Page 52 - 《橡塑技术与装备》英文版2026年3期

HINA R&P TECHNOLOGY AND EQUIPMENT

subsequent processing flows and logistics can be planned more downtime and maintenance costs. Equipment operation data
effectively. By integrating learning methods (such as random is collected through sensors, and machine learning techniques
forests or gradient boosting trees) to analyze various by- such as time series analysis or anomaly detection algorithms
products (such as fuel oil, carbon black, metals, etc.) during are employed to predict potential equipment failures, enabling
the pyrolysis process, their yield and quality can be predicted. preventive maintenance of the equipment. This not only
Research institutions in Japan are developing AI models to reduces the duration of unplanned downtime but also lowers
predict the characteristics of different by-products generated maintenance costs.
during the pyrolysis process, including the purity of oil This section explains how AI can predict potential
products and the quality of carbon black. This helps to enhance equipment failures, schedule maintenance in advance, reduce
the value of by-products and guide subsequent processing downtime, and enhance the operational efficiency of factories
flows. through data collected by Internet of Things (IoT) sensors.
This case study demonstrates the significant role of AI 4.7 Case and analysis of resource recovery
in predicting and classifying by-products generated during the rate prediction
pyrolysis process. By utilizing deep learning models to predict Based on historical data, regression models are used to
the type and quality of these by-products, more efficient and predict the resource recovery rates of different types of waste
environmentally friendly recycling and utilization can be plastics under specific processes, helping decision-makers
achieved, such as reusing carbon black as fuel or material. optimize recycling procedures and resource allocation, and
4.5 Cases and analysis of quality control and guiding production planning. Researchers at the University of
real-time monitoring Tokyo are developing a technology that utilizes AI for waste
Machine learning models can monitor and predict plastic classification and pyrolysis process optimization. By
changes in product quality during the pyrolysis process, such analyzing vast amounts of data, the AI system can predict
as the purity of oil products and the particle size distribution optimal pyrolysis conditions, thereby improving resource
of carbon black, thereby enabling online quality control and recovery rates and reducing energy consumption. Some
avoiding the production of unqualified products. Computer research institutions and companies in the United States
vision and deep learning models, such as convolutional neural are exploring the application of AI in the pyrolysis process,
networks (CNN), are used to monitor changes in product analyzing real-time data to optimize key parameters such as
quality in real-time during the pyrolysis process. Through temperature, pressure, and residence time, in order to increase
image analysis, the system can quickly identify and predict the oil yield and reduce the generation of harmful by-products.
characteristics of by-products, such as the purity of fuel oil and Some projects in France are studying the application of AI
the particle size distribution of carbon black, thus achieving in waste management, including waste plastic classification,
online quality control. Some industrial projects in the United pretreatment, and pyrolysis process optimization, to achieve
States utilize AI technology to monitor the pyrolysis process in efficient resource recovery and waste treatment.
real-time, analyzing sensor data to predict and adjust process This section explains how AI can be used to conduct
parameters. more precise classification and pretreatment of different types
This case study demonstrates the implementation of AI of waste plastics, remove impurities and enhance the purity of
quality control and real-time monitoring to achieve optimal plastics, thereby improving the recovery rate of oil and other
resource recovery efficiency and energy utilization. valuable by-product resources during the pyrolysis process and
4.6 Case and analysis of equipment fault increasing the added value of the products.
prediction 4.8 Case studies and analysis of environme
By monitoring equipment operation data and utilizing ntal impact assessments
machine learning models to predict potential equipment Utilize machine learning models to assess the
failures, maintenance can be carried out in advance, reducing environmental impacts of various pyrolysis processes,

·6· Vol.52，No.3