201 results about "Sample quality" patented technology

The invention belongs to the technical field of electrical equipment detection, and discloses a defect image enhancement method fusing reasoning and generation, which integrates visible light, infrared and laser radar data through a multi-modal feature fusion network, breaks through the limitation that a contrast file CN114281093A only depends on a visible light image, and improves the detection accuracy. The dynamic attention mechanism can flexibly deploy visual, spatial and semantic feature weights according to defect types, key features can still be captured in complex environments such as strong light and shielding, and meanwhile, the spatial form of the defects is analyzed by means of three-dimensional point cloud; by means of the design, missing detection caused by insufficient characteristics of tiny parts such as hardware fittings and pins is effectively avoided. Aiming at the problem of distortion of a sample generated by a traditional data enhancement method in a comparison file, the sample quality is guaranteed through double mechanisms of reasoning constraint and physical verification, defect features output by a reasoning model directly constrain feature distribution of the generated sample, and meanwhile, a material mechanics rule is introduced to verify the physical rationality of the generated sample.

The invention relates to the field of computer vision algorithms, in particular to a defect detection data screening method based on a Pontryagin maximum principle, which comprises the following steps of: training a training data set and testing an evaluation data set to obtain a model reference index; based on a preset proxy data set, respectively calculating definition, labeling integrity and distribution deviation degree indexes to obtain an initial quality weight vector; iteratively updating the basic model parameters and reversely iteratively updating the target vector to obtain a sample quality score; a training scoring device scores and sorts full samples of the training data set; and dividing the sorted full samples into a plurality of candidate screening intervals, and screening high-quality data to train a final model. According to the method, the entropy weight method is adopted to weight the three dimensions to obtain the initial quality weight, so that the initial weight of the sample can reflect the own basic quality difference, the problem that the traditional uniform weight ignores the sample quality difference is avoided, and the accuracy of the sample quality score is further improved.

The invention discloses a tea review expert decision support method and system based on a knowledge graph, and relates to the technical field of tea quality evaluation, and the method comprises the steps: collecting tea sample data and expert review data, calculating an expert score consistency index, and constructing an expert decision knowledge graph; analyzing an expert score distribution rule to obtain a rule mapping weight, mapping the standardized features to a knowledge graph and generating a feature transmission path; correcting a transmission result according to the rule adaptability and the conflict degree to obtain a tea sample quality score and a feature analysis chart; and finally, converting the feature analysis graph into a process control rule sequence, determining an execution priority based on the quality score, and generating a process parameter adjustment scheme. According to the invention, the intelligent optimization of the tea production process is realized, and the stability and consistency of the tea quality are improved.

The invention provides a steel plate surface defect image generation method based on a de-noising diffusion probability model and a visual Transform, relates to the technical field of industrial surface defect detection, and aims to solve the problems of low quality of generated samples, lack of detail richness and insufficient global dependency relationship modeling capability in the conventional defect image generation method. The performance improvement of a defect detection algorithm is limited; according to the generation method provided by the invention, by introducing the ViT model, the capability of capturing the global dependency relationship in the image by the model is remarkably enhanced, the limitation of the traditional DDPM in modeling the global structure of the image is overcome, the diversity and quality of the generated image are effectively improved, and the method is suitable for further improving the modeling precision of a complex defect structure. A channel-space self-attention module is designed, and the CSSAM of the module can dynamically adjust the attention between different space regions and channels, so that the details of complex defects are finely simulated, and the authenticity and high fidelity of a generated image on the aspects of a global structure and local textures are ensured.

Diffusion models are machine learning algorithms implemented as neural network-based denoisers that are uniquely trained to generate high-quality data from an input lower-quality data. However, for complex datasets, the samples generated by a diffusion model can still fail to reproduce the quality and diversity of the training data, due to approximation errors made by the finite-capacity network. Diffusion guidance addresses this issue by steering the sampling process away from a less desired model, toward a preferred one. However, current guidance methods rely on a single additional denoiser and manual tuning of guidance weights, which is suboptimal for large, complex models and leads to inefficiencies. The present disclosure employs a mixture of denoisers to guide a diffusion model, which can increase the expressiveness of the guidance and substantially improve sample quality and diversity in the output of the diffusion model.

The invention relates to a unified system for the extraction and purification of biomolecules using a novel Cloudy-Precipitated (CP) Buffer and an integrated extraction device. The CP Solution remains cloudy even when heated and enables single- or multi-step workflows for isolating nucleic acids without requiring multiple buffers or organic solvents. It contains optimized concentrations of detergents, chaotropic agents, salts, and adsorbents to efficiently lyse cells and minimize processing steps. The associated device incorporates a heating and filtration system that allows separation of target molecules from inhibitors or impurities based on molecular weight. It can operate as a single- or double-step configuration with various filter membrane options. This system reduces time, labor, and environmental impact while enhancing sample quality for molecular diagnostics, research, and therapeutic applications.

The invention discloses a gene sequencing data management method for whole genome methylation sequencing, and relates to the technical field of gene sequencing data management.The method comprises the steps that original gene sequencing data are obtained and preprocessed, and a regional methylation matrix is generated; calculating a technical difference measure and a phenotype difference measure of each region unit based on the region methylation matrix and the sample metadata, identifying a technical methylation bias domain, and grading the technical methylation bias domain into a steady-state technical methylation bias domain and an event-driven technical methylation bias domain according to the occurrence frequency of the technical methylation bias domain in multiple batches; based on a linear adjacency relation and a co-bias relation, constructing a technical methylation bias super-domain and classifying the technical methylation bias super-domain; performing regional shielding on the methylation bias super-domain of the steady state technology, performing normalization correction on the methylation bias super-domain of the event-driven technology, and generating a corrected regional methylation matrix; calculating a sample quality index based on the corrected matrix and performing quality grading; according to the method, technical bias can be identified and corrected, and the methylation data quality and analysis reliability are improved.

The invention provides an image data set construction method and system based on multistage collaboration and a model training method. The image data set construction method comprises the steps of obtaining an initial image pair set; respectively estimating the average depth of pixel points corresponding to foreground objects in the source image and the target image and the consistency of backgrounds; based on the average depth and background consistency result obtained through estimation, screening image pairs meeting a depth change condition and a background consistency condition from the initial image pair set to form a geometric-level image pair set; based on a large visual language model, screening out image pairs with perceptible depth differences of foreground objects from the geometric-level image pair set to form a semantic-level image pair set, and generating natural language texts corresponding to the screened image pairs; and determining a depth difference quantization label of the image pair based on the average depth, and taking the natural language text as a depth difference text label, thereby obtaining an image data set. The method can ensure large-scale production and improve sample quality.

The present application relates to the field of electric digital data processing, and particularly relates to a joint training method and system based on real data and synthetic data, the method comprising: obtaining a mixed sample set composed of real data and synthetic data; constructing a global correlation graph based on the mixed sample set, extracting global authenticity and global closeness, and adaptively determining an optimal number of neighbors for synthetic sample quality evaluation; for each synthetic sample, constructing a neighbor subgraph by K-neighbor method using the optimal number of neighbors, extracting subgraph authenticity and subgraph closeness, and obtaining a quality score by a fusion algorithm; giving a fixed weight to real samples and a corresponding quality score to synthetic samples as a weight, constructing a weighted loss function based on sample weights, and completing joint training of the model. The present application improves the model training effect through adaptive number of neighbors, multi-dimensional quality evaluation and weighted differentiated training.

Provided is a moisture meter including a mass measurement unit for measuring the mass of a specimen, a heater for heating the specimen, a control arithmetic unit configured to control the heating unit to heat the specimen until a change in the mass of the specimen becomes not more than a predetermined threshold and calculate the moisture content of the specimen, a timepiece for measuring a heating time, and a storage unit. The control arithmetic unit measures the moisture content of a standard substance for inspection and a heating time, diagnoses the presence / absence of an abnormality in the heating unit by comparing a measured heating time with a predetermined time as a set heating time stored in the storage unit in advance.

The application discloses a zero-shot station behavior recognition method based on dynamic and static cooperative support set optimization, and aims to solve the problems of low sample quality and insufficient support set information mining in a video behavior recognition task. The method realizes zero-shot behavior recognition in a station scene through a key frame cascaded support set construction module and a dynamic and static cooperative weighted prediction module. The key frame cascaded support set construction module and the dynamic and static cooperative weighted prediction module constitute a zero-shot learning framework. In the key frame cascaded support set construction module, a text-to-image and image-to-video generation model is used to generate static key frame images and dynamic video support set samples for each behavior category. In the dynamic and static cooperative weighted prediction module, rich semantic information in the above samples is fully utilized, and dynamic fusion is performed on the static key frame and the dynamic video through a dynamic and static cooperative strategy, so that high-precision identification of key actions in the station scene is realized.

The invention relates to the technical field of medical equipment, and particularly discloses a support-membrane-free single-hole carrying net used in a biomedical electron microscope ultrathin section technology. A support-membrane-free single-hole carrying net used in a biomedical electron microscope ultrathin slicing technology comprises a single-hole carrying net body, and a bonding layer is arranged on the single-hole carrying net body; the raw materials of the bonding layer comprise chloroprene rubber and toluene. The chloroprene rubber has good initial viscosity, electron beam stability and solubility, the single-hole carrying net body is coated with the chloroprene rubber, the single-hole carrying net with high adhesive force and without the supporting film is obtained, in this way, slice fishing can be well conducted, the quality and thickness of a sample are not increased, and an obtained image is also clear.

The invention discloses a fault sample generation method based on oil and gas field seismic data, and belongs to geophysical exploration and machine learning data engineering.The technical scheme includes the steps that seismic exploration data, well data and actual geological data of an actual work area are obtained; establishing an initial structure model based on seismic exploration data, well data and actual geological data; iteratively adjusting the initial structure model according to a forward modeling result to obtain a calibration structure model; performing process simulation based on the calibration structure model, and outputting a plurality of geologic bodies according to time steps; and converting the geologic body into a forward modeling parameter model, carrying out seismic forward modeling to generate a seismic data body, and carrying out data enhancement and sample amplification. The method has the advantages that compared with other sample generation methods, the generated samples better conform to the actual geological law, so that the sample quality is improved, and meanwhile, the number of the samples can be greatly increased by simulating the forward modeling process of the actual body.

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 invention relates to resource exploration and environment monitoring, and particularly discloses a device and a method for collecting bubble gas in geothermal water, and the device comprises a gas collection probe, a sample tank and an integrated control host. The gas collection probe is immersed in a water body, and a hydrophobic gas-permeable membrane assembly in the gas collection probe can selectively permeate and enrich bubble gas. The integrated control host comprises a control unit, a power unit and a sensing unit, the power unit drives gas to directionally transfer between the gas collection probe and the sample tank along a preset gas path, and the sensing unit acquires key parameters in the collection process in real time. And the control unit regulates and controls the power unit to operate according to the parameters, so that gas collection, treatment and storage circulation are automatically executed. According to the device, through cooperation of the structures, automation and intelligentization of the gas collection process are achieved, and the sampling efficiency, the operation standardization degree and the consistency of the sample quality are effectively improved.

The invention discloses a digestive tract tumor biopsy sampling device convenient for puncture and sampling for medical oncology, which relates to the technical field of tumor biopsy sampling and comprises a biopsy device, a limiting seat and a base. According to the alimentary canal tumor biopsy sampling device convenient to puncture and sample for the medical oncology, tumor tissue can be accurately contained through the sampling groove, fragmentation of a tumor tissue sample is avoided, it is guaranteed that the tissue sample meets the tissue form needed by detection, the sample quality is guaranteed, the outer sleeve needle covers the sampling groove after sampling, the sample can be fixed and cannot fall off, and the sampling efficiency is improved. The tissue is prevented from remaining in the body or being polluted during withdrawing, and meanwhile the risks of bleeding and tissue damage in the puncture process are reduced; enough samples can be obtained through one-time operation, the repeated puncture frequency is reduced, the operation time is shortened, the pain of a patient is relieved, and the operation efficiency is improved; and the modularized puncture needle design allows medical staff to quickly adjust the puncture depth according to individual differences, and the flexibility is high.

The application belongs to the technical field of power equipment detection, and discloses a defect image enhancement method fusing reasoning and generation, which integrates visible light, infrared and laser radar data through a multi-modal feature fusion network, breaks through the limitation of relying only on visible light images in the contrast file CN114281093A, and dynamically adjusts the weights of visual, spatial and semantic features according to the defect type, so that key features can still be captured in complex environments such as strong light and shielding, and the spatial form of the defect is analyzed with the help of three-dimensional point cloud; this design effectively avoids the missed detection of small components such as hardware pins due to insufficient features; in view of the problem that the generated samples are distorted in the traditional data enhancement method in the contrast file, the sample quality is guaranteed through a double mechanism of "reasoning constraint + physical verification", the defect features output by the reasoning model directly constrain the feature distribution of the generated samples, and the physical rationality of the generated samples is verified by introducing the law of material mechanics.

The invention provides a preparation method of an EBSD sample of an aluminum-plastic composite film, the aluminum-plastic composite film comprises an aluminum foil and polymer films coated on two side surfaces of the aluminum foil, the preparation method of the EBSD sample of the aluminum-plastic composite film comprises the following steps: providing the aluminum-plastic composite film with the width of less than or equal to 10mm and the length of less than or equal to 15mm; taking the aluminum-plastic composite film as a positive electrode and a lead plate as a negative electrode, and performing first electrolysis to obtain a first sample; performing second electrolysis on the first sample in the swinging state to peel off the polymer film to obtain a second sample; and carrying out third electrolysis on the second sample to obtain an EBSD sample of the aluminum-plastic composite film. On the basis of an integrated sample preparation mode, plastic film stripping and aluminum foil surface optimization are achieved at the same time in the electrolysis process, the purposes of simplifying the aluminum plastic film aluminum foil EBSD sample preparation steps and improving the sample quality and the sample preparation efficiency are achieved, and therefore efficient EBSD analysis in material scientific research and industrial detection is remarkably improved.

This utility model provides an automated mineral sample dispensing and retention system, relating to the technical field of mineral sample dispensing and retention processes. It includes a conveying unit for conveying original sample boxes and wide-mouth sample bags respectively; a weighing unit for weighing the dispensed and retained samples; a placement unit for placing the dispensed and retained samples; and a moving unit for picking up the original sample boxes and wide-mouth sample bags from the conveying unit and sequentially moving them to the weighing and placement units. This system boasts a high degree of automation, enabling automated dispensing of mineral samples by number, retention of remaining samples, and traceability of remaining sample quality. It ensures the accuracy of mineral sample dispensing and retention information, and allows for fully automated batch processing, benefiting the occupational health and safety of workers, improving work efficiency and quality. It effectively overcomes the technical shortcomings of existing technologies that require manual operation at each stage, have low automation levels, and suffer from significant human error in sample management.

The present application relates to a kind of thermal analysis and material characterization technical field, it is a kind of material pyrolysis behavior simulation method, device, electronic equipment and storage medium based on chemical and element composition, including input target sample's basic data to composition-parameter mapping model, obtain the predicted parameter of each thermal decomposition stage, wherein composition-parameter mapping model is obtained using multiple samples for deep learning;Using the predicted parameter of each thermal decomposition stage generates the simulated thermogravimetric curve of target sample quality with temperature or time change, and numerical differentiation is carried out to simulated thermogravimetric curve, obtains simulated differential thermogravimetric curve.The present application can simulate to obtain thermogravimetric curve and differential thermogravimetric curve according to chemical composition, element composition and thermogravimetric experiment condition without real thermogravimetric experiment, break the original thermogravimetric curve for the limitation of obtaining dependence on thermogravimetric experiment, eliminate the cost and time required for thermogravimetric experiment, reduce the cost of thermogravimetric curve acquisition, improve acquisition efficiency.

The application discloses a reproductive secretion sampling device and relates to the technical field of medical diagnosis sampling, and aims to solve the technical problem that cervical canal secretion sampling is easily affected by external environment. The application solves the pain point of traditional sampling through multidimensional design, guarantees sample quality and operation convenience, and greatly reduces the pollution risk through the whole sealing structure, from the extension of the sampling cotton swab, sampling to falling into the container, and automatic sealing of the sealing cover. The fixed part and the loose card structure realize convenient installation and stable fixation of the sampling cotton swab, and prevent the sampling cotton swab from falling off during sampling. The multi-angle sealing design of the sealing cover prevents the storage liquid from pouring when the sealing cover is inclined. The adjustable pressure receiving strip is suitable for sampling containers with different diameters, and the universality is enhanced. The moisture-absorbing inner layer of the port cover and the elastic return structure reduce the interference of vaginal secretion, improve the sample purity, and overall meet the sterile, stable and adaptive clinical sampling requirements, and guarantee the accuracy of the detection result.

The invention discloses a landslide susceptibility evaluation method and system based on an optimized non-landslide sample selection strategy, and belongs to the field of geological disaster risk analysis. The method aims at solving the technical problem that the evaluation precision of the landslide susceptibility is not high due to unreasonable selection of traditional non-landslide samples. According to the technical scheme, the method comprises the steps that multi-source geographic information data and historical landslide data of a research area are collected and processed in a unified mode; extracting evaluation factors and constructing a susceptibility evaluation index system; performing susceptibility pre-partitioning based on an information quantity value method, and selecting a non-landslide sample in a low-susceptibility region; combining the landslide sample with the primarily selected non-landslide sample, eliminating a noise sample by adopting density-based spatial clustering analysis, and supplementing and generating a balanced optimized sample set by utilizing a synthetic minority oversampling technology; and inputting the optimized sample set into a support vector machine model for training and susceptibility evaluation drawing. By optimizing the sample selection strategy, the sample quality and the accuracy of the model evaluation result are improved.

The invention provides a method and a device for determining the content of free oil in a shale sample. Acquiring a shale sample of the target area, and performing fluid occurrence state recovery treatment on the shale sample to obtain a target shale sample; performing constant-temperature evaporation treatment on the target shale sample, and collecting sample quality data and nuclear magnetic resonance monitoring data at a plurality of time points in the constant-temperature evaporation process; constructing a time-porosity curve according to the sample quality data and the nuclear magnetic resonance monitoring data; determining a porosity interval corresponding to the free oil evaporation stage according to the time-porosity curve; first nuclear magnetic resonance monitoring data at a starting time point and second nuclear magnetic resonance monitoring data at an ending time point are extracted from the nuclear magnetic resonance monitoring data at the multiple time points, and the variation of the content of the clay adsorbed water is calculated; and determining the free oil content of the target shale sample according to the porosity interval and the variation of the clay adsorbed water content. Therefore, the accuracy and reliability of the determination result of the free oil content are improved.

The utility model discloses a convenient, reliable and economical carbon fiber multifilament manual gum dipping and sample preparation device, which relates to the technical field of gum dipping and sample preparation devices and comprises a sliding mechanism, the sliding mechanism comprises a guide rail, two sides of the guide rail are connected with sliding blocks in a sliding manner, the sliding blocks can be fixed at any position of the guide rail through locking knobs, and the sliding blocks are fixed on the guide rail through the locking knobs. A bottom plate is arranged at the top of the sliding block, and a fiber multifilament fixing mechanism is fixedly connected to the top of the bottom plate; the fiber multifilament fixing mechanism comprises two triangular support plates, the two triangular support plates are of a plane symmetry structure, the two triangular support plates are fixed to the bottom plate through bolts, three winding shafts are arranged at the front ends of the triangular support plates, and the rear portions of the triangular support plates are designed to be of an L-shaped structure. According to the utility model, the difficulty of manual sample preparation is reduced, the sample preparation quality is improved, and the rejection rate is greatly reduced, so that the sample preparation efficiency is higher.

The invention provides a training method of a gel quality parameter prediction model and a quality parameter prediction method, and the training method comprises the steps: obtaining a plurality of groups of first gel attribute information after preprocessing; each group of first gel attribute information comprises sample quality parameter information corresponding to sample gel and sample spectrum information under the corresponding sample quality parameter information; and taking the sample spectrum information of each group as the input of a preset model, and taking the corresponding sample quality parameter information as the output of the preset model, so as to train and obtain the gel quality parameter prediction model for predicting the target quality parameter information corresponding to any gel. The gel quality parameter prediction model trained through the method is high in fitting precision and excellent and stable in prediction capacity, the application requirement of near-infrared quantitative analysis is met, and rapid, lossless, stable and accurate detection of gel key quality attributes can be achieved.

The invention relates to a method for analyzing carbon content in a liquid biological sample and a carbon content element analyzer. The method comprises the following steps: unfreezing the biological sample to a liquid state; transferring the biological sample into a glove box; a tin cup is taken to be heated, and the tin cup is put into a glove box after being heated; a biological sample is taken and placed in a tin cup, and the mass of the biological sample in the tin cup is weighed; sealing the opening of the tin cup; the tin cup is moved to a sample disc of an element analyzer, the biological sample is subjected to high-temperature oxidation in the element analyzer to generate corresponding gas, the gas is dried and then subjected to carbon content analysis, the carbon content in the biological sample is obtained, and in the drying process, the mixture is adopted as a drying agent for drying. The drying tube is arranged in the elemental analyzer, the drying agent is optimized, the sample quality is controlled, the influence of moisture on carbon content analysis is reduced, the sample does not need pretreatment such as freeze-drying, and the sample pretreatment time is saved.

The invention relates to an on-line measuring device and a measuring method for sample abrasion loss under a single variable condition, an impact particle swarm generation unit generates particle swarms with different impact speeds, and a measuring box is used for placing a sample and is provided with a temperature adjusting unit, a gas guide unit and a measuring unit for on-line measurement of sample quality. And the controller controls and cooperates to complete the abrasion loss test of the high-temperature oxidation sample under the variable temperature condition and the corrosion atmosphere. According to the method, the erosion rate data is calculated more accurately by monitoring the sample quality in real time, the prediction logic of the damage period caused by erosion is obtained by mastering the relationship between particle erosion and steel sheet quality change under different working conditions, and a theoretical basis is provided for equipment maintenance; the erosion weight loss of the steel sheet sample under the conditions of different particle components, impact angles and medium flow rates at high temperature can be tested, and repeated erosion experiments are performed on a single sample by testing the mass of the erosion steel sheet in real time, so that the change rule of the surface damage degree and the weight loss of the sample along with time under different working conditions is obtained.

This application relates to an automated sample quality control method and a sample analysis system. The method is used in a sample analysis system, which includes at least one analysis module. The method includes: determining a first quality control item for the first analysis module; acquiring status information of a quality control sample matching the first quality control item, the status information of the quality control sample including available quantity information and location information; determining a first quality control list based on the available quantity information of the quality control sample, the first quality control list being used to instruct the first analysis module to use the quality control sample to perform testing for the first quality control item; detecting the availability status of a detection reagent matching the first quality control list; and retrieving the quality control sample to execute the first quality control list based on the location information of the quality control sample. The technical solution provided by this application can improve scheduling efficiency.

The invention discloses a sampling system for geological experiment testing and a method thereof, and relates to the technical field of geological sampling, a plurality of lifting hydraulic cylinders are installed on one side of the top end of a bottom support limiting frame at equal intervals, a lifting clamping frame is installed at the top ends of the lifting hydraulic cylinders, and a rotating motor is installed at the bottom end of the inner side of the lifting clamping frame through a motor base; according to the device, cooling liquid is injected into the inner side of a serial cooling cavity through a draw-off pump and a press-fit sealing pipe, backflow treatment is carried out in cooperation with a backflow limiting pipe, and the backflow treatment effect is improved; the forced cooler is matched to perform forced cooling on the cooling liquid in the synchronous cooling barrel, internal circulation and cutting synchronous treatment are utilized, the situation that the cooling liquid is directly injected into the stratum, and consequently a sample is polluted by the cooling liquid in the sampling process is avoided, meanwhile, the stratum structure is prevented from being damaged by high temperature, and the sampling accuracy and the sample quality are guaranteed.

The invention discloses a sample preparation method and system meeting in-situ irradiation TEM and APT samples. The method comprises the following steps: pretreating a nuclear structure material; according to a pre-trained machine learning model, processing equipment parameters when the femtosecond laser processing technology is adopted for processing are obtained; carrying out rough machining on the nuclear structure material by adopting a femtosecond machining technology to obtain a rough sample; the rough samples are screened, finish machining is conducted on the rough samples meeting the screening standard through the FIB technology, and machining equipment parameters of FIB finish machining are obtained according to a pre-trained machine learning model; wherein the training of the machine learning model comprises the following steps: collecting historical data of sample preparation, including successful and failed cases, recording material characteristic parameters, processing equipment parameters and environmental condition parameters of each case as input variables, recording a sample preparation success rate and a sample quality index as output variables, and preprocessing the historical data; and training according to the preprocessed data to obtain a machine learning model.