737 results about "Predicting performance" patented technology

The invention relates to the technical field of cross-border payment optimization, and discloses a cross-border payment fund routing optimization method and system, and the method comprises the steps: receiving a cross-border payment transaction request, analyzing the transaction parameters of the cross-border payment transaction request, and querying and obtaining an initial available payment channel set and corresponding channel information according to the transaction parameters; according to the transaction parameters, identifying payment routing capability applicable to the transaction and matching a special routing rule of the transaction, and according to the payment routing capability and the special routing rule relationship, performing three-level routing rule filtering processing on the available payment channel set to obtain a remaining available payment channel set; three layers of historical performance index data are obtained, a three-layer performance index model is constructed by performing weighted fusion on the three layers of performance index data, performance indexes of all dimensions are predicted, comprehensive scores of all channels are calculated in combination with customer levels, and the optimal payment channel is selected for each cross-border transaction through the method. The transaction success rate is improved, the cost is reduced, and the user experience is optimized.

The invention provides a method and equipment for predicting the heat exchange performance of a buried pipe ground source heat pump thermal camouflage system. The method comprises the steps that S1, a multi-physics field coupling heat exchange mathematical model of a buried pipe heat exchange system is established; s2, based on the multi-physics field coupling heat exchange mathematical model, a control equation and boundary conditions are constructed; s3, on the basis of the control equation and the boundary conditions, a multi-physics field numerical model is constructed, and simulation and calibration are carried out; and S4, based on the calibrated numerical model, designing a simulation test and carrying out main control factor analysis to obtain a prediction effect of the heat exchange performance. By establishing a rock-soil body-fluid multi-physical field coupling model, the interaction mechanism of underground heat conduction, working medium flow and ground surface heat radiation is completely represented, and the hiding efficiency of the thermal camouflage system is remarkably improved.

The present disclosure discloses a multi-modal brain network calculation method, apparatus, device, and storage medium. The method is configured to train a brain disease prediction model. After the brain region structural feature and the brain region functional feature are separately extracted from magnetic resonance diffusion tensor imaging data and brain functional magnetic resonance data, a graph representation diffusion learning network is used to separate the universal feature and the unique feature in the brain region structural feature and the brain region functional feature. And then, multi-modal universal and unique feature fusion is implemented based on an alignment algorithm and adaptive weighting technology. Thus, complementary information between the multi-modal data is fully mining. The model can learn an effective feature of a related disease in a training process, and a finally obtained brain region disease prediction model has higher precision and better prediction effect.

Platforms, systems, and methods for genetic generalization in synthetic biology development. According to one aspect, there is provided a method for predicting performance associated with genetic edits, the method comprising: receiving, by a platform, information about a strain of a microorganism, wherein the information about the strain comprises information describing a plurality of genetic edits to a base strain of the microorganism; generating, by the platform, a set of genetic embeddings based on the information about the strain, wherein the generating comprises processing the information about the strain using one or more embedding models, wherein each of the one or more embedding models: receives the information about the strain of the microorganism as input; and applies computational transformations to the input using a corresponding embedding model to generate a multi-dimensional vector representation for each of the plurality of genetic edits.

The invention provides a reinforced concrete structure performance degradation intelligent prediction and durability limit state evaluation method. The method comprises the following steps: establishing a random variable probability model of environment and structure parameters; a convolutional neural network is fused to construct a chloride ion diffusion intelligent prediction model, and Monte Carlo sampling is adopted to analyze probability distribution of the initial corrosion time of the steel bar, so that prediction of the steel bar de-blunt time is realized; establishing a steel bar time-varying corrosion model to obtain the change condition of the steel bar corrosion loss rate along with time; further, an incremental static analysis method is adopted, and a multi-scale finite element model of the reinforced concrete structure under different corrosion rate conditions is established through random sampling; obtaining a critical load value in a limit state, and obtaining a failure probability curve under different corrosion degrees; finally, the bearing capacity failure time of the reinforced concrete structure is obtained by defining a bearing capacity reduction coefficient, and evaluation of the durability limit state of the reinforced concrete structure in the corrosion state is achieved.

The invention relates to the technical field of gas turbines, and discloses a geometric parameter optimization method, system and equipment for turbine blades and a medium. The method comprises the following steps: modeling the turbine blade according to geometric parameters, working condition parameters and material parameters of the turbine blade to obtain a plurality of first input samples with first fidelity and a plurality of second input samples with second fidelity; obtaining performance parameters of the turbine blade corresponding to each first input sample and each second input sample as a first response sample and a second response sample; and training the gas turbine performance prediction network, and predicting the performance parameters of the target turbine blade through the trained gas turbine performance prediction network according to the geometric parameters, the working condition parameters and the material parameters of the target turbine blade. And optimizing the geometric parameters of the target turbine blade by taking the maximum reliability and robustness of the target turbine blade indicated by the predicted performance parameters as a target.

The invention provides a traffic control system based on multi-modal data fusion, and belongs to the technical field of traffic control. Comprising a multi-source data acquisition and self-adaptive preprocessing module, a modal difference identification module, a modal difference modeling analysis module and a strategy fusion module, the multi-source data acquisition and self-adaptive preprocessing module is used for constructing a multi-modal original data set, the modal difference identification module is used for constructing a difference index sequence, and the modal difference modeling analysis module is used for analyzing the multi-modal original data set. The modal difference modeling analysis module is used for predicting a position difference coefficient, a speed difference coefficient and a behavior state difference coefficient of the k-type traffic target, and the strategy fusion module is used for constructing a strategy fusion coefficient. According to the system, the position, speed and behavior state differences of the traffic participation targets are analyzed, and finally the prediction effect of the traffic condition is improved through strategy fusion.

The invention provides a bearing fault variable working condition diagnosis method based on a multi-scale convolutional network and MAML, and relates to the technical field of equipment fault diagnosis. The method comprises the following steps: firstly, introducing fast Fourier transform to pre-process an original time domain vibration signal; secondly, a fault diagnosis model based on a multi-scale convolutional network and MAML is applied; and then, an internal and external circulation updating method based on model-independent element learning is adopted, so that the model can quickly adapt to a new task, and a relatively good prediction effect can be achieved only through a small amount of fine adjustment. According to the method, multi-scale feature extraction and meta-learning are creatively combined, the generalization ability of the model under variable working conditions is remarkably improved, the problem that a traditional fault diagnosis method depends on a single working condition and a large sample size is effectively solved, and the method is particularly suitable for small-sample and multi-working-condition bearing fault diagnosis scenes on an industrial site.

The invention relates to the field of intelligent transportation, and discloses an intelligent monitoring and management system and method for an expressway and electronic equipment, and the system collects global spatial-temporal data of the expressway to construct digital twins synchronized with the physical world; in the twinborn body, performing prediction and deduction based on a space-time causal map to identify potential risks; responding to the risk, generating an optimal intervention strategy through anti-fact deduction and executing the optimal intervention strategy, and recording a predicted intervention effect of the optimal intervention strategy; and after intervention, comparing a real traffic state with a prediction effect, calculating an anti-fact error, and carrying out dynamic self-correction on the space-time causal map according to the anti-fact error. According to the invention, links of perception, prediction, decision making, execution and feedback are fused into a self-adaptive control loop, and a self-correction mechanism based on an anti-fact error is introduced, so that the system can continuously learn and self-evolve from interaction with the physical world, and the problems of model solidification and poor adaptability of a traditional traffic management system are solved.

The invention discloses a sensor-based GIL pipe gallery expansion joint on-line monitoring system and method, and relates to the technical field of power equipment monitoring, and the method comprises the steps: collecting key state information and environmental parameters of an expansion joint through the deployment of an expansion joint sensor and an environmental sensor, and carrying out the preprocessing; performing deep fusion on expansion joint sensor data and environment sensor data, constructing a comprehensive state evaluation and prediction model, and analyzing the influence of different factors including environment, mechanical faults and material aging on the expansion joint state; correcting state parameters for environmental factors; for mechanical faults, fault types, positions and severity are identified, and corresponding early warning is triggered; aiming at material aging, monitoring and predicting performance degradation and residual life in real time, and formulating a maintenance strategy; the model performance is evaluated and updated regularly, an early warning threshold value is adjusted dynamically, and self-adaptive early warning is achieved; a data integration and visualization platform is developed, and visual display and interactive analysis of monitoring data are achieved.

The invention belongs to the technical field of aero-engine performance testing, particularly relates to a physical constraint-based aero-engine state prediction model construction method and system, and aims to solve the problems of high calculation complexity and poor data quality of an existing physical information model. The method comprises the following steps: acquiring historical operation data of the aero-engine and a physical constraint rule set of engineering simplification; predicting performance parameters by adopting a deep learning model; constructing a total loss function formed by weighting a data loss item and a physical loss item to train the model; wherein the physical loss item is generated based on the deviation degree of the predicted performance parameter and the engineering simplified physical constraint rule set, and is used for replacing the complex partial differential equation constraint. According to the method, the engineering simplified physical rule is introduced, so that the calculation overhead of model training is remarkably reduced, the model is effectively guided to learn the characteristics conforming to the physical rule, and the accuracy and generalization ability of the prediction model are remarkably improved under the condition of limited data.

The invention provides a virtual-real fusion engineering training method and system, and relates to the field of educational informatization and engineering practical training. According to the method, training knowledge points are extracted from multi-course resources through natural language processing, and a cross-course associated knowledge network is constructed; learning behaviors of students are collected to generate learning state vectors, and a personalized training path is planned by using a reinforcement learning algorithm. Students complete simulation and predict performance indexes on the digital twin platform, and design is automatically deployed to entity hardware for entity verification after the performance indexes are qualified; during operation, multi-modal data such as sensors and images are synchronously collected, stability, accuracy and compliance scores are output through a fusion model, an evaluation result is fed back to a knowledge network to dynamically adjust nodes and edge weights, training path closed-loop optimization is achieved, and the individuation, effectiveness and intelligent level of engineering education are remarkably improved.

The invention discloses an agricultural load prediction method and system based on multivariate time sequence decoupling multi-modal learning, and belongs to the technical field of agricultural load prediction. Comprising the steps of collecting historical agricultural load and meteorological data; decomposing historical agricultural load and meteorological data by using multivariate variational mode decomposition to obtain cycle, trend and residual mode components; and respectively constructing a time convolutional neural network, a bidirectional gating cycle unit and a support vector regression model for the decomposed period, trend and residual modal component data set, and fully mining feature information of each mode after decomposition, thereby realizing accurate prediction of agricultural load. According to the method, the potential nonlinear space-time coupling relationship between the agricultural load and the meteorological factor is captured, the prediction effect in a seasonal periodic fluctuation scene of the agricultural load and a long-term trend and agricultural load abnormal scene is improved, the agricultural load prediction precision is improved, and a support is provided for reliable and stable operation of a power grid.

The invention provides a CNN-PINN-based gas turbine combustion chamber performance prediction method, belongs to the technical field of combustion chamber combustion, and aims to solve the problems of high resource consumption and high grid dependence of the conventional CFD calculation at present, and the method comprises the steps: S1, collecting historical operation data of a combustion chamber, generating simulation data through numerical simulation, and carrying out the calculation of the simulation data; performing comparative analysis on the simulation data and the collected experimental operation data to prepare a data set; s2, preprocessing the data set; s3, establishing a CNN-PINN neural network prediction model according to the data characteristics and the target; s4, establishing a loss function containing a data error term and a physical information error term; s5, training and evaluating the established model by using the preprocessed data; and S6, performing performance prediction on target data by using the trained model, and performing reverse normalization on a prediction result to obtain prediction data corresponding to multiple targets.

The invention provides a WaveNet-BiLSTM-ATA oil well yield prediction method based on PSO optimization, and the method is characterized in that the method comprises the following steps: 1) collecting oil well time sequence characteristic data; 2) data cleaning and preprocessing; 3) executing data formatting and standardization operation to adapt to model test requirements; 4) dividing a training set and a test set according to oil well models; 5) screening important features of the oil well by adopting an isolated forest and a Pearson's correlation coefficient method, and constructing a daily oil production data set; 6) sequentially inputting the oil well characteristics in the step 5) into a WaveNet module and a BiLSTM module, introducing an ATA mechanism at the same time, and building a prediction model; 7) optimizing model hyper-parameters by using a PSO algorithm; and 8) carrying out grouping performance verification by adopting an independent test, completing model evaluation by calculating MAE, RMSE and MAPE indexes, and outputting a prediction result at the same time. According to the method, the WaveNet local feature extraction capability, the BiISTM bidirectional time sequence modeling advantage and the PSO efficient parameter optimization characteristic are fused, so that a relatively good prediction effect is obtained in a specific oil well test.

The invention discloses an artificial intelligence-based intelligent regulation and control method and system for a performance test of a valve actuating mechanism, and belongs to the technical field of industrial equipment automation control. The method comprises the following steps: firstly, collecting multi-source time sequence data in a testing process of a valve actuating mechanism, constructing a long short-term memory (LSTM) network model, and further predicting performance index degradation of the valve actuating mechanism; secondly, test parameter dynamic optimization of reinforcement learning RL is carried out, an optimal strategy is solved, and function updating is carried out; and finally, fault early warning and diagnosis are carried out, closed-loop feedback regulation and control are realized, and a closed-loop intelligent regulation and control framework for the performance test of the valve actuating mechanism is formed. The method is realized based on a data acquisition module, an edge calculation module, an intelligent decision module and an early warning diagnosis module. According to the invention, the fault early warning and diagnosis capability of the valve actuating mechanism can be improved, and the test regulation can be adjusted in real time according to the optimal strategy generated by the valve actuating mechanism, so that the adaptive test is realized.

The invention relates to a runway visual range prediction method based on multi-modal fusion, and the method comprises the following steps: S1, constructing a time-space matched data set which comprises Himawari-9 satellite thermal infrared channel brightness temperature image data, airport site meteorological element data and airport observation RVR / MOR data, carrying out the cleaning work of the data, and dividing the data into a training set, a verification set and a test set; s2, constructing an RVR / MOR forecasting model based on a Cross ViViT model, training model parameters by using a training set and a verification set, adjusting and optimizing the parameters, and performing precision evaluation on the model by using a test set; s3, mapping the high-dimensional semantic representation presented by the constructed RVR / MOR prediction model into a specific RVR prediction value for utilization; the problems that at present, only a traditional statistical model or a depth model based on a single mode is relied on, a complex RVR generation mechanism is often difficult to describe accurately, and particularly, response lag or prediction distortion phenomena easily occur in sudden weather events, so that the prediction effect is affected are solved.

The invention discloses a magnetic anomaly data de-noising method and de-noising system based on step-by-step U-Net, and relates to the technical field of geophysical exploration, and the method comprises the steps: carrying out the forward modeling calculation based on a magnetic anomaly model in a simulation region, and obtaining a synthetic magnetic anomaly data set; a U-Net step-by-step denoising network model is constructed; designing a self-adaptive multi-stage training mechanism based on mean square error minimization; training the U-Net step-by-step denoising network model through the training set, taking parameters of the U-Net step-by-step denoising network model as initial weights, and performing fine tuning on the U-Net step-by-step denoising network model by utilizing supervised learning; the verification set is used for verifying the effect, and the effectiveness of the whole step-by-step denoising network model is verified according to the prediction effect of the test set; a forward noise adding process and a reverse noise removing process are designed, network behaviors are automatically adjusted according to noise steps, the problems of high training cost and poor performance when a deep learning model processes colored noise are effectively solved, and the problem of low signal-to-noise ratio is better solved.

The invention belongs to the technical field of analog integrated circuit design automation, and discloses an analog integrated circuit cross-process performance prediction method and system based on Transform transfer learning, and the method comprises the steps: 1, carrying out the preprocessing of analog integrated circuit sample data; 2, serializing the sample data according to a preset sequence, mapping each parameter into vector representation, adding a global abstract vector, and splicing according to a fixed sequence to form an input sequence; 3, encoding the input sequence to obtain an output sequence; 4, inputting the global features into the regression head network, and outputting a performance index prediction value; 5, training is carried out, network parameters obtained through pre-training of the source technology are migrated to the target technology, fine adjustment is carried out on part of network parameters of a target technology data set, and cross-technology performance prediction is achieved; and 6, performing forward reasoning on operational amplifier design parameters under a given target process according to the steps 2-4 to obtain a performance prediction result. The method improves the stability and efficiency of performance prediction.

A machine learning (ML) based index advisor is provided to help optimize database systems for better cost and performance. The index advisor considers both the performance of queries and the cost of maintaining the indexes. It also provides performance and storage estimates, as well as explanations for the recommendations that are generated. The index advisor generates an index recommendation by generating a set of candidate indexes and applying a trained ML model to operations in the workload and each candidate index to determine a predicted performance benefit. The index advisor determines a total performance benefit for each candidate index.

The invention relates to the technical field of high-alkali coal unit coordination control, and discloses a high-alkali coal unit coordination control system performance comprehensive evaluation method and system, and the method comprises the steps: collecting operation parameter data through a plurality of sensors disposed at key positions of a unit to generate a performance data set, and receiving real-time data through an evaluation server to construct a performance matrix; the performance matrix is processed, features are extracted in combination with a performance data set, performance is predicted according to the operation state gradient and control parameter information, performance index space-time distribution features are output through a comprehensive evaluation model, and the performance data set is updated; and dynamically adjusting unit coordination control according to the evaluation indexes, wherein the operations comprise valve opening, fuel supply and the like. The system comprises a plurality of sensors, an evaluation server, a performance evaluation index determination module and a dynamic adjustment module. According to the invention, comprehensive and dynamic evaluation and accurate control of the performance of the coordinated control system of the high-alkali coal unit are realized, and the safety, stability and economical efficiency of unit operation are improved.

The invention relates to a time sequence prediction method based on time-frequency double-domain decomposition and multi-cycle feature fusion, and the method comprises the following steps: carrying out the normalization operation of input historical data, and carrying out different feature extraction operations for different data sets. Specifically, a trend term is firstly extracted using a moving average in a time domain. Then, a trend term is extracted by applying a selection mode of an adaptive spectrum rarefaction mechanism in a frequency domain, and meanwhile, noise is isolated into a residual term; for a multi-period data set, trend terms are decomposed in a recursive manner, and period modes are separated from short to long time scales. And for each decomposed feature item, learning and predicting by using a prediction module based on a linear model or a multi-layer perceptron, and finally fusing prediction components of each feature and performing inverse normalization operation to obtain a prediction result. According to the method, novel characteristic decomposition processing is carried out on the input data, more details are provided for future prediction, and a better prediction effect is achieved.

The invention relates to a TL-AEAT-BIGRU post-compression yield prediction method based on data joint driving under knowledge constraint, which adopts a CWGAN-GP model based on conditional constraint to perform data enhancement on a small amount of multi-source data, adopts a Pearson + mRmR correlation analysis algorithm to determine main control factors influencing post-compression yield, and performs prediction on the post-compression yield by using a TL-AEAT-BIGRU model. And classifying reservoir categories through a main component analysis method based on the reservoir classification standard established by the former. Correlation experience knowledge between the post-compression yield and the input main control factors is combined with the TL-AEAT-BiGRU post-compression yield prediction model, and the screened main control factors are used as input for post-compression yield prediction. The result shows that compared with other yield prediction models, the model has the best performance in the post-pressure yield prediction of the research area. An ablation experiment shows that each module of the model contributes to improvement of the yield prediction effect, so that the yield prediction accuracy of the model is effectively improved, the problem that the yield prediction precision of the reservoir after pressure is not high under the condition that the sample size is small is effectively solved, and the method has guiding significance on yield analysis of an oil and gas well.

The invention discloses an aluminum-based material rolling process optimization method based on a response surface method and machine learning, which comprises the following steps: collecting related data of an aluminum-based material rolling process, and establishing a rolling process database; determining key rolling process parameters influencing the performance of the aluminum-based material based on the database; a mathematical model used for describing the key rolling process parameters and the relation between the interaction of the key rolling process parameters and the aluminum-based material performance indexes is established, and prediction results of the aluminum-based material performance indexes under different key rolling process parameter combinations are calculated according to the mathematical model. Aluminum-based materials with performance indexes meeting preset requirements and key rolling process parameter combinations corresponding to the aluminum-based materials are predicted to serve as a preliminary optimization result data set; and constructing a back propagation artificial neural network model optimized by a genetic algorithm to predict aluminum-based material performance indexes corresponding to different key rolling process parameter combinations, and screening out an optimal key rolling process parameter combination and an aluminum-based material performance index corresponding to the optimal key rolling process parameter combination from the aluminum-based material performance indexes. According to the method, the quality of the aluminum-based material is remarkably improved through the proposed collaborative optimization strategy.

The invention provides an intelligent storage location optimization method and system, and belongs to the technical field of storage management, and the method comprises the steps: obtaining the data information of commodities in a storage system, constructing a storage location optimization model, comprising a comprehensive objective function taking maximization of space utilization rate and sorting efficiency and minimization of storage cost as objectives and constraint conditions based on physical rules and business rules; solving the storage location optimization model by using an optimization algorithm, and generating a storage location distribution scheme and prediction performance data; executing the storage location allocation scheme, and collecting actual storage location data and actual operation performance data; comparing the actual storage location data with the storage location allocation scheme to obtain a scheme execution coincidence rate; comparing the actual operation performance data with the predicted performance data to obtain a performance deviation rate; and based on the scheme execution coincidence rate and the performance deviation rate, adjusting the constraint condition of the storage location optimization model and / or the weight coefficient of the comprehensive objective function to obtain an updated storage location optimization model, and carrying out storage location allocation.

The invention relates to the technical field of automobile design, in particular to an intelligent generation type design method of an anti-collision beam, which comprises the following steps: firstly, performing experimental analysis on the anti-collision beam to obtain a section image comprising the anti-collision beam and performance response data corresponding to the section image; performing image recognition and text extraction on the section image; fusing the recognized image and the extracted text by using a multi-modal multi-layer fusion model to obtain a multi-modal design variable; training the constructed conditional variation network by using the multi-modal design variables and the performance response data corresponding to the multi-modal design variables to obtain a conditional variation model; generating design variables by using the conditional variation model, and predicting performance response data corresponding to the variables; optimal performance response data are screened out from the obtained performance response data, and then the optimal design scheme of the anti-collision beam to be designed is obtained. According to the method, the optimal design scheme is determined by using the conditional variation network, and the efficiency and the precision of optimization design are effectively improved.

The invention discloses an ultra-low interfacial tension surfactant screening method and system based on AI4S, and relates to the technical field of oil and gas exploitation. The method comprises the following steps: constructing a database covering molecular structures of typical surfactants; constructing a prediction model for predicting the surface tension; generating candidate molecules by using a generative artificial intelligence model comprising a variational auto-encoder and a generative adversarial network; inputting the candidate molecules and the experimental conditions into the trained prediction model to predict the interfacial tension of the prediction model under the target condition, and screening the candidate molecules with the predicted performance reaching the standard; and performing experimental determination on the screened candidate molecules until molecules with target performance are obtained, and feeding back experimental results to a database to form closed-loop optimization of data, prediction and experiments. The research and development efficiency and success rate of the chemical oil-displacing agent can be remarkably improved, the research and development period is shortened, the research and development cost is reduced, and effective technical support is provided for improving the oil and gas recovery efficiency and guaranteeing energy safety.

A system and a method are disclosed for a tool that generates a provider score corresponding to a predicted performance of a provider based on data of claims involving the provider. For a given claim, the tool provides the data as input into a supervised machine learning model and receives as output from the supervised machine learning model a predicted performance of the claim. The tool also inputs the data of the claim into an unsupervised machine learning model that is selected based on a stage of claim processing that the claim belongs to and receives as output from the unsupervised machine learning model an identification of a cluster of candidate claims to which the claim belongs. The tool combines the outputs of the supervised machine learning model and the unsupervised machine learning model to generate the provider score.

The invention discloses an ultrasonic water meter flow measurement and correction system based on Internet of Things, which relates to the technical field of water meter monitoring and comprises a sensing unit, a working data processing unit, a factor determination and analysis unit, a user management unit and an abnormity alarm unit. According to the invention, the sensing unit is used for multi-dimensionally collecting the working related data of the ultrasonic water meter, the working data processing unit is used for real-time processing and interference suppression, the quality of the obtained working data of the ultrasonic water meter can be improved, and the factor determination and analysis unit is used for determining the working data of the ultrasonic water meter by screening the performance related characteristics of the water meter, predicting the performance change trend and adjusting model parameters. The user management unit is convenient for a user to inquire data; the abnormal alarm unit ensures stable operation of equipment; the flow measurement precision can be improved on the whole, the system adapts to complex working conditions, intelligent management and fault early warning are realized, and powerful data support is provided for water affair management.

One or more embodiments of the invention provide a training method and device of a semiconductor structure prediction model and an optical critical dimension measurement method and device. The training method comprises the steps that an original training sample is acquired, the original training sample comprises an original spectral value and multiple sets of target parameters corresponding to a semiconductor structure, and the following steps are executed circularly until conditions are met: the original training sample is adopted as a training sample used for first-round training to train a structure prediction model; after training is completed, whether the precision of the structure prediction model reaches the standard or not is judged; under the condition that the target parameters do not reach the standard, the prediction effect of the current round of each group of target parameters is determined; for each group of target parameters, under the condition that the prediction effect of the current round is superior to the prediction effect of the previous round, updating model weights corresponding to the target parameters; determining an incremental training sample used in the next round of training; and determining the original training sample and the incremental training sample as training samples used in the next round of training to continue to train the structure prediction model.