85 results about "Sample space" patented technology

The invention discloses a small sample modeling stability evaluation method and system based on Bootstrap resampling, and particularly relates to the technical field of computers and data intellectualization, the method comprises the following steps: completing data preprocessing and stable stage identification under a unified time base, and forming a segment set; an average block length is used as a decision quantity, an SBB is optimized to construct a sample space, and an optimal block length is determined in combination with variance consistency and nominal coverage rate consistency criteria; carrying out re-sampling training around short / medium / long scales, and collecting layered indexes such as prediction, parameters and features; stability calculation and empirical coverage rate calibration are completed based on intra-scale statistics and inter-scale weighted mixing, and a pseudo-stationary diagnosis score is output; and finally, engineering judgment and backspacing optimization are carried out according to the coverage rate, the correlation maintenance and the risk threshold. The system records random seeds, block metadata and model version generation evidence pointers, has the characteristics of traceability and reverifiability, and is suitable for scenes of production lines, network traffic, finance and the like.

The invention discloses a flow field prediction-oriented full-process automatic joint reduced-order modeling method, which comprises the following steps of: specifying a target physical field parameter space, and randomly generating a sample space according to a Latin hypercube sampling method; constructing a full-process automatic simulation tool chain, driving target physical field numerical calculation and generating a training data set; carrying out singular value decomposition-based intrinsic orthogonal decomposition on the output physical field data, and only retaining first r main feature components to construct a reduced-order data set; constructing a multi-input multi-output full-connection feedforward neural network, and modeling and training a nonlinear mapping relation between input parameters and reduced-order features; new working condition parameters are input, reduced-order features are predicted through the trained neural network, distribution of a target physical field is reconstructed according to a singular value decomposition reduction matrix, and more flexible and reliable technical support is provided for reducing the training cost of a reduced-order model and improving simulation efficiency.

The invention belongs to the technical field of water conservancy and hydropower engineering, and particularly relates to a computer vision-based dam seepage field rapid calculation method, which comprises the following steps of: acquiring geological attribute parameters and upstream and downstream water level working condition parameters, and constructing a sample space; constructing a data set by using a dam seepage numerical model considering the spatial variability of the permeability coefficient based on the sample space; and establishing an improved conditional deep convolutional generative adversarial network model based on the data set to realize rapid calculation of the dam seepage field. The problems that in an existing dam seepage behavior analysis and research method, a numerical simulation method occupies too many computing resources and consumes long time, most agent models only pay attention to local scattered measuring points for modeling, and overall seepage field modeling at a key section cannot be achieved are solved. However, most of the current image generation researches based on computer vision neglect the generation precision problem of high-frequency details of the image, and the high-frequency edge information of the image cannot be accurately generated.

The invention relates to a harmonic reducer fault diagnosis method based on double-branch feature fusion. The harmonic reducer fault diagnosis method comprises the following steps: synchronously acquiring multi-sensor data of a harmonic reducer acquired by vibration, acoustic emission and current sensors; processing the multi-sensor data through a one-dimensional convolutional neural network to obtain data in a two-dimensional matrix form; the data in the two-dimensional matrix form is input into a pre-trained model, and the processing process of the model is as follows: feature extraction is performed on the data in the two-dimensional matrix form through an MCWGraphKAN branch feature extraction module and a GBiMama branch feature extraction module to obtain a first feature and a second feature, the first feature and the second feature are fused and then mapped to a sample space through a full-connection layer, and the data in the two-dimensional matrix form are obtained; and fault classification is realized. The method achieves the fault diagnosis of the harmonic reducer through a deep learning method, and is used for solving the problems that a single sensor cannot fully express the operation state of the harmonic reducer, and the fault feature extraction of a single-branch network is insufficient.

The invention discloses a CKM-CNN-based power system transient stability assessment method, system, equipment and medium, and relates to the technical field of power system assessment, and the method comprises the steps: collecting time sequence electrical quantity data after power system fault disturbance, constructing a transient sample set, carrying out the standardization processing and marking of a sample, and carrying out the calculation of the transient stability of the power system; a transient sample set is divided through a clustering algorithm, based on a sample cluster division result, an adaptive oversampling technology is adopted to carry out directional enhancement on an unstable sample, an equalized sample space is generated, a multilayer convolutional neural network is utilized to carry out hierarchical feature extraction on the equalized sample space, a high-dimensional feature vector is generated, and according to the high-dimensional feature vector, a high-dimensional feature vector is generated. And outputting a transient stable state probability evaluation result through the classification decision-making layer. According to the method, a generalization sample set is constructed through standardization and labeling, the scarce instability sample identification capability is improved by combining two-stage clustering and ADASYN oversampling, and the system dynamic state is accurately captured by utilizing 1D-CNN hierarchical feature extraction, so that efficient and robust evaluation of the transient stability of the power system is realized.

The invention relates to a frozen coal train uniform temperature rise optimization method based on Morris-CNN fusion. The method comprises the steps that model structure parameters and electromagnetic parameters are determined; a Morris method is adopted, and parameters are selected; obtaining a sample space; obtaining a plurality of grid units; obtaining the temperature value of the temperature measuring point; calculating temperature rise efficiency and uniformity indexes; forming a sample data set to train a convolutional neural network; searching values of five parameters corresponding to the optimal temperature rise efficiency and the optimal uniformity; and the final optimal temperature rise efficiency and the optimal uniformity are obtained. According to the method, a structure optimization design based on the combination of Morris global sensitivity analysis and the convolutional neural network is adopted, the method combines a mathematical method and a statistical method, modeling and analysis are carried out on the influence of multiple variables, and the efficiency and the effect are improved; according to the method, the temperature rise uniformity and the temperature rise efficiency are optimized by applying a multi-objective optimization method, the gradient descent characteristic of the convolutional neural network is utilized, the uniformity and the temperature rise efficiency are balanced, and the optimal parameters are obtained to guide the quick and uniform temperature rise design of the frozen coal train.

The invention provides a gear wind resistance power loss prediction method and system, computer equipment and a medium, and belongs to the field of aero-engine power transmission, and the method comprises the steps: obtaining physical parameters of a to-be-predicted gear; according to the physical parameters of the to-be-predicted gear, gear structure parameters are extracted, tooth surface torque parameters and fluid mechanics parameters are calculated, and the gear structure parameters, the tooth surface torque parameters and the fluid mechanics parameters of the to-be-predicted gear are consistent with the same type of parameters in the sample space in sequence and dimension; the gear structure parameters, the tooth surface torque parameters and the fluid mechanics parameters of the to-be-predicted gear are input into a pre-trained power loss prediction model, the wind resistance power loss of the to-be-predicted gear is obtained, and the power loss prediction model is obtained through a sample space training deep learning model. Compared with numerical simulation, the method has the advantage that the power loss prediction efficiency is remarkably improved.

The invention provides a building energy consumption multi-area cooperative control method, device and equipment and a storage medium, and can be applied to the technical field of artificial intelligence and building energy consumption control. The method comprises the steps that temperature evaluation information and energy consumption evaluation information of a sample space area in a sample building are processed, an initial evaluation result of a sample environment temperature in a preset time period is obtained, and the temperature evaluation information is determined according to sample temperature information of the sample space area; the energy consumption evaluation information is determined by sample power information, sample energy consumption information and cost information of the sample space region; obtaining a control agent based on the sample information, the initial evaluation result, and loss functions and fusion strategies of respective strategy models of the sample space region and the sample association region; and on the basis of the state information of the space area in the building and the associated area information, action information of the valve opening degree of the control equipment is obtained, control operation is executed on the valve opening degree on the basis of the action information, and energy consumption control information of the space area is obtained.

The invention provides an optimized design method for an impact type rotating wheel bucket blade structure based on entropy production analysis. The method comprises the following steps: firstly, generating a sample space through parametric modeling and optimal Latin hypercube experimental design of a rotating wheel; then, acquiring internal flow field data through numerical simulation; identifying a key loss area based on entropy yield flow loss; then, a Gaussian process regression agent model is constructed by taking the geometric dimension parameters of the bucket blade as input and hydraulic efficiency and entropy production as output, multi-objective optimization is carried out by adopting an improved NSGA-II algorithm, and the accuracy of the agent model is improved through a dynamic updating strategy; and finally, carrying out numerical simulation verification on the Pareto optimal solution, and screening out a runner design scheme with excellent comprehensive performance. The flow loss can be effectively reduced, the hydraulic efficiency and the operation stability of the impact type runner are improved, and technical support is provided for optimization of the hydraulic performance of the impact type runner.

The present invention discloses a user operation trajectory analysis method, apparatus, device, and storage medium, which are applied to the field of computer interaction technology. The method comprises: obtaining operational behavior information of users accessing system pages; the operational behavior information includes at least: page information, information about operational components corresponding to the page interactions, and information about regions corresponding to the operational components; constructing a dynamic trajectory sample space based on the operational behavior information; characterizing and stratifying the dynamic trajectory sample space to obtain stratification results; fitting the user operation trajectory based on the stratification results to obtain a target operation trajectory. The present invention characterizes and stratifies the dynamic trajectory sample space and fits the user operation trajectory based on the stratification results to extract high-frequency operational behavior trajectories of different types of users, thereby greatly improving the accuracy of data analysis.

The invention provides a balance calibration modeling method and system based on sample errors and space optimization. The method comprises the steps of obtaining a calibration data set and dividing the calibration data set into a sample set and a verification set; calculating a load space weight and a load complexity weight of the sample set, and obtaining a comprehensive weight based on the load space weight and the load complexity weight; and constructing a polynomial model of the sample set, and solving a coefficient matrix in the polynomial model based on the comprehensive weight to obtain a final balance calibration model. The method further comprises the following steps: verifying and evaluating the balance calibration model by using the sample set and the verification set, and optimizing the balance calibration model according to an evaluation result. According to the balance calibration modeling method cooperating with the sample space distance and the load complexity weight provided by the invention, nonlinear system errors introduced by calibration load complexity are fully considered, so that the influence of error distribution of calibration samples under various load combinations on modeling is more uniform and reasonable; and meanwhile, the problem of sample space occupation ratio is also solved.

This invention discloses a method for constructing a multiscale geological model based on a convolutional conditional neural process. The method comprises: building and training a convolutional conditional neural process model; loading conditional data and generating a grid to be simulated; inputting the grid to be simulated and the conditional data into the trained convolutional conditional neural process model; obtaining a current spatial probability distribution map based on the conditional data and its spatial distribution; and, based on the conditional data and the spatial probability distribution map, converging the variable range of the generated result to a sample space using statistical methods; and saving the final result to complete the simulation. The invention has the beneficial effect of significantly improving the reconstruction capability and efficiency of geological heterogeneous models.

The application discloses a kind of federal learning adversarial sample detection methods based on variational bayesian network, comprising the following steps: training model is issued to each participant of federal learning;Each participant initiates private intersection operation two by two, so that each participant obtains respective sample intersection;The sample intersection obtained by each participant is intersected, to obtain common sample space;Each participant filters target local sample from respective local sample according to common sample space, uses respective target local sample of each participant to train model, to obtain respective gradient information of each participant;Gradient information sent by each participant is received, and the uncertainty of gradient information distribution density is calculated;Adversarial sample in the local sample of each participant is determined according to the uncertainty of gradient information distribution density.The application can complete the task of adversarial sample detection in federal learning system without directly accessing private training data.

The application provides a building energy consumption multi-region collaborative control method, device, equipment and storage medium, which can be applied to the fields of artificial intelligence and building energy consumption control technology. The method comprises the following steps: processing temperature evaluation information and energy consumption evaluation information of a sample space region in a sample building to obtain an initial evaluation result of a sample environment temperature within a preset time period, the temperature evaluation information being determined by sample temperature information of the sample space region, and the energy consumption evaluation information being determined by sample power information, sample energy consumption information and cost information of the sample space region; obtaining a control intelligent agent based on sample information, the initial evaluation result, a loss function of a respective strategy model of the sample space region and a sample associated region, and a fusion strategy; obtaining action information of a valve opening degree of a control device based on state information and associated region information of a space region in the building, performing a control operation on the valve opening degree based on the action information, and obtaining energy consumption control information of the space region.

The application discloses a static voltage stability boundary calculation method and system based on a deep neural network. The static voltage stability boundary point sample set is obtained by calling the continuation power flow method to detect the PV curve nose point for multiple times; the static voltage stability boundary model of a complex high-dimensional power system is established by taking the power growth mode as input features and taking the maximum power growth as an output variable; since the uniformity of the sample has a great influence on the model, the "baffle method" which can uniformly generate feature sets in the sample space is proposed; the static voltage stability boundary sample is generated based on the "baffle method" in the IEEE9 and IEEE39 node systems, and compared with the Monte Carlo random sample, the uniformity and efficiency of the sample generated by the "baffle method" are higher; the static voltage stability boundary model of the system is constructed based on the deep neural network, and compared with the boundary detected by the continuation power flow method, the accuracy of the model is verified, and compared with the time domain simulation boundary, the effectiveness of the model is verified.

The application provides a turbine blade thermal barrier coating force-thermal coupling reliability evaluation method, which comprises the following steps: measuring and constructing a density distribution function of a thermal barrier coating microstructure and thermodynamic parameters, completing macroscopic complex turbine blade thermal barrier coating temperature field and strain field simulation, constructing a Monte Carlo sample space of random parameters at different positions of the blade, establishing a micro-scale force-thermal coupling finite element model and obtaining a life data set, building and training a deep learning agent model, and completing turbine blade thermal barrier coating reliability evaluation and sensitivity analysis. The reliability evaluation method provided by the application comprehensively considers the influence of the thermal barrier coating microstructure factor and the force-thermal coupling failure mechanism, solves the problems of high dimension, high nonlinear calculation cost and the like in the reliability evaluation of the complex blade thermal barrier coating, improves the accuracy and efficiency of the reliability evaluation, and provides an evaluation means for the safe application and optimized design of the thermal barrier coating.

The invention discloses a transient multi-physics field reduced-order reconstruction method and device driven by a time sequence working condition, and relates to the technical field of digital twinning and calculation simulation acceleration. The method comprises the following steps: performing Latin hypercube sampling on monitoring working condition parameters to obtain a working condition parameter combination covering a sample space; establishing a high-fidelity finite element model to obtain geometric topology information; establishing a working condition parameter amplitude curve, performing batch finite element calculation in combination with the high-fidelity finite element model to obtain transient multi-physics field response data, and constructing a graph data structure; the training graph neural network auto-encoder comprises an encoder and a decoder; and training a time sequence neural network, establishing a mapping relation from time sequence working condition parameters to low-dimensional latent variables, and realizing rapid prediction and reconstruction from the time sequence working condition parameters to transient multi-physical field whole-field distribution in combination with a graph neural network decoder. According to the method, online rapid prediction of the transient physical field of the high-temperature turbine component can be realized, and the method has the advantages of high prediction precision, fast calculation response, strong cross-working-condition generalization capability and the like.

The invention provides an oil and gas block well seismic label capacity expansion method, system, device and medium, and belongs to the field of geology, geophysics and machine learning, and the method comprises the steps: dynamically delimiting a core work area based on a real well, and laying candidate points in a gridding manner; pCA dimension reduction and parallel prediction of a plurality of models are adopted to obtain a preliminary screening virtual well point pool; candidate points contradictory to real data are eliminated by setting a real well protection radius and a threshold value, and box separation and capacity limiting are performed according to a sand thickness interval, so that sample space and attribute distribution balance is ensured; and finally, through multi-round iterative extrapolation, sedimentary phased directional attenuation constraint is introduced, double verification is performed on newly added samples, and the coverage range of reliable samples is gradually expanded. According to the method, hundreds of times of expansion of the well seismic label is realized, the generated virtual sample has both geological rationality and statistical diversity, deep learning model training can be effectively supported, the accuracy and generalization ability of reservoir parameter prediction in a less-well area are remarkably improved, and a reliable data basis is provided for oil-gas exploration decision.

This invention discloses a method for high-dimensional uncertainty aggregation modeling of power systems based on Gaussian mixture models (GMMs). The method comprises two stages: characterization of the original uncertainty distribution and GMM aggregation modeling. The GMM aggregation modeling specifically includes three steps: determining the dimension of aggregation variables, calculating aggregation parameters, and integrating aggregation results. This method achieves probability-preserving aggregation of high-dimensional uncertainty variables in power systems while significantly reducing the required sample size, effectively solving the problem of the large sample size required by Monte Carlo sampling when processing high-dimensional data. This invention can be applied to the dimensionality reduction aggregation of high-dimensional variables in uncertain power systems with a high proportion of renewable energy integration, achieving sample space reduction while maintaining probabilistic characteristics, and can be further applied to power system risk assessment, reliability analysis, and operation planning.

This invention provides a CO2 emission concentration prediction method based on a Bayesian optimized principal-complement model, belonging to the field of CO2 emission concentration detection technology. The method includes: acquiring the original dataset of CO2 emission concentrations; selecting optimization parameters based on the Bayesian optimization algorithm and initializing the optimization parameters to obtain a sample space; constructing an objective function and calculating the objective value for the original dataset and the sample space to obtain the objective function value; optimizing the objective function and obtaining the optimal parameter combination through the optimized objective function; obtaining the final prediction model based on the optimal parameter combination, and predicting CO2 emission concentrations using the final prediction model. This method solves the problem of a single prediction algorithm, achieves collaborative optimization of coupled hyperparameters, and avoids the tediousness and uncertainty of manual parameter tuning.

The invention belongs to the technical field of power distribution network optimization scheduling, and discloses a multi-energy micro-grid improved scheduling method, system, equipment and medium, and the method comprises the following steps: obtaining random factors of a multi-energy micro-grid, generating a sampling scene based on a Latin hypercube sampling method, and carrying out the clustering processing of the sampling scene through combining with a hierarchical clustering method; constructing a multi-energy micro-grid optimal scheduling model, and solving the multi-energy micro-grid optimal scheduling model to obtain an optimal scheduling scheme; and scheduling the multi-energy micro-grid, and continuously optimizing a scheduling scheme based on a rolling optimization method according to data newly generated in the operation process of the multi-energy micro-grid. According to the method, the scene generation effect is better by adopting Latin hypercube sampling, so that the simulation scene can fully cover the sample space, and the problem of lack of random factor scenes is solved.

The invention relates to the technical field of label noise of table data, a distance measurement-based method and a neural network-based method are currently common label noise screening methods, and the methods are difficult to distinguish wrongly labeled samples and difficult samples with fuzzy categories near a decision boundary; according to the table data noise recognition and correction method based on Stein scoring, the difference of the attribution degree between sample and data feature distribution and label export distribution is analyzed, and the logarithmic probability density gradient of a sample is estimated through a diffusion model; moving the sample along a gradient represented by a scoring function by using a gradient method until the sample is converged to a centroid, forming a moving track moving from an original position of a sample space to a high-density centroid of data of a domain, and calculating the data feature distribution by comparing the differences of the directions and lengths of the moving track of the sample in the data feature distribution and the distribution exported by the label. Samples of potential tag errors in the tabular data are identified and tags thereof are corrected.

The invention belongs to the field of power system operation reliability evaluation, and relates to a power system robust load shedding decision-making method oriented to a wide-area sample space, which comprises the following steps: acquiring a current power grid state diagram of a power system, utilizing the trained OOD detection model based on hypergraph and substructure enhancement to judge the scene type to which the current power grid state diagram belongs according to the current power grid state diagram of the power system; if the scene type is an in-distribution scene, using a load shedding prediction model based on a graph neural network to make a load shedding decision according to a current power grid state graph of the power system; if the scene type is an out-of-distribution scene, using a traditional reliability evaluation model to make a load shedding decision according to a current power grid state diagram of the power system; according to the method, the OOD detection model is adopted to identify the scene type to which the power grid state diagram belongs, so that efficient prediction of the scene in the distribution and safety guarantee of the scene outside the distribution can be considered, and the real-time and reliability requirements of load shedding judgment in actual operation of the power system are met.

The invention relates to the technical field of silicon controlled rectifier testing, in particular to a silicon controlled rectifier impact resistance testing method and system. The method comprises the following steps: acquiring a zero-crossing signal monitoring moment and a driving signal monitoring moment, acquiring delay duration controlled by a relay each time, and synchronously acquiring environment monitoring data at the monitoring moments; calculating a time extension error coefficient based on the delay duration, calculating an environment error coefficient based on the environment monitoring data, and forming a coefficient vector; collecting all coefficient vectors, calculating a time delay controllable coefficient of each time of relay control, constructing a two-dimensional sample space, and calculating error influence expressions corresponding to different time delays; carrying out statistical analysis on the error influence performance data of all the silicon controlled elements, and determining an error influence performance threshold value; and comparing the error influence performance at each monitoring moment with an error influence performance threshold value, and evaluating the impact resistance of the silicon controlled element. The accuracy of the impact resistance test result of the silicon controlled rectifier is remarkably improved.

A method for simulating a quantum-capacitance response of a material configuration comprises (a) constructing a non-interacting Hamiltonian for the material configuration based on input data; (b) computing a natural-orbitals basis for each of a plurality of parts of the material configuration under the non-interacting Hamiltonian; (c) projecting the non-interacting Hamiltonian in the natural-orbitals basis to obtain a non-interacting quantum-mechanical description for each part; (d) constructing an interacting Hamiltonian by adding an electron-interaction term to the non-interacting Hamiltonian for each of the plurality of parts; (e) for each of a plurality of representative points in a sample space of at least one tunable parameter of the material configuration, using a sums-of-Gaussians procedure to assemble a basis of Gaussian states for approximating low-energy eigenstates of the material configuration under the interacting Hamiltonian; (f) for each of a plurality of vicinities of representative points in the sample space, combining bases of Gaussian states assembled for nearby representative points to form an extended basis; and (g) forecasting the quantum-capacitance response within the sample space using the extended basis.

A method for simulating a quantum-capacitance response of a material configuration comprises (a) constructing a non-interacting Hamiltonian for the material configuration based on input data; (b) computing a natural-orbitals basis for each of a plurality of parts of the material configuration under the non-interacting Hamiltonian; (c) projecting the non-interacting Hamiltonian in the natural-orbitals basis to obtain a non-interacting quantum-mechanical description for each part; (d) constructing an interacting Hamiltonian by adding an electron-interaction term to the non-interacting Hamiltonian for each of the plurality of parts; (e) for each of a plurality of representative points in a sample space of at least one tunable parameter of the material configuration, using a sums-of-Gaussians procedure to assemble a basis of Gaussian states for approximating low-energy eigenstates of the material configuration under the interacting Hamiltonian; (f) for each of a plurality of vicinities of representative points in the sample space, combining bases of Gaussian states assembled for nearby representative points to form an extended basis; and (g) forecasting the quantum-capacitance response within the sample space using the extended basis.

The present application relates to thyristor testing technical field, specifically to a kind of thyristor impact resistance test method and system.The method includes the following steps: obtaining zero-crossing signal monitoring moment and driving signal monitoring moment, obtain the delay length of each relay control, and synchronously obtain the environmental monitoring data of monitoring moment;Based on the delay length calculation time extension error coefficient, based on environmental monitoring data calculation environmental error coefficient, and form coefficient vector;Collect all coefficient vectors, calculate the time delay controllable coefficient of each relay control, construct two-dimensional sample space, calculate the error influence performance corresponding to different time delay;Statistical analysis is carried out to the error influence performance data of all thyristor elements, and the threshold of error influence performance is determined;The error influence performance of each monitoring moment is compared with error influence performance threshold, and the impact resistance of thyristor element is evaluated.The present application realizes the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of the accuracy of

The application discloses an Optuna-XGBoost tunnel stability prediction-based hyperparameter optimization method and relates to the technical field of tunnel construction, and solves the problem that the existing technology is difficult to capture the complex deformation mode by using the posterior probability criterion as the deformation anomaly point strategy, is prone to misjudgment, and influences the accuracy and generalization ability of prediction; the application comprises the following steps: according to the proportion of real data and simulation data in the stability comprehensive index category and the joint fissure rock mass tunnel stability sample database, a training set and a verification set are divided to obtain a sample space; an XGBoost classifier is trained by using an Optuna Bayesian optimization framework based on TPE and the training set; category weights and data source weights are constructed and are distributed to the training samples; after multiple rounds of training, the performance effect is verified by using the verification set to determine the optimal hyperparameters of the XGBoost classifier; and the application improves the applicability and robustness of the prediction model under the condition of complex joint fissure surrounding rock.

The application discloses a kind of experimental scheme design method and device of multi-target variable, the method includes: obtaining the multiple characteristic variables of constituting experimental scheme, utilize all characteristic variables to constitute original sample space;The value of each characteristic variable in the original sample space is pre-sampled, to obtain statistical quantity sampling result;Based on statistical quantity sampling result, each characteristic variable in the sample space after pre-processing is carried out cluster analysis, to obtain target sampling point, utilize all target sampling points to constitute the sample space after pre-processing;Using pre-stored sampling algorithm, the sample space after pre-processing is sampled, to obtain target sample space, the sample in the target sample space is used for experimental sample in experimental scheme design.The method and device provided by the application use the sample data set of reduced space to provide data support for experimental scheme design, so that the experimental method of bayesian optimization is applied to large sample space becomes possible.

The invention discloses a small sample model construction method for receiver interference response prediction, and the method comprises the following steps: 1) constructing a sample space from a plurality of single typical interference signal data sets, and forming task distribution of different interference response data; 2) respectively reading input signals of data in a support set and a query set in task distribution, and carrying out wavelet transformation on the input signals to obtain low-frequency, intermediate-frequency and high-frequency wavelet coefficients; 3) constructing a time convolution network TCN model as a prediction model; step 4) model training; the invention provides a modeling method for receiver interference response with few samples, and the trained model can be used for predicting interference response under other interference conditions. The modeling method for receiver interference response has the advantages that the modeling method for receiver interference response with few samples is provided, and the trained model can be used for predicting interference response under other interference conditions.