1062 results about "Hyper parameters" patented technology

The invention provides an intranet service quality optimization method and system based on deep reinforcement learning, and relates to the technical field of artificial intelligence, and the method comprises the steps: collecting multi-dimensional performance index data, carrying out the preprocessing, constructing an analytic hierarchy model, outputting the performance score of each layer, integrating the performance score of each layer, and outputting an intranet service quality score. Constructing a simulation environment, adjusting hyper-parameters of the adaptive deep reinforcement learning model, continuously learning in an actual intranet environment, determining multi-dimensional rewards, performing multi-target optimization, and generating a multi-target optimization result; and setting a hierarchical performance index threshold, if a hierarchical optimization strategy is triggered, adjusting a network topology structure, adjusting resource allocation and service configuration, adjusting weights of different indexes in the service quality evaluation model, updating a reward function, solving an optimal service decision, generating an optimal service decision sequence, and executing the optimal service decision sequence.

The invention discloses a data exception traceability tracking and positioning method, system and device and a storage medium. Relates to the technical field of power system data management. At present, the problems of insufficient data anomaly detection accuracy, low traceability positioning efficiency, incomplete knowledge association and the like in analysis and application links of power marketing data are solved. The method comprises the following steps: carrying out preprocessing and feature extraction on power marketing data, constructing a data association relationship network through a clustering analysis and statistical method, analyzing the difference of data features, analyzing the propagation path of an abnormal point identified in a data stream by adopting a path reasoning technology, and constructing a data abnormity tracing positioning model. A semi-supervised learning method is adopted to improve the accuracy of the model, result optimization and hyper-parameter adjustment are carried out, and multi-condition factors influencing business data information are introduced into the model as hyper-parameters. According to the technical scheme, the anomaly detection and traceability positioning capability of the power marketing data is greatly improved.

The invention discloses a power load prediction method based on multi-source heterogeneous data feature fusion, system equipment and a medium, and relates to the technical field of power system load prediction, and the method comprises the steps: obtaining a power load prediction data set, and carrying out the preprocessing; performing input feature processing on the static data, the graph data and the time series data based on a multi-layer perceptron, a graph convolutional network and a gating loop unit network, and fusing the processed features based on a feature splicing method to obtain comprehensive feature information; based on full-connection neural network construction, taking the comprehensive features as input, inputting training set data into a model, and performing iterative optimization according to a model effect to adjust model parameters and hyper-parameters; and based on the parameter optimization adjustment model, using the evaluated optimal model to predict the power load in the future time period. The method is superior to a traditional method in precision and stability, can better adapt to complex and changeable load prediction scenes, and has wide application prospects and remarkable actual value.

The technical scheme of the invention relates to the technical field of gas detection, in particular to a hazardous chemical storage leakage positioning and tracing method based on a gas array. The method comprises the following steps: collecting gas concentration, components and environmental physical field parameters through a three-dimensional heterogeneous sensor array, and generating a multi-modal data set; and matching the noise mode feature library in the knowledge base by using a transfer learning algorithm, dynamically correcting the baseline drift of the sensor, and outputting calibrated data. A closed-loop adaptive mechanism adjusts sensor parameters and network hyper-parameters through meta reinforcement learning, neural architecture searches and optimizes a model structure, and a differential evolution algorithm updates fluid mechanics boundary conditions. Through cooperation of physical constraint and data driving, the problem of signal distortion in a complex environment is solved, the leakage source positioning precision and robustness are improved, and the method is suitable for the field of hazardous chemical substance storage safety monitoring.

The invention discloses a vehicle intelligent driving decision optimization method based on deep learning. The method belongs to a vehicle intelligent driving decision technology. Firstly, various sensors are installed on a vehicle to collect data, and data sets are divided according to a proportion after cleaning and marking. A model containing an encoder-decoder and an attention mechanism is constructed, the weight is initialized, and a hyper-parameter and anti-overfitting strategy is set. And training the model by using a specific algorithm, calculating a loss value and evaluating and adjusting according to a verification set. After the model is integrated to the system, a preliminary candidate decision is determined according to a probability threshold, the decision is optimized in combination with a vehicle surrounding environment information set and a risk assessment function, and decision probabilities are redistributed in real time according to environment changes by means of an environment perception feedback mechanism. The method overcomes the defects of the traditional technology, utilizes the advantages of deep learning, effectively improves the driving decision precision, enhances the adaptability of the system to complex road conditions and environments, and guarantees the safe and efficient operation of intelligent driving.

Disclosed are a multilingual event causality identification method and a multilingual event causality identification system based on meta-learning with knowledge. The method includes: partitioning a to-be-processed multilingual dataset into a plurality of sub-datasets; tagging target events corresponding to all data samples in the sub-datasets; obtaining background knowledge for the various events of different language types and combining the background knowledge with a corresponding data sample; basic input samples of all non-target languages in the basic input sample constituting a plurality of basic datasets, and all the remaining basic input samples constituting a target dataset; partitioning support sets and query sets; constructing a multilingual causal identifier; training the multilingual causal identifier based on the support sets and query sets in the plurality of basic datasets to obtain an optimal hyper-parameter; and training the multilingual causal identifier by using the support set in the target dataset, to obtain an optimal non-hyper-parameter.

The invention discloses a life prediction method based on health index construction and neural network fusion, and belongs to the technical field of equipment state monitoring and predictive maintenance. According to the method, through multi-source degradation feature extraction, common dynamic principal component analysis (CDPCA) dimensionality reduction, health index construction and normalization, deep learning multi-model modeling, integrated learning fusion and Bayesian optimization hyper-parameter optimization, online health assessment and residual life prediction of the equipment part degradation process are realized. Specifically, the method comprises the following steps: firstly, extracting time domain, frequency domain and time-frequency domain features from a sensor acquisition signal, and performing dimension reduction through CDPCA to obtain effective degradation characterization; then, weighting the main features to construct a health index (HI) curve, optimizing the weight through a genetic algorithm, and then performing normalization; a plurality of neural network models such as CNN, Bi-GRU, Bi-RNN, Bi-LSTM and SRNN are constructed based on the normalized HI sequence, and degradation trend modeling is realized; inputting the output results of the neural networks into an integrated learning module for fusion optimization; and finally, carrying out automatic optimization on the key hyper-parameters of the model by utilizing Bayesian optimization. In the equipment operation process, a normalized HI curve can be calculated in real time and input into the fusion model, and the residual life estimation value of the part is dynamically output. According to the method, high-precision, high-robustness and online life prediction can be provided under complex working conditions, the safety and reliability of equipment operation and maintenance are improved, and the method has wide engineering application value.

The invention discloses a lithium battery charge state estimation method based on a Bayes-TLCO optimization deep neural network, and belongs to the technical field of battery state monitoring. The method comprises the following steps: firstly, preprocessing a lithium battery charging and discharging data set; then, constructing a deep neural network model comprising a convolutional neural network, a long-short-term memory network and a multi-head attention mechanism, dynamically optimizing hyper-parameters of the model by using a Bayesian optimization-assisted termite life cycle optimization algorithm, introducing Bayesian optimization during iteration stagnation in a TLCO algorithm iteration process, and finally obtaining a termite life cycle optimization model; fitting historical data through a Gaussian process to construct a search empirical model, generating high-value sampling points, and accelerating model hyper-parameter convergence to a globally optimal solution; and finally, estimating the state of charge of the lithium battery. The method breaks through the limitation of a single algorithm, achieves the high-precision estimation of the state of charge of the lithium battery under a complex working condition, effectively improves the model training efficiency, is suitable for electric vehicles, energy storage systems and other scenes, and provides a key technical support for the intelligent upgrading of battery management.

The invention relates to the technical field of multi-modal medical image registration optimization methods, in particular to a medical image registration optimization method and system, and the method comprises the following steps: inputting a medical image data set comprising an MRI image data set, a CT image data set and a PET image data set into a trained deformable registration network; obtaining a deformation field by using the deformable registration network; obtaining a final registration result of the two groups of medical image data based on the deformation field, obtaining an optimized final registration result through an iterative optimization algorithm, and obtaining a quality evaluation coefficient at the same time; constructing an optimized objective function by using the quality evaluation coefficient, and realizing the adjustment of the hyper-parameter of the deformable registration network through the back propagation of the optimized objective function; prediction registration is carried out by using the deformable registration network after hyper-parameter adjustment, so that an optimized registration result is obtained, the position and the shape of a convolution kernel are adaptively adjusted based on a feature extraction network of deformable convolution, and nonlinear deformation of a local anatomical structure is better captured.

The invention provides a tunnel excavation ground surface settlement prediction method and system based on a machine learning hybrid model, and relates to the technical field of tunnel engineering and machine learning crossing, and the method comprises the steps: obtaining the multi-source heterogeneous information of a target tunnel, and constructing a ground surface settlement data set; a Transform-BiLSTM hybrid model is constructed, the robustness of the algorithm in a noise environment is enhanced based on a VMD (variational mode decomposition) algorithm, hyper-parameters are adaptively adjusted and optimized by using a PSO (particle swarm optimization) algorithm based on a ground surface settlement data set, the model prediction precision is maximized, and a ground surface settlement prediction model is obtained; and analyzing decision logic of the ground surface settlement prediction model through the SHAP value, and outputting interpretable engineering guidance suggestions. By constructing a machine learning hybrid model, high-precision and real-time prediction of ground surface settlement in the whole process of tunnel excavation is realized. The precision and generalization ability of the model are improved, the characterization ability of complex spatial-temporal characteristics is enhanced, and overfitting is avoided; and the interpretability is optimized, and the influence of key parameters on a prediction result is quantified, so that construction parameter adjustment is guided.

The invention relates to an advanced gastric cancer survival prediction system based on Lasso regression, Cox regression and an interpretable machine learning technology, and belongs to the technical field of medical artificial intelligence and intelligent decision support. According to the system, by collecting multi-modal clinical data (including demographic information, TNM staging, treatment modes, tumor grading and the like) of a patient, survival-related variables are screened by adopting Lasso regression and a Cox proportional risk model, and an optimized feature set is constructed. Based on the feature set, the system integrates various mainstream machine learning algorithms (such as XGBoost, Random Forest, SVM, Logistic regression and the like) to construct a prediction model, compares the performance of each model, and selects a model with an optimal effect as a main model. And hyper-parameter tuning is performed on the model through grid search and cross validation, so that the precision and generalization ability of the model are improved. An SHAP interpretability analysis method is introduced into the system, transparent interpretation is carried out on a model output result from the global level and the individual level, and the importance and directional effect of all variables in survival prediction are determined. Finally, the model is deployed on a terminal device, a doctor is supported to automatically output the survival probability and an explanation result after inputting patient information, and a reference basis is provided for clinical treatment decision and personalized management. The system has the advantages of high prediction precision, high interpretability, convenience in use, sustainable optimization and the like, is suitable for clinical aid decision-making scenes, and has good application prospects and popularization values.

The invention discloses a drug target activation and inhibition relation prediction method based on a depth map neural network, and aims to improve the modeling precision and prediction performance of an activation or inhibition action mechanism between a drug and a target. According to the method, on the basis of a fine-grained graph interaction modeling mechanism, multi-scale structural characteristics of drug molecules and three-dimensional space structural information of protein residue levels are fused, and a heterogeneous interaction graph between drugs and proteins is constructed. The method comprises the following steps: firstly, acquiring a drug-target sample with an activation / inhibition tag through a public database, predicting a protein structure by utilizing AlphaFold2, and constructing a protein residue map and a drug molecular map; multi-scale structure semantic representation is obtained through sub-graph decomposition, atomic-scale feature extraction and graph neural network coding of drug graph features; protein graph node features are combined with context embedding generated by a pre-training language model, DSSP coding, secondary structure spectrum and atomic structure features are constructed, and edge features are designed based on the geometrical relationship between residues. Then, based on constraints such as spatial distance and biochemical similarity, a fine-grained mapping relation between drug atoms and protein residues is established, an interaction graph is constructed, and coding is carried out through a GraphSAGE network; and finally, fusing the interacted multi-source embedding, and completing the prediction of the activation / suppression relationship through a multi-layer perceptron. A cross entropy loss function, an Adam optimizer and hyper-parameter grid search are adopted in model training; in the evaluation stage, five-fold cross validation and an independent test set are adopted, and indexes such as the accuracy rate, the recall rate, the F1 score, the specificity and the Morse correlation coefficient are used for comprehensively evaluating the performance of the model. Experimental results show that compared with an existing method, the method has the advantages that the prediction accuracy and mechanism interpretability are remarkably improved, and the method has good generalization ability and application prospects and is suitable for multiple fields of drug action mechanism research, new drug discovery and the like.

The invention provides a sunlight greenhouse intelligent environment prediction and regulation method comprising the following steps: collecting greenhouse data and meteorological data outside a greenhouse, combining to form a greenhouse environment data set, and carrying out data noise processing, screening, correlation analysis and normalization processing to obtain a prediction data set; constructing and training a greenhouse temperature and humidity prediction model comprising an input layer, a convolutional neural network layer and a gating circulation unit neural network layer; inputting the prediction data set into a greenhouse temperature and humidity prediction model, and training, verifying and adjusting hyper-parameters of the greenhouse temperature and humidity prediction model to obtain a greenhouse temperature and humidity prediction result; and constructing a greenhouse environment control module, and adjusting the greenhouse environment based on the optimal temperature and humidity information of crop growth and the greenhouse temperature and humidity prediction result. According to the intelligent prediction and control method, the greenhouse control precision and management efficiency can be remarkably improved, modernized and intelligent development of agricultural production is achieved, crop growth is promoted, and losses caused by extreme weather are reduced.

The invention provides a reinforcement learning path planning method for multiple unmanned vehicles in an intelligent factory. The method comprises the following steps: building a feeding path planning simulation environment; taking each unmanned vehicle as an intelligent agent, interacting the intelligent agent with the environment to extract information, and constructing a state space, an action space and a reward function of the Markov decision model according to the information; building an IPPO algorithm network, and training the strategy of the Markov decision model based on the IPPO algorithm network and the reward function; during training, a strategy network loss function with a rollback strategy is obtained through improvement by combining cutting loss with KL divergence in a mode of introducing hyper-parameters; and determining the current state at each moment according to the actual current position of the unmanned vehicle, inputting the current state into the strategy network, and outputting the current action to be executed by the unmanned vehicle. The method is high in convergence speed, stable in strategy updating, high in exploration capacity, capable of effectively reducing path blockage and high in environment adaptability.

The invention discloses a quasi-brittle material damage field inversion method based on a physical information neural network. The method comprises the steps of data acquisition and processing; presetting initial damage field data, and taking the modulus of each unit in the initial damage field data as an independent to-be-inverted parameter; taking the position information of each node in the space as the input of a neural network, processing through a hidden layer of the neural network, and taking displacement data corresponding to the position information as network output data; constructing a loss function of a neural network according to the preprocessed displacement data, network output data and a mechanical law; and performing training optimization on the initial damage field data by using a loss function of a minimized neural network, optimizing hyper-parameters of the neural network by adjusting weights of control item loss, boundary item loss and data item loss until a preset convergence condition is met, and determining target damage field data. According to the method, the dependence of a neural network model on a data set is remarkably reduced, and the inversion result is ensured to have relatively high physical interpretability.

The invention discloses a dynamic weight aggregation federated learning method and system for credit risk assessment, and the method comprises the steps: firstly, calculating the cosine similarity of each client model and the last gradient of a global model, so as to assess the consistency of model parameters; secondly, the distribution difference is quantified by calculating the data statistic difference between each client data and the global data, so that the diversity of the data is evaluated; further, a contribution degree comprehensively considering model parameter similarity and data distribution difference is obtained by combining a hyper-parameter with cosine similarity and data distribution difference; and finally, dynamically adjusting the weight of each client in global model updating according to the contribution degree of each client, so that the model pays more attention to the client with higher data quality, thereby improving the overall model performance. The federated learning method is especially suitable for processing non-independent identically distributed data, optimizes the generalization ability of the model, and is suitable for credit risk assessment in the financial field.

The invention discloses a wind power plant station AGC parameter optimization method based on a GWO-LSTM neural network. The method comprises the steps that a regional power grid frequency model containing a wind power plant station is established; constructing an LSTM neural network model, obtaining power grid system state data according to the regional power grid frequency model, and initializing the LSTM neural network model; performing optimization adjustment on hyper-parameters of the LSTM neural network model by using a GWO algorithm to obtain a GWO-LSTM neural network model, and performing offline training on the GWO-LSTM neural network model; a state vector and a control vector of a power system are obtained, prediction time domain parameters of model prediction control are calculated by adopting a GWO-LSTM neural network model, and optimal parameter combinations under different working conditions are obtained, so that the parameters of model prediction control are adaptively adjusted so as to output an optimal frequency control instruction, and the frequency stability of a wind power plant station under different disturbances is ensured.

The invention discloses a tight reservoir three-dimensional crustal stress field modeling method based on an improved neural network. The tight reservoir three-dimensional crustal stress field modeling method comprises the steps that S1, a unified-format multi-source geological physical data tensor set is constructed; s2, constructing a frequency domain hierarchical enhancement-SIREN implicit neural network structure based on the unified format multi-source geological physical data tensor set; s3, inputting the candidate hyper-parameter configuration into the frequency domain hierarchical enhancement-SIREN implicit neural network to complete one-time model training; s4, aiming at each candidate hyper-parameter configuration, initializing an inner-layer population of the black widow optimization algorithm, completing second model training, and obtaining an optimal model parameter of the frequency domain hierarchical enhancement-SIREN implicit neural network; s5, tight reservoir fracturing parameter optimization and real-time safety window adjustment are achieved. According to the method, the continuous stress field can be quickly generated at the resolution of 1 m, real-time well section updating and fracturing scheme optimization are supported, and the fracturing transformation effect, the fracturing safety margin and the reliability of economic productivity prediction are remarkably improved in practical application.

The invention discloses a water environment dynamic pollution monitoring and atmospheric diffusion prediction method based on deep learning, and the method comprises the following steps: S1, collecting multi-source monitoring data of a water environment and an atmospheric environment, and constructing a pollution time series data set and an atmospheric meteorological data set; s2, constructing a variational auto-encoder model; s3, jointly optimizing a variational auto-encoder model structure and hyper-parameters by using a hosting crab optimization algorithm; s4, the optimized variational auto-encoder model is used for training, and potential variables of the pollution state are extracted; s5, constructing a diffusion modeling network, and outputting a diffusion prediction result; and S6, comparing a diffusion prediction result with a pollution threshold value, and triggering early warning information output when the diffusion prediction result exceeds the standard. According to the invention, intelligent monitoring of the water environment pollution state and accurate prediction of the diffusion behavior of pollutants to the atmosphere are realized, and efficient decision support is provided for environment early warning and treatment.

The invention relates to the field of electric power system operation, and discloses an operation cost prediction and adjustment method suitable for a multi-scene electric power system, and the method comprises the following steps: data collection: collecting basic data related to the operation cost of the electric power system; storage management: constructing a hybrid architecture, storing the basic data, establishing a data quality management system, and regularly maintaining the data; fusion analysis: extracting and standardizing key features of the basic data, building a basic data fusion model based on a deep learning framework, and outputting fusion features; and verification optimization: dividing a training set and a test set by adopting cross validation, adjusting hyper-parameters and a structure of the fusion model, and finely adjusting the model in real time. Effective storage and high-quality maintenance of data are ensured through data acquisition and storage management, accuracy and adaptability of a model are guaranteed through fusion analysis and verification optimization, various requirements of the system are balanced through multi-objective optimization and evaluation utilization, and scientific evaluation and collaborative utilization of flexible resources are achieved.

The invention provides a sequential network traffic prediction method and system based on swarm intelligence parameter optimization, and relates to the technical field of network traffic prediction. The method comprises the following steps: acquiring indexes such as throughput packet loss rate and round-trip delay of a target link by using a network probe, and performing deletion filling normalization and multi-scale decomposition to obtain a standardized traffic sequence; calculating information entropy, constructing a traffic complexity feature vector, and dividing a training set and a verification set; constructing a hybrid depth prediction model composed of a one-dimensional convolutional network and a gating cycle unit, and establishing a hyper-parameter search space; using particle swarm optimization and entropy-driven inertia weight adjustment and mutation probability mapping to reconstruct a speed and position updating strategy, and iteratively outputting a global optimal hyper-parameter; and generating a benchmark prediction result according to full-amount training, extracting a residual error, training a nonlinear residual error compensation model to carry out superposition correction and reverse normalization, obtaining a final flow prediction result, and improving prediction precision and generalization ability.

The invention relates to the field of bearing fault detection, in particular to a bearing fault diagnosis method based on time-frequency enhancement CNN-Transformer, which comprises the following steps: acquiring a bearing vibration signal, and preprocessing the bearing vibration signal; selecting a window size, an interval step length and an overlapping rate, and dividing the preprocessed bearing vibration signal data into a training data set, a verification data set and a test data set by adopting a sliding window; an FFT-CNN-Transform network model is constructed, and the FFT-CNN-Transform network model is constructed; an Adam optimizer and a cross entropy loss function are adopted to train an FFT-CNN-Transform network model, hyper-parameters of the FFT-CNN-Transform network model are adjusted through a verification data set, and the performance of the FFT-CNN-Transform network model is evaluated through the test data set; and by drawing a confusion matrix, the classification effect of the FFT-CNN-Transform network model on different fault categories is displayed. According to the method, key features of a time domain and a frequency domain can be captured at the same time, and data features are compared with fault features, so that more accurate fault prediction is realized, and the accuracy and reliability of bearing fault diagnosis under complex working conditions are remarkably improved.

The invention relates to the technical field of subgrade settlement prediction, in particular to a subgrade settlement prediction method based on deep learning, electronic equipment and a medium, the subgrade settlement prediction method provided by the invention adopts a subgrade settlement prediction model for prediction, and the model adopts the combination of a convolutional neural network and a bidirectional long-short term memory network. Spatio-temporal features are effectively captured, and the prediction precision and stability of the model are improved; by introducing a channel attention enhancement mechanism module, dynamic weighting is performed on different feature channels, the selection ability of the model to key features is improved, and the sensitivity of the model to subtle changes in data and the capture ability of the model to global features are enhanced; the model is trained through the genetic algorithm, the efficiency is improved, and the genetic algorithm can automatically find an optimal combination in a complex hyper-parameter space, so that the model can show good generalization ability on different roadbed settlement data sets, and the prediction precision of the model is improved.

The invention discloses a transformer current rising test data analysis method based on deep learning, and relates to the technical field of power transformation equipment, and the method comprises the following steps: S1, data collection and preprocessing; s2, performing feature engineering; s3, dividing a data set; s4, model selection and architecture design; s5, model training; s6, performing model verification and hyper-parameter tuning; s7, testing and evaluating the model; s8, performing anomaly detection and adaptive learning; s9, result interpretation and model interpretability; and S10, continuously monitoring and updating the model. According to the transformer current rising test data analysis method based on deep learning, based on deep learning, in combination with technologies of feature engineering, model selection and architecture design, anomaly detection and adaptive learning and the like, the accuracy and robustness of a transformer current rising test are improved through comprehensive data analysis; the generalization ability of the model for unseen data can be improved, and the model can adapt to environmental changes in real time.

The invention belongs to the technical field of robots, and discloses a feature fusion and refining embedding sparse Bayesian learning method and system for robot curved surface milling contour error monitoring, and the method comprises the steps: carrying out the structural fusion of feature vectors composed of technological parameters, robot rigidity, cutting force and tracking errors into a dictionary matrix; and a sparse Bayesian learning method is combined to realize robot processing error monitoring. During model training, the posterior distribution of the weight variables can be obtained, and the expectation and variance of the posterior distribution of the weight coefficients are estimated by maximizing the posterior probability. Wherein the variance and the hyper-parameter influencing the weight sparsity are estimated by maximizing an edge likelihood function, that is, optimization is carried out by the expectation of a joint likelihood function under maximized weight posterior distribution, and the minimum value solution can be solved for the converted loss function to optimize the hyper-parameter. The posterior expectation and variance of the model weight and the hyper-parameter are iteratively updated in the training process, and the iteration process is terminated according to a set change threshold value so as to realize the sparsification of the model weight vector and the construction of the sparse model. According to the method, robot milling error monitoring of curved surface parts with different features under the variable poses is achieved, and the part measurement cost is effectively reduced.

The invention discloses a charging station loss prediction method based on an LSTM deep learning method, and relates to the technical field of charging stations, and the method comprises the following steps: 1, collecting data including real-time power load data, a power grid topological structure and time sequence data; 2, feature engineering: carrying out preprocessing and feature extraction on the data, and extracting expert knowledge features; step 3, establishing an LSTM model in combination with a PPO algorithm; 4, performing hyper-parameter tuning on the LSTM model through a random search method; and step 5, the optimized miniaturized model is then deployed to each charging station, and local real-time energy loss prediction is realized. In order to further improve the prediction accuracy, expert knowledge is creatively integrated in a feature engineering part, multi-dimensional loss feature recognition is carried out, and the performance of the model in processing a complex nonlinear relation is improved by combining the input of an electric power system expert, recognizing factors such as equipment aging and the like.

The invention belongs to the field of food detection, and discloses a pork tenderloin freshness multi-modal fusion detection method based on deep learning, which comprises the following steps: making a sample of a pork tenderloin freshness classification model, collecting HSI data, ultraviolet fluorescence image data, electronic nose data and LF-NMR data in the pork tenderloin storage process, and preprocessing the data; respectively designing feature extraction modules for the ultraviolet fluorescence image data and the HSI data; designing a data enhancement module based on a variational auto-encoder to enhance the sample freshness boundary recognition capability of each category; designing an improved dung beetle optimization algorithm to enhance the hyper-parameter optimization ability of the model; a UV-CNN module, an HSI-CNN module, an AAVAE module and an EDBO optimization algorithm are introduced into a BiLSTM network structure, and a CNN-AAVAE-EDBO-BiLSTM classification model of the freshness of the pork tenderloin is constructed; and testing and evaluating the accuracy of the pig tenderloin classification model.

The invention relates to the technical field of tunnel engineering monitoring, in particular to a tunnel abnormal data early warning method based on an LSTM model. Comprising the following steps: S1, data preprocessing; s2, building a model, wherein the structure of the model comprises two layers of LSTM networks and one layer of full connection layer; s3, model hyper-parameters are selected; s4, network training: performing parameter iteration training through optimization calculation and loss calculation; s5, establishing a dynamic prediction model: continuously inputting measured data, and comparing the measured data with a mean square error standard value to judge whether parameters are retrained or not; and S6, early warning judgment: calculating a difference value between a predicted value and a measured value of the LSTM model, comparing the difference value with a preset threshold value, and carrying out graded early warning. Compared with the prior art, the method has the advantages that the prediction curve is more stable, the calculation standard deviation is larger, the prediction confidence interval range is wider, the change of the data trend can be predicted in advance, the change of the data trend can be identified in advance, and the early warning advancement is better.

The invention provides a method for estimating large language model training and reasoning time consumption, a hyper-parameter configuration reference is provided for large model training fine tuning, and a calculator can help a user to simplify the large model training process through automatic analysis and giving of appropriate parallel configuration requirements; a user does not need to manually research and try different configurations, but can select proper configurations according to suggestions of the calculator; according to the method, the computing power and time cost required by large model training are estimated, and a user can better understand and evaluate the cost and benefit of large model training by estimating the core index of large model training, so that the user can more easily try to use, and large model vertical application landing is promoted; gPU models and large models can be flexibly added, and the calculator not only can simulate training indexes under given configuration, but also can provide reference for a user to customize a model training plan. The user can adjust the configuration parameters in the training plan according to the estimation result of the calculator so as to achieve a better training effect and a better resource utilization rate.

The invention discloses a water turbine wear fault diagnosis method based on a deep learning model, and belongs to the technical field of water turbine fault signal processing. Comprising the following steps: acquiring a water turbine wear fault sound signal; a Tent chaotic mapping function is adopted to optimize an initial population of a white shark optimization algorithm WSO; introducing a long and short term memory (LSTM) network model to solve the limitation of a convolutional neural network (CNN) in processing a long sequence problem, and constructing a CNN-LSTM water turbine wear fault diagnosis model; an improved white shark optimization algorithm IWSO is introduced into the CNN-LSTM model, and hyper-parameters of the model are optimized, so that the diagnosis precision is improved, and the calculation amount is reduced; and importing the collected fault signal as an input into the model for training and testing to obtain a final diagnosis result. Diagnosis and identification of the water turbine wear fault signal are completed through the steps, the deep learning model has good robust performance, the diagnosis precision of the water turbine wear fault can be improved, and the deep learning model serves as a beneficial supplement of an existing water turbine state monitoring and fault diagnosis system.