Fluid sample loading dynamic calibration method and system based on sensor

Abstract

The invention relates to the technical field of automatic fluid treatment, and provides a fluid sample loading dynamic calibration method and system based on a sensor, and the method comprises the following steps: after executing calibration operation on a bioreactor, obtaining a first sensing signal output in a transition fluid disturbance observation stage by controlling the sensor of a fluid sample loading unit; acquiring a second sensing signal output in a stable sample physical property evaluation stage by controlling a sensor of a fluid sample loading unit; performing feature comparison on the first sensing signal and the second sensing signal, and determining a disturbance compensation amount according to a comparison result; performing feature comparison on the second sensing signal and a historical reference sensing signal, and determining a drift compensation amount according to a comparison result; according to the disturbance compensation amount and the drift compensation amount, differential compensation adjustment is carried out on driving parameters or sensor interpretation logic of the fluid sample loading unit. The method has the effect of continuously and accurately reflecting the sample loading volume of the undisturbed sample state.

Description

Technical Field

[0001] The present invention relates to the technical field of automated fluid processing, and in particular to a sensor-based fluid loading dynamic calibration method and system. Background Art

[0002] In applications such as biomanufacturing or long-term environmental monitoring, automated fluid loading systems serve as key components connecting bioreactors to subsequent analytical units, and their performance directly impacts the accuracy and reliability of the entire system. These systems typically require the regular extraction of trace samples for analysis with extremely high accuracy and reproducibility over long periods of continuous operation. The loading unit consists of a precision pump, sampling piping, and process monitoring sensors. Although the system performs an initial flow calibration at startup, the physical and chemical properties of the culture fluid change as the bioprocess progresses, and components such as pumps and piping can also experience performance drift due to long-term operation.

[0003] Therefore, the existing sensor-based fluid loading dynamic calibration method is difficult to effectively identify and compensate for the disturbance caused by the calibration operation itself to the sample microenvironment and its impact on the sensor response. It may become the source of introducing new volume measurement deviations and cannot continuously provide a loading volume that accurately reflects the undisturbed sample state.

[0004] In view of the above problems, the existing technology is in urgent need of improvement. Summary of the Invention

[0005] The purpose of this application is to provide a sensor-based fluid loading dynamic calibration method and system, which has the advantage of being able to effectively identify and compensate for the disturbance caused by the calibration operation itself to the sample microenvironment and its impact on the sensor response, thereby providing a loading volume that continuously and accurately reflects the undisturbed sample state.

[0006] This application provides a sensor-based fluid loading dynamic calibration method, the technical solution is as follows:

[0007] After performing a calibration operation on the bioreactor, a first sensor signal output during a transitional fluid disturbance observation phase is obtained by controlling a sensor of the fluid loading unit;

[0008] By controlling the sensor of the fluid loading unit, a second sensor signal outputted during the evaluation phase of the stable sample physical properties is obtained;

[0009] Performing a feature comparison on the first sensor signal and the second sensor signal, and determining a disturbance compensation amount based on the comparison result;

[0010] Performing a feature comparison between the second sensor signal and a historical reference sensor signal, and determining a drift compensation amount based on the comparison result;

[0011] According to the disturbance compensation amount and the drift compensation amount, differential compensation adjustment is performed on the driving parameters of the fluid loading unit or the sensor interpretation logic to complete the dynamic calibration of the fluid loading.

[0012] Through the above scheme, the disturbance caused by the calibration operation itself to the sample microenvironment and its impact on the sensor response can be effectively identified and compensated, thereby providing a sample volume that continuously and accurately reflects the undisturbed sample state.

[0013] Furthermore, the present application also proposes that, according to the above-mentioned sensor-based fluid loading dynamic calibration method, the driving parameters of the fluid loading unit or the sensor interpretation logic are differentially compensated and adjusted according to the disturbance compensation amount and the drift compensation amount to complete the dynamic calibration of the fluid loading, including the following steps:

[0014] Extracting the time characteristics and morphological characteristics of the first sensor signal; extracting the time characteristics and morphological characteristics of the second sensor signal;

[0015] Determining a reference morphological feature based on the morphological feature of the second sensing signal; comparing the morphological feature of the first sensing signal with the reference morphological feature to obtain a comparison result;

[0016] According to the comparison result, the time feature of the first sensor signal is corrected, or the proportion of the time feature of the first sensor signal in the calculation is modified;

[0017] Re-performing the process of determining the disturbance compensation amount and the process of determining the drift compensation amount according to the modified time characteristic of the first sensor signal or the proportion of the time characteristic in the calculation;

[0018] According to the re-determined disturbance compensation amount and drift compensation amount, differential compensation adjustment is performed on the driving parameters or sensor interpretation logic of the fluid loading unit to complete the dynamic calibration of the fluid loading.

[0019] Through the above solution, the accuracy of disturbance compensation and drift compensation is further improved.

[0020] Furthermore, the present application also proposes that, according to the above-mentioned sensor-based fluid loading dynamic calibration method, after performing a calibration operation on the bioreactor, the step of obtaining a first sensor signal output during the transitional fluid disturbance observation phase by controlling the sensor of the fluid loading unit includes:

[0021] After completing the calibration operation of the bioreactor, the fluid loading unit is controlled to inject a standard disturbance liquid into the inlet of the bioreactor to form a transitional fluid disturbance observation stage;

[0022] During the transitional fluid disturbance observation phase, continuous output data is collected through the pressure sensor and flow rate sensor of the fluid sampling path, and the data is subjected to time series filtering and window smoothing to obtain the first sensing signal.

[0023] Through the above scheme, the method of obtaining the disturbance signal is clarified and the signal quality is improved.

[0024] Furthermore, the present application also proposes that, according to the above-mentioned sensor-based fluid loading dynamic calibration method, the step of correcting the time characteristic of the first sensor signal or modifying the proportion of the time characteristic of the first sensor signal in the calculation according to the comparison result includes:

[0025] extracting interference pattern features from the morphological features of the first sensing signal;

[0026] comparing the interference pattern characteristics with a preset reference pattern to determine whether the change in morphological characteristics is attributable to the calibration operation performed;

[0027] When it is determined that the change in the morphological feature is due to the calibration operation that has been performed, the time feature of the first sensing signal is corrected, or the proportion of the time feature of the first sensing signal in the calculation is modified.

[0028] Through the above scheme, it is possible to determine whether the change in morphological features is due to the calibration operation, thereby avoiding misjudgment.

[0029] Furthermore, the present application also proposes that, according to the above-mentioned sensor-based fluid loading dynamic calibration method, the step of comparing the interference pattern characteristics with a preset reference pattern to determine whether the change in morphological characteristics is attributed to the performed calibration operation includes:

[0030] comparing the interference pattern characteristics with a preset reference pattern to determine whether the change in morphological characteristics is attributable to the calibration operation performed;

[0031] When the determination result is that the deviation is due to the calibration operation, the time feature of the first sensor signal is corrected, or the proportion of the time feature of the first sensor signal in the calculation is modified, and the interference pattern feature corresponding to the calibration operation is obtained as the deviation profile;

[0032] Update the baseline model based on the deviation appearance.

[0033] Through the above scheme, the benchmark model can dynamically adapt to the actual situation and improve the accuracy of judgment.

[0034] Furthermore, the present application also proposes that, according to the above-mentioned sensor-based fluid loading dynamic calibration method, the step of updating the reference pattern according to the deviation appearance includes:

[0035] The deviated appearance is weightedly combined with the baseline pattern before the update to generate the updated baseline pattern.

[0036] Through the above solution, a specific benchmark mode update mechanism is provided.

[0037] Furthermore, the present application also proposes that, according to the above-mentioned sensor-based fluid loading dynamic calibration method, the steps of weightedly combining the deviation appearance with the pre-updated reference pattern to generate an updated reference pattern include:

[0038] Obtain a historical deviated appearance sequence consisting of a plurality of deviated appearances;

[0039] Generate trend profiles based on historical deviation profile sequences;

[0040] Determine the difference between the trend appearance and the baseline pattern before the update as the difference degree;

[0041] According to the degree of difference, the weights used for weighted combination are adjusted;

[0042] The deviated appearance is weightedly combined with the baseline pattern before the update according to the adjusted weight for weighted combination to generate an updated baseline pattern.

[0043] Through the above scheme, the baseline model update is made more robust and intelligent.

[0044] Furthermore, the present application also proposes that, according to the above-mentioned sensor-based fluid loading dynamic calibration method, the steps of generating a trend profile based on a historical deviation profile sequence include:

[0045] The reception history deviates from the appearance sequence;

[0046] Smoothing the historical deviation appearance sequence;

[0047] Generate trend profile based on the smoothed historical deviation profile sequence.

[0048] Through the above solution, the accuracy of trend appearance is improved.

[0049] Furthermore, the present application also proposes that, according to the above-mentioned sensor-based fluid loading dynamic calibration method, after executing the step of obtaining a historical deviation appearance sequence consisting of a plurality of deviation appearances, the method further includes:

[0050] For each deviant appearance in the historical deviant appearance sequence, compare it with the baseline pattern;

[0051] Based on the comparison results, determine whether the deviated appearance is an abnormal data point;

[0052] Eliminate abnormal data points in the historical deviation appearance sequence.

[0053] Through the above scheme, the reliability of the historical deviation appearance sequence is improved.

[0054] This application also provides a sensor-based fluid loading dynamic calibration system, the technical solution is as follows:

[0055] A sensor-based fluid loading dynamic calibration system for performing sensor-based fluid loading dynamic calibration, comprising:

[0056] A first sensor signal acquisition module is configured to acquire a first sensor signal output during a transitional fluid disturbance observation phase by controlling a sensor of a fluid loading unit after performing a calibration operation on the bioreactor;

[0057] A second sensor signal acquisition module is used to acquire a second sensor signal output during the evaluation phase of the physical properties of the stable sample by controlling the sensor of the fluid loading unit;

[0058] a disturbance compensation amount acquisition module, configured to compare the characteristics of the first sensor signal and the second sensor signal, and determine the disturbance compensation amount according to the comparison result;

[0059] a drift compensation amount acquisition module, configured to compare the characteristics of the second sensor signal with the historical reference sensor signal and determine the drift compensation amount based on the comparison result;

[0060] The dynamic calibration execution module is used to perform differential compensation adjustments on the driving parameters or sensor reading logic of the fluid loading unit according to the disturbance compensation amount and the drift compensation amount, so as to complete the dynamic calibration of the fluid loading.

[0061] Through the above solution, a hardware / software carrier for implementing the above method is provided.

[0062] From the above, it can be seen that the sensor-based fluid loading dynamic calibration method and system provided in the present application obtains the transition fluid disturbance signal and the stable sample signal, compares and determines the disturbance compensation amount and the drift compensation amount, and performs differentiated compensation adjustment. It can effectively identify and compensate for the disturbance caused by the calibration operation itself to the sample microenvironment and its impact on the sensor response, thereby providing a loading volume that continuously and accurately reflects the undisturbed sample state. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of a method for dynamic calibration of fluid loading based on a sensor in one embodiment of the present invention;

[0064] Figure 2 This is one of the flow charts of a method for dynamic calibration of fluid loading based on a sensor in another embodiment of the present invention;

[0065] Figure 3 This is a second flow chart of a method for dynamic calibration of fluid loading based on a sensor in another embodiment of the present invention;

[0066] Figure 4 This is a third flow chart of a method for dynamic calibration of fluid loading based on a sensor in another embodiment of the present invention;

[0067] Figure 5 This is a fourth flow chart of a method for dynamic calibration of fluid loading based on a sensor in another embodiment of the present invention;

[0068] Figure 6 This is a fifth flow chart of a method for dynamic calibration of fluid loading based on a sensor in another embodiment of the present invention;

[0069] Figure 7 This is a sixth flow chart of a method for dynamic calibration of fluid loading based on a sensor in another embodiment of the present invention;

[0070] Figure 8 This is a seventh flow chart of a method for dynamic calibration of fluid loading based on a sensor in another embodiment of the present invention;

[0071] Figure 9 This is a system block diagram of a sensor-based fluid loading dynamic calibration system according to another embodiment of the present invention;

[0072] Description of reference numerals:

[0073] 1. Sensor-based fluid loading dynamic calibration system; 11. First sensor signal acquisition module; 12. Second sensor signal acquisition module; 13. Disturbance compensation acquisition module; 14. Drift compensation acquisition module; 15. Dynamic calibration execution module. DETAILED DESCRIPTION

[0074] The technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. The components of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for which protection is claimed, but merely represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of this application.

[0075] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and should not be understood as indicating or implying relative importance.

[0076] When traditional existing automated fluid loading systems perform long-term, high-precision micro-sampling analysis on bioreactors that cultivate engineered strains whose physiological characteristics show significant changes in response to external stimuli, if the calibration liquid introduced during the calibration process to ensure operational clarity and that is slightly different from the environment within the reactor causes temporary physiological or chemical state changes in the sample microenvironment near the sampling port that are difficult to directly detect through conventional sensors, thereby affecting the representativeness of subsequent samples or sensor response, there is a problem that the calibration process becomes the source of new volume measurement deviations, making it difficult to ensure that dynamic calibration can continuously provide a loading volume that accurately reflects the undisturbed sample state in the reactor under the requirements of long-term continuous operation and analysis with a narrow range of concentration changes of the detected object.

[0077] For example, consider a continuous-flow culture vessel used to produce a specific protein therapeutic. An engineered strain exhibits a significant physiological response to osmotic pressure or trace chemical concentrations. An automated fluid sampling system periodically extracts nanoliter to microliter samples from the reactor for analysis. The system employs a sensor-based dynamic calibration method, using sterile saline as the standard calibration fluid. During the calibration step, a very small volume of calibration fluid is expelled from the end of a sampling needle and then aspirated back to ensure the sampling port is unobstructed. Despite the small volume, the calibration fluid may have an osmotic pressure that differs by several parts per million from that of the culture fluid or may lack a trace nutrient present in the culture fluid. This subtle difference causes a transient localized shift in osmotic pressure or a decrease in nutrient concentration in the microenvironment near the sampling port. The first sample drawn immediately after the calibration procedure may experience a temporary physiological shift in the cells near the sampling port due to this localized shift in osmotic pressure, such as a slight shift in cell membrane permeability or metabolite release rate. Sensors used for dynamic calibration, such as those that detect liquid interfaces based on conductivity, may not directly identify these changes in the cell's physiological state because the conductivity difference between the calibration solution and the culture medium is small, or the sensor only considers the time it takes for the liquid column to pass through. The system performs dynamic calibration calculations based on this normal sensor signal and adjusts the pump drive parameters. This adjustment is intended to compensate for physical drift in the pump or tubing, but because the sensor signal is indirectly affected by the changes in the sample microenvironment introduced by the calibration, the dynamic calibration calculations may also unconsciously compensate for these slight deviations from the true state of the sample.

[0078] In this regard, this application proposes a sensor-based fluid loading dynamic calibration method, combined with Figure 1 Shown, including:

[0079] S1, after performing a calibration operation on the bioreactor, obtaining a first sensor signal output during a transitional fluid disturbance observation phase by controlling a sensor of a fluid loading unit;

[0080] S2, obtaining a second sensor signal outputted during the evaluation phase of the physical properties of the stable sample by controlling the sensor of the fluid loading unit;

[0081] S3, performing a feature comparison between the first sensor signal and the second sensor signal, and determining a disturbance compensation amount based on the comparison result;

[0082] S4, performing a feature comparison between the second sensor signal and a historical reference sensor signal, and determining a drift compensation amount based on the comparison result;

[0083] S5, according to the disturbance compensation amount and the drift compensation amount, differential compensation adjustment is performed on the driving parameters of the fluid loading unit or the sensor interpretation logic to complete the dynamic calibration of the fluid loading.

[0084] After the calibration operation is performed on the bioreactor, the first sensor signal outputted during the transitional fluid disturbance observation phase is obtained by controlling the sensor of the fluid loading unit. The transitional fluid disturbance observation phase refers to the period after the calibration operation is completed when the fluid state near the sampling port changes briefly due to the interaction between the calibration liquid and the fluid in the reactor. The purpose is to capture the impact of the instantaneous or short-term disturbance caused by the calibration operation itself on the sensor signal. The second sensor signal outputted during the stable sample physical property evaluation phase is obtained by controlling the sensor of the fluid loading unit. The stable sample physical property evaluation phase refers to the period after the influence of the transitional fluid disturbance has basically subsided, when the fluid state is obtained from the main area of ​​the reactor and represents the fluid state not directly affected by the calibration operation. The purpose is to obtain a relatively stable signal for evaluating the drift of the device itself and as a disturbance comparison benchmark. The first sensor signal is the signal collected during the transitional fluid disturbance observation phase, which reflects the comprehensive response of the device physical state to the fluid disturbance introduced by the calibration operation, and can be collected by a pressure sensor, a flow rate sensor, an optical sensor, a conductivity sensor, etc. The second sensor signal is the signal collected during the stable sample physical property evaluation phase, which mainly reflects the impact of the device physical state on the stable sample physical property. The response to a given fluid state can be collected by pressure sensors, flow rate sensors, optical sensors, conductivity sensors, etc. The historical reference sensor signal refers to the sensor signal collected and stored by the system in a previous known stable state, and its purpose is to serve as a reference for evaluating the physical drift of the equipment; feature comparison refers to analyzing the sensor signal, extracting its key features, and comparing the same or different features of different signals, such as comparing the rise time, peak value, integrated area, waveform shape, etc. of the signal waveform, with the purpose of quantifying the difference between the signals; the disturbance compensation amount is based on the difference between the first sensor signal and the second sensor signal. The numerical value or parameter determined by the characteristic comparison result represents the degree of influence of the fluid disturbance introduced by the calibration operation on the sensor signal or the sampling process; the drift compensation amount is the numerical value or parameter determined based on the characteristic comparison result between the second sensor signal and the historical reference sensor signal, which represents the degree of influence of the drift of the performance of the physical components of the equipment over time on the sensor signal or the sampling process; differentiated compensation adjustment refers to the targeted and varying degrees of correction of the driving parameters or sensor interpretation logic of the fluid sampling unit based on the determined disturbance compensation amount and drift compensation amount, with the aim of simultaneously offsetting the influence of calibration disturbance and equipment drift.

[0085] The solution of the present application first controls the sensor of the fluid loading unit to obtain a first sensor signal output during the transient fluid disturbance observation phase after performing a calibration operation on the bioreactor. This signal reflects the physical drift of the device itself and the transient disturbance of the fluid microenvironment near the sampling port caused by the calibration operation. Subsequently, after the disturbance has largely subsided, the sensor of the fluid loading unit is again controlled to obtain a second sensor signal output during the stable sample physical property evaluation phase. This signal primarily reflects the response of the device's physical drift to the stable fluid state. Next, a feature comparison is performed between the first and second sensor signals. By comparing the difference between these two signals, the impact of the disturbance introduced by the calibration operation on the sensor signal can be identified and quantified, thereby determining the disturbance compensation amount. Simultaneously, a feature comparison is performed between the second sensor signal and a historical reference sensor signal. By comparing the difference between the signal in the current stable state and the historical reference signal, the drift of the device's performance over time can be identified and quantified, thereby determining the drift compensation amount. Finally, based on the determined disturbance compensation amount and drift compensation amount, differential compensation adjustments are made to the drive parameters or sensor interpretation logic of the fluid loading unit. This differential adjustment specifically offsets disturbances caused by calibration and device drift, ensuring that subsequent sample loading accurately captures the target volume. This entire process forms a closed-loop dynamic calibration mechanism that continuously monitors and corrects various sources of deviation during the loading process.

[0086] In some preferred embodiments, the present application is implemented as follows: Assume that a fluid loading unit uses a micropump for fluid transport and is equipped with an optical sensor for detecting fluid flow and a pressure sensor for monitoring line pressure. After performing a calibration operation on the bioreactor, such as injecting a small amount of standard calibration fluid, the micropump is immediately activated to perform a short simulated loading process. Simultaneously, the output data of the optical sensor and the pressure sensor are collected. This data is preprocessed to form a first sensor signal. After a period of time to ensure that the influence of the calibration fluid has subsided, the micropump is activated again to perform a simulated loading process. The output data of the optical sensor and the pressure sensor are collected to form a second sensor signal. Features of the first and second sensor signals are extracted, such as the rise time and peak pressure of the pressure waveform, and the liquid surface transit time of the optical signal. These features of the first and second sensor signals are compared, the difference between them is calculated, and the disturbance compensation amount is determined based on the difference. For example, if the peak pressure of the first signal is lower than that of the second signal, it may indicate that the calibration fluid has reduced the local viscosity, and the driving force of the pump needs to be increased. Simultaneously, the system extracts features from the second sensor signal and compares them with corresponding features from a historical baseline sensor signal stored in the system. The difference is calculated, and the amount of drift compensation is determined based on this difference. For example, if the optical transit time of the second signal is longer than the historical baseline, this may indicate decreased pump efficiency and the need to increase the pump's run time. Finally, based on the calculated disturbance and drift compensation amounts, the micropump's drive voltage, pulse width, or run time are adjusted, or the sensor's threshold for interpreting the liquid level or calibration curve is modified to achieve dynamic calibration of the sample volume.

[0087] Optional, combined Figure 2 As shown, S5 performs differential compensation adjustment on the driving parameters or sensor reading logic of the fluid loading unit according to the disturbance compensation amount and the drift compensation amount to complete the dynamic calibration of the fluid loading, including the following steps:

[0088] S51, extracting the time characteristics and morphological characteristics of the first sensor signal; extracting the time characteristics and morphological characteristics of the second sensor signal;

[0089] S52, determining a reference morphological feature based on the morphological feature of the second sensing signal; comparing the morphological feature of the first sensing signal with the reference morphological feature to obtain a comparison result;

[0090] S53, correcting the time feature of the first sensor signal or modifying the proportion of the time feature of the first sensor signal in the calculation according to the comparison result;

[0091] S54, re-performing the process of determining the disturbance compensation amount and the process of determining the drift compensation amount based on the modified time characteristic of the first sensor signal or the proportion of the time characteristic in the calculation;

[0092] S55 , performing differential compensation adjustment on the driving parameters or sensor reading logic of the fluid loading unit according to the re-determined disturbance compensation amount and drift compensation amount, so as to complete the dynamic calibration of the fluid loading.

[0093] Among them, the time feature refers to the description of the regularity of the change of the sensor signal over time, which can be represented by parameters such as the time point of signal occurrence, duration, rise time, and fall time; the morphological feature refers to the shape and structural properties of the sensor signal waveform, which can be represented by parameters such as the peak value, width, slope, and waveform symmetry of the signal; the reference morphological feature refers to the morphological feature used as a comparison standard, which can be determined by statistically averaging, modeling, or selecting typical values ​​of the morphological features of the second sensor signal obtained during the physical property evaluation phase of multiple stable samples; correcting the time feature of the first sensor signal refers to adjusting the time axis or key time point data of the first sensor signal through an algorithm or model to correct the time deviation introduced by external disturbances; modifying the proportion of the time feature of the first sensor signal in the calculation refers to In the subsequent algorithm model for calculating the disturbance compensation amount and the drift compensation amount, the weight coefficient of the time characteristic parameter of the first sensor signal is adjusted to reduce or increase its influence on the calculation result; the process of determining the disturbance compensation amount refers to a series of algorithm steps for calculating the numerical value or adjustment amount that needs to be compensated for the instantaneous disturbance caused by the calibration operation based on the sensor signal characteristics; the process of determining the drift compensation amount refers to a series of algorithm steps for calculating the numerical value or adjustment amount that needs to be compensated for the performance drift caused by the long-term operation of the equipment based on the sensor signal characteristics; differentiated compensation adjustment refers to targeted and differentiated modifications to the driving parameters of the fluid loading unit (such as the pump driving voltage, number of pulses, and running time) or the sensor judgment logic (such as signal threshold, gain, and judgment algorithm) based on the calculated disturbance compensation amount and drift compensation amount.

[0094] In some preferred embodiments, the specific implementation is as follows: first, the time characteristics of the first sensor signal output during the transition fluid disturbance observation phase are extracted, such as the time point at which the signal peak occurs, and the morphological characteristics, such as the width and peak height of the signal waveform. At the same time, the time characteristics and morphological characteristics of the second sensor signal output during the stable sample physical property evaluation phase are extracted. Based on the morphological characteristics of multiple second sensor signals, for example, by calculating their average waveform width and peak height, a reference morphological characteristic is determined. The morphological characteristics (waveform width and peak height) of the currently acquired first sensor signal are compared with the reference morphological characteristics, for example, by calculating the Euclidean distance or correlation coefficient between them. If the comparison result shows that the morphological characteristics of the first sensor signal are significantly different from the reference morphological characteristics, then, based on the degree of difference, the peak time point of the first sensor signal is corrected, or when the disturbance compensation amount is subsequently calculated, the weight of the peak time point in the calculation model is reduced. Then, the disturbance compensation amount and drift compensation amount are recalculated using the corrected peak time point or the modified weight, as well as other signal characteristics. Finally, based on the recalculated disturbance compensation and drift compensation, for example, if the disturbance compensation indicates that the sample volume needs to be increased, the number of pump drive pulses is increased; if the drift compensation indicates that the sensor judgment threshold needs to be adjusted, the judgment threshold is modified to complete the dynamic calibration of fluid loading.

[0095] Optional, combined Figure 3 As shown, after S1 performs a calibration operation on the bioreactor, the step of obtaining a first sensor signal output during the transitional fluid disturbance observation phase by controlling the sensor of the fluid loading unit includes:

[0096] S11, after completing the calibration operation of the bioreactor, controlling the fluid loading unit to inject a standard disturbance liquid into the inlet of the bioreactor to form a transitional fluid disturbance observation stage;

[0097] S12, in the transitional fluid disturbance observation phase, continuously output data is collected through the pressure sensor and the flow rate sensor of the fluid sampling path, and the data is subjected to time series filtering and window smoothing processing to obtain a first sensing signal.

[0098] Among them, the standard perturbation liquid refers to a liquid with known physical and chemical properties, such as physiological saline, specific buffer solution or pure water, which is used to simulate or amplify the impact of the calibration operation on the fluid environment. Its purpose is to provide a controllable fluid sample that reflects the instantaneous state after calibration; the transitional fluid perturbation observation phase refers to a short time window after the completion of the calibration operation and before the sample fluid is completely stable. During this period, the impact of the calibration operation on the local fluid environment may not have completely dissipated. This phase is defined and extended by injecting standard perturbation liquid for observation. Its purpose is to isolate and capture the instantaneous impact of the calibration operation on the fluid state; the pressure sensor and flow rate sensor of the fluid loading path refer to devices used to monitor the pressure and flow rate changes of the fluid in real time when it flows in the pipeline. The pressure Sensors can detect fluid resistance or driving pressure, and flow rate sensors can detect the speed or flow rate of fluid passing through. Their purpose is to fully obtain dynamic physical information during the fluid sampling process; continuous output data refers to the data stream that the sensor continuously collects and outputs at a set sampling frequency over a period of time, rather than discrete or single-point data. Its purpose is to capture the complete process of fluid state changes over time; timing filtering and window smoothing refer to the application of digital signal processing technology to continuously collected time series data. Timing filtering is used to remove high-frequency noise or interference of specific frequencies, and window smoothing (such as moving average) is used to reduce data fluctuations and highlight data trends. Its purpose is to improve the signal-to-noise ratio and stability of the sensor signal so that it can more accurately reflect the actual fluid state changes.

[0099] In some preferred embodiments, specifically, after completing the bioreactor calibration operation, the precision pump of the fluid loading unit is controlled to inject a specific volume of physiological saline into the bioreactor inlet as a standard disturbance liquid, forming a transitional fluid disturbance observation phase, which lasts for a specific time. During this specific time, a piezoresistive pressure sensor and an ultrasonic flow sensor installed in the fluid loading passage continuously collect pressure and flow rate data at a set sampling frequency. The collected data stream is input into a processing unit, where it first undergoes Kalman filtering to remove high-frequency noise, and then undergoes moving average smoothing with a specific window size to ultimately obtain a first sensor signal.

[0100] Optional, combined Figure 4 As shown, the step S53 of correcting the time feature of the first sensor signal or modifying the proportion of the time feature of the first sensor signal in the calculation according to the comparison result includes:

[0101] S531, extracting interference pattern features from the morphological features of the first sensing signal;

[0102] S532, comparing the interference pattern characteristics with a preset reference pattern to determine whether the change in the morphological characteristics is due to the performed calibration operation;

[0103] S533: When it is determined that the change in the morphological feature is due to the calibration operation that has been performed, the time feature of the first sensing signal is corrected, or the proportion of the time feature of the first sensing signal in the calculation is modified.

[0104] Among them, the interference pattern feature refers to a pattern extracted from the morphological features of the first sensor signal, reflecting the signal change caused by the non-calibration operation. It can be implemented by signal decomposition, feature extraction algorithm or pattern recognition technology, with the purpose of decomposing the change of morphological features into different components, so as to more easily distinguish the change caused by the calibration operation from the change caused by other factors; the preset reference pattern refers to a pattern representing the typical change of morphological features caused by the calibration operation, which can be implemented by historical calibration operation data statistics, theoretical model construction or expert experience setting, with the purpose of providing a reference standard for judging whether the current morphological feature change is consistent with the expected result of the calibration operation; comparing the interference pattern feature with the preset reference pattern to determine whether the change of morphological features is attributed to the performed calibration operation refers to comparing the similarity or difference between the interference pattern feature and the reference pattern, and determining whether the current morphological feature change is consistent with the typical change caused by the calibration operation by setting a threshold or classifier. It can be implemented by distance calculation, correlation analysis or machine learning classification, with the purpose of judging whether the change of morphological features is indeed attributed to the calibration operation before correcting the time feature of the first sensor signal.

[0105] In some preferred embodiments, the interference pattern feature can be extracted from the morphological feature of the first sensor signal by a wavelet analysis method, the morphological feature curve of the first sensor signal is subjected to multi-scale decomposition, and the high-frequency components under a specific scale are extracted as interference pattern features, and these high-frequency components can reflect the instantaneous disturbance or noise in the signal. The preset reference pattern can be when the system is first installed or regularly maintained, multiple calibration operations are performed under standard conditions, the morphological features of the first sensor signal obtained after each calibration operation are recorded, and the corresponding wavelet high-frequency components are extracted as a historical interference pattern feature sample set, and then the average pattern or median pattern of the sample set is calculated as the initial reference pattern. The interference pattern feature is compared with the preset reference pattern, and when determining whether the change of the morphological feature is due to the calibration operation performed, the correlation coefficient between the currently extracted interference pattern feature and the reference pattern can be calculated. If the correlation coefficient is greater than a preset threshold value, it is determined that the change of the morphological feature is due to the calibration operation performed. When it is determined that the change in morphological characteristics is attributed to the calibration operation that has been performed, the amplitude of the time characteristics of the first sensor signal can be adjusted according to the degree of difference between the interference pattern characteristics and the reference pattern, such as the inverse of the correlation coefficient, or the weight coefficient of the time characteristics of the first sensor signal when calculating the disturbance compensation amount can be adjusted. For example, the higher the correlation coefficient, the greater the weight.

[0106] Optional, combined Figure 5 As shown, the step S532 of comparing the interference pattern characteristics with the preset reference pattern to determine whether the change in the morphological characteristics is due to the performed calibration operation includes:

[0107] S5321, comparing the interference pattern characteristics with a preset reference pattern to determine whether the change in the morphological characteristics is due to the performed calibration operation;

[0108] S5322: When the determination result is that the deviation is due to the calibration operation, correct the time feature of the first sensing signal, or modify the weight of the time feature of the first sensing signal in the calculation, and obtain interference pattern features corresponding to the calibration operation as the deviation profile;

[0109] S5323: Update the reference model according to the deviated appearance.

[0110] The interference pattern feature refers to a specific waveform or pattern extracted from the morphological characteristics of the first sensor signal that reflects the potential impact of the calibration operation. This feature can be extracted using waveform analysis, spectrum analysis, or machine learning models, with the goal of capturing the impact of the calibration operation on the sensor signal morphology. The preset baseline pattern refers to a reference pattern used for comparison with the interference pattern feature, representing the morphological characteristics that the system should have under normal calibration operation or the morphological characteristics under ideal conditions. This feature can be set based on statistical averaging of historical data, modeling, or expert experience, with the goal of providing a reference standard for determining the impact of the calibration operation. The deviation profile refers to the interference pattern feature corresponding to the calibration operation, obtained when the change in morphological characteristics is determined to be attributable to the calibration operation. This feature records the specific impact of the specific calibration operation on the sensor signal morphology, with the goal of providing data for subsequent baseline pattern updates. Updating the baseline pattern refers to the process of adjusting or correcting the preset baseline pattern based on the obtained deviation profile. This feature can be achieved using methods such as weighted averaging, exponential smoothing, or trend analysis based on historical sequences. This allows the baseline pattern to adapt to changes in the system over time and the cumulative effects of the calibration operation itself, thereby improving the accuracy of subsequent recognition.

[0111] In some preferred embodiments, comparing the interference pattern features with a preset reference pattern can be achieved by calculating the Euclidean distance between the two, setting a distance threshold, and determining that the change in morphological features is attributable to the calibration operation performed when the calculated distance is greater than the threshold. When the determination result is that the change is attributable to the calibration operation, modifying the time feature of the first sensor signal can be achieved by linearly stretching or compressing the time series of the first sensor signal so that it is aligned with the reference pattern on the time axis. Alternatively, modifying the proportion of the time feature of the first sensor signal in the calculation can be achieved by multiplying the time feature by a preset attenuation factor, such as 0.7, when subsequently calculating the disturbance compensation amount. Obtaining the interference pattern features corresponding to the calibration operation as the deviation appearance can be achieved by storing the currently extracted interference pattern feature data in a history record list. Updating the reference pattern based on the deviation appearance can be achieved by using a moving average method, averaging the latest deviation appearance with the most recent N deviation appearances, and then weighted averaging the average value with the old reference pattern to generate a new reference pattern.

[0112] Optionally, the step of updating the reference mode according to the deviation appearance in step S5323 includes:

[0113] The deviated appearance is weightedly combined with the baseline pattern before the update to generate the updated baseline pattern.

[0114] Among them, the deviation appearance refers to the interference pattern characteristics extracted from the morphological characteristics of the first sensor signal after the calibration operation is performed, which are different from the preset reference pattern, reflecting the specific change pattern that the calibration operation may introduce to the sensor signal morphology; the reference pattern refers to the preset pattern used for comparison with the interference pattern characteristics, which represents the sensor signal morphological characteristics of the system under an ideal or normal state, and can be established by means of historical data averaging, statistical models or expert experience; weighted combination refers to a mathematical operation that linearly superimposes two or more numerical values ​​or vectors according to a preset weight ratio, which can be implemented in the form of the formula W1*A+W2*B, where A and B are the quantities to be combined, W1 and W2 are the corresponding weights, and usually satisfy W1+W2=1. Its purpose is to give different influences according to their reliability or importance when fusing different information sources.

[0115] In some preferred embodiments, the deviation appearance is weightedly combined with the reference pattern before the update to generate an updated reference pattern, and a linear weighting method can be used specifically. For example, a weight factor alpha can be set, and its value range is between 0 and 1. The updated reference pattern can be calculated using the formula "reference pattern_new=alpha*deviation appearance+(1-alpha)*reference pattern_old". Among them, "reference pattern_old" is the reference pattern before the update, "deviation appearance" is the interference pattern feature obtained in this calibration operation, and "reference pattern_new" is the calculated updated reference pattern. The value of the weight factor alpha can be determined according to the characteristics of the system, the noise level of the data, or the desired update speed. For example, if you want the reference pattern to be more sensitive to new deviation appearances, you can increase the value of alpha; if you want the reference pattern to be more stable and insensitive to single data, you can reduce the value of alpha. In this way, the degree of influence of new information on the update of the reference pattern can be flexibly controlled.

[0116] Optional, combined Figure 6 As shown, the steps of weighted combination of the deviated appearance and the pre-updated reference pattern to generate the updated reference pattern include:

[0117] A1, obtain a historical deviated appearance sequence consisting of several deviated appearances;

[0118] A2, generates trend appearance based on historical deviation appearance sequence;

[0119] A3, determining the difference between the trend appearance and the baseline pattern before the update as the difference degree;

[0120] A4, adjusting the weights used for weighted combination according to the degree of difference;

[0121] A5: Perform a weighted combination of the deviated appearance and the pre-updated reference pattern according to the adjusted weights for weighted combination to generate an updated reference pattern.

[0122] The historical deviation appearance sequence refers to a chronological collection of deviation appearances generated by several calibration operations performed at different time points. This can be implemented as a data list stored in memory or a database, with the purpose of providing historical data for analyzing the long-term impact trends of calibration operations. The trend appearance refers to the long-term, stable change pattern reflecting the impact of calibration operations on the sample microenvironment, obtained by analyzing or modeling the historical deviation appearance sequence. This can be implemented using statistical methods such as moving average, exponential smoothing, or regression analysis, with the purpose of extracting representative change trends from discrete deviation appearances. The difference refers to the degree of deviation in features between the trend appearance and the baseline pattern before the update. This can be implemented by calculating the distance, similarity, or the difference between the two specific statistical indicators, with the purpose of quantifying the gap between the cumulative impact of the calibration operation and the current baseline pattern. Adjusting the weights used for weighted combination refers to dynamically changing the proportion of the deviation appearance and the baseline pattern before the update in the weighted combination according to the size of the difference. This can be implemented using a preset weight adjustment function or lookup table, with the purpose of enabling the updated baseline pattern to more reasonably integrate the latest deviation information with historical trend information.

[0123] In some preferred embodiments, a historical sequence of deviation appearances can be stored in a circular buffer, which records the deviation appearances generated by the most recent N calibration operations. When generating a trend appearance, a moving average calculation can be performed on the historical sequence of deviation appearances in the circular buffer to obtain a smooth trend appearance. When determining the degree of difference, the Euclidean distance between the trend appearance and the pre-update baseline pattern along key feature dimensions can be calculated. Using this Euclidean distance as the degree of difference, a preset nonlinear function can be used to adjust the weights of the deviation appearance and the baseline pattern in the weighted combination. For example, when the degree of difference is small, the deviation appearance can be weighted relatively high to quickly respond to the latest changes. When the degree of difference is large, the deviation appearance can be weighted relatively low, while the baseline pattern can be weighted relatively high to prevent abnormal deviation appearances from overly influencing the baseline pattern while ensuring that the update direction is consistent with the long-term trend. Finally, based on the adjusted weights, the deviation appearance and the pre-update baseline pattern are combined using a linear weighting formula to obtain the updated baseline pattern.

[0124] Optional, combined Figure 7 As shown, A2 generates trend appearance based on the historical deviation appearance sequence, including the following steps:

[0125] A21, receiving history deviation appearance sequence;

[0126] A22, smoothing the historical deviation appearance sequence;

[0127] A23 generates trend appearance based on the smoothed historical deviation appearance sequence.

[0128] Among them, the historical deviation appearance sequence refers to a data set of deviation appearances obtained in multiple calibration operations arranged in chronological order, which can be stored in system memory, database or file; smoothing processing refers to the application of an algorithm to time series data to reduce noise and short-term fluctuations and reveal potential trends or periodicity. It can be achieved by moving average, exponential smoothing, Savitzky-Golay filtering and other methods. Its purpose is to filter out random errors and highlight long-term change trends; trend appearance refers to the representative pattern or vector extracted by analyzing the historical deviation appearance sequence, which reflects the long-term evolution direction and amplitude of the deviation pattern. Its purpose is to provide a reference that is more stable and more representative of the long-term behavior changes of the system than a single deviation appearance.

[0129] In some preferred embodiments, the present application is specifically implemented as follows: after each calibration operation, the system stores the acquired deviation appearance in a circular buffer or database to form a time series containing several recent deviation appearances, that is, a historical deviation appearance sequence. The moving average method can be used to smooth the historical deviation appearance sequence. For example, for each data point in the historical deviation appearance sequence, the average value of the data point and several previous and subsequent data points is calculated as the smoothed value of the point. Alternatively, exponential smoothing can be used to give recent data a higher weight. Based on the smoothed historical deviation appearance sequence, a trend appearance is generated. The last point of the smoothed sequence can be used as the current trend appearance, or a linear regression can be performed on the smoothed sequence to combine the slope and intercept of the regression line into a characteristic vector of the trend appearance.

[0130] Optional, combined Figure 8 As shown, after executing step A1 to obtain a historical deviated appearance sequence consisting of a plurality of deviated appearances, the following steps are further included:

[0131] A11, for each deviant appearance in the historical deviant appearance sequence, compare it with the reference pattern;

[0132] A12, based on the comparison results, determines whether the deviated appearance is an abnormal data point;

[0133] A13, excludes abnormal data points in the historical deviation appearance sequence.

[0134] Among them, comparing each deviant appearance in the historical deviant appearance sequence with the baseline pattern means performing an operation of comparing each data point in the historical deviant appearance sequence, that is, each deviant appearance, with the currently used baseline pattern. This can be achieved by calculating the difference measure between the deviant appearance and the baseline pattern, such as calculating the Euclidean distance, correlation coefficient or structural similarity index between them, the purpose of which is to quantify the degree of deviation between each deviant appearance and the baseline pattern representing the normal or expected pattern; determining whether the deviant appearance is an abnormal data point means applying a preset rule or statistical method to determine whether the deviant appearance deviates significantly from the baseline pattern based on the difference measure obtained by the comparison, and thus is marked as abnormal. This can be achieved by setting a fixed threshold. When the difference metric exceeds the threshold, it is judged as an anomaly, or statistical analysis based on historical data is used, such as calculating the mean and standard deviation of the historical difference metric, and judging the points beyond the specific standard deviation range as anomalies. Its purpose is to identify outlier data that do not conform to the normal pattern and may be caused by accidental factors or errors; excluding abnormal data points in the historical deviation appearance sequence means removing the deviation appearance determined as an abnormal data point from the historical deviation appearance sequence or marking it as invalid, so that it does not participate in the subsequent trend appearance generation process. It can be achieved by creating a new sequence that does not contain abnormal data points, or skipping the data points marked as abnormal during the processing process. Its purpose is to purify historical data and improve the data quality used to generate trend appearances.

[0135] In some preferred embodiments, specifically, after obtaining a historical deviation appearance sequence consisting of several deviation appearances, the system can traverse each deviation appearance in the sequence. For the currently processed deviation appearance, the Euclidean distance between it and the currently stored reference pattern is calculated, and the Euclidean distance can be used as a comparison result to measure the degree of deviation. Then, according to the preset abnormality judgment rules, for example, if the calculated Euclidean distance exceeds a fixed multiple of the historical Euclidean distance average value (such as three times the standard deviation), the deviation appearance is judged to be an abnormal data point. Finally, when constructing a sequence for generating trend appearances, the system can skip all deviation appearances that are judged to be abnormal data points and only use non-abnormal deviation appearances to perform subsequent smoothing and trend appearance generation steps.

[0136] Through the above technical solution, abnormal data points in the historical deviation pattern sequence can be effectively identified and eliminated, thereby improving the data quality used to generate trend patterns, so that the generated trend patterns can more accurately reflect the actual drift trend of the system, and then optimize the update process of the benchmark model, enhance the accuracy and stability of dynamic calibration, and reduce the adverse effects of occasional abnormal data on long-term calibration effects.

[0137] A sensor-based fluid loading dynamic calibration system for performing sensor-based fluid loading dynamic calibration, combined with Figure 9 As shown, the sensor-based fluid loading dynamic calibration system 1 includes:

[0138] A first sensor signal acquisition module 11 is configured to acquire a first sensor signal output during a transitional fluid disturbance observation phase by controlling a sensor of a fluid loading unit after performing a calibration operation on the bioreactor;

[0139] A second sensor signal acquisition module 12 is used to acquire a second sensor signal output during the stable sample physical property evaluation phase by controlling a sensor of the fluid loading unit;

[0140] The disturbance compensation amount acquisition module 13 is used to compare the characteristics of the first sensor signal and the second sensor signal, and determine the disturbance compensation amount according to the comparison result;

[0141] a drift compensation amount acquisition module 14 for performing a feature comparison between the second sensor signal and a historical reference sensor signal, and determining a drift compensation amount based on the comparison result;

[0142] The dynamic calibration execution module 15 is used to perform differential compensation adjustment on the driving parameters or sensor interpretation logic of the fluid loading unit according to the disturbance compensation amount and the drift compensation amount, so as to complete the dynamic calibration of the fluid loading.

[0143] The first sensor signal acquisition module refers to a processing unit used to collect sensor data at a specific time point or stage, and can be implemented using an integrated circuit, microprocessor, or dedicated signal processing hardware. The second sensor signal acquisition module refers to a processing unit used to collect sensor data at another specific time point or stage, and can be implemented using hardware or software structures similar to the first sensor signal acquisition module. The disturbance compensation acquisition module refers to a processing unit used to analyze and compare the characteristics of two sets of sensor signals and calculate a value reflecting the deviation introduced by the calibration operation. It can be implemented using a software program running a specific algorithm or a digital signal processor. The drift compensation acquisition module refers to a processing unit used to analyze and compare the current sensor signal with historical data and calculate a value reflecting the performance degradation of the device. It can be implemented using a combination of a data storage unit and a computing unit. The dynamic calibration execution module refers to a control unit used to adjust the operating parameters or data processing method of the fluid loading unit based on the calculated compensation value. It can be implemented using a microcontroller or programmable logic device that outputs control signals. The fluid loading unit refers to the actuator responsible for actually extracting or delivering the fluid, and can be implemented using a precision syringe pump, peristaltic pump, or microfluidic pump. A sensor is a device used to sense the physical or chemical properties of a fluid and output a corresponding signal. It can be implemented as a pressure sensor, flow sensor, conductivity sensor, or optical sensor. The transient fluid disturbance observation phase refers to the brief period after the calibration operation is completed when the calibration liquid and the fluid in the reactor mix or interact. The stable sample physical property evaluation phase refers to the period after the transient phase ends and the fluid near the sampling port returns to a relatively stable state. Historical baseline sensor signals refer to sensor signal data collected and stored by the system during previous normal operation or standard conditions, which serves as a reference for evaluating device drift. Drive parameters refer to the input signals or configuration values ​​that control the actions of the fluid loading unit. These can be implemented as pump drive voltage, pulse width, step frequency, or run time. Sensor interpretation logic refers to the algorithm or rules that the system uses to interpret the raw sensor signal to determine the fluid state or volume. This can be implemented using threshold judgment, signal integration, or pattern recognition algorithms. Differentiated compensation adjustment refers to the application of differentiated and refined corrections to the fluid loading unit based on two different compensation amounts: disturbance compensation and drift compensation.

[0144] The solution of this application achieves dynamic calibration of the fluid loading process through a modular system design, solving the problem of new volume measurement deviations that may be introduced during the calibration process. Specifically, after the system performs a calibration operation on the bioreactor, the first sensor signal acquisition module first collects the sensor signal during the transitional fluid disturbance observation phase to capture the transient impact that the calibration operation may have on the local fluid state. Subsequently, the second sensor signal acquisition module collects the sensor signal during the stable sample physical property evaluation phase to reflect the physical properties of the sample in a relatively stable state. The disturbance compensation acquisition module quantifies the impact of the disturbance introduced by the calibration operation on the sensor response by comparing the sensor signal characteristics of these two phases and calculates the value that needs to be compensated. At the same time, the drift compensation acquisition module compares the sensor signal during the stable phase with the pre-stored historical baseline sensor signal to evaluate the performance drift caused by the long-term operation of the equipment and calculate the corresponding drift compensation amount. Finally, the dynamic calibration execution module comprehensively utilizes these two independent compensation amounts to make differentiated adjustments to the drive parameters or sensor interpretation logic of the fluid loading unit. This dual compensation mechanism enables the system to simultaneously correct for both the device's own physical drift and potential perturbations to the sample microenvironment caused by the calibration procedure, ensuring that the final loading volume accurately reflects the fluid state within the reactor unaffected by the calibration. In this way, the system not only compensates for errors introduced by device aging but also prevents the calibration process itself from becoming a new source of error, thereby improving the accuracy and reliability of fluid loading volumes during long-term operation.

[0145] In some preferred embodiments, the present application is implemented as follows. The fluid loading unit can be a precision syringe pump driven by a stepper motor, whose driving parameters include the stepper motor's pulse count and frequency. The sensors can be miniature pressure sensors and flow rate sensors installed in the syringe pump pipeline. The first sensor signal acquisition module and the second sensor signal acquisition module can be integrated into a microcontroller. They control analog-to-digital converters (ADCs) to collect the output voltage signals of the pressure and flow rate sensors during different time windows after calibration and convert the analog signals into digital signals. The disturbance compensation acquisition module and the drift compensation acquisition module can be implemented as software programs running on the microcontroller. The disturbance compensation acquisition module can analyze the waveform characteristics (e.g., peak value, falling edge slope, and stable value) of the pressure and flow rate signals during the transition phase (e.g., 0-5 seconds after calibration) and the stable phase (e.g., 10-15 seconds after calibration) and calculate the difference between the two as the disturbance compensation value. The drift compensation acquisition module can compare the signal stability value during the stable phase with the historical average stability value stored in non-volatile memory to calculate the drift compensation value that reflects the change in pump or pipeline efficiency. The dynamic calibration execution module can also be a software program on the microcontroller. Based on the calculated disturbance compensation and drift compensation, it adjusts the number or frequency of pulses sent to the stepper motor driver or modifies the pressure / flow rate thresholds used to determine the liquid interface. For example, if a temporary pressure increase (disturbance) caused by calibration and a flow rate decrease (drift) due to device aging are calculated, the system can simultaneously fine-tune the number of pulses to compensate for the drift and temporarily adjust the sensor judgment threshold to adapt to the signal characteristics of the disturbance phase. Alternatively, it can perform a weighted correction on the data from the disturbance phase when calculating the final sample loading volume.

[0146] Through the above technical solution, the present application provides a sensor-based fluid loading dynamic calibration system, which can effectively identify and compensate for the disturbance of the sample microenvironment introduced by the calibration operation itself and the sample volume drift caused by the aging of the physical components of the equipment. This enables the system to continuously provide a sample volume that accurately reflects the state of the undisturbed sample in the reactor during long-term continuous operation, avoiding the calibration process from becoming a new source of volume measurement deviation. Especially for the cultivation of biological systems that are sensitive to environmental changes, this system can ensure that the collected samples are more representative, thereby improving the reliability of subsequent analysis results and supporting more accurate biological process monitoring and control.

[0147] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A sensor-based fluid loading dynamic calibration method, characterized in that: include: After performing a calibration operation on the bioreactor, a first sensor signal output during a transitional fluid disturbance observation phase is obtained by controlling a sensor of the fluid loading unit; By controlling the sensor of the fluid loading unit, a second sensor signal outputted during the evaluation phase of the stable sample physical properties is obtained; performing a feature comparison on the first sensing signal and the second sensing signal, and determining a disturbance compensation amount according to the comparison result; performing a feature comparison between the second sensor signal and a historical reference sensor signal, and determining a drift compensation amount based on the comparison result; According to the disturbance compensation amount and the drift compensation amount, differential compensation adjustment is performed on the driving parameters or sensor reading logic of the fluid loading unit to complete the dynamic calibration of the fluid loading.

2. A sensor-based fluid loading dynamic calibration method according to claim 1, characterized in that: The step of performing differential compensation adjustment on the driving parameters or sensor reading logic of the fluid loading unit according to the disturbance compensation amount and the drift compensation amount to complete the dynamic calibration of the fluid loading includes: extracting the time characteristics and morphological characteristics of the first sensor signal; extracting the time characteristics and morphological characteristics of the second sensor signal; determining a reference morphological feature based on the morphological feature of the second sensing signal; and comparing the morphological feature of the first sensing signal with the reference morphological feature to obtain a comparison result; According to the comparison result, correcting the time feature of the first sensor signal, or modifying the proportion of the time feature of the first sensor signal in the calculation; re-performing the process of determining the disturbance compensation amount and the process of determining the drift compensation amount according to the modified time characteristic of the first sensor signal or the proportion of the time characteristic in the calculation; According to the re-determined disturbance compensation amount and drift compensation amount, differential compensation adjustment is performed on the driving parameters or sensor interpretation logic of the fluid loading unit to complete the dynamic calibration of the fluid loading.

3. The method for dynamic calibration of fluid loading based on a sensor according to claim 1, characterized in that: After the bioreactor is calibrated, the step of obtaining a first sensor signal output during the transitional fluid disturbance observation phase by controlling a sensor of the fluid loading unit includes: After completing the calibration operation of the bioreactor, controlling the fluid loading unit to inject a standard disturbance liquid into the inlet of the bioreactor to form a transitional fluid disturbance observation stage; During the transitional fluid disturbance observation phase, continuous output data is collected through the pressure sensor and flow rate sensor of the fluid sampling path, and the data is subjected to time series filtering and window smoothing to obtain the first sensing signal.

4. The method for dynamic calibration of fluid loading based on a sensor according to claim 2, characterized in that: The step of correcting the time feature of the first sensor signal or modifying the proportion of the time feature of the first sensor signal in the calculation according to the comparison result includes: extracting interference pattern features from the morphological features of the first sensing signal; comparing the interference pattern characteristics with a preset reference pattern to determine whether the change in the morphological characteristics is due to the calibration operation performed; When it is determined that the change in the morphological feature is due to the calibration operation that has been performed, the time feature of the first sensing signal is corrected, or the proportion of the time feature of the first sensing signal in the calculation is modified.

5. A sensor-based fluid loading dynamic calibration method according to claim 4, characterized in that: The step of comparing the interference pattern characteristics with a preset reference pattern to determine whether the change in the morphological characteristics is due to the performed calibration operation includes: comparing the interference pattern characteristics with a preset reference pattern to determine whether the change in the morphological characteristics is due to the calibration operation performed; When the determination result is that the deviation is due to the calibration operation, correcting the time feature of the first sensor signal, or modifying the proportion of the time feature of the first sensor signal in the calculation, and obtaining the interference pattern feature corresponding to the calibration operation as the deviation appearance; The reference model is updated according to the deviated appearance.

6. A sensor-based fluid loading dynamic calibration method according to claim 5, characterized in that: The step of updating the reference mode according to the deviation appearance includes: The deviated appearance is weightedly combined with the reference pattern before updating to generate the updated reference pattern.

7. A sensor-based fluid loading dynamic calibration method according to claim 6, characterized in that: The step of performing weighted combination of the deviated appearance and the pre-updated reference pattern to generate the updated reference pattern comprises: Acquire a historical deviated appearance sequence consisting of a plurality of the deviated appearances; generating a trend profile based on the historical deviation profile sequence; determining a difference between the trend profile and a pre-update baseline pattern as a degree of difference; adjusting the weights used for weighted combination according to the degree of difference; The deviated appearance is weightedly combined with the reference pattern before updating according to the adjusted weight for weighted combination to generate the updated reference pattern.

8. A sensor-based fluid loading dynamic calibration method according to claim 7, characterized in that: The step of generating a trend appearance based on the historical deviation appearance sequence includes: receiving the historical deviation appearance sequence; performing smoothing processing on the historical deviation appearance sequence; A trend appearance is generated based on the smoothed historical deviation appearance sequence.

9. The sensor-based fluid loading dynamic calibration method according to claim 7, characterized in that: After executing the step of obtaining a historical deviated appearance sequence consisting of a plurality of the deviated appearances, the method further includes: For each deviated appearance in the historical deviated appearance sequence, comparing it with the reference pattern; According to the comparison result, determining whether the deviated appearance is an abnormal data point; The abnormal data points in the historical deviation appearance sequence are excluded.

10. A sensor-based fluid loading dynamic calibration system for performing sensor-based fluid loading dynamic calibration, characterized in that: include: A first sensor signal acquisition module is configured to acquire a first sensor signal output during a transitional fluid disturbance observation phase by controlling a sensor of a fluid loading unit after performing a calibration operation on the bioreactor; A second sensor signal acquisition module is used to acquire a second sensor signal output during the evaluation phase of the physical properties of the stable sample by controlling the sensor of the fluid loading unit; a disturbance compensation amount acquisition module, configured to compare features of the first sensing signal and the second sensing signal, and determine a disturbance compensation amount based on the comparison result; a drift compensation amount acquisition module, configured to compare the characteristics of the second sensor signal with a historical reference sensor signal, and determine a drift compensation amount based on the comparison result; The dynamic calibration execution module is used to perform differential compensation adjustment on the driving parameters or sensor reading logic of the fluid loading unit according to the disturbance compensation amount and the drift compensation amount, so as to complete the dynamic calibration of the fluid loading.

Images