Internal mixing online self-cleaning method based on torque prediction cooperation

Abstract

The invention discloses an internal mixing online self-cleaning method based on torque prediction synergy, and relates to the technical field of rubber mixing, torque curves, acoustic emission waves and vibration signals are synchronously collected through multi-source sensing, deposition activation coefficients are calculated, and deposition risks are recognized early; low-amplitude torsional vibration desorption is adopted, and the low-amplitude torsional vibration is excited by the rotor wing end to safely and efficiently crack sediments, so that additional wear is avoided; directional jet flow is utilized, the nozzle pressure and the jet flow duration time are automatically adjusted according to the deslagging efficiency coefficient, rapid deslagging is achieved, and the cleaning efficiency is improved; receiving the deposition activation coefficient and the slag discharge amount data, correcting the torque baseline in real time and optimizing the prediction precision; and in combination with focused ultrasound and low-temperature plasma composite pulse, the residual hard shell is quickly desorbed, and a deposition activation coefficient is fed back and updated. And the running efficiency and the service life of the internal mixer can be remarkably improved, energy consumption and abrasion are reduced, and meanwhile, cleaning agent and waste liquid discharge is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rubber mixing and mixing, in particular to an online self-cleaning method for an internal mixer based on torque prediction cooperation. BACKGROUND

[0002] In rubber internal mixing production, high-abrasion fillers such as carbon black, glass fiber and silicon carbide continuously collide with the rotor wing end and the cavity wall under the action of strong shear and high temperature, gradually forming a hard deposition layer, blocking the original gap and significantly increasing the friction resistance, and the torque curve baseline is raised accordingly. The conventional cleaning rhythm is difficult to timely remove such deposits, resulting in phenomena such as energy consumption rising, mixing uniformity decreasing and rotor edge wear accelerating. Once the deposition layer is formed, it will capture undispersed particles and change the flow field of the cavity, causing more severe local thermal shear and amplifying the adhesion of carbon black decomposition products, thereby causing torque drift that cannot be corrected by traditional frequency conversion speed regulation. In order to reduce the number of shutdowns, the industry has introduced digital twin torque model to predict the trend, and has tried to use non-contact methods such as plasma or ultrasonic vibration for online cleaning, but such solutions still have limited practical cases in high-abrasion systems, making the equipment have to be periodically shut down for inspection, and the production rhythm fluctuates.

[0003] The layer-by-layer accumulation of high-abrasion fillers on the rotor surface causes the torque prediction model to quickly lose reference, the algorithm misjudges the material state and delays the online self-cleaning action, and the deposition continues to thicken, causing the motor load to rise and the temperature of the contact surface between the rotor and the cavity wall to rise. The rough surface is more likely to embed new particles, eventually forming a vicious cycle of energy consumption, wear and prediction inaccuracy. If this process is continuously ignored, it can only be stopped by manual sandblasting or chemical immersion.

[0004] Therefore, the present application provides an online self-cleaning method for an internal mixer based on torque prediction cooperation. SUMMARY

[0005] (I) Technical problems solved

[0006] In view of the deficiencies of the prior art, the present application provides an online self-cleaning method for an internal mixer based on torque prediction cooperation, which synchronously collects torque curves, acoustic emission waves and vibration signals through multi-source sensing, calculates deposition activation coefficients and early identifies deposition risks; uses low-amplitude torsional vibration to safely and efficiently crack the deposits by exciting low-amplitude torsional vibration from the rotor wing end, avoiding additional wear; uses directional jets to automatically adjust the jet pressure and jet duration according to the deslagging efficiency coefficient, achieving rapid deslagging and improving cleaning efficiency; receives deposition activation coefficients and deslagging amount data, and real-time corrects the torque baseline and optimizes the prediction accuracy; combined with focused ultrasound and low-temperature plasma complex pulse, the running efficiency and service life of the internal mixer can be significantly improved, and the energy consumption and wear can be reduced, solving the technical problems described in the background art.

[0007] (ii) Technical Solution

[0008] To achieve the above object, the application realizes the above object by the following technical solution: an online self-cleaning method for a mixer based on torque prediction cooperation, comprising: collecting torque curves, acoustic emission waves and vibration signals in real time, calculating a deposition activation coefficient and marking a high-risk section as a trigger basis for low-amplitude torsional vibration detachment;

[0009] According to the high-risk section marking, low-amplitude torsional vibration is excited at the rotor wing end, the deposits are cracked through shear stress and fatigue fracture mechanism, and the vibration amplitude is controlled in real time to protect the rotor;

[0010] According to the high-risk section marking information, a directional jet is used, and the jet pressure and jet duration are automatically adjusted by real-time calculation of the deslagging efficiency coefficient;

[0011] Using a digital twin torque model, combining the deposition activation coefficient and the deslagging amount, the torque baseline is corrected in real time and the end point determination threshold is adjusted, so that the torque prediction continuously fits the equipment state;

[0012] When the torque prediction still shows a local high-risk section, a focused ultrasound and low-temperature plasma combined pulse is applied to quickly detach the residual hard shell, and the updated deposition activation coefficient is fed back.

[0013] Further, the torque curve, acoustic emission wave and vibration signal of the mixer rotor are synchronously collected;

[0014] And the difference between the maximum value and the long-term average value of the torque and the long-term average value are calculated as the torque component, the sum of the power spectral density of the acoustic emission wave in the specified frequency range is calculated as the acoustic emission component, and the difference between the maximum amplitude and the average amplitude and the average amplitude is calculated as the vibration component.

[0015] Further, the torque component, acoustic emission component and vibration component are multiplied by a preset normalization coefficient and added to obtain the deposition activation coefficient; when the deposition activation coefficient exceeds a preset threshold, a marking information of the high-risk section is generated, including the starting time and the duration; and the marking information is output.

[0016] Further, the rotor vibration amplitude is monitored in real time by a displacement sensor, the monitoring value is compared with a preset vibration amplitude, the output power of the vibration exciter is adjusted according to the comparison result, so that the vibration amplitude is kept within the allowable error range; and the vibration frequency, vibration amplitude, actual vibration amplitude, vibration duration, rotor displacement response and acceleration response data are recorded.

[0017] Further, the jet pressure is measured by a micro-pressure sensor and converted into a momentum value, an erosion momentum signal is generated by integral normalization processing, and a particle scattering signal is generated by normalization processing of the scattering spectrum area difference measured by a light scattering sensor.

[0018] Further, the erosion momentum signal and the particle scattering signal are input into a radial basis neural network to calculate a fusion coefficient; the jet port pressure and the jet duration are adjusted according to the fusion coefficient to optimize the deslagging effect; and the deslagging amount data is measured and recorded by a mass flow meter.

[0019] Further, a torque baseline value is calculated by receiving the deposition activation coefficient and the deposition removal amount, the digital twin torque model parameters are adjusted using the deposition activation coefficient, and an end point determination threshold is calculated according to the torque baseline value change.

[0020] Further, a torque prediction value is calculated by the digital twin torque model in combination with the real-time collected torque curve, acoustic emission wave and vibration signal; when the torque prediction value exceeds the end point determination threshold, it is marked as a high-risk section and a reinforcement cleaning is notified; and the digital twin torque model parameters are adjusted according to the deviation between the actual torque and the torque prediction value.

[0021] Further, a focused ultrasound wave and a low-temperature plasma composite pulse are synchronously applied in the high-risk section by an ultrasonic transducer and a plasma generator.

[0022] The dual mechanisms of acoustic wave fragmentation and plasma bombardment are used to detach the residual hard shell, and after cleaning, a new deposition activation coefficient is calculated using multi-source sensing data and compared with the original deposition activation coefficient.

[0023] Further, when the new deposition activation coefficient is less than the original deposition activation coefficient, the original deposition activation coefficient is updated; and when the new deposition activation coefficient is not less than the original deposition activation coefficient, the ultrasonic power or the plasma pulse frequency is adjusted and the cleaning is repeated.

[0024] (III) Beneficial effects

[0025] The present application provides an online self-cleaning method for an internal mixer based on torque prediction cooperation, which has the following beneficial effects:

[0026] By synchronously collecting the torque curve, acoustic emission wave and vibration signal, the deposition activation coefficient is calculated to mark the high-risk section, which significantly improves the comprehensiveness and accuracy of the deposit identification, and can early discover potential risks in the early stage of deposit formation. By accurately positioning the high-risk section, the device load intensification and performance decline caused by further thickening of the deposit layer are effectively avoided.

[0027] The low-amplitude torsional vibration of the rotor wing end is excited, the shear stress and fatigue fracture mechanism are used to crack the deposit, and the safety of the cleaning process is ensured, avoiding damage to the rotor surface. The step effectively cracks the deposit, creating favorable conditions for directional jet deslagging.

[0028] The directional jet flow combines the real-time calculation of the deslagging efficiency coefficient to automatically adjust the nozzle pressure and jet duration, realizing the efficient discharge of sediment fragments. The calculation of the deslagging efficiency coefficient combines the erosion momentum signal and particle scattering signal and is mapped through a radial basis neural network, ensuring the accuracy and adaptability of the deslagging process, and improving the cleaning efficiency.

[0029] The real-time correction of the torque baseline and the adjustment of the end point determination threshold make the torque prediction model continuously fit the equipment operation state. This real-time calibration mechanism significantly improves the accuracy of torque prediction and provides a reliable trigger basis for reinforcement cleaning. The application of digital twinning enhances the adaptability and accuracy of the entire method.

[0030] The focused ultrasound and low-temperature plasma composite pulse technology is used to quickly desorb residual hard shell in high-risk sections, and the updated deposition activation coefficient is fed back to the transmission, forming a closed-loop optimization. This composite cleaning technology ensures the thoroughness and continuity of the cleaning process, and through data feedback and cooperation, further improves the overall efficiency.

[0031] In summary, online self-cleaning of the mixer rotor surface deposits is realized, which significantly reduces the energy consumption of the equipment, delays the rotor wear, stabilizes the production rhythm, and reduces the discharge of cleaning agents and waste liquid, providing comprehensive support for long-term efficient operation of the mixer. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 The figure is a process flow diagram of the online self-cleaning method of the mixer based on torque prediction coordination of the present application. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of protection of the present application.

[0034] Please refer to Figure 1 The present application provides an online self-cleaning method for a mixer based on torque prediction coordination, which comprises,

[0035] Step one, synchronously collect the torque curve, acoustic emission wave and vibration signal of the mixer rotor; calculate the ratio of the difference between the maximum value and the long-term average value of the torque to the long-term average value as the torque component; calculate the sum of the power spectral density of the acoustic emission wave in the specified frequency range as the acoustic emission component; calculate the ratio of the difference between the maximum amplitude and the average amplitude of the vibration to the average amplitude as the vibration component; multiply the torque component, acoustic emission component and vibration component by the preset normalization coefficient respectively and add them to obtain the deposition activation coefficient; when the deposition activation coefficient exceeds the preset threshold, generate the marking information of the high-risk section, including the starting time and the duration; pass the marking information to the next step.

[0036] The step one includes the following contents:

[0037] Step 101, data acquisition

[0038] In the data acquisition stage, the formation of the rotor surface deposits of the mixer is comprehensively monitored by synchronously acquiring three key signals. The first signal is the torque curve, which records the torque change of the mixer rotor in real time by using a torque sensor. The abnormal increase of the torque value reflects the phenomenon that the deposits increase the friction resistance. The second signal is the acoustic emission wave, which captures the acoustic signal generated by the friction between the rotor and the cavity wall through an acoustic emission sensor. The abnormal change of the acoustic wave in the high frequency band indicates that the phenomenon of micro-cracks or particle embedding may occur on the surface of the deposits. The third signal is the vibration signal, which measures the vibration frequency and amplitude of the rotor by using a vibration sensor. The abnormal peak value or frequency offset of the vibration shows the influence of the deposits accumulation on the dynamic balance of the rotor. The synchronous acquisition of these three signals provides multi-dimensional data sources for subsequent feature extraction.

[0039] Step 102, feature extraction

[0040] In the feature extraction stage, the three signals collected are processed to quantify the characteristics of the deposits. For torque curve analysis, the maximum value and the long-term average value of the torque are calculated. The long-term average value is the stable torque value obtained after the rotor runs for a period of time without deposits, and the maximum value is the torque peak value recorded in the current monitoring period. For acoustic emission wave analysis, power spectrum analysis is performed on the acoustic emission signal. The sum of the power spectral density in the specified frequency range is calculated to represent the concentration degree of acoustic energy. The frequency range is determined in advance according to the characteristics of the rotor material. For vibration signal analysis, the maximum amplitude and the average amplitude of the vibration are extracted. The average amplitude is the average level of the vibration in the monitoring period, and the maximum amplitude is the vibration peak value in the period. The extraction of these features provides a quantitative basis for the calculation of the subsequent deposition activation coefficient.

[0041] Step 103, deposition activation coefficient calculation

[0042] In the deposition activation coefficient calculation stage, a comprehensive index is calculated based on the characteristics of the three signals to assess the deposition risk. The specific calculation process is divided into three components.

[0043] First, the calculation method of the torque component is to take the difference between the maximum torque value and the long-term average value, and then divide it by the long-term average value. The result reflects the degree of influence of the deposition on the friction resistance.

[0044] Second, the calculation method of the acoustic emission component is to use the sum of the power spectral density in the specified frequency range directly as the component value. This value represents the strength of the acoustic characteristics of the deposition micro-cracks or friction.

[0045] Finally, the calculation method of the vibration component is to take the difference between the maximum amplitude and the average amplitude of the vibration, and then divide it by the average amplitude. The result indicates the degree of mechanical imbalance caused by the deposition. Finally, the three components are multiplied by their respective normalization coefficients and added together to obtain the deposition activation coefficient. The normalization coefficients are adjusted according to the characteristics of the equipment and experimental data to ensure that each component contributes equally to the deposition activation coefficient.

[0046] Step 104, high-risk section marking

[0047] In the high-risk section marking stage, risk judgment is made based on the calculation results of the deposition activation coefficient.

[0048] A threshold value is set in advance. When the deposition activation coefficient exceeds this threshold value, it is considered that there is a high risk of deposition in the current time period. The marking information includes the start time and duration of the high-risk section, where the start time is the time point when the deposition activation coefficient first exceeds the threshold value, and the duration is the length of time when the coefficient remains above the threshold value. These marking information provides clear execution basis for subsequent processing steps.

[0049] In the data transmission stage, the marking information of the high-risk section, including the start time and duration, is transmitted to the processing unit of the next step. These information is used to guide the implementation of low-amplitude torsional vibration detachment, ensuring that the detachment action is accurately performed on the high-risk section, thereby effectively removing the deposition on the rotor surface.

[0050] By synchronously collecting torque curves, acoustic emission waves, and vibration signals, the formation of deposition is evaluated from three dimensions: macroscopic friction resistance, microscopic crack characteristics, and dynamic balance influence. The rationality of this multi-dimensional monitoring method lies in the fact that a single signal may be disturbed by noise or equipment operating state, while the collaborative analysis of three signals can verify each other, improving the accuracy and comprehensiveness of deposition identification. It can detect deposition problems earlier and more comprehensively, avoiding the decline of equipment performance due to the omission of risks in single-dimensional monitoring.

[0051] Through the calculation of the deposition activation coefficient, the abnormal rising trend of the deposition activation coefficient can be detected at the initial stage of the formation of the deposition, thereby achieving early warning. If the deposition is not treated in time, the deposition will gradually thicken, increase the load of the equipment and prolong the cleaning time. By identifying the risk in time, the system can prevent the deposition from deteriorating, reduce the potential damage in the operation of the equipment and prolong the service life.

[0052] The marking of the high-risk section is based on the judgment that the deposition activation coefficient exceeds the threshold value, and the specific starting time and duration are recorded. The reason for this marking method is that the accurate definition in the time dimension can provide a clear target section for subsequent detachment actions, avoiding blind operation. Reducing invalid cleaning actions, improving detachment efficiency, and ensuring the pertinence of resource use.

[0053] Data acquisition, feature extraction and deposition activation coefficient calculation are all carried out in real time to ensure that high-risk section marking information can be generated and transmitted in real time. The reason for this real-time design is that the state of the deposition may change rapidly during the operation of the internal mixer, and delayed processing cannot meet the needs of online self-cleaning. Timely provision of risk information ensures that the detachment action is synchronized with the state of the deposition, thereby improving the response capability of the self-cleaning system.

[0054] In specific implementation, first, the torque sensor, acoustic emission sensor and vibration sensor are calibrated to set the long-term average of the torque, the specified frequency range of the acoustic emission and the normal amplitude range of the vibration. Then, the normalization coefficient is adjusted through historical deposition data to ensure that the deposition activation coefficient is sensitive to risk and has a low false positive rate. During monitoring, the three signals are synchronously collected every second and the deposition activation coefficient is calculated. When it exceeds the preset threshold, a high-risk section mark is generated, and the mark information is transmitted to the subsequent step. This implementation ensures the stability and practicality of the technical logic.

[0055] Step two, real-time monitoring of rotor vibration amplitude by displacement sensor, comparing the monitoring value with the preset vibration amplitude, adjusting the output power of the vibration exciter according to the comparison result to keep the vibration amplitude within the allowable error range, and recording the vibration frequency, vibration amplitude, actual vibration amplitude, vibration duration, rotor displacement response and acceleration response data.

[0056] The step two includes the following contents:

[0057] Step 201, vibration parameter determination

[0058] In the vibration parameter determination stage, the specific time period and section of the vibration action are determined according to the high-risk section mark information transmitted in step one. The vibration parameters include vibration frequency, vibration amplitude and vibration duration.

[0059] The selection of vibration frequency is based on the natural frequency of the rotor material, which is set to a value close to the natural frequency to enhance the efficiency of deposit cracking by resonance effect. The natural frequency is determined by the rotor geometry and material properties, which can be determined in advance by experiment or finite element analysis. The vibration amplitude is set to low amplitude vibration, and the specific value is determined based on the hardness of the rotor material and the experimental determination of the bond strength of the deposit, to ensure that the rotor surface is not damaged while cracking the deposit. The vibration duration matches the duration of the high-risk section, ensuring that the vibration covers the entire marked section.

[0060] In the vibration excitation stage, the rotor wing end is excited by a low-amplitude torsional vibration for a specified period of time by a vibration exciter, which can use a piezoelectric ceramic or electromagnetic driver to generate a torsional vibration wave according to the preset vibration frequency and vibration amplitude. The torsional vibration wave propagates along the rotor wing end and acts on the interface between the deposit and the rotor to achieve the cracking of the deposit.

[0061] Step 202, deposit cracking mechanism

[0062] In the deposit cracking mechanism stage, low-amplitude torsional vibration achieves the release of the deposit through shear stress effect, fatigue fracture effect and micro-crack propagation. Shear stress effect manifests as shear stress generated by vibration at the interface between the deposit and the rotor. When the shear stress exceeds the bond strength of the deposit, the deposit cracks along the shear direction. Fatigue fracture effect periodically changes the stress state through continuous vibration, accumulates fatigue damage, and eventually leads to the fracture of the deposit. Micro-crack propagation is promoted by vibration energy, which causes the propagation of micro-cracks in the deposit and ultimately leads to the disintegration of the entire deposit.

[0063] Step 203, amplitude control

[0064] In the amplitude control stage, the vibration amplitude is monitored and adjusted in real time to ensure that the rotor surface does not wear to ensure safety. Real-time monitoring uses a displacement sensor to measure the actual vibration amplitude and compares it with the preset vibration amplitude. Define an allowable error range. If the actual vibration amplitude exceeds the preset vibration amplitude plus or minus the allowable error range, adjust the output power of the vibration exciter. If the actual vibration amplitude exceeds the upper limit of the preset vibration amplitude plus the allowable error, reduce the output power of the vibration exciter; if the actual vibration amplitude is lower than the lower limit of the preset vibration amplitude minus the allowable error, increase the output power of the vibration exciter; if the actual vibration amplitude is within the range of the preset vibration amplitude plus or minus the allowable error, maintain the output power of the vibration exciter unchanged.

[0065] In the data recording stage, the vibration frequency, vibration amplitude, actual vibration amplitude and vibration duration during the vibration process, as well as the displacement and acceleration response data of the rotor are recorded. These data provide support for the directional jet slag removal in step three and the torque model calibration in step four, ensuring the data continuity and traceability of the entire self-cleaning process.

[0066] According to the high-risk section marking information of step one, only the necessary section is excited to vibrate, reducing energy waste and unnecessary disturbance to the rotor. Deposits are mainly formed in high-risk sections, and targeted treatment can effectively improve cleaning efficiency. The advantage is to reduce unnecessary vibration to the equipment, while ensuring the targeting of the cleaning action.

[0067] Low amplitude vibration combined with real-time amplitude control ensures that the rotor surface is not damaged when the deposits are cracked, prolonging the service life of the equipment. High amplitude vibration may cause rotor surface wear or fatigue damage, while low amplitude vibration under real-time control can effectively avoid such problems. The integrity of the equipment can be protected, and maintenance costs can be reduced. By using a vibration frequency close to the natural frequency of the rotor material, the shear stress is amplified through resonance effect, improving the efficiency of deposit cracking. Resonance can concentrate vibration energy on the deposits, accelerating the cracking process. Cleaning time can be shortened and production efficiency can be improved.

[0068] The cracked deposit fragments create conditions for the directional jet blast of step three, and the data record supports the torque model optimization of step four. The data and state transfer between steps can achieve coordinated optimization of the self-cleaning process, which can improve the intelligent level of the overall system and ensure the persistence and stability of the cleaning effect.

[0069] In a specific implementation, first, according to the rotor material properties and deposit hardness, the vibration frequency, vibration amplitude and allowed error range are preset. Then, the high-risk section marking information of step one is received to determine the vibration start time and duration. During the specified time period, the vibration exciter excites torsional vibration according to the preset vibration frequency and vibration amplitude, and the displacement sensor monitors the actual vibration amplitude in real time. According to the comparison of the actual vibration amplitude and the preset vibration amplitude, the output power of the vibration exciter is adjusted. During the vibration process, the vibration frequency, vibration amplitude, actual vibration amplitude, vibration duration, and rotor displacement and acceleration response data are recorded and transmitted to steps three and four. This implementation ensures the stability and practicality of the technical logic.

[0070] Through the above steps, precise, efficient and safe cracking of the rotor wing end deposits is achieved, providing support for the stable operation of the internal mixer under high abrasion filler formulations.

[0071] Step three, measure the jet pressure through the micro-pressure sensor and convert it to momentum value, generate the erosion momentum signal through integral normalization processing; measure the scattering spectrum area difference through the light scattering sensor and generate the particle scattering signal through normalization processing; input the erosion momentum signal and particle scattering signal into the radial basis neural network to calculate the fusion coefficient; adjust the nozzle pressure and jet duration according to the fusion coefficient to optimize the blast effect; measure and record the blast quantity data through the mass flow meter.

[0072] The step three includes the following contents:

[0073] Step 301, calculation logic of fusion coefficient

[0074] The fusion coefficient is a numerical value calculated by comprehensively analyzing the erosion momentum signal and the particle scattering signal, which is used to represent the deslagging effect. The process of obtaining the erosion momentum signal is as follows: first, measure the pressure value of the jet acting on the deposit through the micro-pressure sensor, convert the pressure value into the momentum value of the jet, then calculate the momentum accumulation along the predicted action range of the jet, and finally normalize the accumulation result to generate the erosion momentum signal.

[0075] The process of obtaining the particle scattering signal is as follows: measure the scattering spectrum of the deposit particles removed by the jet through the light scattering sensor, calculate the difference of the spectrum area, and normalize the difference value to generate the particle scattering signal. Then, the erosion momentum signal and the particle scattering signal are input into the radial basis neural network. The radial basis neural network maps these two signals through a series of pre-set radial basis functions, each radial basis function has a specific center position and expansion range, and finally outputs the fusion coefficient through weighted summation. The value range of the fusion coefficient is set between 0 and 1, the closer the value is to 1, the better the deposit removal effect.

[0076] Step 302, adjustment logic of jet parameters

[0077] The fusion coefficient is used to dynamically adjust the nozzle pressure and jet duration of the jet to ensure that the deslagging effect reaches the expected value. The adjustment of the nozzle pressure is based on a pre-set initial pressure value, according to the deviation between the fusion coefficient and the target value 1, combined with a pressure adjustment coefficient for increase or decrease processing. When the fusion coefficient is low, it indicates that the deposit removal effect is poor, and the nozzle pressure is increased to enhance the jet force; when the fusion coefficient is close to 1, it indicates that the removal effect is good, and the nozzle pressure is maintained or slightly reduced to save energy. The adjustment of the jet duration is based on a pre-set initial time value, according to the deviation between the fusion coefficient and the target value 1, combined with a time adjustment coefficient for extension or shortening processing. When the fusion coefficient is low, the jet duration is extended to increase the action time; when the fusion coefficient is close to 1, the jet duration is maintained or slightly shortened to improve efficiency. In this way, the jet parameters can be adjusted in real time according to the fusion coefficient.

[0078] During the execution of the directional jet, the mass of the removed deposit is measured in real time by the mass flow meter, and this data is recorded as the deslagging amount. The deslagging amount data reflects the actual amount of removed deposit, which is used for calibration of the digital twin torque model in the subsequent steps. The recording process directly obtains data from the mass flow meter and stores it, ensuring the accuracy and traceability of the data.

[0079] The directional jet acts only on the rotor surface area that needs to be cleaned according to the high-risk section marking information transmitted in step one, avoiding the ineffective action of the jet on irrelevant areas, thereby reducing energy consumption. By calculating the fusion coefficient in real time and adjusting the nozzle pressure and jet duration accordingly, the jet parameters can be dynamically optimized according to the actual situation of sediment removal, improving the efficiency of slag removal. By fusing the erosion momentum signal and particle scattering signal using a radial basis neural network, the slag removal effect can be accurately evaluated, avoiding repeated operations or insufficient removal due to improper parameter settings. By effectively removing the deposits on the rotor surface, the surface roughness is reduced, reducing the likelihood of new particles adhering or embedding, thereby slowing down the wear rate of the equipment and extending its service life.

[0080] In specific implementation, first set the initial value of the nozzle pressure, the initial value of the jet duration, the pressure adjustment coefficient, the time adjustment coefficient and the related parameters of the radial basis neural network, including the center position, the expansion range and the weight value of the radial basis function. Then, receive the high-risk section marking information transmitted in step one to determine the starting time and specific area range of the jet action.

[0081] During the jet execution process, the erosion momentum signal is collected in real time by the micro-pressure sensor, the particle scattering signal is collected in real time by the light scattering sensor, and the fusion coefficient is calculated using a radial basis neural network. According to the fusion coefficient value, adjust the nozzle pressure and jet duration to perform directional jet operation, while measuring and recording the slag removal amount by the mass flow meter, and transmitting the data to step four for subsequent processing.

[0082] Step four, calculate the torque baseline value by receiving the deposition activation coefficient and the amount of sediment removal, adjust the digital twin torque model parameters using the deposition activation coefficient, calculate the end point determination threshold value according to the change of the torque baseline value, combine the real-time collected torque curve, acoustic emission wave and vibration signal to calculate the torque prediction value by the digital twin torque model, when the torque prediction value exceeds the end point determination threshold value, mark it as a high-risk section and notify the reinforcement cleaning, adjust the digital twin torque model parameters according to the deviation between the actual torque and the torque prediction value.

[0083] The step four includes the following contents;

[0084] Step 401, torque baseline correction

[0085] The processing technology logic of torque baseline correction is based on the reference torque value of the device under the influence of no sediment. First, the current torque baseline value, i.e. the torque reference value of the device in the ideal state, is obtained; then, the recorded sediment removal amount in step three is obtained, which represents the total amount of sediment removed during the cleaning process; then, the sediment removal amount is multiplied by a predetermined influence coefficient, which reflects the reduction of the torque baseline for each unit of sediment removal; finally, the current torque baseline value is subtracted from the product to obtain a new torque baseline value as the corrected reference torque reference. The whole process aims to dynamically update the torque baseline according to the actual situation of sediment removal.

[0086] The removal of sediment reduces the resistance of the device during operation, thereby changing the torque reference value of the device under the influence of no sediment. If the torque baseline is not corrected, the model will not accurately reflect the running state of the device after cleaning, which may cause the subsequent torque prediction to deviate from the actual value. By introducing the amount of sediment removal and the influence coefficient for calculation, the torque baseline can be quantitatively adjusted to keep it consistent with the actual state of the device.

[0087] The accuracy of torque prediction can be improved, and the corrected torque baseline can truly reflect the running characteristics of the device after cleaning, ensuring that the input data of the digital twin torque model matches the physical state of the device and avoiding prediction errors caused by reference value deviation.

[0088] Step 402, model parameter adjustment

[0089] The processing technology logic of model parameter adjustment takes the sediment activation coefficient as the core, which is used to dynamically optimize the internal parameters of the digital twin torque model. First, the sediment activation coefficient is obtained from step one, which is a sediment risk indicator calculated based on device operation data; then, the sediment activation coefficient is input into the digital twin torque model as an input variable; then, the model adjusts the parameters related to the influence of sediment, such as the contribution weight of sediment to torque increment, according to the value of the sediment activation coefficient, so that these parameters can reflect the current state of the sediment; finally, the updated parameters are applied to the torque calculation process of the model to adapt to the influence of sediment changes on device operation.

[0090] The presence and changes of sediment will directly affect the running torque of the device, and the digital twin torque model needs to adjust its internal parameters according to the real-time sediment state to ensure the accuracy of the prediction results. The sediment activation coefficient, as a quantitative indicator of sediment risk, can provide dynamic adjustment basis for the model, enabling the model to adapt to different sediment states.

[0091] The adaptability of the digital twin torque model can be enhanced. By adjusting the parameters in real time according to the deposition activation coefficient, the model can maintain high prediction accuracy even when the amount of deposits changes. This not only improves the reliability of torque prediction, but also provides a guarantee for the accuracy of subsequent cleaning decisions.

[0092] Step 402, endpoint determination threshold adjustment

[0093] The processing technology logic of endpoint determination threshold adjustment dynamically updates the torque standard for determining whether reinforcement cleaning is needed based on the change of torque baseline. First, the current endpoint determination threshold is obtained, which is the torque reference value for triggering step five reinforcement cleaning. Then, the torque baseline drop is calculated, which is the difference between the corrected torque baseline value and the corrected torque baseline value. Next, the torque baseline drop is multiplied by a pre-set adjustment coefficient, which represents the degree of influence of torque baseline change on the endpoint determination threshold. Finally, the current endpoint determination threshold is subtracted by the product to obtain the new endpoint determination threshold as the adjusted cleaning trigger standard.

[0094] As the torque baseline decreases, the operating state of the equipment after cleaning changes, and the original endpoint determination threshold may no longer be applicable to the new operating conditions. If the threshold is not adjusted according to the change of torque baseline, it may cause the cleaning action to be triggered too early or too late, which cannot meet the actual needs of the equipment. By introducing the torque baseline drop and the adjustment coefficient, the endpoint determination threshold can be updated synchronously with the equipment state. The trigger condition of reinforcement cleaning can be optimized. The adjusted endpoint determination threshold can accurately reflect the torque demand of the equipment after cleaning, avoiding insufficient cleaning or excessive cleaning due to improper threshold setting, thereby improving the efficiency and pertinence of the cleaning process.

[0095] Step 403, torque prediction and feedback

[0096] The processing technology logic of torque prediction and feedback calculates the future torque value through the digital twin torque model, and optimizes the model parameters using the feedback mechanism. First, collect real-time operating data of the equipment, including torque curve, acoustic emission wave and vibration signal. Then, combine these real-time data with the updated torque baseline and adjusted model parameters, and input them into the digital twin torque model. Next, the model calculates the future torque prediction value based on the input data, which represents the torque change trend that may occur under the current state of the equipment. Then, compare the torque prediction value with the new endpoint determination threshold. If the prediction value exceeds the threshold, mark it as a high-risk section and notify step five to perform reinforcement cleaning. Finally, compare the deviation between the actual measured torque value and the prediction value, and adjust the model internal parameters according to the deviation size to make the prediction result closer to the actual value.

[0097] By predicting the torque trend in real time, potential risk sections in the equipment operation can be found in time to ensure that the cleaning action is triggered when necessary. At the same time, feedback adjustment is made using the deviation between the actual torque and the predicted value to continuously optimize the prediction ability of the model to adapt to the dynamic changes of the equipment state.

[0098] The precise triggering of the cleaning action and the continuous improvement of the model performance are achieved, the torque prediction value provides a forward-looking basis for decision-making, so that the reinforcement cleaning can be efficiently performed on the high-risk sections, and the feedback mechanism improves the long-term prediction accuracy of the model through deviation correction, thereby enhancing the stability and reliability of the entire system.

[0099] In specific implementation, first, the influence coefficient of the deposit removal amount on the torque baseline and the adjustment coefficient of the end point determination threshold need to be set in advance, and the internal parameters of the digital twin torque model are initialized. Then, the deposit activation coefficient is obtained from step one, and the deposit removal amount is obtained from step three. Based on the deposit removal amount and the influence coefficient, a new torque baseline is calculated, and the model parameters are adjusted using the deposit activation coefficient. Subsequently, a new end point determination threshold is calculated according to the change of the torque baseline and the adjustment coefficient. Then, the torque prediction value is calculated by the digital twin torque model in combination with the real-time collected torque curve, acoustic emission wave and vibration signal, and the updated parameters. If the prediction value exceeds the new end point determination threshold, it is marked as a high-risk section and the reinforcement cleaning in step five is notified. Finally, the model parameters are adjusted according to the deviation between the actual torque and the prediction value to form a closed-loop optimization.

[0100] Step four realizes real-time correction of the torque baseline, dynamic adjustment of the model parameters, synchronous update of the end point determination threshold, and precise feedback of the torque prediction, ensuring that the digital twin torque model can continuously adapt to the equipment operating state to provide reliable technical support for online self-cleaning, thereby improving the accuracy and stability of the system.

[0101] Step five, through the ultrasonic transducer and the plasma generator, the focused ultrasonic wave and the low-temperature plasma composite pulse are applied synchronously in the high-risk section, the residual hard shell is detached by using the dual mechanisms of acoustic wave fragmentation and plasma bombardment, after cleaning, the new deposit activation coefficient is calculated using multi-source sensing data and compared with the original deposit activation coefficient, if the new deposit activation coefficient is less than the original deposit activation coefficient, the original deposit activation coefficient is updated, if the new deposit activation coefficient is not less than the original deposit activation coefficient, the ultrasonic power or the plasma pulse frequency is adjusted and the cleaning is repeated;

[0102] The step five includes the following contents:

[0103] Step 501, reinforcement cleaning trigger condition

[0104] The decision is made based on the torque prediction value and the end point determination threshold value, wherein the future torque prediction value representing the torque change trend that may occur under the current operating state of the equipment is calculated first by the digital twin torque model; then the end point determination threshold value, which is a cleaning trigger standard dynamically adjusted according to the sediment removal amount and the sediment activation coefficient, is obtained; then the future torque prediction value is compared with the end point determination threshold value; when the future torque prediction value exceeds the end point determination threshold value, it is determined that the influence of the residual hard shell on the operation of the equipment reaches a significant degree, triggering the reinforced cleaning action. Through the real-time state and the prediction trend of the equipment, it is determined whether further cleaning operation is needed.

[0105] The continuous existence of the residual hard shell reduces the operating efficiency of the equipment and affects the accuracy of torque prediction. By comparing the future torque prediction value with the dynamically adjusted end point determination threshold value, the cleaning demand can be identified in time to ensure that the equipment is processed in a high-risk operating state. Finally, the pertinence and timeliness of the cleaning action are improved, and the reinforced cleaning can be started before the residual hard shell significantly affects the performance of the equipment, avoiding the performance degradation and energy consumption increase caused by delayed cleaning.

[0106] Step 502, focused ultrasound parameter setting

[0107] The optimization is based on the physical characteristics of the residual hard shell, wherein the ultrasonic frequency is determined according to the thickness and material characteristics of the residual hard shell to match the wavelength of the sound wave with the characteristic size of the residual hard shell, so as to improve the focusing efficiency of the sound wave energy; then the ultrasonic power is set according to the bonding strength of the residual hard shell to ensure that the ultrasonic wave provides sufficient energy to make the residual hard shell crack, while avoiding damage to the rotor surface; finally, the ultrasonic action time is aligned with the duration of the high-risk section to ensure that the cleaning covers the entire high-risk section. The whole process realizes the efficient detachment of the residual hard shell through precise parameter setting.

[0108] Focused ultrasound can concentrate sound wave energy on the residual hard shell and induce the expansion of internal micro-cracks, thereby realizing detachment. By adjusting the ultrasonic frequency and power according to the characteristics of the residual hard shell, the detachment effect can be maximized and the equipment surface can be protected. Finally, the cleaning efficiency and safety are improved, and the precise parameter setting ensures that the ultrasonic wave effectively acts on the residual hard shell, quickly completes the detachment, and at the same time avoids damage to the rotor surface, prolonging the service life of the equipment.

[0109] Step 503, low-temperature plasma parameter setting

[0110] The processing technology logic of low-temperature plasma parameter setting is optimized based on the surface characteristics of residual hard shell and ultrasonic synergy. Among them, first, the plasma density is set according to the surface adhesion of the residual hard shell, so as to control the activity of the plasma and optimize the modification effect on the surface of the residual hard shell; then the pulse frequency of the plasma is coordinated with the ultrasonic frequency to form a synchronous pulse mode, which improves the composite cleaning efficiency; finally, the pulse duration is set according to the desorption requirement of the residual hard shell, so as to ensure the thoroughness of the cleaning process. The whole process weakens the adhesion of the residual hard shell to the rotor through plasma bombardment, and assists ultrasonic desorption.

[0111] Low-temperature plasma weakens the adhesion of residual hard shell through surface bombardment, and forms a complement with the fragmentation effect of ultrasonic waves to improve the desorption efficiency. Through the pulse mode coordinated with the ultrasonic frequency, the composite cleaning effect is further enhanced, and the comprehensiveness and thoroughness of cleaning are finally improved. The synergy of low-temperature plasma and focused ultrasound can more effectively remove residual hard shell, reduce the omission in the cleaning process, and ensure the cleanliness of the equipment surface.

[0112] Step 504, composite pulse application

[0113] The processing technology logic of composite pulse application is based on the synchronous execution of the marking information of high-risk sections. Among them, during the duration of the high-risk section, the focused ultrasonic waves are applied through the ultrasonic transducer, and the low-temperature plasma is applied through the plasma generator at the same time, forming a composite pulse effect; the focused ultrasonic waves concentrate on the residual hard shell, producing internal stress to make it fragment; the low-temperature plasma bombards the surface of the residual hard shell, weakening its adhesion to the rotor and cleaning the surface residues; the whole process realizes the rapid desorption of the residual hard shell through the dual mechanism of ultrasonic fragmentation and plasma bombardment.

[0114] The composite pulse technology fully utilizes the advantages of ultrasonic waves and low-temperature plasma, and complements the different aspects of residual hard shell, improving the cleaning efficiency. Through synchronous application in the high-risk section, the pertinence and timeliness of cleaning action are ensured. The cleaning time can be significantly shortened and the cleaning effect can be improved. The synergy of the dual mechanism makes the residual hard shell desorb rapidly, reduces the equipment downtime, and at the same time ensures the thoroughness of cleaning, maintains the stability of equipment performance.

[0115] Step 505, deposition activation coefficient update

[0116] The process logic of the deposition activation coefficient update is evaluated and fed back based on the state of the cleaned equipment. Among them, after the reinforcement cleaning is completed, the deposition activation coefficient is recalculated using the multi-source sensing data in step one, including torque curve, acoustic emission wave and vibration signal, to obtain a new deposition activation coefficient; then the new deposition activation coefficient is compared with the original deposition activation coefficient; if the new deposition activation coefficient is less than the original deposition activation coefficient, it indicates that the cleaning is effective, and the new deposition activation coefficient is fed back to step one to update the original deposition activation coefficient; if the new deposition activation coefficient is not less than the original deposition activation coefficient, the ultrasonic power or the plasma pulse frequency is adjusted, and the cleaning is repeated until the new deposition activation coefficient is less than the original deposition activation coefficient. The whole process realizes the closed-loop optimization of the cleaning effect through data feedback.

[0117] In use, the deposition activation coefficient as a quantitative indicator of the risk of deposition can intuitively reflect the cleaning effect, and by comparing with the original deposition activation coefficient, it can be judged whether the cleaning achieves the expected effect, and accordingly the cleaning parameters are adjusted or the cleaning is repeated to ensure thorough cleaning. Finally, the continuous optimization and data closed loop of the cleaning process are realized.

[0118] By feeding back the updated deposition activation coefficient to step one, the cleaning strategy can be dynamically adjusted to adapt to the changing needs of high-abrasion formulations and improve the intelligent level of online self-cleaning.

[0119] In specific implementation, first, the focused ultrasonic parameters and low-temperature plasma parameters are set according to the characteristics of the residual hard shell, and the high-risk section marking information transmitted from step four is received to determine the starting time and duration of the reinforcement cleaning. In the high-risk section, the composite pulse is applied synchronously by the ultrasonic transducer and the plasma generator to perform reinforcement cleaning. After the reinforcement cleaning is completed, the deposition activation coefficient is recalculated using multi-source sensing data, and compared with the original deposition activation coefficient. If the new deposition activation coefficient is less than the original deposition activation coefficient, it is fed back to step one to update the original deposition activation coefficient; if the new deposition activation coefficient is not less than the original deposition activation coefficient, the cleaning parameters are adjusted and the cleaning is repeated until the condition is met.

[0120] Those skilled in the art can appreciate that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Professionals can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0121] Those skilled in the art can clearly understand the specific working process of the system, the device and the unit described above can refer to the corresponding process in the foregoing method embodiments for description convenience and brevity, and details are not described herein.

[0122] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are merely illustrative, for example, the division of the units is only some logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0123] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected to achieve the purpose of the embodiment scheme according to actual needs.

[0124] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A self-cleaning method for in-line mixing based on torque prediction synergy, characterized in that: include, Torque curves, acoustic emission waves, and vibration signals are acquired in real time. The deposition activation coefficient is calculated and high-risk sections are marked as the triggering basis for low-amplitude torsional vibration desorption. Based on the high-risk section markings, low-amplitude torsional vibrations were induced at the rotor blade tip to lyse the deposits through shear stress and fatigue fracture mechanisms, and the vibration amplitude was controlled in real time to protect the rotor. Based on the high-risk section marking information, directional jetting is adopted, and the nozzle pressure and jetting duration are automatically adjusted by calculating the slag discharge efficiency coefficient in real time; By using a digital twin torque model, combined with the deposition activation coefficient and slag discharge volume, the torque baseline is corrected in real time and the endpoint judgment threshold is adjusted, so that the torque prediction continuously matches the equipment status. When the torque prediction still shows local high-risk sections, a combined pulse of focused ultrasound and low-temperature plasma is applied to rapidly desorb the residual hard shell, and the updated deposition activation coefficient is fed back.

2. The online self-cleaning method for mixing based on torque prediction synergy according to claim 1, characterized in that: Simultaneously acquire the torque curve, acoustic emission waves, and vibration signals of the internal mixer rotor; The ratio of the difference between the maximum torque and the long-term average to the long-term average is calculated as the torque component, the sum of the power spectral density of the acoustic emission wave within the specified frequency range is calculated as the acoustic emission component, and the ratio of the difference between the maximum vibration amplitude and the average amplitude to the average amplitude is calculated as the vibration component.

3. The online self-cleaning method for mixing based on torque prediction synergy according to claim 2, characterized in that: The deposition activation coefficient is obtained by multiplying the torque component, acoustic emission component, and vibration component by preset normalization coefficients and then summing them. When the deposition activation coefficient exceeds a preset threshold, the marking information of high-risk sections is generated, including the start time and duration. Output the tagging information.

4. The online self-cleaning method for internal mixing based on torque prediction synergy according to claim 3, characterized in that: The rotor vibration amplitude is monitored in real time by a displacement sensor. The monitored value is compared with the preset vibration amplitude. The output power of the vibration exciter is adjusted according to the comparison result to keep the vibration amplitude within the allowable error range. Record the vibration frequency, vibration amplitude, actual vibration amplitude, vibration duration, rotor displacement response, and acceleration response data.

5. The online self-cleaning method for mixing based on torque prediction synergy according to claim 4, characterized in that: The jet pressure is measured by a micro-pressure sensor and converted into a momentum value. After integration and normalization, an erosion momentum signal is generated. The difference in the scattering spectrum area is measured by a light scattering sensor and normalized to generate a particle scattering signal.

6. The online self-cleaning method for mixing based on torque prediction synergy according to claim 5, characterized in that: The erosion momentum signal and particle scattering signal are input into a radial basis neural network to calculate the fusion coefficient; the nozzle pressure and jet duration are adjusted according to the fusion coefficient to optimize the slag discharge effect; and the slag discharge data are measured and recorded by a mass flow meter.

7. The online self-cleaning method for mixing based on torque prediction synergy according to claim 6, characterized in that: The torque baseline value is calculated by receiving the deposition activation coefficient and the amount of sediment removed. The parameters of the digital twin torque model are adjusted using the deposition activation coefficient, and the endpoint determination threshold is calculated based on the change in the torque baseline value.

8. The online self-cleaning method for mixing based on torque prediction synergy according to claim 7, characterized in that: The torque prediction value is calculated by combining the real-time acquired torque curve, acoustic emission wave and vibration signal through a digital twin torque model; When the predicted torque value exceeds the endpoint determination threshold, it is marked as a high-risk section and reinforcement cleaning is notified. The parameters of the digital twin torque model are adjusted according to the deviation between the actual torque and the predicted torque value.

9. The online self-cleaning method for mixing based on torque prediction synergy according to claim 8, characterized in that: A combined pulse of focused ultrasound and cryogenic plasma is applied simultaneously in high-risk areas using an ultrasonic transducer and a plasma generator. The residual hard shell was removed by a dual mechanism of acoustic fragmentation and plasma bombardment. After cleaning, the activation coefficient of the new deposition was calculated using multi-source sensing data and compared with the activation coefficient of the original deposition.

10. The online self-cleaning method for mixing based on torque prediction synergy according to claim 9, characterized in that: When the activation coefficient of the new deposition is less than the activation coefficient of the original deposition, the activation coefficient of the original deposition is updated. When the activation coefficient of the new deposition is not less than that of the original deposition, adjust the ultrasonic power or plasma pulse frequency and repeat the cleaning process.

Images