A method for detecting the lower column liquid level of a fractionating column

Abstract

This invention discloses a method for detecting the liquid level at the bottom of a fractionation column, belonging to the field of air separation. Addressing the problem that traditional single differential pressure measurements under cryogenic conditions are prone to failure due to ice blockage or vaporization of the pressure tapping pipe and cannot be maintained online, this method constructs a heterogeneous redundant measurement system combining contact differential pressure detection and non-contact external ultrasonic detection. By simultaneously acquiring two raw signals, density dynamic compensation and sound velocity compensation based on a thermodynamic database are performed respectively to calculate high-precision first and second liquid level values. Subsequently, an adaptive Kalman filter algorithm is introduced to dynamically adjust process noise based on feed and discharge flow rates and estimate observation noise in real time based on residuals, optimally fusing the two liquid level values. This invention achieves highly reliable fusion of multi-source data and system logic self-healing, significantly improving the accuracy, stability, and intelligence level of fractionation column bottom liquid level measurement.

Description

Technical Field

[0001] This application belongs to the field of air separation, specifically relating to a method for detecting the liquid level at the bottom of a distillation column. Background Technology

[0002] With the continuous evolution of air separation technology, the fractionation column, as the core equipment for component separation in air separation units, plays a crucial role in ensuring production continuity and product quality through accurate acquisition of its internal operating parameters. The liquid level at the bottom of the fractionation column, as a key indicator characterizing the material balance and cooling distribution within the column, directly relates to the stability and energy efficiency of the air separation process. Traditional monitoring methods primarily rely on physical pressure sensing technology, measuring the pressure gradient at specific points to monitor the rise and fall of the liquid level in real time.

[0003] Among these technologies, the liquid level detection method at the bottom of the fractionation column is a core component of air separation automation control. This technology primarily focuses on achieving continuous measurement of the liquid level in cryogenic environments, providing decision support for the central control system. By converting and processing pressure signals, the system can detect subtle changes in the distillation process within the column, thereby ensuring the stable accumulation and rational distribution of liquid air at the bottom of the column.

[0004] However, existing measurement schemes based on differential pressure principles face severe challenges in practical operation. The cryogenic medium within the pressure-sensing pipeline is prone to phase change bubbles due to environmental disturbances, or freezing and blockage due to moisture intrusion, resulting in slow data feedback or a falsely constant state. Simultaneously, the fractionation tower is located inside a tightly sealed, insulated box, making online maintenance and replacement virtually impossible without interrupting production. Furthermore, the lack of effective redundancy and cross-validation in single-dimensional detection means the system lacks self-diagnostic capabilities when faced with abnormal instrument fluctuations, making it difficult to provide a reliable level reference in complex fault scenarios.

[0005] Therefore, a method for detecting the liquid level at the bottom of a distillation column is desired. Summary of the Invention

[0006] The purpose of this invention is to provide a method for detecting the liquid level at the bottom of a distillation column, which can effectively solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for detecting the liquid level at the bottom of a distillation column, comprising: Steps S1 to S6: Step S1: Deploy the detection branch, which includes a contact differential pressure detection branch and a non-contact external ultrasonic detection branch; Step S2: The original differential pressure signal of the medium inside the tower is collected through the contact differential pressure detection branch, and the first liquid level value representing the liquid level height is calculated by combining the synchronously acquired real-time operating parameters inside the tower. Step S3: Through the non-contact external ultrasonic detection branch, ultrasonic pulses are emitted and received into the medium inside the tower, the flight time of the ultrasonic waves in the medium is measured, and the second liquid level value representing the liquid level height is calculated by combining the pre-established sound velocity compensation model. Step S4: The first liquid level value and the second liquid level value are used as observation variables and input into an adaptive Kalman filter for multi-source data fusion to obtain the optimal liquid level estimate used to characterize the true liquid level in the tower. Step S5: While performing data fusion, the fault diagnosis logic runs in parallel. By analyzing the signal characteristics of the first liquid level value and the second liquid level value, the fault type of the detection branch is identified in real time, and an early warning is triggered or the system switches to a single-path reliable operation mode based on the diagnosis results.

[0008] Furthermore, the step of deploying the non-contact external ultrasonic detection branch in step S1 specifically includes: installing multiple ultrasonic transducers at equal intervals along an axis perpendicular to the ground on the outer wall surface of the lower column of the distillation tower to form a vertical sensor array. The total number of ultrasonic transducers installed is determined based on the total height of the lower column of the distillation tower to ensure effective coverage of the full range of liquid levels. Each ultrasonic transducer is filled with a special low-temperature resistant acoustic coupling medium between itself and the metal tower wall. The coupling medium can maintain viscoelasticity and low acoustic attenuation characteristics in a cryogenic environment to maximize the transmission efficiency of the acoustic signal.

[0009] Furthermore, the step of calculating the first liquid level value in step S2 further includes a density dynamic compensation process: Based on the pressure and temperature values ​​in the real-time operating parameters inside the tower, the real-time density value of the medium inside the tower under the current operating conditions is determined by looking up tables and interpolation calculations in the pre-set air component thermodynamic property database in the system's non-volatile memory. By using additional temperature sensors arranged on the positive pressure side pressure tap and the negative pressure side pressure tap, the reference temperature of the liquid column in the pressure tap is obtained, and the density value of the liquid column in the pressure tap is calculated using a thermodynamic property database. The density value of the liquid column inside the pressure tapping tube is used to correct the original differential pressure signal to eliminate the measurement error caused by the temperature difference between the inside and outside of the cold box. Based on the principles of hydrostatics, a high-precision first liquid level value is calculated using the corrected differential pressure value and the real-time density value of the medium inside the tower.

[0010] Furthermore, the step of calculating the second liquid level value in step S3 further includes a dynamic sound velocity compensation process: The temperature of the tower wall is monitored in real time by temperature sensors placed on the outside of the tower wall, and the propagation speed of ultrasonic waves in the tower wall at the current tower wall temperature is determined according to the pre-stored temperature-sound velocity correspondence of the tower wall material. Based on the pre-measured tower wall thickness, coupling layer thickness, and corresponding sound velocity, calculate the fixed time offset introduced by the ultrasonic wave in the propagation path due to passing through the tower wall and coupling layer twice. Based on the temperature value and medium composition in the real-time operating parameters inside the tower, the actual propagation speed of ultrasound in the current liquid medium is searched and determined in the pre-set thermodynamic property database. The precise second liquid level value is calculated using the ultrasonic flight time, fixed time offset, and actual sound velocity in the medium.

[0011] Furthermore, the fusion process of the adaptive Kalman filter in step S4 further includes adaptive adjustment of process parameter coupling: The feed and discharge flow rates of the fractionation tower are collected in real time, and the net flow rate change rate, which characterizes the fluctuation of material within the tower, is calculated. The process noise variance in the Kalman filter is dynamically corrected based on the net flow rate change, so that the filter trusts the real-time observations more when the flow rate fluctuates drastically. Based on the observation residuals of the first and second liquid level values ​​at the current moment, a recursive algorithm is used to update the observation noise covariance matrix in real time to adaptively evaluate the confidence level of the two sensors under the current operating conditions.

[0012] Furthermore, the fault diagnosis logic in step S5 specifically includes ice blockage fault diagnosis: Real-time sliding variance monitoring is performed on the calculated first liquid level value; If, within multiple consecutive time windows, the sliding variance of the first liquid level value remains below the preset minimum threshold, exhibiting an extremely smooth horizontal straight line characteristic, while the second liquid level value maintains normal dynamic fluctuations, then it is determined that the pressure tap of the contact differential pressure detection branch has an ice blockage fault.

[0013] Furthermore, the fault diagnosis logic in step S5 specifically includes vaporization interference diagnosis: Perform spectral analysis on the acquired differential pressure raw signal to extract the total energy within the preset high-frequency band; Compare the current total high-frequency energy with the baseline energy stored when the system was initially put into operation and the process was stable; If the current total high-frequency energy exceeds the set multiple of the reference energy, and within the same time window, the variance of the first liquid level value exceeds the preset large jump threshold, then it is determined that there is bubble interference caused by medium vaporization in the pressure tapping tube.

[0014] Furthermore, the fault diagnosis logic in step S5 specifically includes sensor adhesion status diagnosis: The signal-to-noise ratio (SNR) of ultrasonic echo signals is monitored in real time. The SNR is defined as the ratio of the amplitude of the main peak of the echo to the amplitude of the noise floor. If the signal-to-noise ratio is lower than the preset signal-to-noise ratio threshold in multiple consecutive measurements, it is determined that the acoustic coupling medium is aging or the sensor is loose, resulting in a decrease in sound wave transmission efficiency.

[0015] Furthermore, the steps in step S5 to switch to single-path reliable operation mode include: When the fault diagnosis logic determines that the contact differential pressure detection branch is faulty, the system automatically shields the data of the faulty branch and uses only the second liquid level value obtained by the non-contact external ultrasonic detection branch as the final output. When a fault is detected in the non-contact external ultrasonic detection branch, the system uses only the first liquid level value obtained from the contact differential pressure detection branch as the final output, ensuring that the main control system always obtains a continuous and reliable liquid level reference signal.

[0016] Furthermore, it also includes hardware self-test and initialization steps performed during the system startup phase, which are executed before step S1: A current loop continuity test is performed on the differential pressure transmitter. Its status is determined by applying a test current to the transmitter and detecting the loop response. A static impedance test is performed on the ultrasonic transducer. By measuring its impedance value in the unexcited state and comparing it with the standard range, it is determined whether there is any abnormality. The central processing unit performs memory integrity verification by writing and reading back preset data to check the memory status; Only after all three self-tests above pass will the system initialize the filter and enter real-time monitoring mode.

[0017] In summary, this application includes at least one of the following beneficial technical effects: 1. This invention constructs a dual-dimensional measurement system with completely different physical principles by simultaneously deploying a contact-type differential pressure detection branch and a non-contact external ultrasonic detection branch at the bottom of the fractionation tower. This heterogeneous redundancy design effectively compensates for the inherent defects of a single measurement method. Even in extreme conditions where contact measurement fails due to ice blockage or vaporization in the pressure tapping pipeline, the non-contact ultrasonic detection branch can still penetrate the tower wall to obtain the true liquid level. This completely solves the industry problem of difficulty in online maintenance of fractionation towers when they are sealed in a cold storage box, ensuring the long-term safe operation of the air separation unit.

[0018] 2. This invention, by introducing an adaptive Kalman filter algorithm, can analyze the statistical characteristics of signals from different sensors in real time and dynamically allocate weights based on residual covariance. This multi-source data fusion technology effectively eliminates the inherent random noise and environmental disturbances of a single sensor, resulting in a more stable and accurate optimal liquid level estimate compared to traditional single-path measurements. Simultaneously, based on dynamic compensation for pressure, temperature, and component density within the column, as well as sound velocity compensation, it eliminates systematic errors caused by changes in medium properties under cryogenic conditions, significantly improving measurement accuracy and providing high-quality data support for refined distillation control.

[0019] 3. The diagnostic logic built into this invention can accurately determine various typical faults such as ice blockage in the pressure tapping pipe, medium vaporization, and sensor loosening through variance analysis and spectral feature identification of the liquid level signal. This intelligent self-diagnostic function allows operators to grasp the instrument's operating status immediately, overcoming the difficulty of distinguishing between true and false data when anomalies occur in traditional technologies. After identifying a local fault, the system can automatically perform gain adjustment and mode switching, achieving logical self-healing of the measurement function through weight tilting, greatly reducing the risk of misoperation caused by false alarms from instruments and improving the overall intelligence level of the automation system. Attached Figure Description

[0020] Figure 1 This is an overall schematic diagram of a method for detecting the liquid level at the bottom of a distillation column; Figure 2 This is a schematic diagram illustrating the core principles of multi-source data fusion; Figure 3 This is a flowchart of density dynamic compensation and sound velocity compensation calculations; Figure 4 It is a logic diagram for fault diagnosis and identification and operation mode switching. Detailed Implementation

[0021] 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.

[0022] The method for detecting the liquid level at the bottom of a distillation column disclosed in this application includes the following implementation steps: First, in step S1, before performing the actual liquid level detection, the physical deployment of the detection branch needs to be completed. This is the foundation for the implementation of the entire method. The specific operation steps of step S1 are as follows: Step S101: Deploy the pressure tapping point for the differential pressure detection branch. Physically open and weld pipes to the container wall of the lower column of the fractionation tower. Connect the positive pressure tapping pipe to the liquid collection area at the bottom of the lower column, which is located 200mm to 500mm above the liquid outlet at the bottom of the tower. The specific value is determined based on the tower diameter and bottom structure to ensure that it avoids the area where impurities may accumulate at the bottom drain outlet.

[0023] Connect the negative pressure tap to the gas phase zone at the top of the lower column of the fractionation tower. This zone should be located above the uppermost distillation tray and at least 300 mm below the gas phase outlet to ensure that the collected pressure signal can accurately reflect the static pressure of the gas above the liquid phase and avoid the gas flow directly impacting the pressure tap.

[0024] Step S102: Install the pressure tapping lines. When the two sets of pressure tapping lines leave the tower body and enter the internal space of the cold storage box, they must maintain a fixed downward tilt angle, which is set to be 5° to 15° downward relative to the horizontal plane.

[0025] This design utilizes gravity to allow any condensate that may form inside the tube to automatically flow back into the tower, effectively reducing measurement deviations caused by liquid accumulation inside the pressure tapping tube and ensuring accurate pressure transmission. The pressure tapping tube is made of austenitic stainless steel, such as 06Cr19Ni10, to withstand cryogenic environments.

[0026] This completes the setup of the contact-type differential pressure measurement pathway. The above steps, by standardizing the connection point location, installation angle, and material of the pressure tap, ensure the reliability of the original differential pressure signal acquisition from a physical structural perspective, laying the hardware foundation for subsequent high-precision liquid level calculations.

[0027] Step S103: Deploy the sensor array with externally attached ultrasonic detection branches. This process is carried out entirely outside the tower body and does not involve damage to the pressure vessel. Multiple ultrasonic transducers are installed at equal intervals on the outer wall surface of the lower distillation tower along an axis perpendicular to the ground, forming a vertical sensor array.

[0028] The specific number of transducers is not a fixed value, but rather depends on the total height of the lower column of the fractionation column. To determine: When When the distance is less than 5 meters, install 4 transducers; when When the distance is 5 to 10 meters, install 6 units; when For distances greater than 10 meters, install 8 transducers. The spacing between adjacent transducers is... Divide by the number of transducers to ensure effective coverage of the full range of liquid levels.

[0029] Step S104 ensures ultrasonic wave transmission and sensor protection. A layer of dedicated low-temperature acoustic coupling medium must be filled between each ultrasonic transducer and the metal tower wall. This medium is a low-temperature coupling grease formulated with silicone oil as a base and fumed silica thickener. Its operating temperature range covers -200℃ to +80℃, and it maintains viscoelasticity and low acoustic attenuation characteristics even in cryogenic environments of -196℃, ensuring maximum transmission efficiency of the acoustic signal between the metal wall and the sensor interface.

[0030] The transducer housing is made of 304 or 316L stainless steel and its protection level must reach IP67 or IP68 to resist the long-term intrusion of the perlite dust filled in the cold box and the possible humid environment.

[0031] Step S105: Sensor array operating mechanism and transducer selection strategy. During system operation, the central processing unit dynamically determines the ultrasonic transducers to participate in the measurement based on the current liquid level prediction or historical fused liquid level values. The specific selection strategy is as follows: The entire measurement range of the fractionation column is divided into several measurement sub-intervals equal to the number of transducers. Each transducer corresponds to a sub-interval near its installation height. When the liquid level is within a certain sub-interval, the system prioritizes activating the transducer corresponding to that sub-interval for ultrasonic transmission and reception to ensure optimal echo signal strength and the highest signal-to-noise ratio. If the echo signal quality of that transducer is lower than a preset threshold, adjacent transducers are activated sequentially upwards or downwards for measurement until a valid signal meeting the quality requirements is obtained.

[0032] When the liquid level crosses different sub-intervals, the system employs a hysteresis switching strategy to avoid frequent transducer switching caused by liquid level fluctuations near the sub-interval boundaries. Each transducer is in a silent mode when inactive, not participating in excitation or reception, to reduce system power consumption and minimize mutual interference. The echo signal processing for all transducers uses the same time-of-flight measurement and sound velocity compensation algorithm to ensure measurement consistency.

[0033] In addition, during the system self-test or periodic calibration phase, the central processing unit can activate all transducers in sequence. By comparing the measurement results of adjacent transducers at the same liquid level, the redundancy verification and fault location of the sensor array can be achieved, further improving the reliability of the measurement system.

[0034] In summary, step S1 constructs a non-contact redundant measurement system. Through an external array and a dedicated coupling medium, it acquires raw ultrasonic signals that are completely different from the differential pressure method without damaging the tower structure, providing independent hardware support for subsequent data fusion and fault diagnosis.

[0035] For step S2, the signal acquisition stage, the core task is to simultaneously acquire two raw signals through the sensing terminal and perform preprocessing to provide a reliable data foundation for subsequent liquid level calculation. The specific implementation steps are as follows: Step S201: High-speed acquisition of differential pressure signal. The sensing terminal integrates a high-precision analog-to-digital conversion module with a resolution of not less than 16 bits, which continuously samples the 4 to 20 mA current signal output by the differential pressure transmitter.

[0036] The sampling frequency is set to 200Hz based on the maximum possible rate of change in liquid level to ensure complete capture of the transient characteristics of liquid level fluctuations and avoid information loss due to insufficient sampling rate. The sampled digital values ​​are temporarily stored in a local buffer via the SPI interface.

[0037] Step S202, excitation and reception of ultrasonic signals: The central processing unit sends excitation pulse signals to the ultrasonic transducer array arranged on the tower wall through the general input / output port according to the preset pulse repetition frequency.

[0038] Each excitation pulse is a square wave with an amplitude of 5V and a width of 5 microseconds. The pulse repetition frequency is set to 100Hz to ensure a sufficient measurement refresh rate while avoiding signal aliasing due to excessive transmission. Each transducer immediately switches to receiving mode after excitation. The weak echo signal received is conditioned by a preamplifier and a variable gain amplifier, and then digitized by an analog-to-digital converter at a sampling rate of 10MHz to form complete echo waveform data.

[0039] Step S203, hardware synchronization of the two signals: In order to achieve time axis alignment during subsequent data fusion, the system adopts a hardware synchronization triggering mechanism: the timer module of the central processing unit outputs a rising edge synchronization signal to the external trigger input pin of the analog-to-digital conversion module while generating ultrasonic excitation pulses.

[0040] The synchronization signal forces the sampling clock of the differential pressure signal to be strictly aligned with the transmission time of the ultrasonic pulse. Through this hardware connection method, the time deviation between the two signals can be controlled within 1 microsecond, which is far better than the 1 millisecond deviation tolerance required by the subsequent fusion algorithm.

[0041] Step S204: Transmission of raw data. The differential pressure sampling sequence and ultrasonic echo waveform data obtained synchronously are locally buffered and then uploaded in real time to the central processing unit via industrial Ethernet using the Modbus TCP protocol, or via a redundantly configured PROFIBUS DAP fieldbus. The communication protocol adopts a standard industrial protocol, and the data packets contain timestamp information to ensure data integrity during transmission.

[0042] Step S205: Digital filtering preprocessing. After receiving the raw data, the central processing unit immediately performs digital filtering to remove interference. For the differential pressure signal, a 50th-order finite impulse response low-pass filter is designed based on a 200Hz sampling frequency, using a Hamming window function and a cutoff frequency set to 20Hz to suppress 50Hz power frequency interference and its higher harmonics generated by high-power frequency converters and other equipment. At the same time, a dual second-order infinite impulse response notch filter with a center frequency of 50Hz and a quality factor Q=10 is used to further attenuate the power frequency components.

[0043] For the ultrasonic echo signal, a bandpass filter with a passband range of 1.9MHz to 2.1MHz was designed based on the transducer center frequency of 2MHz to filter out out-of-band random noise. The signal-to-noise ratio of the filtered signal was significantly improved, laying the foundation for subsequent accurate calculations of flight time and differential pressure values.

[0044] Through step S2 above, the synchronous acquisition, transmission, and purification of the two raw signals—differential pressure and ultrasonic—are completed, resulting in a high-quality data stream with time alignment and good noise suppression. Next, the central processing unit will calculate the first liquid level value H1 and the second liquid level value H2 based on this data.

[0045] For step S3, which involves calculating the first liquid level value, the core task is to obtain a high-precision liquid level value based on the original differential pressure signal, combined with the pressure and temperature data inside the tower, through dynamic density compensation. The specific implementation steps are as follows: Step S301: Obtain real-time operating parameters. The central processing unit reads the filtered raw differential pressure signal from the data buffer. Simultaneously, the pressure value P output by the pressure transmitter inside the tower and the temperature value T output by the temperature transmitter are retrieved and collected synchronously.

[0046] The pressure transmitter has a range of 0 to 1.6 MPa and an accuracy of 0.075 class; the temperature transmitter uses Pt100 platinum resistance thermometer, with a measurement range of -200℃ to 50℃ and an accuracy of ±0.1℃. These parameters are converted into digital values ​​through an analog input module and then stored in a designated register.

[0047] Step S302: Construct a thermodynamic property database. Pre-set the thermodynamic property database of air components in the system's non-volatile memory. This database is calculated based on NISTREFPROP software and covers the entire operating range of the fractionation column in actual operation: pressure range from 0.1MPa to 1.0MPa, pressure step 0.01MPa; temperature range from -200℃ to -170℃, temperature step 1℃.

[0048] The database is stored in the form of a two-dimensional table, with each grid point corresponding to a medium density value. The unit is kg / m³. For non-grid point conditions, it can be obtained through subsequent interpolation.

[0049] Step S303: Bilinear interpolation is used to calculate the real-time density. The central processing unit locates four neighboring grid points in the two-dimensional table of the database based on the current measured pressure P and temperature T. , , , And read the corresponding density values ​​ρ11, ρ12, ρ21, and ρ22.

[0050] Then, the real-time density ρ under the current operating condition is calculated using a bilinear interpolation algorithm. The calculation formula is as follows: First, perform linear interpolation in the temperature direction:

[0051] Then perform linear interpolation in the pressure direction:

[0052] By performing the above two interpolation steps, the accurate density value of the medium at the current pressure and temperature can be obtained.

[0053] Step S304, correction of liquid column density in the pressure tapping pipe: Considering the possible difference between the ambient temperature inside the cold storage box and the temperature inside the tower, which may cause the liquid column density in the pressure tapping pipe to be inconsistent with the medium density inside the tower, thus introducing measurement error, an additional temperature sensor T1 is arranged after the root valve along the positive pressure side pressure tapping pipe, close to the transmitter side, and a temperature sensor T2 is also arranged on the negative pressure side pressure tapping pipe.

[0054] The central processing unit reads T1 and T2, and takes their average value as the reference temperature of the liquid column in the pressure tapping tube. Since the medium inside the pressure tapping pipe has the same composition as the medium inside the tower, and its pressure is approximately equal to the tower pressure P, the database established in step S302 is used to determine the pressure P and temperature. Perform bilinear interpolation again to obtain the density of the liquid column inside the pressure tapping tube. Corrected differential pressure value Calculated based on the principles of fluid statics:

[0055] in This represents the vertical height difference of the pressure taps, a value measured and stored in the system during installation. Due to the actual inclined installation of the pressure taps, The vertical distance between the pressure taps at both ends of the pressure tapping tube and the transmitter.

[0056] Step S305: Calculate the first liquid level value. After completing the density correction, calculate the first liquid level value H1 using the basic formula of hydrostatics:

[0057] In the formula: H1 is the first liquid level value after density compensation, in meters; The differential pressure value is corrected by the liquid column in the pressure tapping tube, in Pascals; ρ is the real-time density of the medium inside the tower calculated in step S303, in kilograms per cubic meter; g is the local gravitational acceleration, taken as 9.8 meters per second squared, or can be precisely set according to the standard value found in the local latitude.

[0058] Step S306, secondary compensation for ambient temperature fluctuations: To further eliminate the residual impact of ambient temperature fluctuations on measurement accuracy, the system introduces a regression compensation model based on historical data. This model makes minor adjustments to H1 based on the rate of change of the additional temperature sensors T1 and T2.

[0059] in This is the real-time average of T1 and T2. Here, H1 is the reference temperature during calibration, and k is the compensation coefficient. k is obtained by recording a set of H1 and the actual liquid level reference value at different ambient temperatures (e.g., obtained through manual gauging or from a known calibration point) during initial system commissioning and process stability. Then, least squares linear regression is used to fit H1 and... The relationship is given by the slope k. This coefficient ranges from -0.001 to 0.001 meters per degree Celsius and is updated periodically during the system's self-calibration phase.

[0060] In summary, through step S3 above, the central processing unit completes in-depth processing and compensation of the differential pressure signal, obtaining a high-precision first liquid level value. This value not only eliminates the influence of changes in medium density under operating conditions, but also corrects for errors caused by differences in ambient temperature within the pressure tapping tube, providing accurate and reliable input for subsequent data fusion with ultrasonic liquid level values.

[0061] For step S4, which involves calculating the second liquid level value, the core task is time-domain analysis based on the ultrasonic echo signal. This involves accurately measuring the flight time and performing multi-factor compensation to obtain a high-precision liquid level value. The specific implementation steps are as follows: In step S401, an ultrasonic pulse is emitted and an echo signal is received. The central processing unit sends a high-voltage excitation pulse to the selected ultrasonic transducer through a general-purpose input / output port. The pulse amplitude is 5V and the width is 5 microseconds, which excites the transducer to generate an ultrasonic pulse.

[0062] The ultrasonic pulse penetrates the tower wall and enters the liquid medium, where it is reflected at the liquid-gas interface. The reflected echo penetrates the tower wall again and is received by the same transducer. The transducer immediately switches to receiving mode after excitation. The weak echo signal it receives is amplified by a preamplifier and then digitized by an analog-to-digital converter at a sampling rate of 10MHz to form complete echo waveform data.

[0063] Step S402: Measure the ultrasonic time-of-flight (ToF). The processor analyzes the digitized echo waveform and records the time interval from the pulse emission moment to the moment the main peak of the reflected echo from the liquid surface is received; this is the time-of-flight (ToF). To accurately locate the main peak moment, a peak detection algorithm is used: search for the point of maximum amplitude of the echo waveform within a preset time window; the time corresponding to this point is the main peak moment of the echo.

[0064] Step S403: Construct a temperature-sound velocity compensation model. Since the propagation speed of sound waves in the metal tower wall is significantly affected by temperature, it is necessary to compensate for the changes in sound velocity caused by variations in tower wall temperature. First, the longitudinal wave sound velocity of the tower wall material (e.g., 06Cr19Ni10 austenitic stainless steel) at different temperatures is experimentally measured, and a temperature-sound velocity relationship table is established. The temperature range covers -200℃ to +50℃, with a temperature step of 10℃.

[0065] The relation table is pre-stored in non-volatile memory as a two-dimensional array. Meanwhile, a temperature sensor is placed on the outer side of the tower wall near the ultrasonic transducer. The tower wall temperature is monitored in real time. The central processing unit processes the measured temperature... The current sound velocity at the tower wall is obtained by linear interpolation in the temperature-velocity of sound table. .

[0066] Step S404: Calculate the fixed time offset introduced by the tower wall and coupling layer. The ultrasonic wave needs to pass through the tower wall and coupling layer twice in its propagation path, and this part of the path introduces a fixed time offset. . Due to the thickness of the tower wall Coupling layer thickness And the corresponding speed of sound.

[0067] Tower wall thickness The coupling layer thickness is measured and stored in the system during installation using an ultrasonic thickness gauge. The thickness is typically 0.1mm to 0.3mm, guaranteed by the transducer installation process, with the nominal value of 0.2mm being used.

[0068] Tower wall sound speed The sound velocity of the coupling layer is obtained from step S403. Based on the physical properties of the coupling medium (low-temperature coupling grease), its value is approximately 1500 m / s. Therefore, the formula for calculating the fixed time offset is:

[0069] Step S405: Calculate the speed of sound in the medium. The propagation speed of ultrasound in a liquid medium (liquid air, liquid oxygen, or liquid nitrogen) is also affected by the medium's temperature and composition. Therefore, based on the medium temperature... (Measured by the temperature transmitter in step S301) and the medium composition (obtained by an online analyzer or preset typical values, such as oxygen content 21% and nitrogen content 79%), the corresponding sound velocity value is looked up in a pre-established thermodynamic property database. .

[0070] The database is calculated based on NIST REFPROP software and covers a complete operating range of pressures from 0.1 to 1.0 MPa, temperatures from -200°C to -170°C, and oxygen content from 0 to 100%. The sound velocity values ​​are stored in a three-dimensional table.

[0071] Step S406: Calculate the second liquid level value H2 based on the time of flight (ToF) and fixed time offset. and medium sound velocity The second liquid level value H2 is calculated using the following formula:

[0072] In the formula: H2 is the second liquid level value, in meters; is the speed of sound in the liquid medium, in meters per second; ToF is the measured time of flight, in seconds. A fixed time offset, in seconds, is introduced for the tower wall and coupling layer.

[0073] Step S407: Eliminating Clutter Interference Using an Adaptive Threshold Algorithm. To eliminate clutter interference generated by complex gas-liquid interfaces and internal components (such as support grids) within the tower, an adaptive threshold algorithm is used during echo peak detection. The algorithm flow is as follows: Calculate the energy envelope of the echo signal, that is, perform a Hilbert transform on the echo signal to obtain the envelope E(t).

[0074] Set the initial detection threshold Th = α· ,in α is the global maximum value of the envelope, and α is the scaling factor, typically taken as 0.5.

[0075] Starting from the pulse emission moment, search backwards to find the first envelope point that exceeds the threshold Th, and record the time of that point. .

[0076] exist Within each preset window width (e.g., a window width of 50 sampling points), find the local maximum value. The time corresponding to this maximum value is the moment of the main peak of the echo.

[0077] If no valid echo is found within the preset maximum search time range, the threshold α is lowered to 0.3, and the search is repeated; if it is still ineffective, it is determined that there is no echo, and an invalid value is output.

[0078] Through steps S401 to S407 above, the central processing unit completes the precise processing of the ultrasonic echo signal to obtain a high-precision second liquid level value H2. This value fully considers the influence of temperature on sound velocity, corrects the fixed delay introduced by the tower wall and coupling layer, and effectively suppresses clutter interference through an adaptive threshold algorithm, providing accurate and reliable input for subsequent data fusion with the differential pressure method liquid level value.

[0079] For step S5, which involves data fusion, the core task is to use the first liquid level value H1 and the second liquid level value H2 calculated in steps S3 and S4 as observed variables, and fuse them through an adaptive Kalman filter to obtain the optimal liquid level estimate. The specific implementation steps are as follows: Step S501: Establish a Kalman filter model, using the first liquid level value H1 and the second liquid level value H2 as observation variables, and construct the state equation and observation equation of the discrete-time linear system. The state variable X(k) represents the actual liquid level at time k, and the state transition matrix A takes a value of 1 because the liquid level changes slowly within adjacent sampling intervals, which can be approximated as a random walk. The observation matrix H is a 2x1 matrix, with all values ​​taking a value of 1, meaning that each observation is a direct measurement of the actual liquid level and is therefore noisy. The system's state equation and observation equation are expressed as follows:

[0080] Where X(k) is the liquid level scalar at time k, A=1 is the state transition matrix, W(k) is the process noise, assumed to be zero-mean Gaussian white noise, and its variance Q(k) needs to be adjusted in real time according to process fluctuations, and Z(k) is the 2D observation vector. , Let V(k) be the observation matrix, and let V(k) be the 2D observation noise vector. Assuming it is zero-mean Gaussian white noise, its covariance matrix R(k) is a 2×2 diagonal matrix. The diagonal elements correspond to the observation noise variances of the two sensors, respectively.

[0081] Step S502: Initialize the filter by setting the initial state estimate value during system startup. And the initial error covariance P(0|0). Take the first liquid level value at the moment of start-up of the differential pressure detection branch. P(0|0) is set according to the sensor's factory accuracy, for example, it is taken as a diagonal matrix with diagonal elements of (0.01m)², which represents the uncertainty of the initial estimate.

[0082] Step S503: Adaptive adjustment of process noise covariance. Process noise reflects the random changes in liquid level caused by fluctuations in feed and discharge flow rates. Real-time acquisition of fractionation column feed flow rate. and discharge flow rate Calculate the rate of change of net flow. ,in This is the rated flow rate. The process noise variance Q(k) is dynamically corrected according to the following formula:

[0083] in The variance of the basic process noise is taken as (0.001m)², and α is an empirical proportionality coefficient with a value of [value missing]. This formula results in greater flow fluctuations, larger process noise variance, reduced filter confidence in the model, and greater reliance on observations.

[0084] Step S504: Adaptive estimation of observation noise covariance and Kalman filtering recursion; after obtaining the observation vector Z(k) at the current time, calculate the observation residual vector. To estimate the observation noise covariance in real time, the diagonal elements R1(k) and R2(k) of R(k) are updated using an exponentially weighted moving average method:

[0085] Where β is the forgetting factor, typically ranging from 0.05 to 0.1, and e1(k) and e2(k) are the residuals of the first and second liquid level values, respectively.

[0086] Initially, R1(0) and R2(0) are set according to the sensor accuracy, for example, taken as (0.005m)². Then, the standard recursive steps of Kalman filtering are executed: first, a time update is performed, predicting the current state and covariance based on the previous state estimate and error covariance; then, the Kalman gain is calculated, and the weights of the observed values ​​are determined using the predicted covariance and the current observation noise covariance; finally, a measurement update is performed, using the current observations to correct the state estimate and obtain the optimal estimate. .

[0087] These standard recursive formulas are well known to those skilled in the art, and their mathematical expressions will not be elaborated here.

[0088] Step S505: Output the fusion result, including the updated state estimate. The optimal liquid level estimate at time k is output to the control system for subsequent monitoring and decision-making.

[0089] In summary, through step S5 above, adaptive Kalman filtering fusion of the liquid level measurements from two heterogeneous sensors is achieved. This process dynamically adjusts the process noise based on the inlet and outlet flow rates and estimates the observed noise in real time based on the residuals. This ensures that the fusion result can automatically adapt to changes in operating conditions and sensor status, maintaining high accuracy and reliability at all times, providing a precise data foundation for subsequent fault diagnosis and process control.

[0090] For step S6, namely fault diagnosis and early warning, the core task is to execute fault diagnosis logic in parallel while data fusion is being performed, to identify various instrument faults in real time and trigger corresponding early warning and handling measures. The specific implementation steps are as follows: Step S601, Differential pressure detection branch ice blockage fault diagnosis, the system performs real-time variance monitoring on the first liquid level value H1, calculates the sliding variance value Var(H1) within 100 consecutive sampling points, and the time window corresponds to 0.5 seconds based on the 200Hz sampling frequency.

[0091] If Var(H1) is less than the preset threshold of 0.0001 square meters for five consecutive time windows, that is, H1 exhibits an extremely smooth horizontal straight line characteristic, while the second liquid level value H2 still maintains normal dynamic fluctuations and its variance is greater than 0.001 square meters, then it is determined that the pressure tapping pipe of the differential pressure detection branch has an ice blockage fault.

[0092] This phenomenon is caused by the pressure tapping line being blocked by ice crystals, preventing the pressure difference from being transmitted with changes in liquid level, resulting in a constant transmitter output.

[0093] Step S602, differential pressure detection branch vaporization interference diagnosis. The acoustic characteristics of vaporization interference are: when the low-temperature medium in the pressure tapping tube vaporizes due to local temperature rise, the bursting of bubbles will generate a wide-bandwidth, unsteady acoustic emission signal, the energy of which is mainly concentrated in the mid-to-high frequency range. According to the bubble bursting dynamics theory and experimental measurements, for air separation media such as liquid oxygen and liquid nitrogen, the acoustic pressure energy generated by bubble bursting has a significant characteristic peak in the frequency range of 50Hz to 500Hz. This frequency range is distinct from the low-frequency disturbances generated by the normal process fluid flow in the tower (usually below 20Hz), and also forms a frequency domain separability with the main frequency and harmonics of the large rotating machinery (such as air compressors and booster compressors) in the air separation unit (usually integer multiples of the equipment rotation speed frequency, commonly seen in narrowband peaks below 10Hz and above 100Hz).

[0094] The system uses the raw differential pressure signal Spectral analysis was performed, and a Fast Fourier Transform with a Hanning window was used to convert the time-domain signal to the frequency domain, extracting the total energy within the 50Hz to 500Hz frequency band. To eliminate environmental background noise interference, under the initial commissioning and stable process conditions, and with confirmation of no vaporization faults in the pressure tapping pipe, the differential pressure signal was continuously collected for 10 minutes to establish a background noise reference spectrum. The energy distribution characteristics of each frequency band (including the 50Hz to 500Hz band) were recorded as... And store it in non-volatile memory.

[0095] During real-time diagnostics, the system first performs spectral analysis on the current differential pressure signal to extract the real-time energy in the 50Hz to 500Hz frequency band. Simultaneously, the energy of non-sensitive frequency bands adjacent to this frequency band and without bubble characteristics (such as 30Hz to 48Hz, 520Hz to 600Hz) is extracted as a reference for environmental background noise. .like Exceed The set multiplier (typically 3x), and Significantly higher than (Typical multiple is 5 times or more), and if the variance of the first liquid level value H1 in the same time window exceeds 0.01 square meters (i.e., there is a large and irregular jump), then it is determined that there is bubble interference caused by medium vaporization in the pressure tapping tube.

[0096] Furthermore, the system adopts an adaptive frequency band adjustment mechanism: if the background noise energy in the 50Hz to 500Hz frequency band is found to increase significantly due to changes in equipment status (such as frequency converter carrier frequency adjustment or compressor operating conditions) during continuous operation, the system automatically performs dynamic segmentation of the frequency band, eliminates narrowband interference frequency points whose energy proportion exceeds the preset threshold, and retains only the remaining continuous frequency band as the gasification characteristic analysis window, thereby avoiding misjudgment caused by changes in industrial environmental noise.

[0097] Step S603: Diagnose the bonding status of the external ultrasonic detection branch sensor. The system monitors the signal-to-noise ratio (SNR) of the ultrasonic echo signal in real time, defined as the ratio of the echo peak amplitude to the noise floor amplitude. The noise floor is obtained by statistically analyzing the signal amplitudes of the first 100 sampling points of the transmitted pulse. A signal-to-noise ratio threshold is set. If the SNR is below 15 dB in 10 consecutive measurements, it is determined that the acoustic coupling medium is aging or the sensor is loose, resulting in a decrease in sound wave transmission efficiency.

[0098] Step S604: Warning Trigger and Operating Mode Switching. Once any fault type identified in steps S601 to S603 is detected, the system immediately triggers the warning mechanism. The central processing unit, according to a predefined fault status word encoding format, sends fault codes (e.g., ice blockage fault code 0x01, vaporization interference code 0x02, sensor loosening code 0x03) to the air separation unit's distributed control system via industrial Ethernet using the Modbus TCP protocol or PROFIBUS DP fieldbus. Simultaneously, the system automatically switches to a single-source reliable operating mode. If a differential pressure branch fault is detected, only the ultrasonic level value H2 is used as the output; if an ultrasonic branch fault is detected, only the differential pressure level value H1 is used as the output, ensuring that the level reference signal obtained by the main control system is continuous and accurate. All fault diagnosis information, including fault type, occurrence timestamp, and fault branch identifier, is stored in non-volatile memory for historical retrieval and fault tracing.

[0099] Through the above steps S6, the system achieves real-time diagnosis and automatic handling of typical faults in the two types of sensors, significantly improving the reliability of the measurement system.

[0100] In practical applications, the method of this invention also includes a complete initialization and self-calibration process. During system startup, a hardware self-test procedure is first executed: for differential pressure transmitters, the continuity of the current loop is determined by applying a known test current to the transmitter and detecting the current response in the loop; for ultrasonic transducers, the static impedance value of the transducer in the unexcited state is measured and compared with the standard impedance range; if it exceeds the normal range, the excitation impedance is determined to be abnormal; for the processing unit, the integrity of the memory is verified by writing a set of preset data to a memory address and then reading it back for verification. Only after all three self-tests pass can the system enter the real-time monitoring state.

[0101] The initial state vector of the adaptive Kalman filter is directly taken as the first liquid level value of the differential pressure detection branch at the moment of startup. The initial error covariance matrix is ​​set according to the sensor's factory accuracy specifications: the differential pressure transmitter has an accuracy class of 0.075, corresponding to a maximum error of 0.075% of the range. When the range is 1.6 MPa, the maximum error is 1.2 kPa, which translates to a liquid level error of approximately 0.012 meters. Therefore, the diagonal elements of the initial covariance matrix are taken as the square of 0.012 meters, i.e., 1.44 × 10⁻⁶. -4 Square meters; the accuracy of ultrasonic level gauges is typically 0.5% of the measuring range or ±2mm, taking the larger value of 0.5%, meaning an error of 0.05 meters for a 10-meter measuring range. Therefore, the element corresponding to the ultrasonic branch in the initial covariance matrix is ​​taken as the square of 0.05 meters, i.e., 2.5 × 10⁻⁶. - ³ square meters. This initialization setting can shorten the time it takes for the filter to reach a stable convergence state.

[0102] When the fractionation column is shut down or the liquid level is at a known calibration point, such as full or empty, the system automatically performs a periodic self-calibration step. At this time, process conditions are stable, and the true liquid level value is known. The system simultaneously records the first liquid level value output from the differential pressure detection branch and the second liquid level value output from the ultrasonic detection branch, calculating the static deviation between the two. This is calculated by subtracting the true liquid level value from the first liquid level value and subtracting the true liquid level value from the second liquid level value, respectively, yielding their respective zero-point offsets. These offsets are updated and stored in non-volatile memory as zero-point calibration parameters and used in subsequent calculations to correct the corresponding liquid level measurements, thereby eliminating zero-point drift that may occur during long-term operation.

[0103] In summary, this invention provides a method for detecting the liquid level at the bottom of a distillation column. By constructing a complete technical solution integrating hardware deployment, signal processing, data fusion, and intelligent diagnosis, it achieves high-precision and high-reliability measurement of the liquid level inside the column under cryogenic conditions.

[0104] This method employs a heterogeneous redundancy design that combines contact differential pressure detection with non-contact external ultrasonic detection, fundamentally overcoming the problem that traditional single measurement methods are prone to failure and cannot be maintained online under extreme conditions such as pressure pipe ice blockage and medium vaporization.

[0105] At the signal processing level, density dynamic compensation and sound velocity compensation are used to eliminate the influence of operating condition changes on the two measurement principles, respectively. An adaptive threshold algorithm is used to effectively suppress clutter interference, ensuring the calculation accuracy of the two liquid level values. On this basis, an adaptive Kalman filter algorithm is introduced to dynamically adjust the process noise according to the feed and discharge flow rates and estimate the observation noise in real time based on the residual, achieving optimal fusion of the two measurement values ​​and significantly improving the stability and reliability of the output results.

[0106] Meanwhile, the system's built-in fault diagnosis logic can identify typical faults such as ice blockage, vaporization interference, and sensor loosening in real time, and automatically trigger warnings and mode switching to ensure the continuity of measurement functions.

[0107] The entire method is logically rigorous and interconnected, which not only solves the long-standing pain points in existing technologies, but also achieves a comprehensive improvement in measurement accuracy, system reliability and intelligence, providing a practical and advanced technical solution for the level detection of distillation towers in the field of air separation.

[0108] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects.

[0109] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for detecting the liquid level at the bottom of a distillation column, characterized in that, Includes the following steps S1 to S6: Step S1: Deploy the detection branch, which includes a contact differential pressure detection branch and a non-contact external ultrasonic detection branch; Step S2: The original differential pressure signal of the medium inside the tower is collected through the contact differential pressure detection branch, and the first liquid level value representing the liquid level height is calculated by combining the synchronously acquired real-time operating parameters inside the tower. Step S3: Through the non-contact external ultrasonic detection branch, ultrasonic pulses are emitted and received into the medium inside the tower, the flight time of the ultrasonic waves in the medium is measured, and the second liquid level value representing the liquid level height is calculated by combining the pre-established sound velocity compensation model. Step S4: The first liquid level value and the second liquid level value are used as observation variables and input into an adaptive Kalman filter for multi-source data fusion to obtain the optimal liquid level estimate used to characterize the true liquid level in the tower. Step S5: While performing data fusion, the fault diagnosis logic runs in parallel. By analyzing the signal characteristics of the first liquid level value and the second liquid level value, the fault type of the detection branch is identified in real time, and an early warning is triggered or the system switches to a single-path reliable operation mode based on the diagnosis results.

2. The method for detecting the liquid level at the bottom of a distillation column according to claim 1, characterized in that, The specific steps of deploying the non-contact external ultrasonic detection branch in step S1 include: installing multiple ultrasonic transducers at equal intervals along an axis perpendicular to the ground on the outer wall surface of the lower column of the fractionation tower to form a vertical sensor array. The total number of ultrasonic transducers installed is determined based on the total height of the lower column of the distillation tower to ensure effective coverage of the full range of liquid levels. Each ultrasonic transducer is filled with a special low-temperature resistant acoustic coupling medium between itself and the metal tower wall. The coupling medium can maintain viscoelasticity and low acoustic attenuation characteristics in a cryogenic environment to maximize the transmission efficiency of the acoustic signal.

3. The method for detecting the liquid level at the bottom of a distillation column according to claim 2, characterized in that, The step of calculating the first liquid level value in step S2 further includes a density dynamic compensation process: Based on the pressure and temperature values ​​in the real-time operating parameters inside the tower, the real-time density value of the medium inside the tower under the current operating conditions is determined by looking up tables and interpolation calculations in the pre-set air component thermodynamic property database in the system's non-volatile memory. By using additional temperature sensors arranged on the positive pressure side pressure tap and the negative pressure side pressure tap, the reference temperature of the liquid column in the pressure tap is obtained, and the density value of the liquid column in the pressure tap is calculated using a thermodynamic property database. The density value of the liquid column inside the pressure tapping tube is used to correct the original differential pressure signal to eliminate the measurement error caused by the temperature difference between the inside and outside of the cold box. Based on the principles of hydrostatics, a high-precision first liquid level value is calculated using the corrected differential pressure value and the real-time density value of the medium inside the tower.

4. The method for detecting the liquid level at the bottom of a distillation column according to claim 2, characterized in that, The step of calculating the second liquid level value in step S3 further includes a dynamic sound velocity compensation process: The temperature of the tower wall is monitored in real time by temperature sensors placed on the outside of the tower wall, and the propagation speed of ultrasonic waves in the tower wall at the current tower wall temperature is determined according to the pre-stored temperature-sound velocity correspondence of the tower wall material. Based on the pre-measured tower wall thickness, coupling layer thickness, and corresponding sound velocity, calculate the fixed time offset introduced by the ultrasonic wave in the propagation path due to passing through the tower wall and coupling layer twice. Based on the temperature value and medium composition in the real-time operating parameters inside the tower, the actual propagation speed of ultrasound in the current liquid medium is searched and determined in the pre-set thermodynamic property database. The precise second liquid level value is calculated using the ultrasonic flight time, fixed time offset, and actual sound velocity in the medium.

5. The method for detecting the liquid level at the bottom of a distillation column according to claim 1, characterized in that, The fusion process of the adaptive Kalman filter in step S4 further includes adaptive adjustment of process parameter coupling: The feed and discharge flow rates of the fractionation tower are collected in real time, and the net flow rate change rate, which characterizes the fluctuation of material within the tower, is calculated. The process noise variance in the Kalman filter is dynamically corrected based on the net flow rate change, so that the filter trusts the real-time observations more when the flow rate fluctuates drastically. Based on the observation residuals of the first and second liquid level values ​​at the current moment, a recursive algorithm is used to update the observation noise covariance matrix in real time to adaptively evaluate the confidence level of the two sensors under the current operating conditions.

6. The method for detecting the liquid level at the bottom of a distillation column according to claim 1, characterized in that, The fault diagnosis logic in step S5 specifically includes ice blockage fault diagnosis: Real-time sliding variance monitoring is performed on the calculated first liquid level value; If, within multiple consecutive time windows, the sliding variance of the first liquid level value remains below the preset minimum threshold, exhibiting an extremely smooth horizontal straight line characteristic, while the second liquid level value maintains normal dynamic fluctuations, then it is determined that the pressure tap of the contact differential pressure detection branch has an ice blockage fault.

7. The method for detecting the liquid level at the bottom of a distillation column according to claim 1, characterized in that, The fault diagnosis logic in step S5 specifically includes vaporization interference diagnosis: Perform spectral analysis on the acquired differential pressure raw signal to extract the total energy within the preset high-frequency band; Compare the current total high-frequency energy with the baseline energy stored when the system was initially put into operation and the process was stable; If the current total high-frequency energy exceeds the set multiple of the reference energy, and within the same time window, the variance of the first liquid level value exceeds the preset large jump threshold, then it is determined that there is bubble interference caused by medium vaporization in the pressure tapping tube.

8. The method for detecting the liquid level at the bottom of a distillation column according to claim 1, characterized in that, The fault diagnosis logic in step S5 specifically includes sensor fit status diagnosis: The signal-to-noise ratio (SNR) of ultrasonic echo signals is monitored in real time. The SNR is defined as the ratio of the amplitude of the main peak of the echo to the amplitude of the noise floor. If the signal-to-noise ratio is lower than the preset signal-to-noise ratio threshold in multiple consecutive measurements, it is determined that the acoustic coupling medium is aging or the sensor is loose, resulting in a decrease in sound wave transmission efficiency.

9. The method for detecting the liquid level at the bottom of a distillation column according to claim 1, characterized in that, The steps in step S5 to switch to single-path reliable operation mode include: When the fault diagnosis logic determines that the contact differential pressure detection branch is faulty, the system automatically shields the data of the faulty branch and uses only the second liquid level value obtained by the non-contact external ultrasonic detection branch as the final output. When a fault is detected in the non-contact external ultrasonic detection branch, the system uses only the first liquid level value obtained from the contact differential pressure detection branch as the final output, ensuring that the main control system always obtains a continuous and reliable liquid level reference signal.

10. The method for detecting the liquid level at the bottom of a fractionation column according to claim 1, characterized in that, It also includes a hardware self-test and initialization step performed during the system startup phase, which is executed before step S1: A current loop continuity test is performed on the differential pressure transmitter. Its status is determined by applying a test current to the transmitter and detecting the loop response. A static impedance test is performed on the ultrasonic transducer. By measuring its impedance value in the unexcited state and comparing it with the standard range, it is determined whether there is any abnormality. The central processing unit performs memory integrity verification by writing and reading back preset data to check the memory status; Only after all three self-tests above pass will the system initialize the filter and enter real-time monitoring mode.

Images