of the last two calender rollers (11) of the calender set (100), an interface platform (27), an unwinding tension sensor (23), a tension swing roller (24), an unwinding deviation correction assembly (25) and an unwinding ultrasonic sensor (26) are sequentially arranged along the unwinding direction of the metal foil, and unwinding guide rollers (28) are arranged between adjacent components.
15. The pole piece forming and integrating machine of claim 7, wherein: The unwinding set (200) is provided with a diaphragm unwinding shaft (22), and a plurality of unwinding guide rollers (28) are arranged between the diaphragm unwinding shaft (22) and the nip (10) of the last two calender rollers (11) of the calender set (100).
16. The pole piece forming and integrating machine of claim 7, wherein: The winding set (300) is provided with a winding shaft (31), between the winding shaft (31) and the nip (10) of the last two calender rollers (11) of the calender set (100), a first winding tension sensor (32), a first swing roller assembly (33), a tension blocking assembly (34), a second winding tension sensor (35), a second swing roller assembly (36), a winding deviation correction assembly (37) and a winding ultrasonic sensor (38) are sequentially arranged along the winding direction of the pole piece, and winding guide rollers (39) are arranged between adjacent components.
Citation Information
Patent Citations
Roll gap control method and system based on BP (back propagation) neural network and PID (proportion-integration-differentiation) parameter optimization
CN103639211A
Material weighing major data detection and packaging intelligent control system
CN114397809A