A method and system for cycle time coordination control of a high-speed plastic cup production line

Abstract

This invention relates to the field of intelligent manufacturing control technology for plastic products, and discloses a cycle time collaborative control method and system for a high-speed plastic cup production line. The method includes: acquiring a complete area image of the plastic cup and analyzing it to obtain defect-sensitive feature vectors; using an EWMA control chart and an OS-ELM model to predict defect evolution trends; adaptively adjusting injection molding process parameters; mapping process intervention information to a discrete event simulation model; performing cycle time collaborative adjustment; and updating model parameters online based on execution feedback. Compared to existing technologies that rely on manual sampling, especially under high-speed continuous production conditions of plastic cups, which cannot achieve adaptive intervention of injection molding process parameters, this application improves the quality stability of the high-speed plastic cup production line by constructing a continuous control chain of image feature analysis, process parameter adjustment, and cycle time collaborative control, thus achieving preventative adjustment of the injection molding process.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing control technology for plastic products, and in particular to a cycle time collaborative control method and system for a high-speed production line for plastic cups. Background Technology

[0002] Currently, in high-speed injection molding production lines for plastic cups, product quality control and production line cycle time coordination are two core challenges. Traditional quality control methods largely rely on offline or online inspection of the molded plastic cups, and then adjusting injection molding process parameters (such as temperature, pressure, and time) after the fact based on the inspection results. This method suffers from response lag and cannot predict or preventively intervene in the evolution of defects. Furthermore, adjustments to process parameters often alter the injection cycle, thereby affecting the cycle time matching of subsequent stages such as cooling, conveying, rejection, and stacking.

[0003] Current technologies for coordinating production line cycle times often employ fixed timing control or simple feedback adjustments, lacking the ability to globally optimize for dynamically changing cycle time disturbances caused by quality interventions. Especially under high-speed production conditions, even minor cycle time variations in a single section can be rapidly amplified, leading to backlogs in the production line buffer, equipment idling, or incorrect product rejection, severely impacting production efficiency and costs. Therefore, there is an urgent need for a technology capable of real-time, collaboratively adjusting the operating cycle times of each section of the production line while simultaneously implementing predictive quality interventions. This would enable a dynamic balance between quality and efficiency under high-speed, continuous production conditions, thereby improving the overall intelligence and economic benefits of high-speed plastic cup production lines. Summary of the Invention

[0004] To address the aforementioned technical shortcomings, the purpose of this invention is to propose a cycle time collaborative control method for a high-speed plastic cup production line. This method aims to solve the technical problem that existing technologies rely on manual sampling inspection, especially under high-speed continuous production conditions of plastic cups, and cannot achieve adaptive intervention of injection molding process parameters.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a cycle time coordination control method for a high-speed production line of plastic cups.

[0006] The cycle time coordination control method for a high-speed plastic cup production line includes: Step S10: Obtain a complete image of the plastic cup, and perform defect-related feature analysis on the complete image of the plastic cup to obtain a defect-sensitive feature vector; Step S20: Based on the defect-sensitive feature vector, the evolution trend of molding defects in plastic cups is predicted online using EWMA control charts and OS-ELM models to obtain defect trend assessment results; Step S30: Based on the defect trend assessment results, PCA principal component analysis and Mamdani fuzzy PID control are used to adaptively adjust the injection molding process parameters to obtain process intervention information; Step S40: Map the process intervention information to a preset discrete event simulation model, and use the rolling time domain optimization method to perform cycle time coordination adjustment on the rejection section, cooling section, conveying section and stacking section to obtain cycle time coordination control instructions; Step S50: Based on the actual execution feedback of the rhythmic coordinated control command, the model parameters are updated online using a closed-loop parameter update method based on Markov decision process and Q-learning to obtain the model parameter self-update result.

[0007] Preferably, step S10, which involves acquiring an image of the entire region of the plastic cup and performing defect-related feature analysis on the image to obtain a defect-sensitive feature vector, specifically includes: Step S101: Using a high-speed linear array camera and pulse light source that are synchronously triggered with the injection molding machine's mold opening and closing signals and the conveyor belt encoder pulses, images of the cup mouth area and the cup body sidewall are acquired when the plastic cup passes through the detection station. The positions are aligned and the areas are stitched together according to the displacement corresponding to the conveyor belt encoder pulses to obtain the complete area image of the plastic cup. Step S102: The Sobel operator is used to extract the gradient features of the cup rim edge, the Gabor filter is used to extract the texture direction features of the cup body, and the local binary mode is used to extract the local texture abrupt change features of the cup wall, so as to obtain a multi-region defect candidate feature set. Step S103: Perform weighted fusion on the multi-region defect candidate feature set to obtain the defect-sensitive feature vector, wherein the defect-sensitive feature vector is expressed according to the following formula:

[0008] in, For defect-sensitive feature vectors, The gradient features of the cup rim obtained by the Sobel operator are shown below. The cup body texture direction features are obtained by Gabor filter. The abrupt changes in local texture of the cup wall are obtained from the local binary pattern. , and These are the defect sensitivity weight coefficients corresponding to the gradient features of the cup rim edge, the texture direction features of the cup body, and the local texture abrupt change features of the cup wall.

[0009] Preferably, step S20, which involves using an EWMA control chart and an OS-ELM model to predict the evolution trend of molding defects in plastic cups online based on the defect-sensitive feature vector, and obtaining the defect trend assessment result, specifically includes: Step S201: Select key defect feature components related to rim flash, shrinkage marks, or tearing of the cup wall from the defect-sensitive feature vector, and calculate the exponentially weighted moving average using an EWMA control chart. The exponentially weighted moving average is expressed by the following formula:

[0010] in, It is the index-weighted moving average of the current production cycle. These are the key defect feature components of the current production cycle. It is the index-weighted moving average of the previous production cycle. For smoothing coefficients; Step S202: Input the exponentially weighted moving average sequence of multiple consecutive production cycles into the OS-ELM model to predict the trend of defect feature changes in multiple future production cycles, and obtain the future defect prediction sequence. Step S203: Based on the exponentially weighted moving average, the future defect prediction sequence, and the preset qualified product characteristic control limits, generate the defect trend assessment result, wherein the defect trend assessment result includes a parameter adjustment demand intensity factor and a predicted quality status, and the parameter adjustment demand intensity factor is expressed according to the following formula:

[0011] in, Adjust the demand intensity factor for the parameters. The degree of defect characteristic deviation in the current production cycle. The proportion of future defect prediction sequences that exceed the preset control limit. The trend warning level is based on the EWMA control chart. , and These are the weighting coefficients corresponding to the degree of deviation of defect characteristics, the proportion of predicted over-limits, and the trend warning level, respectively.

[0012] Preferably, in step S202, the OS-ELM model takes the real-time updated exponentially weighted moving average sequence as input, and while keeping the hidden layer input weights and hidden layer bias parameters unchanged, corrects the output weight parameters online through a recursive least squares update method, so that the OS-ELM model updates the future defect prediction sequence as the feature distribution drifts caused by the fluctuation of high-speed production cycle.

[0013] Preferably, step S30, which involves adaptively adjusting the injection molding process parameters using PCA principal component analysis and Mamdani fuzzy PID control based on the defect trend assessment results to obtain process intervention information, specifically includes: Step S301: Extract parameters from the defect trend assessment results to adjust the demand intensity factor and predict the quality status, and combine the defect sensitive feature vector to perform PCA principal component projection to obtain the defect principal component response results. Determine the current defect type based on the defect principal component response results. The current defect type includes at least one of cup rim flash, cup body shrinkage, cup wall tear, or contour deformation. Step S302: Input the parameter adjustment demand intensity factor and the current defect type into the Mamdani fuzzy PID controller, and adjust the proportional coefficient, integral coefficient and derivative coefficient according to the preset fuzzy rule library so that different defect types correspond to different injection molding process parameter adjustment directions; Step S303: Defuzzify the output of the Mamdani fuzzy PID controller using the centroid method, and calculate the process parameter adjustment amount based on the error between the current injection molding process parameters and the target injection molding process parameters to obtain the process intervention information. The process parameter adjustment amount is expressed according to the following formula:

[0014] in, This refers to the adjustment amount of process parameters. This represents the error between the current injection molding process parameters and the target injection molding process parameters during the current production cycle. This represents the error between the current injection molding process parameters and the target injection molding process parameters in the previous production cycle. This is the integral error term obtained by accumulating errors from consecutive production cycles. , and These are the proportional coefficient, integral coefficient, and derivative coefficient after adjustment by the Mamdani fuzzy PID controller.

[0015] Preferably, step S40, which maps the process intervention information to a preset discrete event simulation model and uses a rolling time-domain optimization method to perform cycle time coordination adjustment on the rejection section, cooling section, conveying section, and stacking section to obtain cycle time coordination control instructions, specifically includes: Step S401: Construct a discrete event simulation model based on the process intervention information and the defect trend assessment results. The discrete event simulation model maps each in-process plastic cup to a simulation entity with a product number, predicted quality status, expected arrival timestamps for each process stage, and process adjustment association markers. Step S402: When the predicted quality status indicates the presence of predicted defective products, based on the estimated arrival timestamps of the simulated entity at each work section, the optimal trigger time for the rejection section is calculated using a rolling time-domain optimization method. The optimal trigger time for the rejection section is determined according to the following formula:

[0016] in, To eliminate the optimal trigger time for the work section, The removal trigger time needs to be optimized. To predict the estimated time for defective products to arrive at the rejection section, This refers to the fluctuation in production line cycle time. To eliminate the cache backlog caused by the triggered action, , and These are the cycle time disturbance weighting coefficients corresponding to the expected arrival time deviation, production line cycle time fluctuation, and buffer backlog, respectively. Step S403: Based on the change in injection cycle caused by the process intervention information, the running time of the cooling section, conveying section and stacking section is proportionally fine-tuned, and the optimal trigger time of the rejection section is combined with the running time proportional fine-tuning result into a cycle time collaborative control command.

[0017] Preferably, step S50, which involves updating the model parameters online using a closed-loop parameter update method based on Markov decision processes and Q-learning, based on the actual execution feedback of the rhythmic coordinated control command, to obtain the self-updating results of the model parameters, specifically includes: Step S501: Collect the actual execution results, actual defect detection results, actual cycle time fluctuation results, and process parameter adjustment response results of the cycle time coordinated control command, and construct the actual execution results, actual defect detection results, actual cycle time fluctuation results, and process parameter adjustment response results into state samples, action samples, and reward samples in the Markov decision-making process; Step S502: Construct a reward function based on the actual defect rate, cycle time variance, process parameter adjustment range, and false rejection rate, wherein the reward function is expressed by the following formula:

[0018] in, This is the reward value for the current production cycle. This represents the actual defect rate. For the beat variance, For the adjustment range of process parameters, False rejection rate , , and These are the penalty weighting coefficients corresponding to the actual defect rate, cycle time variance, process parameter adjustment range, and false rejection rate, respectively. Step S503: The Q-learning algorithm is used to update the action value function based on the state samples, action samples, and reward samples. The output weight parameters of the OS-ELM model, the fuzzy rule weights of the Mamdani fuzzy PID controller, and the beat disturbance weight coefficients of the discrete event simulation model are then corrected online based on the updated action value function to obtain the self-updating results of the model parameters.

[0019] This invention also provides a cycle time coordination control system for a high-speed plastic cup production line, comprising: The image feature parsing module is used to acquire a complete image of the plastic cup, perform defect-related feature parsing on the complete image of the plastic cup, and obtain a defect-sensitive feature vector. The defect trend prediction module is used to predict the evolution trend of plastic cup molding defects online based on the defect sensitive feature vector, using EWMA control chart and OS-ELM model, and obtain defect trend evaluation results. The process parameter adjustment module is used to adaptively adjust the injection molding process parameters based on the defect trend assessment results using PCA principal component analysis and Mamdani fuzzy PID control to obtain process intervention information. The cycle time coordination control module is used to map the process intervention information to a preset discrete event simulation model, and to make the discrete event simulation model use a rolling time domain optimization method to perform cycle time coordination adjustment on the rejection section, cooling section, conveying section and stacking section to obtain cycle time coordination control instructions. The closed-loop parameter update module is used to update the model parameters online based on the actual execution feedback of the clockwise coordinated control command, using a closed-loop parameter update method based on Markov decision process and Q-learning, to obtain the self-updated model parameters.

[0020] The present invention also provides a cycle time coordination control device for a high-speed plastic cup production line. The cycle time coordination control device for the high-speed plastic cup production line includes: a memory, a processor, and a cycle time coordination control program for the high-speed plastic cup production line stored in the memory and executable on the processor. When the cycle time coordination control program for the high-speed plastic cup production line is executed by the processor, the above-described method is implemented.

[0021] The present invention also provides a computer program product, the computer program product including a cycle time coordination control program for a high-speed plastic cup production line, the cycle time coordination control program for the high-speed plastic cup production line implementing the above method when executed by a processor.

[0022] The beneficial effects of this invention are as follows: 1. This invention integrates defect trend prediction (EWMA+OS-ELM), adaptive adjustment of process parameters (PCA+Mamdani fuzzy PID), and multi-segment cycle time rolling optimization (discrete event simulation + rolling time domain optimization) to construct a closed-loop control chain of "quality prediction-process intervention-cycle time coordination". This enables intelligent decision-making that connects micro-defect characteristics to macro-production line operating status, effectively solving the technical problem of dynamic coordination between quality control and production cycle time in high-speed production lines.

[0023] 2. This invention introduces a closed-loop parameter update mechanism based on Markov decision process and Q-learning, which can optimize the prediction model, control rules and optimization weights online according to the actual execution effect of the cycle control command (such as actual defect rate, cycle variance, and false rejection rate). This enables the entire collaborative control system to have self-learning and self-adaptive capabilities, continuously adapt to changes and drifts in the production environment, and thus improve the overall performance indicators of the production line in a long-term and stable manner. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the first embodiment of a cycle time coordination control method for a high-speed plastic cup production line according to the present invention. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0026] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1: As Figure 1 The diagram shown is a flowchart illustrating the first embodiment of a cycle time coordination control method for a high-speed plastic cup production line according to the present invention.

[0028] In the first embodiment, the cycle time coordination control method for a high-speed plastic cup production line includes: Step S10: Obtain a complete image of the plastic cup, and perform defect-related feature analysis on the complete image of the plastic cup to obtain a defect-sensitive feature vector; The "complete area image of the plastic cup" in this step refers to a complete digital image covering key quality inspection areas such as the cup rim and sidewalls, acquired and stitched together through a synchronous triggering mechanism. "Defect-related feature analysis" specifically involves using the Sobel operator, Gabor filter, and Local Binary Pattern (LBP) to extract multi-dimensional features such as the rim edge gradient, cup texture direction, and local texture abrupt changes on the cup wall. The resulting "defect-sensitive feature vector" is a numerical representation that integrates the above multi-region features and assigns different defect sensitivity weights. It serves as the direct input for subsequent defect trend prediction, quantitatively reflecting the potential defect morphology and severity of the current plastic cup.

[0029] The technical advantage of this step lies in providing the production line with a real-time, objective, and quantitative initial characterization of product quality through high-precision image acquisition and multi-feature fusion analysis. It transforms complex visual defect information into structured feature vectors, laying a precise and reliable data foundation for subsequent data-driven prediction and decision-making, and is the primary step in achieving predictive quality control.

[0030] Compared to traditional manual visual inspection or simple image comparison of a single area, this step significantly improves the comprehensiveness and accuracy of defect detection through multi-area synchronous acquisition and feature fusion. In particular, it has a stronger feature capture capability for defects of different locations and types, such as cup rim flash, cup body shrinkage marks, and cup wall scratches, avoiding the problem of missed detection due to incomplete detection areas.

[0031] For example, during a production cycle, a high-speed linear scan camera simultaneously acquires images of a plastic cup. Analysis reveals that the gradient feature (Gs) at the cup rim edge is abnormally high, while the texture feature (Gg) of the cup body is normal. After weighted fusion, the component corresponding to the cup rim defect in the defect-sensitive feature vector Fd is significantly prominent, providing a clear signal input for the next step of the system to predict the evolution trend of the "cup rim flash" defect.

[0032] Step S20: Based on the defect-sensitive feature vector, the evolution trend of molding defects in plastic cups is predicted online using EWMA control charts and OS-ELM models to obtain defect trend assessment results; The "EWMA control chart" in this step is a statistical tool used to monitor changes in process quality. It calculates a moving average (Et) by exponentially weighting historical critical defect characteristic components to smooth random fluctuations and highlight trends. The "OS-ELM model" (Online Sequence Extreme Learning Machine) is a neural network capable of online incremental learning, used to predict future changes in defect characteristics based on the time series output of the EWMA. The "Defect Trend Assessment Results" not only include predictions of whether future defects will exceed limits, but also generate quantitative indicators such as the "Parameter Adjustment Demand Intensity Factor (Ra)" and "Predicted Quality Status." These indicators directly indicate whether process parameters need adjustment and the urgency of such adjustment.

[0033] The technical advantage of this step lies in shifting the detection of defects in plastic cup molding from "post-production inspection" to "pre-production prediction." By combining EWMA's robust trend monitoring with OS-ELM's online sequence prediction capabilities, it is possible to detect quality degradation trends in advance and provide a forward-looking decision-making basis for process adjustments, thus potentially allowing for intervention before defects accumulate in large numbers.

[0034] Compared to traditional methods that rely solely on static threshold alarms or fixed models for quality assessment, the EWMA-OS-ELM combined model employed in this step can adapt to the characteristic distribution drift caused by production cycle fluctuations and update the prediction model online. This makes trend predictions more closely aligned with actual production dynamics, improving the timeliness and accuracy of predictions and reducing false alarms and missed alarms.

[0035] For example, the EWMA value (Et) of the local texture mutation feature (Gl) of the cup wall was observed to show a slow upward trend over the last 10 production cycles. Based on this sequence, the OS-ELM model predicted that this value might exceed the control limit within the next 5 cycles. Therefore, a defect trend assessment result was generated, in which the parameter adjustment demand intensity factor Ra increased due to a clear upward trend (high Lt) and the prediction of exceeding the limit (Pt>0). The predicted quality status was marked as "potential cup wall tear risk", which triggered the subsequent process parameter adjustment process.

[0036] Step S30: Based on the defect trend assessment results, PCA principal component analysis and Mamdani fuzzy PID control are used to adaptively adjust the injection molding process parameters to obtain process intervention information; The "PCA Principal Component Analysis" in this step is used to reduce the dimensionality and extract features from the defect trend assessment results (such as Ra) and defect-sensitive feature vectors to obtain the "Defect Principal Component Response Results." Its main function is to more clearly determine the currently dominant defect type (such as flash on the cup rim, shrinkage marks on the cup body, etc.). "Mamdani Fuzzy PID Control" is a method that combines fuzzy logic with PID control. Based on fuzzy inputs such as the "parameter adjustment demand intensity factor" and the "current defect type," it dynamically adjusts the proportional (Kp), integral (Ki), and derivative (Kd) coefficients of the PID controller through a preset fuzzy rule base. The "process intervention information" is ultimately reflected in the calculated specific "process parameter adjustment amount (Δut)," such as the adjustment values ​​for injection pressure, holding time, or mold temperature.

[0037] The technical advantage of this step lies in achieving adaptive and precise adjustment of injection molding process parameters. It intelligently matches different control strategies (by adjusting PID coefficients) based on the type and severity of defects, making process adjustments no longer a "one-size-fits-all" approach, but rather targeted and flexible, thereby more effectively suppressing the further development of specific defects.

[0038] Compared to the limitations of traditional PID control, which has fixed parameters and responds the same way to all defect types, this step uses PCA to identify defect types and combines Mamdani fuzzy inference to dynamically tune PID parameters. This allows the control system to perform the same adjustments as an experienced engineer, focusing on rapid suppression (increasing Kd) for "cup rim flash" and eliminating steady-state error (adjusting Ki) for "cup body shrinkage marks". This significantly improves the effectiveness and adaptability of process parameter adjustments.

[0039] For example, when PCA determines that the current primary defect type is "shrinkage mark on the cup body," and the intensity factor Ra for parameter adjustment needs is large, the Mamdani fuzzy rule base may output a set of adjusted PID parameters: appropriately increasing the integral coefficient Ki to strengthen the correction of historical errors, while fine-tuning the proportional coefficient Kp. The process parameter adjustment amount Δut, calculated based on these parameters and the current temperature error, is essentially an instruction to increase the mold heating temperature, aiming to improve the melt shrinkage compensation effect and eliminate shrinkage marks.

[0040] Step S40: Map the process intervention information to a preset discrete event simulation model, and use the rolling time domain optimization method to perform cycle time coordination adjustment on the rejection section, cooling section, conveying section and stacking section to obtain cycle time coordination control instructions; The "discrete event simulation model" in this step is a computer model that abstracts the production line as a series of discrete events (such as plastic cups arriving at the cooling section and leaving the rejection section). Each plastic cup in production is mapped to a "simulation entity" with attributes such as product number, quality status, and timestamp. "Rolling time-domain optimization" is an optimization method that performs optimization calculations at each decision point based on the current state and predictions of the future finite time domain (such as the arrival sequence of the next 10 products), and only implements the optimal decision at the current moment, then rolls forward over time. The optimization objective is to perform "cycle time coordination adjustment" on the "rejection section, cooling section, conveying section, and stacking section," and its output "cycle time coordination control instructions" specifically include the optimal trigger time for the rejection section and fine-tuning of the runtime of other sections.

[0041] The technical advantage of this step lies in transforming the process intervention information generated in the preceding steps (which may alter the injection molding cycle) into coordinated control instructions for the cycle time of subsequent sections of the entire production line. Through simulation and rolling optimization, it minimizes the disturbances to the overall cycle time of the production line caused by process adjustments and rejection actions, while ensuring that predicted defective products are accurately rejected, thus maintaining the smooth operation of the production line.

[0042] Compared to traditional production line control methods that rely on independent timing or simple linkage between different sections, this step utilizes discrete event simulation and rolling optimization to globally consider the cascading effects of a single process adjustment on multiple subsequent sections and work-in-process inventory, proactively calculating the coordinated solution that minimizes overall cycle time disturbance. This avoids problems such as waiting in subsequent sections, overflowing buffers, or equipment idling caused by local adjustments, achieving globally optimal or suboptimal cycle time control in dynamic environments.

[0043] For example, process intervention information caused the injection molding cycle to increase by 0.1 seconds. Discrete event simulation model pre-runs revealed that without adjustment, the subsequent cooling time would be insufficient, and when a product predicted as defective arrived at the rejection section, it would be too close to a normal product, potentially causing the rejection robot to be unable to react in time and reject it incorrectly. After rolling time-domain optimization, the output instructions were: to slightly increase the cooling section duration by 0.08 seconds, to precisely set the rejection trigger time of the predicted defective product to a specific time window after its arrival and before the arrival of the next normal product, and to fine-tune the conveyor belt speed accordingly. In this way, both quality was ensured and cycle time fluctuations were minimized.

[0044] Step S50: Based on the actual execution feedback of the rhythmic coordinated control command, the model parameters are updated online using a closed-loop parameter update method based on Markov decision process and Q-learning to obtain the model parameter self-update result.

[0045] This step constitutes the closed-loop feedback and self-learning link of the entire method. The "actual execution feedback of the cycle time coordinated control command" includes actual execution results (such as actual action time of each section), actual defect detection results (compared with predictions), actual cycle time fluctuation results, and process parameter adjustment response results (actual process parameter changes). These data are constructed as state, action, and reward samples in the "Markov decision process." The "reward function (rt)" is a negative index (penalty term) that integrates the actual defect rate (Qt), cycle time variance (Vt), process parameter adjustment magnitude (Δut), and false rejection rate (Mt), and its value reflects the comprehensive performance of the current control strategy. The "Q-learning algorithm" learns which action (such as adjusting a model parameter) to take in a specific state to maximize long-term cumulative rewards by iteratively updating the "action value function." Finally, the "model parameter self-update result" is an online correction of key parameters such as OS-ELM output weights, fuzzy PID rule weights, and discrete event simulation optimization weights.

[0046] The technical advantage of this step is that it enables the entire cycle time coordination control system to have the ability to learn and self-optimize online. It no longer relies on fixed models and parameters, but can automatically adjust the parameters of its internal model based on feedback from actual production results. This allows its prediction, control, and optimization capabilities to continuously improve with the accumulation of production data, becoming increasingly adaptable to the specific operating conditions of the production line.

[0047] Compared to traditional control systems that require offline calibration and debugging of parameters and are difficult to adaptively adjust with equipment wear and changes in raw material batches, this step introduces a closed-loop update mechanism based on reinforcement learning, which enables "never-stop" online fine-tuning of parameters. It can cope with various slow or sudden changes in the production environment, always maintain high control performance, and reduce the cost of later maintenance and re-debugging.

[0048] For example, after a period of operation, it was found that although defect prediction was accurate, the false rejection rate (Mt) increased due to overly aggressive tick optimization. When updating the action value function, the Q-learning algorithm gradually learns that parameter adjustment strategies leading to a high false rejection rate have lower value. In the next parameter update, it may automatically increase the weight of the false rejection rate penalty term (η4) in the reward function, or adjust the tick perturbation weight coefficient (μ3) regarding cache backlog (Nb) in the discrete event simulation model. This allows subsequent optimization strategies to be more cautious in dealing with actions that might cause false rejections while maintaining quality, thereby gradually reducing the false rejection rate.

[0049] Example 2: Furthermore, the present invention provides a cycle time coordination control system for a high-speed plastic cup production line, employing the cycle time coordination control method for a high-speed plastic cup production line described in the above embodiments, which can solve the technical problem of cycle time coordination control in a high-speed plastic cup production line. The beneficial effects of the cycle time coordination control system for a high-speed plastic cup production line provided by the present invention are the same as those of the cycle time coordination control method for a high-speed plastic cup production line provided in the above embodiments, and other technical features in the cycle time coordination control system for a high-speed plastic cup production line are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0050] Example 3: This invention provides a rhythm coordination control device for a high-speed plastic cup production line. The device includes: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor, which are then executed to enable the at least one processor to perform the rhythm coordination control method for a high-speed plastic cup production line described in Example 1. The rhythm coordination control device for a high-speed plastic cup production line in this invention may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This rhythm coordination control device for a high-speed plastic cup production line is merely an example and should not limit the functionality or scope of the invention. A cycle time coordination control device for a high-speed plastic cup production line may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in a read-only memory or a program loaded from a storage device into a random access memory. The random access memory also stores various programs and data required for the operation of the cycle time coordination control device for the high-speed plastic cup production line. The processing unit, read-only memory, and random access memory are interconnected via a bus. An I / O interface is also connected to the bus. Typically, the following systems can be connected to the I / O interface: input devices including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices including, for example, magnetic tapes, hard disks, etc.; and communication devices. The communication device allows the cycle time coordination control device for the high-speed plastic cup production line to communicate wirelessly or wiredly with other devices to exchange data. Although a cycle time coordination control device for a high-speed plastic cup production line with various systems is shown in the figures, it should be understood that implementation or possession of all the systems shown is not required. It can be implemented alternatively or have more or fewer systems.

[0051] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described cycle time coordination control method for a high-speed plastic cup production line. The computer program product provided by this invention can solve the technical problem of cycle time coordination control for a high-speed plastic cup production line. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the cycle time coordination control method for a high-speed plastic cup production line provided in the above embodiments, and will not be repeated here.

[0052] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a read-only memory. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0053] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0054] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the present invention and its equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A cycle time coordination control method for a high-speed plastic cup production line, characterized in that, The methods include: Step S10: Obtain a complete image of the plastic cup, and perform defect-related feature analysis on the complete image of the plastic cup to obtain a defect-sensitive feature vector; Step S20: Based on the defect-sensitive feature vector, the evolution trend of molding defects in plastic cups is predicted online using EWMA control charts and OS-ELM models to obtain defect trend assessment results; Step S30: Based on the defect trend assessment results, PCA principal component analysis and Mamdani fuzzy PID control are used to adaptively adjust the injection molding process parameters to obtain process intervention information; Step S40: Map the process intervention information to a preset discrete event simulation model, and use the rolling time domain optimization method to perform cycle time coordination adjustment on the rejection section, cooling section, conveying section and stacking section to obtain cycle time coordination control instructions; Step S50: Based on the actual execution feedback of the rhythmic coordinated control command, the model parameters are updated online using a closed-loop parameter update method based on Markov decision process and Q-learning to obtain the model parameter self-update result.

2. The cycle time coordination control method for a high-speed plastic cup production line as described in claim 1, characterized in that, Step S10, which involves acquiring a complete image of the plastic cup and performing defect-related feature analysis on the complete image to obtain a defect-sensitive feature vector, specifically includes: Step S101: Using a high-speed linear array camera and pulse light source that are synchronously triggered with the injection molding machine's mold opening and closing signals and the conveyor belt encoder pulses, images of the cup mouth area and the cup body sidewall are acquired when the plastic cup passes through the detection station. The positions are aligned and the areas are stitched together according to the displacement corresponding to the conveyor belt encoder pulses to obtain the complete area image of the plastic cup. Step S102: The Sobel operator is used to extract the gradient features of the cup rim edge, the Gabor filter is used to extract the texture direction features of the cup body, and the local binary mode is used to extract the local texture abrupt change features of the cup wall, so as to obtain a multi-region defect candidate feature set. Step S103: Perform weighted fusion on the multi-region defect candidate feature set to obtain the defect-sensitive feature vector, wherein the defect-sensitive feature vector is expressed according to the following formula: in, For defect-sensitive feature vectors, The gradient features of the cup rim obtained by the Sobel operator are shown below. The cup body texture direction features are obtained by Gabor filter. The abrupt changes in local texture of the cup wall are obtained from the local binary pattern. , and These are the defect sensitivity weight coefficients corresponding to the gradient features of the cup rim edge, the texture direction features of the cup body, and the local texture abrupt change features of the cup wall.

3. The cycle time coordination control method for a high-speed plastic cup production line as described in claim 1, characterized in that, Step S20, which involves using the defect-sensitive feature vector to predict the evolution trend of molding defects in plastic cups online using EWMA control charts and OS-ELM models, and obtaining the defect trend assessment results, specifically includes: Step S201: Select key defect feature components related to rim flash, shrinkage marks, or tearing of the cup wall from the defect-sensitive feature vector, and calculate the exponentially weighted moving average using an EWMA control chart. The exponentially weighted moving average is expressed by the following formula: in, It is the index-weighted moving average of the current production cycle. These are the key defect feature components of the current production cycle. It is the index-weighted moving average of the previous production cycle. For smoothing coefficients; Step S202: Input the exponentially weighted moving average sequence of multiple consecutive production cycles into the OS-ELM model to predict the trend of defect feature changes in multiple future production cycles, and obtain the future defect prediction sequence. Step S203: Based on the exponentially weighted moving average, the future defect prediction sequence, and the preset qualified product characteristic control limits, generate the defect trend assessment result, wherein the defect trend assessment result includes a parameter adjustment demand intensity factor and a predicted quality status, and the parameter adjustment demand intensity factor is expressed according to the following formula: in, Adjust the demand intensity factor for the parameters. The degree of defect characteristic deviation in the current production cycle. The proportion of future defect prediction sequences that exceed the preset control limit. The trend warning level is based on the EWMA control chart. , and These are the weighting coefficients corresponding to the degree of deviation of defect characteristics, the proportion of predicted over-limits, and the trend warning level, respectively.

4. The cycle time coordination control method for a high-speed plastic cup production line as described in claim 3, characterized in that, In step S202, the OS-ELM model takes the real-time updated exponentially weighted moving average sequence as input. While keeping the hidden layer input weights and hidden layer bias parameters unchanged, the output weight parameters are corrected online by recursive least squares update, so that the OS-ELM model updates the future defect prediction sequence as the feature distribution drifts caused by the fluctuation of high-speed production cycle.

5. The cycle time coordination control method for a high-speed plastic cup production line as described in claim 1, characterized in that, Step S30, which involves adaptively adjusting the injection molding process parameters using PCA principal component analysis and Mamdani fuzzy PID control based on the defect trend assessment results to obtain process intervention information, specifically includes: Step S301: Extract parameters from the defect trend assessment results to adjust the demand intensity factor and predict the quality status, and combine the defect sensitive feature vector to perform PCA principal component projection to obtain the defect principal component response results. Determine the current defect type based on the defect principal component response results. The current defect type includes at least one of cup rim flash, cup body shrinkage, cup wall tear, or contour deformation. Step S302: Input the parameter adjustment demand intensity factor and the current defect type into the Mamdani fuzzy PID controller, and adjust the proportional coefficient, integral coefficient and derivative coefficient according to the preset fuzzy rule library so that different defect types correspond to different injection molding process parameter adjustment directions; Step S303: Defuzzify the output of the Mamdani fuzzy PID controller using the centroid method, and calculate the process parameter adjustment amount based on the error between the current injection molding process parameters and the target injection molding process parameters to obtain the process intervention information. The process parameter adjustment amount is expressed according to the following formula: in, This refers to the adjustment amount of process parameters. This represents the error between the current injection molding process parameters and the target injection molding process parameters during the current production cycle. This represents the error between the current injection molding process parameters and the target injection molding process parameters in the previous production cycle. This is the integral error term obtained by accumulating errors from consecutive production cycles. , and These are the proportional coefficient, integral coefficient, and derivative coefficient after adjustment by the Mamdani fuzzy PID controller.

6. The cycle time coordination control method for a high-speed plastic cup production line as described in claim 1, characterized in that, Step S40, which maps the process intervention information to a preset discrete event simulation model and uses a rolling time-domain optimization method to perform cycle time coordination adjustment on the rejection section, cooling section, conveying section, and stacking section to obtain cycle time coordination control instructions, specifically includes: Step S401: Construct a discrete event simulation model based on the process intervention information and the defect trend assessment results. The discrete event simulation model maps each in-process plastic cup to a simulation entity with a product number, predicted quality status, expected arrival timestamps for each process stage, and process adjustment association markers. Step S402: When the predicted quality status indicates the presence of predicted defective products, based on the estimated arrival timestamps of the simulated entity at each work section, the optimal trigger time for the rejection section is calculated using a rolling time-domain optimization method. The optimal trigger time for the rejection section is determined according to the following formula: in, To eliminate the optimal trigger time for the work section, The removal trigger time needs to be optimized. To predict the estimated time for defective products to arrive at the rejection section, This refers to the fluctuation in production line cycle time. To eliminate the cache backlog caused by the triggered action, , and These are the cycle time disturbance weighting coefficients corresponding to the expected arrival time deviation, production line cycle time fluctuation, and buffer backlog, respectively. Step S403: Based on the change in injection cycle caused by the process intervention information, the running time of the cooling section, conveying section and stacking section is proportionally fine-tuned, and the optimal trigger time of the rejection section is combined with the running time proportional fine-tuning result into a cycle time collaborative control command.

7. The cycle time coordination control method for a high-speed plastic cup production line as described in claim 1, characterized in that, Step S50, which involves updating the model parameters online using a closed-loop parameter update method based on Markov decision processes and Q-learning, based on the actual execution feedback of the rhythmic coordinated control command, to obtain the self-updated model parameters, specifically includes: Step S501: Collect the actual execution results, actual defect detection results, actual cycle time fluctuation results, and process parameter adjustment response results of the cycle time coordinated control command, and construct the actual execution results, actual defect detection results, actual cycle time fluctuation results, and process parameter adjustment response results into state samples, action samples, and reward samples in the Markov decision-making process; Step S502: Construct a reward function based on the actual defect rate, cycle time variance, process parameter adjustment range, and false rejection rate, wherein the reward function is expressed by the following formula: in, This is the reward value for the current production cycle. This represents the actual defect rate. For the beat variance, For the adjustment range of process parameters, False rejection rate , , and These are the penalty weighting coefficients corresponding to the actual defect rate, cycle time variance, process parameter adjustment range, and false rejection rate, respectively. Step S503: The Q-learning algorithm is used to update the action value function based on the state samples, action samples, and reward samples. The output weight parameters of the OS-ELM model, the fuzzy rule weights of the Mamdani fuzzy PID controller, and the beat disturbance weight coefficients of the discrete event simulation model are then corrected online based on the updated action value function to obtain the self-updating results of the model parameters.

8. A cycle time coordination control system for a high-speed plastic cup production line, applied to the cycle time coordination control method for a high-speed plastic cup production line according to any one of claims 1 to 7, characterized in that, The cycle time coordination control system of the high-speed plastic cup production line includes: The image feature parsing module is used to acquire a complete image of the plastic cup, perform defect-related feature parsing on the complete image of the plastic cup, and obtain a defect-sensitive feature vector. The defect trend prediction module is used to predict the evolution trend of plastic cup molding defects online based on the defect sensitive feature vector, using EWMA control chart and OS-ELM model, and obtain defect trend evaluation results. The process parameter adjustment module is used to adaptively adjust the injection molding process parameters based on the defect trend assessment results using PCA principal component analysis and Mamdani fuzzy PID control to obtain process intervention information. The cycle time coordination control module is used to map the process intervention information to a preset discrete event simulation model, and to make the discrete event simulation model use a rolling time domain optimization method to perform cycle time coordination adjustment on the rejection section, cooling section, conveying section and stacking section to obtain cycle time coordination control instructions. The closed-loop parameter update module is used to update the model parameters online based on the actual execution feedback of the clockwise coordinated control command, using a closed-loop parameter update method based on Markov decision process and Q-learning, to obtain the self-updated model parameters.

9. A cycle time coordination control device for a high-speed plastic cup production line, characterized in that, The cycle time coordination control device of the high-speed plastic cup production line includes: a memory, a processor, and a cycle time coordination control program for the high-speed plastic cup production line stored in the memory and executable on the processor. When the cycle time coordination control program for the high-speed plastic cup production line is executed by the processor, it implements a cycle time coordination control method for a high-speed plastic cup production line according to any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a cycle time coordination control program for a high-speed plastic cup production line. When the cycle time coordination control program for the high-speed plastic cup production line is executed by a processor, it implements a cycle time coordination control method for a high-speed plastic cup production line according to any one of claims 1 to 7.

Images