Production method of degradable plastic

Abstract

The invention provides a degradable plastic production method which comprises the following steps: acquiring an optimal raw material proportioning scheme, and controlling an intelligent batching system to perform automatic batching; temperature and humidity parameters of raw material storage and conveying links are obtained, and raw material pretreatment process parameters are dynamically adjusted in combination with time sequence data analysis; performing quality detection on the preprocessed raw materials, identifying defects and impurities of the raw materials by adopting machine vision and a deep learning algorithm, and removing unqualified raw materials; various raw materials are accurately weighed through an industrial robot and mixed according to the optimized formula proportion; a thermal infrared imager is adopted to monitor temperature field distribution of a plastic extrusion forming mold in real time, and finite element simulation analysis is combined to optimize mold structure design and a heating and refrigerating system; the extrusion molding process is monitored on line, material component changes are analyzed through a chemometrics model, and if degradation products exceed the standard, extrusion process parameters are adjusted or a mold is replaced.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to a method for producing degradable plastics. Background Art

[0002] As a new type of environmentally friendly material, biodegradable plastics face technical contradictions in the production process of increasing output and efficiency. On the one hand, biodegradable plastics need to achieve large-scale production under the premise of ensuring performance and quality to meet the growing market demand; on the other hand, traditional production processes are difficult to adapt to the special properties of biodegradable plastics, resulting in low production efficiency and high costs.

[0003] Specifically, degradable plastics are easily affected by environmental factors such as temperature and humidity during the production process, resulting in unstable product performance. At the same time, the ratio of degradable plastic raw materials requires precision, and the quality of raw materials from different batches varies greatly, which brings difficulties to production. In addition, degradable plastics are prone to degradation during the extrusion molding process, resulting in a decrease in product strength, while increasing the extrusion temperature may cause material denaturation.

[0004] Therefore, to achieve high yield and high efficiency of biodegradable plastics, it is necessary to innovate and optimize production processes and equipment. An intelligent batching system is introduced to achieve precise control of raw material ratios. Summary of the invention

[0005] The present invention provides a method for producing a degradable plastic, which mainly comprises: Obtain the performance requirements of biodegradable plastic products, establish a raw material formula database, use machine learning algorithms to analyze historical production data, and obtain the optimal raw material ratio scheme; obtain the optimal raw material ratio scheme, and control the intelligent batching system for automatic batching; obtain the temperature and humidity parameters of the raw material storage and transportation links, and dynamically adjust the raw material pretreatment process parameters in combination with time series data analysis; conduct quality inspection on the pretreated raw materials, use machine vision and deep learning algorithms to identify raw material defects and impurities, and eliminate unqualified raw materials; use industrial robots to accurately weigh various raw materials and mix them according to the optimized formula ratio; use infrared thermal imagers to monitor the temperature field distribution of plastic extrusion molding molds in real time, and combine finite element simulation analysis to optimize the mold structure design and heating and cooling system; monitor the extrusion molding process online, use chemometric models to analyze material component changes, and adjust the extrusion process parameters or replace the mold if the degradation product exceeds the standard; collect data from the entire production process, use digital twin technology to build a virtual model of the production line, and find the best combination of process parameters through simulation optimization to guide actual production. Furthermore, the method of obtaining the performance requirements of the biodegradable plastic products, establishing a raw material formula database, and using a machine learning algorithm to analyze historical production data to obtain the optimal raw material ratio scheme includes: establishing a raw material formula database to record the performance parameters and ratio information of various raw materials; collecting historical production data, including raw material types, ratios, process parameters and product performance; using data preprocessing technology to clean and standardize the raw material formula database and historical production data to remove outliers and missing values; extracting key features from the raw material formula database and historical production data through feature engineering methods; selecting a machine learning algorithm to establish a prediction model for predicting product performance of different raw material ratio schemes; using a genetic algorithm to search for the optimal raw material ratio scheme in the raw material formula database, and optimizing the ratio scheme through crossover and mutation operations until the ratio with the best predicted product performance is found. Furthermore, the temperature and humidity parameters of the raw material storage and transportation links are obtained, combined with time series data analysis, to dynamically adjust the raw material pretreatment process parameters, including: deploying temperature and humidity sensors to collect the temperature and humidity parameters of the raw material environment in real time; transmitting the collected raw material temperature and humidity time series data to the data analysis platform, and cleaning, denoising and normalizing the data; using feature engineering methods to extract key features from historical raw material temperature and humidity data, including statistics, temperature and humidity change trends and fluctuation cycle time series characteristics; using time series analysis algorithms to establish a prediction model for temperature and humidity parameters and raw material quality; according to the established prediction model, calculating the raw material quality prediction value under the current raw material temperature and humidity parameters, and comparing it with the set quality threshold; if the predicted quality is lower than the threshold, triggering an early warning; when the predicted quality deviates from the optimal range, using an optimization algorithm to search for the optimal raw material pretreatment process parameter combination so that the predicted quality reaches the target value. Furthermore, the quality inspection of the pretreated raw materials, the use of machine vision and deep learning algorithms to identify raw material defects and impurities, and the removal of unqualified raw materials include: obtaining pretreated raw material image data, inputting the image data into a pre-built convolutional neural network model for feature extraction and classification recognition, and obtaining position coordinate information of raw material defects and impurities; using an image segmentation algorithm to segment the raw material image into a foreground and background according to the position coordinates of the defects and impurities, and obtaining a binary segmented image of the raw material defect area and the impurity area; calculating the raw material defect area ratio and impurity area ratio by counting the white pixels in the segmented image; comparing the ratio value with a preset threshold, and if the defect area ratio or the impurity area ratio exceeds the threshold, the raw material is determined to be unqualified; controlling a manipulator or a pneumatic device to remove unqualified products from the production line; for qualified raw materials that are not removed, extracting and saving their image features, and establishing an image feature database for qualified raw materials. Furthermore, the method of accurately weighing various raw materials by an industrial robot and mixing them according to an optimized formula ratio includes: obtaining target weighing data of various raw materials according to a preset formula ratio and sending it to the control system of the industrial robot; accurately weighing the actual weight of each raw material by a weighing sensor of the industrial robot, and comparing the actual weighing data with the target weighing data in real time; automatically adjusting, supplementing or reducing the amount of corresponding raw materials added according to the deviation value between the actual weighing and the target weighing, through the control system of the industrial robot, until the ratio requirement is met; adding the weighed raw materials to a designated mixing device through the industrial robot according to the optimized formula ratio; starting the mixing device to fully mix the added raw materials, and automatically controlling the mixing time according to preset parameters; sampling the mixed materials, obtaining key quality parameters of the mixture through online quality detection equipment, and transmitting the detection data to the formula optimization system in real time. Furthermore, the use of an infrared thermal imager to monitor the temperature field distribution of the plastic extrusion molding die in real time, combined with finite element simulation analysis, to optimize the mold structure design and heating and cooling system, includes: using an infrared thermal imager to monitor the temperature field of the plastic extrusion molding die in real time to obtain temperature distribution data on the mold surface; converting the obtained temperature distribution data into a grid data format that can be recognized by finite element simulation software through an interpolation algorithm, and then inputting it into the finite element simulation analysis software to perform numerical simulation and analysis on the temperature field of the mold; judging whether the temperature field distribution of the mold is uniform based on the results of the finite element simulation analysis, and determining the area where the temperature field distribution is uneven; for the area where the temperature field distribution is uneven, optimizing the mold structure, adjusting the mold wall thickness, and optimizing the flow channel layout to improve the temperature field distribution; according to the optimized mold structure, adjusting and optimizing the electric heating rod and cooling water channel layout of the mold accordingly to achieve precise control of the mold temperature field. Furthermore, the online monitoring of the extrusion molding process uses a chemometric model to analyze changes in material components, and if the degradation products exceed the standard, the extrusion process parameters are adjusted or the mold is replaced, including: establishing a finite element model of the temperature field distribution of the extrusion molding process according to the material components and process parameters, and simulating and analyzing the influence of the temperature field distribution on the material components; using an online near-infrared spectrometer to monitor the extrusion molding process in real time, obtaining near-infrared spectrum data of the material components, and transmitting the data to a computer for analysis and processing; using a chemometric method to establish a quantitative relationship model between the material components and the near-infrared spectrum data, optimizing the model through a cross-validation method, and realizing rapid prediction of the material components; in the extrusion molding process, near-infrared spectrum data is obtained in real time and input into the chemometric model to predict the content of the material components; comparing the predicted value with the preset qualified range of the material components to determine whether the degradation products exceed the standard; if the predicted material components do not meet the requirements, according to the temperature field distribution model, an optimization algorithm is used to find the optimal combination of process parameters such as extrusion temperature, extrusion pressure, and extrusion speed to control the temperature field distribution and reduce material degradation. Furthermore, the data of the entire production process is collected, and a virtual model of the production line is constructed using the digital twin technology. The optimal combination of process parameters is found through simulation optimization to guide actual production, including: according to the real-time data of each process of the production line, including raw material ratio, pretreatment parameters, extrusion parameters, etc., a digital twin model of the production line is constructed using relevant software; production data is acquired in real time through relevant protocols, and the digital twin model is updated in real time using the simulation function of the relevant software to ensure that the model is synchronized with the actual production line status; in the updated digital twin model, different combinations of process parameters are set, and simulation optimization is performed using the simulation function of the relevant software to evaluate the product quality and production efficiency under each parameter combination; the optimal combination of process parameters is determined through analysis of simulation results; the optimal combination of process parameters obtained by optimization is applied to the actual production line, and production data is monitored in real time, and statistical methods such as quality control charts are used to determine whether the product quality meets expectations; if the product quality does not meet expectations, the digital twin model is returned, and the process parameters are adjusted using the design optimization algorithm according to the quality deviation, and simulation optimization is performed again until the optimal combination of parameters that meets quality requirements is found.

[0006] The technical solution provided by the embodiment of the present invention may have the following beneficial effects: The present invention discloses an intelligent production method for degradable plastics. The method establishes a raw material formula database and applies a machine learning algorithm to analyze historical data to obtain an optimal raw material ratio scheme. During the production process, the present invention uses an Internet of Things sensor to monitor the temperature and humidity parameters of the raw material storage and transportation links in real time, and dynamically adjusts the pretreatment process. At the same time, machine vision and deep learning algorithms are used to perform quality inspection on the raw materials and remove unqualified products. In the plastic extrusion molding stage, the present invention uses an infrared thermal imager to monitor the temperature field distribution of the mold, and optimizes the mold design in combination with finite element simulation analysis. In addition, an online near-infrared spectrometer is used to monitor the changes in material components during the extrusion process in real time, and the process parameters are adjusted in time. Finally, the present invention uses digital twin technology to construct a virtual model of the production line, and finds the optimal combination of process parameters through simulation optimization. This intelligent production method significantly improves the quality stability and production efficiency of degradable plastic products. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 This is a flow chart of a method for producing a degradable plastic according to the present invention. In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0008] like Figure 1 In this embodiment, a method for producing a degradable plastic may specifically include: S101. According to the performance requirements of biodegradable plastic products, a raw material formula database is established, and the historical production data is analyzed using machine learning algorithms to obtain the optimal raw material ratio scheme. The optimal raw material ratio scheme is input into the intelligent batching system for subsequent automated batching control.

[0009] According to the performance requirements of biodegradable plastic products, a raw material formula database is established to record the performance parameters and ratio information of various raw materials, and historical production data is collected at the same time, including raw material types, ratios, process parameters and product performance. Data preprocessing technology is used to clean and standardize the raw material formula database and historical production data, remove outliers and missing values, and ensure data quality. Through feature engineering methods, key features such as raw material types, ratios, process parameters, etc. are extracted from the raw material formula database and historical production data to construct input features of the machine learning model. Suitable machine learning algorithms, such as support vector regression or random forest regression, are selected to establish a prediction model to predict product performance of different raw material ratio schemes. Genetic algorithms are used to search for the optimal raw material ratio scheme in the raw material formula database, and the ratio scheme is continuously optimized through operations such as crossover and mutation until the ratio with the best predicted product performance is found. The optimal raw material ratio scheme obtained is converted into batching control parameters, including the amount, order and time of various raw materials, etc., to form detailed batching control instructions and transmit them to the intelligent batching system. The intelligent batching system controls the automatic raw material feeding device according to the received batching control instructions, and accurately feeds various raw materials into the mixing equipment according to the specified feeding amount, feeding sequence and feeding time. During the feeding process, the feeding amount is monitored in real time by sensors, and the feeding device is adjusted through the feedback control mechanism to ensure that the actual feeding amount is consistent with the batching control instructions, thereby achieving accurate and automatic batching and ensuring the stability and consistency of product performance.

[0010] Specifically, the optimization of raw material formula for degradable plastic products is a complex process involving multiple steps and technologies. First, it is key to establish a raw material formula database. This database contains the performance parameters and ratio information of various raw materials, such as the tensile strength and bio-based content of polylactic acid (PLA). At the same time, historical production data is collected, such as the degradation cycle of the product obtained by extrusion molding at 180°C using a formula of 70% PLA and 30% starch. Data preprocessing is an important step to ensure the accuracy of the model. For example, for raw material content data, there may be outliers such as 150% records, which need to be identified and eliminated by statistical methods. Standardization processing can unify data of different dimensions, such as tensile strength (MPa) and bio-based content (%), to the same scale for subsequent modeling. Feature engineering is the process of extracting key information. For example, the type of raw material can be converted into one-hot encoding, PLA can be represented as [1,0,0], and starch can be represented as [0,1,0]. Process parameters such as temperature and pressure can be directly used as numerical features. These processed features constitute the input of the machine learning model. It is crucial to choose the appropriate machine learning algorithm. Support vector regression is suitable for processing high-dimensional features and can effectively predict product performance corresponding to different formulas. Random forest regression can handle nonlinear relationships and provide feature importance ranking, such as it may be found that temperature has a greater impact on product performance than pressure. Genetic algorithms are used to search for the optimal solution in a large formula space. For example, the initial population may contain 10 different formula schemes. Through crossover operations, a formula of 70% PLA and 30% starch is combined with a formula of 60% PLA and 40% modified starch, which may produce a new formula of 65% PLA and 35% modified starch. Mutation operations may randomly replace starch in a formula with cellulose to explore new performance possibilities. The conversion of optimal formulas into control parameters is the key to achieving precise production. If the optimal formula is 65% PLA and 35% modified starch, with a total weight of 100kg, the PLA dosage needs to be set to 65kg and the modified starch to 35kg. The order of addition may be to add PLA first and then modified starch, and the time for each raw material may be 2 minutes. The intelligent batching system executes these precise instructions. For example, the screw feeder for PLA may be set to a feeding rate of 32.5kg / min for 2 minutes. During this process, the weight sensor monitors the feeding amount in real time. If the actual feeding rate is detected to be 31.8kg / min, the control system will automatically increase the screw speed to ensure that the final feeding amount is accurate to 65kg. This intelligent and precise batching method not only improves production efficiency, but more importantly, ensures the consistency of product performance. Through data-driven optimization and precise control, the problem of unstable product performance caused by fluctuations in raw material ratios can be significantly reduced, providing a strong guarantee for the quality of biodegradable plastic products.

[0011] S102. In the process of raw material storage and transportation, IoT sensors collect temperature and humidity parameters in real time, and combine them with time series data analysis to dynamically adjust the raw material pretreatment process parameters, including drying temperature and time.

[0012] By deploying temperature and humidity sensors in the raw material storage and transportation links, the temperature and humidity parameters of the raw material environment are collected in real time. The collected raw material temperature and humidity time series data are transmitted to the data analysis platform, and the data is cleaned, denoised and normalized to ensure data quality. Using feature engineering methods, key features are extracted from historical raw material temperature and humidity data, such as statistical quantities such as mean, variance, maximum value, minimum value, and time series features such as the change trend and fluctuation cycle of temperature and humidity. Using time series analysis algorithms such as ARIMA or LSTM neural network, a prediction model for temperature and humidity parameters and raw material quality is established. Through methods such as cross-validation and grid search, the model hyperparameters are optimized to improve the prediction accuracy. According to the established prediction model, the raw material quality prediction value under the current raw material temperature and humidity parameters is calculated and compared with the set quality threshold. If the prediction quality is lower than the threshold, an early warning is triggered and the relevant personnel are notified by email, SMS, etc. When the prediction quality deviates from the optimal range, an optimization algorithm such as gradient descent or genetic algorithm is used to search for the optimal raw material pretreatment process parameter combination, such as drying temperature and time, so that the prediction quality reaches the target value. The optimized drying temperature and time parameters are passed to the PLC control system. By adjusting the heater power and conveyor belt speed, the operating status of the drying equipment is adjusted in real time to achieve dynamic optimization control of the raw material pretreatment process. After each batch of raw material pretreatment is completed, the quality parameters of the processed raw materials, such as moisture content, impurity rate, etc., are collected and fed back to the data analysis platform. Retrain and optimize the prediction model using the newly collected data to improve the generalization ability and robustness of the model. Use data visualization tools such as Tableau or PowerBI to design intuitive dashboards and reports to display real-time temperature and humidity monitoring data and quality prediction results of various links such as raw material warehousing, transportation and pretreatment, provide decision support for production managers, and achieve full-process visualization control of raw material quality.

[0013] Specifically, in the production of biodegradable plastics, raw material quality control is crucial. By deploying temperature and humidity sensors to monitor the raw material environment in real time, quality problems can be effectively prevented. For example, a DHT22 digital temperature and humidity sensor is installed in the PLA raw material warehouse to collect data every 5 minutes. After preprocessing, these data can be used to build a prediction model. Data preprocessing is a key step to ensure the accuracy of the model. For example, the temperature data of a batch of PLA raw materials has an abnormal value of 45°C, which is much higher than the normal range of 20-30°C, which may be caused by sensor failure. Such abnormal values ​​are identified and eliminated by the median method to improve data quality. In addition, standardizing the temperature (°C) and relative humidity (%) to the range of 0-1 helps with subsequent modeling. Feature engineering is the process of extracting useful information from raw data. For temperature and humidity time series data, statistics such as daily mean and standard deviation can be calculated. For example, the average daily temperature of the PLA raw material warehouse in a certain month was 25.3°C, and the standard deviation was 1.2°C, reflecting the stability of temperature. In addition, the periodic characteristics of temperature and humidity can be extracted by Fourier transform, such as the obvious day and night fluctuations in temperature. Time series analysis algorithms can effectively predict raw material quality. The ARIMA model is suitable for data with linear trends and seasonality, while the LSTM neural network can capture complex nonlinear relationships. For example, using the temperature and humidity data of the past 30 days, the LSTM model can predict the moisture content changes of PLA raw materials in the next 7 days. By optimizing the number of hidden layers and learning rate of LSTM through grid search, the prediction error can be controlled within ±0.5%. Early warning and optimization control based on prediction results are the key to realizing intelligent production. Assume that the optimal moisture content of PLA raw materials is 0.5% and the early warning threshold is 0.8%. When the model predicts that the moisture content reaches 0.75%, the system automatically sends an early warning email to the quality control personnel. At the same time, the optimization algorithm calculates the optimal drying parameters: temperature 80℃, time 4 hours, and passes these parameters to the PLC control system to automatically adjust the operating status of the drying equipment. Continuous optimization is the guarantee for maintaining efficient operation of the system. After each batch of raw materials is processed, its actual moisture content is measured, such as 0.48%, and this data is fed back to the model. Through online learning, the model can continuously adapt to slight changes in raw material characteristics, such as changes in environmental humidity caused by seasonal changes. This closed-loop optimization ensures the long-term effectiveness of the prediction model. Visualization provides intuitive support for decision-making. For example, using a heat map to display the temperature distribution of the raw material warehouse can quickly identify potential hot spots. Trend charts can show historical changes in moisture content and future forecasts, helping managers to adjust production plans in a timely manner. Through this full-process visual control, the quality of raw materials for biodegradable plastic production is fully guaranteed, providing a stable and reliable raw material foundation for subsequent processing.

[0014] S103. Perform quality inspection on the raw materials after pretreatment, use machine vision and deep learning algorithms to identify raw material defects and impurities, remove unqualified raw materials, and ensure the stability of the quality of raw materials entering production.

[0015] The pre-processed raw material image data is obtained, and the image data is input into the pre-built convolutional neural network model for feature extraction and classification recognition to obtain the position coordinate information of raw material defects and impurities. According to the position coordinates of defects and impurities, the grabcut algorithm is used to segment the raw material image foreground and background to obtain a binary segmented image of the raw material defect area and impurity area. By counting the white pixels in the segmented image, the raw material defect area ratio and impurity area ratio are calculated, and the ratio values ​​are compared with the preset threshold. If the defect area ratio or impurity area ratio exceeds the threshold, the raw material is judged to be unqualified. For raw materials judged to be unqualified, the manipulator or pneumatic device is controlled to remove them from the production line, and the number of removed raw materials is counted and recorded for subsequent raw material quality analysis. For qualified raw materials that are not removed, their image features such as color, texture, and shape are extracted and saved, and an image feature database of qualified raw materials is established for subsequent raw material quality stability analysis and production parameter optimization. During the raw material quality inspection process, if the defect rate or impurity rate of multiple batches of raw materials exceeds the preset threshold, an early warning message will be triggered to notify the quality management personnel to conduct a traceability analysis of the quality problems of the raw material supplier, and require the supplier to provide corrective and preventive measures and track the implementation of the rectification. Regularly perform K-means clustering analysis on the image feature database of qualified raw materials, optimize the weight parameters of the convolutional neural network model according to the cluster center, and fine-tune it through the back propagation algorithm to improve the accuracy of raw material defect and impurity identification, ensuring the continuous optimization and upgrading of the raw material quality inspection system.

[0016] Specifically, in the production of biodegradable plastics, raw material quality inspection is a key link. Through image processing and deep learning technology, efficient and accurate quality control can be achieved. First, obtain the pre-processed raw material image data, which are usually taken by high-resolution industrial cameras and cover all angles of the raw materials. Input these images into a pre-trained convolutional neural network model, such as ResNet50 or VGG16, for feature extraction and classification recognition. The model output contains the location coordinate information of defects and impurities, with pixel-level accuracy. After obtaining the location coordinates, the GrabCut algorithm is used to segment the raw material image foreground and background. GrabCut is an iterative graph cut algorithm that can effectively separate the target object and the background. In PLA particle detection, the defect and impurity areas can be regarded as the foreground, and the rest as the background. The algorithm outputs a binary segmented image, in which white pixels represent defect or impurity areas. By calculating the number of white pixels, the defect area ratio and impurity area ratio can be obtained. For example, the defect area ratio of a batch of PLA raw materials is 2.5%, and the impurity area ratio is 1.8%. Compare these values ​​with preset thresholds, such as a defect ratio threshold of 3% and an impurity ratio threshold of 2%. If the threshold is exceeded, it is judged as a defective product. For defective products, a manipulator or pneumatic device is used to remove them. This process is coordinated by the PLC control system to ensure accuracy and speed. The number of raw materials removed will be recorded for subsequent analysis. For example, if the rejection rate reaches 5% on a certain day, which is much higher than the normal level of 1-2%, it may indicate quality problems of upstream suppliers. The image features of qualified raw materials, such as color histogram, gray level co-occurrence matrix, shape descriptors, etc., are extracted and stored in the database. These data are used for raw material quality stability analysis and production parameter optimization. By comparing the feature data of different batches, slight changes in the properties of raw materials can be discovered in time. If the defect rate or impurity rate of multiple batches of raw materials exceeds the standard, the system will trigger an early warning. The quality management personnel will then carry out traceability analysis, which may involve multiple links such as the supplier's production process and transportation conditions. Suppliers are required to provide detailed corrective and preventive measures and accept follow-up audits. Regular K-means clustering analysis of qualified raw material image features can reveal potential changes in raw material quality. The cluster center reflects the typical characteristics of the raw material and can be used to optimize the neural network model. The back-propagation algorithm fine-tunes the model weights to improve the accuracy of defect and impurity identification. This continuous optimization mechanism ensures that the detection system can adapt to dynamic changes in raw material characteristics, such as color changes caused by seasonal changes.

[0017] S104, the intelligent batching system uses industrial robots to accurately weigh various raw materials according to the input optimal ratio, mix them according to the optimized formula ratio, and realize automatic control of the raw material ratio.

[0018] According to the preset formula ratio, the target weighing data of various raw materials are obtained and sent to the control system of the industrial robot. The actual weight of each raw material is accurately weighed through the weighing sensor of the industrial robot, and the actual weighing data is compared with the target weighing data in real time. According to the deviation between the actual weighing and the target weighing, the control system of the industrial robot automatically adjusts the amount of the corresponding raw materials to be added or reduced until the ratio requirement is met. The weighed raw materials are put into the designated mixing equipment through the industrial robot according to the optimized formula ratio. The mixing equipment is started to fully mix the put raw materials, and the mixing time is automatically controlled according to the preset parameters to ensure the mixing uniformity. The mixed materials are sampled, and the key quality parameters of the mixture are obtained through the online quality detection equipment, and the detection data are transmitted to the formula optimization system in real time. Based on the mixture quality detection results, the formula optimization system uses the least squares support vector machine regression algorithm to establish a mathematical model between the mixture quality parameters and the formula ratio, and dynamically optimizes and adjusts the original formula according to the ratio correction value output by the model to generate a better ratio scheme. The optimized proportioning scheme is sent to the industrial robot control system, and the target weighing data of the raw materials is updated to guide the intelligent proportioning process of the next batch, so as to achieve continuous optimization of the formula and stable control of product quality.

[0019] Specifically, the core of the intelligent batching system lies in accurate weighing and dynamic optimization. Taking the production of degradable plastics as an example, the system first obtains the target weighing data of each raw material according to the preset formula. Assume that a batch requires 70kg of polylactic acid (PLA), 5kg of toughening agent, and 3kg of plasticizer. These data are sent to the industrial robot control system to guide the robot to perform actual weighing. The industrial robot is equipped with a high-precision weighing sensor that can accurately measure the weight of the raw materials. During the weighing process, the system compares the actual weight with the target weight in real time. If the actual weight of PLA is 69.8kg, the system will automatically adjust and add 0.2kg to meet the standard. This real-time correction ensures the accuracy of the raw material ratio and lays the foundation for subsequent processing. After weighing, the industrial robot puts the raw materials into the mixing equipment. The mixing time is preset according to the material properties. For example, the PLA mixture may take 15 minutes to ensure uniformity. During the mixing process, parameters such as temperature and humidity inside the equipment are monitored in real time to prevent material degradation or agglomeration. After mixing, the system performs online quality inspection. Taking the melt index (MFI) as an example, assume that the target MFI is 6g / 10min. If the test result is 5.8g / 10min, which is slightly lower than the target value, this data will be transmitted to the formula optimization system. The formula optimization system uses the least squares support vector machine regression algorithm to establish a relationship model between MFI and raw material ratio. This algorithm can effectively handle nonlinear relationships and is suitable for complex polymer systems. Model analysis may show that a slight increase in the proportion of plasticizer can increase MFI. Based on this, the system generates a new formula: PLA69.8kg, toughener 5kg, plasticizer 3.2kg. The optimized formula is sent to the control system to guide the next batch of production. This closed-loop optimization mechanism can continuously improve product quality stability. For example, after multiple optimizations, the MFI fluctuation range may be reduced from ±0.5g / 10min to ±0.2g / 10min, significantly improving product consistency. In addition, the system also records the impact of environmental factors on the formula. For example, in seasons with high humidity, PLA may need pre-drying treatment, and the system will automatically adjust the drying parameters to ensure the stability of the raw material state. This all-round intelligent control not only improves production efficiency, but also greatly reduces human errors, providing strong support for the large-scale production of high-quality biodegradable plastics.

[0020] S105, infrared thermal imager real-time monitoring of the temperature field distribution of plastic extrusion molding die, combined with finite element simulation analysis, optimize the mold structure design and heating and cooling system to make the temperature field distribution more uniform.

[0021] Infrared thermal imager is used to monitor the temperature field of plastic extrusion molding die in real time to obtain the temperature distribution data of the die surface. The obtained temperature distribution data is converted into a grid data format that can be recognized by finite element simulation software through interpolation algorithm, and then input into finite element simulation analysis software to perform numerical simulation and analysis on the temperature field of the mold. According to the results of finite element simulation analysis, it is judged whether the temperature field distribution of the mold is uniform and the area with uneven temperature field distribution is determined. For the area with uneven temperature field distribution, the mold structure is optimized, such as adjusting the mold wall thickness, optimizing the flow channel layout, etc., to improve the temperature field distribution. According to the optimized mold structure, the electric heating rod and cooling water channel layout of the mold are adjusted and optimized accordingly to achieve accurate control of the mold temperature field. The support vector regression (SVR) algorithm is used to establish the mapping relationship between the mold temperature field distribution and factors such as mold structure parameters, electric heating rod power, and cooling water channel layout. The established SVR model is used to optimize the mold structure and heating and cooling system parameters to obtain an improved design scheme. The optimized mold structure and heating and cooling system parameters are input into the finite element simulation software for simulation verification to ensure that the optimized mold temperature field distribution meets the requirements, thereby achieving the stability of the plastic extrusion molding process and improving product quality.

[0022] Specifically, infrared thermal imagers play a key role in monitoring the temperature field of plastic extrusion molds. This advanced device can capture the thermal radiation from the mold surface and convert it into a high-precision temperature distribution map. For example, in a typical plastic bottle cap mold, the thermal imager may show that the temperature near the injection point is higher (about 180°C), while the temperature at the edge of the mold is lower (about 140°C). This temperature gradient may cause product quality problems such as warping or internal stress. The acquired temperature data needs to be converted into a gridded format through an interpolation algorithm so that it can be input into the finite element analysis software. Commonly used interpolation methods include Kriging and inverse distance weighted method. Taking a 100×100mm mold surface as an example, it may need to be divided into 10,000 1×1mm grid cells, each of which is assigned a temperature value. This fine grid division can accurately reflect the subtle changes in the temperature field. The finite element analysis software uses this data for numerical simulation to reveal the temperature distribution inside the mold. The analysis results may show that there are "hot spots" in some corners, with temperatures 20°C higher than the surrounding area. This non-uniformity may cause uneven plastic flow in the mold, affecting the dimensional accuracy of the product. The mold structure can be optimized for the problems found. For example, cooling water channels can be added to the hot spot area, or the mold wall thickness can be adjusted. In an actual case, the mold wall thickness at the corner was increased from 10mm to 15mm, which effectively reduced the local temperature and controlled the temperature difference with the surrounding area to within 5°C. The heating and cooling system of the mold also needs to be adjusted accordingly. It may be necessary to rearrange the electric heating rods so that they are more evenly distributed on the mold surface. The design of the cooling water channel is also crucial, and a spiral or parallel layout can be used to ensure the consistency of the cooling effect. Through these optimizations, the maximum deviation of the mold surface temperature can be reduced from the original 40°C to within 10°C. The application of the support vector regression (SVR) algorithm makes the optimization of the mold temperature field more accurate. By collecting a large amount of historical data, SVR can establish a nonlinear relationship model between mold structure parameters, heating power, cooling water flow and other factors and temperature distribution. This model can predict, for example, how increasing the heating power by 1kW will affect the temperature of various parts of the mold. Using this model, engineers can quickly evaluate the effects of different optimization schemes without time-consuming actual experiments. Finally, the optimized design needs to be verified by finite element simulation. The simulation results may show that the optimized mold reduces the temperature field fluctuation by 70% and improves the product size consistency by 50% during the injection molding process. This significant improvement not only improves product quality, but also reduces the scrap rate, thereby improving production efficiency and economic benefits. Through this closed-loop optimization process, plastic extrusion molding technology has made a qualitative leap, laying the foundation for the large-scale production of high-precision and high-consistency plastic products.

[0023] S106. Analyze the influence of temperature field distribution on material components, monitor the extrusion molding process in real time with online near-infrared spectrometer, analyze the changes in material components with chemometric model, and adjust the extrusion temperature, pressure, speed and other process parameters or replace the mold when the degradation products exceed the standard.

[0024] According to the material components and process parameters, a finite element model of the temperature field distribution of the extrusion molding process is established to simulate and analyze the influence of the temperature field distribution on the material components. An online near-infrared spectrometer is used to monitor the extrusion molding process in real time, obtain the near-infrared spectrum data of the material components, and transmit the data to a computer for analysis and processing. Using chemometric methods such as partial least squares regression, a quantitative relationship model between the material components and the near-infrared spectrum data is established, and the model is optimized by cross-validation and other methods to achieve rapid prediction of the material components. During the extrusion molding process, near-infrared spectrum data is obtained in real time and input into the chemometric model to predict the content of the material components. The predicted value is compared with the preset qualified range of the material components to determine whether the degradation products exceed the standard. If the predicted material components do not meet the requirements, the optimal combination of process parameters such as extrusion temperature, extrusion pressure, extrusion speed, etc. is found according to the temperature field distribution model through optimization algorithms such as genetic algorithms to control the temperature field distribution and reduce material degradation. If the material composition requirements cannot be met after adjusting the process parameters, the molding process under different mold models and structures is simulated using mold flow analysis software such as Moldflow according to the rheological properties of the material and the molding process requirements, and the optimal mold solution with uniform melt filling, reasonable temperature distribution, and stable pressure distribution is selected to optimize the molding process. Feedback the optimized process parameters and mold information to the production management system, adjust the equipment parameters, update the operation instructions, and guide the operators to make adjustments to form a closed-loop control. At the same time, the predicted values ​​of material components, key process parameters, product quality data, etc. before and after optimization are compared and analyzed, the optimization effect is summarized, and an optimization report is formed to provide reference and guidance for subsequent production optimization to ensure stable product quality and continuous improvement.

[0025] Specifically, in the extrusion process, the influence of temperature field distribution on material composition is crucial. Establishing a finite element model can simulate this complex process and provide a theoretical basis for optimizing process parameters. For example, in polyethylene extrusion, the model may show that the temperature at the front end of the screw reaches 200°C, while it drops to 180°C at the outlet. This temperature gradient may cause local overheating of the material, triggering degradation reactions and producing low molecular weight substances. The application of near-infrared spectroscopy technology makes it possible to monitor material components in real time. In a typical PVC pipe production line, a near-infrared probe installed at the outlet of the extruder can collect spectral data once a second. These data reflect the changes in the content of each component in the material, such as the content of plasticizers and stabilizers. By establishing a partial least squares regression model, complex spectral information can be converted into intuitive component content prediction values. In practical applications, the accuracy of the prediction model is crucial. For example, for a PVC formula containing 10% plasticizer, the model prediction error must be controlled within ±0.5% to meet product quality requirements. To achieve this accuracy, it may be necessary to collect hundreds of samples for modeling and optimize the model parameters through methods such as cross-validation. When the prediction results show that the material composition deviates from the preset range, the system automatically triggers process parameter optimization. Suppose that the hydrogen chloride content in PVC is detected to be excessive, indicating that severe thermal degradation has occurred. At this time, the optimization algorithm may recommend reducing the extrusion temperature from 180°C to 170°C, while increasing the screw speed from 50rpm to 60rpm to reduce the residence time of the material in the high temperature zone. However, process parameter adjustment may not solve all problems. In this case, mold optimization becomes necessary. Through mold flow analysis software, the effect of different mold designs on material flow can be simulated. For example, changing the flow channel cross-section of the mold from circular to elliptical may make the melt shear stress distribution more uniform, thereby reducing local overheating. The optimized process parameters and mold design need to be verified in actual production. For example, before optimization, the plasticizer content in the product fluctuated between 9.5% and 10.5%; after optimization, the fluctuation range was reduced to 9.8% to 10.2%, significantly improving product consistency. This improvement not only improves product quality, but also reduces raw material waste and reduces production costs. By establishing such a closed-loop control system, companies can achieve precise control and continuous optimization of the extrusion molding process. This not only improves product quality stability, but also provides valuable data support for new product development and accelerates innovation. In the long run, this data-driven optimization method will become a key means for the plastics processing industry to enhance its competitiveness.

[0026] S107. Collect data from the entire production process, use digital twin technology to build a virtual model of the production line, and use simulation optimization to find the best combination of process parameters, including raw material ratio, pretreatment parameters, extrusion parameters, etc., to guide actual production and improve product quality stability.

[0027] According to the real-time data of each process of the production line, including raw material ratio, pretreatment parameters, extrusion parameters, etc., the digital twin model of the production line is constructed using SiemensNX software. The production data is obtained in real time through the OPCUA protocol, and the digital twin model is updated in real time using the CAE simulation function of SiemensNX to ensure that the model is synchronized with the actual production line status. In the updated digital twin model, different combinations of process parameters are set, including raw material ratio, pretreatment temperature, extrusion speed, etc., and the CAE simulation function of SiemensNX is used for simulation optimization to evaluate the product quality and production efficiency under each parameter combination. The best process parameter combination is determined by analyzing the simulation results. The best process parameter combination obtained by optimization is applied to the actual production line. By monitoring the production data in real time, statistical methods such as quality control charts are used to determine whether the product quality meets expectations. If the product quality does not meet expectations, the digital twin model is returned, and the process parameters are adjusted according to the quality deviation using design optimization algorithms such as genetic algorithms, and simulation optimization is performed again until the best parameter combination that meets the quality requirements is found. Continuously collect production process data, including raw material properties, process parameters, product quality data, etc., use machine learning algorithms such as support vector machines (SVM), etc., to establish a prediction model between process parameters and product quality, and regularly (such as weekly) use newly collected data to update and optimize the model. Based on the optimized prediction model, adjust the process parameters in the digital twin model, continuously improve the production process, and improve product quality stability and production efficiency.

[0028] Specifically, the application of digital twin technology in plastic extrusion production lines provides a powerful tool for optimizing production processes. The digital twin model built by SiemensNX software can reflect the status of the production line in real time and provide a reliable basis for optimizing process parameters. For example, on a production line with an annual output of 5,000 tons of PVC pipes, the digital twin model can accurately simulate the entire process from raw material ratio to finished product discharge. Through the OPCUA protocol, the model can collect more than 1,000 data points from the production line per second, including key parameters such as screw temperature, pressure, and speed. Using the CAE simulation function of SiemensNX, engineers can test different combinations of process parameters in a virtual environment. For example, by adjusting the preheating zone temperature from 140°C to 160°C and the extrusion speed from 20m / min to 30m / min, the effects of these changes on product dimensional stability and surface quality can be observed. The simulation results show that when the preheating zone temperature is set to 155°C and the extrusion speed is 25m / min, the production efficiency can be increased by 15% while ensuring product quality. Applying the optimized parameters to actual production, quality control personnel can use XR control charts to monitor key quality indicators. Assuming that the outer diameter tolerance requirement of the pipe is ±0.1mm, the sample data before optimization shows that the outer diameter variation range is ±0.15mm. After applying the optimization parameters, the outer diameter variation range is reduced to ±0.08mm, which significantly improves product consistency. If the actual production results deviate from the expected, such as insufficient uniformity of the pipe wall thickness, engineers can return to the digital twin model for further optimization. Through genetic algorithms, hundreds of parameter combinations can be automatically generated and evaluated in the model. For example, the algorithm may recommend increasing the mold temperature from 180°C to 185°C, while fine-tuning the temperature distribution of each section of the screw to improve melt fluidity and uniformity. Continuously collected production data provides the basis for establishing a predictive model. Using the support vector machine (SVM) algorithm, a nonlinear relationship model between process parameters and product quality can be established. For example, by analyzing 100,000 sets of data collected in 3 months, the model can predict product performance under given process parameters, such as impact strength, heat deformation temperature, etc. This predictive capability allows engineers to evaluate the feasibility of new formulas or processes before actual production, greatly shortening the product development cycle. By continuously optimizing digital twin models and prediction algorithms, companies can achieve continuous improvement in production processes. In one actual case, a PVC pipe manufacturer used this method to increase its product qualification rate from 98% to 99.5% within 6 months, while reducing energy consumption by 8% and increasing raw material utilization by 3%. This not only improves the company's economic benefits, but also reduces environmental impact, reflecting the profound impact of digital transformation on the manufacturing industry.

[0029] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the concept of the present application. For example, the above features are replaced with the technical features with similar functions disclosed in the present application (but not limited to) to form a technical solution.

Claims

1. A method for producing degradable plastics, characterized in that: include: Obtain the performance requirements of biodegradable plastic products, establish a raw material formula database, use machine learning algorithms to analyze historical production data, and obtain the optimal raw material ratio solution; Obtain the optimal raw material ratio solution and control the intelligent batching system for automatic batching; Obtain temperature and humidity parameters of raw material storage and transportation, and dynamically adjust raw material pretreatment process parameters based on time series data analysis; Conduct quality inspection on pre-treated raw materials, use machine vision and deep learning algorithms to identify raw material defects and impurities, and remove unqualified raw materials; Industrial robots are used to accurately weigh various raw materials and mix them according to the optimized formula ratio; Use infrared thermal imager to monitor the temperature field distribution of plastic extrusion mold in real time, and combine finite element simulation analysis to optimize the mold structure design and heating and cooling system; Online monitoring of the extrusion molding process, using chemometric models to analyze changes in material components, and adjusting extrusion process parameters or replacing molds if degradation products exceed the standard; Collect data from the entire production process, use digital twin technology to build a virtual model of the production line, find the best combination of process parameters through simulation optimization, and guide actual production.

2. The method for producing a degradable plastic according to claim 1, characterized in that: The method of obtaining the performance requirements of the degradable plastic products, establishing a raw material formula database, and using a machine learning algorithm to analyze historical production data to obtain the optimal raw material ratio scheme includes: Establish a raw material formula database to record the performance parameters and ratio information of various raw materials; Collect historical production data, including raw material types, mixing ratios, process parameters and product performance; Use data preprocessing technology to clean and standardize the raw material formula database and historical production data to remove outliers and missing values; Extract key features from the raw material formula database and historical production data through feature engineering methods; Select machine learning algorithms and build prediction models to predict product performance with different raw material ratios; Genetic algorithms are used to search for the best raw material ratio in the raw material formula database. Through crossover and mutation operations, the ratio is optimized until the ratio with the best predicted product performance is found.

3. The method for producing a degradable plastic according to claim 1, characterized in that: The method of obtaining the temperature and humidity parameters of the raw material storage and transportation links, combining the time series data analysis, and dynamically adjusting the raw material pretreatment process parameters includes: Deploy temperature and humidity sensors to collect temperature and humidity parameters of the environment where the raw materials are located in real time; The collected raw material temperature and humidity time series data are transmitted to the data analysis platform, and the data are cleaned, denoised and normalized; Using feature engineering methods, key features are extracted from historical raw material temperature and humidity data, including statistics, temperature and humidity change trends, and fluctuation cycle time series characteristics; Use time series analysis algorithm to establish a prediction model for temperature and humidity parameters and raw material quality; According to the established prediction model, the predicted value of raw material quality under the current raw material temperature and humidity parameters is calculated and compared with the set quality threshold; If the prediction quality is lower than the threshold, an early warning is triggered; When the predicted quality deviates from the optimal range, the optimization algorithm is used to search for the best combination of raw material pretreatment process parameters so that the predicted quality reaches the target value.

4. The method for producing a degradable plastic according to claim 1, characterized in that: The quality inspection of the pre-treated raw materials, using machine vision and deep learning algorithms to identify raw material defects and impurities, and removing unqualified raw materials, includes: Obtain preprocessed raw material image data, input the image data into a pre-built convolutional neural network model for feature extraction and classification recognition, and obtain the location coordinate information of raw material defects and impurities; According to the position coordinates of defects and impurities, the image segmentation algorithm is used to segment the foreground and background of the raw material image to obtain the binary segmentation image of the raw material defect area and impurity area; By counting the white pixels in the segmented image, the defect area ratio and impurity area ratio of the raw material are calculated; Compare the percentage value with the preset threshold value. If the defect area percentage or impurity area percentage exceeds the threshold value, the raw material is judged as unqualified. Control the robot or pneumatic device to remove defective products from the production line; For qualified raw materials that are not rejected, their image features are extracted and saved, and an image feature database of qualified raw materials is established.

5. The method for producing a degradable plastic according to claim 1, characterized in that: The industrial robot accurately weighs various raw materials and mixes them according to the optimized formula ratio, including: According to the preset formula ratio, the target weighing data of various raw materials are obtained and sent to the control system of the industrial robot; The weighing sensor of the industrial robot can accurately weigh the actual weight of each raw material, and compare the actual weighing data with the target weighing data in real time; According to the deviation between the actual weighing and the target weighing, the control system of the industrial robot automatically adjusts the amount of corresponding raw materials to be added or reduced until the required ratio is met; The weighed raw materials are placed into the designated mixing equipment by industrial robots according to the optimized formula ratio; Start the mixing equipment to fully mix the raw materials, and the mixing time is automatically controlled according to the preset parameters; The mixed materials are sampled, the key quality parameters of the mixture are obtained through online quality testing equipment, and the test data are transmitted to the formula optimization system in real time.

6. The method for producing a degradable plastic according to claim 1, characterized in that: The method uses an infrared thermal imager to monitor the temperature field distribution of the plastic extrusion molding die in real time, combines finite element simulation analysis, and optimizes the die structure design and heating and cooling system, including: Use infrared thermal imager to monitor the temperature field of plastic extrusion mold in real time and obtain the temperature distribution data on the mold surface; The acquired temperature distribution data is converted into a grid data format that can be recognized by the finite element simulation software through an interpolation algorithm, and then input into the finite element simulation analysis software to perform numerical simulation and analysis on the temperature field of the mold; According to the results of finite element simulation analysis, determine whether the mold temperature field distribution is uniform and determine the area where the temperature field distribution is uneven; For areas with uneven temperature field distribution, optimize the mold structure, adjust the mold wall thickness, and optimize the flow channel layout to improve the temperature field distribution; According to the optimized mold structure, the electric heating rod and cooling water channel layout of the mold are adjusted and optimized accordingly to achieve precise control of the mold temperature field.

7. The method for producing a degradable plastic according to claim 1, characterized in that: The online monitoring of the extrusion molding process uses a chemometric model to analyze changes in material components, and if the degradation product exceeds the standard, the extrusion process parameters are adjusted or the mold is replaced, including: According to the material composition and process parameters, a finite element model of the temperature field distribution in the extrusion molding process is established to simulate and analyze the influence of the temperature field distribution on the material composition; An online near-infrared spectrometer is used to monitor the extrusion molding process in real time, obtain near-infrared spectrum data of material components, and transmit the data to a computer for analysis and processing; Using chemometric methods, a quantitative relationship model between material components and near-infrared spectral data is established, and the model is optimized through cross-validation methods to achieve rapid prediction of material components; During the extrusion process, near-infrared spectroscopy data is acquired in real time and input into the chemometric model to predict the material component content; Compare the predicted value with the preset qualified range of material components to determine whether the degradation products exceed the standard; If the predicted material composition does not meet the requirements, the optimal combination of process parameters such as extrusion temperature, extrusion pressure, extrusion speed, etc. is found through optimization algorithm according to the temperature field distribution model to control the temperature field distribution and reduce material degradation.

8. The method for producing a degradable plastic according to claim 1, characterized in that: The whole production process data is collected, and a virtual model of the production line is constructed using digital twin technology. The best combination of process parameters is found through simulation optimization to guide actual production, including: Based on the real-time data of each process of the production line, including raw material ratio, pretreatment parameters, extrusion parameters, etc., a digital twin model of the production line is constructed using relevant software; Acquire production data in real time through relevant protocols, and use the simulation function of relevant software to update the digital twin model in real time to ensure that the model is synchronized with the actual production line status; In the updated digital twin model, different combinations of process parameters are set, and simulation optimization is performed using the simulation function of relevant software to evaluate product quality and production efficiency under each parameter combination; Determine the best process parameter combination through simulation result analysis; Apply the best process parameter combination obtained through optimization to the actual production line, monitor production data in real time, and use statistical methods such as quality control charts to determine whether product quality meets expectations; If the product quality does not meet expectations, the digital twin model is returned, and the process parameters are adjusted using the design optimization algorithm based on the quality deviation, and the simulation optimization is performed again until the best parameter combination that meets the quality requirements is found.

Images