50 results about "Univariate" patented technology

The invention discloses a microservice system time sequence anomaly detection method and device, equipment, a medium and a product, and the method comprises the steps: carrying out the preprocessing of original multivariate time sequence data, converting the data into a sliding window sequence, carrying out the channel independent division of the sliding window sequence into a univariate time sequence of a plurality of channels, carrying out the periodic coding, and carrying out the calculation of the univariate time sequence; the method comprises the following steps: generating a periodic coding sequence, carrying out feature analysis on the periodic coding sequence, obtaining local features and global features, carrying out feature fusion on the local features and the global features, decoding, generating reconstructed time sequence data, and carrying out mean square error calculation on the reconstructed time sequence data and original multivariate time sequence data. According to the method, the abnormal score of each time step is obtained, anomaly detection is carried out based on the abnormal scores, semantic differences of different variables in multivariate time sequence data are accurately distinguished, relation characteristics of the different variables in the micro-service system are fully extracted, and abnormal fluctuation in the time sequence of the micro-service system is accurately detected.

The invention discloses a univariate ultra-short-term wind power prediction method based on two-stage trend decomposition, and the method comprises the steps: carrying out the multiple times of trend decomposition of historical wind power time series data, and generating a macroscopic trend component, a mesoscale trend component and a residual component; performing exponential distribution initialization causal convolution kernel extraction on the decomposed macroscopic trend component and the mesoscale trend component to extract multi-scale trend characteristics, and keeping time sequence causality by adopting a proportional normalized adaptive weight; residual modeling is enhanced by adopting a loop reconstruction attention mechanism, and residual component dynamic features are obtained through sequence splicing and double residual connection; and performing linear processing on each component and performing result fusion to generate an ultra-short-term wind power prediction value. And precise prediction of ultra-short-term wind power can be realized.

This invention provides a method and system for univariate time series prediction of process parameters based on knowledge transfer, belonging to the field of process industry technology. The method includes: collecting basic univariate time series information and preprocessing the basic information to obtain training basic information; pre-training a knowledge-embedded DKN network based on the training basic information to obtain denoising parameters; performing secondary training of the knowledge-embedded DKN network based on the denoising parameters and the basic information to obtain a trained network; and performing time series prediction under unstable operating conditions based on the trained network. This invention solves the problems of generalization performance degradation in traditional data-driven univariate time series prediction models under uncertain input scenarios, and the slow convergence and false saturation of traditional DKN networks due to the time-varying characteristics of process industry operating conditions.

The invention relates to a time sequence prediction method based on an adaptive multi-scale Transform, and belongs to the field of time sequence prediction. The method comprises the following steps: carrying out preprocessing and seasonal trend decomposition on multivariable time series data; predicting a trend term based on a linear model; splitting a seasonal item into a univariate sequence, performing fast Fourier transform, and extracting main periodic components; carrying out multi-scale fragment division on each variable time sequence, and distributing a corresponding global Token; on the basis of a cross attention mechanism, correlation among variables is calculated; on the basis of a self-attention mechanism, calculating the correlation of the multi-scale sequence fragments of the variables; constructing a clustering distributor based on the global Token; and performing weighted fusion on the multi-scale output of each variable, and combining with a trend term prediction result to generate a complete prediction result.

The invention provides a plateau rainfall prediction method, system and device based on intelligent similarity correction, and relates to the technical field of short-term climate prediction.The method comprises the steps that mode rainfall data of a month to be predicted, multivariable historical factor data corresponding to the mode rainfall data and global historical annual rainfall error data are obtained, determining standard grid point data of the wind field, the height field and the air temperature and effective historical annual rainfall error data; inputting the standard grid point data and the effective historical annual rainfall error data into a pre-trained error prediction model to obtain a prediction error field output by the error prediction model; and based on the prediction error field, correcting the mode precipitation data to obtain the final predicted precipitation. According to the method, multivariable information is integrated, the nonlinear synergistic effect relationship is mined more deeply, the description capability of the rainfall anomaly driving mechanism of the to-be-predicted region is enhanced, and compared with a traditional single-variable or linear scheme, the accuracy of similar selection and rainfall season prediction is effectively improved.

The invention relates to the technical field of multi-source data fusion, and discloses a multi-source information unified measurement method and system for complex equipment, and the method comprises the steps: calculating a sample overall variance of a data sample set of the complex equipment as a parameter of a Gaussian kernel function, and obtaining an uncertainty model; by utilizing uncertainty modeling, calculating the contribution degree of each data sample to each preset focal element to obtain a contribution degree set, and performing evidence modeling of uncertain parameters to obtain a univariate evidence structure; calculating an evidence distance sequence among the evidence bodies in the test information and the expert information, constructing a credibility index sequence according to the evidence distance sequence, and further obtaining a weighted multi-source data set through weight configuration; fusing the weighted multi-source data set to obtain multi-source uncertainty fused data; and constructing a multi-source information unified measurement model by using the univariate evidence structure and the multi-source uncertainty fusion data. Unified measurement of incomplete, multi-source and related uncertain parameters of a complex system can be realized.

The invention discloses a multi-period non-stationary multi-hydrological variable spatial valuation method, and relates to the technical field of hydrogeology. Comprising the steps of obtaining a univariate time sequence; combining the univariate time sequences of all observation stations in the drainage basin to obtain a multivariate random time sequence; calculating a long-term average level of each observation station through a space-time random function to obtain a long-term trend; subtracting the multivariable random time sequence of each observation station from the corresponding long-term trend to obtain a random residual error; according to the random residual error, determining a covariance function and a variation function between any two observation stations in the drainage basin; and on the basis, a multi-period non-stationary multivariable space valuation model is constructed by combining a multivariable Kriging model theory, and multi-hydrological variable valuation is performed on the drainage basin to be measured. According to the invention, internal correlation between variables can be revealed, and the prediction precision and reliability can be effectively improved.

Some embodiments of the present disclosure provide a manner for an automatic theorem solver to answer a query. Ahead of time, data that supports columns is received. The data is converted to a data structure. Sets of univariate and multivariate morphisms are then determined and the numbers of morphisms in the sets may be reduced in accordance with various metrics. Additionally, the morphisms may be used to generate chains of morphisms. A plurality of equations may be selected for a category. Upon receiving the morphisms, chains of morphisms and selected equations, the automatic theorem solver may be ready to receive a query. The automatic theorem solver may then determine an answer to the query and present the answer.

The disclosed embodiments are directed toward cryptographic methods and variants thereof based on homomorphic encryption enabling the evaluation of real-valued functions on encrypted data, in order to allow carrying out homomorphic processing on encrypted data more broadly and efficiently.

The invention provides a multi-scale time sequence feature extraction method based on a local linear layer. The method comprises the following steps: S1, carrying out standardization processing on an input univariate or multivariate time sequence; s2, setting a plurality of time windows with different sizes, wherein the size of each window corresponds to one scale; s3, under a single scale, retaining a local connection structure in a time dimension by adopting a mask mechanism, and independently extracting local features of time steps in each window in the scale through linear transformation; s4, under a single scale, sharing a linear weight in a channel dimension, and carrying out special extraction on all variables by adopting the same linear weight matrix; and S5, stacking the initial sequence and the feature sequence extracted under each scale along the dimension of the scale to form a multi-scale feature sequence of each variable, and outputting the multi-scale feature sequence to be used by a subsequent model. The invention provides a novel neural network layer named as a local linear layer, local modeling and multi-scale feature extraction can be realized, parameter efficiency and modeling capability can be improved, and the method is suitable for time sequence modeling tasks.

The invention provides a building energy consumption time sequence missing data interpolation method, equipment and a program product, and relates to the technical field of computers. According to the building energy consumption time sequence missing data interpolation method disclosed by the invention, the target input sequence is constructed based on the univariate energy consumption time sequence data of the target building, and the subsequent model training is carried out, so that the target data interpolation model obtained after training is used for carrying out missing data interpolation; the method does not need to depend on multivariable information such as external meteorological data or building attributes, and the applicability in an actual engineering scene is expanded. Moreover, the statistical constraint term is embedded in the total loss function of the adversarial learning joint training, so that the internal structure characteristics and the global mode of the building energy consumption data are effectively maintained, and the interpolation stability under the condition of high missing rate or long missing interval is remarkably improved. Moreover, the calculation of multiple times of tensor expansion and singular value decomposition does not need to be carried out, and the calculation cost of missing data interpolation is reduced.

The invention discloses a univariate time sequence probability prediction method and device and a storage medium, and belongs to the technical field of industrial large models, and the method comprises the steps: carrying out the sliding window segmentation of a univariate time sequence, and obtaining an original data window sequence; performing physical constraint frequency domain enhancement operation on the original data window sequence to obtain an enhanced window sequence; extracting a time domain feature and a frequency domain feature of the enhanced window sequence and splicing the time domain feature and the frequency domain feature into a joint feature vector; inputting the joint feature vector into a pre-constructed Lags-Lama decoder architecture, outputting a hidden state vector, inputting the hidden state vector into a parameterized generalized extreme value distribution GEV output layer, and generating a probability prediction result of a future time point; according to the method, the problems of poor zero sample generalization ability, insufficient long sequence dependence modeling and data distribution offset in univariate time sequence prediction are solved.

A prediction system for a building including a processing circuit including a processor and memory, the memory having instructions stored thereon that, when executed by the processor, cause the processing circuit to receive data relating to a plurality of components, the data indicating performance of the plurality of components, generate, based on the received data, a univariate prediction model and a multivariate prediction model, generate, using the received data, one or more predicted operational parameters for the plurality of components corresponding to a future time period, and execute at least one of the univariate prediction model or the multivariate prediction model on the one or more predicted operational parameters to predict a fault associated with at least one of the plurality of components to occur during the future time period.

This application provides a method, system, medium, and device for real-time prediction of marine engine operating parameters based on the DLinear algorithm. This application replaces explicit sequence decomposition with feature decoupling, simplifying multivariate prediction into multiple univariate prediction problems through feature decomposition, and uses linear mapping instead of complex time series modeling. This method simplifies the Dlinear algorithm, making it suitable for time series prediction of parameters with weak correlations. For parameters with stronger correlations, the original Dlinear algorithm structure can be used. This application collects parameter data from marine engines across different load ranges, establishing a feature decoupling prediction model and a trend / seasonal bilinear prediction model. This avoids the impact of full-load testing on engine lifespan, solves the problem of insufficient effective data samples for modeling, and can quickly capture the nonlinear characteristics of multivariate co-evolution.

The invention relates to a cryptographic method and variants thereof based on homomorphic encryption enabling the evaluation of real-valued functions on encrypted data, in order to allow carrying out homomorphic processing on encrypted data more broadly and efficiently.

The invention relates to a cloud resource automatic scaling system and a cloud resource automatic scaling method. The system comprises: a data acquisition module configured to be used for collecting multivariable historical index data and current state data of one or more container instances from one or more container instances; and a prediction module which is in communication connection with the data acquisition module and is configured to receive the multivariable historical index data and the current container instance state, predict a single-variable future index based on the multivariable historical index data, and send the predicted single-variable future index to the data acquisition module. Generating a scaling strategy based on the future index of the single variable and the current state data; and the scaling control module is in communication connection with the prediction module and is configured to receive the scaling strategy and generate a control instruction for executing scaling operation on the one or more container instances based on the scaling strategy. According to the invention, excellent prediction precision and perspectiveness can be provided.

Some embodiments of the present disclosure provide a manner for an automatic theorem solver to answer a query. Ahead of time, data that supports columns is received. The data is converted to a data structure. Sets of univariate and multivariate morphisms are then determined and the numbers of morphisms in the sets may be reduced in accordance with various metrics. Additionally, the morphisms may be used to generate chains of morphisms. A plurality of equations may be selected for a category. Upon receiving the morphisms, chains of morphisms and selected equations, the automatic theorem solver may be ready to receive a query. The automatic theorem solver may then determine an answer to the query and present the answer.

This invention relates to the field of anomaly detection technology, specifically to a multi-industry-adaptive technical foundation detection method and system. It is used to perform basic state detection on multivariate operational data generated by equipment, components, process units, or production line links in multiple industry scenarios. The method involves: offline, acquiring historical datasets with state annotations, establishing a candidate detection algorithm library, and training, validating, and evaluating the performance of each candidate algorithm to form associated metadata; extracting univariate statistical features, multivariate correlation features, anomaly distribution features, and overall structural features from each historical dataset, and obtaining a unified meta-feature representation through screening, aggregation, and embedding transformation; training a multi-output performance prediction model using the unified meta-feature representation as input and the candidate algorithm performance vector as output; online, extracting the meta-features to be detected from the industry object to be detected and inputting them into the model to obtain the predictive detection performance of each candidate algorithm, selecting a single target detection component or determining multiple algorithms and their integrated weights, and outputting at least one of anomaly score and state label to obtain the anomaly state detection result. This invention can solve the problem that anomaly detection algorithms are difficult to adapt quickly and accurately based on human experience when there are diverse types of industry objects, large differences in operational data structures, and frequent changes in operating conditions in multi-industry scenarios.

A method (500) includes receiving a query (20) to identify anomalies in multivariate time series data (152), which includes endogenous variables (152D) and exogenous variables (152X). The method includes determining the effect of exogenous variables on endogenous variables (164). The method includes identifying univariate time series data (152U) and using the univariate time series data to train one or more models (172). The method includes determining the expected value (152E) of a given time series value and determining the difference between the expected value and the given time series value. The method includes determining that the difference between the expected value and the given time series value of a particular time series value satisfies a threshold. In response, the method includes determining that a particular time series data value is anomalous and reporting the anomalous value to a user (12) (152A).

The invention belongs to the technical field of data analysis and statistical process control, and discloses an abnormal value screening and control limit determining method for finite single variable reliability data based on a normal quantile, which specifically comprises the following steps: acquiring a finite single variable reliability data set to be analyzed; sorting the finite single variable reliability data sets, and calculating an empirical cumulative distribution function value of each data point based on the sorted finite single variable reliability data sets; calculating a standard normal distribution quantile corresponding to each empirical cumulative distribution function value; and establishing a function relation model between a standard normal distribution quantile and an observed value of the sorted finite univariate reliability data set, substituting a target cumulative distribution function value into the function relation model, and carrying out extrapolation calculation to obtain an extreme quantile, so that normal, non-normal and even multi-modal distribution data can be effectively processed.

Computer vision systems and methods for detecting anomalous building models are provided. The systems and methods can detect anomalies in building models using one or more of an independent univariate Gaussian algorithm, a multivariate Gaussian algorithm, a combination of a multivariate Gaussian algorithm for continuous features and a frequency histogram algorithm for discrete features, and / or a bin frequency model. The system automatically processes computerized models to determine anomalies, and indicates whether the models are accurate and whether correction is required.

The invention discloses a horizontal and vertical seismic oscillation combined selection method based on a multivariable return period, and the method comprises the steps: determining the annual exceeding probability of a site, namely a target return period, and selecting a condition strength parameter and a target horizontal and vertical strength parameter; according to the basic information of the site, obtaining a random seismic directory of the site through Open Quake, and determining the occurrence rate of an earthquake year; constructing multivariate Gaussian mixture distribution based on a site random seismic directory; randomly extracting a large number of simulation spectrums from the multivariate Gaussian mixture distribution; calculating a multivariable recurrence period of each simulation spectrum, and selecting the simulation spectrum with the recurrence period within + / -5% of the target recurrence period as a target spectrum; after the target spectrum distribution is known, a seismic oscillation record set meeting the error requirement is selected from the target database; different from a single-variable return period based on a single-dimensional strength parameter, a horizontal and vertical seismic oscillation selection method based on a multivariable return period can select a group of horizontal and vertical seismic oscillation records with danger consistency.

The invention discloses a multivariable time sequence prediction method based on independent component analysis and a time sequence basic model, and the method comprises the steps: firstly carrying out the independent component analysis demixing processing of original multivariable time sequence data, obtaining a plurality of statistical independent source signals, and recording a mixed matrix, a mean value parameter and a whitening matrix generated in the demixing process; respectively inputting each independent source signal into a pre-trained time sequence basic model to carry out univariate time sequence prediction; and finally, sequentially carrying out whitening inverse transformation and reverse centralization processing on a prediction result by utilizing the recorded hybrid matrix, whitening matrix and mean value parameters, and reconstructing a multivariable prediction result consistent with the original data distribution. The method does not need to carry out any modification or task specific fine adjustment on the structure or parameters of the time sequence basic model, completely retains the universal time sequence modeling capability obtained in the pre-training stage, and can be directly applied to a target industrial scene.

A method, system, and computer program product for an interpretable, feature-based post-hoc black box explainer for univariate time series forecasters are provided. The method receives a set of time series forecasting predictions. The set of time series forecasting predictions are generated from a set of black-box models trained with an initial data set. The method generates a set of features based on at least a portion of the initial data set. A set of surrogate models are trained based on the set of time series forecasting predictions and at least a portion of the set of features. A subset of surrogate models is selected. Based on the subset of surrogate models, the method generates one or more explanation outputs for time series forecasting predictions of the set of black-box models.

The invention relates to a computer-implemented method (200) for training an artificial intelligence-based model (M) and to a computer-implemented method (600) for supporting an operation on a patient's eye, comprising such a model (M), which is used to calculate and / or provide refractive output data (933) on the basis of input data (931), and in particular to create nomograms. Nomograms from the prior art may either have an accuracy in need of improvement or, if the accuracy is improved, may generate implausible profiles of the nomogram. According to the invention, more accurate nomograms based on the output data (933) are made possible without implausible states. This is made possible by a cost function (915) which comprises a non-linear output value (917), representing a measure of the prediction error, of a univariate linear regression function and at least one non-linear penalty term (919). The invention further relates to a trained model (M) or a weighting matrix (923) of such a model (M), to a computer program product, to a data processing device (800), to a computer-readable medium, and to a laser therapy device.

The disclosed embodiments are directed toward cryptographic methods and variants thereof based on homomorphic encryption enabling the evaluation of real-valued functions on encrypted data, in order to allow carrying out homomorphic processing on encrypted data more broadly and efficiently.

The invention discloses a time sequence prediction method based on a visual pre-training model, and the method comprises the steps: carrying out the interpolation and normalization processing of input data, and dividing the data into a univariable sequence and a multivariable sequence; converting the univariate sequence into an image through segmentation, rendering and alignment operations, and reconstructing the converted image by using the pre-training MAE to obtain a predicted value corresponding to the univariate sequence; converting the multivariable sequence into an image through clustering, rendering and alignment operations, reconstructing the converted image by using the pre-training MAE, and restoring the reconstructed image to obtain a predicted value corresponding to the multivariable sequence; and respectively endowing different learnable weight parameters for the predicted value corresponding to the single-variable sequence and the predicted value corresponding to the multivariable sequence, and dynamically integrating time features and space features. According to the method, the cooperative relationship among the multivariate sequences is considered, the influence of interpolation errors on the model can be reduced, and the prediction performance is improved.

A computer-implemented method (200) for training an artificial intelligence-based model (M) and a computer-implemented method (600) for supporting surgery on a patient's eye are provided, comprising such a model (M) which serves to calculate and / or provide refractive output data (933) based on input data (931), and in particular to generate nomograms. Prior art nomograms may either have insufficient accuracy or, if the accuracy is improved, generate implausible nomogram progressions. According to the invention, more accurate nomograms based on the output data (933) are enabled without implausible states.This is made possible by a cost function (915) comprising a nonlinear output value (917) of a univariate linear regression function representing a measure of the prediction error and at least one nonlinear penalty term (919). Furthermore, a trained model (M) or a weighting matrix (923) of such a model (M), as well as a computer program product, a data processing device (800), a computer-readable medium, and a laser therapy device are described.

The present invention discloses a method, apparatus, device, medium and product for detecting time series anomalies in a microservice system. The method comprises: preprocessing original multivariate time series data and converting it into a sliding window sequence, dividing the sliding window sequence into a plurality of univariate time series channels by channel-independent division, performing period encoding, generating a periodic coding sequence, performing feature analysis on the periodic coding sequence, obtaining local features and global features, fusing the local features with the global features, decoding the features, generating reconstructed time series data, calculating the mean square error between the reconstructed time series data and the original multivariate time series data, obtaining anomaly scores for each time step, performing anomaly detection based on the anomaly scores, accurately distinguishing semantic differences between different variables in the multivariate time series data, fully extracting relational features between different variables in the microservice system, and accurately detecting abnormal fluctuations in the time series of the microservice system.