111 results about "Baseline drift" patented technology

The invention discloses a gas concentration detection circuit and method based on differential measurement and reference compensation, and the circuit comprises at least one active gas sensitive element, the electrical parameters of which change along with the concentration of a target gas; the sensitivity of the reference gas sensitive element to the target gas is at least lower than that of the active gas sensitive element by a preset percentage threshold value; the active gas sensitive element and the reference gas sensitive element are connected in a differential pair form to form a differential signal generation unit; the differential amplification circuit is used for amplifying the differential signal output by the differential signal generation unit; the output signal of the differential amplification circuit is used for representing the concentration information of the target gas. According to the invention, environmental interference and baseline drift are directly and effectively suppressed from the hardware level, the stability and selectivity of gas concentration measurement are improved, and meanwhile, the holding circuit is simple in structure, low in cost and convenient for large-scale application.

The invention discloses a method for detecting tea saponin in a tea extract, and relates to the technical field of chemical analysis and computer science, and the method comprises the following steps: eliminating high-frequency noise of tea extract data by using a Gaussian filtering algorithm, eliminating baseline drift by using a baseline correction algorithm, and generating a tea extract sample data stream; the method comprises the following steps: performing dynamic gradient separation through a chromatographic signal simulation algorithm on the basis of a tea extract sample data stream to generate a virtual chromatographic peak sequence, calculating a peak area through a sliding window integration algorithm on the basis of the virtual chromatographic peak sequence, and generating chromatographic signal data by using a wavelet transform threshold denoising algorithm, the chromatographic signal data are input into the deep convolutional neural model, the feature pyramid network is obtained through the multi-scale feature fusion algorithm, the feature information of the chromatographic peaks is extracted from different scales by constructing the deep convolutional neural model and combining the multi-scale feature fusion algorithm, and therefore the recognition precision of the feature peaks is improved.

The invention relates to the technical field of liquid chromatographic analysis, and discloses a data acquisition method and system of a liquid chromatograph. The method comprises the following steps: extracting chromatograms without solvent residues and with known solvent residues from a historical database, and training by using a self-encoding neural network and a convolutional neural network to generate a baseline reference model and a ghost peak early warning model; obtaining solvent parameters of the previous analysis method, and calculating the residual solvent concentration through residual solvent information; inputting the residual solvent concentration into the baseline reference model and the ghost peak early warning model, and calculating to obtain a baseline drift prediction correction value and a ghost peak prediction risk area; pre-correcting the chromatographic signal acquired in real time to obtain a purified chromatographic signal; performing qualitative and quantitative analysis on the purification signal according to dynamic allocation resources of the ghost peak risk area to generate final chromatographic data; the data interference of the mobile phase residue is effectively eliminated, and the accuracy of data acquisition is improved.

The invention relates to the technical field of gene detection, and discloses a mycobacterium tuberculosis drug resistance gene detection data analysis system which comprises a data import module, a signal correction module, a feature extraction module, a multi-stage interpretation module, a result fusion module and a report generation module. The system performs baseline drift correction, abnormal point detection and nonlinear normalization on a fluorescence intensity value output by a detector, extracts multi-dimensional features such as peak intensity, a mutation ratio, a consistency coefficient and an in-batch difference index, and constructs a dynamic threshold function to realize wild type and mutation type layered judgment. A drug resistance risk conclusion is generated through drug resistance characteristic matrix mapping, a standardized detection report is output, and intelligent analysis of a detection result and automatic report generation are achieved. Through multi-stage signal correction, multi-dimensional feature extraction and dynamic threshold interpretation, intelligent analysis and standardized report output of a mycobacterium tuberculosis drug resistance gene detection result are realized, and data judgment accuracy and clinical application efficiency are improved.

The invention relates to the technical field of heating and ventilation fault recognition, in particular to a heating and ventilation system fault recognition method based on operation state offset modeling and a control system. The method comprises the following steps: acquiring steady-state operation data of the heating and ventilation system and extracting actual parameters of equipment; inputting the environment and load parameters into the lightweight integrated tree base line model, and outputting an equipment operation parameter reference value; performing baseline drift correction on the reference value in combination with the accumulated operation duration of the equipment, calculating a dynamic difference value between an actual value and a calibration reference value, and constructing an operation state offset characteristic matrix; and inputting the matrix into a preset fault diagnosis rule base, matching a hidden fault in combination with the dynamic tolerance interval and the multi-dimensional Boolean logic, and giving an alarm. Performing reference compensation by using the accumulated operation time length in combination with a linear or exponential progressive attenuation model; on the premise of not consuming high computing power to re-train an underlying algorithm model, the interference of natural physical aging of the equipment on residual calculation is eliminated.

The application discloses a kind of detection device and method of copper alloy melt purity under vacuum environment, belong to copper alloy purity detection technical field.Detection device includes vacuum smelting cavity module, laser-induced breakdown spectroscopy detection module, temperature monitoring module, central data processing and control module and man-machine interactive display module.Detection method includes system initialization, laser-induced breakdown spectroscopy multi-pulse acquisition, spectrum adaptive baseline compensation, inclusion distribution uniformity quantification and purity comprehensive evaluation.Through segmented polynomial iterative threshold truncation algorithm, adaptive eliminate blackbody radiation and plasma continuous radiation superposition in vacuum high-temperature environment and form complex nonlinear baseline drift;Inclusion distribution uniformity index and concentration purity sub-index are weighted and fused, realize the comprehensive quantitative characterization of melt purity concentration dimension and uniformity dimension, and realize the real-time online quality determination of refining process by comparing with qualified threshold.

The invention relates to a method for rapidly determining the content of K.2. 2.2. 2 in an FDG card sleeve kit leacheate, and belongs to the technical field of radiopharmaceutical analysis. The method comprises the following steps: selecting a 1110 + / -5 cm <-1 > peak as a quantitative characteristic peak of K.2.2. 2, wherein the peak corresponds to symmetric stretching vibration of a C-O-C ether bond in a K.2.2. 2 molecule; selecting a 2253 + / -3cm <-1 > peak as an internal standard peak, wherein the peak corresponds to C = N triple bond stretching vibration of acetonitrile in the leacheate; preparing a K.2. 2.2. 2 gradient solution with the concentration of 0.05 to 5 mg / mL, and fixing the concentration of K2CO3 to be 0.5 to 2 mg / mL; the method comprises the following steps: acquiring a spectrum by adopting an ATR technology, setting the resolution ratio to be 4cm <-1 > and the scanning frequency to be 32, and establishing a K.2. 2.2. 2 linear model according to the ratio of A1110 / A2253; eliminating baseline drift by adopting a second derivative spectrum; and a PLS algorithm is adopted to correct trace interference of K2CO3.

The invention relates to the technical field of electromagnetic flow meters, in particular to an online calibration method and system for an electromagnetic flow meter and the electromagnetic flow meter. The online calibration method analyzes an error source of baseline drift on the basis of the basic principle of the electromagnetic flow meter, establishes an error estimation function, and determines the error estimation function when the flow is stable. Various coefficients of an error estimation function are adjusted by continuously using newly obtained baseline offset, excitation coil resistance and electrode liquid connection resistance, so that the error estimation function can track the state change of the electromagnetic flowmeter, online calibration is carried out on the electromagnetic flowmeter by utilizing each flow stabilization period, and the error estimation precision is continuously improved. After an estimation error is calculated through the error estimation function optimized through the steps, the measured induced electromotive force can be corrected, baseline drift is eliminated, and therefore on the premise that the normal measurement process of the electromagnetic flowmeter is not interrupted, key parameters such as baseline offset can be calibrated through a gap with stable flow.

The invention provides a breathing mode classification method and system based on a CNN-LSTM neural network, and the method comprises the steps: collecting body surface point cloud data in a non-shielding state and an arm shielding state in a human body breathing process, and extracting a breathing motion feature on the basis of extracting the breathing motion feature; the invention provides an optimization method for abnormal respiratory movement characteristics of a significant respiratory movement area, solves the interference of environmental noise and newborn clinical characteristics on respiratory signals, and comprises the following steps of: firstly, partitioning a thoracic and abdominal voxel model, and extracting the respiratory movement characteristics of the significant area based on KPCA (Kernel Principal Component Analysis); motion artifacts are removed through a Savitzky-Golay filter, body motion interference is inhibited by using a method of fusing a peak threshold method and a local variance threshold, and baseline drift is removed by using a grey wolf optimization algorithm. According to the method, a CNN-LSTM neural network is constructed to classify four breathing modes of normal, rapid, slow and pause, model parameters and evaluation indexes are determined, effective classification and recognition of the breathing modes are achieved, and the monitoring precision and clinical application value are improved.

This invention provides a handheld method and device for detecting vegetation index of crop leaves, relating to the field of plant phenotyping technology. The method includes: controlling a contact linear array image sensor to scan the leaf of the crop to be tested line by line in the shaded leaf channel to obtain multi-channel reflectance grayscale digital quantities corresponding to each band; sequentially performing radiation baseline drift suppression, fluorescence cross-channel crosstalk compensation, and crosstalk bias correction on the multi-channel reflectance grayscale digital quantities to obtain multi-channel response data; determining the vegetation index of the leaf of the crop to be tested based on the multi-channel response data, and extracting the leaf phenotypic parameters of the leaf of the crop to be tested; determining the growth status of the leaf of the crop to be tested based on the vegetation index and leaf phenotypic parameters. This method can suppress the influence of ambient light interference, fluorescence crosstalk, and multi-band crosstalk on the measurement results, improve the accuracy of the vegetation index and the reliability of the leaf phenotypic parameters, and is beneficial for high-frequency, continuous, single-person handheld operation under field conditions.

The invention belongs to the technical field of intelligent operation and maintenance and data quality control, and particularly relates to a transformer oil gas chromatographic image data defect diagnosis method and system.The method includes the steps that chromatographic images are received and subjected to standardized preprocessing, and a chromatographic signal sequence is reconstructed; establishing a mapping relation between a chromatographic peak and a gas component by using an OCR (Optical Character Recognition) technology; baseline drift, fluctuation and negative peaks are identified through baseline correction; locating a chromatographic peak based on the corrected signal, and identifying abnormities such as peak disappearance, ghost peak, peak separation degree difference and bulge; performing multi-label peak shape classification on the monochromatic spectrum peak sequence through a convolutional neural network model, and identifying six types of peak shape anomalies such as a leading peak and a trailing peak; and finally, fusing the expert knowledge graph to generate a structured diagnosis report containing cause analysis and processing measures. According to the method, full-process automation from image input to report output is realized, the objectivity and accuracy of chromatographic data quality evaluation are remarkably improved, and a reliable data basis is provided for transformer fault diagnosis.

Embodiments of the present application relate to a kind of electrocardiosignal processing method and device, the method includes: receiving first electrocardiosignal;Filtering processing generates second electrocardiosignal;Baseline drift elimination processing is respectively carried out to first, second electrocardiosignal, and third, fourth electrocardiosignal is generated;R point identification processing is carried out to fourth electrocardiosignal, and first R point sequence is generated;Electrocardiosignal segment intercepting processing is carried out to third, fourth electrocardiosignal, and first, second electrocardiosignal segment sequence is generated;Heart beat classification processing is carried out to first electrocardiosignal segment sequence, and first segment type sequence is generated;Interference classification processing is carried out to second electrocardiosignal segment sequence, and second segment type sequence is generated;Segment type fusion processing is carried out to first, second segment type sequence, and third segment type sequence is generated;First R point sequence and third segment type sequence are as electrocardiosignal processing result output.The processing efficiency of electrocardiosignal processing can be improved in the present application and the stability of processing quality is guaranteed.

This invention provides a data fusion method for multi-parameter environmental sensor chips based on artificial intelligence, belonging to the field of agricultural product growth environment monitoring technology. This invention constructs a continuous data sequence using a multi-parameter environmental sensor chip, evaluates and preprocesses the data quality using anomaly jump matrices and interference anomaly matrices, constructs a multi-dimensional fusion algorithm to calculate convergence, stability, and disorder indices to quantify the data fusion state, and employs an adaptive calibration algorithm combined with jump compensation and anomaly compensation matrices to dynamically compensate for sensor baseline drift and sensitivity attenuation in real time. The number of attention approximation features is dynamically adjusted through a feature adjustment function, ultimately achieving classification prediction and quality assessment of agricultural product growth environment. This invention solves the technical problem that baseline drift and sensitivity attenuation in multi-parameter environmental sensors significantly reduce data fusion accuracy over time during long-term operation.

The application discloses a self-calibration wrist surface myoelectric gesture recognition method, system, device and medium, comprising the following steps: first, multi-channel acquisition of the wrist surface myoelectric signal of the wearer when performing gestures; then, baseline drift, filtering, noise reduction and action segment judgment are performed on the signal, and 3D time domain features are extracted; then, the features are input into a CNN-LSTM-Attention network model for offline pre-training; finally, a small amount of labeled data of a to-be-tested user is collected, the initial accuracy of the model is verified, if the initial accuracy is lower than a threshold value, online self-calibration is performed on the model by using KNN pseudo-labeling and stability regularization term, and the model is used for real-time gesture recognition until a preset accuracy is reached. The method adopts the CNN-LSTM-Attention model to fuse local, time sequence and context features, and introduces an online self-calibration mechanism, so that the adaptability and recognition accuracy of the model to new users are effectively improved.

This invention discloses a rapid near-infrared analysis method for coal quality based on constrained learning and Adaptive Deep Belief Network (ADBN). First, a near-infrared spectrometer is used to acquire diffuse reflectance spectra of coal samples. Each sample is acquired 3-5 times and averaged to reduce random errors, and the true values ​​of corresponding coal quality parameters are determined. Next, the spectral data is preprocessed to eliminate noise, baseline drift, and scattering interference. A constrained learning algorithm is used to select feature bands strongly correlated with coal composition, constructing constrained spectral feature vectors. An Adaptive Deep Belief Network (ADBN) is then constructed. The spectral feature vectors are input into the model, incorporating physical constraints on coal quality parameters, spectral feature correlation constraints, and regularization constraints into the loss function. Finally, the trained model is used to predict the spectral composition of unknown coal samples, obtaining the coal composition content. This method effectively removes redundant bands and noise, solving the redundancy problem of traditional feature selection. ADBN can dynamically adjust the network structure to adapt to the inherent laws of coal quality, improving prediction accuracy and reliability.

The embodiment of the application provides a cascade anti-interference circuit suitable for an electrocardio monitoring device, which comprises a sliding mean filter module, a notch filter module, a lifting wavelet decomposition module, a threshold calculation module, a threshold processing module and a lifting wavelet reconstruction module; while ensuring a small circuit scale, the ECG signal collected by the wearable electrocardio monitoring device is subjected to hierarchical noise reduction processing, and high signal-to-noise ratio ECG signal output is realized; for the ECG signal collected by the wearable electrocardio monitoring device, first, sliding mean filtering is performed to filter out baseline drift noise caused by human respiratory movement and other activities; the power frequency interference caused by the wired and wireless connection of the electromagnetic environment around the device is suppressed by using the notch filter module; finally, wavelet noise reduction is realized by using the lifting wavelet decomposition and reconstruction and threshold noise reduction method, the electromyographic interference caused by the autonomous or unconscious movement of the wearer is suppressed, and high signal-to-noise ratio ECG signal is output, which is used for physiological parameter extraction and electrocardio diagnosis.

The invention discloses an infrared spectrum signal processing algorithm based on a TDLAS technology, and belongs to the technical field of infrared spectrum detection. The algorithm does not need to independently carry out zero gas or reference gas detection, and comprises the following steps: collecting an infrared spectrum original signal of a gas to be detected; fourier transform is carried out on the original signal, and a first harmonic and a second harmonic are separated; determining a non-absorption wavelength interval based on the second harmonic, fitting a drift compensation reference line in combination with the signal of the first harmonic in the interval, and further fitting a full-wavelength signal baseline; and calculating a difference value between the original signal and the baseline to obtain an absorption signal, and selectively carrying out noise reduction optimization. According to the method, the detection steps are simplified, baseline drift caused by detection time difference is eliminated, the baseline accuracy and the absorption signal extraction precision are improved, and the method is suitable for scenes such as gas concentration detection and environment monitoring and has high practical value.

The application discloses an intelligent health monitoring system based on electroencephalogram signals, and relates to the technical field of intelligent medical treatment, comprising the following steps: constructing a user personalized neural baseline model through small sample calibration and meta-learning; carrying out daily monitoring based on the model, and dynamically updating the model by using an online Bayesian algorithm to track long-term physiological drift of the individual; when significant neural baseline drift is detected, automatically matching and triggering intervention measures; constructing an anti-fact causal inference model to quantify the net effect of the intervention measures on the neural baseline; generating a personalized health insight report based on the net effect; and driving the iterative evolution of the neural baseline model through reinforcement learning by using closed-loop data containing drift, intervention measures and net effect, so that the application realizes intelligent health management with long-term self-adaptation, accurate quantitative intervention effect and self-evolution ability.

The invention relates to an aerosol optical tweezer formant trajectory identification method and system based on iterative clustering, belongs to the technical field of spectrum formant identification, and solves the problem that in the prior art, when aerosol spectrum data with a very long time sequence or extremely high data point density is analyzed, the spectrum data is not accurate. And the problems of difficult trajectory tracking and large computing resource consumption caused by noise and baseline drift are solved. The method comprises the following steps: acquiring multi-frame aerosol spectrum data, and constructing a time-peak position data set; according to the time-peak data set, based on a segmented anchoring rule, determining at least one anchoring trajectory and a corresponding time-peak data subset by adopting a dynamic trajectory tracking algorithm; based on the anchoring trajectory, calculating a relative wavelength difference value, carrying out clustering analysis, and determining a corresponding segmented trajectory; a formant trajectory is determined based on the anchoring trajectory and the segmentation trajectory. The method achieves the automatic and high-precision recognition of a plurality of formant tracks of long-sequence spectral data based on the stable physical constraint of the relative distance between formants.

The application discloses a kind of based on environmental correlation and the living body detection method of Transformer, comprising the following steps: S1: obtaining video data containing face;S2: intercepting the video data of five blocks of face region;S3: the average value of the pixel value of G channel in each region is taken as the only pixel value of this region;S4: remove linear trend from the pixel signal of different time in the same region in order to eliminate baseline drift;S5: the data of the same region is normalized and pretreated;S6: the generated signal wave is fast fourier transform;S7: filter out the frequency less than 50 and greater than 180.It can effectively realize the function of living body detection, avoid false body attack, based on environmental correlation and Transformer, so that living body detection can better judge and detect whether it is living body, distinguish the characteristics of false body and living body, and accurate and stable results can be obtained.

This application relates to a sensor data processing method and electronic device. The method includes: identifying effective movements during the movement of the measured object based on a displacement data stream and the movement type of the measured object; analyzing the effective movements to obtain motion parameters; performing a linear transformation on the motion parameters; and generating an effectiveness analysis result. The displacement data stream is obtained by sequentially filtering, differentiating, and integrating the original data stream. The original data stream includes raw data obtained by the target sensor based on sampling during the movement of the measured object, and the raw data is correlated with the displacement in a preset direction. Using this application, the problems of baseline drift and noise in sensor data can be solved, enabling the evaluation of motion effectiveness and providing reliable data support for the quality analysis of athletes' core movements.

This invention relates to the field of data processing technology, and more specifically, to an intelligent processing method for multi-source monitoring data of high-risk chemical products. The method includes: acquiring the concentration sequence of high-risk chemical gases and synchronous micrometeorological parameters within the target monitoring area; constructing an environmental non-stationarity intensity based on the characteristics of the micrometeorological parameters; calculating a baseline drift tendency coefficient by combining historical trends to assess the probability that concentration changes are attributable to environmental drift; dynamically correcting the background baseline using the baseline drift tendency coefficient to generate net leakage characteristics; and introducing the net leakage characteristics into a cumulative sum control chart algorithm to construct an adaptive damping cumulative quantity to achieve leakage monitoring of high-risk chemical products. This invention accurately assesses the impact of micrometeorology on background concentrations, scientifically distinguishes between environmental drift and actual leakage, effectively eliminates interference and reduces false alarms, dynamically adjusts the cumulative intensity to ensure stable detection under different environments, improves monitoring accuracy and reliability, and provides efficient protection for chemical safety.

The invention belongs to the technical field of information security data processing, and particularly relates to a knowledge graph-based network security situation awareness method, which comprises the following steps of: extracting entity interaction logic in network traffic data, and constructing a topological association graph structure; obtaining threat diffusion intensity based on the alarm weight of the active path, the connection closeness and the total number of nodes; a saturation control function is utilized, a compensation correction value is obtained according to the real-time background flow load and the historical average safety reference flow, and self-adaptive suppression of reference drift generated by service fluctuation is achieved; and based on the compensation correction value and the diffusion intensity difference value, obtaining a whole network situation index in combination with the sudden change gain, and executing closed-loop adaptive control of firewall banning and equipment parameter adjustment. According to the method, the problems that threat diffusion intensity is difficult to evaluate and false alarms are easily generated due to service fluctuation interference in the prior art are solved, and capture of network threat instantaneous outbreak characteristics and resource allocation are realized.

This invention provides a method for identifying polysaccharides from traditional Chinese medicine (TCM) based on multispectral fusion and machine learning, applicable to the rapid and non-destructive identification of polysaccharides from different types and origins. The method first uses infrared and Raman spectroscopy to acquire and construct a raw spectral dataset of TCM polysaccharide samples. The dataset is then preprocessed using standardization to eliminate noise, baseline drift, and other non-sample-related interferences. Effective spectral information is then filtered through feature extraction, followed by feature-level complementary fusion of the infrared and Raman spectral features. Based on the fused feature set, a machine learning classification model is constructed and trained using a cross-validation strategy to achieve accurate identification of the type and origin of TCM polysaccharides. Furthermore, the SHAP interpretability analysis tool is used to analyze the key spectral feature bands upon which the model's discrimination decisions rely, and to analyze the correlation between these key bands and specific functional groups and chemical bond vibrations in the TCM polysaccharide molecules, providing chemical mechanism support for the identification results.

The application discloses a kind of processing method, system, electronic equipment and storage medium of distributed signal array, belong to signal processing technical field.The application utilizes empirical mode decomposition method to process the multiple original signal curves obtained by distributed sensor, the original data curve is carried out empirical mode decomposition, obtains several different intrinsic mode components and a residual steady-state quantity;Determine the optimal trend item in several different intrinsic mode components and a residual steady-state quantity, and the original data curve is subtracted to obtain the data curve after eliminating baseline drift;Multiple data curves after eliminating baseline drift are interpolated to obtain newly encrypted data curves.The method eliminates the error superposition caused by the distributed sensor array structure when interpolating data later, can more clearly locate the pipe wall defect position and accurately depict the defect shape, so as to realize high-resolution in-pipe detection.

The application is suitable for the technical field of touch screen performance detection, and provides a performance detection method and system of visual interactive touch screen based on AI technology, which comprises the following steps: after multi-point touch detection is performed on a display unit to be detected, if the measurement data of the capacitive sensor obtained is abnormal, the current touch working mode is determined, and the initial gain amplitude parameter prepared for the capacitive sensor and the historical processing record of the noise conditioning device used for processing the measurement data of the capacitive sensor are obtained. The initial gain amplitude parameter is dynamically corrected by combining the first correction factor and the second correction factor, so that the adaptive optimization of the processing capacity of the noise conditioning device is realized. The first correction factor is used for evaluating the noise baseline drift condition and reflecting the long-term noise change trend, and the second correction factor is used for analyzing the change condition of the calibration error and evaluating the real-time correction effect of the noise conditioning device.

The present application belongs to the technical field of fault diagnosis, and particularly relates to a belt conveyor bearing fault diagnosis method and system based on temperature and vibration signal analysis. The method comprises the following steps: removing outliers and correcting baseline drift of the collected temperature and vibration signals respectively; using kurtosis and envelope entropy to optimize variational mode decomposition to extract fault components and generate time-frequency diagrams, and using temperature change rate to map weighting coefficients for reconstruction and high-frequency gain compensation; splicing the weighted time-frequency diagram and the two-dimensional temperature diagram into a three-dimensional tensor, inputting the residual network of the band coordinate attention and the deformable convolution branch, and using the temperature energy proportion to adjust the joint loss output to output the diagnosis result. The present application effectively enhances the multi-modal feature expression capability of the model under complex variable working conditions by establishing the physical correlation between temperature change and high-frequency vibration, and significantly improves the diagnosis accuracy and generalization performance.

This invention belongs to the field of battery safety monitoring and health status assessment technology, specifically relating to a battery thermal runaway early warning method and system based on spatiotemporal features and transfer learning. The method includes: simultaneously acquiring real-time mechanical vibration signals and spatial temperature matrix signals of the target battery; performing time alignment and preprocessing to extract a 17-dimensional spatiotemporal feature vector containing accumulated mechanical features and spatial thermodynamic features; inputting this vector into a pre-trained spatiotemporal bidirectional long short-term memory model to obtain an initial predicted value of the remaining thermal runaway time; extracting robust system biases using early few-sample calibration data, performing truncated mean bias correction and physical trend smoothing; and triggering an early warning when the final predicted time is less than a preset safety threshold. This invention eliminates the need for target domain retraining, effectively eliminates baseline drift, and exhibits extremely high early warning accuracy and robustness in cross-chemical system transfer.

The present invention relates to the technical field of data processing. Disclosed are a data-analysis-based resolution improvement method and system for a mass spectrometer. The method comprises: training a resolution factor model; on the basis of an original mass spectrum and an original fragment map, determining basic information of a compound to be tested, determining a target signal-to-noise ratio of the original mass spectrum on the basis of a mass spectrum of a standard compound, and performing denoising processing on the original mass spectrum by means of a denoising method; on the basis of the denoised original mass spectrum and the mass spectrum of the standard compound, determining a baseline reference region; identifying features of peaks in the mass spectrum, separating overlapped peaks by means of the features of the peaks, and performing clustering analysis on all the peaks on the basis of a resolution factor; and performing dimension reduction processing on a class of peaks having data dimensions exceeding a dimension threshold value. In this way, data inconsistency caused by baseline drifts is reduced or avoided. Finally, the resolution of a mass spectrometer is improved by means of the separation of overlapped peaks, the clustering of peaks, and dimension reduction processing of the peaks, thereby ensuring the efficiency of resolution improvement.