PMMA polymerization reaction viscosity real-time monitoring system and method based on multi-mode sensing and edge intelligent calculation
By combining multimodal sensing with edge intelligent computing, the problems of lag and anti-interference in PMMA polymerization reaction monitoring were solved, achieving high-precision and rapid reaction process control, and improving production efficiency and product quality.
Patent Information
- Application Number
- CN202510882011.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-27
- Publication Date
- 2025-10-31
AI Technical Summary
Existing methods for monitoring the viscosity of PMMA polymerization reactions suffer from severe hysteresis and insufficient anti-interference capabilities, resulting in large measurement errors, an inability to gain a deep understanding of the reaction process, and impacts production efficiency and product quality.
By combining a multimodal sensor array (resonant viscometer, multispectral imaging module, vibration sensor, and distributed temperature and pressure sensor) with edge intelligent computing, and through spatiotemporal data alignment and a lightweight CNN prediction model, real-time monitoring and precise control are achieved.
It significantly improves viscosity measurement accuracy from ±15% to ±3%, increases response speed by 50 times, adapts to different reaction conditions, extends sensor life, reduces maintenance costs, and promptly identifies anomalies and avoids production accidents.
Abstract
Description
Technical Field
[0001] This invention relates to the field of PMMA polymerization reaction viscosity monitoring technology, specifically to a real-time monitoring system and method for PMMA polymerization reaction viscosity based on multimodal sensing and edge intelligent computing. Background Technology
[0002] In existing technologies, viscosity monitoring during PMMA polymerization mainly relies on offline sampling or single-modal online sensors (such as rotational viscometers), which have significant drawbacks:
[0003] Firstly, offline detection cannot reflect the reaction process in real time, resulting in a lag in the adjustment of process parameters, which seriously affects production efficiency and product quality.
[0004] Secondly, single sensors are easily affected by factors such as temperature and pressure fluctuations, resulting in large measurement errors, typically exceeding ±15%, which makes it difficult to meet the requirements of high-precision production.
[0005] Third, traditional methods have not established a dynamic correlation model between molecular structure evolution (such as branching degree and molecular weight distribution) and viscosity, which makes it impossible to deeply understand the reaction process and limits the precise control of polymerization reactions.
[0006] Therefore, the present invention aims to solve the problems of serious lag and insufficient anti-interference ability of traditional monitoring methods. Summary of the Invention
[0007] The purpose of this invention is to provide a real-time monitoring system and method for the viscosity of PMMA polymerization reaction based on multimodal sensing and edge intelligent computing, so as to solve the problems of serious lag and insufficient anti-interference ability of traditional monitoring methods mentioned in the background art.
[0008] To achieve the above objectives, the present invention provides the following technical solution: a real-time monitoring system for the viscosity of PMMA polymerization reaction based on multimodal sensing and edge intelligent computing, comprising:
[0009] Multimodal sensor array: includes a resonant viscometer, a multispectral imaging module, a vibration sensor, and a distributed temperature and pressure sensor, used for accurate monitoring of the viscosity of PMMA polymerization reaction;
[0010] Edge computing unit: integrates a spatiotemporal data alignment module with a lightweight CNN prediction model;
[0011] Intelligent feedback actuator: includes an adaptive compensation controller and a visual human-machine interface; wherein, the compensation controller generates temperature and pressure coupling adjustment commands based on reaction kinetic equations; the human-machine interface displays molecular structure evolution maps and three-dimensional viscosity field distribution in real time.
[0012] Furthermore, the resonant viscometer incorporates an MXene / PMMA composite electrode with a biomimetic micropillar structure on its surface.
[0013] Furthermore, the multispectral imaging module is equipped with a 405nm-1550nm dual-band light source and a high-resolution CCD array.
[0014] Furthermore, the spatiotemporal data alignment module employs an improved particle filter algorithm to achieve a multi-source data synchronization error of ≤0.5ms.
[0015] Furthermore, the CNN prediction model is compressed to less than 5MB using knowledge distillation techniques.
[0016] Furthermore, the distributed temperature and pressure sensors are arranged at equal intervals on the inner wall of the reactor to collect temperature and pressure data of the reaction system in real time.
[0017] Furthermore, the MXene / PMMA composite electrode of the resonant viscometer enhances sensing sensitivity by increasing charge trapping capability.
[0018] Furthermore, the method for real-time monitoring of PMMA polymerization reaction viscosity based on multimodal sensing and edge intelligent computing includes the following steps:
[0019] The turbidity changes in the reaction system were captured by a multispectral imaging module, and molecular chain entanglement features were extracted by combining a Gabor filter.
[0020] The vibration sensor is used to collect the fluid shear stress signal, and the high-frequency noise component is separated by Fourier transform;
[0021] A multimodal feature fusion matrix is constructed, and an attention mechanism is used to dynamically allocate the weights of each sensor.
[0022] A viscosity prediction model is trained based on a transfer learning framework, and predicted values are generated by inputting historical data and real-time sensing data.
[0023] When the monitored viscosity deviates from the set threshold, the reactor temperature control system is triggered to perform PID regulation.
[0024] This invention provides a real-time viscosity monitoring system and method for PMMA polymerization reaction based on multimodal sensing and edge intelligent computing, which has the following advantages: Through cross-validation of multimodal data, the viscosity measurement error is reduced from ±15% in traditional methods to ±3%, significantly improving measurement accuracy; the edge computing unit achieves a 10ms-level response speed, 50 times faster than cloud-based solutions, enabling timely reflection of changes in the reaction process and providing the possibility for real-time control; through transfer learning technology, the system can adapt to different reaction conditions, such as bulk polymerization and suspension polymerization, improving the system's versatility and flexibility; the application of MXene composite electrodes extends the sensor lifespan to over 2000 high-temperature reaction cycles, reducing equipment maintenance costs and improving system reliability. Detailed Implementation
[0025] This invention provides a technical solution: a real-time monitoring system for the viscosity of PMMA polymerization reaction based on multimodal sensing and edge intelligent computing, comprising:
[0026] Multimodal sensing array: Includes a resonant viscometer, a multispectral imaging module, a vibration sensor, and distributed temperature and pressure sensors, used for precise monitoring of the viscosity of PMMA polymerization reaction. The resonant viscometer has a built-in MXene / PMMA composite electrode with a biomimetic micropillar structure on the electrode surface. The multispectral imaging module is equipped with a 405nm-1550nm dual-band light source and a high-resolution CCD array. The distributed temperature and pressure sensors are equidistantly arranged on the inner wall of the reactor to collect temperature and pressure data of the reaction system in real time. The MXene / PMMA composite electrode of the resonant viscometer improves sensing sensitivity by enhancing charge trapping capability.
[0027] Edge computing unit: integrates a spatiotemporal data alignment module and a lightweight CNN prediction model. The spatiotemporal data alignment module adopts an improved particle filter algorithm to achieve a multi-source data synchronization error of ≤0.5ms. The CNN prediction model is compressed to less than 5MB through knowledge distillation technology.
[0028] The intelligent feedback actuator comprises an adaptive compensation controller and a visual human-machine interface. The compensation controller generates temperature and pressure coupling adjustment commands based on reaction kinetic equations. The human-machine interface displays molecular structure evolution maps and three-dimensional viscosity field distributions in real time. Furthermore, the spatiotemporal data alignment module employs an improved particle filtering algorithm to achieve a multi-source data synchronization error ≤0.5ms.
[0029] A method for real-time monitoring of the viscosity of PMMA polymerization reaction based on multimodal sensing and edge intelligent computing includes the following steps:
[0030] The turbidity changes in the reaction system were captured by a multispectral imaging module, and molecular chain entanglement features were extracted by combining a Gabor filter.
[0031] The vibration sensor is used to collect the fluid shear stress signal, and the high-frequency noise component is separated by Fourier transform;
[0032] A multimodal feature fusion matrix is constructed, and an attention mechanism is used to dynamically allocate the weights of each sensor.
[0033] A viscosity prediction model is trained based on a transfer learning framework, and predicted values are generated by inputting historical data and real-time sensing data.
[0034] When the monitored viscosity deviates from the set threshold, the reactor temperature control system is triggered to perform PID regulation.
[0035] Six temperature / pressure sensors are equidistantly arranged on the inner wall of a 50L reactor to comprehensively monitor the temperature and pressure distribution within the reactor. A multispectral imaging module is installed on the top of the reactor to capture optical information such as turbidity changes in the reaction system. A resonant viscometer is positioned at the discharge port with a sampling frequency of 100Hz to acquire viscosity data at high frequency. The edge computing unit loads a pre-trained model, with initial training data including PMMA samples of five molecular weight distributions. After the reaction starts, the system generates a viscosity-molecular weight correlation thermogram in real time. When the monitored value exceeds a threshold, the initiator feed rate is automatically adjusted to achieve real-time control of the reaction process, ensuring that the reaction proceeds within the set viscosity range.
[0036] Interference factors, such as excessive initiator, are injected during the mid-reaction phase to simulate abnormal conditions in the production process. Through vibration spectrum analysis, the system can identify abnormalities and trigger alarms within 30 seconds, while traditional methods typically require 5 minutes to detect anomalies. This invention's monitoring system achieves this 5-minute lead time, gaining valuable time for timely handling of abnormal situations and effectively preventing product quality problems and production accidents caused by these abnormalities.
[0037] In summary, the PMMA polymerization reaction viscosity real-time monitoring system and method based on multimodal sensing and edge intelligent computing provided by this invention significantly improves the accuracy and real-time performance of PMMA polymerization reaction viscosity monitoring through the innovative application of technologies such as multimodal sensing fusion and edge intelligent computing. It also enhances the system's adaptability to different reaction conditions and provides strong technical support for process quality control in fine chemical production.
[0038] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only for the purpose of helping to understand the method and core ideas of the present invention. The above descriptions are only preferred embodiments of the present invention. It should be noted that due to the limitations of textual expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of the present invention.
Claims
1. A real-time monitoring system for the viscosity of PMMA polymerization reaction based on multimodal sensing and edge intelligent computing, characterized in that, include: Multimodal sensor array: includes a resonant viscometer, a multispectral imaging module, a vibration sensor, and a distributed temperature and pressure sensor, used for accurate monitoring of the viscosity of PMMA polymerization reaction; Edge computing unit: integrates a spatiotemporal data alignment module with a lightweight CNN prediction model; Intelligent feedback actuator: includes an adaptive compensation controller and a visual human-machine interface; wherein, the compensation controller generates temperature and pressure coupling adjustment commands based on reaction kinetic equations; the human-machine interface displays molecular structure evolution maps and three-dimensional viscosity field distribution in real time.
2. The PMMA polymerization reaction viscosity real-time monitoring system based on multimodal sensing and edge intelligent computing according to claim 1, characterized in that, The resonant viscometer has a built-in MXene / PMMA composite electrode with a biomimetic micropillar structure on the electrode surface.
3. The PMMA polymerization reaction viscosity real-time monitoring system based on multimodal sensing and edge intelligent computing according to claim 2, characterized in that, The multispectral imaging module is equipped with a dual-band light source of 405nm-1550nm and a high-resolution CCD array.
4. The PMMA polymerization reaction viscosity real-time monitoring system based on multimodal sensing and edge intelligent computing according to claim 3, characterized in that, The spatiotemporal data alignment module employs an improved particle filter algorithm to achieve a multi-source data synchronization error of ≤0.5ms.
5. The PMMA polymerization reaction viscosity real-time monitoring system based on multimodal sensing and edge intelligent computing according to claim 4, characterized in that, The CNN prediction model is compressed to less than 5MB using knowledge distillation techniques.
6. The PMMA polymerization reaction viscosity real-time monitoring system based on multimodal sensing and edge intelligent computing according to claim 5, characterized in that, The distributed temperature and pressure sensors are arranged at equal intervals on the inner wall of the reactor to collect temperature and pressure data of the reaction system in real time.
7. A real-time monitoring system for the viscosity of PMMA polymerization reaction based on multimodal sensing and edge intelligent computing according to claim 6, characterized in that, The MXene / PMMA composite electrode of the resonant viscometer enhances sensing sensitivity by increasing charge trapping capability.
8. A method for real-time monitoring of PMMA polymerization reaction viscosity based on multimodal sensing and edge intelligent computing as described in claim 7, characterized in that, Includes the following steps: The turbidity changes in the reaction system were captured by a multispectral imaging module, and molecular chain entanglement features were extracted by combining a Gabor filter. The vibration sensor is used to collect the fluid shear stress signal, and the high-frequency noise component is separated by Fourier transform; A multimodal feature fusion matrix is constructed, and an attention mechanism is used to dynamically allocate the weights of each sensor. A viscosity prediction model is trained based on a transfer learning framework, and predicted values are generated by inputting historical data and real-time sensing data. When the monitored viscosity deviates from the set threshold, the reactor temperature control system is triggered to perform PID regulation.