Automatic control system for flotation liquid level based on rheological model and cross-correlation analysis

Abstract

The present application relates to supercritical fluid extraction recovery and industrial automation control field, specifically to the floatation liquid level automatic control system based on rheological model and cross-correlation analysis, including: receiving the real-time working condition data of extraction container collected by multimodal sensor array; calling the pre-stored Peng-Robinson equation of state model, calculating the theoretical static liquid level height without foam and boiling state under the current working condition; calling the pre-stored fluid rheology parameter library, generating the theoretical disturbed liquid level waveform; performing difference calculation: if the cross-correlation coefficient is greater than or equal to the preset confidence threshold, sending the start instruction to the deslagging actuator to execute the physical deslagging action; otherwise, performing digital low-pass filtering processing on the real-time observed liquid level height and keeping the feeding state; the present application solves the problem of reference surface drift caused by continuous boiling or severe environmental fluctuations in the supercritical extraction process, and makes the liquid level control based on rigorous thermodynamics.

Description

Technical Field

[0001] This invention relates to the field of supercritical fluid extraction and recovery and industrial automation control technology, specifically to an automatic flotation liquid level control system based on rheological models and cross-correlation analysis. Background Technology

[0002] In the multi-stage flotation process for supercritical fluid extraction and waste battery electrolyte recovery, the inside of the extraction container is in a high-pressure environment with drastic thermodynamic fluctuations. The fluid inside the container not only experiences normal liquid level rises and falls, but also physical foam accumulation, gas-liquid boiling and flash evaporation induced by pressure fluctuations, and electromagnetic noise interference from the sensor itself.

[0003] Existing liquid level control schemes mostly rely on traditional sensors to directly read signals and process them using simple smoothing filters based on historical data. However, this approach struggles to identify the physical state from simple numerical values ​​amidst complex signals that mix actual liquid level, foam reflection, and boiling expansion. In particular, traditional filtering cannot eliminate systematic biases caused by artificially inflated volumes due to continuous boiling, making it impossible for the system to accurately distinguish whether liquid level fluctuations are caused by physical foam accumulation or non-physical gas-liquid boiling. This easily leads to misjudgments in control logic, such as mistaking boiling for foam and executing incorrect mechanical scraping, or erroneously triggering a feed stop due to false high liquid level signals caused by noise, severely hindering production continuity and safety. Therefore, how to construct an automatic control mechanism that can accurately distinguish between physical foam, boiling expansion, and sensor noise based on physical rheological mechanisms has become an urgent technical problem to be solved. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides an automatic flotation liquid level control system based on rheological models and cross-correlation analysis. Specifically, the technical solution of this invention is as follows:

[0005] An automatic flotation liquid level control system based on rheological models and cross-correlation analysis includes a multimodal sensor array, a slag discharge actuator, and a controller electrically connected to the multimodal sensor array and the slag discharge actuator; the controller is configured to perform the following steps:

[0006] It receives real-time operating data of the extraction vessel collected by a multimodal sensor array. The real-time operating data includes temperature values, pressure values, feed and discharge mass flow rates, and time-series real-time observed liquid level height with fluctuations.

[0007] The pre-stored Peng-Robinson equation of state model is invoked, and the theoretical static liquid level height under the current working conditions of no foam and no boiling is calculated based on temperature and pressure values.

[0008] The pre-stored fluid rheology parameter library is called, and a preset foam layer thickness model and flash volume expansion model are superimposed on the theoretical static liquid level height to generate the theoretical disturbed liquid level waveform.

[0009] Perform differential calculation: Subtract the theoretical static liquid level height from the real-time observed liquid level height in the time series to obtain the actual residual sequence, and subtract the theoretical static liquid level height from the theoretical disturbed liquid level waveform to obtain the theoretical disturbance sequence;

[0010] Calculate the cross-correlation coefficient between the actual residual sequence and the theoretical interference sequence;

[0011] In response to a cross-correlation coefficient greater than or equal to a preset confidence threshold, the liquid level fluctuation is determined to be caused by physical foam accumulation, and a start command is sent to the slag discharge actuator to perform physical slag scraping action;

[0012] In response to the cross-correlation coefficient being less than the confidence threshold, it is determined that the liquid level fluctuation is caused by non-physical sensor noise. Digital low-pass filtering is performed on the real-time observed liquid level height while maintaining the feeding state.

[0013] Optionally, the controller performs the following steps when calculating the theoretical static liquid level height:

[0014] Using Henry's Law and a pre-defined solubility constant, the theoretical volume of solute dissolved in the solvent is calculated.

[0015] Calculate the theoretical density of the liquid mixture based on the temperature and pressure values;

[0016] Based on the integral values ​​of the influent and effluent mass flow rates, the theoretical dissolved volume, and the theoretical density, the physical height of the pure liquid phase within the geometric container, excluding gas phase voids, is calculated and used as the theoretical static liquid level.

[0017] Optionally, the fluid rheology parameter library includes surface tension attenuation factors and viscosity coefficients; when generating the theoretically disturbed liquid level waveform, the controller specifically performs the following steps:

[0018] The theoretical half-life of bubble rupture is calculated based on the surface tension decay factor.

[0019] Constructing a non-Newtonian fluid support force model using viscosity coefficient;

[0020] By combining the theoretical half-life and the non-Newtonian fluid support force model, the virtual thickness of the foam layer that varies with time is calculated, and this virtual thickness is superimposed on the theoretical static liquid level to form a foam layer thickness model.

[0021] Optionally, the fluid rheology parameter library also includes a supercritical fluid density fluctuation factor; when generating the theoretically disturbed liquid level waveform, the controller specifically performs the following steps:

[0022] Monitor the rate of change of pressure values ​​over time. When the rate of change exceeds the preset pressure drop threshold, trigger the flash evaporation simulation logic.

[0023] Based on the supercritical fluid density fluctuation factor, the liquid phase volume expansion rate caused by local pressure reduction is calculated.

[0024] The volume expansion rate is converted into the corresponding imaginary liquid level height, and this imaginary liquid level height is superimposed on the theoretical static liquid level height to form a flash volume expansion model.

[0025] Optionally, the controller performs the following steps when calculating the cross-correlation coefficient:

[0026] Project the actual residual sequence and the theoretical disturbance sequence into the same time window;

[0027] Calculate the waveform slope values ​​of the two sequences at corresponding time points;

[0028] Calculate the Pearson correlation coefficient between the waveform slope values ​​of the two sequences, and use this Pearson correlation coefficient as the cross-correlation coefficient.

[0029] Optionally, when performing digital low-pass filtering, the controller specifically executes the following steps:

[0030] Identify signal components in the real residual sequence that cannot be aligned in the time domain with the theoretical interference sequence;

[0031] The signal components are labeled as random electromagnetic interference noise;

[0032] A moving average filtering algorithm is used to remove signal components that are marked as random electromagnetic interference noise, generating smoothed true liquid level data to prevent the system from stopping the feeding action due to false high liquid level signals.

[0033] Optionally, the multimodal sensor array includes an array of pressure transmitters and an RF admittance level gauge mounted at different heights in the extraction vessel;

[0034] The real-time liquid level height is calculated by weighting the static pressure difference data measured by the pressure transmitter array and the capacitance change data measured by the radio frequency admittance level gauge.

[0035] Optionally, the flotation process system is a waste battery electrolyte recovery system, and the fluid system in the extraction container is a binary or ternary mixture of supercritical carbon dioxide and electrolyte.

[0036] The interference injection simulation module is used to simulate the liquid level fluctuation characteristics caused by carbon dioxide boiling or lithium salt crystallization entrainment.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. This system abandons the traditional approach of relying on historical data for smoothing. By introducing a state equation model, it calculates the theoretical static liquid level height in a foam-free and boiling-free state based on real-time thermodynamic parameters. This innovation provides an absolute physical truth reference for liquid level monitoring, solves the problem of reference surface drift caused by continuous boiling or drastic environmental fluctuations during supercritical extraction, and establishes liquid level control on a rigorous thermodynamic basis.

[0039] 2. This system adopts a synthesis-comparison logic and actively constructs a foam layer thickness and flash expansion model using fluid rheology parameters, transforming expert experience into mathematical waveforms. By performing cross-correlation analysis between real residuals and theoretical physical characteristics, it achieves a leap from black-box statistics to white-box physical determinism, accurately distinguishing between physical foam accumulation and non-physical pressure flash interference, and greatly reducing the false alarm rate.

[0040] 3. By calculating the correlation between the waveform slopes of the real sequence and the theoretical sequence, this system can identify the dynamic trend of liquid level changes rather than simply the numerical amplitude. This topological mapping-based technology enables the system to accurately lock physical properties even when physical properties such as foam growth rate and burst half-life fluctuate significantly. This effectively filters out random electromagnetic interference and sensor drift, ensuring identification stability under complex high-pressure environments.

[0041] 4. This system utilizes a multimodal sensor array and an inverse variance weighted fusion algorithm. The system can dynamically adjust data weights based on the real-time signal-to-noise ratio and is coupled with targeted digital filtering strategies. In high-risk scenarios such as waste battery electrolyte recycling, this mechanism can not only reliably capture liquid level details, but also effectively prevent erroneous shutdowns or feeding stops triggered by false high liquid level signals. While ensuring equipment safety, it significantly improves the continuity of industrial production. Attached Figure Description

[0042] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0043] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0045] Example 1:

[0046] Please see Figure 1 An automatic flotation liquid level control system based on rheological models and cross-correlation analysis includes a multimodal sensor array, a slag discharge actuator, and a controller electrically connected to the multimodal sensor array and the slag discharge actuator; the controller is configured to perform the following steps:

[0047] It receives real-time operating data of the extraction vessel collected by a multimodal sensor array. The real-time operating data includes temperature values, pressure values, feed and discharge mass flow rates, and time-series real-time observed liquid level height with fluctuations.

[0048] The pre-stored Peng-Robinson equation of state model is invoked, and the theoretical static liquid level height under the current working conditions of no foam and no boiling is calculated based on temperature and pressure values.

[0049] The pre-stored fluid rheology parameter library is called, and a preset foam layer thickness model and flash volume expansion model are superimposed on the theoretical static liquid level height to generate the theoretical disturbed liquid level waveform.

[0050] Perform differential calculation: Subtract the theoretical static liquid level height from the real-time observed liquid level height in the time series to obtain the actual residual sequence, and subtract the theoretical static liquid level height from the theoretical disturbed liquid level waveform to obtain the theoretical disturbance sequence;

[0051] Calculate the cross-correlation coefficient between the actual residual sequence and the theoretical interference sequence;

[0052] In response to a cross-correlation coefficient greater than or equal to a preset confidence threshold, the liquid level fluctuation is determined to be caused by physical foam accumulation, and a start command is sent to the slag discharge actuator to perform physical slag scraping action;

[0053] In response to the cross-correlation coefficient being less than the confidence threshold, it is determined that the liquid level fluctuation is caused by non-physical sensor noise. Digital low-pass filtering is performed on the real-time observed liquid level height while maintaining the feeding state.

[0054] This embodiment provides an automatic flotation level control system based on rheological models and cross-correlation analysis. The system aims to address the pain point that traditional level detection methods in supercritical fluid environments are unable to distinguish between physical foam accumulation, gas-liquid boiling, and sensor false noise.

[0055] The system in this embodiment includes a multimodal sensor array, a slag removal actuator, and a controller electrically connected to the multimodal sensor array and the slag removal actuator. The multimodal sensor array refers to a set of sensing devices capable of simultaneously acquiring fluid thermodynamic parameters and interfacial physical properties, with the aim of constructing a comprehensive flow field state observation view. The slag removal actuator refers to a mechanical device, such as a scraper or overflow valve, used for physically removing surface scum or foam, and is controlled by the controller's commands. The controller is configured to run specific physical calculation and logical judgment programs, specifically executing the following steps:

[0056] The controller receives real-time operating data of the extraction container collected by the multimodal sensor array;

[0057] Real-time operating data specifically includes:

[0058] Temperature value : The temperature of the fluid inside the vessel, collected in real time by thermocouples;

[0059] Pressure value The pressure inside the vessel is collected in real time by a pressure transmitter.

[0060] Infeed and discharge mass flow rates : The fluid inflow and outflow rate measured by a mass flow meter;

[0061] Real-time observation of liquid level height including fluctuations in time series This is the raw liquid level signal directly read by the sensor, which is a mixture of the actual liquid level, foam layer reflection, false expansion caused by boiling, and electromagnetic noise.

[0062] The controller invokes a pre-stored state equation model;

[0063] The Peng-Robinson equation of state model refers to a semi-empirical equation used to calculate the thermodynamic state of non-ideal fluids. In this embodiment, it is innovatively used to calculate the zero potential energy surface of the liquid level.

[0064] Temperature values ​​collected by sensors and pressure values The controller calculates the theoretical static liquid level height under the current operating conditions, where there is no foaming and no boiling. ;

[0065] This approach abandons the erroneous practice of obtaining baseline values ​​through historical liquid level smoothing filtering, a common method in traditional techniques. This is because, during supercritical fluid extraction, smoothing of historical data cannot eliminate persistent systematic biases, such as volume expansion caused by continuous boiling. Instead, it utilizes the EOS equation to calculate... It represents the physical truth value of a physically pure and static layered interface, providing an absolute physical reference frame for subsequent difference calculations;

[0066] The controller calls upon a pre-stored library of fluid rheological parameters, based on the theoretical static liquid level height calculated above. Based on this, a pre-set foam layer thickness model and flash volume expansion model are superimposed to generate the theoretical disturbed liquid level waveform. ;

[0067] This process employs a synthesis-comparison logic; the system does not directly identify foam, but rather uses rheological parameters from an expert knowledge base, such as surface tension and viscosity, to actively synthesize a waveform in digital space that a sensor would have if foam were present.

[0068] The controller performs the following mathematical operations:

[0069] Real-time monitoring of liquid level height using time series Subtract the theoretical static liquid level height To obtain the actual residual sequence This sequence represents all signal components that deviate from the ideal still liquid surface in reality, including real abnormal foam, real disturbing boiling, and meaningless noise.

[0070] The theoretical disturbed liquid level waveform Subtract the theoretical static liquid level height The theoretical interference sequence was obtained. This sequence contains only theoretical features caused purely by physical mechanisms of foaming or boiling, and does not contain any random environmental noise.

[0071] The controller calculates the actual residual sequence. With theoretical interference sequence Cross-correlation coefficients between them;

[0072] When the cross-correlation coefficient is greater than or equal to a preset confidence threshold, such as 0.95, the system determines that the liquid level fluctuation is caused by physical foam accumulation; at this time, the controller sends a start command to the slag discharge actuator to perform physical slag scraping action.

[0073] This means that the waveform characteristics in reality are in high agreement with the theoretical physical model, confirming the existence of the bubble; this judgment method is based on white-box physical determinism rather than black-box statistical probability, effectively avoiding erroneous actions caused by environmental interference.

[0074] If the cross-correlation coefficient is less than the confidence threshold, the system determines that the liquid level fluctuation is caused by non-physical sensor noise such as sensor drift or electromagnetic interference. At this time, the controller performs digital low-pass filtering on the real-time observed liquid level height and maintains the feeding state without triggering slag discharge.

[0075] By using physical principles as a sieve, all signal components that do not conform to the laws of fluid physics are filtered out, ensuring the continuity of production and avoiding erroneous shutdowns or feeding stops.

[0076] Example 2:

[0077] When calculating the theoretical static liquid level, the controller performs the following steps:

[0078] Using Henry's Law and a pre-defined solubility constant, the theoretical volume of solute dissolved in the solvent is calculated.

[0079] Calculate the theoretical density of the liquid mixture based on the temperature and pressure values;

[0080] Based on the integral values ​​of the influent and effluent mass flow rates, the theoretical dissolved volume, and the theoretical density, the physical height of the pure liquid phase within the geometric container, excluding gas phase voids, is calculated and used as the theoretical static liquid level.

[0081] This embodiment further defines the controller's calculation of the theoretical static liquid level height. The specific logic; to ensure the uniformity of physical dimensions and accurate conversion of fluid volume, the controller pre-stores the effective cross-sectional area constant of the extraction container. and system startup time Initial fluid mass inside the vessel at that time This value is obtained through static weighing or liquid level initialization calibration;

[0082] During the calculation process, the controller specifically performs the following steps:

[0083] Using Henry's Law and a pre-defined solubility constant, calculate the theoretical solubility volume of a solute, such as CO2, in a solvent like an electrolyte. ;

[0084] Based on temperature values and pressure values The theoretical density of the liquid mixture was calculated using mixing rules (Mixing Rules). ;

[0085] The controller is based on the integral value of the inlet and outlet mass flow rates. With initial fluid mass Calculate the total cumulative mass within the system;

[0086] The physical height of the pure liquid phase, excluding gas phase voids, within a geometric container can be calculated using the following physical formula. The calculation formula is as follows:

[0087]

[0088] in, The effective cross-sectional area of ​​the extraction container; The initial fluid mass at system startup; This is the current calculation time; The time variable for integral calculation; , They are time points The feed and discharge mass flow rates; This is the theoretical density of the liquid mixture; This is the theoretical volume of solute that dissolves.

[0089] To ensure the homogeneity of physical dimensions, the initial fluid mass is... After combining with the mass accumulation term, divide by the theoretical density Converted to fluid volume, and the theoretical dissolved volume The total liquid volume is obtained by summing the liquid volume with the fluid volume; finally, the total liquid volume is divided by the effective cross-sectional area of ​​the container. This allows us to obtain a dimensionally uniform theoretical liquid level height; and to construct a liquid level reference that strictly conforms to the laws of physical conservation and dimensional homogeneity, eliminating the reference surface drift error caused by operating condition fluctuations or missing initial integral values.

[0090] Example 3:

[0091] The fluid rheology parameter library includes surface tension attenuation factors and viscosity coefficients; when generating the theoretical disturbed liquid level waveform, the controller specifically performs the following steps:

[0092] The theoretical half-life of bubble rupture is calculated based on the surface tension decay factor.

[0093] Constructing a non-Newtonian fluid support force model using viscosity coefficient;

[0094] By combining the theoretical half-life and the non-Newtonian fluid support force model, the virtual thickness of the foam layer that varies with time is calculated, and this virtual thickness is superimposed on the theoretical static liquid level to form a foam layer thickness model.

[0095] This embodiment further defines the method for generating the foam layer thickness model;

[0096] The fluid rheology parameter library contains surface tension attenuation factors for specific systems. and viscosity coefficient The above parameters are not static constants, but are dynamically obtained through the online calibration module: during the steady-state phase at the beginning of each feed, the system uses small perturbation tests to invert the current fluid viscosity coefficient. Meanwhile, based on the real-time collected temperature values... and pressure values The surface tension attenuation factor is determined by using a pre-stored physical property lookup table. Temperature compensation corrections were performed to ensure that the model parameters matched the current complex mixed electrolyte conditions.

[0097] The controller performs the following steps to generate a foam model:

[0098] Half-life calculation: based on surface tension decay factor Calculate the theoretical half-life of bubble bursting. The specific calculation formula is as follows: ,in This is a preset stability reference constant that includes time dimension transformation. For example, take an experience index. ;

[0099] Support force modeling: using viscosity coefficient Construct a non-Newtonian fluid support force model; specifically, calculate the maximum stable stacking height of the foam layer. To achieve this, the calculation formula is: ,in This is the fluid support force coefficient;

[0100] Virtual thickness superposition: Combining theoretical half-life and support force model, calculate the virtual thickness of the foam layer as it changes over time. :

[0101]

[0102] This virtual thickness is superimposed on the theoretical static liquid level. A foam layer thickness model is formed; by converting the experience and knowledge of senior process experts regarding flotation foam layers into the aforementioned determined mathematical function, the system can accurately simulate the reflective signal characteristics that the sensor should have.

[0103] Example 4:

[0104] The fluid rheology parameter library also includes supercritical fluid density fluctuation factors; when generating the theoretically disturbed liquid level waveform, the controller specifically performs the following steps:

[0105] Monitor the rate of change of pressure values ​​over time. When the rate of change exceeds the preset pressure drop threshold, trigger the flash evaporation simulation logic.

[0106] Based on the supercritical fluid density fluctuation factor, the liquid phase volume expansion rate caused by local pressure reduction is calculated.

[0107] The volume expansion rate is converted into the corresponding imaginary liquid level height, and this imaginary liquid level height is superimposed on the theoretical static liquid level height to form a flash volume expansion model.

[0108] This embodiment further defines the generation method of the flash volume expansion model to handle boiling-like interference;

[0109] The fluid rheology parameter library also includes supercritical fluid density fluctuation factors. ;

[0110] The controller performs the following steps:

[0111] Real-time monitoring of pressure values rate of change over time When the rate of change exceeds the preset pressure drop threshold That is, when a rapid depressurization or pressure fluctuation occurs, the flash evaporation simulation logic is triggered;

[0112] Based on supercritical fluid density fluctuation factor Calculate the liquid phase volume expansion coefficient caused by local pressure reduction. The dimensionless ratio is calculated using the following formula:

[0113]

[0114] in This is the expansion sensitivity coefficient. This refers to the reference density value of the fluid under standard conditions or the system's preset calibration density constant; if That is, after the pressure fluctuation returns to stability, the controller does not immediately... Instead of forcing a zero-level reset, a first-order inertial process of bubble dissipation is introduced:

[0115]

[0116] in The discrete sampling time interval or control period of the system. The time constant for bubble dissipation is used; this correction avoids non-physical step drops in the simulated waveform and ensures that the liquid level drop process at the end of the flash evaporation stage is consistent with the actual physical characteristics.

[0117] The controller calculates the flash height increment based on the currently estimated theoretical static liquid level. :

[0118]

[0119] The calculated flash height increment Superimposed to the theoretical static liquid level height To form a flash volume expansion model :

[0120]

[0121] By establishing a clear linear expansion relationship, the system can distinguish between physical foam and flash expansion; foam needs to be discharged, while flash expansion is a temporary artificial volume increase caused by fluctuations in operating conditions; by simulating boiling characteristics, the system can identify whether the current liquid level rise is caused by boiling in subsequent comparisons, thereby avoiding misoperation.

[0122] Example 5:

[0123] When calculating the cross-correlation coefficient, the controller performs the following steps:

[0124] Project the actual residual sequence and the theoretical disturbance sequence into the same time window;

[0125] Calculate the waveform slope values ​​of the two sequences at corresponding time points;

[0126] Calculate the Pearson correlation coefficient between the waveform slope values ​​of the two sequences, and use this Pearson correlation coefficient as the cross-correlation coefficient.

[0127] This embodiment further defines the specific calculation method for the cross-correlation coefficient, and adopts the residual topology mapping technique;

[0128] The controller performs the following steps:

[0129] Time window projection: Projecting the real residual sequence and theoretical interference sequence Project onto the same time window, such as the most recent 30-second data window;

[0130] Execution of temporal sliding alignment: Given the randomness of the starting point of physical bubble generation, the controller sets the maximum time delay search range within the current time window. Theoretical interference sequence ,That The waveform at that moment begins in the actual residual sequence. Perform time delay scanning; calculate different time delays The cross-correlation coefficients are calculated, and the largest cross-correlation coefficient is selected as the final criterion, thereby eliminating the phase misalignment caused by the uncertainty of the bubble triggering time.

[0131] Slope feature extraction: Calculate the waveform slope values ​​of the two sequences at corresponding time points. ;

[0132] Waveform slope value: refers to the first derivative of the sequence as a function of time. It reflects the trend and speed of liquid level change and is more representative of the dynamic characteristics of the physical process than the amplitude alone.

[0133] Pearson correlation calculation: Calculate the Pearson correlation coefficient between the waveform slope values ​​of two sequences, and use this coefficient as the final cross-correlation coefficient;

[0134] By comparing slopes rather than absolute values, the system can identify the shape similarity of waveforms. As long as the fluctuation trend of the real waveform, such as the rate of foam growth and bursting, is consistent with the theoretical simulation, the system can accurately determine its physical properties even if there are differences in amplitude, which greatly improves the robustness of identification.

[0135] Example 6:

[0136] When the controller performs digital low-pass filtering, it specifically executes the following steps:

[0137] Identify signal components in the real residual sequence that cannot be aligned in the time domain with the theoretical interference sequence;

[0138] The signal components are labeled as random electromagnetic interference noise;

[0139] A moving average filtering algorithm is used to remove signal components that are marked as random electromagnetic interference noise, generating smoothed true liquid level data to prevent the system from stopping the feeding action due to false high liquid level signals.

[0140] This embodiment further defines the specific execution logic of the digital low-pass filter;

[0141] When the system determines that the signal is noise, the controller executes:

[0142] Signal recognition: Identifying real-world residual sequences Unable to match theoretical interference sequences Signal components aligned in the time domain;

[0143] Noise labeling: The above signal components are labeled as random electromagnetic interference noise;

[0144] Smoothing: A sample-and-hold and moving average strategy is employed; when the current sampling point is marked as noise, the controller replaces the value of that point with the valid liquid level value from the previous moment. Or the current theoretical calculation value Then, the replaced values ​​are input into the moving average queue for filtering calculation to generate smoothed real liquid level data;

[0145] This processing method prevents time axis misalignment caused by direct rejection or lowering of the mean due to zeroing, effectively preventing the system from erroneously triggering the stop feeding action due to false high liquid level signals, and ensuring the stability of the control system.

[0146] Example 7:

[0147] The multimodal sensor array includes an array of pressure transmitters and an RF admittance level gauge installed at different heights in the extraction vessel;

[0148] The real-time liquid level height is calculated by weighting the static pressure difference data measured by the pressure transmitter array and the capacitance change data measured by the radio frequency admittance level gauge.

[0149] This embodiment specifically illustrates the composition of the multimodal sensor array and the data fusion method;

[0150] The multimodal sensor array includes an array of pressure transmitters and an RF admittance level gauge installed at different heights in the extraction vessel;

[0151] Real-time monitoring of liquid level height It is obtained through data fusion, and the calculation formula is as follows:

[0152]

[0153] in, Height calculated based on static pressure difference data measured by pressure transmitter array;

[0154] Height calculated based on capacitance change data measured by a radio frequency admittance level gauge;

[0155] These are weighting coefficients dynamically assigned based on the signal-to-noise ratio under the current operating conditions; the controller calculates the past performance of the two sensors in real time. Measurement variance over a period of time and The coefficients are calculated using the inverse variance weighting method:

[0156]

[0157] Using the above formula, when the variance of a sensor increases, its weight automatically decreases, thereby capturing the details of liquid level fluctuations more comprehensively and robustly, and providing more reliable raw data for subsequent algorithm analysis.

[0158] Example 8:

[0159] The flotation process system is a waste battery electrolyte recovery system, and the fluid system in the extraction container is a binary or ternary mixture of supercritical carbon dioxide and electrolyte.

[0160] The interference injection simulation module is used to simulate the liquid level fluctuation characteristics caused by carbon dioxide boiling or lithium salt crystallization entrainment.

[0161] This embodiment illustrates the specific application scenario of this system;

[0162] This flotation process system is a waste battery electrolyte recovery system, in which the fluid system in the extraction container is a binary or ternary mixture of supercritical carbon dioxide and electrolyte.

[0163] In this scenario, the interference injection simulation module is specifically used to simulate the liquid level fluctuation characteristics caused by viscosity changes in the flash evaporation model corresponding to carbon dioxide boiling or the foam model corresponding to lithium salt crystallization entrainment.

[0164] For the specific high-risk, high-pressure, and complex operating scenario of waste battery recycling, a customized liquid level control solution is provided, which solves the industry problem that traditional sensors completely fail in such systems due to drastic changes in physical properties.

[0165] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An automatic flotation liquid level control system based on rheological models and cross-correlation analysis, characterized in that, This includes a multimodal sensor array, a slag discharge actuator, and a controller electrically connected to the multimodal sensor array and the slag discharge actuator; the controller is configured to perform the following steps: It receives real-time operating data of the extraction vessel collected by a multimodal sensor array. The real-time operating data includes temperature values, pressure values, feed and discharge mass flow rates, and time-series real-time observed liquid level height with fluctuations. The pre-stored Peng-Robinson equation of state model is invoked, and the theoretical static liquid level height under the current working conditions of no foam and no boiling is calculated based on temperature and pressure values. The pre-stored fluid rheology parameter library is called, and a preset foam layer thickness model and flash volume expansion model are superimposed on the theoretical static liquid level height to generate the theoretical disturbed liquid level waveform. Perform differential calculation: Subtract the theoretical static liquid level height from the real-time observed liquid level height in the time series to obtain the actual residual sequence, and subtract the theoretical static liquid level height from the theoretical disturbed liquid level waveform to obtain the theoretical disturbance sequence; Calculate the cross-correlation coefficient between the actual residual sequence and the theoretical interference sequence; In response to a cross-correlation coefficient greater than or equal to a preset confidence threshold, the liquid level fluctuation is determined to be caused by physical foam accumulation, and a start command is sent to the slag discharge actuator to perform physical slag scraping action; When the cross-correlation coefficient is less than the confidence threshold, it is determined that the liquid level fluctuation is caused by non-physical sensor noise. Digital low-pass filtering is performed on the real-time observed liquid level height while maintaining the feeding state. When calculating the theoretical static liquid level, the controller performs the following steps: Using Henry's Law and a pre-defined solubility constant, the theoretical volume of solute dissolved in the solvent is calculated. Calculate the theoretical density of the liquid mixture based on the temperature and pressure values; Based on the integral values ​​of the influent and effluent mass flow rates, the theoretical dissolved volume, and the theoretical density, the physical height of the pure liquid phase within the geometric container, excluding gas phase voids, is calculated and used as the theoretical static liquid level. The calculation formula is as follows: ； in, The effective cross-sectional area of ​​the extraction container; The initial fluid mass at system startup; This is the current calculation time; The time variable for integral calculation; , They are time points The feed and discharge mass flow rates; This is the theoretical density of the liquid mixture; This is the theoretical volume of solute that dissolves. The fluid rheology parameter library includes surface tension attenuation factors and viscosity coefficients; when generating the theoretical disturbed liquid level waveform, the controller specifically performs the following steps: The theoretical half-life of bubble rupture is calculated based on the surface tension decay factor. Constructing a non-Newtonian fluid support force model using viscosity coefficient; By combining the theoretical half-life and the non-Newtonian fluid support force model, the virtual thickness of the foam layer as a function of time is calculated, and this virtual thickness is superimposed on the theoretical static liquid level to form a foam layer thickness model. The fluid rheology parameter library also includes supercritical fluid density fluctuation factors; when generating the theoretically disturbed liquid level waveform, the controller specifically performs the following steps: Monitor the rate of change of pressure values ​​over time. When the rate of change exceeds the preset pressure drop threshold, trigger the flash evaporation simulation logic. Based on the supercritical fluid density fluctuation factor, the liquid phase volume expansion rate caused by local pressure reduction is calculated. The volume expansion rate is converted into the corresponding imaginary liquid level height, and this imaginary liquid level height is superimposed on the theoretical static liquid level height to form a flash volume expansion model. When calculating the cross-correlation coefficient, the controller performs the following steps: Project the actual residual sequence and the theoretical disturbance sequence into the same time window; Calculate the waveform slope values ​​of the two sequences at corresponding time points; Calculate the Pearson correlation coefficient between the waveform slope values ​​of the two sequences, and use this Pearson correlation coefficient as the cross-correlation coefficient.

2. The automatic flotation liquid level control system based on rheological model and cross-correlation analysis according to claim 1, characterized in that, When the controller performs digital low-pass filtering, it specifically executes the following steps: Identify signal components in the real residual sequence that cannot be aligned in the time domain with the theoretical interference sequence; The signal components are labeled as random electromagnetic interference noise; A moving average filtering algorithm is used to remove signal components that are marked as random electromagnetic interference noise, generating smoothed true liquid level data to prevent the system from stopping the feeding action due to false high liquid level signals.

3. The automatic flotation liquid level control system based on rheological model and cross-correlation analysis according to claim 1, characterized in that, The multimodal sensor array includes an array of pressure transmitters and an RF admittance level gauge installed at different heights in the extraction vessel; The real-time liquid level height is calculated by weighting the static pressure difference data measured by the pressure transmitter array and the capacitance change data measured by the radio frequency admittance level gauge.

4. The automatic flotation liquid level control system based on rheological model and cross-correlation analysis according to claim 1, characterized in that, The flotation process system is a waste battery electrolyte recovery system, and the fluid system in the extraction container is a binary or ternary mixture of supercritical carbon dioxide and electrolyte. The interference injection simulation module is used to simulate the liquid level fluctuation characteristics caused by carbon dioxide boiling or lithium salt crystallization entrainment.

Images