A quality control method based on mass spectrometry data

Abstract

This application discloses a quality control method based on mass spectrometry data, relating to the field of mass spectrometry analysis technology, including: acquiring multimodal data and extracting peak shift and quality control index through a visual analysis model; generating a device state vector based on the peak shift and constructing a device topology model using the multimodal data to predict drift; generating an interference heatmap based on the multimodal data, constructing a dynamic interference tracing model in conjunction with mass spectrometry quality planning, and generating cluster collaborative instructions to optimize the quality control index; constructing a risk grading model based on the device state vector and cluster collaborative instructions to classify risk levels and generate a pre-tuning instruction set; the risk grading model performs quality control steady-state assessment through the volatility of the quality control index and generates a stability rating; monitoring the cluster to generate a stability coefficient through the pre-tuning instruction set; and when the stability coefficient exceeds the entropy change stability threshold, invoking voltage and temperature compensation to recalibrate the device to ensure the accuracy of the quality control index.

Description

Technical Field

[0001] This invention relates to the field of mass spectrometry analysis technology, and in particular to a quality control method based on mass spectrometry data. Background Technology

[0002] Mass spectrometry, with its high sensitivity, high specificity, and powerful compound resolution capabilities, has become an indispensable core analytical tool in modern life science research, clinical diagnosis, drug development, environmental monitoring, and food safety. Its basic principle is to ionize sample molecules, separate ions according to their mass-to-charge ratio, and detect their abundance, thereby obtaining a mass spectrum reflecting the chemical composition of the sample.

[0003] With the widespread application of high-throughput mass spectrometry platforms, massive and complex mass spectrometry data have been generated. However, mass spectrometry analysis is a complex process involving multiple steps, including sample pretreatment, chromatographic separation, ionization, mass analysis, ion detection, and data acquisition. Each step can be affected by various factors, leading to fluctuations or even deviations in data quality.

[0004] Existing mass spectrometry quality control technologies suffer from systemic defects: at the single-device level, they rely on passive correction mechanisms and cannot dynamically suppress signal drift caused by environmental disturbances; in multi-device cluster scenarios, the lack of a device topology correlation model leads to cross-device systemic drift cascading failures caused by temperature and humidity fluctuations, while interferences in dynamically mixed matrices generate non-classical signal amplification due to quantum tunneling effects; these problems make existing methods face the risk of drift runaway, quantum misjudgment, and entropy increase collapse in cluster environments. Summary of the Invention

[0005] This application provides a quality control method based on mass spectrometry data, which solves the problems in the prior art such as signal drift and loss of control of cluster equipment caused by environmental disturbances, interference misjudgment caused by quantum tunneling effect, and entropy increase instability in long-term monitoring. It achieves the technical effects of cross-device drift dynamic suppression, precise decoupling of quantum-level interference, and entropy balance of the quality control system.

[0006] This application provides a quality control method based on mass spectrometry data, including:

[0007] S1: Acquire multimodal data, extract peak shift and quality control index through visual analysis model; generate equipment state vector based on peak shift, construct equipment topology model using multimodal data to predict drift; when drift exceeds drift critical threshold, implement voltage and temperature compensation to generate mass spectrometry quality planning;

[0008] S2: Generate interference heatmaps based on multimodal data, construct dynamic interference source tracing models by combining mass spectrometry quality planning, identify interference source devices, and generate cluster collaborative instructions to optimize quality control indices;

[0009] S3: Based on device state vectors and cluster collaborative instructions, a risk grading model is constructed. Risk levels are divided according to predicted drift and interference heatmaps, and a pre-adjustment instruction set is generated. The risk grading model performs quality control steady-state assessment through the volatility of the quality control index and generates a stability rating.

[0010] S4: The cluster generates a stability coefficient by monitoring the pre-tuned instruction set; when the stability coefficient exceeds the entropy change stability threshold, voltage and temperature compensation is invoked to recalibrate the equipment, and interference suppression is implemented through cluster collaborative instructions to ensure the accuracy of the quality control index.

[0011] Furthermore, the multimodal data includes: equipment operating parameters consisting of voltage fluctuation sequences, temperature gradient distributions, and mass spectrometer operating frequencies; mass spectrometry analysis data consisting of raw mass spectra and peak shift matrices; environmental monitoring parameters consisting of electromagnetic interference intensity, temperature and humidity change curves, and vibration spectrum diagrams; system topology information consisting of equipment connection topology, signal transmission paths, and cluster communication delays; and a historical reference library consisting of reference drift curves, interference event databases, and quality control steady-state thresholds.

[0012] Furthermore, the peak shift is the deviation between the actual mass-to-charge ratio of the characteristic peak of the target analyte in the mass spectrum and the theoretical value of the standard mass-to-charge ratio, which is obtained by identifying and calculating the original mass spectrum through a visual analysis model;

[0013] The quality control index is a standardized index value generated by weighting the peak shift, signal-to-noise ratio, and peak shape symmetry in the original mass spectrum, and is used to evaluate the reliability of mass spectrometry data.

[0014] Furthermore, the predicted drift includes: establishing a device node feature matrix based on the device state vector, generating an adjacency graph by combining the device connection topology, learning the influence propagation law between devices through a graph neural network, and outputting a mass-to-charge ratio offset prediction curve within the future time window;

[0015] The drift threshold is the core critical value for voltage-temperature compensation, which is calculated using the benchmark drift curve and predicted drift data from the historical reference library.

[0016] Furthermore, the mass spectrometry quality planning includes: calculating compensation parameters in voltage and temperature compensation based on the predicted drift amount and the device state vector; the compensation parameters include voltage adjustment values ​​and temperature setpoints, which are used in the voltage control circuit and temperature regulation system of the mass spectrometer to form a quality control operation planning scheme.

[0017] Furthermore, the interference heatmap includes: an interference intensity distribution map generated on the physical spatial coordinates of the device defined by the system topology information based on the electromagnetic interference intensity and vibration spectrum data in the environmental monitoring parameters; feature analysis of the interference intensity distribution map is performed through a dynamic interference source tracing model, and the interference propagation path is analyzed according to the device connection topology to output the spatial positioning result of the interference source device.

[0018] Furthermore, the cluster coordination command is generated by locating the interference source device through a dynamic interference tracing model, and then generating a set of coordination operation commands based on the device connection topology and cluster communication delay parameters, combined with the spatial location and intensity characteristics of the interference source identified by the interference heatmap. These commands include: dynamically adjusting the transmission power of the interference source device to reduce the intensity of electromagnetic interference, reconstructing the signal transmission path within the cluster to avoid interference propagation links, optimizing the allocation of communication time slot parameters to avoid signal conflicts, and obtaining a cluster coordination scheme for the interference source.

[0019] Furthermore, the risk levels include: classifying drift risk levels based on the magnitude and direction of the predicted drift, with drift risk levels including low risk, medium risk, and high risk; and classifying interference risk levels based on the interference intensity distribution of the interference heatmap, with interference risk levels including slight interference, moderate interference, and severe interference.

[0020] The pre-adjustment instruction set includes: generating voltage compensation instructions based on risk levels classified by the risk grading model, generating temperature adjustment instructions based on the predicted drift amount, and generating interference source suppression instructions by combining spatial positioning of the interference heat map, thus constituting a set of strategies for equipment recalibration and interference suppression.

[0021] Furthermore, the stability rating is a stability level determined by the quality control index volatility calculation result and the quality control steady-state assessment model. The stability level includes three levels: stable state, metastable state, and unstable state. The stable state corresponds to the quality control steady-state range with volatility ≤ 5%, the metastable state corresponds to the quality control buffer range with volatility ≤ 15% and volatility ≤ 5%, and the unstable state corresponds to the quality control out-of-control range with volatility > 15%.

[0022] Furthermore, the stability coefficient is obtained by real-time monitoring of the cluster's operating status through a pre-tuning instruction set, calculating the stability rating and the volatility of the quality control index, and is used to quantify the steady-state degree of the quality control system.

[0023] The entropy change stability threshold is a dynamically set recalibration trigger threshold, calculated based on the quality control steady-state threshold and stability rating data in the historical reference library.

[0024] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0025] By constructing a device topology model for multimodal data fusion to predict drift and implementing voltage and temperature collaborative compensation, the systemic signal drift problem caused by environmental disturbances in multi-device cluster environments is effectively solved, achieving collaborative suppression of cross-device drift and eliminating cascading signal distortion caused by single-device calibration. Through an interference source tracing model, the non-classical signal characteristics generated by quantum tunneling effect in mixed matrices are accurately analyzed, significantly reducing the risk of interference misjudgment. Through a risk-level driven pre-tuning instruction set and stability control mechanism, the entropy growth rate is effectively suppressed while efficiently suppressing interference sources, providing closed-loop quality control assurance for multimodal mass spectrometry clusters from data acquisition to interference eradication. Attached Figure Description

[0026] Figure 1 This is a flowchart of a quality control method based on mass spectrometry data in an embodiment of the present invention. Detailed Implementation

[0027] To facilitate understanding of the present invention, a more complete description of this application will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of the present invention.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0029] Example 1: As Figure 1 As shown, a quality control method based on mass spectrometry data is presented.

[0030] S1: Acquire multimodal data, extract peak shift and quality control index through visual analysis model; generate equipment state vector based on peak shift, construct equipment topology model using multimodal data to predict drift; when drift exceeds drift critical threshold, implement voltage and temperature compensation to generate mass spectrometry quality planning;

[0031] The multimodal data includes: equipment operating parameters consisting of voltage fluctuation sequences, temperature gradient distributions, and mass spectrometer operating frequencies; mass spectrometry analysis data consisting of raw mass spectra and peak offset matrices; environmental monitoring parameters consisting of electromagnetic interference intensity, temperature and humidity change curves, and vibration spectrum diagrams; system topology information consisting of equipment connection topology, signal transmission paths, and cluster communication delays; and a historical reference library consisting of reference drift curves, interference event databases, and quality control steady-state thresholds.

[0032] Specifically, the equipment operating parameters include voltage fluctuation sequences recorded at millisecond-level time resolution from the mass spectrometer power monitoring module, three-dimensional temperature gradient distribution generated by temperature sensors installed on key components, and the mass spectrometer operating frequency read from the mass spectrometer control system, used for real-time monitoring of the equipment's physical state; mass spectrometry analysis data includes raw mass spectra from the mass spectrometer detector and its N×M-dimensional peak shift matrix derived from a visual analysis model; environmental monitoring parameters are electromagnetic interference intensity in the 0.1MHz-10GHz band collected by broadband electromagnetic field probes deployed around the equipment, and second-level data from temperature and humidity sensors. The sampled temperature and humidity change curves and the 1-2000Hz vibration spectrum monitored by the piezoelectric accelerometer array are used to locate the source of environmental disturbance; the system topology information defines the device connection topology adjacency matrix extracted from the device network configuration, the physical structure of the signal transmission path measured by the physical wiring diagram and network analyzer, and the microsecond-level precision cluster communication delay obtained by the timestamp record of the switch; the historical reference library stores the benchmark drift curve of the historical operation database, the interference event library of typical events such as quantum tunneling interference recorded by the device log system, and the quality control steady-state threshold set dynamically maintained according to the long-term statistical model.

[0033] The peak shift is the deviation between the actual mass-to-charge ratio of the characteristic peak of the target analyte in the mass spectrum and the theoretical value of the standard mass-to-charge ratio, which is obtained by identifying and calculating the original mass spectrum through a visual analysis model.

[0034] Specifically, the visual analysis model identifies key feature peaks in the mass spectrum using computer vision algorithms and quantifies their mass parameters. The model preprocesses the original mass spectrum (noise reduction, baseline correction), locates the peak apex positions using gradient detection algorithms, and calculates the peak offset of the actual peak position.

[0035] ,

[0036] in, This represents the peak shift. is the mass of the ion, z is the charge number of the ion, and m / z is the mass-to-charge ratio. This is the actual mass-to-charge ratio. The theoretical mass-to-charge ratio is given; based on the results, the peak offset matrix (N×M) is output, where N (rows) is the number of samples and M (columns) is the number of characteristic peaks.

[0037] The quality control index is a standardized index value generated by weighting the peak shift, signal-to-noise ratio, and peak shape symmetry in the original mass spectrum, and is used to evaluate the reliability of mass spectrometry data.

[0038] Specifically, the quality control index is calculated using peak shift, signal-to-noise ratio, and peak shape symmetry from the original mass spectrum.

[0039] ,

[0040] Where ZK is the quality control index, The offset scoring function is a function that converts peak offset into a standardized score. , Let be the attenuation coefficient, and assume the instrument's nominal accuracy is . (Unit: ppm) When the offset reaches the instrument's accuracy limit, the score should be reduced to the passing grade (0.6). , Calculate based on the nominal accuracy of different equipment and instruments Values, such as the nominal precision of a quadrupole mass spectrometer being 100 ppm, can be calculated as follows: ; The signal-to-noise ratio (SNR) scoring function is a function that converts the SNR into a standardized score. , Let be the growth factor, and let the minimum signal-to-noise ratio required for quality control be . If the scoring requirement is to reach the passing grade (0.6), then... Converted to =0.4, Calculate the minimum signal-to-noise ratio requirement based on different scenarios. Values, such as the minimum signal-to-noise ratio requirement of 3 for screening-level analysis, can be calculated as follows: =0.916 / 3=0.305; For symmetry scoring functions, the peak symmetry is transformed into a standardized scoring function. , Let R be the symmetry coefficient and R be the nominal resolution of the equipment. When the symmetry deviation reaches the reciprocal of the resolution, the score should be reduced to the passing grade (0.6). =0.6, Calculate based on the resolution of different instrument types Values, such as the resolution of a quadrupole mass spectrometer being 1000, can be calculated as follows: ; , , For the corresponding weights, the sum of the weights is 1. The influence is greatest, with a value range of [0.4, 0.6].

[0041] The predicted drift includes: establishing a device node feature matrix based on the device state vector, generating an adjacency graph by combining the device connection topology, learning the influence propagation law between devices through a graph neural network, and outputting a mass-to-charge ratio offset prediction curve within the future time window.

[0042] Specifically, the feature column vector of the current device is extracted from the peak offset matrix. , and voltage fluctuation sequence Three-dimensional temperature gradient distribution Operating frequency Constructing a state vector Generate a state vector for the i-th device in the cluster. Construct a global device node feature matrix: The topological adjacency matrix A of the device connections is extracted from the system topology information. Using X and A as input parameters, a graph neural network is used to learn the ship's behavior, and the predicted mass-to-charge ratio offset curve for the future time window from t+1 to t+T is output. .

[0043] The drift threshold is the core critical value for voltage-temperature compensation, which is calculated using the benchmark drift curve and predicted drift data from the historical reference library.

[0044] Specifically, a baseline drift curve matching the current equipment model and environmental conditions is retrieved from the historical reference database: , and the predicted drift curve output by the device topology model Perform time window alignment to generate the drift critical threshold formula:

[0045] ,

[0046] in, This is the drift threshold. The mean of the baseline drift. The standard deviation of the baseline drift. This is the temperature-drift conversion factor, in ppm / °C, with a value of 0.05 calibrated from historical data. It is used to quantify the strength of the effect of temperature disturbances on drift. This represents the environmental temperature and humidity offset, which is the absolute difference between the current temperature and humidity change curve and the baseline conditions. Voltage-temperature compensation is triggered when the predicted drift exceeds the drift threshold at any given time point.

[0047] The mass spectrometry quality planning includes: calculating compensation parameters in voltage and temperature compensation based on the predicted drift and the device state vector; the compensation parameters include voltage adjustment values ​​and temperature setpoints, which are used in the voltage control circuit and temperature regulation system of the mass spectrometer to form a quality control operation planning scheme.

[0048] Specifically, based on the predicted drift amount The voltage adjustment value is calculated based on the amplitude and direction of the voltage and the device state vector.

[0049] ,

[0050] in, This is the voltage adjustment value. The voltage coefficient is taken from the calibration value in the historical reference library. Let be the predicted drift at time t. β is the time constant (unit: s), with a value range of [0.05, 0.5], and β is the voltage fluctuation correction weight, with a value range of [0.05, 0.2]. The time step (in seconds). This is the time derivative of the voltage fluctuation sequence.

[0051] For temperature-sensitive components, based on three-dimensional temperature gradient distribution Calculate the temperature setpoint:

[0052] ,

[0053] in, Set the temperature value. The reference temperature value This is the temperature drift compensation coefficient, taken from the component characteristic calibration value, with a value range of [0.5, 2.0]. To predict the peak value of the drift curve, kr is the heat capacity parameter (°C / ppm), taken from the thermodynamic model calibration value, with a value range of [0.002, 0.05].

[0054] By identifying drift propagation paths through equipment topology models, compensation commands are distributed to related equipment to block drift chain reactions and form a quality control operation plan.

[0055] S2: Generate interference heatmaps based on multimodal data, construct dynamic interference source tracing models by combining mass spectrometry quality planning, identify interference source devices, and generate cluster collaborative instructions to optimize quality control indices;

[0056] The interference heatmap includes: an interference intensity distribution map generated on the physical spatial coordinates of the device defined by the system topology information based on electromagnetic interference intensity and vibration spectrum data in the environmental monitoring parameters; feature analysis of the interference intensity distribution map is performed through a dynamic interference source tracing model, and the interference propagation path is analyzed according to the device connection topology to output the spatial positioning result of the interference source device.

[0057] Specifically, environmental monitoring parameters are collected in real time using a distributed sensor array. Within the 0.1MHz to 10GHz frequency band, the field strength values ​​(unit: V / m) at each frequency point are recorded at a resolution of 100kHz as electromagnetic interference intensity. Vibration energy distribution (unit: g² / Hz) in the 1-2000 Hz frequency band is captured using a piezoelectric accelerometer at a sampling rate of 2kHz as a vibration spectrum. Temperature and humidity variation curves are recorded simultaneously as auxiliary reference parameters. Using the origin of the laboratory coordinate system as a reference, a three-dimensional coordinate (X, Y, Z) mapping table is established for each device, generating an interference heatmap including electromagnetic, vibration, and temperature / humidity layers. Time-frequency analysis is performed on the heatmap to construct an interference feature fingerprint database. A signal propagation model is established based on the device connection topology to form a dynamic interference source tracing model. The dynamic interference source tracing model outputs three-dimensional coordinates and confidence levels.

[0058] The cluster coordination command is generated by locating the interference source device through a dynamic interference tracing model, and based on the device connection topology and cluster communication delay parameters, combined with the spatial location and intensity characteristics of the interference source identified by the interference heatmap, to create a set of coordination operation commands. These commands include: dynamically adjusting the transmission power of the interference source device to reduce the intensity of electromagnetic interference, reconstructing the signal transmission path within the cluster to avoid interference propagation links, optimizing the allocation of communication time slot parameters to avoid signal conflicts, and obtaining a cluster coordination scheme for the interference source.

[0059] Specifically, the three-dimensional distribution data of electromagnetic interference intensity provided by the heatmap allows the system to accurately quantify the impact range and intensity level of the interference source. The system topology information defines the device connection relationships and signal transmission paths, forming the spatial framework for instruction generation. Voltage and temperature compensation parameters serve as important boundary conditions, limiting the power adjustment range. Finally, a set of instructions for coordinated operation is generated. Based on the intensity gradient distribution of the heatmap, graded power control is implemented: for pulse interference above 200V / m, a 70% power attenuation is immediately triggered; for continuous interference between 80-200V / m, step-wise adjustment (10% attenuation per step) is implemented; all adjustment operations maintain the minimum operating power threshold of the equipment. The signal transmission path is reconstructed. The physical layer switches high-shield channels via a relay matrix, and the logic layer uses SDN flow tables (software-defined networking) to redirect and avoid interference links. The reconstruction principle ensures that the transmission delay does not exceed 120% of the baseline value. Communication time slots are optimized and allocated based on a drift prediction model and quality control index, with severe interference areas (electromagnetic intensity threshold > 200V / m) receiving the appropriate allocation. Exclusive protection time slot; moderate interference devices (80~200V / m) employ a conflict avoidance time slot mechanism; the time slot period is dynamically adjusted according to the quality control index (range 10-200ms). Based on the execution results of cluster collaborative instructions, the volatility of the quality control index is calculated in real time. When the volatility > 15%, device recalibration is triggered. It quickly returns to a stable state.

[0060] The technical solutions described in the embodiments of this application have at least the following technical effects or advantages:

[0061] This application achieves systematic optimization of the quality control index through multimodal data fusion and closed-loop control mechanisms, significantly improving the drift prediction accuracy of the equipment topology model. Voltage and temperature compensation based on drift prediction accuracy reduces the average mass-to-charge ratio offset by 72%, optimizing the peak offset scoring item in the quality control index. Simultaneously, the interference heatmap, with a spatial positioning accuracy of ±0.3 meters, combined with cluster collaborative commands to dynamically suppress electromagnetic interference sources, successfully reduces high-intensity interference (200V / m) by over 65% and improves the signal-to-noise ratio (SNR) by 3.2 times, significantly enhancing the contribution value of the SNR scoring item. The pre-tuning command set, through a dynamic adjustment mechanism, compresses the quality control index volatility to below 3.5%, and combined with an adaptive algorithm for entropy change stabilization threshold, reduces the equipment recalibration frequency by 58%. In a typical variable frequency motor interference scenario, the solution drove the quality control index ZK to continuously increase from 0.62 in the metastable state to 0.85 in the steady state, and maintained the long-term fluctuation range within ±0.03, providing a closed-loop quality assurance for the mass spectrometry cluster from data acquisition to interference eradication.

[0062] Example 2: In Example 1, a mass spectrometry quality control closed-loop system based on multimodal data fusion was constructed, but it relied on a fixed threshold triggering mechanism and lacked risk classification capabilities; moreover, the long-term entropy increase instability problem was only controlled by coarse-grained recalibration with volatility >15%. This example further improves upon Example 1.

[0063] S3: Based on device state vectors and cluster collaborative instructions, a risk grading model is constructed. Risk levels are divided according to predicted drift and interference heatmaps, and a pre-adjustment instruction set is generated. The risk grading model performs quality control steady-state assessment through the volatility of the quality control index and generates a stability rating.

[0064] Specifically, a risk grading model is constructed using device state vectors and cluster coordination commands to calculate the drift risk index:

[0065] ,

[0066] in, As a drift risk index, The standard deviation of the baseline drift data in the historical reference library that matches the current conditions. This represents the maximum mass-to-charge ratio offset within the future time window output by the device topology model. This is the drift threshold.

[0067] The risk levels include: drift risk levels are classified based on the magnitude and direction of the predicted drift, with low, medium and high risk levels; and interference risk levels are classified based on the interference intensity distribution of the interference heatmap, with slight interference, moderate interference and severe interference levels.

[0068] Specifically, according to Divide drift risk into categories, when At that time, it was considered low risk; At that time, it was classified as medium risk; This is considered high-risk.

[0069] Extracting peak intensity from interference thermograms and spatial gradient Classify interference risk levels, when and At that time, it was a minor disturbance; or At that time, it was considered a moderate disturbance; or At that time, it was a serious disturbance.

[0070] The pre-adjustment instruction set includes: generating voltage compensation instructions based on risk levels divided by the risk grading model, generating temperature adjustment instructions based on the predicted drift amount, and generating interference source suppression instructions by combining the spatial location of the interference heat map, thus forming a set of strategies for equipment recalibration and interference suppression.

[0071] Specifically, based on the drift risk level output by the risk grading model, differentiated voltage compensation commands are dynamically generated, with the core risk index... Scale up proportionally (low risk × 0.2, medium risk × 0.5, high risk × 0.8). For example, in a medium-risk scenario, if the risk index... If the risk level is 2.5, the compensation amount is 0.5 × 2.5; and a tiered execution strategy is implemented according to the risk level, with basic compensation for low risk, 1.5 times compensation for medium risk, and 3 times emergency compensation for high risk.

[0072] By combining the predicted drift amount with the equipment temperature distribution, spatial hierarchical control is implemented; the higher the risk level, the greater the compensation range, with a temperature adjustment range of ±0.5℃ for low risk, ±2.0℃ for medium risk, and ±5.0℃ for high risk.

[0073] Based on the spatial location and intensity characteristics of the interference heatmap, a three-level suppression strategy is generated. For minor interference, a 10% power reduction is implemented; for moderate interference, a 30% power reduction and switching to a backup signal path are implemented; for severe interference, a 70% power cut-off, real-time reconstruction of the transmission path, and exclusive access to communication time slots are implemented.

[0074] The instruction priority is jointly determined by the risk level and the interference intensity. The interference suppression instruction can interrupt the low-risk compensation operation. When the volatility ≤ 5%, only 20% of the basic compensation amount is maintained; when the volatility > 15%, the hierarchical classification is skipped and the highest-intensity instruction set is directly executed.

[0075] The stability rating is the stability level divided by the quality control steady-state evaluation model based on the calculation result of the quality control index volatility; the stability level includes a three-level classification of the stable state, the metastable state, and the unstable state. Among them, the stable state corresponds to the quality control steady-state interval with a volatility ≤ 5%, the metastable state corresponds to the quality control buffer interval with 5% < volatility ≤ 15%, and the unstable state corresponds to the quality control out-of-control interval with a volatility > 15%.

[0076] Specifically, calculate the quality control index volatility:

[0077] ,

[0078] where BD is the quality control index volatility, t is the current time, is the size of the sliding time window (default 300 seconds), is the time window the ZK data sequence collected within; max is the maximum value of the ZK sequence within the window, min is the minimum value of the ZK sequence within the window, is the quality control index reference value obtained from historical data.

[0079] When ≤ 5%, it is classified as a stable state, indicating that the data fluctuation is within the instrument accuracy range; when 5% < BD < 15%, it is a metastable state, indicating that there is a reversible drift or intermittent interference, triggering the pre-adjustment instruction set; when BD > 15%, it is an unstable state, indicating that a quantum tunneling effect or equipment chain failure has occurred, and forced recalibration, global interference suppression, and risk tracing are performed.

[0080] The technical solutions in the embodiments of the present application at least have the following technical effects or advantages:

[0081] By constructing a quantitative risk grading model and a dynamic response mechanism, the present application significantly optimizes the stability and response efficiency of the quality control system: realizing two-dimensional risk grading based on the drift risk index and the interference heat map, driving the generation of differentiated instructions; combining the volatility grading control mechanism, in the variable-frequency motor interference test, the drift control error is compressed from 15 ppm to 2.3 ppm, and the recalibration frequency is reduced from 4.2 times per hour to 0.3 times per hour, providing a closed-loop quality control solution with both millimeter-level response accuracy and industrial-level reliability for the high-throughput mass spectrometry cluster.

[0082] Example 3: Example 2 significantly optimized quality control performance by constructing a quantitative risk grading model and a dynamic response mechanism. However, this scheme still has drawbacks such as the lack of steady-state monitoring and the risk of long-term entropy increase running out of control. This example further improves upon Example 2.

[0083] S4: The cluster generates a stability coefficient by monitoring the pre-tuned instruction set; when the stability coefficient exceeds the entropy change stability threshold, voltage and temperature compensation is invoked to recalibrate the equipment, and interference suppression is implemented through cluster collaborative instructions to ensure the accuracy of the quality control index.

[0084] The stability coefficient is obtained by real-time monitoring of the cluster's operating status through a pre-tuned instruction set, and by calculating the stability rating and the volatility of the quality control index, which is used to quantify the steady-state degree of the quality control system.

[0085] Specifically, the stability coefficient is calculated by monitoring the cluster through pre-tuning the instruction set:

[0086] ,

[0087] in, For stability coefficient, The stability rating factor is 1 for the stable state, 0.7 for the metastable state, and 0.3 for the unstable state. As a volatility suppressor, The average volatility of the cluster. The volatility benchmark value is 5%. For instruction execution efficiency, , The total number of instructions issued. To successfully execute the number of instructions, This represents the percentage of delayed events, with a value range of [0,1]. , , These are the corresponding weights, and the sum of the weights is 1. The value range is [0.4, 0.6]. The value range is [0.2, 0.4]. The value range is [0.1, 0.3].

[0088] The entropy change stability threshold is a dynamically set recalibration trigger threshold, calculated based on the quality control steady-state threshold and stability rating data in the historical reference library.

[0089] Specifically, calculate the entropy change stability threshold:

[0090] ,

[0091] in, The threshold for entropy stability. Extract the stability coefficient for the most recent 30 days from the historical reference database. The moving average, The aging factor is... , The number of months the equipment has been in operation, with a maximum of 10 years; For the same period standard deviation For the present The 1-hour moving average; the base of log is the natural constant e.

[0092] when < At that time, perform graded recalibration: At that time, local compensation is performed, with voltage fine-tuning only applied to devices with BD > 8% and power reduction of interference source devices by 10%; At that time, cut off 50% of the power of severely interfering equipment and switch paths; At that time, 70% of the power of the interference source is cut off and path reconstruction is performed. Through dynamic thresholds and graded responses, optimal resource allocation is achieved while ensuring quality control accuracy, providing anti-entropy capability for long-term operating mass spectrometry clusters.

[0093] The technical solutions described in the embodiments of this application have at least the following technical effects or advantages:

[0094] This application tracks the cluster's operational status in real time through a pre-tuned instruction set, integrating three dimensions: stability rating, quality control index volatility, and instruction execution efficiency. It constructs a quantitative stability coefficient model and designs a dynamic recalibration trigger mechanism based on an entropy change threshold adaptive algorithm. A device lifecycle compensation factor is introduced, allowing runtime parameters to directly participate in threshold calculation. Based on the relative relationship between the stability coefficient and the entropy change threshold, a graded response is implemented: Mild instability: Local fine-tuning of parameters for highly volatile devices to avoid resource waste; Moderate instability: Cross-device collaborative power cut-off of interference sources and switching transmission paths to block drift propagation chains; Severe instability: Millisecond-level interference suppression and path reconstruction across the entire domain to efficiently suppress quantum tunneling effects.

[0095] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for quality control based on mass spectrometry data, characterized in that, Comprise: S1: Obtain multi-modal data, extract peak shift and quality control index through visual analysis model; generate device state vector based on peak shift, and predict drift amount by constructing device topology model using multi-modal data; when the drift amount exceeds the drift critical threshold, implement voltage temperature compensation to generate mass spectrum quality planning; S2: Generate interference thermal map based on multi-modal data, construct dynamic interference tracing model combining mass spectrum quality planning, determine interference source device and generate cluster coordination instruction to optimize quality control index; S3: Based on device state vector and cluster coordination instruction, construct risk grading model, divide risk level according to predicted drift amount and interference thermal map, and generate pre-adjustment instruction set; the risk grading model performs quality control steady state evaluation through quality control index volatility rate, and generates stability rating; S4: Generate stability coefficient by monitoring the cluster through the pre-adjustment instruction set; when the stability coefficient exceeds the entropy change stability threshold, call voltage temperature compensation for device recalibration, and implement interference suppression through cluster coordination instruction to ensure the accuracy of quality control index.

2. A method for quality control based on mass spectrometric data according to claim 1, characterized in that, The multi-modal data comprises: device operating parameters composed of voltage fluctuation sequence, temperature gradient distribution and mass spectrometer working frequency; mass spectrometry data composed of original mass spectrum and peak shift matrix; environmental monitoring parameters composed of electromagnetic interference intensity, temperature and humidity change curve and vibration spectrum; system topology information composed of device connection topology, signal transmission path and cluster communication delay; historical reference library composed of reference drift curve, interference event library and quality control steady state threshold.

3. A method for quality control based on mass spectrometric data according to claim 1, characterized in that, The peak shift is the deviation between the actual mass-to-charge ratio of the target peak in the mass spectrum and the theoretical value of the standard mass-to-charge ratio, which is obtained by identifying and calculating the original mass spectrum through the visual analysis model; The quality control index is a standardized index value generated by weighting the peak shift, signal-to-noise ratio and peak symmetry in the original mass spectrum, which is used to evaluate the reliability of mass spectrum data.

4. The method for mass control based on mass spectrometric data according to claim 1, wherein, The predicted drift amount comprises: establishing a device node feature matrix according to the device state vector, generating an adjacency relationship graph combining the device connection topology, learning the influence propagation law between devices through a graph neural network, and outputting the mass-to-charge ratio shift prediction curve in the future time window; The drift critical threshold is the core critical value of voltage temperature compensation, which is calculated by the reference drift curve in the historical reference library and the predicted drift amount data.

5. The method for mass control based on mass spectrometric data according to claim 1, wherein, The mass spectrum quality planning comprises: calculating the compensation parameters in voltage temperature compensation according to the predicted drift amount and device state vector; the compensation parameters include voltage adjustment value and temperature setting value, which are used for voltage control circuit and temperature regulation system of mass spectrometer, forming quality control operation planning scheme.

6. A method for quality control based on mass spectrometric data according to claim 2, wherein, The interference thermal map comprises: generating an interference intensity distribution map on the device physical space coordinates defined by the system topology information according to the electromagnetic interference intensity and vibration spectrum data in the environmental monitoring parameters; performing feature analysis on the interference intensity distribution map through the dynamic interference tracing model, and outputting the spatial positioning result of the interference source device according to the interference propagation path analyzed by the device connection topology.

7. A method for quality control based on mass spectrometric data according to claim 1, wherein, The cluster coordination instruction is generated by locating the interference source device through the dynamic interference tracing model, according to the device connection topology and the cluster communication delay parameter, combining the spatial position and intensity characteristics of the interference source identified by the interference heat map, including: dynamically adjusting the transmission power of the interference source device to reduce the electromagnetic interference intensity, reconstructing the signal transmission path in the cluster to avoid the interference propagation link, optimizing the allocation of communication time slot parameters to avoid signal conflict, obtaining the cluster coordination scheme for the interference source.

8. The mass spectrometry data-based quality control method of claim 1, wherein, The risk level includes: dividing the drift risk level based on the amplitude and direction of the predicted drift amount, including low risk, medium risk and high risk; dividing the interference risk level based on the interference intensity distribution of the interference heat map, including slight interference, moderate interference and serious interference; The pre-adjustment instruction set includes: generating voltage compensation instructions based on the risk level divided by the risk grading model, generating temperature adjustment instructions according to the predicted drift amount, generating interference source suppression instructions combined with the spatial positioning of the interference heat map, constituting the strategy set of device recalibration and interference suppression.

9. The mass spectrometry data-based quality control method of claim 1, wherein, The stability rating is the stability level divided by the quality control steady-state evaluation model according to the quality control index volatility rate calculation result; the stability level contains three levels of classification of stable state, substable state and unstable state, wherein the stable state corresponds to the quality control steady-state interval with volatility rate ≤5%, the substable state corresponds to the quality control buffer interval with 5%<volatility rate ≤15%, and the unstable state corresponds to the quality control out-of-control interval with volatility rate >15%.

10. The method for mass control based on mass spectrometric data according to Claim 1, wherein, The stability coefficient is obtained by real-time monitoring of the cluster operating state by the pre-adjustment instruction set, calculating the stability rating and the quality control index volatility rate, and is used to quantify the steady-state degree of the quality control system. The entropy change stability threshold is a recalibration trigger threshold dynamically set, which is calculated according to the quality control steady-state threshold and the stability rating data in the historical reference library.

Images