Intelligent Lamination and Tension Control System and Method for Multi-Wedge Belt Production Line of Washing Machine

Abstract

This invention relates to the field of speed control technology, specifically to an intelligent lamination and tension control system and method for multi-ribbed belt production lines in washing machines. The system includes: a multi-dimensional state information extraction module, a real-time collaborative control module, a control performance evaluation module, and a fuzzy rule self-tuning module. In this invention, by calculating tension error and lamination position deviation, a basis for dynamic adjustment is provided, eliminating reliance on preset fixed parameters. By invoking a fuzzy logic inference engine, tension error, error change rate, and position deviation are fused and processed to achieve synchronous and precise control of the transmission roller speed and the position of the correction mechanism. This effectively replaces the slow-responding mechanical tension device, solving the problems of tension fluctuation and lamination deviation. Furthermore, a long-term comprehensive performance index is introduced to evaluate the control effect. When performance deteriorates, a gradient descent algorithm is immediately triggered to self-tune the fuzzy rule parameters, continuously optimizing the control strategy and ensuring the system maintains high stability and high accuracy.

Description

Technical Field

[0001] This invention relates to the field of speed control technology, and in particular to an intelligent lamination and tension control system and method for multi-wedge belt production lines for washing machines. Background Technology

[0002] Speed ​​control technology is a key component of industrial automation and production processes. Its core focus is on precisely adjusting and maintaining the speed of moving parts in mechanical equipment or production lines through specific control systems and algorithms. This technology systematically encompasses the entire closed-loop or open-loop control process, from sensor detection, signal processing, controller computation to actuator driving, to ensure production process stability, product quality consistency, and maximized production efficiency. Specifically, the intelligent lamination and tension control system for traditional multi-ribbed belt production lines for washing machines refers to a technical solution for controlling the speed and tension of the base belt during the lamination process in multi-ribbed belt manufacturing. Traditional systems typically use motors with preset fixed parameters to drive rollers, and apply and maintain the tension of the base belt during the lamination process using mechanical tension devices, such as spring-loaded pressure rollers or counterweight rollers.

[0003] Existing technologies use preset fixed parameters to drive the motor and rely on mechanical devices such as springs or counterweights to apply tension. This approach has drawbacks in actual operation. Fixed parameters cannot cope with changes in material properties or working conditions during production, resulting in inflexible speed control. Mechanical tension devices are slow to respond and have inertia, making it difficult to quickly compensate for tension fluctuations. This technical solution also lacks a real-time monitoring and adjustment mechanism for the actual position of the baseband during the lamination process. When disturbances occur, the system cannot coordinate the adjustment of speed and tension, nor can it correct positional deviations. This directly leads to unstable baseband tension and decreased lamination accuracy, which in turn affects the product consistency and production stability of multi-wedge belts. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and propose an intelligent lamination and tension control system and method for multi-wedge belt production lines for washing machines.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent lamination and tension control system for a multi-wedge belt production line for washing machines, comprising:

[0006] The multi-dimensional state information extraction module is used to collect voltage signals from tension sensors and image data from industrial cameras, calculate tension error, tension error change rate and bonding position deviation, and transmit the tension error, tension error change rate and bonding position deviation to the real-time collaborative control module, and transmit the tension error and bonding position deviation to the control performance evaluation module.

[0007] The real-time collaborative control module is used to generate a collaborative adjustment signal consisting of speed increase / decrease and translational displacement based on the tension error, the rate of change of the tension error and the overlapping position deviation, by calling a fuzzy logic inference engine; to regulate the transmission roller servo motor connected to it according to the speed increase / decrease in the collaborative adjustment signal; and to regulate the correction mechanism actuator connected to it according to the translational displacement.

[0008] The control performance evaluation module is used to calculate a long-term comprehensive performance index based on the time series of the tension error and the bonding position deviation, compare the long-term comprehensive performance index with the performance degradation judgment threshold to generate a performance degradation trigger instruction, and transmit it to the fuzzy rule self-tuning module.

[0009] The fuzzy rule self-tuning module is used to call the gradient descent algorithm to adjust the center value parameter of the membership function according to the performance degradation trigger instruction, and update the fuzzy logic inference engine with the center value parameter.

[0010] As a further aspect of the present invention, the tension error specifically refers to the difference between the target tension set value and the actual tension measurement value; the tension error change rate includes the difference between the current error value and the error value of the previous sampling period; the overlapping position deviation specifically refers to the difference between the preset overlapping center and the actual overlapping center; the coordinated adjustment signal includes a transmission control command and a correction control command; the performance degradation trigger command includes an optimization start signal and a target module address; and the center value parameter specifically refers to the error membership function center, the error change rate membership function center, and the deviation membership function center.

[0011] As a further aspect of the present invention, the specific function of the multi-dimensional state information extraction module is as follows:

[0012] A signal processing submodule is set up to acquire the voltage signal of the tension sensor, filter and calibrate the voltage signal of the tension sensor, and obtain the actual tension measurement value.

[0013] The tension error is calculated based on the target tension setting value and the actual tension measurement value.

[0014] The current error value and the error value of the previous sampling period are obtained, and the rate of change of tension error is calculated.

[0015] An image processing submodule is set up to acquire the image data from the industrial camera, identify the edges of the multi-wedge strip through edge detection and morphological learning operations, and determine the actual overlapping center.

[0016] The bonding position deviation is calculated based on the difference between the preset bonding center and the actual bonding center;

[0017] A data integration submodule is set up to encapsulate the tension error, the rate of change of the tension error, and the bonding position deviation, and transmit them to the real-time collaborative control module and the control performance evaluation module.

[0018] As a further aspect of the present invention, the specific function of the real-time collaborative control module is as follows:

[0019] An input fuzzification submodule is set up to obtain the tension error, the rate of change of the tension error and the overlap position deviation, and convert them into fuzzy language variables based on the preset membership function;

[0020] A fuzzy inference submodule is set up to call the fuzzy rule library, perform logical inference on the fuzzy linguistic variables, and obtain fuzzy control output;

[0021] An output defuzzification submodule is set up to perform defuzzification calculations on the fuzzy control output using the centroid method, so as to obtain the speed increase / decrease and the translational displacement.

[0022] A control execution submodule is set up to combine the speed increase / decrease and the translational displacement into the coordinated adjustment signal, and drive the transmission roller servo motor and the correction mechanism actuator.

[0023] As a further aspect of the present invention, the specific function of the control performance evaluation module is as follows:

[0024] A data caching submodule is set up to collect and store the time series of the tension error and the bonding position deviation within a preset evaluation period;

[0025] A submodule for calculating indicators is set up to perform calculations based on the time series using formulas. Calculate the aforementioned long-term comprehensive performance index;

[0026] in, This represents the aforementioned long-term comprehensive performance index. This represents the total number of sampling points within the evaluation period. The index representing the sampling point. Representing the The tension error at each sampling point Representing the The overlap position deviation of each sampling point Represents the tension error weighting coefficient. Represents the overlapping deviation weighting coefficient;

[0027] A performance determination submodule is set up to compare the long-term comprehensive performance index with the preset performance degradation determination threshold. When the index exceeds the performance degradation determination threshold, a performance degradation trigger instruction is generated.

[0028] As a further aspect of the present invention, the specific function of the fuzzy rule self-tuning module is as follows:

[0029] A command receiving submodule is configured to receive the performance degradation trigger command and lock the target module address;

[0030] A gradient calculation submodule is set up to calculate the long-term comprehensive performance index. Given the objective function, the gradient descent algorithm is used to calculate the sensitivity gradient of the objective function to the center value parameter.

[0031] The parameter update submodule is configured to, based on the sensitivity gradient, update... Calculate the updated center value parameters;

[0032] in, Representing the The first input variable The membership function in The center value parameter at time, represent The center value parameter at time, Represents the number of iterations. Represents the learning rate. The objective function represents the central value parameter. The partial derivatives;

[0033] The engine update submodule is configured to write the updated center value parameter into the parameter storage area of ​​the fuzzy logic inference engine.

[0034] As a further aspect of the present invention, the process by which the image solving submodule determines the actual overlapping center is specifically as follows:

[0035] The industrial camera image data is subjected to grayscale conversion and Gaussian filtering to obtain a preprocessed image;

[0036] The Canny operator is used to extract the multi-wedge edge contours from the preprocessed image;

[0037] The bilateral straight line equations of the multi-wedge edge contour are located using Hough line detection or contour fitting algorithms.

[0038] Calculate the centerline of the bilateral straight line equation and determine the position coordinates of the centerline as the actual overlapping center.

[0039] As a further aspect of the present invention, the fuzzy rule base invoked by the fuzzy inference submodule specifically includes:

[0040] Based on the fuzzy linguistic variables of the tension error and the rate of change of the tension error, a set of tension control rules for regulating the increase or decrease of the speed is generated;

[0041] The tension control rule set includes:

[0042] If the tension error is positive and the rate of change of the tension error is zero, then the increase or decrease in speed is negative.

[0043] Based on the fuzzy linguistic variables of the overlapping position deviation, a set of correction control rules for regulating the translational displacement is generated;

[0044] The set of corrective control rules includes:

[0045] If the overlap position deviation is positive, then the translational displacement is negative.

[0046] As a further aspect of the present invention, the process of generating the performance degradation judgment threshold in the performance judgment submodule is specifically as follows:

[0047] During the initial calibration phase of the intelligent bonding and tension control system on the multi-ribbed belt production line of the washing machine, at least one batch of the long-term comprehensive performance indicators were obtained. Historical data constitutes the baseline performance dataset;

[0048] Calculate the statistical mean and standard deviation of all long-term composite performance metrics in the baseline performance dataset;

[0049] The statistical average plus a preset multiple of the standard deviation is used as the performance degradation judgment threshold.

[0050] A method for intelligent lamination and tension control in a multi-ribbed belt production line for washing machines, the method being implemented based on the aforementioned intelligent lamination and tension control system for a multi-ribbed belt production line for washing machines, includes the following steps:

[0051] S1: Collect voltage signals from tension sensors and image data from industrial cameras, calculate the tension error, the rate of change of tension error, and the bonding position deviation, and transmit the tension error, the rate of change of tension error, and the bonding position deviation to the real-time collaborative control step, and transmit the tension error and the bonding position deviation to the control performance evaluation step;

[0052] S2: Based on the tension error, the rate of change of the tension error and the overlap position deviation, a fuzzy logic inference engine is invoked to generate the coordinated adjustment signal consisting of the speed increase / decrease and the translational displacement, and the transmission roller servo motor is adjusted according to the speed increase / decrease in the coordinated adjustment signal, and the correction mechanism actuator is adjusted according to the translational displacement.

[0053] S3: Based on the time series of the tension error and the bonding position deviation, calculate the long-term comprehensive performance index, compare the long-term comprehensive performance index with the performance degradation judgment threshold to generate the performance degradation trigger instruction, and pass it to the fuzzy rule self-tuning step.

[0054] S4: According to the performance degradation trigger instruction, call the gradient descent algorithm to adjust the center value parameter of the membership function, and update the fuzzy logic inference engine with the center value parameter.

[0055] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0056] In this invention, tension sensor signals and industrial camera image data are collected in real time to calculate tension error and lamination position deviation, providing a basis for dynamic adjustment. This eliminates the reliance on preset fixed parameters. By calling a fuzzy logic inference engine, tension error, error change rate, and position deviation are fused and processed to generate a coordinated adjustment signal. This enables synchronous and precise control of the transmission roller speed and the position of the correction mechanism, effectively replacing the slow-responding mechanical tension device and solving the problems of tension fluctuation and lamination deviation. Furthermore, a long-term comprehensive performance index is introduced to evaluate the control effect. When performance deteriorates, the gradient descent algorithm is immediately triggered to self-tune the fuzzy rule parameters, continuously optimizing the control strategy and ensuring that the system maintains high stability and high accuracy. Attached Figure Description

[0057] Figure 1 This is a flowchart of the overall control architecture of the system of the present invention;

[0058] Figure 2 This is a flowchart of the multidimensional state information extraction process of the present invention;

[0059] Figure 3 This is a flowchart of the real-time collaborative control process of the present invention;

[0060] Figure 4 This is a flowchart of the control performance evaluation and decision-making process of the present invention;

[0061] Figure 5 This is a flowchart of the fuzzy rule self-tuning process of the present invention. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this invention clearer, the software-based technical solution is described in detail below with reference to system architecture diagrams and embodiments. It should be understood that the specific embodiments described herein are only for explaining the technical solutions of this invention and do not constitute a limitation on the scope of protection.

[0063] In the description of this invention, the system architecture relationships or data processing flows indicated by terms such as "layer," "module," "interface," "data flow," "client," and "server" are all defined based on the architecture diagram or flowchart corresponding to the embodiments. This way of describing is only used to clearly illustrate the logical relationships between the elements in the technical solution, and not to limit the physical deployment form. The term "multiple" includes two or more technical units, including but not limited to multiple data nodes, processing threads, service instances, or functional components and other scalable elements. The specific number is determined according to the actual business scenario and needs to be specifically specified.

[0064] Please see Figure 1 and Figure 2 This invention provides a technical solution: an intelligent lamination and tension control system for a multi-ribbed belt production line for washing machines, comprising:

[0065] The multi-dimensional state information extraction module is used to collect voltage signals from tension sensors and image data from industrial cameras, calculate tension error, tension error change rate and bonding position deviation, and transmit tension error, tension error change rate and bonding position deviation to the real-time collaborative control module, and transmit tension error and bonding position deviation to the control performance evaluation module.

[0066] Tension error specifically refers to the difference between the target tension set value and the actual tension measurement value. Tension error change rate includes the difference between the current error value and the error value of the previous sampling period. Overlap position deviation specifically refers to the difference between the preset overlap center and the actual overlap center.

[0067] The specific functions of the multidimensional state information extraction module are as follows:

[0068] A signal processing submodule is set up to acquire the voltage signal of the tension sensor, filter and calibrate the voltage signal of the tension sensor, and obtain the actual tension measurement value.

[0069] The tension error is calculated based on the target tension set value and the actual tension measurement value.

[0070] Obtain the current error value and the error value of the previous sampling period, and calculate the rate of change of tension error;

[0071] An image processing submodule is set up to acquire image data from an industrial camera, identify the edges of the multi-wedge strip through edge detection and morphological learning operations, and determine the actual bonding center.

[0072] The bonding position deviation is calculated based on the difference between the preset bonding center and the actual bonding center;

[0073] A data integration submodule is set up to encapsulate tension error, tension error change rate and bonding position deviation, and transmit them to the real-time collaborative control module and the control performance evaluation module.

[0074] The image solving submodule determines the actual overlay center as follows:

[0075] The industrial camera image data is subjected to grayscale conversion and Gaussian filtering to obtain a preprocessed image;

[0076] The Canny operator is used to extract the multi-wedge edge contours in the preprocessed image;

[0077] The bilateral straight line equations of the multi-wedge edge contour are located using Hough line detection or contour fitting algorithms.

[0078] Calculate the centerline of the bilateral straight line equation and determine the position coordinates of the centerline as the actual overlapping center.

[0079] After the signal processing submodule starts, it first acquires continuous voltage signals from the tension sensor associated with the tension roller of the transmission system. The sampling frequency is set to 100Hz, meaning voltage data is collected every 0.01 seconds. At a specific sampling instant, the five consecutive raw voltage values ​​acquired are [2.98V, 3.05V, 2.99V, 3.03V, 3.01V]. A moving average method is used to filter this set of data. The calculation process is as follows: The filtered voltage signal value is then obtained. Subsequently, this voltage value is converted according to a pre-defined calibration formula. This calibration formula is obtained by measuring the sensor voltage output under different tensions using a standard force gauge during the production line commissioning phase and performing linear regression fitting.

[0080] Table 1. Tension Sensor Calibration Data:

[0081]

[0082] As shown in Table 1, the conversion formula between tension (T) and voltage (U) is obtained through linear fitting of the calibration data. Substituting the filtered voltage value of 3.012V into the formula, the actual tension measurement value is calculated: .

[0083] Next, based on the target tension value (e.g., 155N) set for a specific model of multi-wedge belt according to the production process, the tension error is calculated. Its calculation is as follows: =Target tension setpoint - Actual tension measurement value= .

[0084] To calculate the rate of change of tension error ( The system needs to obtain the tension error value from the previous sampling period (time t-1). Assuming the tension error calculated at time t-1 is 5.625 N, the rate of change of the tension error in the current sampling period (time t) is: =Current error value - Error value from the previous sampling period= .

[0085] Meanwhile, the image processing module acquires image data from an industrial camera (1920×1080 pixels resolution, 12mm lens focal length) mounted directly above the lamination station. First, the acquired color image is converted to an 8-bit grayscale image. Then, a 5×5 Gaussian kernel is used to convolve the grayscale image, resulting in a smoothed image. After processing, the Canny operator is called for edge detection. The Canny operator's internal execution includes: calculating the image gradient, performing non-maximum suppression along the gradient direction (which thins the edges), and applying dual thresholds (e.g., a low threshold of 50 and a high threshold of 150) to connect the edges. After completion, a binarized edge contour image with clear wedges is obtained.

[0086] Next, the Hough line detection algorithm is used to process the binarized image. The algorithm performs a voting process in the parameter space to identify the lines representing the two main edges of the multi-wedge band. The detection result is two line equations; for example, in the image coordinate system, the equation of the left edge line is... The equation of the right edge line is Therefore, the actual overlapping center position of the multi-wedge band is calculated as follows: Actual overlapping center = (left edge position + right edge position) / 2 = Pixel.

[0087] According to process requirements, the preset overlay center of the multi-wedge tape should be located at the horizontal center of the image, i.e., at 960 pixels. Based on this, the overlay position deviation is calculated ( ): =Actual Overlap Center - Preset Overlap Center= 964 pixels - 960 pixels = 4 pixels. A pixel-to-millimeter conversion is performed. Through preliminary calibration (placing a calibration board within the camera's field of view), it is determined that 1 pixel corresponds to 0.05mm. Therefore, the physical value of the overlap position deviation is: Pixels mm / pixel = mm.

[0088] Finally, the data integration submodule encapsulates the three calculated core state parameters into a single data packet. This data packet contains: {Tension Error: 5.706N, Tension Error Change Rate: 0.081N, Overlapping Position Deviation: 0.2mm}. This data packet is split into two paths: a complete data packet containing all three parameters is transmitted to the real-time collaborative control module; while the data packet containing only the tension error and overlapping position deviation {Tension Error: 5.706N, Overlapping Position Deviation: 0.2mm} is transmitted to the control performance evaluation module.

[0089] Please see Figure 1 and Figure 3 The real-time collaborative control module is used to generate a collaborative adjustment signal consisting of speed increase / decrease and translational displacement based on tension error, tension error change rate and bonding position deviation by calling a fuzzy logic inference engine. The module controls the transmission roller servo motor connected to it according to the speed increase / decrease in the collaborative adjustment signal, and controls the correction mechanism actuator connected to it according to the translational displacement.

[0090] The coordinated adjustment signals include drive control commands and correction control commands;

[0091] The specific functions of the real-time collaborative control module are as follows:

[0092] The input fuzzification submodule is set up to obtain tension error, tension error change rate and overlay position deviation, and convert them into fuzzy linguistic variables based on the preset membership function;

[0093] Set up a fuzzy inference submodule to call the fuzzy rule library, perform logical inference on fuzzy linguistic variables, and obtain fuzzy control output;

[0094] Set up an output defuzzification submodule to perform defuzzification calculations on the fuzzy control output using the centroid method, and obtain the speed increase / decrease and translational displacement.

[0095] A control execution submodule is set up to combine the speed increase / decrease and translational displacement into a coordinated adjustment signal, and drive the transmission roller servo motor and the correction mechanism actuator.

[0096] The fuzzy rule base called by the fuzzy inference submodule specifically includes:

[0097] Based on fuzzy linguistic variables of tension error and the rate of change of tension error, a set of tension control rules for regulating the increase or decrease of speed is generated;

[0098] The tension control rule set includes:

[0099] If the tension error is positive and the rate of change of the tension error is zero, then the increase or decrease in speed is negative.

[0100] Based on the fuzzy linguistic variables of the overlap position deviation, a set of correction control rules for regulating the translational displacement is generated;

[0101] The corrective control rule set includes:

[0102] If the deviation in the overlapping position is positive, then the translational displacement is negative.

[0103] After receiving the data packet {tension error: 5.706N, tension error change rate: 0.081N, bonding position deviation: 0.2mm} transmitted by the multi-dimensional state information extraction module, the real-time collaborative control module starts its internal functions.

[0104] First, the input fuzzification submodule processes the three received precise values. This submodule internally defines a membership function for each input variable, mapping the precise values ​​to the membership degrees of the fuzzy linguistic variables. Using tension error (… Taking 'PS' as an example, its fuzzy set is defined as {Negative Large (NB), Negative Small (NS), Zero (ZE), Positive Small (PS), Positive Large (PB)}. Each fuzzy set is defined by a triangular or trapezoidal membership function. For example, the membership function center value of the 'Positive Small (PS)' fuzzy set is 5N, the width is 4N, and the domain is [3N, 7N]; the center value of the 'Positive Large (PB)' fuzzy set is 10N, the width is 5N, and the domain is [7.5N, 12.5N]. The current tension error is 5.706N, which falls within the domains of both 'PS' and 'PB' fuzzy sets. According to the membership function calculation, its membership degree to 'PS' is... The membership degree of 'PB' is Outside this range, the value is 0. However, if the domain is adjusted, for example, if 'PS' is and 'PB' is , then the membership degree of 5.706N for 'PS' is . The membership degree of 'PB' is We adopt the latter setting. Similarly, regarding the rate of change of tension error... (Definition of fuzzy sets and) Similarly, but with a narrower range, for example, PS is [0.05, 0.15], and its membership degree to 'PS' is 0.69; regarding the overlap position deviation (For example, a fuzzy set PS is [0.1, 0.3]), its membership degree to 'PS' is 0.5.

[0105] Next, the fuzzy reasoning submodule performs logical reasoning based on a pre-set fuzzy rule base. This rule base contains two rule sets. The tension control rule set is used to generate speed increases and decreases, and its rule form is "If..." A and If the value is B, then the increase or decrease in speed is C. For example, a rule is: "If..." For PS and If the value is PS, then the speed increase or decrease is NS (negative small). The trigger strength of this rule is... Another rule "If..." For PB and If it is PS, then the speed increase or decrease is NB (negative large), and its trigger strength is This rule is triggered. The system summarizes all triggered rules and their strengths. The correction control rule set is used to generate translational displacement, and its rule form is "If..." If the value is D, then the translational displacement is E. For example, the rule: "If..." If the value is PS, then the translational displacement is NS (negative small). The trigger strength of this rule is... .

[0106] Then, the output defuzzification submodule performs defuzzification calculations on the results of the fuzzy inference to obtain precise control output values. This module uses the center of gravity method. For speed increases and decreases, it is assumed that the inference triggered two rules, resulting in strengths of... The NS output and strength are The ZE (zero) output. Let the center value of the output fuzzy set corresponding to NS be -0.05 m / s, and the center value corresponding to ZE be 0 m / s. Then the precise velocity increase / decrease is calculated as follows: For the translational displacement, assuming the inference yields an NS output of strength 0.5 with a center value of -0.1mm, and no other rules are triggered, the precise translational displacement is calculated as follows: .

[0107] Finally, the control execution submodule combines the precise values ​​obtained after defuzzification into a coordinated adjustment signal. This signal includes drive control commands and correction control commands. Specifically: {speed increase / decrease: -0.0123m / s, translational displacement: -0.1mm}. This module sends the speed increase / decrease of -0.0123m / s to the driver of the drive roller servo motor, and the driver correspondingly reduces the motor speed, thus reducing the running speed of the multi-ribbed belt. At the same time, it sends the translational displacement of -0.1mm to the actuator of the correction mechanism (such as a stepper motor or piezoelectric ceramic actuator), driving it to control the guide roller to make a small translation.

[0108] Please see Figure 1 and Figure 4 The control performance evaluation module is used to calculate the long-term comprehensive performance index based on the time series of tension error and bonding position deviation, compare the long-term comprehensive performance index with the performance degradation judgment threshold to generate a performance degradation trigger instruction, and pass it to the fuzzy rule self-tuning module.

[0109] Performance degradation trigger instructions include the tuning start signal and the target module address;

[0110] The specific functions of the control performance evaluation module are as follows:

[0111] Set up a data caching submodule to collect and store the time series of tension error and bonding position deviation within a preset evaluation period;

[0112] Set up an indicator calculation submodule to perform calculations based on time series data using formulas. Calculate long-term comprehensive performance indicators;

[0113] in, Represents long-term comprehensive performance indicators. This represents the total number of sampling points within the evaluation period, where i represents the index of the sampling point. This represents the tension error at the i-th sampling point. This represents the overlap position deviation of the i-th sampling point. Represents the tension error weighting coefficient. Represents the overlapping deviation weighting coefficient;

[0114] A performance determination submodule is set up to compare long-term comprehensive performance indicators with preset performance degradation determination thresholds. When the performance degradation threshold is exceeded, a performance degradation trigger instruction is generated.

[0115] In the performance assessment submodule, the process for generating the performance degradation assessment threshold is as follows:

[0116] During the initial calibration phase of the intelligent lamination and tension control system on the multi-ribbed belt production line for washing machines, long-term comprehensive performance indicators for at least one batch were obtained. Historical data constitutes the baseline performance dataset;

[0117] Calculate the statistical mean and standard deviation of all long-term composite performance metrics in the baseline performance dataset;

[0118] The statistical mean plus a preset multiple of the standard deviation is used as the threshold for judging performance degradation.

[0119] After receiving the time series data of {tension error, bonding position deviation} continuously transmitted by the multi-dimensional state information extraction module, the control performance evaluation module starts its evaluation process.

[0120] First, the data caching submodule collects and stores data according to a preset evaluation period. The evaluation period is set to 10 seconds, and the sampling frequency is 10Hz; therefore, N=100 sampling points will be collected and stored within one evaluation period. A circular buffer is allocated within the module to store these 100 consecutive pairs of (…). , ) data. For example, the first few items of a stored data sequence might be: {( , =(5.71N, 0.20mm)},{( , =(5.65N, 0.18mm)},{( , =(5.68N, 0.21mm)},…,{( , = (5.75N, 0.19mm)}.

[0121] Table 2. Error Sample Data Table During the Evaluation Period:

[0122] As shown in Table 2, after data collection is completed, the indicator calculation submodule, based on all 100 sampling points in the cache, calculates the index using the formula... Calculate long-term comprehensive performance indicators .

[0123] in, This represents the long-term comprehensive performance index, where N represents the total number of sampling points within the evaluation period, and i is the index of the sampling point from 1 to N. It is the tension error value at the i-th sampling point. It is the overlap position deviation value of the i-th sampling point. The weighting coefficient for tension error acts on the sum of the squares of the tension errors. The weighting coefficients for the overlay position deviations act on the sum of the squares of the overlay position deviations.

[0124] Weighting coefficient and The determination is based on experimental verification. During the system debugging phase, multiple weight combinations were set (e.g., (0.5, 0.5), (0.6, 0.4), (0.7, 0.3)), and the following conditions were met: Under each weighted configuration, a batch (e.g., 500 strips) of multi-wedge belts was produced, and the finished products underwent quality inspection, including tension uniformity and dimensional accuracy after lamination. The pass rate of each batch was calculated. The pass rates of the products under each weighted configuration were compared, and the group with the highest pass rate was selected. Experimental data showed that when... , At that time, the overall product qualification rate reached 99.7%, the highest among all groups, therefore it was determined that... , .

[0125] Assuming that, after calculation, the sum of squares of tension errors within the evaluation period is... The sum of squares of the overlap position deviations is Substitute these values ​​into the formula: .

[0126] Next, the performance determination submodule will calculate the long-term comprehensive performance index. The performance is compared with a preset performance degradation threshold. This threshold is generated as follows: During the initial calibration phase of the system, 10 batches of products are produced consecutively, each batch corresponding to an evaluation period, and 10 long-term comprehensive performance indicators are calculated. Historical data constitutes the baseline performance dataset. For example, the 10 values ​​are: [1850.5, 1885.2, 1863.7, 1890.1, 1845.9, 1877.6, 1855.4, 1892.3, 1868.0, 1881.3]. First, calculate the statistical mean of this dataset: Secondly, calculate the sample standard deviation. The calculation process is as follows: First, calculate the square of the difference between each data point and the mean. Add all the squared differences together and divide by the total number of data points to obtain the variance. Finally, take the square root of the variance to obtain the sample standard deviation. The performance degradation threshold is set as a preset multiple (e.g., 2 times) of the average value plus the standard deviation, i.e., threshold = Because of the currently calculated If the value exceeds the performance degradation threshold of 1902.78, the performance determination submodule determines that the system performance has deteriorated.

[0127] Accordingly, the module generates a performance degradation trigger instruction, which includes a tuning start signal (a boolean flag with a value of TRUE) and the target module address (a memory address or identifier pointing to the fuzzy rule self-tuning module). This instruction is then passed to the fuzzy rule self-tuning module.

[0128] Please see Figure 1 and Figure 5 The fuzzy rule self-tuning module is used to call the gradient descent algorithm to adjust the center value parameter of the membership function according to the performance degradation trigger instruction, and update the fuzzy logic inference engine with the center value parameter;

[0129] The central value parameters are specifically the error membership function center, the error rate of change membership function center, and the deviation membership function center;

[0130] The specific functions of the fuzzy rule self-tuning module are as follows:

[0131] Set up an instruction receiving submodule to receive performance degradation trigger instructions and lock the target module address;

[0132] A gradient calculation submodule is set up to calculate long-term comprehensive performance indicators. Given the objective function, the gradient descent algorithm is used to calculate the sensitivity gradient of the objective function to the central value parameter;

[0133] The parameter update submodule is used to adjust parameters based on the sensitivity gradient. Calculate the updated center value parameters;

[0134] in, The parameter represents the center value parameter of the j-th membership function of the i-th input variable at time k+1. The parameter represents the center value at time k, where k represents the number of iterations. Represents the learning rate. Represents the objective function For the center value parameter The partial derivatives;

[0135] The Engine Update submodule is configured to write the updated center value parameters into the parameter storage area of ​​the fuzzy logic inference engine.

[0136] In standby mode, the fuzzy rule self-tuning module continuously monitors for performance degradation trigger commands. Once it receives a command from the control performance evaluation module, its internal processing flow is activated.

[0137] First, the instruction receiving submodule parses the received instruction. It reads that the tuning start signal is TRUE and locks the target module address contained in the instruction, which points to the fuzzy logic inference engine parameter storage area in the real-time collaborative control module.

[0138] Next, the gradient calculation submodule starts its computation. It uses the long-term integrated performance index calculated by the control performance evaluation module. The objective function is to compute the objective function. The sensitivity gradient, i.e., the partial derivative, of the central value parameter of the fuzzy membership function. Here, Represents the i-th input variable (e.g., i=1 corresponds to tension error) The partial derivative is calculated using the chain rule in the gradient descent algorithm. It is assumed that the parameter with the greatest impact on performance degradation is the center value of the membership function of the 'positive minimum (PS)' membership function of the tension error. Its current value is 5N, and the calculated sensitivity gradient is: .

[0139] Subsequently, the parameter update submodule performs iterative parameter updates based on the calculated sensitivity gradient. The update formula is as follows: .

[0140] in, It is the center value parameter for the next iteration. The center value parameter of the current iteration, k represents the iteration number. It's the learning rate. It is the objective function For the center value parameter The partial derivatives of .

[0141] Learning rate The settings were determined through experimental verification. During implementation, a series of learning rate values ​​(e.g., 0.01, 0.001, 0.0001) were used for testing. The effects of different learning rates after several iterations were observed. The rate of decrease of the value and whether oscillations occur. Experiments show that when At that time, the parameter adjustment process was stable and the convergence speed was acceptable; This can cause systemic oscillations. The convergence is too slow. Therefore, choose... .

[0142] Substitute the current value and the calculation result into the formula, and... Update: N.

[0143] The calculation results determined that the center of the 'positive small' fuzzy set of tension error was adjusted from 5N to 4.95498N.

[0144] Finally, the engine update submodule performs the final write operation of the parameters. It directly writes the calculated new center value parameter 4.95498N into the parameter storage area of ​​the fuzzy logic inference engine via the target module address, overwriting the original center value 5N. Subsequently, the real-time collaborative control module will use this updated membership function center value during fuzzification processing. The entire self-tuning process completes one iteration. This module will continue this process until… The value falls back below the performance degradation threshold, or reaches the preset maximum number of iterations.

[0145] A method for intelligent lamination and tension control in a multi-ribbed belt production line for washing machines, wherein the method is based on the aforementioned intelligent lamination and tension control system for a multi-ribbed belt production line for washing machines, and includes the following steps:

[0146] S1: Collect voltage signals from tension sensors and image data from industrial cameras, calculate tension error, tension error change rate, and lamination position deviation, and transmit tension error, tension error change rate, and lamination position deviation to the real-time collaborative control step, and transmit tension error and lamination position deviation to the control performance evaluation step.

[0147] S2: Based on tension error, tension error change rate and overlapping position deviation, the fuzzy logic inference engine is called to generate a coordinated adjustment signal consisting of speed increase / decrease and translational displacement. The transmission roller servo motor is adjusted according to the speed increase / decrease in the coordinated adjustment signal, and the correction mechanism actuator is adjusted according to the translational displacement.

[0148] S3: Based on the time series of tension error and bonding position deviation, calculate the long-term comprehensive performance index, compare the long-term comprehensive performance index with the performance degradation judgment threshold to generate a performance degradation trigger instruction, and pass it to the fuzzy rule self-tuning step.

[0149] S4: Based on the performance degradation trigger instruction, call the gradient descent algorithm to adjust the center value parameter of the membership function, and update the fuzzy logic inference engine with the center value parameter.

[0150] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the scope of protection defined by the claims of the present invention.

Claims

1. A multi-V belt production line intelligent covering and tension control system for a washing machine, characterized in that, The system comprises: A multi-dimensional state information extraction module is configured to collect tension sensor voltage signals and industrial camera image data, calculate tension error, tension error change rate and lamination position deviation, and transmit the tension error, the tension error change rate and the lamination position deviation to a real-time collaborative control module and transmit the tension error and the lamination position deviation to a control performance evaluation module; The real-time collaborative control module is configured to call a fuzzy logic inference machine to generate a collaborative adjustment signal composed of speed increase / decrease and translation displacement based on the tension error, the tension error change rate and the lamination position deviation, and regulate a transmission roller servo motor connected thereto according to the speed increase / decrease in the collaborative adjustment signal and regulate a deviation correction mechanism actuator connected thereto according to the translation displacement; The control performance evaluation module is configured to calculate a long-term comprehensive performance index based on time series of the tension error and the lamination position deviation, compare the long-term comprehensive performance index with a performance degradation judgment threshold to generate a performance degradation trigger instruction, and transmit the performance degradation trigger instruction to a fuzzy rule self-tuning module; The fuzzy rule self-tuning module is configured to adjust a center value parameter of a membership function by calling a gradient descent algorithm according to the performance degradation trigger instruction, and update the fuzzy logic inference machine with the center value parameter; The specific function of the fuzzy rule self-tuning module is implemented as: A setting instruction receiving submodule is configured to receive the performance degradation trigger instruction and lock a target module address; A setting gradient calculation submodule is configured to take the long-term comprehensive performance index as a target function and calculate a sensitivity gradient of the target function to the center value parameter by using the gradient descent algorithm; The setting parameter updating sub-module is configured to update the center value parameter according to the sensitivity gradient by calculating an updated center value parameter. wherein, a center value parameter of the jth membership function representing the ith input variable at the kth iteration, a center value parameter of the kth iteration, k representing an iteration number, representing a learning rate, representing a partial derivative of the objective function J with respect to the center value parameter ; A setting engine updating submodule is configured to write the updated center value parameter into a parameter storage area of the fuzzy logic inference machine.

2. The intelligent covering and tension regulating system for a washing machine V-ribbed belt production line according to claim 1, characterized in that, The tension error specifically refers to the difference between a target tension set value and an actual tension measured value, the tension error change rate includes the difference between a current error value and a last sampling period error value, the lamination position deviation specifically refers to the difference between a preset lamination center and an actual lamination center, the collaborative adjustment signal includes a transmission control instruction and a deviation correction control instruction, the performance degradation trigger instruction includes a tuning start signal and a target module address, and the center value parameter specifically refers to an error membership function center, an error change rate membership function center and a deviation membership function center.

3. The intelligent covering and tension regulating system for a washing machine V-ribbed belt production line according to claim 2, characterized in that, The specific function of the multi-dimensional state information extraction module is implemented as: A setting signal processing submodule is configured to obtain the tension sensor voltage signals, filter and calibrate the tension sensor voltage signals to obtain the actual tension measured value; The tension error is calculated based on the target tension set value and the actual tension measured value; The tension error change rate is calculated by obtaining the current error value and the last sampling period error value; An image solver module is configured to obtain the industrial camera image data, identify a multi-V-ribbed belt edge through edge detection and morphological learning operation, and determine the actual lamination center; The lamination position deviation is calculated based on the difference between the preset lamination center and the actual lamination center; A data integration submodule is arranged to encapsulate the tension error, the tension error rate of change and the lamination position deviation, and deliver them to the real-time collaborative control module and the control performance evaluation module.

4. The intelligent covering and tension regulating system for a washing machine V-ribbed belt production line according to claim 1, characterized in that, The specific function of the real-time collaborative control module is implemented as: An input fuzzification submodule is arranged to acquire the tension error, the tension error rate of change and the lamination position deviation, and convert them into fuzzy language variables based on a preset membership function; A fuzzy reasoning submodule is arranged to call a fuzzy rule base, and perform logical reasoning on the fuzzy language variables to obtain fuzzy control outputs; An output defuzzification submodule is arranged to perform defuzzification calculation on the fuzzy control outputs by using the center of gravity method to obtain the speed increment / decrement and the translation displacement; A control execution submodule is arranged to combine the speed increment / decrement and the translation displacement into the collaborative adjustment signal, and drive the transmission roller servo motor and the deviation correction mechanism actuator.

5. The intelligent overlaying and tension regulating system for a washing machine V-ribbed belt production line according to claim 1, characterized in that, The specific function of the control performance evaluation module is implemented as: A data caching submodule is arranged to collect and store the time series of the tension error and the lamination position deviation within a preset evaluation period; The index calculation submodule is configured to calculate the long-term synaptic performance index based on the time series through a formula ​ Wherein, J represents the long-term performance index, N represents the total number of sampling points in the evaluation period, i represents the index of the sampling point, The tension error of the i th sampling point, The lamination position deviation of the i th sampling point, The tension error weight coefficient, The lamination deviation weight coefficient; A performance determination submodule is arranged to compare the long-term comprehensive performance index with a preset performance degradation determination threshold, and generate the performance degradation trigger instruction when the long-term comprehensive performance index is greater than the performance degradation determination threshold.

6. The intelligent covering and tension regulating system for a washing machine V-ribbed belt production line according to claim 3, characterized in that, The process of determining the actual lamination center by the image solver module is specifically as follows: Perform grayscale and Gaussian filtering on the industrial camera image data to obtain a preprocessed image; Extract the multi-V-ribbed belt edge profile in the preprocessed image by using the Canny operator; Locate the double-side straight line equation of the multi-V-ribbed belt edge profile by using the Hough straight line detection or contour fitting algorithm; Calculate the center line of the double-side straight line equation, and determine the position coordinates of the center line as the actual lamination center.

7. The intelligent overlaying and tension regulating system for a washing machine V-ribbed belt production line according to claim 4, characterized in that, The fuzzy rule base called by the fuzzy reasoning submodule specifically includes: Based on the fuzzy language variables of the tension error and the tension error rate of change, generate a tension control rule set for regulating the speed increment / decrement; The tension control rule set includes: If the tension error is positive and large and the tension error rate of change is zero, the speed increment / decrement is negative and large; Based on the fuzzy language variables of the lamination position deviation, generate a deviation correction control rule set for regulating the translation displacement; The deviation correction control rule set includes: If the lamination position deviation is positive and large, the translation displacement is negative and large.

8. The intelligent covering and tension regulating system for a washing machine V-ribbed belt production line according to claim 5, characterized in that, In the performance determination submodule, the generation process of the performance degradation determination threshold is specifically as follows: In the calibration stage of the initial operation of the intelligent lamination and tension regulation system for the washing machine multi-V-ribbed belt production line, obtain the historical data of the long-term comprehensive performance index J of at least one batch to form a baseline performance data set; Calculate the statistical mean and standard deviation of all long-term comprehensive performance indexes in the baseline performance data set; Add a preset multiple of the standard deviation to the statistical mean as the performance degradation determination threshold.

9. The intelligent covering and tension control method for a multi-vee belt production line of a washing machine, characterized in that, The method is used for realizing the intelligent covering and tension regulation system of the washing machine multi-rib belt production line according to any one of claims 1-8, comprising the following steps: S1: collecting tension sensor voltage signals and industrial camera image data, calculating the tension error, the tension error change rate and the covering position deviation, and transmitting the tension error, the tension error change rate and the covering position deviation to the real-time collaborative control step, and transmitting the tension error and the covering position deviation to the control performance evaluation step; S2: based on the tension error, the tension error change rate and the covering position deviation, calling a fuzzy logic inference machine to generate the collaborative adjustment signal composed of a speed increment / decrement and a translation displacement, and regulating the driving roller servo motor according to the speed increment / decrement in the collaborative adjustment signal and regulating the deviation correction mechanism actuator according to the translation displacement; S3: based on the time series of the tension error and the covering position deviation, calculating a long-term comprehensive performance index, comparing the long-term comprehensive performance index with the performance degradation judgment threshold to generate the performance degradation trigger instruction, and transmitting to the fuzzy rule self-tuning step; S4: according to the performance degradation trigger instruction, calling a gradient descent algorithm to adjust the center value parameter of the membership function, and updating the fuzzy logic inference machine with the center value parameter.

Images