Intelligent Separation and Detection Method for Multiphase Mixtures at Inspection Stations

Abstract

This invention relates to the field of forensic identification technology, specifically to an intelligent separation and detection method for multiphase mixtures at inspection stations. The method includes: transmitting broadband ultrasonic pulses to the multiphase mixture sample to be tested to construct an attenuation spectrum matrix; extracting the volume-weighted particle size distribution density function; calculating a mapping table between the particle size distribution span index and particle size and volume fraction; determining the number of stages and dividing the particle size sub-intervals based on the particle size distribution span index; calculating the equivalent separation acceleration to construct a set of staged separation parameters; performing separation, calculating the single-stage separation efficiency and performing online correction, and collecting the separation products of each stage; performing component detection on the separation products of each stage to obtain the component confidence vector, projecting the maximum value to obtain the global detection confidence, and triggering an alarm when a threshold is reached. This invention eliminates the systematic detection blind spot caused by the excessively large particle size span in samples with a wide particle size distribution.

Description

Technical Field

[0001] This invention relates to the field of forensic identification technology, specifically to an intelligent separation and detection method for multiphase mixtures at checkpoints. Background Technology

[0002] In law enforcement scenarios such as customs, border crossings, and security checkpoints, seized liquid samples are often not single homogeneous liquids, but multiphase mixtures containing various substances such as aqueous phase, oil phase, and solid particles. Accurately separating and detecting the target components from these complex mixtures within the limited time available at the scene is the core technical challenge for forensic identification at checkpoints.

[0003] Existing technologies typically employ a process of fixed-parameter separation followed by end-of-pipe detection. Operators inject samples into a separation device with preset parameters such as centrifugation speed, membrane pore size, or sedimentation time. After separation, the products of each phase are then subjected to spectral or chromatographic detection. This process is stable when the sample properties are relatively uniform and has been applied in some port laboratories.

[0004] When the droplet size distribution of the dispersed phase in the mixture under test varies greatly, a fixed-parameter separation scheme can create a systematic detection blind zone. Specifically, for samples with a large droplet size distribution, the droplet size can span multiple orders of magnitude from submicron to hundreds of micrometers. If the separation parameters are set with the goal of handling large droplets, submicron-sized droplets, due to insufficient separation driving force, will penetrate with the continuous phase, carrying the target component directly into the waste liquid end, leading to missed detection. If the separation intensity is increased with the goal of handling microdroplets, large droplets will coalesce and clog the inlet end, causing a sharp drop in separation throughput or even device failure. This contradiction is particularly prominent in samples from inspection stations. Due to the diverse disguise methods of smuggled goods and the significant differences in the degree of emulsification of mixtures, a single fixed parameter cannot cover the entire particle size range, resulting in an uncontrollable risk of systematic missed detections in the test results. Summary of the Invention

[0005] To address the problem that fixed-parameter separation schemes generate systematic detection blind spots across the entire particle size range when the dispersed phase particle size distribution of multiphase mixtures is too large, this invention provides an intelligent separation and detection method for multiphase mixtures at inspection stations.

[0006] This invention provides an intelligent separation and detection method for multiphase mixtures at inspection stations, comprising: transmitting broadband ultrasonic pulses to the multiphase mixture sample to be inspected and acquiring the acoustic attenuation coefficient sequence to construct an attenuation spectrum matrix; performing inversion calculation on the attenuation spectrum matrix to extract the volume-weighted particle size distribution density function, and then calculating a mapping table between the particle size distribution span index and particle size and volume fraction; determining the number of stages based on the particle size distribution span index, dividing the particle size axis into multiple particle size sub-intervals corresponding to the number of stages; for each particle size sub-interval, calculating the corresponding equivalent separation acceleration based on the inverse solution of Stokes' sedimentation law; and... A particle size and volume fraction mapping table constitutes a set of separation parameters for staged separation. Separation is performed according to the set of separation parameters. The inlet and outlet turbidity of each stage separation chamber are collected in real time to calculate the single-stage separation efficiency. In response to the single-stage separation efficiency being lower than the efficiency lower limit threshold, the equivalent separation acceleration is corrected online, and the separation products of each stage are collected. Component detection is performed on the separation products of each stage to obtain the component confidence vector. The maximum value of the component confidence vector of each stage is projected to obtain the global detection confidence. In response to the global detection confidence reaching the detection confidence threshold, the presence of the target component is determined and an alarm is triggered.

[0007] The objective characterization of sample particle size distribution is obtained by acoustic attenuation spectrum inversion, and the particle size distribution span is used as the physical basis for determining the number of stages. The original single-stage separation with fixed parameters is extended to multi-stage series separation with adaptive parameters, thereby eliminating the systematic detection blind zone caused by the large particle size span in samples with wide particle size distribution.

[0008] Preferably, the volume-weighted particle size distribution density function is extracted by inverting the attenuation spectrum matrix, and then the particle size distribution span index and the mapping table of particle size and volume fraction are calculated. This includes: taking the attenuation spectrum matrix as input, solving by minimizing the sum of squared residuals between the measured attenuation and the model predicted attenuation, and extracting the volume-weighted particle size distribution density function; obtaining the characteristic particle size values ​​corresponding to the cumulative volume distribution reaching different preset proportions based on the volume-weighted particle size distribution density function; calculating the difference between the coarse end characteristic diameter and the fine end characteristic diameter, and using the ratio of the difference to the median diameter in the volume as the particle size distribution span index; and statistically analyzing the volume fraction corresponding to each particle size interval to finally obtain the mapping table of particle size and volume fraction.

[0009] By minimizing the sum of squared residuals, the accuracy of the particle size distribution density function extraction is ensured, providing a reliable data foundation for the subsequent calculation of classification parameters.

[0010] Preferably, determining the number of stages based on the particle size distribution span index includes: obtaining the single-stage effective particle size coverage ratio corresponding to the membrane pore size adjustment range of a single-stage separation chamber, and converting the single-stage effective particle size coverage ratio into a single-stage coverage threshold; setting multiple stage boundaries based on the single-stage coverage threshold; comparing the particle size distribution span index with the stage boundaries, and determining the corresponding number of stages based on the comparison results.

[0011] Preferably, for each particle size sub-range, the equivalent separation acceleration is calculated based on the inverse solution of Stokes' sedimentation law, including: obtaining the dynamic viscosity of the continuous phase, the density of the dispersed phase, and the density of the continuous phase of the multiphase mixture sample to be tested; and, based on the preset target sedimentation velocity and the target lower limit of the particle size corresponding to the particle size sub-range, combined with the dynamic viscosity of the continuous phase, the density of the dispersed phase, and the density of the continuous phase, the equivalent separation acceleration is calculated in reverse, so that the equivalent separation acceleration is negatively correlated with the square of the target lower limit of the particle size.

[0012] By using Stokes' law to calculate sedimentation velocity in a forward direction and then applying it in reverse to derive the required separation acceleration for each stage, the driving force for separation at each stage is determined by the sample physical properties and the device geometric parameters.

[0013] Preferably, the inlet and outlet turbidity of each separation chamber are collected in real time to calculate the single-stage separation efficiency. In response to the single-stage separation efficiency being lower than the lower efficiency threshold, the equivalent separation acceleration is corrected online, including: calculating the ratio of inlet turbidity to outlet turbidity, and calculating the single-stage separation efficiency based on the ratio; conducting batch separation experiments using standard samples during the calibration stage, and statistically analyzing the distribution mean and standard deviation under normal operating conditions to set the lower efficiency threshold; in response to the single-stage separation efficiency being lower than the lower efficiency threshold, increasing the equivalent separation acceleration by a preset step size until the single-stage separation efficiency reaches the lower efficiency threshold or the device safety upper limit, thus completing the online correction of the equivalent separation acceleration.

[0014] By using online turbidity monitoring and threshold determination based on calibration statistical characteristics, real-time perception of the separation efficiency at each stage and online parameter correction are achieved.

[0015] Preferably, the equivalent separation acceleration is increased in increments of a preset step size, including: measuring the rotational speed and efficiency response curves of each separation chamber; extracting the minimum equivalent acceleration increment corresponding to when the efficiency change exceeds the standard deviation; and using the minimum equivalent acceleration increment as the preset step size to control the increase of the equivalent separation acceleration.

[0016] Preferably, the separation products at each stage are collected, including: recording the actual efficiency value for extreme conditions where the equivalent separation acceleration reaches the upper limit of the device safety and the single-stage separation efficiency still does not reach the lower limit threshold of efficiency; marking the separation products at the corresponding level as partially separated and marking the confidence downgrade warning in the final test report; collecting the dispersed phase droplets trapped at each stage into independent sample bottles and collecting the continuous phase from the final stage outlet to finally obtain the separation products at each stage.

[0017] Preferably, component confidence vectors are obtained by performing component detection on the products separated at each stage, including: for each stage of retentates enriched with dispersed phase droplets, Raman spectroscopy is used to perform component fingerprint comparison to obtain the corresponding spectral matching confidence; for continuous phase products, ion mobility spectrometry is used to detect dissolved target substances to obtain the corresponding spectral matching confidence; and the spectral matching confidence of multiple target components in each stage of products is summarized to construct the component confidence vector.

[0018] Preferably, the global detection confidence is obtained by projecting the maximum value of the confidence vector of each component level, including: extracting multiple spectral matching confidence values ​​corresponding to the same target component from the confidence vector of each component level; selecting the maximum value among the multiple spectral matching confidence values; and using the maximum value as the global detection confidence value corresponding to the target component.

[0019] Preferably, in response to the global detection confidence level reaching the detection confidence threshold, determining the presence of the target component and triggering an alarm includes: calculating the minimum confidence level and standard deviation of all known positive samples on the calibrated sample set; subtracting three times the standard deviation from the minimum confidence level to obtain the detection confidence threshold; comparing the global detection confidence level with the detection confidence threshold, and in response to the global detection confidence level being not less than the detection confidence threshold, determining the presence of the target component and triggering an alarm.

[0020] This invention extends the original single-stage separation with fixed parameters to multi-stage series separation with adaptive parameters, thereby eliminating the systematic detection blind zone caused by the large particle size span in samples with a wide particle size distribution. The driving force of each separation stage is derived by inverse solution of Stokes' law, the grading boundary is determined by the inherent geometric parameters of the device, and the online correction threshold and step size are generated by the statistical characteristics of the calibration data.

[0021] For extreme samples that have not been fully separated even when the device reaches its safety limit, a downgrade labeling method is used instead of blocking to ensure continuous output of test results, which effectively improves the robustness of the device and the reliability of test results under complex field sample conditions. By independently detecting the products of each separation stage and taking the maximum value projection for comprehensive judgment, the spatial distribution information of components obtained from multi-stage separation is used for the final judgment, so that an alarm can be triggered as long as effective enrichment is achieved at any stage, thereby eliminating the risk of missed judgment due to incomplete stage separation. Attached Figure Description

[0022] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent upon reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the invention are illustrated by way of example and not limitation, and like or corresponding reference numerals denote like or corresponding parts, wherein: Figure 1 This invention relates to a flowchart of an intelligent separation and detection method for multiphase mixtures at inspection stations.

[0023] Figure 2 This invention relates to a two-dimensional planar scatter plot comparing the relationship between particle size distribution span index and overall separation efficiency. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0026] This invention discloses an intelligent separation and detection method for multiphase mixtures at inspection stations, referring to... Figure 1 This includes steps S1-S4: S1. Construct the attenuation spectrum matrix and extract the particle size distribution features.

[0027] In an optional embodiment, the multiphase mixture sample to be tested is injected into the front-end phase characterization unit. A broadband ultrasonic pulse with a frequency range covering 1 MHz to 100 MHz is emitted to the sample through the acoustic attenuation spectrum sensor array built into the unit, and the acoustic attenuation coefficient sequences at different frequencies are collected to construct an attenuation spectrum matrix. ,in This represents the number of frequency sampling points. The time sampling frame number is used. Since the scattering cross-section of droplets of different sizes exhibits different responses to sound waves with frequency, by covering the frequency band from 1 MHz to 100 MHz, droplets ranging from submicron to hundreds of micrometers fall into the effective scattering response range, thus ensuring that subsequent inversion calculations have sufficient frequency domain information to support them.

[0028] Based on the physical mapping relationship between the attenuation of sound waves at a specific frequency and the scattering cross section of droplets of corresponding particle size, a regularized inversion method based on the ECAH acoustic scattering model is used to analyze the attenuation spectrum matrix. Inversion calculations are performed. Specifically, using the attenuation spectral matrix... Using the particle size distribution density function as input and the particle size distribution density function as output, the solution is obtained by minimizing the sum of squared residuals between the measured attenuation and the model-predicted attenuation, thus extracting the volume-weighted particle size distribution density function of the dispersed phase. Then the median diameter of the volume was calculated. Fine end characteristic diameter and coarse end feature diameter ,in , and These represent the particle size values ​​corresponding to cumulative volume distributions of 10%, 50%, and 90%, respectively, in units of... .

[0029] Based on this, the particle size distribution span index is calculated according to the following formula:

[0030] The value reflects the breadth of the dispersed phase particle size distribution and is a core input parameter for determining the number of fractions. Simultaneously, the volume fraction corresponding to each particle size interval is statistically analyzed to construct a discretized particle size and volume fraction mapping table. This serves as the basis for the input data in subsequent calculations of the classification parameters.

[0031] It should be noted that there are many other ways to achieve wide particle size range characterization in this field. Those skilled in the art can choose different acoustic or optical particle size measurement methods according to actual needs. ECAH model inversion is only a preferred implementation method.

[0032] Thus, the volume-weighted particle size distribution density function covering the entire particle size range can be extracted from complex multiphase mixtures. The span index provides physical input for determining the number of subsequent grades and parameters at each grade.

[0033] S2. A set of separation parameters is constructed based on the particle size span index.

[0034] In an optional embodiment, with Value as a grade number The basis for determination is that the membrane pore size adjustment range of a single-stage separation chamber is set as follows: Define the single-stage effective particle size coverage ratio This value is an inherent geometric parameter of the device and can be obtained through actual measurement during factory calibration. The particle size range that a single-stage separation chamber can effectively cover corresponds to approximately Order of magnitude, convert it to Contribution amount, to obtain the single-level coverage threshold The grading boundaries are as follows: and For the standard centrifuge-membrane composite separation chamber used in this scheme, The typical measured range makes Approximately 1.5, corresponding to the following grading rules: When When the particle size distribution is concentrated, a single-stage separation can cover the main particle size range. At this time, the processor will take... ;when When the value is in the range of medium width, it indicates that the distribution is of medium width, at which point the processor will take... ;when When this occurs, it indicates that the distribution spans multiple orders of magnitude, at which point the processor will take... .

[0035] Sure Then, the particle size axis was divided into equal logarithmic intervals. Sub-intervals , From 1 to Corresponding to level 1 to level 2 in sequence The separation units are arranged from coarse to fine. For each sub-interval, the minimum equivalent separation acceleration required for the droplets in that size range is solved using Stokes' sedimentation law. The original Stokes sedimentation velocity formula is:

[0036] At the target settlement velocity Under constraints, for the minimum processing particle size of this stage Solve for the required equivalent separation acceleration The corresponding relationship is:

[0037] in, The dynamic viscosity of the continuous phase is expressed in Pa·s. and These are the densities of the dispersed phase and the continuous phase, respectively, in kg / m³. This represents the lower limit of the target particle size for this stage, in meters (m). The target settling velocity is determined by the ratio of the designed residence time of the sample in the separation chamber to the characteristic dimension of the separation chamber. From this relationship, it can be seen that... and The square of the acceleration is negatively correlated, which makes the acceleration at each level increase monotonically from the coarse end to the fine end.

[0038] Furthermore, With centrifuge speed The conversion relationship between them is as follows:

[0039] in, Let be the radius of gyration of the separation chamber. Using this conversion relationship, the calculated value will be... Converted into a speed setpoint that the controller can write. The calculated values ​​from each stage... The sequence and the corresponding particle size sub-region boundaries together constitute the hierarchical separation parameter set. And distribute it to the separate execution unit.

[0040] Thus, by using Stokes' law to calculate the settling velocity in the forward direction and then applying it in reverse to derive the required separation acceleration step by step, the driving force for separation at each stage is determined by the sample physical properties and the device geometric parameters.

[0041] S3. Perform multi-stage separation and correct the separation acceleration online.

[0042] The sample flows sequentially from the first coarse end through The separation execution unit is composed of a series of programmable centrifuge-membrane composite separation chambers. The centrifugal speed and membrane pore size of each chamber are determined according to the parameter set corresponding to the stage. and The controller is programmed to retain dispersed phase droplets within the corresponding particle size range at each stage, while continuous phase and droplets with finer particle sizes penetrate into the next stage.

[0043] In an optional embodiment, online turbidity sensors are deployed at the inlet and outlet of each separation chamber to collect inlet turbidity data in real time. and outlet turbidity The unit is NTU. Since turbidity is linearly proportional to the volume concentration of the dispersed phase, the turbidity is defined according to the following relationship: Single-stage separation efficiency:

[0044] This reflects the degree to which the stage retains droplets within the target particle size range. A higher value indicates more complete retention, while a lower value indicates a deviation between the separation parameters of this stage and the actual droplet size.

[0045] During the calibration phase before the device is put into use, batch separation experiments are conducted using standard samples of known particle sizes covering each target particle size range, and statistics are compiled for each level. Distribution mean under normal operating conditions and standard deviation The efficiency threshold that will trigger online correction will be set according to the following formula:

[0046] Response at any level The controller determines that there is a deviation between the separation parameters of this stage and the actual droplet size, and triggers online correction of the parameters for this stage: By step size Increment, until Or until the device reaches its safety limit. By measuring the rotational speed and efficiency response curves of each separation chamber, the efficiency change exceeding... The minimum equivalent acceleration increment corresponding to the time is used as the online correction step size. .

[0047] against Reaching the device's safety limit Not yet reached Under extreme operating conditions, the controller records the actual efficiency value of that stage, marks the product of that stage as partially separated, and adds a corresponding confidence level downgrade warning in the final test report to ensure the process continues to the next stage. The dispersed phase droplets retained at each stage are collected into separate sample vials, and the continuous phase is collected from the final stage outlet, yielding a total of... The separated products are then used for subsequent testing.

[0048] In this way, through online turbidity monitoring and threshold determination based on calibration statistical characteristics, real-time perception of the separation efficiency at each stage and online parameter correction are achieved.

[0049] S4. Obtain the component confidence vector to determine the target component.

[0050] In an optional embodiment, for The separated products were subjected to component analysis. For each stage of the retained products enriched with dispersed phase droplets, Raman spectroscopy was used to scan and the component fingerprints were compared using molecular vibrational characteristic peaks. For the continuous phase products, ion mobility spectrometry was used to detect the dissolved target compounds. The results of each stage of analysis were expressed as a component confidence vector. It means that among them The total number of components in the target component database. For the first The first product in the grade The spectral matching confidence level of the target component ranges from 0 to 1.

[0051] When making a comprehensive judgment, the maximum value of the result at each level is projected:

[0052] Projecting the maximum value of the detection results at each level, the highest spectral matching confidence level for the same target component at each level is calculated. This serves as the global detection confidence level for the component, ensuring that an alarm is triggered once effective enrichment is achieved at any level.

[0053] Furthermore, the minimum confidence level and standard deviation of all known positive samples are statistically analyzed on the calibration sample set. The minimum confidence level minus three times the standard deviation is taken as the detection confidence threshold. Then, the global detection confidence level is compared with the detection confidence threshold. In response to Determine the target component An alarm is triggered if the condition is present. The final output includes the detection results for each target component, the corresponding enrichment level, the confidence level, and the actual efficiency status of the separated products at each level, forming a complete detection report.

[0054] like Figure 2The figure shown is a two-dimensional scatter plot comparing the relationship between particle size distribution span index and overall separation efficiency in an embodiment of the present invention. It can be seen that the sample points of the existing method are concentrated in the medium span range and the overall efficiency is relatively low, while the sample points of the three grades of the present invention maintain high separation efficiency in the wide span range, indicating that the present invention has a stable and efficient separation capability for different particle size distribution spans.

[0055] In this way, by independently detecting the products of each separation stage and taking the maximum value projection for comprehensive judgment, the spatial distribution information of components obtained from multi-stage separation is used for the final judgment, so that an alarm can be triggered as long as effective enrichment is achieved at any stage, thereby eliminating the risk of missed judgment due to incomplete separation.

Claims

1. A method for intelligent separation and detection of multiphase mixtures at inspection stations, characterized in that, include: Broadband ultrasonic pulses are emitted to the multiphase mixture sample to be tested and the acoustic attenuation coefficient sequence is collected to construct an attenuation spectrum matrix. The attenuation spectrum matrix is ​​inverted to extract the volume-weighted particle size distribution density function, and then the particle size distribution span index and the mapping table of particle size and volume fraction are calculated. The number of stages is determined based on the particle size distribution span index. The particle size axis is divided into multiple particle size sub-intervals corresponding to the number of stages. For each particle size sub-interval, the equivalent separation acceleration is calculated based on the inverse solution of Stokes' sedimentation law. The set of stage separation parameters is constructed by combining the particle size and volume fraction mapping table. Separation is performed according to the set of staged separation parameters. The inlet and outlet turbidity of each stage separation chamber are collected in real time to calculate the single-stage separation efficiency. In response to the single-stage separation efficiency being lower than the efficiency lower limit threshold, the equivalent separation acceleration is corrected online, and the separation products of each stage are collected. The components of each separation product are detected to obtain a component confidence vector. The global detection confidence is obtained by projecting the maximum value of each component confidence vector. When the global detection confidence reaches the detection confidence threshold, the presence of the target component is determined and an alarm is triggered.

2. The intelligent separation and detection method for multiphase mixtures at inspection stations according to claim 1, characterized in that, The step of inverting the attenuation spectrum matrix to extract the volume-weighted particle size distribution density function, and then calculating the mapping table between the particle size distribution span index and the particle size and volume fraction, includes: Using the attenuation spectrum matrix as input, the volume-weighted particle size distribution density function is extracted by minimizing the sum of squared residuals between the measured attenuation and the model-predicted attenuation. Based on the volume-weighted particle size distribution density function, the characteristic particle size values ​​corresponding to the cumulative volume distribution reaching different preset ratios are obtained; Calculate the difference between the coarse end characteristic diameter and the fine end characteristic diameter, and use the ratio of the difference to the volume median diameter as the index of the particle size distribution span; The volume fraction corresponding to each particle size range is statistically analyzed to obtain the particle size and volume fraction mapping table.

3. The intelligent separation and detection method for multiphase mixtures at inspection stations according to claim 1, characterized in that, The determination of the number of grades based on the particle size distribution span index includes: Obtain the single-stage effective particle size coverage ratio corresponding to the membrane pore size adjustment range of the single-stage separation chamber, and convert the single-stage effective particle size coverage ratio into a single-stage coverage threshold. Multiple tier boundaries are set based on the single-level coverage threshold; The particle size distribution span index is compared with the grading boundary, and the corresponding grading number is determined based on the comparison result.

4. The intelligent separation and detection method for multiphase mixtures at inspection stations according to claim 1, characterized in that, The calculation of the equivalent separation acceleration for each particle size sub-range, based on the inverse solution of Stokes' sedimentation law, includes: Obtain the dynamic viscosity of the continuous phase, the density of the dispersed phase, and the density of the continuous phase of the multiphase mixture sample to be tested; Based on the preset target settling velocity and the target lower limit of the particle size corresponding to the particle size sub-range, combined with the dynamic viscosity of the continuous phase, the density of the dispersed phase and the density of the continuous phase, the equivalent separation acceleration is calculated in reverse, so that the equivalent separation acceleration is negatively correlated with the square of the target lower limit of the particle size.

5. The intelligent separation and detection method for multiphase mixtures at inspection stations according to claim 1, characterized in that, The real-time acquisition of inlet and outlet turbidity of each separation chamber to calculate single-stage separation efficiency, and the online correction of the equivalent separation acceleration in response to the single-stage separation efficiency being lower than the efficiency lower limit threshold, including: Calculate the ratio of the inlet turbidity to the outlet turbidity, and calculate the single-stage separation efficiency based on the ratio; During the calibration phase, standard samples were used to conduct batch separation experiments, and the distribution mean and standard deviation under normal operating conditions were statistically analyzed to set the efficiency lower limit threshold. In response to the single-stage separation efficiency being lower than the lower efficiency threshold, the equivalent separation acceleration is increased by a preset step size until the single-stage separation efficiency reaches the lower efficiency threshold or the device safety upper limit, thus completing the online correction of the equivalent separation acceleration.

6. The intelligent separation and detection method for multiphase mixtures at inspection stations according to claim 5, characterized in that, The step of increasing the equivalent separation acceleration by a preset step size includes: The rotational speed and efficiency response curves of each separation chamber were measured; The minimum equivalent acceleration increment corresponding to the change in extraction efficiency exceeding the standard deviation; The minimum equivalent acceleration increment is used as the preset step size to control the increase of the equivalent separation acceleration.

7. The intelligent separation and detection method for multiphase mixtures at inspection stations according to claim 1, characterized in that, The collection of products from each separation stage includes: For extreme operating conditions where the equivalent separation acceleration reaches the upper limit of device safety and the single-stage separation efficiency still does not reach the lower limit threshold of efficiency, the actual efficiency value is recorded; The corresponding level of separated products are marked as partially separated, and a confidence downgrade warning is marked in the final test report accordingly. The dispersed phase droplets retained at each stage are collected into separate sample vials, and the continuous phase is collected from the final stage outlet to obtain the separation products of each stage.

8. The intelligent separation and detection method for multiphase mixtures at inspection stations according to claim 1, characterized in that, The step of obtaining component confidence vectors by performing component detection on the separated products at each stage includes: For each stage of the trapped material enriched with dispersed phase droplets, Raman spectroscopy scanning was used to perform component fingerprinting to obtain the corresponding spectral matching confidence level. For continuous phase products, ion mobility spectrometry is used to detect dissolved target substances to obtain the corresponding spectral matching confidence level; The confidence scores of spectral matching for multiple target components in each stage of the product are summarized to construct the component confidence vector.

9. The intelligent separation and detection method for multiphase mixtures at inspection stations according to claim 1, characterized in that, The process of projecting the maximum value of the confidence vectors of each level of components to obtain the global detection confidence includes: Extract multiple spectral matching confidence scores corresponding to the same target component from the confidence vectors of the components at each level; The maximum value is selected from the multiple spectral matching confidence scores; The maximum value is taken as the global detection confidence level corresponding to the target component.

10. The intelligent separation and detection method for multiphase mixtures at inspection stations according to claim 1, characterized in that, The response to the global detection confidence level reaching the detection confidence threshold, determining the presence of the target component and triggering an alarm, includes: Calculate the minimum confidence level and standard deviation of all known positive samples on the calibrated sample set; The detection confidence threshold is obtained by subtracting three times the standard deviation from the minimum confidence value. By comparing the global detection confidence level with the detection confidence level threshold, and in response to the global detection confidence level being no less than the detection confidence level threshold, it is determined that the target component exists and the alarm is triggered.

Images