567 results about "Predictive value" patented technology

The invention relates to the technical field of pipeline safety monitoring, and discloses an online lossless real-time monitoring system for micro-strain of an in-service natural gas pipeline. A micro-strain data acquisition unit of the system acquires a micro-strain data set on the surface of the in-service natural gas pipeline in real time. And the three-dimensional strain field reconstruction unit receives the data set and executes three-dimensional strain field reconstruction processing to generate strain distribution characteristics of the pipeline. And the life prediction model analysis unit calls a pre-trained life prediction model to carry out nonlinear analysis processing on the strain distribution characteristics, and outputs a residual life prediction value and a key risk area identifier of the pipeline. The environmental factor compensation unit performs environmental factor compensation correction processing on the residual life prediction value to generate a corrected residual life prediction value. And the maintenance strategy generation unit generates a pipeline maintenance strategy set according to the key risk area identifier. According to the invention, real-time and accurate evaluation and intelligent maintenance decision support of the health condition of the pipeline are realized.

The invention relates to the technical field of disaster prevention and reduction of a power system, and discloses an icing risk early warning method based on multi-model fusion and residual time sequence characteristic analysis, which comprises the following steps: collecting meteorological data of a line area in real time, removing abnormal values through secondary judgment of a Pauta criterion and a trend, and standardizing; adopting a TEROL algorithm to screen high-weight key features; running SWD-BP, MUL-GRNN and ELM models in parallel, constructing a dynamic weight by combining DSI, an independence weight method and an entropy weight method, and calculating a final meteorological predicted value; generating a prediction residual signal, extracting time domain features such as a mean value and a peak value, and constructing a residual feature matrix through a sliding window; and inputting an LSTM model to process a time sequence dependency relationship, and judging an icing risk level. According to the method, meteorological prediction is optimized through multi-model dynamic fusion, and deviation is analyzed and corrected in combination with residual time sequence characteristics, so that the problem of weak generalization ability of a single model is effectively solved, and the accuracy of icing risk early warning is obviously improved.

The invention relates to the technical field of medical data processing, in particular to a lung cancer risk early warning system based on remote four diagnosis information, which comprises an acquisition module, a generation module, a calculation module, an early warning module and a correction module. According to the method, through multi-modal time sequence fusion and causal map driven closed-loop regulation and control, the more stable, interpretable and adaptive early warning capability for the lung cancer risk is realized, heterogeneous evidences from observation, listening and interrogation are aggregated into a dynamic alarm index according to time and causal paths, instantaneous noise is filtered by using trend consistency and adversarial test, and the early warning effect is improved. Therefore, false alarms caused by short-time fluctuation are remarkably reduced while the sensitivity to real abnormity is kept; the method achieves balance among false alarm reduction, missed diagnosis risk reduction, positive prediction value improvement and user sampling burden optimization, and effectively solves the problem of insufficient early warning accuracy caused by risk identification lag due to dependence on single static medical record information.

The invention belongs to the technical field of metallurgical industry process intelligent control and prediction, and discloses a self-adaptive mixed intelligent prediction method and system for smelting endpoint parameters of an electric arc furnace. Acquiring smelting process data of the electric arc furnace; constructing a dual-drive prediction system comprising a mechanism model and a data drive model; calculating a decision coefficient of a prediction value and an actual measurement value of the theoretical model, and counting an effective historical data volume; constructing a machine learning prediction model, and selecting a modeling algorithm according to the effective historical data volume; selecting a hybrid prediction strategy based on the decision coefficient and the effective historical data volume; predicting and outputting an end point carbon content predicted value and an end point temperature predicted value according to a hybrid prediction strategy; predictive deviation threshold value judgment and execution control are conducted, and the electric arc power, the oxygen blowing flow, the feeding speed or the cooling water flow are adjusted. Accurate prediction and dynamic optimization control of the electric arc furnace end point parameters are achieved, and the smelting quality, the energy utilization rate and the production stability are remarkably improved.

The invention discloses a power equipment health evaluation method and system based on time sequence analysis and probability modeling, equipment and a medium, and relates to the technical field of power equipment state monitoring and fault prediction, and the method comprises the steps: obtaining and preprocessing multi-source operation data of power equipment, outputting a predicted value and a confidence interval of a future parameter through a time sequence prediction model, calculating a residual sequence of an actual observation value and a predicted value, fitting distribution through a probability distribution model, establishing a statistical characteristic model of a normal operation state of the equipment, performing anomaly judgment, calculating a health degree index of the equipment based on a deviation degree and a dynamic weight of a monitoring parameter and weighted accumulation, and dividing equipment state grades according to the index. Quantitative evaluation of the health state of the equipment is realized. According to the method, accurate quantification and early abnormity identification of the health state of the power equipment are realized, a reliable basis is provided for predictive maintenance, and the intelligent level and the safety guarantee capability of power grid operation and maintenance are remarkably improved.

The invention discloses a bridge multi-source damage evolution prediction method based on a knowledge graph, and relates to the technical field of bridge loss prediction, and the method comprises the steps: employing the combined modeling capability of a multi-source monitoring feature set Mon, a damage evolution stage node set Dag, a bridge damage knowledge graph Gra and a causal chain Chain; the prediction process of bridge damage evolution is converted from single-moment numerical prediction to a structured reasoning process oriented to multi-dimensional features, multiple damage stages and a chain type trigger relation. By means of a time slice feature vector predicted value VecPred and a stage category identification predicted value LabPred obtained through time sequence diagram convolution reasoning, which stage the damage may evolve to at the next moment can be revealed in advance in the bridge operation process, and whether the damage may be further propagated along a chain path or not is clearly predicted in combination with a chain risk result Risk output by a causal chain Chain. The chain type damage evolution can be recognized in advance in the actual operation scene of the bridge.

The invention relates to the technical field of time sequence data processing and deep learning, in particular to a time sequence filling method and system based on a coarse-to-fine filling normal form. In the invention, a preprocessing module is used for sampling an input sequence to obtain a subsequence rk with the length of Lk, and a regression prediction module is combined with a causal mask to generate a prediction value r'k with the same structure and the same length as the rk based on all subsequences {r1... rk} generated by the preprocessing module; the correction module performs up-sampling on the predicted value r'k to obtain an up-sampling sequence r ''k with the length of T; supplementing missing data of the original sequence based on the r ''k to obtain a corrected sequence X (k + 1); and traversing k = 1... K to obtain a correction sequence X (K + 1) as a final repair completion time sequence. The method overcomes the defects that the time sequence filling mode in the prior art does not consider the unfixed missing rate and missing value block distribution, and is beneficial to improving the filling accuracy.

The invention relates to the technical field of medical image diagnosis, and discloses a cardiovascular disease diagnosis model construction method based on image processing. The method comprises the steps that cardiac medical image data of a target patient is collected, a standardized data set is generated through preprocessing, and a morphological and hemodynamic feature set is extracted; establishing a heart state evolution characteristic spectrum according to a characteristic dynamic evolution rule, dividing a pathological state space, and calculating the characteristic distribution density of a historically diagnosed case; acquiring real-time image data of a patient to be diagnosed, and constructing a real-time diagnosis feature vector; mapping the vector to a pathological state space, and calculating a space matching degree to generate a pathological association index; and combining the association index and the two types of feature sets to construct a heart pathology probability prediction model, outputting a pathology probability prediction value and generating a hierarchical diagnosis suggestion. According to the method, through multi-dimensional feature analysis and space matching analysis, precise and graded diagnosis of the cardiovascular diseases is realized, and an efficient and feasible technical path is provided for diagnosis of the cardiovascular diseases.

The invention provides an electric quantity prediction method and system fusing physical constraint factors, and relates to the technical field of electric quantity prediction. Historical load, weather, electricity price and calendar data are collected, and a key feature set is constructed through preprocessing and feature selection; a prediction model with the physical information neural network as the core is constructed, the prediction model comprises a recursion sub-module used for short-term prediction and a trend sub-module used for long-term prediction, and a physical constraint loss item based on a physical rule is introduced into model training so as to enhance the generalization ability; a multi-time granularity modeling framework is adopted, uncertainty quantization is achieved through a Monte Carlo Dropout or Bayesian neural network, and a confidence interval of a predicted value is output; and finally, causal reasoning is carried out through a Shapley value algorithm and anti-fact simulation, and key influence factors are identified. According to the method, the precision, stability and interpretability of electric quantity prediction are effectively improved, and reliable support is provided for power grid dispatching and decision making.

The invention provides a heart rate and breath prediction method, a program product, electronic equipment and a storage medium. The method comprises the following steps: acquiring an original signal sequence acquired from a human body through an ultra-wideband radar; determining a thoracic cavity displacement signal according to a mapping relation between the phase change and thoracic cavity displacement; performing multi-domain feature extraction on the thoracic cavity displacement signal to obtain multi-domain features; the multi-domain features comprise time domain features, frequency domain features and time-frequency domain features; on the basis of the multi-domain features, prediction values are obtained through a pre-trained prediction model, and the prediction values comprise the heart rate and the respiratory rate in the future time period. The robustness of the algorithm is enhanced through multi-dimensional feature fusion, high precision is still kept in an electromagnetic interference or multi-target scene, and the reliability of heart rate and respiratory rate prediction in a complex environment is improved. The physical signal advantage of the ultra-wideband radar is complementary with the environmental adaptability of the neural network, and the prediction accuracy of the heart rate and respiratory rate is improved.

The invention discloses a foundation pit supporting strength self-adaptive adjusting method and system based on real-time monitoring data, and relates to the technical field of engineering control, and the method comprises the following steps: collecting monitoring data of a foundation pit supporting structure in real time; preprocessing the monitoring data to obtain standardized time series data; inputting the standardized time sequence data into a pre-trained LSTM prediction model, and outputting a displacement prediction value sequence; calculating a prediction state index based on the displacement prediction value sequence; with minimization of the maximum prediction state index and minimization of the total support strength adjustment cost as targets, a multi-target optimization model is constructed, and the optimal prestress adjustment amount of each intelligent support execution mechanism is solved; and the optimal prestress adjusting quantity is converted into a control instruction, and a corresponding intelligent supporting execution mechanism is driven to act. The foundation pit supporting strength can be automatically and accurately adjusted according to data monitored in real time.

The invention discloses a bone infection and drug resistance prediction method and system fusing a knowledge graph and a graph convolutional network. The method comprises the following steps: preprocessing bone infection multi-source heterogeneous data to obtain a standardized data feature matrix; constructing a bone infection knowledge graph based on the matrix and obtaining a knowledge embedding matrix, and fusing the two to generate a medical semantic constraint fusion feature matrix; key medical variables are screened, the maximum information coefficient (MIC) of the key medical variables is calculated, and an adjacent matrix is constructed in combination with medical association strength factors; inputting the fusion feature matrix and the adjacent matrix into a GCN spatial feature extraction module and a BiGRU time sequence dependence capture module to obtain spatial and time sequence features, and fusing the spatial and time sequence features into a space-time fusion feature matrix; and inputting the result into a double-task prediction module, and outputting a bone infection and drug resistance grading probability prediction value. According to the invention, accurate and rapid prediction of bone infection prediction and drug resistance grading can be realized.

The invention provides an intelligent prediction and management method for the usage amount of service consumables, and relates to the technical field of data processing, and the method comprises the steps: processing multi-dimensional data, and dividing a consumable region into patient one-to-one consumables and shared consumables according to the consumption characteristics of the consumables, so as to obtain a consumable classification result; according to one-to-one consumables of patients in the consumable classification result, a first dosage predicted value is obtained by combining real-time outpatient registration data; according to the shared consumables in the consumable classification result, a second dosage predicted value is obtained in combination with operation habit parameters and consumable specification variables, and the habit parameters reflect stable preferences of specific departments, medical groups and medical staff when the shared consumables are used; the consumable specification variable refers to a physical attribute characteristic influencing the single consumption of the shared consumable. According to the method, the consumable consumption prediction precision is improved, and the clinical supply shortage risk and the expired consumable waste are reduced.

The invention discloses a fermentation strain yield prediction method based on artificial intelligence, particularly relates to the field of biological fermentation production, and is used for solving the problems that in the prior art, due to the fact that the yield is difficult to directly map in concentration prediction, monitoring indexes are normal, the final yield deviates from expectation, and reliable strain comparison and production management decision cannot be supported. A historical batch process observation sequence and a material account sequence are aligned to generate a conversion efficiency characterization sequence to form a yield learning sample training yield prediction model, and the conversion efficiency characterization sequence is generated on batches and input into the model to obtain a yield prediction value. And calculating an accounting volume trajectory consistency parameter and a conversion contribution steady-state proportion parameter, inputting the parameters into a coefficient analysis model to obtain a decision credible coefficient, generating a yield risk judgment result for a yield prediction value based on the decision credible coefficient, and realizing a new-batch online prediction record.

The invention provides a method and a system for processing content measurement data of multi-index components in folium cortex eucommiae, and relates to the technical field of traditional Chinese medicine quality control and data analysis crossing. The method comprises the following steps: aligning standardized content data of a current batch with data of a corresponding batch in a predicted content data set; obtaining an actual detection value and predicted value pairing sequence of each index; the deviation ratio of each data pair in the actual detection value and predicted value pairing sequence of each index is calculated, and a deviation ratio analysis result is formed; and based on the deviation ratio analysis result, in combination with a pre-stored legal standard threshold and historical batch data, performing comprehensive evaluation through principal component analysis and analytic hierarchy process to obtain a multi-dimensional quality comprehensive evaluation value including specification conformity, component proportional relation and process stability. According to the method, an integrated data processing system is constructed, so that the intelligence and precision level of multi-index quality evaluation of the folium cortex eucommiae is improved.

The invention relates to the technical field of perforation early warning, in particular to an ESD perforation early warning system and method based on deep reinforcement learning. A data acquisition unit of the system is used for acquiring multi-modal data of a current cutting area, a data processing unit is used for performing data processing on the multi-modal data to obtain early warning processing data, and a data fusion unit is used for performing weight distribution on each piece of data in the early warning processing data and then outputting the early warning processing data. Performing fusion processing on each piece of data in the early warning processing data by using the distributed weight to obtain early warning fusion data; the model processing unit performs perforation risk prediction on the early warning fusion data by using a preset deep reinforcement learning model to obtain a risk prediction value; and the perforation early warning unit judges the risk prediction value by using a preset risk model and then sends out perforation early warning information. According to the invention, real-time and accurate auxiliary decision-making in the operation is provided for medical staff, the perforation rate in the ESD operation is obviously reduced, and the operation safety and efficiency are improved.

The invention discloses a sewage water quality intelligent soft measurement method based on big data and a soft measurement model, relates to the technical field of sewage water quality intelligent soft measurement, and aims to solve the problem that the sensitivity of the model is reduced due to inaccurate water quality analysis. According to a regulation and control link, virtual simulation evaluation is carried out according to a risk matching target and a calling decision scheme, an optimal execution scheme is selected, a soft measurement model construction and optimization link is high in pertinence, a model type is selected according to index characteristics, a data set is scientifically divided to avoid overfitting, errors are monitored in real time in training, and model performance is improved through multi-dimensional optimization. An independent test set ensures the model precision and stability, an optimization model is deployed in a real-time prediction stage, a core factor is rapidly extracted to calculate a predicted value, water quality dynamic perception is realized, traditional detection hysteresis is overcome, a change trend is pre-judged in advance, time is bought for sewage treatment regulation and control, and the response speed and prediction reliability are improved.

A computer-implemented method is provided for monitoring and / or controlling a chemical and / or biological process. The method comprises: obtaining (S10) a time series dataset comprising observations, each one of the observations being collected with respect to the chemical and / or biological process at a particular time point, wherein each one of the observations includes values of observed parameters obtained with a spectroscopic method at the particular time point and a value of an analyte parameter obtained with a reference measurement method at the particular time point; obtaining (S20) a prediction model for estimating a predicted value of the analyte parameter in the chemical and / or biological process, the prediction model being trained using at least part of the time series dataset; validating (S30) the prediction model by assessing an ability of the prediction model to predict a difference between an actual value of the analyte parameter which would be obtained with the reference measurement method at a given time point and an expected value of the analyte parameter at the given time point, using a set of observations that is not included in the at least part of the time series dataset used for training the prediction model; determining (S40) whether the prediction model is valid or invalid based on a result of the validating; and monitoring and / or controlling (S50) the chemical and / or biological process when the prediction model is determined to be valid.

The invention discloses a reservoir flood forecasting method based on physical constraint and space-time double-flow coupling. The reservoir flood forecasting method comprises the following steps: S1, acquiring multi-source hydrometeorological data; s2, decomposing the multi-source hydro meteorological data into a historical state sequence and a future driving sequence; s3, performing feature extraction on the historical state sequence data through the physical enhanced long-short term memory network of the historical inertial feature extraction branch to obtain historical inertial features, and performing feature extraction on a future driving sequence through the time domain convolutional network of the future forced feature extraction branch to obtain future forced features; s4, performing weighted fusion on the historical inertial features and the future forced features to generate fusion features; and S5, inputting the fused features into a decoder to obtain a predicted water level increment, and superposing the predicted water level increment to the current water level to obtain a predicted value. The prediction timeliness is improved, the prediction precision of the water recession stage is improved, and the physical consistency of the prediction result is enhanced.

The invention relates to the technical field of intelligent heat supply, and discloses a load prediction energy-saving regulation and control method for secondary network heat supply. Extreme weather is identified through the multi-source meteorological data, and graded early warning signals are generated; collecting and processing data flow of a heat supply pipe network and an indoor sensor; fusing the data flow and the early warning signal, and generating a thermal load prediction value subjected to physical constraint and dynamic compensation by adopting a hybrid prediction model; based on the early warning level and the predicted value, a multi-objective optimization problem is solved in a rolling mode through model prediction control, and a water supply temperature and water pump frequency setting curve is generated; and finally, after instruction smoothing and safety verification, an execution mechanism is driven to complete accurate regulation and control. According to the invention, by constructing a sensing, predicting, optimizing and executing full-link intelligent regulation and control system, the industrial problem that the rapid mutability of extreme weather is not matched with the slow dynamic response of the system is solved, and the regulation and control quality and the energy efficiency level of the heat supply system are remarkably improved.

The invention relates to the crossing field of artificial intelligence technology and bioinformatics, and discloses a pig feed efficiency prediction model and system based on multi-omics data. Modulating a neural differential equation which runs on a priori knowledge graph and is realized by a graph neural network by using the matrix so as to solve and generate a continuous evolution trajectory of an individual physiological state; and finally, aggregating the tracks, combining the constraint matrix, and outputting a feed efficiency prediction value through a second preset model. The invention further provides a corresponding prediction system which comprises a static constraint module, a dynamic core module and a prediction module. According to the method, static genetic constraint and dynamic physiological process simulation are combined, genetic differences among different individuals can be reflected, and the biological consistency and individualization precision of a prediction model are improved.

The invention discloses a method for integrating an industrial field information system and a cloud platform based on edge computing. The method comprises the following steps: locally acquiring original monitoring data of a multi-source sensor of industrial equipment at an edge node; according to the method, the original data of the multi-source sensor is locally acquired and preprocessed through the edge node, and only the standardized time sequence data set and the feature vector are transmitted to the cloud, so that the original data transmission quantity is greatly reduced, and the network bandwidth pressure and the communication cost are relieved; feature engineering is carried out on the edge side in real time, a predictive maintenance model is deployed to carry out local reasoning, and an abnormal probability or an RUL predicted value is output, so that data processing link delay is remarkably shortened, and the real-time performance of health state assessment is improved; a maintenance decision is quickly triggered based on an evaluation result, so that instant response and accurate maintenance of equipment abnormity are realized, and fault expansion caused by cloud processing delay is avoided; meanwhile, an edge and cloud collaborative optimization model is combined with global data continuous iteration, and prediction accuracy and adaptability are improved.

The invention discloses a power load prediction method and device based on quadratic mode decomposition and double-model parallelism, and a medium, and the method comprises the steps: carrying out the fine decomposition of an original load sequence through a quadratic mode decomposition method, carrying out the denoising and reconstruction through combining with a wavelet threshold method, and finally obtaining a series of stable mode components; in a prediction stage, a parallel prediction architecture of the Informer and the BiLSTM is constructed, all modal components are synchronously input, global long-term dependence is captured by using a multi-head probability sparse self-attention mechanism of the Informer, and local short-term dynamic is captured by using the BiLSTM; and carrying out splicing and nonlinear fusion on the heterogeneous features extracted by the two to obtain a final prediction value. Compared with the prior art, cooperative capture and accurate prediction of the multi-scale features of the non-stationary power load sequence are realized through secondary decomposition from coarse to fine and a targeted double-model parallel architecture.

The invention belongs to the technical field of medical equipment, and provides a bipolar electrotome control method and device, electrosurgical equipment and a storage medium, and the method comprises the steps: obtaining an output voltage, an output current and a tissue temperature; determining a feature vector based on the output voltage, the output current and the tissue temperature; inputting a plurality of feature vector values in the feature vectors into a trained full-connection neural network to obtain a closure degree predicted value, a carbonization risk predicted value and a quality score predicted value; determining a dynamic output power based on the closure degree predicted value, the carbonization risk predicted value, a feature vector and a basic output power; based on the dynamic output power and a preset condition, determining a safe output power through a safe output power calculation formula; controlling the closing action of the bipolar electrotome based on the closing degree predicted value, the safe output power and the closing quality score value; the efficiency of controlling the bipolar electrotome to perform the closing action is improved, the problem of excessive carbonization of tissues is reduced, and the operation safety is greatly improved.

The invention discloses an ancient building settlement situation prediction method and system based on decomposition integration, and belongs to the technical field of cultural relic settlement monitoring and deep learning crossing. The method comprises the following steps: acquiring multi-source settlement monitoring data of an InSAR, a GNSS, a traditional sensor and the like, and generating continuous time sequence characteristic data by preprocessing a data format, complementing missing values and performing standardization; inputting the processed data into a pre-training neural network model based on a Transform architecture, capturing a long time sequence dependency relationship by using a multi-head attention mechanism of the pre-training neural network model, and outputting a future settlement predicted value; and setting a grading early warning threshold according to the type of the ancient building, generating early warning information in combination with a prediction result, and outputting a visual chart containing comparison between a true value and a predicted value. According to the method, efficient processing and long-time-sequence accurate prediction of settlement data are realized, and reliable technical support is provided for early discovery and disposal of settlement diseases of cultural relics and buildings.

The invention provides a postoperative bleeding risk monitoring method and system for a hepatobiliary patient. The method comprises the following steps: collecting real-time multi-modal data; constructing an LSTM-CNN hybrid model by using the time-frequency decomposition features, and obtaining a local tissue hypoxia index and a vasomotor function anomaly probability; the low-frequency impedance change rate and the albumin level are fused through a random forest algorithm, and the ascites occurrence probability and the effusion amount predicted value are obtained; and generating a bleeding point positioning coordinate and a thermodynamic diagram risk grade. And calculating a comprehensive bleeding risk probability and positioning a bleeding area. And generating graded early warning signals and recommending personalized treatment schemes or nursing suggestions. According to the invention, the LSTM-CNN hybrid model, the random forest algorithm and the three-dimensional convolutional neural network are adopted to deeply extract different data features, so that the limitation of single index evaluation is avoided. A causal relationship model is established through the Bayesian network, and the comprehensive risk probability calculation preciseness is improved; the early warning threshold is dynamically adjusted by combining the individual characteristics of the patient, and the traditional problem of easy misjudgment is solved.

The invention relates to the technical field of PTA production control, and discloses a PTA mixed feeding control method and system. The system comprises a mechanism modeling module, a data driving prediction module, a dynamic fusion module, a rolling optimization module and an online learning module. The mechanism modeling module constructs a sub-model based on a PTA mixed reaction kinetic equation, and outputs a theoretical feeding constraint value; the data driving prediction module receives temperature, pressure and raw material flow data collected by the real-time sensor and outputs a real-time feeding prediction value. The dynamic fusion module performs weighted fusion on the two values to generate a fused feeding prediction value; the rolling optimization module generates a feeding parameter adjustment instruction through a rolling optimization algorithm. And when the deviation between the adjusted sensor data and the fused predicted value exceeds a preset threshold value, the online learning module performs online fine adjustment on the parameters of the data driving model. The system combines a reaction mechanism and real-time working condition optimization feeding control, reduces manual intervention, and adapts to a complex production environment.

The invention discloses a new energy uncertainty interval fine modeling method based on ELM-Copula. Comprising the following steps: 1) data reading and preprocessing: obtaining a power prediction value and an actual output value by reading historical operation data of a new energy electric field, performing kernel density estimation on the per-unit power prediction value and the actual output value, and calculating an error absolute value according to the per-unit power prediction value and the actual output value; 2) constructing a dynamic Copula function model; 3) measuring goodness of fit of the model; 4) calculating a confidence interval of a prediction error, analyzing uncertainty of power generation power prediction, and giving a confidence interval of a prediction value; according to the method, a multi-type dynamic Copula model is introduced to construct a dynamic dependency structure between a prediction error and power, the ELM is applied to a post-processing stage of a dynamic Copula prediction interval, a correction coefficient is generated by learning historical deviation characteristics, and an original interval is shrunk, so that the prediction precision and practicability are improved on the premise that the coverage rate is not reduced.

A method for determining the predictive value of a feature may include: (a) performing predictive modeling procedures associated with respective predictive models, wherein performing each modeling procedure includes fitting the associated model to an initial dataset representing an initial prediction problem; (b) determining a first accuracy score of each of the fitted models, representing an accuracy with which the fitted model predicts an outcome of the initial prediction problem; (c) shuffling values of a feature across observations included in the initial dataset, thereby generating a modified dataset representing a modified prediction problem; (d) determining a second accuracy score of each of the fitted models, representing an accuracy with which the fitted model predicts an outcome of the modified prediction problem; and (e) determining a model-specific predictive value of the feature for each of the fitted models based on the first and second accuracy scores of the fitted model.

The invention relates to the technical field of structural health monitoring and reliability evaluation, and discloses a structural component fatigue life prediction method based on a CNN-BiGRU-MHSA fusion model, and the method comprises the steps: collecting multi-source heterogeneous signals, such as strain, vibration and load, calculating crack propagation mechanism characteristics in combination with a fracture mechanics theory, and generating a physical-oriented fatigue life label. The method comprises the following steps: constructing a deep fusion network, extracting local spatial features by using a convolutional neural network, capturing a bidirectional time sequence dependency relationship through a bidirectional gating circulation unit, calculating a global feature weight by using a multi-head self-attention mechanism, and generating weighted fusion features; and finally, a predicted value is output through full-connection layer regression, and parameters are optimized based on physical tags. According to the method, effective combination of a physical mechanism and deep learning is realized, the defect of poor interpretability of a pure data driving model is made up, and the accuracy and robustness of fatigue life prediction under complex working conditions are improved.