169 results about "Sample mass" patented technology

The invention provides a joint optimization method for a stacking sequence and a grabbing posture of a neutron measurement system, and relates to the field of data processing. According to the method, multi-source monitoring data of a neutron measurement system are collected, shadow voxels are constructed in combination with a three-dimensional space model, a neutron resonance shadow domain is formed through clustering superposition, and a dangerous concentration cloud cluster and a safe rarefied area are marked; mapping candidate stacking positions and postures according to sample mass distribution, geometric dimensions, centroid positions and irradiation sensitive characteristics, generating a stacking mode and determining a stacking sequence; performing path risk assessment on the grabbing postures, and selecting the grabbing postures of which the crossing risk meets a threshold value; superposing micro-random jitter on the path to suppress resonance phase locking; and abnormal events are monitored in real time during operation, and a risk domain is updated to execute hierarchical unloading control, so that the instantaneous impact and damage risk is reduced. According to the technical scheme, the probability of instantaneous impact risks caused by stacking overall micro-drift and grabbing track overlapping can be effectively reduced, and therefore the safety of the system is guaranteed.

The invention relates to a port crude oil unloading speed regulation and control prediction method based on physical constraints, and belongs to the technical field of crude oil storage and transportation automation control, and the method comprises the following steps: S1, constructing a coupling equation of pipeline outlet pressure from an oil transfer arm to a reservoir area and storage tank flow speed; s2, constructing a physical constraint neural network model; s3, constructing a loss function formula of the physical constraint neural network model based on the coupling equation; and S4, performing adaptive training on the physical constraint neural network model based on historical data, and updating network parameters of the model. According to the method, the physical law of pipeline fluid movement is fused into the loss function in the form of constraint conditions in the model training stage, the degree of dependence of the model on the pure data sample size is remarkably reduced, and accurate regulation and control of the oil discharge rate under the complex working condition are achieved.

The invention belongs to the technical field of uncertainty probability characterization analysis, and discloses a thermoplastic composite material leading edge structure design allowable value probability characterization method and system under a small sample condition, and the method comprises the steps: defining a plurality of candidate probability distribution models; fitting each model based on the original sample data and calculating an AIC value and a BIC value; a dynamic weight factor alpha is calculated according to the sample size n, and then a hybrid information criterion HIC value is calculated; generating a plurality of sample sets through Bootstrap self-service sampling, recalculating the HIC value on each sample set, and counting the selected optimal frequency of each model; and determining an optimal probability distribution model according to the frequency, wherein the optimal probability distribution model is used for representing a design allowable value. According to the method, the dynamic weight factor alpha is introduced, AIC and BIC criteria are effectively unified, optimal balance between prediction precision and model complexity is achieved under the condition of small samples, and engineering practicability and robustness are remarkably improved.

The invention discloses a cognitive function screening system and method based on an MMSE (Minimum Mean Square Error) prediction model, and relates to the technical field of data analysis, the method comprises the following steps: collecting physical examination index data, removing missing records, adopting a multiple interpolation method for interpolation, and obtaining multiple sets of complete data sets; a continuous prediction model is constructed, MMSE continuous prediction values are obtained, and a sensitivity analysis report is generated; constructing a first-stage classification model, adaptively dividing an optimal threshold combination, and dividing a sample into a high-confidence region, a to-be-discriminated region and a low-confidence region; if the sample size of the to-be-discriminated region is higher than a preset training threshold value, constructing an enhanced feature set, and constructing a second-stage classification model; when the prediction probability reaches the optimal re-discrimination threshold value, the classification result in the first stage is corrected, and otherwise, the classification result is maintained; if not, maintaining the classification result; and integrating the classification results to obtain a final classification result of all the samples.

The invention discloses a method for judging and optimizing a raw material ratio in microwave synergistic pyrolysis, and relates to the technical field of solid waste recycling and biomass energy utilization. In order to overcome the defect that in-depth research and engineering application of a microwave pyrolysis process are restricted in the prior art, the technical scheme provided by the invention comprises the following steps: preparing samples from oily sludge and palm kernel shells according to a preset mass ratio, and mixing to obtain a mixed raw material; establishing an inert atmosphere and forming a controlled reaction system; applying microwave heating and collecting a sample mass change curve and temperature change data; comparing the mass change curve with a theoretical curve formed by weighted stacking of a single raw material pyrolysis curve to obtain a difference result; judging the synergistic effect strength among the raw materials according to the difference result, and outputting a synergistic effect judgment result; and determining an optimal ratio based on a synergistic effect judgment result. The method is suitable for raw material ratio judgment and optimization work of solid waste and biomass under the microwave synergistic pyrolysis condition.

The invention relates to a light high-entropy alloy yield strength machine learning prediction method, which comprises the following steps: S1, obtaining an original data set of a light high-entropy alloy sample, the original data set comprising the yield strength of the light high-entropy alloy sample and corresponding experimental process conditions; s2, extracting feature parameters for describing samples based on the original data set; s3, dividing the original data set into a training set and a test set; s4, screening an optimal feature subset from the feature parameters, and calculating a secondary feature set based on the optimal feature subset; and S5, taking the yield strength of the light high-entropy alloy sample as a target variable, taking the secondary feature set as an independent variable, and adopting XGBoost trained by the original data set to construct a quantitative prediction model of the yield strength of the light high-entropy alloy. Compared with the prior art, the method has the advantages that the physical interpretability is enhanced, the generalization ability is improved, the dependence on the sample size is reduced, and the like.

The invention provides a laying hen feed raw material sample screening method and system based on multi-source variability, and relates to the technical field of data processing analysis, and the method comprises the steps: obtaining multi-source attribute data and conventional component content data of raw materials, mapping the attribute data into tensor modal dimensions through multi-source variability tensor construction processing, and obtaining a multi-source variation tensor model; the component data is used as a characteristic component, and a multi-source variability tensor is obtained through decoupling and compression. Variability spectrum decomposition processing is carried out, local rank spectrum decomposition is carried out along producing areas, time and component dimensions, and a variability spectrum vector set is obtained; and identifying a candidate modeling sample set through multi-scale extremum and sparsity screening. And evaluating the contribution degree and sensitivity of the sample to a standard ileum amino acid digestibility prediction equation through a leave-one-out method and sensitivity analysis, and screening out an optimal modeling sample. According to the method, the multi-source variation information of the raw materials can be integrated, the variation spectrum is comprehensively covered with the minimum sample size, and the precision and generalization ability of the prediction model are remarkably improved.

This invention relates to the field of nuclear safety equipment testing technology. It provides a method for optimizing parameters in accelerated reliability testing of nuclear safety-grade valves. The method involves the following steps: establishing a correlation model between accelerated stress and life parameters; calculating the acceleration factor and formulating an initial test plan; deriving the theoretical sample size based on the success-failure test method combined with confidence requirements; constructing an extended coefficient table using the range-extended test method to optimize test parameters; generating multiple accelerated test plans through chi-square distribution verification; eliminating plans that do not meet the life characteristic threshold; determining the optimal sample size, number of failures, and test cycle through comprehensive test cost and risk analysis; and finally outputting a complete test plan including the action interval time. This method solves the problems of limited test samples, insufficient failure data, low confidence levels, and difficulty in adapting accelerated models in existing high-reliability valve technologies.

The technical scheme of the invention discloses a method for judging weights of associated elements of process parameter data and quality data. The invention provides a method for analyzing relevance between process parameter data and quality data and endowing the process parameter data with weights. Aiming at the current situations that the industrial quality data acquisition sample size is small and part of values are easily identified as abnormal values when a traditional method similar to standard deviation is used, the method adopts a grouping comparison strategy when abnormal data are removed in data preprocessing, and effectively retains part of quality characteristics with large deviation. And a small amount of data still having dispute after deletion and selection is subjected to artificial expert review, so that the quality characteristics are further reserved, and the dependence on artificial deletion and selection is reduced. According to the method disclosed by the invention, in order to solve the problems that a model generated by machine learning is relatively poor in interpretability and relatively few in verification, an SHAP algorithm is adopted to carry out association construction on the model, process parameter weights are output, meanwhile, an output result is verified, and the correctness of a weight conclusion is ensured.

This application relates to the field of equipment performance testing and evaluation technology, and provides a method and system for equivalent reduction of equipment test sample size. By constructing a data fusion evaluation model based on actual and numerical test data, the fidelity of actual test response estimation under small sample conditions is improved. An equivalent measure is designed and calculated by fusing the mean square error of the evaluation model from a single data source with a preset test confidence level, and a box plot of the equivalent measure from a single data source is obtained. By connecting the lower quartile and upper quartile of each obtained box plot, equivalent composite confidence intervals for actual and numerical simulation tests are plotted. Based on these equivalent composite confidence intervals, the theoretically required actual test sample size is equivalently reduced to the numerical simulation test sample size, achieving an equivalent reduction of the actual test sample size by the numerical simulation test, thus solving the problem of equipment performance evaluation under insufficient actual test samples.

The utility model discloses an integrated urine specimen collector, including the outer layer urine storage collection pipe, the inside of outer layer urine storage collection pipe is provided with the inner layer urine collector, the opening position of outer layer urine storage collection pipe is connected with the screw cap, is used for carrying out effective sealing to outer layer urine storage collection pipe, still include: the left side fixed connection of inner layer urine collector inside has fixed cylinder. This integrated urine specimen collector adopts two-piece design, and the inner layer pullable urine collection core has self -adaptation caliber function, and the outer layer is equipped with screw cap, and the inner layer is equipped with stretchable folding handle, and the bottom 8ml storage pipe one side has three urine leakage holes, and the other side is no hole, can promote collection efficiency, ensure that the specimen is safe, simplify the operation and accurate control sample amount, and when need to inspect urine, only need to unscrew the upper layer's inspection cover, and the probe is inserted from the hollow place of screw cap and can carry out the inspection, thereby make the inspection operation more simple, accurate.

The invention discloses a data processing method and device, electronic equipment and a storage medium, and relates to the technical field of computers. A specific embodiment of the method comprises the following steps: receiving an experiment trigger instruction, and obtaining corresponding historical sample data to determine experiment sample data and contrast sample data; querying an early stop index type, determining an early stop index based on the early stop index type, and calculating a corresponding early stop threshold value; executing an experimental operation program on the experimental sample data to obtain an experimental result, and calculating a first conversion quantity and a second conversion quantity of the early stop index in the experimental result and the contrast sample data according to the early stop threshold value to determine a change parameter of the early stop index; and in response to the change parameter meeting the early stop condition, stopping the experimental operation program. According to the implementation mode, the problem that time and resource waste is easily caused by a row mode that the experiment result is analyzed only after the execution of the preset sample size is finished or the experiment is executed for the preset time period in the experiment can be solved.

The invention discloses a sample sampling method. The method comprises the following steps: acquiring a first step number when a pipettor triggers an air suction alarm signal; comparing the first step number with the second step number and the third step number; treating the blood film or / and the bubbles by adopting different methods according to the comparison result of the descending steps; and carrying out sampling operation on the sample subjected to blood membrane or / and bubble treatment. The first step number of descending of the pipettor can be judged according to the air suction alarm signal when the blood membranes or / and the bubbles appear, and the blood membranes or / and the bubbles appear at different positions and correspond to different sample sizes in the test tubes, so that the first step number is compared with the second step number and the third step number, and the blood membranes or / and the bubbles in the test tubes are judged according to the comparison result. Different treatment methods are adopted for blood membranes or / and bubbles at different positions, so that the blood membranes or / and bubbles in the sample are automatically treated, the invalid air suction phenomenon is avoided, and the sampling precision of the sample is improved.

The embodiment of the invention provides a sample generation method for group product fault detection and diagnosis combined verification, and the method comprises the steps: screening initial samples stored in a database according to the project demands of a target project, obtaining a first intermediate sample related to the target item; when it is determined that the sample size of the first intermediate sample is insufficient, a second intermediate sample is generated through interpolation; determining fault modes corresponding to each first intermediate sample and each second intermediate sample, and extracting the fault modes meeting a consistency index in the fault modes to obtain a to-be-verified fault mode set; determining a to-be-verified sample corresponding to each fault mode in the to-be-verified fault mode set, and performing sufficiency test on the to-be-verified samples; and according to a sufficiency test result, determining a target sample related to the target item in the to-be-verified samples.

This invention belongs to the field of resource allocation and discloses an emergency resource demand prediction method and apparatus based on WGAN-Stacking. The method includes: acquiring meteorological, geographical, and power grid data; generating fault samples using a Wasserstein generative adversarial network; quantifying the distribution difference between the generated fault samples and real samples based on FID distance and kernel Inception distance to screen target samples whose sample quality reaches a preset threshold, with the quantified distribution difference used to characterize the sample quality of the fault samples; inputting the target samples into a pre-constructed resource demand prediction model to predict the emergency resource demand in different regions; the resource demand prediction model adopts a Stacking ensemble learning framework, using RF, GBDT, and SVM as heterogeneous base learners and FNN as a meta-learner, learning the optimal nonlinear combination strategy between the prediction results of each base learner; and allocating emergency resources according to the prediction results output by the resource demand prediction model.

The application belongs to the technical field of sensor preparation, and particularly relates to an integrated interstitial fluid sampling and electrochemical detection system and a preparation method and application thereof. The integrated interstitial fluid sampling and electrochemical detection system comprises a microneedle array unit, an electrochemical sensor and a micro negative pressure pump. The microneedle array unit has good biocompatibility, is safe and reliable, and will not damage blood vessels and nerves of a patient; the microneedle is short and small, has weak discomfort, does not affect normal activities of a user, and is more conducive to continuous long-term monitoring; the electrochemical sensor can accurately identify a template molecule and transmit a signal; and the micro negative pressure pump can generate vacuum pressure to collect a sample amount meeting detection. The integrated interstitial fluid sampling and electrochemical detection system can be used by being inserted into a subcutaneous or intradermal position of a human body, can realize real-time and integrated sampling-detection, and can detect various biomarkers.

The invention belongs to the technical field of underwater sound target recognition and machine learning, and discloses a small sample ship voiceprint data enhancement and high-precision recognition method, which comprises the following steps of: dividing an effective audio slice into a small sample data set, a large sample data set and an overall data set based on sample sizes of various categories; performing short-time Fourier transform and Mel filtering processing on the effective audio slices to generate a logarithmic Mel spectrogram with a fixed size; constructing and training a convolutional variational auto-encoder to form a reconstructed small sample data set; and combining the small sample data set, the reconstructed small sample data set and the large sample data set, carrying out feature extraction on the combined data set by adopting a BEATs encoder of a frozen parameter to obtain a voiceprint feature vector, training and calling a Light GBM classifier to carry out classification identification, and outputting an identification result of a ship category. According to the method, the overfitting problem of the classifier is remarkably relieved, and the precision and stability of ship voiceprint data recognition can be improved.

The application discloses a mass flow calibration method and system, comprising: connecting a Coriolis mass flow meter to a calibration bench pipeline, and installing an intelligent sensor containing an ultrasonic Doppler array on the upstream side in the axial direction; starting a liquid supply pump to pump calibration fluid, controlling the ultrasonic Doppler array to perform cross-section scanning on the fluid to obtain dynamic flow profile time series data; inputting the dynamic flow profile time series data into a pre-trained fluid relaxation state prediction machine learning model, outputting a steady state arrival time point from the fluid relaxation state prediction machine learning model; triggering a calibration sampling instruction at the steady state arrival time point, obtaining a sampling mass flow value and comparing it with a reference standard flow value to complete calibration. The application solves the problem of systematic error in calibration sampling caused by non-steady state changes in flow pattern during the shear thinning recovery period of high viscosity fluid.

The application provides a liquid separation system, which belongs to the technical field of medical devices. The liquid separation system comprises a first conveying device, a second conveying device, a sub-tube transfer device, a pipetting device and a controller. The first conveying device is provided with a first conveying part, which is used for transferring a mother sample module into the first conveying device or transferring the mother sample module in the first conveying device out. The second conveying device is provided with a second conveying part, which is used for transferring a carrier into the second conveying device or transferring a sub-sample module from the second conveying device. The second conveying device has a sub-tube loading position and a liquid discharging position. The first conveying device has a liquid suction position. The controller is in communication connection with at least the first conveying part, the second conveying part, the sub-tube transfer device and the pipetting device. The liquid separation system provided by the application realizes two independent conveying devices, avoids track blockage caused by the increase of total sample amount after liquid separation, and ensures the liquid separation efficiency and smoothness.

A training method of a marketing prediction model for eliminating time deviation and a precision marketing method relate to the technical field of data processing, samples are grouped by setting time stamps and marketing scheme stamps, experimental samples and contrast samples in the same time group are ensured to be comparable in average value, and the accuracy of marketing prediction is improved. Therefore, the time deviation problem when historical samples are adopted is eliminated, meanwhile, purchase rate target pre-training is carried out on the model to solve the problem that high-latitude features cannot be effectively utilized due to the fact that the unbiased sample size is small, and finally the overall loss is calculated in a multi-task training mode, so that the accuracy of the model is improved. The marketing prediction model can achieve better prediction accuracy under the condition that biased samples and a small number of unbiased samples are adopted, so that the sampling cost caused by only using the unbiased samples is greatly reduced, and both the cost and the effect are achieved. By using the marketing prediction model, accurate putting of marketing schemes can be further realized, and the marketing prediction model has a good practical value.

The application discloses a complex long-acting preparation steady-state pharmacokinetics extrapolation equivalence prediction method, relates to the steady-state bioequivalence technical field, and comprises the following steps: collecting and unifying concentration-time, a drug administration scheme and a covariate, generating a standardized time, an event marker code and a covariate dictionary, setting a deletion weight and a population alignment weight; setting a candidate structure based on population pharmacokinetics, estimating a model and a covariate effect through a quality threshold determination with a weight; constructing a virtual population and performing multiple drug administration simulation to determine the reaching of stability according to the relative change of consecutive two trough concentrations, generating a peak-trough sampling time window; performing non-compartment analysis in the steady-state interval, calculating a dosing interval area, a steady-state peak value and a steady-state trough value, calculating a geometric mean ratio and a confidence interval of two preparations, carrying out efficacy-sample size linkage, and outputting a recommended design; the method can reduce the test burden, shorten the cycle and improve the decision transparency.

The invention discloses a multi-coal-source coal blending data processing method based on a WSMOTE algorithm, and belongs to the technical field of coal processing data processing, and the method comprises the steps: data collection and classification: collecting multi-dimensional data of a raw coal floating and sinking test, a coal blending test and the like, carrying out the preprocessing, dividing the data into a missing data set and a complete data set, carrying out the interpolation optimization of a missing value, and carrying out the calculation of the complete data set. And carrying out adaptive processing according to data distribution types, or carrying out weighted combination normalization according to feature importance, analyzing the number and distribution characteristics of minority class samples, dynamically adjusting the neighbor sample size and the new sample synthesis amount, and completing data enhancement and standardization. According to the method, standardized parameters are dynamically adjusted to adapt to data distribution changes, abnormal values are accurately processed, the combined interpolation model is subjected to weight optimization, errors are reduced, and data integrity is improved. WSMOTE parameters are adaptively combined with data distribution to synthesize samples, data are balanced, and the over-fitting risk is reduced. In practical application, the quality of coal blending model training data can be improved, and optimization of a multi-coal-source coal blending scheme is assisted.

This invention provides a method for predicting the remaining service life of capacitive current transformers under small sample conditions. The method includes the following steps: collecting capacitor aging data and capacitive current transformer detection data; extracting features from the capacitor aging data and remaining service life labels; constructing a wide-range learning model for training; and inputting the detection data features and remaining service life labels into the model for testing. The refined model is then tested and applied to actual power equipment, thereby providing intelligent operation and maintenance support for the power system. This method can effectively predict the remaining service life of capacitive current transformers using a wide-range learning model with a small sample size, overcoming the shortcomings of traditional methods that rely on large-scale datasets for training. Furthermore, the model can be effectively fine-tuned using a small amount of historical data, reducing the "cold start" problem and improving its application efficiency in actual power equipment.

The invention provides a new energy output curve generation method and system, and relates to the technical field of new energy power generation and power system planning. The method comprises the steps of constructing a new energy output characteristic evaluation system, based on historical new energy data, adopting a Markov chain to divide M time state intervals, adopting a Monte Carlo method to divide R longitudinal output intervals, and extracting data to generate an initial curve; and after correction in a wet season and a dry season, performing cyclic screening according to a set constraint condition, and outputting a curve conforming to a multi-time-dimension feature and an extreme scene. The system corresponds to the method. By the adoption of the method, the annual scale curve can be generated only through historical data without depending on meteorological data or massive historical samples, the problems that in the prior art, a scheduling method is not matched in time scale and insufficient in historical sample size are solved, and a reliable data boundary is provided for new energy installation planning and energy storage facility design.

The invention belongs to the technical field of power system automation, and discloses a power grid operation data sample quantitative grading method and system based on a scene method.The method comprises the steps that firstly, power grid steady-state operation risk evaluation indexes and corresponding severity functions are obtained, and a unified risk quantitative standard is established; secondly, determining a comprehensive risk grading label and a sample comprehensive index by combining the steady-state operation data of the power grid, and calibrating a risk grading threshold value; and finally, a multi-dimensional operation risk index calculation system is established, comprehensive indexes of different dimensions are obtained through calculation of the indexes and weight coefficients of all the dimensions, and quantitative classification of the samples is completed in combination with a classification standard and a classification threshold. The method breaks through traditional evaluation dimension limitation, gets rid of probability analysis dependence, determines risk severity, can provide accurately quantified multi-scene samples for power grid regulation and control operation test verification, adapts to continuous time period advanced scheduling demands, improves accuracy and robustness of a regulation and control technology, and guarantees safe and stable operation of a novel power system.

This invention discloses a method for identifying fresh / dried cocoon raw silk based on ultraviolet absorption spectroscopy. The method includes: pre-treating and extracting raw silk samples from fresh and dried cocoons to obtain a test solution; collecting the ultraviolet absorption spectra of the test solution and ultrapure water, and obtaining a corrected ultraviolet absorption spectrum after background signal correction; pre-processing the corrected ultraviolet absorption spectrum to obtain a pre-processed spectrum and its derivative spectrum; performing difference analysis to screen difference scoring indicators, and then establishing a discrimination model for fresh and dried cocoons; obtaining the pre-processed spectrum and its derivative spectrum of the raw silk sample to be identified, inputting them into the discrimination model for processing, and outputting the category to complete the identification of fresh and dried cocoon raw silk. This invention is convenient to operate, uses relatively common instruments, has high accuracy, and the accuracy can be further improved with increasing sample size, making it easy to promote and apply, and meeting the needs of identifying fresh / dried cocoon raw silk.

The invention discloses a sparse sample entropy criterion-based physical characteristic parameter estimation method, which comprises the following steps of: determining a pneumatic system fault factor and a physical constraint thereof, and converting the physical constraint into an integral form; solving an integral form based on a Lagrange direct construction method to obtain a probability density function, and ensuring that the probability density function has physical significance; screening target distribution conforming to physical constraints from a preset general probability density function library, and extracting functions conforming to basic distribution shape requirements in the target distribution to form a function group; dividing the function group into a single-parameter group and a double-parameter group based on the parameter scale, respectively calculating to obtain corresponding entropy information, sorting differential entropy results, and taking a maximum value to obtain most unbiased probability density distribution conforming to the physical constraint. According to the method, high-confidence parameter estimation under the condition that the sample size is scarce is realized, the limitation of data shortage is broken through, and the probability density function can be stably generated only by utilizing physical characteristics such as extreme values and peak positions.

The application relates to the technical field of laboratory automation management and data processing, in particular to a detection consumable allocation method and system based on comprehensive efficiency optimization and a storage medium, a predicted sample quantity in a future early warning period is output by calling a sample quantity prediction model, and a single-specification reservation coefficient is dynamically determined accordingly, so that safe inventory control considering current task execution and future demand shortage prevention is realized; on the basis, a plurality of candidate allocation schemes are generated within the limitation, multi-dimensional efficiency calculation is carried out in combination with a single-specification unit cost, a to-be-detected sample quantity, a fixed control site quantity and real-time residual effective time, a consumable cost performance, actual utilization rate and inventory turnover efficiency obtained are fused into a comprehensive efficiency value for final decision, the limitation of traditional single-dimensional extensive allocation is avoided, on the premise of meeting the absolute well site demand of single detection, the overall high procurement cost is reduced, well site vacancy waste is reduced and near-efficiency period consumables are accelerated.

The present application belongs to the technical field of underwater acoustic target recognition and machine learning, and discloses a small sample ship acoustic print data enhancement and high-precision recognition method. The method divides effective audio slices into small sample data sets, large sample data sets and overall data sets based on the sample amount of each category. Then, the effective audio slices are subjected to short-time Fourier transform and Mel filtering processing to generate fixed-size log Mel spectrograms. Next, a convolutional variational autoencoder is constructed and trained to form a reconstructed small sample data set. After that, the small sample data set, the reconstructed small sample data set and the large sample data set are combined, and the merged data set is subjected to feature extraction by a BEATs encoder with frozen parameters to obtain an acoustic print feature vector. A LightGBM classifier is trained and called for classification and recognition to output the recognition result of the ship category. The method significantly alleviates the overfitting problem of the classifier and can improve the precision and stability of ship acoustic print data recognition.