Ultrasonic transmission-based online viscosity monitoring method for a regenerative polyester homogenizer

Abstract

The application relates to the field of polymer material processing and process detection technology, and discloses a regenerated polyester homogenizer online viscosity monitoring method based on ultrasonic transmission, which realizes non-contact, real-time and accurate monitoring and active regulation of the viscosity of a regenerated polyester melt and significantly improves product batch consistency and production intelligent level by constructing a multi-channel ultrasonic transmission sensor array in the homogenizer, obtaining melt acoustic propagation characteristics in real time, dynamically analyzing intrinsic viscosity based on a sound velocity-attenuation joint inversion model, and classifying and regulating process parameters according to deviation size and duration by closed-loop linkage of viscosity data and stirring, heating and other execution units.

Description

Technical Field

[0001] This invention belongs to the field of polymer material processing and process detection technology, specifically relating to an online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission. Background Technology

[0002] With the large-scale application of recycled polyester in textiles, packaging, and engineering plastics, melt viscosity control during its production process has become a crucial factor determining product performance consistency. Viscosity, as a core indicator reflecting polymer molecular weight distribution and chain structure, directly affects spinning processability, mechanical strength, and thermal stability. In the homogenization reactor process of recycled polyester, the melt needs to achieve component homogenization and molecular chain reconstruction under high temperature, high vacuum, and strong shear conditions, and viscosity is a comprehensive characterization of this dynamic chemical-physical coupling process. Therefore, accurate and real-time monitoring and control of viscosity is a fundamental prerequisite for ensuring the quality stability and process economy of recycled polyester.

[0003] Online viscosity monitoring technology based on the principle of ultrasonic transmission has gradually become an important means of detecting the rheological properties of melts due to its advantages such as non-contact operation, fast response, and high temperature resistance. This method inverts the complex viscosity of the medium by measuring the attenuation and velocity changes of ultrasonic waves propagating in polyester melt, avoiding the drawbacks of traditional rotational viscometers such as easy wear, clogging, and severe hysteresis. However, current ultrasonic monitoring systems are mostly limited to data acquisition and display functions and have not yet achieved deep integration with production processes, making it difficult to translate monitoring results into effective control commands.

[0004] The following problems exist between online viscosity monitoring and process control: operators rely on historical experience or offline laboratory data to manually adjust parameters such as stirring speed and heating power, which cannot achieve dynamic response; at the same time, the response of polyester melt viscosity to process disturbances has obvious inertia. By the time ultrasonic detection shows that the viscosity deviates from the target value, the reaction has already entered an irreversible stage, resulting in large fluctuations in intrinsic viscosity between batches and a decrease in the pass rate; in addition, the homogenization process generally adopts a fixed time sequence or open-loop control strategy, ignoring the impact of raw material batch differences, thermal history accumulation, and flow field non-uniformity on the homogenization effect, making it difficult to guarantee the consistency of the microstructure of each batch of products. Summary of the Invention

[0005] This invention provides an online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission, aiming to solve the technical problems of disconnect between monitoring and control, lag in process response, and uncontrollable homogenization effect in existing technologies. The method constructs a multi-channel ultrasonic transmission sensor array inside the homogenization reactor to acquire the acoustic propagation characteristics of the recycled polyester melt in real time at different spatial locations and time dimensions, and dynamically analyzes the characteristic viscosity of the melt based on a sound velocity-attenuation joint inversion model. Simultaneously, the obtained viscosity data is linked in a closed loop with the process execution unit of the homogenization reactor. Based on a preset viscosity target range and dynamic tolerance band, control commands for stirring rate, heating power, and material residence time are automatically generated, thereby achieving integrated control of real-time perception, accurate judgment, and proactive intervention in the homogenization process of the recycled polyester melt.

[0006] As one embodiment of the present invention, the online viscosity monitoring method for recycled polyester homogenizing reactor based on ultrasonic transmission includes the following steps: At least four ultrasonic transmitting transducers are uniformly arranged circumferentially on the wall of the homogenizing reactor, and the same number of ultrasonic receiving transducers are correspondingly arranged on the inner wall, forming a multi-path cross-transmission sensing network; the ultrasonic transmitting transducers transmit ultrasonic signals with a center frequency of 1.5 MHz and a pulse width of 20 microseconds in a pulse modulation manner, which are captured by the receiving transducers after penetrating the recycled polyester melt; the ultrasonic signals received in each transmission path are subjected to time-domain envelope extraction and frequency-domain spectrum analysis, and the ultrasonic propagation time and energy attenuation coefficient under that path are calculated respectively; based on the propagation time data of all transmission paths, the least squares method is used to fit the spatial distribution field of sound velocity inside the melt; simultaneously, based on the energy attenuation coefficients of all paths, a sensor network is constructed within the melt. The attenuation coefficient spatial distribution field of the part is calculated; the sound velocity spatial distribution field and the attenuation coefficient spatial distribution field are input into a pre-trained acoustic-rheological coupling inversion model to output a three-dimensional distribution map of the intrinsic viscosity of the recycled polyester melt at the current moment; the three-dimensional distribution map of intrinsic viscosity is volume-weighted and averaged to obtain a global intrinsic viscosity value characterizing the homogenization state of the whole melt; the global intrinsic viscosity value is compared with a preset target viscosity value. If the absolute value of the deviation is greater than a preset first threshold, the first-level control strategy is activated to adjust the rotation speed of the stirring paddle at the bottom of the homogenization tank; if the absolute value of the deviation is greater than a preset second threshold and the duration exceeds 30 seconds, the second-level control strategy is activated to adjust the heat medium flow rate of the jacket heating system while adjusting the stirring speed; if the absolute value of the deviation is within the ±0.5% tolerance band of the target viscosity value for 60 consecutive seconds, it is determined that the current batch has reached the homogenization endpoint and the discharge control signal is triggered.

[0007] Furthermore, the construction process of the acoustic-rheological coupled inversion model includes: collecting melt samples from no fewer than 500 batches of recycled polyester production under different temperature, pressure, molecular weight distribution, and impurity content conditions; measuring the intrinsic viscosity of each batch of samples under standard laboratory conditions, and simultaneously recording the sound velocity and attenuation data obtained by an ultrasonic transmission sensor network under the same operating conditions; using the mean sound velocity, standard deviation of sound velocity, mean attenuation, and standard deviation of attenuation as input feature vectors, and the intrinsic viscosity measured in the laboratory as the output label, training the nonlinear mapping relationship using a support vector regression algorithm; the support vector regression algorithm uses a radial basis function kernel, with a penalty factor C of 10 and a kernel function parameter γ of 0.1; after cross-validation, the root mean square error of the model prediction is less than 0.005 dL / g.

[0008] Furthermore, the ultrasonic transmitting and receiving transducers in the multipath cross-transmission sensor network are both made of high-temperature resistant piezoelectric ceramic material with a Curie temperature of not less than 350 degrees Celsius. The encapsulation shell is made of Hastelloy C276 material, with an operating temperature range of 280 degrees Celsius to 320 degrees Celsius. The transducers are sealed to the homogenizing vessel body via flanges, and the connection is equipped with double-layer metal spiral wound gaskets to ensure airtightness under vacuum or slightly positive pressure conditions. All electrical leads of the transducers are led out through high-temperature ceramic insulating sleeves and connected to the signal conditioning module located in the explosion-proof junction box.

[0009] Furthermore, the signal conditioning module performs preamplification, bandpass filtering, and analog-to-digital conversion on the original received signal; the preamplifier has a gain of 40 dB, the bandpass filter has a passband range of 0.8 MHz to 2.2 MHz, the analog-to-digital converter has a sampling frequency of 20 MHz and a quantization bit depth of 16 bits; the digital signal after analog-to-digital conversion is transmitted to the central processing unit, which performs time-domain envelope extraction and frequency-domain spectrum analysis.

[0010] Furthermore, the time-domain envelope extraction employs the Hilbert transform method to construct the analytic signal for each received signal sequence x(n). Where j is the imaginary unit, H[x(n)] represents the Hilbert transform operator, and the envelope signal e(n) is defined as |z(n)|; the ultrasonic propagation time is determined by detecting the initial zero-crossing point of the envelope signal e(n), specifically by finding the time point at which the amplitude of the envelope signal first exceeds three standard deviations of the noise floor; the energy attenuation coefficient α is obtained by formula The calculation is performed, where L is the transmission path length, P0 is the received signal energy along the same path in the reference state, and P is the current received signal energy.

[0011] Furthermore, in the first-level control strategy, the adjustment amount ΔN of the impeller speed is given by the formula... Determine, where dt is the time differential, η target η is the target intrinsic viscosity value. current K represents the current global intrinsic viscosity value. p This is a proportionality coefficient, with a value of 15 revolutions per minute per minute per liter per gram, K. i The integral coefficient is 0.5 revolutions per minute per minute per liter per gram per second.

[0012] Furthermore, in the second-level control strategy, the adjustment of the heat transfer medium flow rate is based on the Arrhenius relationship between melt temperature and viscosity. Where η is the intrinsic viscosity of the melt, exp is the natural exponential function, B is the pre-exponential factor, and E a The activation energy is given by R, the gas constant is given by T, and the absolute temperature is given by T. The required temperature correction ΔT is calculated based on the current viscosity deviation. lm Then through the jacket heat transfer equation The required change in heat load is calculated, and the opening of the heat transfer medium control valve is adjusted accordingly; the heat transfer medium is a mixture of biphenyl and diphenyl ether, Q is the required change in heat load, and U is the heat transfer coefficient.

[0013] Furthermore, the global intrinsic viscosity value is calculated using a volume-weighted average method, which divides the homogenization vessel cavity into several hexahedral grid units. The viscosity value of each unit is obtained by inverse distance weighted interpolation based on the inversion results of at least four transmission paths around it. The unit volume weight is determined according to its proportion in the total melt volume. Finally, the global intrinsic viscosity value is the sum of the products of the viscosity values ​​of each unit and their volume weights.

[0014] Furthermore, in the initial stage of the homogenization process, the system performs a rapid modeling phase: within the first 10 minutes, ultrasonic data is collected and the intrinsic viscosity distribution map is updated at a frequency of once every 5 seconds, while the stirring power consumption and melt temperature change rate are recorded; when the global intrinsic viscosity value change rate is less than 0.1% per minute for three consecutive calculations, the system automatically switches to steady-state monitoring mode, reducing the data acquisition frequency to once every 30 seconds, in order to reduce the computational load and extend the sensor life.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0016] 1. This invention achieves non-contact, non-destructive, real-time online monitoring of the intrinsic viscosity of recycled polyester melt by constructing a multi-path ultrasonic transmission sensing network and an acoustic-rheological coupling inversion model, overcoming the process lag problem caused by traditional offline sampling and detection.

[0017] 2. The viscosity monitoring results are linked with the process execution units such as stirring and heating to form a closed-loop linkage control. The control strategy is triggered in stages according to the magnitude and duration of viscosity deviation, which eliminates the uncertainty of human experience intervention and solves the technical bottleneck of the disconnect between monitoring and control. Through the global viscosity calculation method of volume weighted average and the dynamic tolerance band determination mechanism, the scientific determination of the homogenization endpoint of each batch is ensured, which significantly improves the batch consistency and stability of the intrinsic viscosity of recycled polyester products.

[0018] 3. The entire system has the ability to operate reliably for a long time in the high temperature, high viscosity and strong corrosion environment of polyester melt, providing solid technical support for the intelligent and high-quality control of recycled polyester production. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall technical solution architecture of the online viscosity monitoring method for recycled polyester homogenization kettle based on ultrasonic transmission proposed in this invention.

[0020] Figure 2 This is a schematic diagram of the core principle framework of the acoustic-rheological coupling inversion model in this invention;

[0021] Figure 3 This is a flowchart illustrating the logical flow of the multipath ultrasonic transmission sensing network construction and acoustic characteristic extraction in this invention.

[0022] Figure 4 This is a flowchart illustrating the logical flow of global intrinsic viscosity calculation and homogenization state determination in this invention.

[0023] Figure 5 This is a flowchart illustrating the logical flow of the viscosity deviation-driven hierarchical closed-loop control strategy in this invention.

[0024] Figure 6 This is a schematic diagram of the multi-level interaction and data flow between the melt in the homogenization vessel, the sensor array, and the process execution unit in this invention. Detailed Implementation

[0025] Please refer to the attached document. Figures 1 to 6This invention provides an online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission, aiming to solve the technical problems of disconnect between monitoring and control, lag in process response, and uncontrollable homogenization effect in existing technologies. The method constructs a multi-channel ultrasonic transmission sensor array inside the homogenization reactor to acquire the acoustic propagation characteristics of the recycled polyester melt in real time at different spatial locations and time dimensions, and dynamically analyzes the characteristic viscosity of the melt based on a sound velocity-attenuation joint inversion model. Simultaneously, the obtained viscosity data is linked in a closed loop with the process execution unit of the homogenization reactor. Based on a preset viscosity target range and dynamic tolerance band, control commands for stirring rate, heating power, and material residence time are automatically generated, thereby achieving integrated control of real-time perception, accurate judgment, and proactive intervention in the homogenization process of the recycled polyester melt.

[0026] The method includes the following steps: at least four ultrasonic transmitting transducers are evenly arranged circumferentially on the cylindrical wall of the homogenizing vessel, and the same number of ultrasonic receiving transducers are correspondingly arranged on the inner wall opposite to them, forming a multi-path cross-transmission sensing network.

[0027] The ultrasonic transmitting transducer transmits an ultrasonic signal with a center frequency of 1.5 MHz and a pulse width of 20 microseconds in a pulse modulation manner. After penetrating the recycled polyester melt, the signal is captured by the receiving transducer. Time-domain envelope extraction and frequency-domain spectrum analysis are performed on the ultrasonic signal received along each transmission path to calculate the ultrasonic propagation time and energy attenuation coefficient for that path.

[0028] Based on the propagation time data of all transmission paths, the least squares method is used to fit the spatial distribution field of sound velocity inside the melt; simultaneously, based on the energy attenuation coefficients of all paths, the spatial distribution field of attenuation coefficients inside the melt is constructed. The spatial distribution fields of sound velocity and attenuation coefficients are input into a pre-trained acoustic-rheological coupled inversion model, outputting a three-dimensional distribution map of the intrinsic viscosity of the recycled polyester melt at the current moment. A volume-weighted average is performed on the three-dimensional distribution map of intrinsic viscosity to obtain the global intrinsic viscosity value characterizing the homogenization state of the entire melt in the reactor.

[0029] The global intrinsic viscosity value is compared with the preset target viscosity value. If the absolute value of the deviation is greater than the preset first threshold, the first-level control strategy is activated to adjust the rotation speed of the stirring paddle at the bottom of the homogenizing tank. If the absolute value of the deviation is greater than the preset second threshold and the duration exceeds 30 seconds, the second-level control strategy is activated to adjust the flow rate of the heat medium in the jacket heating system while adjusting the stirring speed. If the absolute value of the deviation is within the ±0.5% tolerance range of the target viscosity value for 60 consecutive seconds, the current batch is determined to have reached the homogenization endpoint, and the discharge control signal is triggered.

[0030] In step S1, at least four ultrasonic transmitting transducers are evenly arranged circumferentially on the wall of the homogenizing vessel, and the same number of ultrasonic receiving transducers are correspondingly arranged on the inner wall, forming a multi-path cross-transmission sensing network. Both the ultrasonic transmitting and receiving transducers are made of high-temperature resistant piezoelectric ceramic material with a Curie temperature of not less than 350 degrees Celsius. The encapsulation shell is made of Hastelloy C276 material, with an operating temperature range of 280 degrees Celsius to 320 degrees Celsius.

[0031] The transducers are sealed to the homogenizing vessel body via flanges, with double-layered spiral wound gaskets at the connection points to ensure airtightness under vacuum or slightly positive pressure conditions. All electrical leads of the transducers are led out through high-temperature ceramic insulating sleeves and connected to a signal conditioning module located in an explosion-proof junction box. The arrangement of the multi-path cross-transmission sensor network ensures that each pair of transmitting-receiving transducers forms an independent transmission path, and these paths spatially intersect, covering multiple areas within the homogenizing vessel cavity, thus comprehensively reflecting the acoustic propagation characteristics of the melt at different locations. The number of transmission paths is determined by the diameter of the homogenizing vessel: 4 pairs of transducers are arranged when the vessel diameter is less than two meters; 6 pairs are arranged when the diameter is between two and three meters; and 8 pairs are arranged when the diameter is greater than three meters. The installation angle intervals for each pair of transducers are 90 degrees, 60 degrees, or 45 degrees to achieve uniform sampling of the melt cross-section.

[0032] In step S2, the ultrasonic transducer transmits an ultrasonic signal with a center frequency of 1.5 MHz and a pulse width of 20 microseconds using pulse modulation. The selection of these signal parameters is based on a balance between the acoustic absorption characteristics of the recycled polyester melt and the required penetration depth. Too high a center frequency would cause the signal to attenuate rapidly in the high-viscosity melt, failing to penetrate the entire reactor; too low a frequency would reduce spatial resolution, making it difficult to capture local viscosity changes. A pulse width of 20 microseconds ensures sufficient energy to penetrate the melt while avoiding multipath interference caused by excessively long pulses. The transmitted signal is generated by a digital signal generator in the central processing unit and then drives the ultrasonic transducer via a power amplifier. The transmission timing uses a polling method, sequentially activating each pair of transmitter-receiver transducers to ensure that signals from each path do not interfere with each other. Each complete path scan cycle does not exceed 500 milliseconds to meet the requirements of real-time monitoring.

[0033] In step S3, time-domain envelope extraction and frequency-domain spectrum analysis are performed on the ultrasonic signals received through each transmission path to calculate the ultrasonic propagation time and energy attenuation coefficient for that path. The signal conditioning module performs pre-amplification, bandpass filtering, and analog-to-digital conversion on the original received signal. The preamplifier has a gain of 40 dB to improve the signal-to-noise ratio of weak signals; the bandpass filter has a passband range of 0.8 MHz to 2.2 MHz to filter out low-frequency mechanical vibration noise and high-frequency electromagnetic interference; the analog-to-digital converter has a sampling frequency of 20 MHz and a quantization bit depth of 16 bits to ensure complete preservation of signal details. The digital signal after analog-to-digital conversion is transmitted to the central processing unit for further processing. Time-domain envelope extraction uses the Hilbert transform method to construct the analytic signal for each received signal sequence x(n). Where j is the imaginary unit, H[x(n)] represents the Hilbert transform operator, and the envelope signal e(n) is defined as |z(n)|; the ultrasonic propagation time is determined by detecting the initial zero-crossing point of the envelope signal e(n), specifically by finding the time point at which the amplitude of the envelope signal first exceeds three standard deviations of the noise floor; the energy attenuation coefficient α is obtained by formula The calculation is performed, where L is the transmission path length, P0 is the received signal energy along the same path under reference conditions, obtained by calibration using standard silicone oil injected into an empty reactor. This calibration process is performed once after each equipment maintenance to ensure the accuracy of the reference energy. P is the current received signal energy.

[0034] In step S4, based on the propagation time data of all transmission paths, the least squares method is used to fit the spatial distribution field of the sound velocity inside the melt. Let there be a total of M transmission paths, and the... The propagation time of the path is t i The path length is L i Then the average speed of sound along that path The homogenization vessel's inner cavity is divided into N hexahedral mesh elements, with the sound velocity in each element being an unknown. A system of linear equations is then established. ,in For an M×N path-cell incidence matrix, element A ik The length of the i-th path passing through the k-th unit represents the proportion of the path's length; x is an N-dimensional sound speed vector; b is an M-dimensional observation vector. b i For the first The M-dimensional observation vectors of the path are given. The sound speed vector is solved using the least squares method. T is the transpose; the sound velocity estimates for each element are obtained, thus constructing the sound velocity spatial distribution field. Simultaneously, in step S5, based on the energy attenuation coefficients of all paths, the attenuation coefficient spatial distribution field inside the melt is constructed. Using the same mesh division and correlation matrix, the attenuation coefficient αi of each path is used as the observation value, and the attenuation coefficient of each element is inverted using the least squares method to form the attenuation coefficient spatial distribution field.

[0035] In step S6, the spatial distribution fields of sound velocity and attenuation coefficient are input into a pre-trained acoustic-rheological coupling inversion model, outputting a three-dimensional distribution map of the intrinsic viscosity of the recycled polyester melt at the current moment. The construction process of the acoustic-rheological coupling inversion model includes: collecting melt samples from no fewer than 500 batches of recycled polyester production under different temperature, pressure, molecular weight distribution, and impurity content conditions; measuring the intrinsic viscosity of each batch under standard laboratory conditions, and simultaneously recording the sound velocity and attenuation data obtained by an ultrasonic transmission sensor network under the same operating conditions; using the mean sound velocity, standard deviation of sound velocity, mean attenuation, and standard deviation of attenuation as input feature vectors, and the laboratory-measured intrinsic viscosity as the output label, training a nonlinear mapping relationship using a support vector regression algorithm. The support vector regression algorithm uses a radial basis function kernel, with a penalty factor C of 10 and a kernel function parameter γ of 0.1. After cross-validation, the root mean square error of the model prediction is less than 0.005 dL / g. In practical applications, the model receives the sound velocity and attenuation coefficient of each grid cell as input and outputs the characteristic viscosity value corresponding to that cell, thereby generating a complete three-dimensional viscosity distribution map.

[0036] In step S7, the three-dimensional intrinsic viscosity distribution map is volume-weightedly averaged to obtain a global intrinsic viscosity value characterizing the homogenization state of the entire melt. The homogenization vessel cavity is divided into several hexahedral grid cells, and the viscosity value of each cell is obtained by inverse distance weighted interpolation of the inversion results of at least four transmission paths around it. The cell volume weight is determined according to its proportion in the total melt volume. Let the viscosity of the k-th cell be η. k The volume is V k The total melt volume is V total Then the global intrinsic viscosity value This calculation method fully considers the volume contribution of the melt in different regions of the vessel, avoiding the bias caused by simple arithmetic averaging.

[0037] In step S8, the global intrinsic viscosity value is compared with a preset target viscosity value, and a graded control strategy is executed. The preset first threshold is 0.8%, and the second threshold is 1.5%. If the first threshold is reached, the first-level control strategy is activated. This involves adjusting the agitator speed. From the formula Determine, where dt is the time differential, η target η is the target intrinsic viscosity value. current K represents the current global intrinsic viscosity value. p This is a proportionality coefficient, with a value of 15 revolutions per minute per minute per liter per gram, K. i This is the integral coefficient, with a value of 0.5 revolutions per minute per minute per liter per second per gram per second. The adjustment range of the agitator speed is limited to between 30 revolutions per minute and 120 revolutions per minute, with each adjustment step not exceeding 5 revolutions per minute. If If the second threshold is reached and the duration exceeds 30 seconds, the second-level control strategy is activated. Simultaneously, the flow rate of the heating medium in the jacket heating system is adjusted. The adjustment of the heating medium flow rate is based on the Arrhenius equation relating melt temperature and viscosity. Where η is the intrinsic viscosity of the melt, exp is the natural exponential function, B is the pre-exponential factor, and E a Let R be the activation energy, R be the gas constant, and T be the absolute temperature. The required temperature correction ΔT is calculated by working backwards from the current viscosity deviation. lm Then through the jacket heat transfer equation Calculate the required change in heat load, and then adjust the opening of the heat transfer medium control valve accordingly. Q is the required change in heat load, and U is the heat transfer coefficient. The heat transfer medium is a biphenyl-diphenyl ether mixture, with an operating temperature range of 250°C to 350°C. If... If the viscosity remains within ±0.5% of the target viscosity for 60 consecutive seconds, the current batch is considered to have reached the homogenization endpoint, triggering the discharge control signal to stop stirring and open the discharge valve.

[0038] In the initial stage of the homogenization process, the system performs a rapid modeling phase. For the first 10 minutes, ultrasonic data is acquired every 5 seconds, and the intrinsic viscosity distribution map is updated, while simultaneously recording stirring power consumption and melt temperature change rate. When the rate of change of the global intrinsic viscosity value calculated for three consecutive times is less than 0.1% per minute, the system automatically switches to steady-state monitoring mode, reducing the data acquisition frequency to once every 30 seconds to reduce computational load and extend sensor lifespan. This mechanism ensures high temporal resolution data during the initial stage of rapid melt state changes, while minimizing unnecessary resource consumption during the stable phase.

[0039] The system comprises a homogenization vessel body, a multi-path ultrasonic transmission sensor network, a signal conditioning module, a central processing unit, and a process execution unit. The homogenization vessel body is a vertical cylindrical container with a bottom stirring impeller and a jacketed heating system. The multi-path ultrasonic transmission sensor network consists of ultrasonic transmitting and receiving transducers evenly arranged circumferentially. The transducers are made of high-temperature resistant piezoelectric ceramic material, encapsulated in a Hastelloy C276 shell, and sealed to the vessel body via flanges. The signal conditioning module, located in an explosion-proof junction box, includes a preamplifier, a bandpass filter, and an analog-to-digital converter, used for preprocessing the received signal. The central processing unit is an industrial-grade embedded computer running a real-time operating system, performing time-domain envelope extraction, frequency-domain spectrum analysis, sound velocity and attenuation field inversion, viscosity distribution calculation, and control command generation. The process execution unit includes a stirring motor frequency converter, a heat medium control valve, and a discharge control valve, receiving commands from the central processing unit and executing corresponding actions. All components are interconnected via industrial Ethernet or fieldbus, forming a complete closed-loop control system.

Claims

1. A method for online viscosity monitoring of a recycled polyester homogenization reactor based on ultrasonic transmission, characterized in that, include: At least four ultrasonic transducers are evenly arranged along the circumferential direction on the cylindrical wall of the homogenizing vessel, and the same number of ultrasonic receiving transducers are correspondingly set at the corresponding inner wall position to form a multi-path cross-transmission sensing network. The ultrasonic transmitting transducer is controlled to emit ultrasonic signals in a pulse modulation manner, which penetrate the recycled polyester melt and are captured by the ultrasonic receiving transducer. For each transmission path, the ultrasonic signal received is subjected to time-domain envelope extraction and frequency-domain spectrum analysis, and the ultrasonic propagation time and energy attenuation coefficient under that path are calculated respectively. Based on the propagation time data of all transmission paths, the least squares method is used to fit the spatial distribution field of sound velocity inside the melt; simultaneously, based on the energy attenuation coefficient of all paths, the spatial distribution field of attenuation coefficient inside the melt is constructed. The spatial distribution field of sound velocity and the spatial distribution field of attenuation coefficient are input into a pre-trained acoustic-rheological coupling inversion model to output a three-dimensional distribution map of the intrinsic viscosity of recycled polyester melt at the current moment. The volume-weighted average of the three-dimensional distribution map of intrinsic viscosity is used to obtain the global intrinsic viscosity value that characterizes the homogenization state of the melt in the whole reactor. The global intrinsic viscosity value is compared with the preset target viscosity value. If the absolute value of the deviation is greater than the preset first threshold, the first-level control strategy is activated to adjust the rotation speed of the stirring paddle at the bottom of the homogenizing tank. If the absolute value of the deviation is greater than the preset second threshold and the duration exceeds 30 seconds, the second-level control strategy is activated, adjusting the stirring speed and the flow rate of the heat medium in the jacket heating system. If the absolute value of the deviation is within the ±0.5% tolerance range of the target viscosity value for 60 consecutive seconds, the current batch is determined to have reached the homogenization endpoint, triggering the discharge control signal. In the first-level control strategy, the adjustment amount ΔN of the impeller speed is given by the formula Determine, where dt is the time differential, η target η is the target intrinsic viscosity value. current K represents the current global intrinsic viscosity value. p This is a proportionality coefficient, with a value of 15 revolutions per minute per minute per liter per gram, K. i This is the integral coefficient, with a value of 0.5 revolutions per minute per minute per liter per second per gram per second; In the second-level control strategy, the adjustment of the heat transfer medium flow rate is based on the Arrhenius relationship between melt temperature and viscosity. Where η is the intrinsic viscosity of the melt, exp is the natural exponential function, B is the pre-exponential factor, and E a The activation energy is given by R, the gas constant is given by T, and the absolute temperature is given by T. The required temperature correction ΔT is calculated based on the current viscosity deviation. lm Then through the jacket heat transfer equation The required change in heat load is calculated, and the opening of the heat transfer medium control valve is adjusted accordingly; the heat transfer medium is a mixture of biphenyl and diphenyl ether, Q is the required change in heat load, and U is the heat transfer coefficient.

2. The online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission according to claim 1, characterized in that, The ultrasonic transmitting and receiving transducers in the multipath cross-transmission sensor network are both made of high-temperature resistant piezoelectric ceramic material; the transducers are sealed to the homogenization vessel body through flanges, and double metal spiral wound gaskets are provided at the connection; the electrical leads of all transducers are led out through high-temperature ceramic insulating sleeves and connected to the signal conditioning module located in the explosion-proof junction box.

3. The online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission according to claim 2, characterized in that, The signal conditioning module performs pre-amplification, bandpass filtering, and analog-to-digital conversion on the original received signal; the digital signal after analog-to-digital conversion is transmitted to the central processing unit, which performs time-domain envelope extraction and frequency-domain spectrum analysis.

4. The online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission according to claim 3, characterized in that, The time-domain envelope extraction employs the Hilbert transform method to construct the analytic signal for each received signal sequence x(n). Where j is the imaginary unit, H[x(n)] represents the Hilbert transform operator, and the envelope signal e(n) is defined as |z(n)|; the ultrasonic propagation time is determined by detecting the initial zero-crossing point of the envelope signal e(n), specifically by finding the time point at which the amplitude of the envelope signal first exceeds three standard deviations of the noise floor; the energy attenuation coefficient α is obtained by formula The calculation is performed, where L is the transmission path length, P0 is the received signal energy along the same path in the reference state, and P is the current received signal energy.

5. The online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission according to claim 1, characterized in that, Based on the propagation time data of all transmission paths, the least squares method is used to fit the spatial distribution field of sound velocity inside the melt, including: dividing the homogenization vessel cavity into N hexahedral mesh elements, establishing an M×N path-element correlation matrix A, where M is the number of transmission paths, and element A... ik Let t represent the proportion of the length of the i-th path that passes through the k-th unit; the propagation time of the i-th path is t. i The path length is L i Then the average speed of sound along that path b is an M-dimensional observation vector. b i Let M be the observation vector of the i-th path; solve for the sound speed vector using the least squares method. T is the transpose; the sound velocity estimates of each unit are obtained, thus constructing the sound velocity spatial distribution field; simultaneously, using the same grid division and correlation matrix, the attenuation coefficient of each path is used as the observation value, and the attenuation coefficient of each unit is inverted by the least squares method to form the attenuation coefficient spatial distribution field.

6. The online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission according to claim 1, characterized in that, The construction process of the acoustic-rheological coupling inversion model includes: collecting melt samples from no less than 500 batches of recycled polyester production under different temperature, pressure, molecular weight distribution and impurity content conditions; For each batch of samples, the intrinsic viscosity was measured under standard laboratory conditions, and the sound velocity and attenuation data were simultaneously recorded by the ultrasonic transmission sensor network under the same working conditions. The mean sound velocity, standard deviation of sound velocity, mean attenuation, and standard deviation of attenuation were used as input feature vectors, and the intrinsic viscosity measured in the laboratory was used as output label. The nonlinear mapping relationship was trained using the support vector regression algorithm. The support vector regression algorithm uses a radial basis function kernel.

7. The online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission according to claim 1, characterized in that, The volume-weighted average of the three-dimensional intrinsic viscosity distribution map is used to obtain the global intrinsic viscosity value characterizing the homogenization state of the entire melt in the reactor, including: The homogenization vessel cavity is divided into several hexahedral grid units. The viscosity value of each unit is obtained by inverse distance weighted interpolation from the inversion results of at least four transmission paths around it. The unit volume weight is determined based on its proportion in the total melt volume; let the viscosity of the k-th unit be η. k The volume is V k The total melt volume is V total Then the global intrinsic viscosity value .

8. The online viscosity monitoring method for a recycled polyester homogenization reactor based on ultrasonic transmission according to claim 1, characterized in that, In the initial stage of the homogenization process, the system performs a rapid modeling phase: In the first 10 minutes, ultrasonic data was collected and the intrinsic viscosity distribution map was updated at a frequency of once every 5 seconds, while the stirring power consumption and melt temperature change rate were recorded. When the rate of change of the global intrinsic viscosity value calculated for three consecutive times is less than 0.1% per minute, the system automatically switches to steady-state monitoring mode and reduces the data acquisition frequency to once every 30 seconds.

Images