Rectification system based on closed loop of linear sensor

Abstract

The invention discloses a linear sensor closed-loop-based deviation rectification system, and relates to the technical field of closed-loop deviation rectification, the linear sensor closed-loop-based deviation rectification system comprises a deviation rectification system, and the deviation rectification system comprises a distributed sensor array, a dynamic cooperative controller, an actuator group, a double-closed-loop control module and a real-time data fusion module which are arranged in the width direction of a deviation-rectified material; the specific system is as follows: the distributed sensor array forms an overlapped detection area through V-shaped staggered layout, and the system has the advantages that through the V-shaped staggered layout of the distributed sensor array, the full width direction of a material is covered, the blind area problem of traditional single-point detection is thoroughly solved, and full-area non-blind-area detection is realized; and in combination with + / -0.01 mm high-precision acquisition of the magnetostrictive sensor and an LMS adaptive filtering algorithm, the comprehensiveness and reliability of offset data are remarkably improved, and accurate input is provided for subsequent control.

Description

Technical Field

[0001] The present invention relates to the technical field of closed-loop deviation correction, and in particular to a deviation correction system based on a linear sensor closed loop. Background Art

[0002] In industrial production processes such as material processing, printing, and coating, materials (such as steel strips, films, composite fiber films, etc.) are prone to lateral or longitudinal deviation during transmission or processing due to factors such as tension fluctuations and mechanical vibrations, resulting in problems such as slitting misalignment, printing registration deviation, and uneven coating, seriously affecting product quality and production efficiency. Traditional correction systems mostly use single-point sensors to detect deviations, which have detection blind spots and are difficult to cover the full width of the material. Control strategies usually rely on static PID algorithms and cannot dynamically adapt to changes in working conditions such as material speed and tension. The actuator activation mode is single (such as full start), resulting in energy redundancy. In addition, there is a lack of effective electromagnetic anti-interference design, and the signal stability is poor. These problems make traditional systems insufficient in correction accuracy, high energy consumption, and weak adaptability in high-speed operation or low-rigidity material scenarios. In response to the above shortcomings, the existing technology urgently needs a new correction system that can achieve full-area blind-spot detection, dynamically optimize control parameters, hierarchical coordinated actuator actions, and have strong anti-interference capabilities, so as to improve the correction efficiency and accuracy under different materials and different working conditions, and meet the modern industry's demand for high-quality and intelligent production. To this end, we propose a correction system based on a linear sensor closed loop. Summary of the Invention

[0003] The object of the present invention is to provide a correction system based on a linear sensor closed loop.

[0004] To achieve the above-mentioned object, the present invention provides the following technical solution: a correction system based on a linear sensor closed loop, comprising a correction system, wherein the correction system comprises: The distributed sensor array, dynamic collaborative controller, actuator group, dual closed-loop control module and real-time data fusion module are arranged along the width of the material to be corrected. The specific system is as follows: The distributed sensor array forms overlapping detection areas through a V-shaped staggered layout; The dual closed-loop control module dynamically adjusts control parameters through the collaboration of inner-loop real-time PID control and outer-loop LSTM prediction optimization; The dynamic collaborative controller hierarchically activates the actuator combination based on the offset pattern and allocates output weights to improve the correction efficiency; The real-time data fusion module integrates multi-sensor data into a visual decision basis to assist the controller in optimizing the correction strategy.

[0005] As a further solution of the present invention: the distributed sensor array is a magnetostrictive linear sensor group, specifically characterized by: Adopting V-shaped staggered layout, the overlap rate of adjacent sensor detection areas is ≥30%, covering the entire area in the width direction of the material; Each set of sensors synchronously collects longitudinal displacement ΔY (unit: mm) and lateral offset ΔX (unit: mm), with a detection accuracy of ≤±0.01mm; The sensor output signal is processed by the adaptive filtering module, which updates the filter coefficients based on the LMS (least mean square) algorithm. The calculation formula is: ; in, For the The filter coefficient vector for the iteration, is the number of iterations the algorithm uses to update the filter coefficients, is the convergence factor (0.01≤ ≤0.1), For the The error signal of the iteration (the difference between the sensor output and the expected signal), For the The input signal vector (raw sensor output signal) of the iteration.

[0006] As a further solution of the present invention, the dual closed-loop control module includes an inner loop control unit, which generates a PID control signal based on the ΔX and ΔY data fed back by the sensor in real time. The control law expression is: ; in, for The control output at the moment (voltage signal to drive the actuator, unit: V), for The offset error at the moment (the difference between the actual position and the target position, unit: mm), is the integral time variable (value range: 0≤ ≤t), for The offset error at the moment (unit: mm), From the initial time 0 to the current time The cumulative error (unit: mm·s) for The first-order time derivative of the offset error (i.e., the rate of change of the offset error, unit: mm / s), is the proportional coefficient (unit: V / mm, proportional gain between control output and current error), is the integral coefficient (unit: V / (mm·s), which controls the integral gain of the output and the accumulated error. is the differential coefficient (unit: V·s / mm, the differential gain of the control output and the error change rate).

[0007] As a further solution of the present invention, the dual closed-loop control module further includes an outer loop prediction unit, which uses an LSTM (long short-term memory) model to predict the offset trend and optimize the PID parameters, which is specifically implemented as follows: Input parameters: Historical offset data , , , , ( is the time window length, =20), set the material tension to (Unit: N), material running speed is (Unit: m / s); Output parameter: offset prediction value within the next 50ms (Unit: mm), proportional coefficient increment (Unit: V / mm), integral coefficient increment (Unit: V / (mm·s)); Prediction value calculation formula: ; The outer loop unit updates the prediction result every 10ms and 、 Feedback is sent to the inner loop unit to achieve dynamic optimization of PID parameters.

[0008] As a further solution of the present invention: the actuator weight distribution mechanism of the dynamic collaborative controller, the actuator output weight calculation formula is: ; in, For the The output weight of each actuator (dimensionless, 0≤ ≤1), is the material stiffness coefficient (unit: N / mm, 2000 for steel and 500 for plastic), For the The offset detected by the sensor corresponding to each actuator (unit: mm), The sum of the offsets detected by all sensors (unit: mm). This weight distribution mechanism makes the actuator output proportional to the offset of the corresponding area, achieving precise correction of each area of the material.

[0009] As a further solution of the present invention: the offset mode hierarchical control strategy of the dynamic cooperative controller is as follows: Set the offset threshold to A Local offset: When ≤A (A is the threshold value, defined as 0.5% of the material width, i.e. A=0.005W, W is the material width, unit: mm), only the two adjacent actuators in the offset area are activated to reduce energy redundancy; Overall offset: When When >A, all actuators on the entire line are activated and weights are allocated according to the formula in claim 5 to ensure synchronous correction of the overall position of the material; The threshold value A can be dynamically adjusted according to the material type through the controller human-machine interface.

[0010] As a further solution of the present invention: the closed-loop response mechanism of the actuator group is composed of a plurality of servo motor-driven correction mechanisms. The servo motor is connected to a high-precision encoder (resolution ≤ 0.005mm). The encoder feeds back position data to the inner loop control unit in real time, forming a position closed loop of "sensor detection-controller calculation-actuator action-encoder feedback". The system response time is ≤ 10ms, ensuring the real-time performance of the correction action. As a further solution of the present invention: the electromagnetic anti-interference design of the system, the electromagnetic interference is as follows: The sensor signal transmission line is wrapped with a Permalloy shielding layer (thickness ≥ 0.1mm), with an electromagnetic interference shielding effectiveness of ≥ 60dB in the 100kHz~10MHz frequency band; A common-mode choke (inductance ≥ 1mH) is connected in series to the controller power input to suppress common-mode interference in the 10kHz~10MHz frequency band and ensure the stability of the control signal.

[0011] By adopting the above technical solution, compared with the prior art, the beneficial effects of the present invention are: 1. This invention uses a V-shaped staggered layout of distributed sensor arrays to cover the entire width of the material, completely solving the blind spot problem of traditional single-point detection and achieving full-area blind spot detection. Combined with the high-precision acquisition of ±0.01mm of magnetostrictive sensors and the LMS adaptive filtering algorithm, the comprehensiveness and reliability of offset data are significantly improved, providing precise input for subsequent control. 2. The present invention uses a dual closed-loop control module to coordinate the inner-loop real-time PID and outer-loop LSTM predictive optimization. The LSTM model is used to predict the future 50ms offset trend (updated every 10ms) based on historical offset data, material tension, and operating speed. PID parameters (such as proportional and integral coefficient increments) are dynamically adjusted to effectively compensate for control lag caused by operating condition changes (such as speed fluctuations and tension mutations). This allows the system to maintain a response time of ≤10ms and stable correction accuracy in high-speed, variable operating conditions. 3. This invention utilizes a hierarchical control strategy through a dynamic collaborative controller: For local deviations (≤0.5% of the material width), only two adjacent actuators are activated, reducing energy redundancy. For global deviations, all actuators are activated, and a weighted distribution mechanism, where output is proportional to the offset in the corresponding area, allows for precise correction of each area of the material. This reduces energy consumption by over 30% compared to the traditional full-activation mode, balancing correction efficiency and energy efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a schematic diagram of the process flow of the correction system in an embodiment of the present invention; Figure 2 This is a schematic diagram of a double closed-loop control process in an embodiment of the present invention; Figure 3 Schematic diagram of the dynamic collaborative control process in an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The specific embodiments of the present invention will be further described below in conjunction with the accompanying drawings. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0014] In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0015] Please see the attached Figure 1 -Attached Figure 3 The present invention provides a correction system based on a linear sensor closed loop, including a correction system, the correction system including: The distributed sensor array, dynamic collaborative controller, actuator group, dual closed-loop control module and real-time data fusion module are arranged along the width of the material to be corrected. The specific system is as follows: The distributed sensor array forms overlapping detection areas through a V-shaped staggered layout; The dual closed-loop control module dynamically adjusts control parameters through the collaboration of inner-loop real-time PID control and outer-loop LSTM prediction optimization; The dynamic collaborative controller activates actuator combinations based on the deviation pattern and allocates output weights to improve the correction efficiency. The real-time data fusion module integrates multi-sensor data into a visual decision-making basis, assisting the controller in optimizing the correction strategy.

[0016] In one embodiment of the present invention, the distributed sensor array is a magnetostrictive linear sensor group, and its specific features are: Adopting V-shaped staggered layout, the overlap rate of adjacent sensor detection areas is ≥30%, covering the entire area in the width direction of the material; Each set of sensors synchronously collects longitudinal displacement ΔY (unit: mm) and lateral offset ΔX (unit: mm), with a detection accuracy of ≤±0.01mm; The sensor output signal is processed by the adaptive filtering module, which updates the filter coefficients based on the LMS (least mean square) algorithm. The calculation formula is: ; in, For the The filter coefficient vector for the iteration, is the number of iterations the algorithm uses to update the filter coefficients, is the convergence factor (0.01≤ ≤0.1), For the The error signal of the iteration (the difference between the sensor output and the expected signal), For the The input signal vector (raw sensor output signal) of the iteration.

[0017] In one embodiment of the present invention, the dual closed-loop control module includes an inner loop control unit that generates a PID control signal based on the ΔX and ΔY data fed back by the sensor in real time. The control law expression is: ; in, for The control output at the moment (voltage signal to drive the actuator, unit: V), for The offset error at the moment (the difference between the actual position and the target position, unit: mm), is the integral time variable (value range: 0≤ ≤t), for The offset error at the moment (unit: mm), From the initial time 0 to the current time The cumulative error (unit: mm·s) for The first-order time derivative of the offset error (i.e., the rate of change of the offset error, unit: mm / s), is the proportional coefficient (unit: V / mm, proportional gain between control output and current error), is the integral coefficient (unit: V / (mm·s), which controls the integral gain of the output and the accumulated error. is the differential coefficient (unit: V·s / mm, the differential gain of the control output and the error change rate).

[0018] In one embodiment of the present invention, the dual closed-loop control module further includes an outer loop prediction unit, which uses an LSTM (long short-term memory) model to predict the deviation trend and optimize the PID parameters, specifically implemented as follows: Input parameters: Historical offset data , , , , ( is the time window length, =20), set the material tension to (Unit: N), material running speed is (Unit: m / s); Output parameter: offset prediction value within the next 50ms (Unit: mm), proportional coefficient increment (Unit: V / mm), integral coefficient increment (Unit: V / (mm·s)); Prediction value calculation formula: ; The outer loop unit updates the prediction result every 10ms and 、 Feedback is sent to the inner loop unit to achieve dynamic optimization of PID parameters.

[0019] In one embodiment of the present invention, the actuator weight allocation mechanism of the dynamic collaborative controller and the actuator output weight calculation formula are as follows: ; in, For the The output weight of each actuator (dimensionless, 0≤ ≤1), is the material stiffness coefficient (unit: N / mm, 2000 for steel and 500 for plastic), For the The offset detected by the sensor corresponding to each actuator (unit: mm), The sum of the offsets detected by all sensors (unit: mm). This weight distribution mechanism makes the actuator output proportional to the offset of the corresponding area, achieving precise correction of each area of the material.

[0020] In one embodiment of the present invention, the offset mode hierarchical control strategy of the dynamic cooperative controller is as follows: Set the offset threshold to A Local offset: When ≤A (A is the threshold value, defined as 0.5% of the material width, i.e. A=0.005W, W is the material width, unit: mm), only the two adjacent actuators in the offset area are activated to reduce energy redundancy; Overall offset: When When >A, all actuators on the entire line are activated and weights are allocated according to the formula in claim 5 to ensure synchronous correction of the overall position of the material; Threshold A can be dynamically adjusted according to the material type through the controller human-machine interface.

[0021] In one embodiment of the present invention: the closed-loop response mechanism of the actuator group, the actuator group is composed of a correction mechanism driven by multiple servo motors, the servo motors are connected to a high-precision encoder (resolution ≤ 0.005mm), and the encoder feeds back position data to the inner loop control unit in real time, forming a position closed loop of "sensor detection-controller calculation-actuator action-encoder feedback", and the system response time is ≤ 10ms, ensuring the real-time nature of the correction action.

[0022] In one embodiment of the present invention: the electromagnetic anti-interference design of the system, the electromagnetic interference is as follows: The sensor signal transmission line is wrapped with a Permalloy shielding layer (thickness ≥ 0.1mm), with an electromagnetic interference shielding effectiveness of ≥ 60dB in the 100kHz~10MHz frequency band; A common-mode choke (inductance ≥ 1mH) is connected in series to the controller power input to suppress common-mode interference in the 10kHz~10MHz frequency band and ensure the stability of the control signal.

[0023] Example 1: Application of the correction system in the scenario of high-speed slitting of steel coil materials Application scenario: In a steel company's steel coil slitting production line, a 2mm thick cold-rolled steel plate runs at a speed of 8m / s and needs to be slit into a 1200mm wide steel strip. Due to the high speed of the steel plate and the high material stiffness ( = 2000N / mm), the traditional single-point detection system is prone to deviation detection lag due to vibration, and the correction system of the present invention is required to achieve high-precision real-time correction; System Configuration: Distributed sensor array: Ten sets of magnetostrictive linear sensors are deployed along the width of the steel plate (1200mm), using a V-shaped staggered layout (150mm spacing between adjacent sensors, 35% overlap), covering the entire width. Each set of sensors simultaneously collects ΔX (lateral offset) and ΔY (longitudinal displacement), with a detection accuracy of ±0.01mm. Actuator group: Equipped with 8 servo motor-driven correction rollers (corresponding to the sensor array partitions), each motor is connected to an encoder with a resolution of 0.005mm and a feedback frequency of 100Hz; Control parameter initialization: PID initial coefficient =1.5V / mm, =0.05V / (mm·s), =0.2V·s / mm; LSTM model time window a=20 (i.e., collecting historical data for the previous 200ms), predicting the offset for the next 50ms; offset threshold A=0.005W=6mm (W=1200mm); Operation process: Sensor detection and filtering: When the steel strip is running, the sensor array collects ΔX in each area in real time (for example, the third group of sensors detects ΔX = 4.2mm, and the fifth group detects ΔX = 5.8mm). The original signal is processed by LMS adaptive filtering (μ = 0.05) to eliminate high-frequency noise (such as 1kHz interference) caused by steel strip vibration, and outputs stable ΔX and ΔY data to the dual closed-loop control module; Dual closed-loop control coordination: Inner loop PID control: Calculate the current offset error based on the target position (the center line of the steel plate is aligned with the slitting tool) = Actual position - target position (such as at a certain moment =3.5mm), according to the PID control law: , output control voltage signal to the dynamic cooperative controller; Outer loop LSTM prediction: LSTM model inputs the offset data 200ms before ( to ), current tension =5000N, speed =8m / s, predict the offset value of the next 50ms =4.1mm (with actual offset error ≤ 0.02mm), and calculate =+0.1V / mm, =+0.01V / (mm·s), fed back to the inner loop to update the PID parameters ( =1.6, =0.06), to compensate for the control lag caused by speed fluctuation; Dynamic coordination of actuators: Since each sensor detects (Single-area offset) The maximum value is 5.8 mm ≤ A = 6 mm (local offset), and the dynamic collaborative controller only activates the two adjacent actuators (corresponding to the second, fourth, and sixth groups of actuators) in the offset area (the third and fifth groups of sensors); According to the weight formula =2000×( / )Calculate the output weight ( =4.2+5.8+…=18mm), the fourth group of actuator weights =2000×5.8 / 18≈644 (normalized to 0.36), driving it to output greater thrust and achieve precise correction in local areas; Closed-loop response and anti-interference: After the actuator moves, the encoder feeds back the position of the correction roller in real time (for example, the displacement of the fourth group of actuators is 0.8mm), and the inner ring unit verifies the error. The thickness is reduced to 0.5mm, and the system response time is 8ms. At the same time, the permalloy shielding layer (thickness 0.2mm) of the sensor signal transmission line effectively suppresses the 2MHz electromagnetic interference generated by the slitting machine (shielding effectiveness 65dB), ensuring the stability of the control signal. Test results: Within 30 minutes of system operation, the steel plate slitting deviation was ≤±0.05mm, a four-fold improvement compared to the traditional system (deviation ±0.2mm). Energy consumption was also reduced by 30% compared to the full actuator activation mode, verifying the real-time performance and energy efficiency advantages in high-speed scenarios. Example 2: Application of the deviation correction system in a plastic film printing production line Application scenario: In a BOPP plastic film printing line of a packaging material factory, a film with a thickness of 0.05mm runs at a speed of 5m / s and a pattern with a width of 800mm needs to be printed. Due to the low stiffness of the film ( =500N / mm), is easily affected by tension fluctuations (±50N), and traditional systems often cause pattern deviation due to detection blind spots or control lags. The system of the present invention is required to achieve flexible correction of low-rigidity materials; System Configuration: Distributed sensor array: 8 sets of magnetostrictive sensors are arranged along the film width (800mm), with a V-shaped staggered layout (adjacent spacing of 100mm, overlap rate of 32%), with a detection accuracy of ±0.01mm; adaptive filtering module =0.08 (more sensitive to film signal noise); Actuator group: equipped with 6 servo motor-driven air-floating correction rollers (to avoid mechanical contact damage to the film), encoder resolution 0.003mm, feedback frequency 150Hz; Control parameter initialization: PID initial coefficient =0.8V / mm (for low stiffness materials, the proportional gain needs to be reduced to avoid overshoot), =0.03V / (mm·s), =0.1V·s / mm; LSTM time window =20 (collect data for the first 200ms), predict offset for the next 50ms; offset threshold A=0.005W=4mm (W=800mm); Operation process: Full-area, blind-spot-free detection: When the film is running, the sensor array uses a V-shaped overlapping layout (for example, the second group of sensors covers the 30-130mm area, and the third group covers the 80-180mm area), eliminating the blind spot of 60-80mm in traditional single-point detection. At a certain moment, the fifth group of sensors detected ΔX = 3.2mm (lateral offset), and the sixth group detected ΔX = 4.5mm (exceeding A = 4mm, triggering the overall offset mode). Dual closed-loop parameter dynamic optimization Inner loop PID control: current error =4.5mm (the target position is that the center line of the film is aligned with the printing roller), calculation: , output control voltage to drive the actuator; Outer loop LSTM prediction: Input historical offset data ( to , maximum offset 3.8mm), current tension =300N (fluctuates to 250N), speed =5m / s, predicting the next 50ms offset =5.1mm, and calculate =+0.2V / mm (increase proportional gain to cope with the increase in offset caused by tension reduction), =+0.005V / (mm·s), after the inner loop is updated =1.0, =0.035; Hierarchical control and weight distribution: Because Group 6 =4.5mm>A=4mm (overall offset), the dynamic collaborative controller activates all 6 actuators in the line, according to the weight formula =500×( c / )( =3.2+4.5+…=12mm), the sixth group of actuator weights =500×4.5 / 12≈187 (normalized to 0.31), driving it to output a larger thrust, and other actuators are distributed proportionally to ensure the overall synchronous correction of the film; Anti-interference and visual assistance: The printing press's inverter generates 500kHz electromagnetic interference. The sensor transmission line's permalloy shield (0.15mm thick) has a shielding effectiveness of 62dB, reducing signal noise from ±0.02mm to ±0.005mm. The real-time data fusion module integrates the ΔX and ΔY data from each sensor into a heat map (red indicates high-drift areas). The operator, through the human-machine interface, can observe significant drift in Group 6 and manually fine-tune the threshold A to 3.5mm (to accommodate variations in film thickness), further improving control accuracy. Test results: During system operation, film printing deviation was ≤±0.03mm, and the pattern overprint pass rate increased from 85% to 98%. The air-floating actuator prevented film scratches, increasing the overall yield by 12%, validating the system's flexible control capabilities for low-rigidity materials. Example 3: Application of the deviation correction system in the composite fiber material coating process Application scenario: In the composite fiber membrane coating line of a new energy battery factory, a PET / ceramic composite membrane with a thickness of 0.1mm runs at a speed of 3m / s and needs to be evenly coated with electrolyte (coating width 600mm). Due to the uneven stiffness of the material in different areas (local areas =800N / mm, overall =600N / mm), the traditional system is prone to uneven coating thickness due to local wrinkles, and the system of the present invention is required to achieve "local-overall" coordinated correction; System Configuration: Distributed sensor array: 6 sets of magnetostrictive sensors are arranged along the membrane width (600mm), with a V-shaped staggered layout (adjacent spacing 80mm, overlap rate 35%), with a detection accuracy of ±0.01mm; adaptive filtering module =0.06 (balanced noise suppression and signal response); Actuator group: Equipped with 4 servo motor-driven elastic correction rollers (to adapt to the flexible deformation of the composite film), encoder resolution 0.002mm, feedback frequency 200Hz; Control parameter initialization: PID initial coefficient =1.2V / mm, =0.04V / (mm·s), =0.15V·s / mm; LSTM time window =20 (collecting data from the first 200ms), predicting the offset of the next 50ms; offset threshold A=0.005W=3mm (W=600mm), which can be dynamically adjusted through the human-machine interface; Operation process: Multi-parameter detection under complex working conditions: When the composite film is running, the sensor array simultaneously collects ΔX (lateral offset) and ΔY (longitudinal displacement, reflecting material tension fluctuations). At a certain moment, the third group of sensors detected ΔX = 2.8mm (local offset) and ΔY = 0.08mm (longitudinal relaxation caused by tension fluctuations). The fourth group detected ΔX = 3.5mm (exceeding A = 3mm, triggering overall offset). Double closed-loop multi-objective control: Inner loop PID control: with coating uniformity as the goal (lateral deviation required ≤±0.02mm), calculate the current error , drives the actuator to adjust the roller pressure; Outer loop LSTM prediction: Input historical offset data ( to , maximum ΔX=3.2mm), current tension =400N (reduced to 380N due to ΔY fluctuation), speed =3m / s, predicting the next 50ms offset =3.8mm, and calculate =+0.15V / mm (enhanced proportional control), =+0.008V / (mm·s) (acceleration integral correction), after the inner loop is updated =1.35, =0.048; Intelligent switching of hierarchical control: For Group 3 =2.8mm≤A=3mm (local offset), only the two adjacent actuators (groups 2 and 4) are activated, and the output weight =800×2.8 / ( =2.8+3.5+…=9mm)≈249 (normalized to 0.28), focusing on correcting the wrinkle area; For Group 4 =3.5mm>A=3mm (overall offset), activate all 4 actuators, press =600×( / )Assign weights (Group 4 =600×3.5 / 9≈233, normalized to 0.26), ensuring the overall position is adjusted synchronously; Closed-loop response and long-term stability: After the actuator is actuated, the encoder feedback indicates that the fourth group of actuators has a displacement of 0.6 mm, and the inner loop unit verifies the error. The system response time is reduced to 0.01mm, and the system response time is 7ms (meeting the coating process requirement of ≤10ms). The common-mode choke (inductance 1.5mH) on the controller power supply side suppresses the 8MHz common-mode interference generated by the coating machine heating device, and the control signal fluctuation is reduced from ±0.1V to ±0.02V, ensuring parameter stability. Test results: Within one hour of system operation, the composite film coating thickness deviation decreased from ±5μm to ±1μm, significantly improving electrolyte uniformity. The local wrinkle correction success rate was 100%, with no coating breakage incidents caused by offset, validating the advantages of "local-global" collaborative control under complex material working conditions. Although the present invention is disclosed above with reference to preferred embodiments, this is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications without departing from the spirit and scope of the present invention. Therefore, any modifications, equivalent variations, and modifications made to the above embodiments in accordance with the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection defined by the claims of the present invention.

Claims

1. A correction system based on a linear sensor closed loop, comprising a correction system, characterized in that: The correction system includes: The distributed sensor array, dynamic collaborative controller, actuator group, dual closed-loop control module and real-time data fusion module are arranged along the width of the material to be corrected. The specific system is as follows: The distributed sensor array forms overlapping detection areas through a V-shaped staggered layout; The dual closed-loop control module dynamically adjusts control parameters through the collaboration of inner-loop real-time PID control and outer-loop LSTM prediction optimization; The dynamic collaborative controller hierarchically activates the actuator combination based on the offset pattern and allocates output weights to improve the correction efficiency; The real-time data fusion module integrates multi-sensor data into a visual decision basis to assist the controller in optimizing the correction strategy.

2. The linear sensor closed-loop correction system according to claim 1, characterized in that: The distributed sensor array is a magnetostrictive linear sensor group, and its specific features are: Adopting V-shaped staggered layout, the overlap rate of adjacent sensor detection areas is ≥30%, covering the entire area in the width direction of the material; Each set of sensors synchronously collects longitudinal displacement ΔY and lateral offset ΔX, with a detection accuracy of ≤±0.01mm; The sensor output signal is processed by the adaptive filtering module, which updates the filter coefficients based on the LMS algorithm. The calculation formula is: ； in, For the The filter coefficient vector of the iteration, is the number of iterations the algorithm uses to update the filter coefficients, is the convergence factor, For the The error signal of the iteration, For the The input signal vector for the iteration.

3. The linear sensor closed-loop correction system according to claim 2, characterized in that: The dual closed-loop control module includes an inner loop control unit that generates a PID control signal based on the ΔX and ΔY data fed back by the sensor in real time. The control law expression is: ； in, for The control output at each moment, for The offset error at time, is the integration time variable, for The offset error at time, From the initial time 0 to the current time The accumulated error, for The first time derivative of the time offset error, is the proportionality coefficient, is the integration coefficient, is the differential coefficient.

4. The linear sensor closed-loop correction system according to claim 3, characterized in that: The dual closed-loop control module also includes an outer loop prediction unit, which uses the LSTM model to predict the offset trend and optimize the PID parameters. The specific implementation is as follows: Input parameters: Historical offset data , , , , set the material tension to , material running speed is ; Output parameter: offset prediction value within the next 50ms , proportional coefficient increment , integral coefficient increment ; Prediction value calculation formula: ; The outer loop unit updates the prediction result every 10ms and 、 Feedback is sent to the inner loop unit to achieve dynamic optimization of PID parameters.

5. The linear sensor closed-loop correction system according to claim 4, characterized in that: The actuator weight distribution mechanism of the dynamic collaborative controller and the actuator output weight calculation formula are as follows: ； in, For the The output weight of each actuator, is the material stiffness coefficient, For the The offset detected by the sensor corresponding to each actuator, The weight distribution mechanism is the sum of the offsets detected by all sensors. It makes the actuator output proportional to the offset of the corresponding area, thus achieving precise correction of each area of the material.

6. The linear sensor closed-loop correction system according to claim 5, characterized in that: The offset mode hierarchical control strategy of the dynamic cooperative controller is as follows: Set the offset threshold to A Local offset: When ≤A, only two adjacent actuators in the offset area are activated to reduce energy redundancy; Overall offset: When When >A, all actuators on the entire line are activated and weights are allocated according to the formula in claim 5 to ensure synchronous correction of the overall position of the material; The threshold value A can be dynamically adjusted according to the material type through the controller human-machine interface.

7. The linear sensor closed-loop correction system according to claim 6, characterized in that: The closed-loop response mechanism of the actuator group is composed of a plurality of servo motor-driven correction mechanisms. The servo motors are connected to high-precision encoders. The encoders feed back position data to the inner loop control unit in real time. The system response time is ≤10ms, ensuring the real-time performance of the correction action.

8. The linear sensor closed-loop correction system according to claim 7, characterized in that: The electromagnetic anti-interference design of the system, the electromagnetic interference is as follows: The sensor signal transmission line is wrapped with a Permalloy shielding layer, and the electromagnetic interference shielding effectiveness in the 100kHz~10MHz frequency band is ≥60dB; A common-mode choke is connected in series at the controller power input to suppress common-mode interference in the 10kHz~10MHz frequency band and ensure the stability of the control signal.

Images